American Journal of PUBLIC HEALTH Contents: July 1 2005, Volume 95, Issue 7 LETTERS

American Journal of PUBLIC HEALTH Contents: July 1 2005, Volume 95, Issue 7 LETTERS:

G. David Batty and Ian J. Deary HEALTH COMMUNICATION, INTELLIGENCE, AND HEALTH DIFFERENTIALS Am J Public Health 2005 95: 1088,. Vicki S. Freimuth and Sandra Crouse Quinn FREIMUTH AND QUINN RESPOND Am J Public Health 2005 95: 1089, 10.2105/AJPH.2005.063396. Lois A. Fingerhut, James Harrison, Yvette Holder, Birthe Frimodt-Møller, Susan Mackenzie, Saakje Mulder, and Ian Scott ADDRESSING THE GROWING BURDEN OF TRAUMA AND INJURY IN LOW- AND MIDDLE-INCOME COUNTRIES Am J Public Health 2005 95: 1089-1090, Herbert K. Abrams LINKING HEALTH TO SOCIAL JUSTICE Am J Public Health 2005 95: 1090, Iman A. Nuwayhid NUWAYHID RESPONDS Am J Public Health 2005 95: 1090-1091, Anne-Emanuelle Birn and Klaudia Dmitrienko THE WORLD BANK: GLOBAL HEALTH OR GLOBAL HARM? Am J Public Health 2005 95: 1091-1092, Jennifer Prah Ruger RUGER RESPONDS Am J Public Health 2005 95: 1092, Erika Bácskai, Ágnes Tallár, and József Gerevich DRINKING AND INTIMATE PARTNER VIOLENCE IN A CHANGING SOCIETY Am J Public Health 2005 95: 1092-1093, Xiao Xu and Jacquelyn Campbell XU AND CAMPBELL RESPOND Am J Public Health 2005 95: 1093, ERRATUM:

ERRATUM Am J Public Health 2005 95: 1093 ERRATUM Am J Public Health 2005 95: 1093. EDITOR'S CHOICE:

Bernard M. Dickens The Challenges and Opportunities of Ethics Am J Public Health 2005 95: 1094. EDITORIALS:

Anne-Emanuelle Birn and Natalia Molina In the Name of Public Health Am J Public Health 2005 95: 1095-1097, Jake McKinstry Using the Past to Step Forward: Fetal Alcohol Syndrome in the Western Cape Province of South Africa Am J Public Health 2005 95: 1097-1099, Jorge Rosenthal, Arnold Christianson, and Jose Cordero Fetal Alcohol Syndrome Prevention in South Africa and Other Low- Resource Countries Am J Public Health 2005 95: 1099-1101, FIELD ACTION REPORT: gfedc Martha W. Moon, JoAnne K. Henry, Karen Connelly, and Phyllis Kirsch Public Health Nurses for Virginia’s Future: A Collaborative Project to Increase the Number of Nursing Students Choosing a Career in Public Health Nursing Am J Public Health 2005 95: 1102-1105, COMMENTARY: gfedc Gordon Marc le Roux "Whistle While You Work": A Historical Account of Some Associations Among Music, Work, and Health Am J Public Health 2005 95: 1106-1109, HEALTH POLICY AND ETHICS: gfedc Jane Galvão Brazil and Access to HIV/AIDS Drugs: A Question of Human Rights and Public Health Am J Public Health 2005 95: 1110-1116, gfedc Moïse Desvarieux, Roland Landman, Bernard Liautaud, Pierre-Marie Girard for the INTREPIDE Initiative In Global Health Antiretroviral Therapy in Resource-Poor Countries: Illusions and Realities Am J Public Health 2005 95: 1117-1122, gfedc Neal Dickert and Jeremy Sugarman Ethical Goals of Community Consultation in Research Am J Public Health 2005 95: 1123-1127, PUBLIC HEALTH THEN AND NOW: gfedc Alexandra Minna Stern STERILIZED in the Name of Public Health: Race, Immigration, and Reproductive Control in Modern California Am J Public Health 2005 95: 1128-1138, CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK: gfedc Howard I. Kushner and Claire E. Sterk The Limits of Social Capital: Durkheim, Suicide, and Social Cohesion Am J Public Health 2005 95: 1139-1143, gfedc Rebecca M. Young and Ilan H. Meyer The Trouble With "MSM" and "WSW": Erasure of the Sexual-Minority Person in Public Health Discourse Am J Public Health 2005 95: 1144-1149, gfedc L. Gary Hart, Eric H. Larson, and Denise M. Lishner Rural Definitions for Health Policy and Research Am J Public Health 2005 95: 1149-1155, gfedc Alicia Ely Yamin The Right to Health Under International Law and Its Relevance to the United States Am J Public Health 2005 95: 1156-1161, PUBLIC HEALTH MATTERS: gfedc Alan Berkman, Jonathan Garcia, Miguel Muñoz-Laboy, Vera Paiva, and Richard Parker A Critical Analysis of the Brazilian Response to HIV/AIDS: Lessons Learned for Controlling and Mitigating the Epidemic in Developing Countries Am J Public Health 2005 95: 1162-1172,

gfedc Laura J. McGough, Steven J. Reynolds, Thomas C. Quinn, and Jonathan M. Zenilman Which Patients First? Setting Priorities for Antiretroviral Therapy Where Resources Are Limited Am J Public Health 2005 95: 1173-1180, RESEARCH AND PRACTICE: gfedc Kate E. Pickett, Jessica Mookherjee, and Richard G. Wilkinson Adolescent Birth Rates, Total Homicides, and Income Inequality In Rich Countries Am J Public Health 2005 95: 1181-1183, gfedc Christine Pace, Ambrose Talisuna, David Wendler, Faustin Maiso, Fred Wabwire-Mangen, Nathan Bakyaita, Edith Okiria, Elizabeth S. Garrett-Mayer, Ezekiel Emanuel, and Christine Grady Quality of Parental Consent in a Ugandan Malaria Study Am J Public Health 2005 95: 1184-1189, gfedc Philip A. May, J. Phillip Gossage, Lesley E. Brooke, Cudore L. Snell, Anna- Susan Marais, Loretta S. Hendricks, Julie A. Croxford, and Denis L. Viljoen Maternal Risk Factors for Fetal Alcohol Syndrome in the Western Cape Province of South Africa: A Population-Based Study Am J Public Health 2005 95: 1190-1199, gfedc Karen M. Emmons, Ann M. Stoddard, Robert Fletcher, Caitlin Gutheil, Elizabeth Gonzalez Suarez, Rebecca Lobb, Jane Weeks, and Judy Anne Bigby Cancer Prevention Among Working Class, Multiethnic Adults: Results of the Healthy Directions–Health Centers Study Am J Public Health 2005 95: 1200-1205, gfedc Maria Melchior, Nancy Krieger, Ichiro Kawachi, Lisa F. Berkman, Isabelle Niedhammer, and Marcel Goldberg Work Factors and Occupational Class Disparities in Sickness Absence: Findings From the GAZEL Cohort Study Am J Public Health 2005 95: 1206-1212, gfedc Gordon S. Smith, Helen M. Wellman, Gary S. Sorock, Margaret Warner, Theodore K. Courtney, Glenn S. Pransky, and Lois A. Fingerhut Injuries at Work in the US Adult Population: Contributions to the Total Injury Burden Am J Public Health 2005 95: 1213-1219, gfedc Alison M. Trinkoff, Meg Johantgen, Carles Muntaner, and Rong Le Staffing and Worker Injury in Nursing Homes Am J Public Health 2005 95: 1220-1225,

gfedc Larkin L. Strong and Frederick J. Zimmerman Occupational Injury and Absence From Work Among African American, Hispanic, and Non-Hispanic White Workers in the National Longitudinal Survey of Youth Am J Public Health 2005 95: 1226-1232, gfedc Donald C. Cole, Selahadin Ibrahim, and Harry S. Shannon Predictors of Work-Related Repetitive Strain Injuries in a Population Cohort Am J Public Health 2005 95: 1233-1237, gfedc Philip A. May, Patricia Serna, Lance Hurt, and Lemyra M. DeBruyn Outcome Evaluation of a Public Health Approach to Suicide Prevention in an American Indian Tribal Nation Am J Public Health 2005 95: 1238-1244, gfedc Debbie A Lawlor, G. David Batty, Susan M.B. Morton, Heather Clark, Sally Macintyre, and David A. Leon Childhood Socioeconomic Position, Educational Attainment, and Adult Cardiovascular Risk Factors: The Aberdeen Children of the 1950s Cohort Study Am J Public Health 2005 95: 1245-1251, gfedc George A. Kaplan, Kristine Siefert, Nalini Ranjit, Trivellore E. Raghunathan, Elizabeth A. Young, Diem Tran, Sandra Danziger, Susan Hudson, John W. Lynch, and Richard Tolman The Health of Poor Women Under Welfare Reform Am J Public Health 2005 95: 1252-1258, gfedc Jesse B. Milby, Joseph E. Schumacher, Dennis Wallace, Michelle J. Freedman, and Rudy E. Vuchinich To House or Not to House: The Effects of Providing Housing to Homeless Substance Abusers in Treatment Am J Public Health 2005 95: 1259-1265, gfedc Catherine Mathews, Sally J. Guttmacher, Alan J. Flisher, Yolisa Mtshizana, Andiswa Hani, and Merrick Zwarenstein Written Parental Consent in School-Based HIV/AIDS Prevention Research Am J Public Health 2005 95: 1266-1269, MARKETPLACE:

MARKETPLACE Am J Public Health 2005 95: 1271-1274. JOB OPPORTUNITIES:

JOB OPPORTUNITIES Am J Public Health 2005 95: 1275-1281.

 LETTERS

HEALTH COMMUNICATION, IQ reflects an individual’s ability to learn, About the Authors INTELLIGENCE, AND HEALTH reason, and solve problems.7 Therefore, plausi- The authors are with the Department of Psychology, Uni- versity of Edinburgh, Edinburgh, Scotland. G. David Batty DIFFERENTIALS ble mechanisms underlying these observations is also with the Medical Research Council Social and that have particular relevance to health com- Public Health Sciences Unit, University of Glasgow, Glas- In a recent editorial, Freimuth and Quinn1 munications—defined by Freimuth and Quinn1 gow, Scotland. Requests for reprints should be sent to G. David Batty, discussed the contribution that health com- as the study of methods that influence deci- MRC Social and Public Health Sciences Unit, University munication might make to eliminating health sions that enhance health—concern the concep- of Glasgow, 4 Lilybank Gardens, Glasgow, UK, G12 8RZ disparities. The mechanisms underlying the tualization of self-health care asaset of cogni- (e-mail: [email protected]). doi:10.2105/AJPH.2005.063370 recently suggested role of intelligence (de- tive tasks. These tasks include the optimal noted here by IQ) in the etiology of several interpretation of health prevention messages important health outcomes are pertinent to and, among persons with existing chronic Acknowledgments G. David Batty is the recipient of a Wellcome Advanced this debate. illness, improved disease management.6 Training Fellowship (071954/Z/03/Z); Ian Deary In long-term follow-up studies, persons In terms of the former, Scottish children holds a Royal Society-Wolfson Research Merit Award. with low scores on written IQ tests adminis- who scored high on tests of intelligence in tered in childhood and early adulthood have 1932 were more likely than others to stop References 8 been found to have elevated rates of all-cause smoking as adults in the 1960s and 1970s. 1. Freimuth VS, Quinn SC. The contributions of mortality,2 coronary heart disease,3 some This observation may be attributable to the health communication to eliminating health disparities. cancers,4 and psychiatric disorders5 than their differential interpretation of antismoking ad- Am J Public Health. 2004;94:2053–2055. higher-scoring counterparts in later life. For vice that appeared for the first time during 2. Whalley LJ, Deary IJ. Longitudinal cohort study of childhood IQ and survival up to age 76. BMJ. the most examined health outcome—all-cause this period. With regard to disease manage- 2001;322:819. mortality—these effects are strong (exceeding ment, the daily management of one’s illness is 3. Batty GD, Mortensen E, Nybo Andersen A-M, those of other psychological characteristics, cognitively demanding; it requires ongoing Osler M. Childhood intelligence in relation to adult cor- such as personality disposition, and on a par learning and independent decisionmaking. onary heart disease and stroke risk: evidence from a with the effects of established risk factors such Persons with low levels of education or health Danish birth cohort study. Paediatr Perinat Epidemiol. In press. as raised blood pressure, social disadvantage, literacy, both of which are related to IQ,7 are 4. Hart CL, Taylor MD, Davey Smith G, et al. Child- and obesity), incremental (apparent across the less likely to appreciate when their condi- hood IQ, social class, deprivation, and their relation- full range of IQ scores, not just confined to tions require medical attention, to be aware ships with mortality and morbidity risk in later life: below-average scores), consistent across a of the appropriate actions to take when they prospective observational study linking the Scottish Mental Survey 1932 and the Midspan studies. Psycho- number of study populations and research do realize they need attention, and to cor- som Med. 2003;65:877–883. groups, and seemingly independent of child- rectly comprehend instructions for self-med- 5. Batty GD, Mortensen EL, Osler M. Childhood in- 6 9 hood social circumstances. ication. They may also fail to seek medical telligence in relation to later psychiatric disorder: evi- advice promptly after a significant illness dence from a Danish birth cohort study. Br J Psychiatry. episode (e.g., myocardial infarction) and re- In press. ceive treatment at a facility most appropriate 6. Batty GD, Deary IJ. Early life intelligence and Letters to the editor referring to a recent Journal 9 adult health. BMJ. 2004;329:585–586. article are encouraged up to 3 months after the to their clinical requirements. Health care advice, treatment protocols, 7. Neisser U, Boodoo G, Bouchard TJ Jr, et al. Intelli- article’s appearance. By submitting a letter to gence: knowns and unknowns. Am Psychol. 1996;51: the editor, the author gives permission for its and disease prevention information may ex- 77–101. publication in the Journal. Letters should not ceed the intellectual capacity of some peo- 8. Taylor MD, Hart CL, Davey Smith G, et al. 9,10 duplicate material being published or submitted ple. If this is the case, and since highly Childhood mental ability and smoking cessation in elsewhere. The editors reserve the right to edit simplified advice might substantially reduce adulthood: prospective observational study linking the and abridge letters and to publish responses. Scottish Mental Survey 1932 and the Midspan stud- the intended impact of the advice, proactive Text is limited to 400 words and 10 refer- ies. J Epidemiol Community Health. 2003;57: ences. Submit online at www.ajph.org for involvement of health care providers in the 464–465. immediate Web posting, or at submit.ajph.org provider–client interaction, as advocated by 9. Gottfredson LS, Deary IJ. Intelligence predicts for later print publication. Online responses Freimuth and Quinn,1 might reduce health health and longevity, but why? Curr Dir Psychol Sci. are automatically considered for print differentials. 2004;13:1–5. publication. Queries should be addressed to 10.Gottfredson LS. Intelligence: is it the epidemiologists’ elusive “fundamental cause” of social class in- the department editor, Jennifer A. Ellis, PhD, G. David Batty, PhD, MSc at [email protected]. equalities in health? J Pers Soc Psychol. 2004;86: Ian J. Deary, PhD, FRCPE 174–199.

1088 | Letters American Journal of Public Health | July 2005, Vol 95, No. 7  LETTERS 

ADDRESSING THE GROWING BURDEN OF TRAUMA AND INJURY IN LOW- AND MIDDLE-INCOME COUNTRIES

In “Addressing the Growing Burden of Trauma and Injury in Low- and Middle- Income Countries,” Hofman et al. report on a meeting sponsored by the National Institutes of Health (NIH).1 The article acknowledges some work by the Centers for Disease Con- trol and Prevention (CDC) but neglects to acknowledge an international activity sponsored by the CDC’s National Center for Health Statistics and cofunded by NIH’s National Institute of Child Health and Human Development. The International Collaborative Effort (ICE) on Injury Statistics has been ongoing since 1994.2 Its primary goal is to provide a forum for international exchange and collaboration among injury researchers who develop and promote international standards for injury data collection and analysis. A secondary goal is to produce products of the highest quality to facilitate the comparability and improved quality of injury data. It is this second goal that is particularly relevant. Epidemiology is listed by Hofman et al. as one of the most critical areas of research: “A broad spectrum of epidemiological and surveillance research is needed to enhance knowledge. . . .”1(p15) Several ICE projects are directly related to this area, including the development of injury indicators and the various reporting frameworks for standardized comparisons of both fatal and nonfatal injury outcomes (see http://www.cdc.gov/nchs/ about/otheract/ice/projects.htm). A discussion of the International Classifica- tion of External Causes of Injury (ICECI) would perhaps have been relevant for the NIH meeting. The ICECI was adopted into the family of classifications by the World Health Organization (WHO) in October 2003. In the 1980s and early 1990s efforts were made to improve upon the International Classification of Diseases classification of external causes of injury for the purposes of injury prevention. Under the auspices of WHO, injury professionals from all over the world worked to develop ICECI, an improved tool for capturing data on the circumstances in which injuries

July 2005, Vol 95, No. 7 | American Journal of Public Health Letters | 1089  LETTERS 

occur. The most recent version is available on the ICECI Web site.3 Countries for which ICECI-based surveillance was reported at the Safety 2004 Conference in Vienna, Austria, include Nicaragua, El Salvador, and Colombia. WHO has also published injury surveillance guidelines (compatible with the ICECI) that are specifically intended for use in settings where resources are scarce.4 The next meeting of the ICE on Injury Statistics will focus on injury statistics in Latin America and will take place in Mexico.

Lois A. Fingerhut, MA James Harrison, MPH Yvette Holder, MPH Birthe Frimodt-Møller, MD Susan Mackenzie, PhD Saakje Mulder, PhD Ian Scott

About the Authors The authors are members of the ICE on Injury Statistics. Lois Fingerhut is with the Centers for Disease Control and Prevention, National Center for Health Statistics, Hyatts- ville, Md. James Harrison is with the Australian Institute of Health and Welfare, National Injury Surveillance Unit, Adelaide, Australia. Yvette Holder is with International Biostatistics and Information Services, Castries, St Lucia. Birthe Frimodt-Møller is with the National Institute of Public Health, Copenhagen, Denmark. Susan Mackenzie is with the Injury and Child Maltreatment Section, Health Canada, Ottawa. Saakje Mulder is with the Consumer Safety Institute, Amsterdam, The Netherlands. Ian Scott is with the Injury and Violence Prevention office, WHO, Geneva, Switzerland. Requests for reprints should be sent to Lois A. Fingerhut, MA, NCHS/OAE, 3311 Toledo Rd, Room 6316, Hyatts- ville, MD 20782 (e-mail: [email protected]). doi:10.2105/AJPH.2005.064469

References 1. Hofman K, Primack A, Keusch G, Hrynkow S. Ad- dressing the growing burden of trauma and injury in low- and middle-income countries. Am J Public Health. 2005;95:13–17. 2. Fingerhut LA. International Collaborative Effort on Injury Statistics: 10 year review. Inj Prev. 2004;10: 264–267. 3. International Classification of External Causes of Injury (ICECI). Version 1.2. Available at: http://www. iceci.org. Accessed May 10, 2005. 4. Holder Y, Peden M, Krug E, Lund J, Gururaj G, Kobusingye O, eds. Injury Surveillance Guidelines. 2001. Available at: http://www.who.int/violence_injury_ prevention/publications/surveillance/surveillance_ guidelines/en. Accessed May 10, 2005.

1090 | Letters American Journal of Public Health | July 2005, Vol 95, No. 7  LETTERS 

most notably structural adjustment programs and interviews, to which a former speech- (which have denuded the social welfare infra- writer for the World Bank president might structure of developing countries in areas have sought access, is disappointing. In failing such as housing, education, health services, to convert the price tags of projects into infla- subsidies, and family transfers); and the im- tion-adjusted dollars—a surprising oversight pact on health of the Bank’s newfound focus for a health economist—Ruger underestimates on the health sector. the impact of past World Bank activities. Ruger repeats the insider’s lament that Overall, this one-sided article fails to eluci- lending policies were perennially subject to date the powerful political and economic the exigencies of Wall Street bondholders, forces motivating World Bank policies and but she overlooks the far larger question of activities and does not provide the carefully the nature and distribution of power at the researched historical analysis we have come World Bank. With votes directly related to to expect from “Public Health Then and shareholding size, World Bank decision- Now” articles. making is profoundly undemocratic, favoring elite interests within wealthy nations (the Anne-Emanuelle Birn, ScD, MA United States alone commands 16.4% of Klaudia Dmitrienko, MA, MHSc votes within the Bank). Any account of the Bank’s evolution ought to consider the impact About the Authors of this governance structure on the roles and Anne-Emanuelle Birn is with the Department of Public activities that the Bank adopts. Health Sciences, University of Toronto, Toronto, Ontario, According to both internal and external where Klaudia Dmitrienko is a doctoral candidate. Requests for reprints should be sent to Anne-Emanuelle observers, the neoliberal policies advocated Birn, ScD, MA, Department of Public Health Sciences, by the Bank and its sister institutions begin- University of Toronto, 1st Floor, McMurrich Bldg, 12 ning in the 1980s have provoked or wors- Queen’s Park Crescent W, Toronto, ON M5S 1A8, Can- ada (e-mail: [email protected]). ened dire economic conditions—and the at- doi:10.2105/AJPH.2005.064733 tendant health effects, such as increased rates of malaria, HIV/AIDS, and tuberculosis—in References 2–5 much of the developing world. This “role 1. Ruger JP. The changing role of the World Bank in of the World Bank in global health” remains global health. Am J Public Health. 2005;95:60–70. unaddressed by Ruger. 2. Musgrove P. The economic crisis and its impact on health and health care in Latin America and the Indeed, the negative impact of structural Caribbean. Int J Health Serv. 19 87;17(3):411–441. adjustment programs on health conditions in 3. Stiglitz JE. Globalization and Its Discontents. New THE WORLD BANK: GLOBAL HEALTH developing countries helped spur the Bank’s York, NY: WW Norton; 2002. OR GLOBAL HARM? focus on health in the late 1980s.6 With its 4. Gloyd S. Sapping the poor: the impact of struc- double-entendre title, the Bank’s influential tural adjustment programs. In: Fort M, Mercer MA, Gish O, eds. Sickness and Wealth: The Corporate Assault In recent years the World Bank has be- 19 93 report Investing in Health hailed the im- on Global Health. Cambridge, Mass: South End Press; come “the world’s largest external funder of portance of health to development while ad- 2004:43–54. health.”1(p61) According to Ruger, this situation vocating the privatization of health services.7 5. Bassett MT, Bijlmakers L, Sanders DM. Profes- reflects the Bank’s increased sensitivity to But the Bank’s approach to health sector sionalism, patient satisfaction and quality of health care: experience during Zimbabwe’s structural adjust- poverty and its growing sophistication— lending has exacerbated poor health outcomes ment programme. Soc Sci Med. 19 97;45(12): beginning under the leadership of US Secre- by reducing access to health services for those 18 45–1852. tary of Defense turned World Bank President unable to pay for care in newly privatized sys- 6. Banerji D. A simplistic approach to health policy (1968–1981) Robert McNamara—about de- tems, which focus on cost recovery.8,9 Recent analysis: the World Bank team on the Indian health sector. Int J Health Serv. 1994;24(1):151–159. velopment theory and practice. Such an untargeted programs aimed at the poorest ignore 7. World Bank. World Development Report 1993: In- critical portrayal befits the World Bank’s own structural deficiencies in social services. vesting in Health. Available at: http://www-wds.worldbank. Web site (a major source for Ruger’s article), In sum, Ruger portrays the Bank’s increas- org/default.jsp?site=wds. Accessed May 2, 2005. but Journal readers should expect more. ing involvement in the health sector as un- 8. Russell S, Gilson L. User fee policies to promote Missing from this officialist version are dis- problematic. Critics are dismissed as a hand- health service access for the poor: a wolf in sheep’s clothing? Int J Health Serv. 19 97;27(2):359–379. cussions of the Bank’s undemocratic gover- ful of cranks rather than as serious academic 10–12 9. Laurell AC, Arellano OL. Market commodities nance and decisionmaking structures; the un- and policy researchers. The author’s re- and poor relief: the World Bank proposal for health. toward human effects of longstanding World liance on official Web sites and published his- Int J Health Serv. 1996;26(1):1–18. Bank pro-privatization policies and practices, tories rather than internal memos, archives, 10. Kim JY, Millen JV, Irwin A, Gershman J, eds.

July 2005, Vol 95, No. 7 | American Journal of Public Health Letters | 1091  LETTERS 

Dying for Growth: Global Inequality and the Health of the Poor. Monroe, Me: Common Courage Press; 2000. 11. Navarro V. Whose globalization? Am J Public Health. 1998;88:742–743. 12.Brand H. The World Bank, the Monetary Fund, and poverty. Int J Health Serv. 1994;24(3):567–578.

1092 | Letters American Journal of Public Health | July 2005, Vol 95, No. 7  LETTERS 

ERRATA In: American Journal of Public Health. 2003;93(4). The photographer’s name was omitted from the cover caption. The cover image was photographed by Suzanne Plunkett.

In: Wolfe WS, Campbell CC, Frongillo EA Jr, Haas JD, Melnik TA. Identifying overweight school children in New York State: prevalence estimates and related characteristics. Am J Public Health. 1994;84:807–813.

A table was printed with incorrect data. Corrected values for TAB LE 2—Sample Distri- butions of Physical Indices, Compared With Age- and Sex-Specific National Refer- ence Percentiles are: Second Graders Fifth Graders (n = 960), (n = 837), Indices and Percentiles % %

Body mass indexa >95th 11.2 8.3 Triceps skinfolda >95th 9.3 7.6

aCompared with the second National Health and Nutrition Examination Survey (NHANES II).15

The article’s conclusions are unaffected by these corrections.

July 2005, Vol 95, No. 7 | American Journal of Public Health Letters | 1093 AMERICAN JOURNAL OF PUBLIC HEALTH

EDITOR-IN-CHIEF Mary E. Northridge, PhD, MPH TECHNICAL DEPUTY EDITOR Jennifer A. Ellis, PhD  EDITOR’S CHOICE FEATURE EDITOR Gabriel N. Stover, MPA ASSOCIATE EDITORS Mary Bassett, MD, MPH Michael R. Greenberg, PhD Sofia Gruskin, JD, MIA Deborah Holtzman, PhD, MSW Said Ibrahim, MD, MPH The Challenges and Opportunities of Ethics Sherman A. James, PhD Stewart J. Landers, JD, MCP Robert Sember Journal readers will not need to be reminded ethic to do no harm, in requiring positive acts Stella M.Yu, ScD, MPH of the responsibility of public health practi- of good when harm is avoidable simply by Roger Vaughan, DrPH, MS tioners and scientists to conduct their prac- inaction. Fourth is the ethical principle of jus- INTERNATIONAL ASSOCIATE EDITORS tices ethically. Nor need they be alarmed that tice, particularly distributive justice directed Daniel Tarantola, MD (Geneva, Switzerland) ethical review of their practices might disclose to the just allocation of risks and benefits. Cesar Gomes Victora, MD, PhD (Pelotas, Brazil) histories of ethical misconduct. Both routine The challenge to public health practitioners, DEPARTMENT EDITORS John Colmers, MPH Government, Politics, and Law and exceptional initiatives in public health are accepted by the American Public Health Asso- usually undertaken ethically. ciation (http://www.apha.org/codeofethics/ Elizabeth Fee, PhD, and Theodore M. Brown, PhD Images of Health Public health practice may nevertheless be ethics.htm), is to apply these principles not just Public Health Then and Now enriched by perceptions of how practice can individually but at the collective (macroethical) Voices From the Past be explicitly responsive to ethical concerns, level. The 4 Belmont principles have equal Bernard M. Dickens, PhD and how the choice of options in the design rank. Since one may be applied to subordinate Health Policy and Ethics Forum and implementation of projects can be en- others, practitioners must justify their ordering Kenneth R. McLeroy, PhD Public Health Matters hanced by a sense of ethical values. Ethical of priorities. For instance, a study in which EDITORIAL BOARD insight into projects may afford opportunities each informed individual freely consents to M. Lyndon Haviland, DrPH (2005), Chair for initiatives that add to the projects’ value participate respects the individuals’ autonomy Frank J. Chaloupka, PhD (2006) and the satisfaction of those involved. but may be considered harmful by the com- Vanessa Northington Gamble, MD, PhD (2006) Ethics is not best applied as a directive munities to which they belong. Similarly, a Michael D. Kogan, PhD (2007) superstructure. Ethics related to issues of study approved by the public to advance a Linda Young Landesman, DrPH, MSW (2006) Bruce Lubotsky Levin, DrPh, MPH (2005) biology, often called bioethics, has attracted community’s well-being may be opposed by Marsha D. Lillie-Blanton, DrPH (2007) justified criticism when its practitioners— individual members whose privacy or self- Kusuma Madamala, MPH (2006) bioethicists—condescendingly pronounce on interest is compromised. Issues of consent, con- Neil Hann, MPH, CHES (2007) the ethics of practice or admonish those confidentiality, benefit, and risk perception may Gregory Pappas, MD, PhD (2005) scientiously undertaking advancement of have to be resolved at a community or demo- Allan Steckler, DrPH (2006) health interests for offending refined ethical cratic level even though individuals’ personal Henrie M.Treadwell, PhD (2005) Terrie F.Wetle, PhD (2007) principles. Ethics is better applied collabora- interests warrant appropriate protection. Siu G.Wong, OD, MPH (2007) tively, equipping practitioners to maximize There are often different ethical approaches Ruth E. Zambrana, PhD (2005) the ethical advantages and minimize the to implementing projects. Those favoring par- STAFF Georges C. Benjamin, MD, FACP ethical costs of their proposals. ticular prioritizations of values may regard Executive Director/Publisher Modern principles of bioethics, which have other approaches not as unethical, but as af- Ellen T. Meyer, Director of Publications their roots in ancient moral philosophy, have fording different weight to ethical principles Nancy Johnson, MA, Executive Editor Dave Stockhoff, MA, Production Editor been popularly distilled in the 1978 Belmont and values. The role of ethical analysis is to Noëlle A. Boughanmi, Associate Production Editor Report of the National Commission for Pro- require proponents to articulate the principles Brian Selzer, Assistant Production Editor tection of Human Subjects of Biomedical and or values at stake and to justify why they have Ashell Alston, Director of Advertising Behavioral Research. These widely respected elevated or subordinated each of them. When Irma Rodenhuis, Graphic Designer core principles offer guiding values for both ethical analysis does not produce agreement, Heather Wildrick, Editorial Assistant research and epidemiological practice. The it can facilitate respectful disagreement and Jim Richardson, Subscriptions Coordinator Dana Jones, Reviews Coordinator first principle, that persons be treated with re- open understanding of ethical options. FREELANCE STAFF Janis Foster, Greg Edmondson, spect, requires that autonomy be accorded a Gary Norton, Frank Higgins, Jan Martin, high value and that persons incapable of self- Bernard M. Dickens, PhD, LLD, FRSC Gretchen Becker,Alisa Guerzon, Copyeditors determination be protected against exploita- Department Editor, Health Policy and Ethics Rebecca Richters, Editorial Proofreader Alison Moore, Chris Filiatreau, tion and abuse. Faculty of Law, Faculty of Medicine, Chrysa Cullather, Proofreaders Second is the duty to maximize good, or and Joint Centre for Bioethics Vanessa Sifford, Michele Pryor, Graphic Designers beneficence. This extends beyond the third University of Toronto Aleisha Kropf, Image Consultant principle, nonmaleficence, the historic medical Toronto, Ontario

1094 | Editor’s Choice American Journal of Public Health | July 2005, Vol 95, No. 7  EDITORIALS

In the Name of History, despite its wrenching pain, immoral. Although by 1920 the whether by altering the environ- Cannot be unlived, but if faced Hardy–Weinberg principle12 ment or by manipulating the Public Health With courage, need not be lived again. had shown the futility of such gene pool.14 ,15 Maya Angelou attempts to alter the gene pool, In correcting the perception “On the Pulse of Morning” eugenics continued to be invoked that American eugenic steriliza- to justify the use of sterilization tion was advanced by a narrow In the late 19th century, eugen- practices on a widening pool of set of actors in an era long ics—a set of ideas about the bio- so-called undesirables, increas- gone, Stern raises a set of dis- logical betterment of human ingly defined as immigrants, concerting issues for public stock—emerged in Britain and the poor, and racial/ethnic health practitioners and advo- was soon incorporated into so- minorities. cates today. The most obvious cial and public health policy in of these, and the most troubling numerous settings. By the 1910s, EUGENICS, CALIFORNIA for its contemporary resonance, advocates of positive eugenics STYLE is the social prejudice that in Catholic Europe and Latin marked activities pursued in America combined with pedi- As Stern shows in this issue,13 the name of public health. The atric and other health reformers perhaps nowhere were these extent to which health and to back family wages, universal ideas and practices as thor- medical policies absorb and preschools, foster care, housing oughly embraced and institu- reflect the dominant class and codes, and school health exams, tionalized as in California, racial logic of the time has been all with the goal of improving where a sterilization law was in well documented. the conditions of childhood and effect from 1909 to 1979. In As Vanessa Gamble has ar- of the human stock more gen- Stern’s analysis, 2 key points gued, the US Public Health erally.1,2 During the 1920s in stand out that are equally unset- Service’s 40-year study of un- the Soviet Union, eugenics was tling. First, Stern shows the con- treated syphilis in 400 Black construed as a form of social tinuities between the tens of men in Tuskegee, Alabama, rep- medicine and supported by the thousands of forced sterilizations resents only 1 episode in a sea official public health agency. Sci- that took place in mental institu- of institutionalized practices entists appealed to eugenics as tions in the early 20th century and daily interactions that has a demonstration of the utility and the coercive sterilization produced a legacy of overtreat- of genetics to public health, (typically in the name of family ment, undertreatment, and while the policy implications of a planning) of Mexican American mistreatment of African Ameri- Bolshevik-style positive eugenics and other immigrant, minority, cans.16 Nayan Shah has likewise for a time coexisted with “envi- and working-class women in documented the portability of ronmentalist” social hygiene more recent decades. Second, the health department’s and approaches.3 she demonstrates that eugenic the public’s scapegoating of the In the better-known applica- sterilization was not promoted Chinese community in San tions of negative eugenic think- and practiced by a handful of Francisco, California, from epi- ing in Protestant settings—in prejudiced public health officials demic to epidemic across 2 cen- Scandinavia,4,5 Great Britain,6–8 and doctors whose activities turies17 (3 if we include the se- the United States,6,9 and, most were left unchecked, but rather vere acute respiratory syndrome infamously, Nazi Germany10 ,11 — was fully backed by numerous [SARS] epidemic of 2003). The various interventions including state and federal agencies and classism of health and medical sterilization were employed to social, political, and academic institutions—in the US context, “breed out” certain “defective” luminaries. Indeed, public health often intertwined with racial human characteristics, initially a and eugenics programs shared and ethnic discrimination— variety of mental conditions and the same general goal: to im- reveals similarly pernicious behaviors defined as criminal or prove the well-being of society, patterns of prejudice.18–21

July 2005, Vol 95, No. 7 | American Journal of Public Health Editorials | 1095  EDITORIALS 

Stern investigates how public tient preferences. In an inversion policies may lead to frustration; if trol has had more conflicted health institutions evolved into a of their intended use, informed institutionally based ethical mea- meanings and consequences for key site of racialization from the consent policies protected not sures are not linked to larger poor and working-class women late 19th into the 20th century.13 patients but doctors, who were movements for social rights, we around the world.26–28 Public health not only repro- believed to be obtaining patients’ may find ourselves developing In the end, we must examine duced larger societal tenets consent properly whether or not ever more detailed ethical codes the ideological footprint left by but also exerted influence well this was actually the case. while endlessly battling class and proponents of eugenics through beyond the profession and prac- Moreover, the racial logic that racial prejudice. their writings, the policies they tice of public health. While girded the sterilization projects of A further jarring issue high- developed, and the profession- sterilization laws may not have the 1920s was transformed and lighted by Stern has to do with als they trained. Early eugeni- mentioned race and class explic- presented as a defense strategy how seemingly well-understood cists generated influential poli- itly, they were racialized and in Madrigal v Quilligan. The public health activities are re- cies that helped embed racial class-oriented in enforcement. judge did not rule against the shaped in various political con- and classist reasoning into pub- Immigrants, particularly those of defendants on the basis of half texts. Although many accounts lic institutions. Stern13 lays the color, were not only sterilized in century–old racial logic, but he portray eugenics as a unitary theoretical groundwork for us to disproportionate numbers but summoned a “clash of cultures” movement informed by conser- see how a belief in a racial and also were marked as inferior by argument that nonetheless rested vative ideas and supported by social hierarchy was at the core the practice. on a belief in racial hierarchy. political counterparts, it was of sterilization projects in the California’s sterilization law above all a technocratic devel- 192 0s and the 1970s and the was justified as a preventive THE PAST AND opment that could be and was passage of Proposition 187 in measure that was at the same THE PRESENT appropriated and refashioned the 1990s. (Proposition 187 time cost saving, desirable for by utopians, social progressives, was approved via referendum the patient and her family, and On one level, Madrigal and nativists, and Nazis. The evolu- by California voters in 1994 to good for the public. Yet, remark- other similarly reasoned deci- tion of eugenic policies in prevent undocumented immi- ably, California state legislators sions alert us to the need for fre- Protestant countries—where grants from receiving public acknowledged that implementa- quent reevaluation of the mea- sterilization was almost univer- benefits or services, including tion of the law might result in le- sures intended to protect the sally adopted as the preferred health care and education; it gitimate legal claims, and in 1917 public—particularly its most means of achieving eugenic was never implemented because they modified the law to protect marginalized members—from goals—is in distinct contrast to of legal challenges.) If we see doctors who carried out state- prejudicial treatment by health the situation in many Catholic these projects as historically sanctioned sterilizations from institutions. In recent decades settings, where a positive eugen- linked, we cannot relegate such legal retaliation.13 “cultural competence” has ics of enhancing prenatal and dangerous approaches to an This explicit recognition of the emerged as a standard of prac- childraising circumstances sub- unprogressive past. potential for violation of patients’ tice that puts the onus on health stituted for sterilization. The past is infinitely complex, rights was perversely echoed in care institutions to take into ac- In linking eugenics to right- but surely not impenetrable. the ruling in Madrigal v Quilligan count differential linguistic abili- wing political agendas, some Public health history teaches (No. 75 Civ 2057 [CD Cal June ties and distinct cultural under- scholars have inaccurately us that scientific and technical 30, 1978], aff’d, 639 F2d 789 standings of family, community, pointed to the end of World developments interact continu- ([9th Cir 1981]) some 60 years and medical authority and deci- War II and the discrediting of ously with the political and later. Despite the existence of in- sionmaking.22,23 Although it is “Nazi science” at the Nurem- social context and that health formed consent practices by the potentially useful in preventing berg trials as the demise of eu- policies and their implementa- early 1970s—developed in part the tragedies of medical mis- genics.25 Yet, as Stern shows for tion both reflect and shape the because of the involuntary steril- communication,24 cultural com- California, eugenics did not dis- political context and social hier- izations of the previous half petence already runs the risk of appear then; support for eu- archy within particular societies. century—the judge in this case becoming just another item in a genic sterilization merged with Facing history with courage interpreted the sterilization of the checklist of requirements to growing concerns about over- compels us to raise questions of Mexican American plaintiffs to allay liability concerns—an item population and family planning. the past based on the pain of be the result of cultural misun- that nonetheless leaves many Birth control, at bottom a tech- the present and to raise ques- derstanding rather than the prod- elements of the power imbal- nocratic measure, was also ap- tions of the present based on the uct of powerful incentive systems ance between patient and insti- propriated differentially by vari- pain of the past. for doctors to perform these pro- tution unaddressed. ous actors. Seized upon as a cedures or of provider manipula- At another level, this process means of freedom for elite and Anne-Emanuelle Birn, ScD, MA tion or of willful ignorance of pa- of reevaluation and refinement of middle-class women, birth con- Natalia Molina, PhD, MA

1096 | Editorials American Journal of Public Health | July 2005, Vol 95, No. 7  EDITORIALS 

About the Authors ence: Eugenics in Germany, France, Brazil, 11.Proctor R. Racial Hygiene: Medicine New York City, 1892. Bull Hist Med. Anne-Emanuelle Birn is with the Depart- and Russia. New York, NY: Oxford Uni- Under the Nazis. Cambridge, Mass: 19 95;69:420–457. ment of Public Health Sciences, University versity Press; 1990. Harvard University Press; 1988. 21.Feldberg GD. Disease and Class: of Toronto, Toronto, Ontario. Natalia 3. Adams M. Eugenics as social medi- 12. Sturtevant A. A History of Genetics. Tuberculosis and the Shaping of Modern Molina is with the Department of Ethnic cine in revolutionary Russia: prophets, New York, NY: Harper & Row; 1965. North American Society. New Brunswick, Studies, University of California, San Diego. patrons, and the dialectics of discipline- Available at: http://www.esp.org/books/ NJ: Rutgers University Press; 1995. Requests for reprints should be sent to building. In: Solomon S, Hutchinson JF, sturt/history (PDF file). Accessed Anne-Emanuelle Birn, ScD, MA, Depart- 22. National Center for Cultural Com- eds. Health and Society in Revolutionary May 1, 2005. petence. Available at: http://gucchd. ment of Public Health Sciences, University Russia. Bloomington: Indiana University 13. Stern AM. Sterilized in the name georgetown.edu/nccc/products.html. Ac- of Toronto, 1st Floor, McMurrich Bldg, 12 Press; 1990:200–223. Queen’s Park Crescent W, Toronto, ON of public health: race, immigration, cessed October 25, 2004. M5S 1A8, Canada (e-mail: ae.birn@ 4. Broberg G, Roll-Hansen N, eds. and reproductive control in modern 23. Cross TL, Bazron BJ, Dennis KW, utoronto.ca). Eugenics and the Welfare State: Steriliza- California. Am J Public Health. 2005; Isaacs MR. Towards a Culturally Compe- This editorial was accepted December 2, tion Policy in Norway, Sweden, Denmark, 95:1128–1138. tent System of Care: A Monograph on and Finland. Lansing: Michigan State 2004. 14 .Pernick MS. Taking better baby Effective Services for Minority Children University Press; 1997. doi: 10.2105/AJPH.2004.058065 contests seriously. Am J Public Health. Who Are Severely Emotionally Disturbed. 5. Weindling P. International eugen- 2002;92:707–708. Washington, DC: Georgetown Univer- ics: Swedish sterilisation in context. sity Center for Child Health and Mental Acknowledgments Scand J Hist. 1999;24:179–197. 15.Pernick MS. Eugenics and public Health Policy, CASSP Technical Assis- The authors thank Nikolai Krementsov, health in American history. Am J Public tance Center; 1989. Mark B. Adams, and the American Pub- 6. Barkan E. The Retreat of Scientific Health. 19 97;87:1767–1772. lic Health Association’s Spirit of 1848 Racism: Changing Concepts of Race in 24.Fadiman A. The Spirit Catches You 16. Gamble VN. Under the shadow coordinating and history committees, Britain and the United States Between the and You Fall Down: A Hmong Child, Her of Tuskegee: African Americans and especially Catherine Cubbin, Luis World Wars. New York, NY: Cambridge American Doctors, and the Collision of health care. Am J Public Health. 19 97; Avilés, and Nancy Krieger, for their University Press; 1992. Two Cultures. New York, NY: Farrar, 87:1773–1778. helpful comments. 7. Mazumdar PMH. Eugenics, Human Straus, & Giroux; 1997. This editorial was based on a presen- Genetics and Human Failings: the Eugen- 17. Shah N. Contagious Divides: Epi- 25. Paul DB. Controlling Human Hered- tation given at the session “Immigrant ics Society, Its Source and Its Critics in demics and Race in San Francisco’s Chi- ity: 1865 to the Present. Atlantic High- California: Inequality and Public Health Britain. London, United Kingdom: Rout- natown. Berkeley: University of Califor- lands, NJ: Humanities Press; 1995. in Historical Perspective” organized by nia Press; 2001. ledge; 1992. 26. Nelson J. Women of Color and the the Spirit of 1848 Caucus History Com- 8. Kevles D. In the Name of Eugenics: 18.Rosenberg CE. Social class and Reproductive Rights Movement. New mittee, 131st Annual Meeting of the Genetics and the Uses of Human Heredity. medical care in nineteenth-century York, NY: New York University Press; American Public Health Association, San New York, NY: Alfred A. Knopf; 1985. America: the rise and fall of the dispen- 2003. Francisco, Calif, November 17, 2003. sary. J Hist Med Allied Sci. 1974;2 9: 9. Leon SM. “Hopelessly entangled 27. Hartmann B. Reproductive Rights 32–54. in Nordic pre-suppositions”: Catholic and Wrongs: The Global Politics of Popu- References participation in the American Eugenics 19. Krieger N, Fee E. Measuring social lation Control and Contraceptive Choice. 1. Stepan NL. The Hour of Eugenics: Society in the 1920s. J Hist Med Allied inequalities in health in the United Boston, Mass: South End Press; 1995. Race, Gender, and Nation in Latin Amer- States: a historical review. Int J Health Sci. 2004;59:3–49. 28. Briggs L. Reproducing Empire: Race, ica. Ithaca, NY: Cornell University Press; Serv. 1996;26:391–418. 10.Weindling P. The survival of eu- Sex, Science, and U.S. Imperialism in Puerto 19 91. genics in 20th century Germany. Am 20. Markel H. “Knocking out the chol- Rico. Berkeley: University of California 2. Adams MB, ed. The Wellborn Sci- J Hum Genet. 19 92;52:643–649. era”: cholera, class, and quarantine in Press; 2002.

July 2005, Vol 95, No. 7 | American Journal of Public Health Editorials | 1097  EDITORIALS 

While the AIDS epidemic in phology, most notably in the face Alcohol Syndrome in the West- Using the Past South Africa has been well publi- and head; and developmental ern Cape Province of South cized and documented in recent difficulties. The most recent sta- Africa,”3 a control group and a to Step years, a lesser-known health cri- tistics on FAS rates in the West- case group of mothers of vary- sis with profound socioeconomic ern Cape Province, reported by ing socioeconomic status were Forward: Fetal implications has been developing the National Institute of Alco- used to examine the factors that for hundreds of years. The West- holism and Alcohol Abuse, show are associated with mothers’ Alcohol ern Cape Province of South that 40.5 to 46.4 per 1000 having children with FAS. Two Africa has the highest rates of children are found by screening of the most telling findings of fetal alcohol syndrome (FAS) in to have FAS.1 Compare this rate the study were that mothers of Syndrome in the world. FAS is caused by ma- with a FAS frequency of 0.5 to lower socioeconomic status ternal alcohol use during preg- 2.0 per 1000 children in the were at higher risk for having the Western nancy and is one of the leading United States or the average rate children with FAS and that the causes of preventable mental and of 0.97 per 1000 children in the study community as a whole Cape Province physical retardation among in- developed world,2 and the differ- had very limited knowledge or fants worldwide.1 Infants suffer- ence is staggering. understanding of FAS and the of South Africa ing from FAS often have low In the population-based study implications of consuming alco- birthweights; physical dysmor- “Maternal Risk Factors for Fetal hol during pregnancy. In addi-

July 2005, Vol 95, No. 7 | American Journal of Public Health Editorials | 1097  EDITORIALS 

tion, this study corroborates made illegal, and after 300 ingly, many of the women sur- poverished conditions will be current research that is finding years of implementation in the veyed in the region reported hard pressed not to perpetuate an increase in FAS in the West- Western Cape Province, the struggling with feelings of de- the cycle. ern Cape Province and in the world is only now beginning to pression and low self-esteem In addition to labor rights, rest of South Africa. The study see its devastating repercussions. and with alcohol abuse.7 With the physical and mental health accurately summarizes contem- The system has become so ritu- alcoholism so entrenched in the of the community must be at- porary drinking patterns as a alized that “it is still apparent culture, both at work and at tended to through structured product of the Western Cape today that alcohol is a favored, home, it is likely difficult for educational interventions. There Province “dop” system and con- valued and expected commodity mothers attempting to maintain are limited resources in the cludes that prevention is needed among many of the local popu- sobriety to get support. Western Cape Province for to combat the FAS epidemic. lation workers, who receive low Almost 50% of pregnant pregnant women and for chil- For prevention to be effective, pay and who live in very hum- mothers in the Western Cape dren with FAS. Primary inter- however, it is essential to under- ble circumstances.”7(p7) To c oun- Province drink alcohol, com- vention methods should focus stand the legacy of the dop sys- teract this mentality, the health pared with 34% of pregnant on the education of mothers tem in the Western Cape Prov- care sector in the Western Cape mothers in the metropolitan and the community at large on ince and to examine why FAS Province has formed the Dop- areas of South Africa.7 It has the direct correlation between rates are so astonishingly high stop Association to educate the been found that mothers of FAS alcohol consumption during in such a small geographic area. community and inform farm- children in the region come pregnancy and FAS. As a sec- workers of their legal rights.5 from families with a history of ondary intervention, it is crucial HISTORICAL CONTEXT The dop system promoted and generations of alcohol abuse that FAS continue to be moni- AND THE DOP SYSTEM sustained a culture of alcohol in- and heavy drinking.7 Mothers of tored and tested for at the earli- take that not only ensured that children with FAS reported hav- est possible age. Timely data In the 1700s European colo- local communities stayed impov- ing 12.6 drinks per week, com- collection documenting current nialists capitalized on the fertile erished, but also had negative pared with 2.4 drinks for con- FAS trends among children in land and climate of South Africa biological, psychological, and trol subjects, and 50% of the region is critical to imple- to create an agricultural econ- social consequences for the pop- mothers of FAS children re- menting effective methods of omy. In the Western Cape Prov- ulation. Nowhere is this more ported drinking more heavily intervention and awareness ince, this meant grape and wine evident than in the effect the while pregnant.8 While the use programs.1 production.4 To pay the farm- system has had on the mothers of alcohol as a coping mecha- An important research project workers, the colonial farmers de- and children of the Western nism for dealing with stress is that could serve to refine inter- veloped what is now known as Cape Province. clearly documented, the future vention methods would be to the dop system. repercussions of this social habit study the social habits of mothers Payment under the dop sys- MOTHERS AND FAS remain uncertain. in the Western Cape Province tem initially consisted of bread, who do not produce offspring tobacco, and wine.5 While this To fully understand why the LONG-TERM with FAS. Mothers of FAS chil- method of payment was not par- FAS rates of the Western Cape IMPLICATIONS IN THE dren in the province reported ticularly unusual in farming Province are the highest in the WESTERN CAPE drinking heavily to cope with a communities, the dop system be- world, it is crucial to examine PROVINCE stressful relationship with an al- came unique to the Western the social conditions in which cohol-abusing man.7 Further- Cape as farmers “institutional- the women of the province live. FAS is now an epidemic in more, mothers of FAS children ized alcohol as a condition of Women constitute roughly 30% the Western Cape Province.8 In were found to consume 97% of service.”6(p60) One incentive for of the commercial agriculture a community dependent on a their alcohol intake on weekends, using the system was that it pro- workforce and are 2 to 3 times low-wage economy that is which is when they spend a sig- vided an ingenious way for more likely than men to be slowly undergoing moderniza- nificant amount of time with farmers to dispose of excess hired as casual laborers.6 More tion, FAS has dangerous impli- their partners.7 This could sug- wine that was deemed unfit to than two thirds of farmworker cations.5 There is strong evi- gest that women who are not drink. As one study noted, it was families live in waged poverty, dence that farm children in the producing FAS children have “reject wine unsuited for the with a household income of less region, who are often malnour- stronger, healthier support net- open market”6(p61) and was sold than $100 per month.6 Women ished or suffering from FAS, works, reducing their reliance on back to the farmers for next to in the Western Cape Province grow up to be low-skilled, mal- alcohol. Researchers are also nothing. face a difficult challenge as both nourished adult farmworkers.6 looking into the possibility of bio- The dop system continues mothers and workers dealing It is a difficult cycle to break, logical and genetic mutations today, despite being recently with waged poverty. Not surpris- and FAS children born into im- caused by generations of alcohol

1098 | Editorials American Journal of Public Health | July 2005, Vol 95, No. 7  EDITORIALS 

use in the region, making chil- abuse of human rights for profit. This editorial was accepted October 8, population of the Western Cape Prov- dren more susceptible to FAS.4 The highest rates of FAS in the 2004. ince, South Africa. Alcohol Clin Exp Res. doi: 10.2105/AJPH.2004.056366 2001;25:1719–1722. FAS has profound socioeco- world are caused by a multitude nomic implications for the future of contributing factors, and a ho- 5. London L. Alcohol consumption References amongst South African farm workers: a health of the Western Cape Prov- listic, comprehensive approach challenge for post-apartheid health sec- 1. Centers for Disease Control and tor transformation. Drug and Alcohol ince. The combination of high will be necessary to begin revers- Prevention. Fetal alcohol syndrome— Dependence. 2000;59(2):199–206. FAS rates and high HIV infection ing a trend that has been devel- South Africa, 2001. MMWR Morb Mor- rates poses a serious threat to the oping for 300 years. tal Wkly Rep. 2003;28(52):660–664. 6. London L. Human rights, environ- family structure and core work- 2. May PA, Gossage JP. Estimating mental justice, and the health of farm the prevalence of fetal alcohol syn- workers in South Africa. Int J Occup En- force in the region. South African Jake McKinstry, MS drome. A summary. Alcohol Res Health. viron Health. 2003;9:59–68. public health authorities must ad- 2001;25:159–167. 7. Viljoen D, Croxford J, Gossage JP, dress the past inequities of the 3. May PA, Gossage JP, Brooke LE, et Kodituwakku PW, May PA. Character- dop system and restructure About the Author al. Maternal risk factors for fetal alcohol istics of mothers of children with fetal syndrome in the Western Cape Province alcohol syndrome in the Western Cape health care systems and strate- At the time the editorial was written, the of South Africa: a population-based Province of South Africa: a case con- author was a Master of Science candidate gies to address the epidemic of study. Am J Public Health. 2005;95: trol study. J Stud Alcohol. 2002;63: in Urban Planning at Columbia Univer- 119 0–1199. 6–17. FAS in the Western Cape Prov- sity’s School of Architecture, Planning, and ince.5 Children with FAS are the Preservation, New York, NY. 4. Viljoen D, Carr LG, Foroud TM, 8. May PA, Brooke L, Gossage JP, et ones who will suffer the conse- Requests for reprints should be sent to Brooke L, Ramsay M, Li TK. Alcohol al. Epidemiology of fetal alcohol syn- Jake McKinstry, 2514 34th Avenue South, dehydrogenase-2*2 allele is associated drome in a South African Community quences of an outmoded farming Seattle, WA 98144 (e-mail: jsm2106@ with decreased prevalence of fetal alco- in the Western Cape Province. Am J system predicated upon the columbia.edu). hol syndrome in the mixed-ancestry Public Health. 2000;90:1905–1912.

July 2005, Vol 95, No. 7 | American Journal of Public Health Editorials | 1099  EDITORIALS 

For decades, South Africa has en- risk and offering effective inter- additional public health efforts Fetal Alcohol dured an epidemic of fetal alco- ventions. Not surprisingly, the will be needed to achieve the ex- hol syndrome (FAS) that requires risk factors found in South Africa pected outcome of reducing or Syndrome immediate and decisive public are not much different from eliminating FAS in South Africa health attention. The prevalence those found in high-income and to sustain success. These ef- Prevention in rates of more than 40 cases per countries.4 forts should include development 1000 children in the Western Developing and implementing of policies needed to address FAS South Africa and the Northern Cape Provinces strategies to reduce alcohol use nationwide, development of on- and more than 20 cases per during pregnancy by targeting going surveillance methods for 1000 children in Gauteng Prov- high-risk women may provide FAS and alcohol use during preg- and Other ince represent some of the high- public health benefits beyond re- nancy, training of health care pro- est rates of FAS in the world.1–3 ducing the incidence of FAS if fessionals and community work- Low-Resource May et al.’s article on maternal strategies for HIV and sexually ers, and community education risk factors for FAS in South transmitted diseases (STDs) can targeting high-risk women. Most Countries Africa4 should be a call to action be integrated.5,6 Bundled preven- important, prevention strategies to identify and apply strategies tion efforts will benefit from effi- should be addressed not only to that can prevent this syndrome. ciencies that can be brought into women of childbearing age, but It should also alert other low- the delivery of primary health also to public health officials, poli- resource nations in which alcohol care and long-term costs of FAS cymakers, health care providers, is available and used by women to society. and communities. Given the high of childbearing age to the possi- May et al. suggest that preven- rate of teenage pregnancies in bility of a similar problem. tion should be accomplished South Africa, school children are May et al. identify risk factors through social improvement and particularly important targets in for FAS, such as binge drinking proven techniques of birth con- any effort to halt alcohol abuse during pregnancy, maternal age, trol, treatment for alcohol abuse, by pregnant women. poor education, poor nutrition, and screening for alcohol use genetic influences, gravidity, and during prenatal services. We POLICYMAKERS poor socioeconomic environ- agree that these are key preven- ment, that provide a starting tion strategies that are directly in- As May et al. point out, FAS point for identifying women at formed by this study. However, is a birth defect that results from

July 2005, Vol 95, No. 7 | American Journal of Public Health Editorials | 1099  EDITORIALS 

exposure of the fetus to alcohol Cape, Northern Cap, and Gauteng and effective health education laypeople accepted by the com- during pregnancy as well as a provinces. However, to assess the and behavioral change tech- munity as reliable sources of problem of substance abuse by magnitude of FAS (local, regional, niques and strategies. Then, in information, including health the mother. As such it is prima- and national) requires an estima- communities in which FAS, HIV, workers, traditional leaders and rily a maternal and child health tion of the prevalence of the dis- and STDs are major health prob- healers, teachers, and represen- (MCH) issue, one that has been ease and of the associated morbid- lems, screening and treatment for tatives of the private sector im- recognized in the South African ity in the population as a whole, as alcohol use and sexual behavior plementing prevention and con- National Policy Guidelines for well as the financial cost to indi- before and during pregnancy will trol programs. the Management and Prevention viduals, families, and society. To allow providers to identify the at- of Genetic Disorders, Birth De- accomplish such an estimate, we risk population early. Interven- WOMEN OF fects and Disabilities.7 Given the suggest that FAS, HIV, and STD tions to influence sexual behav- CHILDBEARING AGE outreach to wide segments of programs develop integrated ior and alcohol use can focus on the population that is possible surveillance systems with simi- education about risk reduction Women of childbearing age through primary health care, lar data collection tools and en- strategies and provision of the are the key element in success- MCH systems should contribute gage in similar approaches for skills and means to negotiate and ful FAS prevention. We need to to FAS prevention. Given the tracking and evaluation of pre- practice safe sex. better understand the many so- common elements among the vention efforts. In low-resource countries like cial and psychological processes populations currently at risk for South Africa, capacity develop- that contribute to risky drinking FAS, HIV, and STDs, such pro- HEALTH CARE ment will have to be achieved and sexual activities in the envi- grams could maximize resources, PROVIDERS in an environment in which pri- ronments in which these women efficiencies, and organizational mary health care workers are al- live, and we must seek to delin- synergies. The preconceptional, Health care providers at all ready overburdened. The knowl- eate personal and societal inter- perinatal, and postnatal periods levels should be trained to edge and skills imparted to them ventions that are both accept- are obvious times for the con- screen for, diagnose, prevent, to address the problems associ- able and realizable.10 centration of efforts for preven- and treat an alcohol-exposed ated with FAS (and HIV and Individual- and group-level in- tion of FAS, HIV, and STDs. pregnancy. Curriculum programs STDs) will need to be succinctly terventions should include edu- Outreach is required to draw and materials tailored to meet packaged, of proven efficacy, and cation and counseling that pro- more at-risk women into these the learning needs of these pro- well marketed. mote safe sex behaviors, planned services, as are efforts to high- fessionals should be developed pregnancies, improved nutrition, light the specific needs of these and used. This has, in part, been COMMUNITY and changes in alcohol consump- women as part of maternal and achieved in South Africa with tion behavior. On the individual child care. The national MCH the Perinatal Education Pro- The social norms and policy level, delaying pregnancy in the program in South Africa should gram’s Birth Defects Manual, environments governing com- women at highest risk, including collaborate with the provincial which has been developed as munities affect the risks and those who already have a child programs to develop a national part of a national genetic educa- protective behaviors of individu- with FAS, is imperative. Educa- populationwide approach, tion program for primary care als. Interventions that address tion and support provided in through appropriate legislation practitioners.8 However, like social network- and community- group settings can promote and and health policy, and an opera- most health care workers’ cur- level phenomena have been ef- reinforce safer behaviors and are tional-level approach that inte- riculum materials, it does not fective in reducing HIV risk especially good for providing in- grates FAS, HIV, and STD pre- comprehensively address risk- among a range of population terpersonal skills training in ne- vention and control into MCH factor intervention techniques or groups in different geographic gotiating and sustaining appropri- delivery services. Such inte- long-term planning. These topics settings.9 Similar approaches ate behavior change. Given the grated programs could stimulate will need to be addressed at the can be implemented for FAS. high rates of teenage pregnancy broad support and would entail national level. For communities, changes in at- in South Africa, schoolchildren well-defined responsibilities, re- Capacity-building of experi- titudes, norms, and practices are a particularly important tar- sources, and budgets. enced staff to conduct training is can be brought about through get in any effort to halt alcohol FASprevention requires a local, essential. Their participation in health communication, social abuse by pregnant women. regional, and national surveillance the development of ideas for marketing, community mobiliza- system for FAS and for alcohol overcoming obstacles will assist tion and organization, and com- CONCLUSIONS use during pregnancy. Researchers them in offering effective, tar- munitywide events. Interven- in South Africa have reported high geted health education. Training tions on community building We must continue to develop FASprevalence rates in some can incorporate role-playing to and social support should enlist new methods of FAS prevention, small communities in the Western develop expertise in appropriate community leaders and other remaining mindful of both the

1100 | Editorials American Journal of Public Health | July 2005, Vol 95, No. 7  EDITORIALS 

biomedical and the behavioral Health Laboratory Service and the Uni- aspects of the problem. For ex- versity of the Witwatersrand, Johannes- burg, South Africa. ample, prevention efforts that Requests for reprints should be sent to promote treatment for alcohol Jorge Rosenthal, PhD, MC, National Cen- use and that incorporate safe-sex ter on Birth Defects and Developmental Disabilities, MS-E86, 1600 Clifton Rd, methods of protection will con- Atlanta, GA 30333 (e-mail: jyr4@ tribute to the prevention of FAS, cdc.gov). HIV, and STDs in a way that This editorial was accepted October 24, 2004. women can control. If such ef- doi: 10.2105/AJPH.2004.057372 forts are successful in decreasing the prevalence of the target dis- Acknowledgment eases, they can show communi- We gratefully acknowledge Mary Mc- ties that prevention efforts can Cauley for her review of this editorial. work and therefore alter the perception that communities have References 1. Centers for Disease Control and about prevention. Prevention. Fetal alcohol syndrome— Using the information that is South Africa, 2001. MMWR Morb Mor- becoming available, we need to tal Wkly Rep. 2003;52(28):660–662. develop new public health inter- 2. Centers for Disease Control and Prevention. Update: trends in fetal alco- ventions for FAS prevention that hol syndrome—Alaska, Arizona, Colo- are mindful of biomedical, be- rado and New York, 1995–1997. havioral, and social realities. MMWR Morb Mortal Wkly Rep. 2002; 51:433–435. These interventions must combine prevention of FAS, HIV, 3. South African–US Consultation on Fetal Alcohol Syndrome Research: Scien- and STDs and must work on all tific Progress and Future Directions. Sum- levels of social organization, mary. Cape Town: Government of South from individuals to whole soci- Africa and US Centers for Disease Con- trol and Prevention; 2003. eties. In pursuing the best prac- 4. May PA, Gossage JP, Brooke LE, tices and most effective interven- et al. Maternal risk factors for fetal al- tions, we must recognize that cohol syndrome in the Western Cape these efforts will succeed at the Province of South Africa: a population- based study. Am J Public Health. 2005; level necessary to prevent FAS, 95:1190–1199. HIV, and STD problems only if 5. Askew I, Berer M. The contribu- they are applied in combination tion of sexual and reproductive health and if fiscal and personnel re- services to the fight against HIV/AIDS: a review. Reprod Health Matters. 2003; source are provided. Further- 11(22):51–73. more, they will succeed only if 6. Fetal Alcohol Syndrome: Guidelines they are linked to other efforts for Referral and Diagnosis. Atlanta, Ga: to address the macro level social Centers for Disease Control and Preven- conditions that contribute to tion; 2004. the disparate vulnerability of 7. National Policy Guidelines for the Management and Prevention of Genetic affected populations. Disorders, Birth Defects and Disabilities. Pretoria, South Africa: South African Jorge Rosenthal, PhD, MC National Department of Health; 2001. Arnold Christianson, MD 8. Woods D. Birth Defects Manual. Cape Town, South Africa: Perinatal Edu- Jose Cordero, MD cation Trust; 2002. 9. AIDS Epidemic Update. Geneva, About the Authors Switzerland: United Nations Programme on HIV/AIDS and the World Health Jorge Rosenthal and Jose Cordero are Organization; December 1999. with the National Center on Birth Defects and Developmental Disabilities, Centers 10.Susser I, Stein Z. Culture, sexuality, for Disease Control and Prevention, At- and women’s agency in the prevention lanta, Ga. Arnold Christianson is with of HIV/AIDS in South Africa. Am J the Division of Human Genetics National Public Health. 2000;90:1042–1048.

July 2005, Vol 95, No. 7 | American Journal of Public Health Editorials | 1101  FIELD ACTION REPORT 

Public Health Nurses for Virginia’s Future: A Collaborative Project to Increase the Number of Nursing Students Choosing a Career in Public Health Nursing

| Martha W. Moon, PhD, RN, MPH, JoAnne K. Henry, EdD, RN, MPH, Karen Connelly, RN, MPA, and Phyllis Kirsch, MS, MPH

A shift in the role of public health practice in the United States faculty member at VCU provided assessment and evaluation of the to population-focused care, together with demographic shifts in- leadership for this project, as did effects of contemporary issues creasing the diversity and age of the population, has created a the director of public health nurs- and health policy on the public’s need for a public health workforce more highly skilled in coming for the Virginia State Depart- health. Despite this class, the munity and population-based practices. Despite this need, few ment of Health and the director school had not been able to place changes have been made in the pattern of field placements for of the VCU Community Nursing many students in population- nursing students, in part because many public health nurses in Organization; support was pro- focused public health settings for population-focused roles are unfamiliar with models of successful vided by the school’s community clinical experience. Nurses in student fieldwork in their areas. health clinical coordinator. these community settings reported We describe the Public Health Nurses for Virginia’s Future proj- The project’s goal was to in- difficulty identifying appropriate ect, a successful project undertaken by nurse educators and public health leaders to increase the number of highly qualified grad- crease the number of nursing clinical placements for students. uates working in state and local health departments. (Am J Public students choosing a career in Many of these nurses had re- Health. 2005;95:1102–1105. doi:10.2105/AJPH.2004.055368) public health nursing. Specific ceived no training in population- objectives were as follows: focused care in their own basic nursing education or more tradi- BACCALAUREATE EDUCATION • To provide experience in public tional public health clinical train- in nursing includes preparation in health department settings for ing, and they were unfamiliar community- and population- a minimum of 76 bachelor of with models of successful student focused care as well as individual- science (BS) students per year, placement in their current fields. focused care, all of which are es- compared with 32 at baseline They reported not having time in sential for today’s public health • To increase by 50% the num- their work to figure out how best practice.1,2 Less than 50% of the ber of students who report to use students. Through this public health nursing workforce upon graduation that they will project, we hoped to increase the in Virginia has this baccalaureate- consider a career in public number of population-focused level preparation. In an effort to health nursing placements and preceptors avail- increase the number of nurses • To increase by more than three- able to students, thus improving choosing to enter the public fold the involvement of state the preparation of future public health field, and to prepare them and local health department health nurses, helping Virginia’s with the required skills needed public health nurses in the Public Health System better meet to successfully fill roles in the Community and Public Health current needs, and preparing for emerging public health system, Nursing course the public health challenges of the the Virginia Commonwealth Uni- next 2 decades. versity (VCU) School of Nursing NEED FOR THE PROGRAM and the Virginia State Depart- METHODS ment of Health established a At VCU, students are required partnership entitled “Public to take a community and public The project employed the fol- Health Nurses for Virginia’s Fu- health nursing course in their sen- lowing 3 strategies to achieve its ture,” funded by the Division of ior year. The course employs an objectives: Nursing of the US Health Re- epidemiological approach to sources and Services Administra- population-focused nursing • Public Health Nursing Advisory tion. The lead community health through community health status Group. The project team

1102 | Field Action Report | Peer Reviewed | Moon et al. American Journal of Public Health | July 2005, Vol 95, No. 7  FIELD ACTION REPORT 

developed a Public Health local health departments for health nursing. In the following KEY FINDINGS Nursing Advisory Group to those who were unable to par- school year, a question was guide changes in the course and ticipate in the live telecasts. added to the student evaluation • Face-to-face discussions with to identify health department forms for community health public health nurses who had mentors for students. This EVALUATION nursing that asked, “Would you never worked with students group comprised public health consider a career in public health about the opportunities for nurse managers from local The project exceeded 2 of the nursing?” (Answer options were working with students, as well health departments and state 3 specific objectives and made yes, no, and unsure.) Fifty-seven as examples of appropriate health department nurses. They excellent progress toward meet- students, or 40% of the students student projects, encouraged provided suggestions for the ing the third. To determine if the surveyed, responded yes. many nurses to become seminar content and for project project’s objectives were attained, preceptors. Face-to-face dis- implementation. the number of students placed in Objective 3 cussions also helped experi- • Outreach to public health nurse population-focused settings (data With the assistance of the Pub- managers. The team conducted recorded by clinical faculty for lic Health Nursing Advisory enced preceptors better use telephone outreach and face-to- baseline and project period) was Group and outreach to public their student trainees. face meetings with public counted; the percentage of stu- health nurses, the team increased • Public health nurses and epi- health nurse managers in the dents reporting that they would the number of public health demiologists working in the field State Health Department and consider a career in public health nurses who served or agreed to were enthusiastic about the throughout the state to increase (data recorded on anonymous serve in the future as preceptors seminar series,which provided the number of field placement course evaluation forms com- for baccalaureate nursing stu- relevant information to their sites and preceptors for bac- pleted by students) was calcu- dents, from 25 in 2002–2003 to work and helped familiarize calaureate nursing students. lated; and the number of public 77 in 2003–2004. A serendipi- them with the methods being Through these meetings, the health nurses willing to serve as tous benefit is that many of the team clarified the roles and re- preceptors (data recorded by clin- preceptors gained an appreciation taught to community and pub- sponsibilities of the preceptors ical faculty for baseline and proj- of the BS degree in nursing and lic health nursing students. and the students; gave concrete ect period) was counted. Figure 1 indicated an interest in returning • Adequate focus on public health examples of service learning graphically illustrates the relative to school for their BS degrees. nursing in the curriculum, and projects that would increase the success of the 3 objectives. The seminar series helped to well-planned and well-managed students’ knowledge of public better prepare current and field experiences, can dramati- RESULTS health nursing and contribute prospective preceptors for their cally increase students’interest to improving the health of the roles by familiarizing them with in public health nursing careers. community; and assured the Objective 1 community assessment as cur- preceptors that clinical faculty The school placed 66 students rently taught to students. Partici- would maintain ongoing com- (compared with 32 the previous pating public health nurses found munication and provide help year) in public health department the content useful not only for supervising appropriate and settings. These students were both their roles as preceptors but also productive clinical placements traditional undergraduate stu- for their continuing education. for students. dents and Registered Nurse (RN) The first seminar had viewers at • Public health topical seminars. to Bachelor of Science Comple- 26 sites throughout the state. The project team provided tion Program students (i.e., RNs Follow-up phone calls and in- quarterly seminars that focused working toward a BS degree). The quiries regarding the series evi- on targeted community assess- students completed a variety of denced a pressing need for addi- ment and were open to stu- service learning projects that in- tional sessions on the topic. dents, public health nurses, and creased their knowledge of public Offering continuing education faculty, a need identified by the health nursing and contributed to units for the seminars provided Public Health Nursing Advisory improving the health of the com- an additional incentive for nurses, Group. The seminars were munity. Examples are provided even though these units are not broadcast live to local health (box page 1105). required for nurses in Virginia. departments through a new statewide video-conferencing Objective 2 DISCUSSION system at the State Health De- In the school year 2002–2003, partment. The sessions were only 2 students stated that they This project demonstrated recorded and distributed to would consider a career in public how to take a straightforward ap-

July 2005, Vol 95, No. 7 | American Journal of Public Health Moon et al. | Peer Reviewed | Field Action Report | 1103  FIELD ACTION REPORT 

proach in solving a public health 90 problem. In the first step of the Baseline project, the project team assessed 80 the problem and identified the Goal 70 problem’s primary barriers to Results resolution. In this case, the prob- 60 lem was identified as an insuffi- 50 cient number of students choosing to enter the public health 40 workforce, as well as inadequate 30 preparation for those students who did enter the public health Number of Students or Preceptors 20 workforce. The greatest barrier 10 to ensuring appropriate public health preparation for these stu- 0 Objective 1: Students Objective 2: Students Objective 3: Number of dents was finding preceptors who placed in population- reporting they will public health nurses focused settings consider a career in public serving as preceptors were willing and able to provide health nursing meaningful practical experience for the students. For the next Objective step of the project, the project team involved key stakeholders FIGURE 1—Relative success of the Public Health Nurses for Virginia’s Future project. (the Public Health Nursing Advi- sory Group), in the design and implementation of the solution. Members of the advisory group measured at baseline and at the used to plan the project, develop with potential and current pre- understood firsthand the prob- conclusion of the intervention. the Public Health Nursing Advi- ceptors. The leadership advisory lems facing public health nurses The collaborative relationship sory Group, organize meetings group can be convened annually in Virginia and appreciated the that developed between the for the advisory board, conduct to review the curriculum, plan opportunity to work with VCU to school of nursing and state and outreach to health department for the following year’s seminars, help shape the educational expe- local public health organizations personnel, develop and present collect additional suggestions, rience of their future colleagues. was a key to the success of the the seminars, attend telephone and review placement recom- Having helped design a solution, project. The school’s service con- and in-person grantee meetings, mendations. Through these ef- they were invested in the proj- tract with the Virginia Depart- and conduct the evaluation and forts, the school of nursing fac- ect’s success. ment of Health covered all public reporting of results. With the ulty will maintain ongoing The third step was to plan and health departments throughout infrastructure in place—most communication with public implement an intervention to the state, greatly facilitating entry notably the leadership advisory health nurses in leadership roles, confront the problem and its as- to new public health settings group and the outreach presenta- management, and field settings, sociated barriers. The interven- once a willing preceptor was tions for potential preceptors— simultaneously improving the tion incorporated multiple meth- identified. The state public health the program can be continued quality of the course content and ods, from face-to-face meetings leadership’s unequivocal support with moderate effort and no ad- clinical experiences for student to statewide educational broad- encouraged local public health ditional funding. nurses. With improved prepara- casts. It focused on opening nursing officials to invest the nec- The success of the program tion and positive field experi- 2-way lines of communication essary time to benefit from the will build on itself. For example, ences, it is our expectation that between faculty and nurses, pro- seminars, the interaction with the the team now has additional more students will choose to viding nurses in the field with school of nursing, and student examples of successful student enter public health nursing the information and tools they placements. projects that can help new pre- careers. need to be effective preceptors ceptors identify similar projects NEXT STEPS for public health nursing stu- within their settings (box page About the Authors dents, and providing counsel to 1105). Some of the outreach Martha W. Moon, JoAnne K. Henry, and the course faculty to ensure that The project was accomplished functions can be delegated to Phyllis Kirsch are with the Virginia Com- the curriculum remained rele- with funding for 10 hours per clinical faculty, who can perform monwealth University School of Nursing, Richmond. At the time of this project, vant and current. Finally, indica- week of faculty time over the outreach duties during the Karen Connelly was with the Virginia De- tors of program success were course of one year. This time was course of their normal contact partment of Health, Richmond.

1104 | Field Action Report | Peer Reviewed | Moon et al. American Journal of Public Health | July 2005, Vol 95, No. 7  FIELD ACTION REPORT 

Requests for reprints should be sent to MODELS OF SUCCESSFUL STUDENT FIELDWORK ARISING FROM THIS PROJECT Martha W. Moon, RN, PhD, MPH, Vir- ginia Commonwealth University School of • Educating new teen mothers about child development.Two students conducted a project to mitigate an identified Nursing, PO Box 980567, Richmond, VA knowledge deficiency regarding child development among teen mothers.The students developed a booklet with information 23298-0567 (e-mail: [email protected]). This article was accepted January 22, about developmental milestones, that included information on when the milestones should be achieved, sources of 2005 delays in achieving those milestones, when to seek professional care because of those delays, suggestions for stimu- lating the child in areas where the child was lagging, and check boxes to mark as the child achieves each milestone. Contributors The students produced the booklet with a magnetized backing so that the young mothers could keep it on their re- M.W. Moon and J.K. Henry originated the idea for the project, secured funding, frigerators and refer to it regularly.The students also developed and presented a seminar to a group of teen mothers provided leadership, and conducted the and fathers, during which the developmental milestones were explained, the booklet was reviewed, and the parents seminar series. M.W. Moon drafted the article. K. Connelly was instrumental in were shown how to use the booklet.The public health nurses in the health department plan to continue using the book- convening the Public Health Nursing let and the seminar in their Resource Mothers Program. Advisory Group and gaining the support of public health nurses throughout the • Designing and conducting fire safety education.A local public health department identified a need for fire safety ed- state. P. Kirsch provided clinical coordi- ucation for children.A student worked with 150 kindergartners in a county school, using the Freddie Fire Safety Cur- nation. All authors reviewed subsequent drafts of the article. riculum from South Carolina (accessible at http://www.llr.state.sc.us/freddie.asp). This curriculum teaches children to identify items that can get hot, to recognize the sound of a smoke detector, to go to an adult if they find matches Acknowledgments or a lighter, to “stop, drop, and roll” if their clothes catch on fire, to crawl low to the ground if they encounter a smoke- This project was funded by the Division filled room, and to recognize a firefighter in protective gear. The student gave the children educational kits with a of Nursing of the US Health Resources and Services Administration from July 1, “homework assignment” to construct or review fire escape routes with their parents.Two weeks later, she conducted 2003, through June 30, 2004 (grant a follow-up session in which she reinforced the lessons and reviewed the fire escape plans the children and their par- D52HP00583). The Virginia Depart- ment of Health provided substantial in- ents had prepared. kind support through the use of its • Conducting targeted community assessment.One student conducted a targeted community assessment to determine statewide video-conferencing system. a city health department’s need for tuberculosis control and prevention activities.The city was considering a new pro- References gram to provide chest x-rays for postpartum mothers with a particular risk profile.While tuberculosis is a significant 1. Gebbie KM, Hwang I. Preparing problem in some Virginia localities,the data revealed that tuberculosis is not a significant problem in the city she stud- currently employed public health nurses ied.This analysis helped the local health department better target resources to more pressing public health needs in for changes in the health system. Am J Public Health. 2000;90:716–721. that locality. 2. Reed J, Moffatt S, King M, et al. Updating child health resource book.Two students worked with a county health department to provide the child health The impact of the nursing shortage on • public health nursing by the Quad division with a comprehensive,up-to-date resource book that would be used by public health nurses and outreach work- Council of Public Health Nursing Orga- ers to refer clients to community resources.The current version had not been updated for 9 years. Using the current nizations, July 2001. Available at: version as a starting point,the students called all listed resources to update information.They surveyed health department http://www.astdn.org/publication_im- pact_nursing_shortage.htm. Accessed personnel to find out what information the staff needed in the resource book and to identify additional resources.They May 5, 2005. used the Internet, telephone directory, and other sources to find more local service providers. The final listings contained detailed information about each service provider, including whether Medicaid was accepted and whether various languages were known in order to attend to monolingual clients.

July 2005, Vol 95, No. 7 | American Journal of Public Health Moon et al. | Peer Reviewed | Field Action Report | 1105  COMMENTARY 

“Whistle While You Work”: A Historical Account of Some Associations Among Music, Work, and Health

Music has long been a | Gordon Marc le Roux, BMedSc Hons uniting force among workers. Music can improve team spirit and provide an enjoy- Music has a power unlike any- ries that highlight the dangers of would sing hymns and carols “in- able diversion, but it is most thing else. It is present in all cul- bygone practices. Workers some- cessant during the day” to pre- useful in expressing the true tures and societies, through hap- times found themselves trapped vent themselves from falling feelings of a sometimes des- piness and hardship. Music in the rapidly spinning leather asleep.11(p14) perate community. crosses all boundaries and unites straps of Victorian factories. Small In the days before the dark Over time, a variety of mu- members of every walk of life, children lost digits trying to ex- clouds of the Industrial Revolu- sical media have emerged to but none more than workers. tract objects blocking the mation descended, laborers could match the prevailing condi- Music in the workplace is likely chinery’s cogs. Miners breathed work from home and pursue tions at work: the folk songs 10 to have been present in the earli- in stone dust and later developed their musical hobbies too. Some of 19th-century handloom est societies. It can be beneficial silicosis.7–9 Over time, many even “made music a special weavers, the songs of indus- for alleviating the monotony of safety features have been de- study.”11(p15) The workers would trial Britain’s trade union members, the workers’ radio repetitive labor. There is no signed to minimize health haz- meet in each others’ homes or programs of the 1940s. clearer illustration of this than ards and accidents in the work- public houses to sing solo or in Associations have arisen the classic scene from the Walt place. Nevertheless, workers’ groups. Furthermore, “[t]hese to encourage and coordinate Disney movie Snow White and psychological state is always a worthy men made a large sacri- musical activities among the Seven Dwarfs in which that consideration, and this is where fice of time and labour in the workers, and public aware- loveable septet sings “Whistle music fits into the history of oc- cultivation of music, principally ness of the hazards of some While You Work.” cupational health. for the love of music.”11(p15) occupations has been pro- Sundstrom summarizes the moted through music. (Am limited research available that MUSIC AS PASTIME SONGS FOR THE J Public Health. 2005;95: examines the benefits of music in WORKERS’ UNIONS 1106–1109. doi:10.2105/AJPH. the workplace.1 Music motivates Some European composers 2004.042564) workers, and many workers find are celebrated for drawing upon Singing in factories would music enjoyable. It decreases the folk music of their countries, soon end, as the noise of the In- boredom and leads to increased such as the 19th-century Czech dustrial Revolution’s machines productivity, perhaps partially Bedrˇich Smetana and the early drowned out the human voice. because people work in time 20th-century Hungarian Béla As Sundstrom notes, some of with the beat. There is some evi- Bartók. However, before the In- the quieter industries did try to dence that music is associated dustrial Revolution other genres boost morale by hiring women with a decrease in errors in man- of music were just as common to sing among the workers.1 ufacturing.1 Music is said to im- as folk music in rural communi- Other industries hired company prove perfomance by increasing ties. Elbourne notes that in some orchestras and formed glee psychological arousal, and vigi- English communities not only clubs so that the workers could lance.2–5 These last 2 points are was folk music popular, but so at least enjoy some positive crucial for decreasing the num- were religious hymns and music experiences.1 ber of occupation-related acci- by Handel.11 The performers of Some trade union songs have dents. Finally, making or listen- such works were often laborers, been traced to the 19th century, ing to music can increase including handloom weavers in particularly those of agricultural psychological well-being, so al- Lancashire.11 These early 19th- workers. Such songs as “Stand though music may not be di- century weavers were always Like the Brave” encouraged la- rectly responsible for preventing “singing away to the click of the borers “to join the Union and accidents it can certainly have shuttle”11(p14) and even developed fight for better conditions.”12 ( p8) beneficial effects on mental “a rather saucy type of song.”11(p6) The great London dock strike of health and mood.6 Child laborers, who were en- 1889 was a successful cry from The history of occupational deavoring to finish their work be- the unskilled laborers who had health includes many horrific sto- fore the Christmas holiday, struggled to retain secure jobs

1106 | Commentary | Peer Reviewed | le Roux American Journal of Public Health | July 2005, Vol 95, No. 7  COMMENTARY 

alongside their more skilled The Workers’ Music Associa- the use of the WMA’s choirs The value of the WMA did counterparts. Subsequently, trade tion (WMA), founded in 1936, and the public (Figure 1). Best not end with World War II. In unions spread and broadened was initially formed to coordi- sellers included Popular Soviet 19 60 and 1961, the variety of their support, indiscriminately, nate the musical activity of work- Songs (selling more than 20 000 trades and professions attracted for all groups of workers.10 Many ers who were members of some copies) and Red Army Songs, to the WMA was diverse, includ- songs grew in response to this 44 choirs and 5 orchestras in the both published during World ing “[e]ngineers, miners, teach- flourishing trade union move- London Choral Union and Coop- War II, as well as the Pocket ers, medical students, electri- ment, including “The Red Flag.” erative.14 ,15 The WMA itself was Song Book (1948) and The cians, clerks, printers . . . and a The railway industry also rooted in communism; many in Shuttle and Cage (1954).14 euphonium-playing plumber.”17 ( p8) developed a satirical repertoire. this group, including its founder In 1940, the WMA founded The WMA exists to this day, Ward, for example, says that the and longtime president, com- the Topic Record Club to distrib- although much of its work now “isolation and loneliness [and] poser Alan Bush (1900–1995), ute records of popular workers’ focuses on promoting musical long hours” of railway workers were members of the British songs (including those of new education among the public. allowed “periods for reflection” Communist Party.15 However, Soviet composers) to its members and the “high risk of death and this political leaning was proba- each month.14 WMA members THE NEW MEDIUM: mutilation” among railway work- bly a reflection of the times; the also formed their own musical RADIO ers prompted a body of song Soviet Union had entered the ensembles, performing concerts to rival that of miners and war, and a sense of “friendship to benefit the war effort, trade During World War II, the weavers.13(p13) He also mentions was fostered” by the fact that the union members, other war radio was essential for instantly the high rates of accidents and Soviet Union “was an ally against workers, and the armed forces.14 disseminating information. It was deaths suffered by railroad Fascism.”15 Consequently, the Direct participation in music- often the harbinger of doom, but guards, shunters, and way men, British developed a fresh interest making usually provides a higher it was also a source of comfort which resulted in such song in Russian music.15 level of satisfaction than simply for many. On June 23, 1940, the themes as “Don’t say you heard The WMA successfully pro- listening to music. Therefore British Broadcasting Corporation it from me,” “Only one killed,” moted Soviet culture through making music is an effective way (BBC) first transmitted a radio and “Done to death.”13(p15) There songs, concerts, lectures, and to allay one’s worries and fears, program that would run for 27 was, not surprisingly, much politi- publications.15 Many booklets of lessen the burden of responsibil- years: “Music While You Work.” cal censorship of these songs, songs, in fact quite “a respect- ity, and achieve a sense of re- It was a twice daily, half-hour which were highly critical of able library of propaganda freshment before returning to program of music “meant spe- the railway companies. For the music,”14(p4) were published for the “real world.” cially for factory workers to lis- workers themselves, it was often ten to as they work.”18 ( p3) Many the case that “to sing meant people can still recall the famous the sack.”13(p15) theme song of this program, The people’s flag is deepest red, “Calling All Workers.” It shrouded oft our martyr’d dead, MUSICAL The BBC explained its dra- And ere their limbs grew stiff and cold, ORGANIZATIONS matic rescheduling of program- Their hearts’ blood dyed its every fold. ming in the article “Radio in Music at work seems to have Wartime: Should It Be Grave or Then raise the scarlet standard high, become less common as indus- Gay?”18 The BBC realized that Within its shade we’ll live and die! tries grew. This decline must radio had grown in popularity Though cowards flinch and traitors sneer, have been the impetus for the because new listeners tuned in We’ll keep the red flag flying here! foundation of labor bands and particularly for the news.18 choirs who no doubt rehearsed Therefore, such programs as rac- With heads uncovered swear we all during their leisure time. In some ing results and sporting commen- To bear it onward till we fall; cases, these groups were highly taries were discontinued in order Come dungeons dark or gallows grim, successful. Perhaps most popular to be sensitive to “the prevailing This song shall be our parting hymn. were the colliery brass bands, mood of the nation.”18 ( p3) Be- 16 which often performed in con- Source. 48 Songs: Community Singing. cause “some say that they are certs and competitions and FIGURE 1—“The Red Flag,” a socialist British labor anthem that giving long hours to war-work achieved extraordinary levels of reflects on the solidarity of this movement during adversity, was and they look to radio for excellence. Unions and associa- inspired by the great London dock strike of 1889 and has re- amusement and diversion to re- 12 tions arose to organize these mained popular to this day. fresh them and help them to ensembles. carry on,”18 ( p3) the BBC sched-

July 2005, Vol 95, No. 7 | American Journal of Public Health le Roux | Peer Reviewed | Commentary | 1107  COMMENTARY 

uled Music While You Work at tion.24 This trend was soon rec- fashion, highlighting issues such About the Author midmorning and midafternoon, ognized by others. In the late as compensation and safety in The author is a medical student at the School of Medicine, University of Birming- where one might expect a dip in 1950s the company Muzak was the hope of remedying the ap- ham, England. concentration.19 formed; it provided music to be palling conditions. The workers Requests for reprints should be sent No recording of any early played in offices to “subtly stimu- also speak of new machinery, to Gordon Marc le Roux, BMedSc Hons, c/o Robert Arnott, Centre for the History episode of Music While You late employees during times better conditions, and better pay. of Medicine, The Medical School, Uni- Work remains in the archives, when they otherwise worked In contrast to the traditional versity of Birmingham, Edgbaston, but the schedules suggest that slowly.”1(p170) folk music of workers, Big Birmingham B15 2TT, United Kingdom (e-mail: [email protected]). a wide variety of music would Hewer’s aim was to educate the This article was accepted September have catered to most tastes. The MUSIC FOR PUBLICITY public about issues that, at the 16, 2004. program featured dance bands time, were as remote to the re- playing popular, high-spirited During the 1950s and 1960s, searchers themselves as to the Acknowledgments music of the time.20 The very 8radio documentaries were pro- public. MacColl writes that for I thank Robert Arnott, director, and particularly Jonathan Reinarz, Wellcome first program featured the duced by Charles Parker, former Parker, meeting the miners was a research fellow, at the Centre for the Organolists, who played the WMA member and BBC em- “shattering experience.”27 Parker History of Medicine, University of organ, drums, and piano.21 ployee. The documentaries in- “confessed to feeling utterly Birmingham Medical School, for their encouragement and critiques of earlier Some cheaper, yet no doubt clude many tape-recorded ac- uneducated in the presence of versions of this article. I am grateful popular, programming alterna- counts from workmen of the [the] miners” and his Panglossian to the staff at the Birmingham City tives included “music of the time, as well as some renditions “view that everything is all right Archives in the Birmingham Central 18(p16) Library for allowing me access to the films, on record.” of traditional workers’ folk in the best of all possible worlds” Charles Parker Archive. Finally, many Geiger and his orchestra, com- songs.25 Many of the program’s was soon dashed.27 The radio thanks to John Jordan, conductor and prising a cello, violins, bass, and themes were set to music by ballad summed up the stark real- archivist at the Workers’ Music Associa- tion, for kindly providing me with copies Hungarian cimbalon, were fre- Ewan MacColl and Peggy Seeger, ity of the coal miner’s life, but it of some interesting material that relates quent contributors,22 performing who “were at the forefront of the also demonstrated the sense of to the archive. love songs, folk dances, and British folksong revival.”25 In unity and pride among these waltzes by Strauss.23 This lively fact, MacColl and Seeger regu- workers. References combination helped workers get larly recorded folk music for 1. Sundstrom E. Music. In: Work Places: The Psychology of the Physical through their shifts. Topic—the WMA’s own record CONCLUSION Environment in Offices and Factories. The bands were instructed to label—which has been called “the Cambridge, United Kingdom: Cam- play medleys rather than individ- significant disseminator for a The associations between bridge University Press; 1986:167–178. ual tunes, in order to keep the growing folk network.”15 To - music and work are interesting 2. Zimny GH and Weidenfeller EW. 20 Effects of music upon GSR and heart- workers’ attention. Further- gether, Parker, MacColl, and and important. The folk tradi- rate. Am J Psychol. 1963;73:311–314. more, the musicians were to keep Seeger created a new genre tion within the workplace was 3. Davenport WG. Arousal theory 25 pace with the “rhythms of the termed the “radio ballad.” drowned out by the noise of the and vigilance: schedules for background workbench” so that production One of these productions, Big Industrial Revolution. When work- stimulation. J Gen Psychol. 1974;91(4): 51–59. would not slow down.20(p23) In Hewer, was transmitted on the ers could no longer sing at work, 4. Fox JG and Embrey ED. Music—an 19 42, the song “Deep in the BBC Home Service on August they formed union ensembles and aid to productivity. Appl Ergonomics. 25 Heart of Texas” was banned from 18, 1961. This documentary performed during their leisure 1972;3(4):202–205. the program because it contained told the story of Britain’s time. During World War II, music 5. Davies DR, Lang L, and Shackle- a participatory handclapping sec- coalminers’ experiences. One sec- was piped into the workplace to ton VJ. The effects of music and task difficulty on performance at a visual tion that tempted laborers to stop tion begins with the words “Coal keep up morale and enhance vigilance task. Br J Psychol. 1973;64(3): 20 1,18,20 work and join in. is a thing that’s cost life to get,” production. Later in the 383–389. Music While You Work proba- and from then on the listener is 20th century, music took on a 6. Terry GR. Office Management and bly minimized the occurrence presented with many accounts of new role, becoming a medium Control. Homewood, IL: Richard D. of accidents by improving alert- ill health and death in the coal through which the public could Irwin; 1975:451. ness and team interaction. The mines: “You’re eating coal, you’re be educated and informed about 7. Engels F. The Condition of the 25 Working Class in England. London: minister of labor, in 1940, wrote breathing coal,” “reduced to the conditions of workers. Penguin Books; 1987:182. that the program also “made the nothing . . . no lungs to breathe,” Music in the workplace has in- 8. Select Report of the Commission- hours pass more quickly and re- “he’s worked in the pits since he creased efficiency, lifted spirits, ers on the Employment of Children sulted in increased produc- was . . . twelve year old . . . he’s and even bound society together. (Trades & Manufactures). 1843; P.P. vol. XIII: 195-199. tion.”24(p60) In fact, some partici- got inside his lungs a good tomb- Music still does all of this, and it 26 9. Hunter D. The Diseases of Occupa- pating factories during the war stone, of solid coal-dust.” The is one force that can help carry tions (6th ed.). London: Hodder and noted a 20% increase in produc- documentary continues in this us forward into the future. Stoughton; 1978:963–964,982.

1108 | Commentary | Peer Reviewed | le Roux American Journal of Public Health | July 2005, Vol 95, No. 7  COMMENTARY 

10. London Metropolitan Archives. Radio Times: J British Broadcasting The Great Dock Strike, 1889. Informa- Corporation.1940;67(873):12. tion leaflet No 19.1999. Available at: 24.Reynolds W. A note on “Music http://www.cityoflondon.gov.uk/NR/ While You Work.” In: BBC Yearbook rdonlyres/C27142E4-DEFC-4161- 1945. London, United Kingdom: British AE95-F4B45716E7D4/0/LH_LMA_ Broadcasting Corporation; 1945:60. dockstrike.PDF. Accessed May 20, 2005. 25. Aston L. The Radio-Ballads 1957–1964 [sleeve notes]. In: MacColl E, 11. Elbourne RP. “Singing away to the Parker C, Seeger P. The Big Hewer: a click of the shuttle”: musical life in the Radio-Ballad About Britain’s Coal Min- handloom weaving communities of Lan- ers [compact disc]. London: Topic Rec- cashire. Local Historian.1976;12(1): ords; 1966. TSCD 804. 13–17. 26. MacColl E, Parker C, Seeger P. The 12. Miller J. More on workers’ songs. Big Hewer: a Radio-Ballad About Bri- Music and Life. January 1962:6–10. tain’s Coal Miners [compact disc]. Lon- Located at: Charles Parker Archive, don: Topic Records; 1966. TSCD 804. Birmingham City Archives, Central Li- brary, Birmingham, United Kingdom. 27. MacColl E. The Big Hewer: a CPA/1/8/5. Radio-Ballad about Britain’s Coal Miners [sleeve notes]. In: MacColl E, Parker C, 13.Ward J. Songs of protest and rail- Seeger P. The Big Hewer: a Radio-Ballad way industrial songs. Music and Life. about Britain’s Coal Miners [compact January 1975:13–15. Located at: disc]. London: Topic Records; 1966. Charles Parker Archive, Birmingham TSCD 804. City Archives, Central Library, Birming- ham, United Kingdom. CPA/1/8/5. 14 . Sahnow W. WMA: Twenty One Years. London, United Kingdom: Work- ers’ Music Association; 1957:4–6. 15.Brocken M. The Battle of the Field: the political context of the hagiog- raphy of the second British folk revival. The Emergence of AIDS Available at: http://www.mustrad.u-net. The Impact on Immunology, com/topic.htm. Accessed May 9, 2005. Microbiology and Public Health 16.Connell J. The Red Flag. In: 48 Songs: Community Singing. London, Edited by Kenneth H. Mayer, MD, United Kingdom: Workers’ Music Asso- ciation; 1960–1963:8. Located at: and H. F.Pizer Charles Parker Archive, Birmingham City Archives, Central Library, Birming- his unique book highlights the lessons learned from and ham, United Kingdom. CPA/1/8/5. about AIDS over the past 20 years, and highlights the 17. Music Their Holiday: The Story of T the W.M.A. Summer School. London, knowledge that may advance worthwhile strategies for United Kingdom: Workers’ Music Asso- combating HIV and AIDS in the future. The 11 chapters in- ISBN 0-87553-176-8 ciation; 1960–1961:8. Located at: clude: The Virus Versus the Immune System, How Charles Parker Archive, Birmingham 2000 ❚ 350 pages ❚ softcover City Archives, Central Library, Birming- $13.95 APHA Members Infectious is Infectious, The Race against Time: The ham, United Kingdom. CPA/1/8/5. $19.95 Nonmembers Challenge for Clinical Trials, Sex and Drugs and the Virus, plus shipping and handling 18.Radio in wartime: should it be and more. grave or gay? Radio Times: J British This book is an ideal reference for Infectious disease spe- Broadcasting Corporation.1940; ❚ ❚ 67(873):3. cialists Epidemiologists Public health practitioners ❚ Clinicians ❚ All those concerned with AIDS. 19.For the forces: Wednesday June 26. Radio Times: J British Broadcasting Cor- poration.1940;67(873):16. ORDER TODAY! 20. Barnard S. On the Radio: Music American Public Health Association Radio in Britain. Milton Keynes, United Publication Sales Kingdom: Open University Press; Web: www.apha.org 1989:23. E-mail: [email protected] 21.For the forces: Sunday June 23. Tel: 888-320-APHA Radio Times: J British Broadcasting FAX: 888-361-APHA EA01J7 Corporation.1940;67(873):7. 22. Home service: Thursday July 4. Radio Times: J British Broadcasting Corporation.1940;68(874):24. 23. Home service: Monday June 24.

July 2005, Vol 95, No. 7 | American Journal of Public Health le Roux | Peer Reviewed | Commentary | 1109  HEALTH POLICY AND ETHICS 

Invoking Rights and Ethics in Research and Practice

Brazil and Access to HIV/AIDS Drugs: A Question of Human Rights and Public Health

| Jane Galvão, PhD

I explore the relationship be- demic.1–7 In this article, I analyze The relative success of the ARV been reported in the country, tween public health and human the links between public health program in Brazil reflects a some- comprising 220783 men and rights by examining the Brazil- 24 and human rights, using the what privileged position compared 89527 women ; of this total, ian government’s policy of free 24 Brazilian government’s policy of to lower-income countries, some approximately 48% have died. and universal access to anti- free and universal access to anti- of which have higher levels of The epidemic is spreading, par- retroviral medicines for people HIV infection. In turn, using the ticularly among the poor, with HIV/AIDS. retroviral medicines (ARVs) for Brazilian government’s manage- women, and those living outside The Brazilian government’s people with HIV/AIDS as an ex- 25 management of the HIV/AIDS ample. Although I refer to the pro- ment of HIV/AIDS as a model the urban centers. In people epidemic arose from initiatives duction of generic versions of may not transfer easily to other aged 15 to 49 years, the esti- 22 in both civil society and the AIDS drugs as well as the role of nations. However, Brazil’s expe- mated prevalence of HIV is governmental sector follow- international pharmaceutical com- rience offers inspiration for finding 0.65%, with approximately ing the democratization of the panies, both topics are explored in appropriate and life-saving solu- 600000 people infected with 23 21 country. The dismantling of au- greater detail elsewhere.8–15 tions in other contexts. To gain HIV. Of this total, approxi- thoritarian rule in Brazil was ac- Globally, ARVs remain be- a wider perspective on Brazil’s mately 200000 know their HIV companied by a strong orien- 26,27 yond the reach of the majority of HIV/AIDS policy—and in particu- status, the majority of which tation toward human rights, people with HIV/AIDS.16–18 Of lar the synergy between health are registered in the public which formed the sociopolitical and human rights—I solicited com- health system and are receiving framework of Brazil’s response the 6 million people worldwide ments from several individuals, treatment. According to figures to the HIV/AIDS epidemic. who needed ARVs in 2003, Even if the Brazilian experi- fewer than 8% were receiving quoted in this article, who work from 2003, of those individuals ence cannot be easily trans- them.18 Although Brazil is consid- for Brazilian and international or- in the public health system, ferred to other countries, the ered a middle-income country,19 ganizations that are currently at 135 000 were undergoing highly 28 model of the Brazilian govern- its government provides ARVs to the forefront of the struggle active antiretroviral therapy. ment’s response may nonethe- its constituents free of charge. To against HIV/AIDS. By reviewing less serve as inspiration for make such a policy viable, the Brazil’s policies and relating other LIMITING THE COSTS OF finding appropriate and life- government has limited the people’s experiences, I hope to ARV THERAPY saving solutions in other na- drugs’ high cost by producing demonstrate the importance of tional contexts. (Am J Public some ARVs domestically and by community mobilization, political Although other countries in Health. 2005;95:1110–1116. doi: will, international solidarity, and fi- 10.2105/AJPH.2004.044313) negotiating with international Latin America have established pharmaceutical companies to im- nancial commitment in the fight programs to improve access to port other ARVs2,20; of the 15 against HIV/AIDS. treatment for people with HIV/ FOR SEVERAL YEARS I HAVE ARVs utilized in the country in AIDS,29 Brazil’s program is the studied Brazil’s management of 2002, 7 were produced in local HIV/AIDS IN BRAZIL most far reaching.30 The program the HIV/AIDS epidemic and the laboratories, either public or pri- provides state-of-the-art ARV ways in which Brazil’s policies vate, and the remainder were In 1980, the first case of AIDS treatment to people in need, free havecontributed to the global purchased on the international in Brazil was registered. By De- of charge, through the public fight against the HIV/AIDS epi- market.2,21 cember 2003, 310310 cases had health system, and the govern-

1110 | Health Policy and Ethics | Peer Reviewed | Galvão American Journal of Public Health | July 2005, Vol 95, No. 7  HEALTH POLICY AND ETHICS 

ment controls the costs of the pro- made access to ARV therapy pos- tion with the patent-holding com- NGOs and human rights advo- gram by encouraging local labora- sible for the Brazilian population. panies that manufacture those cates have linked HIV/AIDS to tories to produce ARVs, which Without it, the price of the drugs medicines. As a result, in January human rights and have demon- decreases Brazil’s need to import would be beyond Brazil’s reach.” 2004, the National AIDS Pro- strated that human rights viola- vital drugs from foreign countries. The Brazilian government’s gram announced that it had suc- tions increase the spread of During his tenure as director of strategy for controlling the costs cessfully negotiated reduced HIV.39–42 For example, people Brazil’s National AIDS Program, of AIDS medicines has not been prices for those drugs and other affected by HIV/AIDS who are Paulo Teixeira understood that without challenges and obstacles. related medicines, thereby antici- living in areas where discrimina- the success and sustainability of In 2001, the country was in- pating savings of almost US$100 tion, stigmatization, and threats Brazil’s policy of access to ARVs volved in international disputes million for 2004.33,34 Because of against individuals with HIV/ depended on effective strategies about its program of access to these price reductions, the esti- AIDS are high are less inclined to control costs. He believes that AIDS medicines.2 In that year, mated costs of ARV treatment to seek testing, thereby postpon- maintaining political will while the World Trade Organization in 2004 were around US$170 ing treatment if available, which fostering new international al- (WTO) accepted a request for a million.33 means that opportunities to de- liances for more favorable regula- panel by the United States, which Concern for human rights com- crease HIV transmission are lost. tions on pharmaceutical-related was challenging Brazil’s patent bined with the urgent need for Gruskin et al. described 3 trade and intellectual property laws, laws that permit the com- access to treatment by people stages through which the rela- will continue to be central. Ac- pulsory license of patents under with HIV/AIDS has bolstered tionship between HIV/AIDS and cording to Teixeira, special conditions. At its heart, wider efforts to lower the costs human rights has proceeded.42 the US challenge questioned of ARVs. Peter Piot, executive In the first stage, 1981–1986, The biggest challenge for the future of the Brazilian National Brazil’s commitment to producing director of the Joint United Na- human rights advocates pitted AIDS Program is the mainte- ARVs nationally; explicitly, how- tions Programme on HIV/AIDS themselves against public health nance and the sustainability of ever, the United States was chal- (UNAIDS), recognized the role of officials who proposed measures the policy of free and universal access to ARVs. This is directly lenging the prospective patent vi- Brazil’s HIV/AIDS policies in fa- such as mandatory testing and linked to a change in the world olations that would occur as a cilitating this development, saying, quarantine to counter the emerg- economic order with regard to result of Brazil’s program. In June ing epidemic. In the second medicines. It will depend not 2001, the United States withdrew The Brazilian experience has stage, beginning around 1987, only on a firm position by played a key part in changing ex- Brazil, in the defense of its pol- its complaint before the WTO. To pectations in the interpretation of officials openly acknowledged icy of production and distribu- date, Brazil has not produced any the World Trade Organization’s that mandatory testing and quar- tion of ARVs, but also on the of its medicines under compul- TRIPS (Trade-Related Aspects antine undermined the efficacy strengthening of international of Intellectual Property Rights) alliances that are beginning to sory licensing, and the ARVs that Agreement. When the Doha of prevention programs. In the form with the World Health are currently produced are those Ministerial Meeting of the World third phase, which started in the Organization, the World Trade medications whose introduction Trade Organization at the end of late 1980s, research developed Organization, the United Na- 2001 declared that the TRIPS tions Special Session on HIV/ predated Brazil’s signing of the Agreement ought not stand in from the idea that vulnerability AIDS, and the Global Fund Trade-Related Aspects of Intellec- the way of AIDS responses, it in was a key to infection. This re- to Fight AIDS, Tuberculosis and tual Property Rights (TRIPS) effect acknowledged the ethical search was then developed and Malaria. It is extremely impor- and practical imperatives repre- tant that these alliances are agreement. In September 2003, sented by Brazil’s generic anti- disseminated by human rights ac- maintained and intensified to a presidential decree was issued retroviral industry. tivists such as the late Jonathan guarantee the conquests already achieved. that facilitated the importation Mann, a central figure for and of generic medicines.31 At the AIDS AND HUMAN RIGHTS staunch defender of human Increased local production of time, according to Brazil’s rights for people living with ARVs has been integral to the MOH,3 imported name-brand Globally, violating the human HIV/AIDS, and by groups such Brazilian strategy. According to ARVs—nelfinavir, lopinavir, rights of people with HIV/AIDS— as the Global AIDS Policy Coali- Eloan Pinheiro, former director and efavirenz—were consuming through stigmatization, discrimi- tion, which was founded and led of Far-Manguinhos, a Brazilian 63% of the budget for acquiring nation, and violence—is increas- by Mann. During this phase, Ministry of Health (MOH) labora- ARVs.28,32 The possibility of ingly recognized as a central Gruskin et al. argued, “it became tory that produces ARVs, “the importing generic medicines im- problem that is impeding the clear that a lack of respect for local production of AIDS drugs proved Brazil’s bargaining posi- fight against AIDS.35–38 AIDS human rights and dignity was a

July 2005, Vol 95, No. 7 | American Journal of Public Health Galvão | Peer Reviewed | Health Policy and Ethics | 1111  HEALTH POLICY AND ETHICS 

major contributor to the HIV/ almost 20 years, Brazil went to health care, including access them with the needed medica- AIDS problem.”42(p326) With the through a process of democrati- to medicines, improved in the tions by suing state or municipal advent of more effective treat- zation, gradually reconstructing country. At that time, the Uni- governments.62,63 In these strug- ments for controlling the effects civil society53 and formulating a fied Health System (Sistema gles, the judiciary proved to be of AIDS, a fourth phase in the re- new social agenda for areas such Único de Saúde [SUS]) was es- an important ally. The judges lation between human rights and as education and health. During tablished.54,56,57 The SUS of- often ruled favorably, citing HIV/AIDS began: the promotion this period, campaigns such as fered free comprehensive health Brazil’s constitution, which guar- of access to treatment. the Movement for Sanitary Re- care to the entire population, re- anteed that every citizen had a Today, access to treatment inform, which sought to democra- gardless of employment status right to health and the state had creasingly is being advocated as tize health policy and establish or access to other forms of a duty to ensure every citizen’s a human right, a viewpoint that health care as a right for all health insurance.2 However, health. Brazilian lawyer Miriam is playing a prominent role in de- Brazilians, began.20,54 At this prior to the establishment of the Ventura pointed to the codifica- velopments to counter the HIV/ time, Brazilian newspapers also SUS, the national agenda had tion of the policy as the “success- AIDS pandemic.43–48 For exam- began describing the emergence included local production of ful result of a model of action ple, at the April 2001 57th of a new disease, as the first some medicines by state labora- adopted by organized civil soci- Session of the Commission on AIDS cases in the country were tories and free distribution of ety.” This mobilization, she con- Human Rights, the United Na- being reported.3,6 certain medicines by the public tinued, “utilized the language of tions High Commissioner for In Brazil, the early HIV/AIDS health system.58 human rights and the strategic Human Rights approved a resolu- movement relied on experi- People with HIV/AIDS were application of national laws . . . tion that makes access to medical enced activists who had orga- among those who benefited from [and] succeeded in placing on drugs in cases of pandemics such nized against the military the new health system; they the political agenda questions as HIV/AIDS a basic human regime; some of these individu- began to receive drugs for oppor- that affect the life of people liv- right.49 Although this resolution als helped create the first non- tunistic infections, and in 1991 ing with HIV/AIDS, and in so did not have the power of law, governmental organizations began to receive zidovudine doing altered public and state it was nonetheless an important (NGOs) and some came to as- (AZT).3,4,59 In November 1996, policies regarding health care.” step toward establishing the right sume roles in local, state, and the access to medicines policy In 1989, Ventura helped to of people living with HIV/AIDS federal government. Not surpris- became firmly established establish Pela VIDDA—which to receive the medicine and ingly, this first generation of when the president of the repub- means “For the Valorization, In- treatment they need. Also in Brazilian activists approached lic signed a law60 that guaran- tegration and Dignity of People 2001, a declaration was ap- the government about the new teed free distribution of medi- with AIDS”—the legal AIDS ser- proved at the Fourth World disease using strategies they cines to people with HIV/AIDS vice for the first group of Brazil- Trade Organization Ministerial had implemented against the throughout the public health sys- ian people with HIV/AIDS. Conference50 that allowed coun- dictatorship, strategies that in- tem.2,3,21,29,59 ARVs, along with Herbert Daniel, founder and tries to apply for compulsory cluded the demand for the de- medicine for malaria, Hansen first president of the group, de- licensing in order to produce mocratization of access to infor- disease, cholera, hemophilia, dia- nounced the denial of the rights necessary medicines in cases of mation and the defense of betes, schistosomiasis, trachoma, of people with HIV/AIDS, which national public health emergen- human rights.4 In Brazil, the re- leishmaniasis, and filariasis, form he termed “civil death.”64 Daniel cies.51 In 2003, UNAIDS reaf- sponse to the HIV/AIDS pan- the category of medicines the was a writer and a militant of the firmed the relevance of human demic arose from initiatives in MOH has deemed “strategic” for gay movement who also fought rights to HIV/AIDS by establish- both civil society and the gov- treating endemic diseases; these against the dictatorship in Brazil ing a Global Reference Group on ernment and followed the pro- medicines in turn are purchased before being forced into exile; he HIV/AIDS and Human Rights.52 cess of democratization4,55—a by the federal government.5,61 died of AIDS in 1992. context with a strong orientation The government’s commit- In Brazil, the participation of HIV/AIDS AND HUMAN toward human rights. ment to provide AIDS medicines civil society had a key role in RIGHTS IN BRAZIL In 1988, with the reorganiza- resulted, in part, from pressure bringing about and sustaining the tion of Brazil’s public health sys- from civil society, where people Brazilian government’s ARV dis- In the early 1980s, after a tem56 and the adoption of with HIV/AIDS sought to force tribution policy.1,7,62,63 The im- military dictatorship that lasted Brazil’s new constitution, access the health system to provide portance of this contribution was

1112 | Health Policy and Ethics | Peer Reviewed | Galvão American Journal of Public Health | July 2005, Vol 95, No. 7  HEALTH POLICY AND ETHICS 

2,21,29 highlighted by Veriano Terto, Jr, quired infection, are tarred with tunistic infections. However, interfere with adherence to executive director of the Brazil- the same brush of stigma. beyond the biomedical and eco- treatment.68–71 Breaking the vicious cycle of so- ian Interdisciplinary AIDS Asso- cial exclusion is therefore cru- nomic arguments for treatment, a Providing better access to ciation (ABIA), an NGO founded cial both to interrupting trans- rights-based approach focused on HIV/AIDS therapy has become in 1986 in Rio de Janeiro: mission and to maximizing the the inequality that fueled the a global initiative as well. In care and support available to virus’s spread.65,66 Current ABIA 2003, the World Health Organi- The participation of organized people living with HIV/AIDS. civil society in access to AIDS This is perhaps the key to the president Richard Parker, chair of zation launched its “3 by 5” treatment goes back to the tremendous impact of Brazil’s the Department of Sociomedical strategy,18, 43,47,72 which aims to 1996 decision to guarantee eighties, when popular pressure Sciences in the Mailman School of extend access to ARV therapy and progressive political forces constitutionally access to ARV were fundamental in creating a therapy. Not only has this deci- Public Health at Columbia Uni- to an additional 3 million people unified public health system sion led to the quadrupling of versity, stated with HIV/AIDS living in devel- the number of Brazilians access- based on the principles of uni- oping countries by the end of versal access, comprehensive- ing these drugs, it also sent the By affirming universal access 72 ness, and participatory decision signal that people living with to treatment for all those in- 2005. The theme for the XV making. If currently we have a HIV/AIDS were valued citi- fected with HIV, Brazilian International AIDS Conference, zens, whose care was a matter system for the distribution of policy has simultaneously reaf- held in Bangkok, Thailand, in medicines in the public health of entitlement, not of privilege. firmed the rights and citizen- network, and legislation that ship of those who otherwise July 2004 was “Access for all,” guarantees this system, this is ABIA was the first Brazilian would be defined primarily by an assertion that it is time to de- based in the values of univer- AIDS NGO to have as its presi- their broader exclusion in liver the message, the medicine, sality and equality in access to Brazilian society. Because of 73 treatment for all epidemics con- dent and founder someone who this, prevention becomes possi- the help, and the hope to all. templated in the health system, disclosed his HIV positive status. ble, not just as a technical ex- In this manner, Brazil, whose and in public participation, Herbert de Souza—known by the ercise in public health, but as human rights advocates lobbied which underlies and accompa- itself the right to health of all nies the Brazilian public health nickname Betinho—was a former citizens. While the broader so- for health rights and whose gov- policies. political exile. He was a hemo- cial inequalities that shape the ernment placed human rights at philiac, as were his 2 brothers; all epidemic have only become the center of its HIV/AIDS pol- Terto continued, saying, more extreme over time, the 3 became infected with HIV strategic approach to AIDS in icy, has been a vital role model. The participation of civil soci- through blood transfusions in the Brazil has thus been able, in a Resources allocated to disease ety also was fundamental for mid 1980s. One of the first im- targeted way, to mitigate their treatment are often seen as com- including solidarity, respect for worst effects, to respond to the portant struggles carried out by peting with resources available for human rights, and the struggle stigma and discrimination so against prejudice and discrimi- ABIA was the promotion of blood often generated by the epi- disease prevention, a dichotomy nation to the response against safety, a tremendous problem at demic, and to recover the sim- leading to a debate regarding pri- AIDS. These points were fun- ple idea of human dignity, that time in Brazil. After the ap- orities in the fight against the damental for amplifying guaranteed by civil rights, as 74 the notion of health beyond proval of the Brazilian constitu- the most powerful way of re- HIV/AIDS pandemic. In the the search for physical well- tion in 1988, it was forbidden to sponding to the reality of 19 9 0s, The World Bank, for ex- being, and technical measures AIDS-related vulnerability. sell blood in Brazil. Betinho’s ample, did not favor a policy of focused only on the treatment of individuals. In this sense, the brothers died in the mid 1980s, Providing better access to providing AIDS medicines to de- demand for universal and free and Betinho died in 1997.4,7 HIV/AIDS therapy in resource- veloping countries,75 including access to medicines should be poor countries is not only a hu- Brazil.76 At the time, the World seen as a question of making 67 real the right to life, and re- LINKING TREATMENT manitarian imperative but also Bank believed that, with limited spect of the basic human rights AND PREVENTION a viable and financially justified resources, funds should be di- of people living with HIV/ course of action in terms of eco- rected to prevention in order to AIDS in Brazil. In Brazil, there are at least 2 nomic costs and social benefits. limit new infections. More re- Piot made a similar observation: important arguments from an eco- This is apparent when consider- cently, however, the World Bank nomic perspective for maintaining ing the Brazilian initiative of has come to emphasize the im- The HIV epidemic is aided by social exclusion—marginalized free access to AIDS medicines: distributing ARVs—with savings portance of combining prevention populations are the most vul- the impact of ARVs in reducing in both lives and financial re- and treatment.77,78 According to nerable to HIV infection, deaths and the significant reduc- sources—along with studies else- Piot, “[there is an] inextricable whether through sex or needle sharing, and people living with tion in hospitalization and treat- where that suggest socioeco- link between prevention and care, HIV/AIDS, however they ac- ment costs associated with oppor- nomic levels of patients do not which operate together as twin

July 2005, Vol 95, No. 7 | American Journal of Public Health Galvão | Peer Reviewed | Health Policy and Ethics | 1113  HEALTH POLICY AND ETHICS 

pillars of a comprehensive AIDS nomic inequalities.81 Second, the ment of an International Centre for contributing comments that have en- response.” He added that “Brazil Brazilian policy regarding access for Technical Cooperation on riched this article. is perhaps the world’s leading ex- to ARVs needs to be analyzed in AIDS in Brazil to help develop- ample of the synergies available terms of the program’s sustaining countries strengthen their re- References 92 1. Galvão J. AIDS no Brasil: A Agenda between prevention and care.” ability and potential transferabil- sponses to HIV/AIDS. Placing de Construção de uma Epidemia. São 2,22,82 But even with Brazil’s success at ity to other countries. Third, this center in Brazil is not Paulo and Rio de Janeiro, Brazil: Edi- offering treatment to people with Brazil must continue to strive for likely coincidental—UNAIDS tora 34, Associação Brasileira Interdisci- plinar de AIDS; 2000. HIV/AIDS, an assessment is positive results. The experience hopes to draw from Brazil’s rich 2. Galvão J. Access to antiretroviral needed to determine how the na- of other countries such as Thai- experience in responding to the drugs in Brazil. Lancet. 2002;360: tion—which is devoting financial land demonstrates that it is not HIV/AIDS pandemic. 18 62–1865. and human resources to both pre- easy to sustain a successful re- Brazil’s experience and related 3. Galvão J. Brazilian policy for the vention and treatment—is or is sponse against the HIV/AIDS initiatives will merit continued at- distribution and production of antiretro- not succeeding in preventing new pandemic.83–85 In Thailand, se- tention as the world confronts viral drugs: a privilege or a right? [in Portuguese.] Cad Saúde Pública. 2002; infections. vere police repression against in- the growing HIV/AIDS pan- 18:213–219. In order to adequately analyze jection drug users has threatened demic. Providing access to life- 4. Galvão J. Brazil and access to the public health system in prevention measures directed at saving medicines and transfer- AIDS medication: public health as a Brazil54,56,57 and explain how that particular population. In any ring technologies will be human right issue. In: Sydow E, Men- donça ML, eds. Human Rights in Brazil Brazil’s AIDS program was de- national program, it is essential challenges, not only for those in- 2002. São Paulo, Brazil: Social Network veloped, I would need to write to prevent past success from volved with HIV/AIDS but also for Justice and Human Rights and another entire article. However, turning into complacency and for the field of public health as a Global Exchange; 2002:181–188. even in Brazil, when deciding inaction and to remain vigilant whole, posing practical and theo- 5. Galvão J, Passarelli CA, Reingold AL, Rutherford GW. Acesso a medicamen- how to first proceed against the in regard to human rights, par- retical questions that will need to tos para AIDS: lições da iniciativa HIV/AIDS pandemic, there was ticularly for the rights of those be answered in the years to brasileira. Divulg Saúde Debate. 2003; great internal conflict within people who are made most come.93 29:11–22. Brazil’s government, as sectors vulnerable.85 6. Galvão J. Community mobilization and access to medicines: the Brazilian that wanted to prioritize HIV/ New developments linking About the Author non-governmental responses for the AIDS were opposed by sectors HIV to national, international, At the time of the study, Jane Galvão was HIV/AIDS epidemic. Text presented that wanted money and re- and human security,86–89 to- with the Institute for Global Health, Uni- at: Harvard Forum on Human Rights in Brazil, Brazil Human Rights Series: sources allocated elsewhere.79 gether with a growing human versity of California, San Francisco, and the School of Public Health, University of Rights to Health; November 3, 2003; Even today, despite international rights orientation toward people California, Berkeley. Boston, Mass. recognition and growing internal with HIV/AIDS, have con- Requests for reprints should be sent to 7. Galvão J. O modelo brasileiro na support, the Brazilian govern- tributed to new recommenda- Jane Galvão, PhD, International Planned promoção ao acesso aos medicamentos Parenthood Federation/Western Hemi- para AIDS. Sociedad Iberoamericana de ment’s HIV/AIDS policy occa- tions from organizations such as sphere Region, 120 Wall St, 9th Fl, New Información Científica [online database]; 90 sionally receives criticism at the World Health Organization York, NY 10005-3902 (e-mail: jgalvao@ March 11, 2004. Available at: http://www. home from those who believe and others91 to combine preven- ippfwhr.org). siicsalud.com/dato/dat036/04309001. This article was accepted November 9, htm. Accessed April 4, 2005. Brazil should spend more money tion, support, treatment, and care 2004. 8. Attaran A, Gillespie-White L. Do in other areas and less money in responding to the HIV/AIDS patents for antiretroviral drugs constrain 80 on HIV/AIDS. pandemic. Although the Brazilian Acknowledgments access to AIDS treatment in Africa? There are 3 points to consider experience has helped move ac- The first draft of this article was sup- JAMA. 2001;286:1886–1892. regarding Brazil’s HIV/AIDS pol- cess to treatment as a basic human ported by the Fogarty International 9. Chien CV. Cheap drugs at what AIDS Training Program (grant 1-D43- icy, lest the positive lessons de- right beyond abstract discussion, price to innovation: does the compul- TW00003) (School of Public Health, sory licensing of pharmaceuticals hurt scribed herein give a simplistic this approach still poses immedi- University of California, Berkeley) from innovation? Berkeley Technol Law J. 2003; image of a genuinely complex re- ate and practical challenges, par- February 2001 to November 2002. 18(1)1–57. I thank Daniel Hoffman for his com- 10.Patently robbing the poor to serve ality. First, the Brazilian program ticularly in terms of maintaining ments and for translating the original text the rich [editorial]. Lancet.2002;360: in response to the HIV/AIDS political will and sustaining finan- from Portuguese, and Kate MacLaughlin 885. epidemic has been successful cial support. To respond to some for further editing the text. I especially wish to thank Richard Parker, Eloan 11. Kapczynski A, Crone ET, Merson M. even though the country is still of these challenges, UNAIDS re- Pinheiro, Peter Piot, Paulo Teixeira, Global health and university patents. marked by profound socioeco- cently announced the establish- Veriano Terto, Jr, and Miriam Ventura Science. 2003;301:1629.

1114 | Health Policy and Ethics | Peer Reviewed | Galvão American Journal of Public Health | July 2005, Vol 95, No. 7  HEALTH POLICY AND ETHICS 

12.Prusoff W. The scientist’s story. 8,000 deaths a day [editorial]. Wash- ratórios de anti-retrovirais [press republic health in Haiti. N Engl J Med. New York Times. March 19, 2001:A4. ington Post. March 26, 2004:A22. lease]. Brasília, Brazil: National AIDS 2004;350:1483–1486. Program; January 15, 2004. Available 13. Shanker D. Brazil, the pharmaceu- 24.Ministry of Health. Boletim Epi- 46.Gruskin S, Tarantola D. Health and at: http://www.aids.gov.br/imprensa/ tical industry and the WTO. J World demiológico AIDS. Brasília, Brazil: Min- Human Rights. Boston, Mass: François- noticiasimpressao.asp?notcod=53234. Intellect Prop. 2002;5(1):53–104. istry of Health; 2004. Available at: Xavier Bagnoud Center for Health and Accessed April 4, 2005. 14 .Vianna JMN. Intellectual property http://www.aids.gov.br/final/biblioteca/ Human Rights; 2000. Working Paper rights, the World Trade Organization boletim_dezembro_2003/bol_dezem- 34. Kaisernetwork.org. Brazil’s Na- 10. Available at: http://www.hsph.har- and public health: the Brazilian perspec- bro_2003.pdf. Accessed April 4, 2005. tional STD/AIDS Programme Announces vard.edu/fxbcenter/FXBC_WP10— tive. Conn J Int Law.2002;17:311–318. 25. Ministry of Health. Response: The Ex- Largest Drug Price Reduction Deals in Gruskin_and_Tarantola.pdf. Accessed perience of the Brazilian AIDS Programme. Five Years. Menlo Park, Calif: Henry J. April 4, 2005. 15. McNeil D Jr. Plan to battle AIDS Kaiser Family Foundation; January Brasília, Brazil: Ministry of Health; 2002. 47. Mukherjee J. Basing treatment on worldwide is falling short. New York 20, 2004. Available at: http://www. Available at: http://www.aids.gov.br/ rights rather than ability to pay: 3 by 5. Times. March 28, 2004:A1;A14. kaisernetwork.org/daily_reports/print_ final/biblioteca/resposta/resp_ingles.pdf. Lancet. 2004;363:1071–1072. 16. Attawell K, Mundy J. Provision of Accessed April 4, 2005. report.cfm?DR_ID=21751&dr_cat=1. Antiretroviral Therapy in Resource-Limited Accessed April 4, 2005. 48.Tarantola D. Building on the Syn- Settings: A Review of Experience Up to 26. Ministério da Saúde lança cam- ergy Between Health and Human Rights: 35. Farmer P. AIDS and Accusation: August 2003. London, England: Health panha com o objetivo de incentivar os A Global Perspective. Boston, Mass: Haiti and the Geography of Blame. Systems Resource Centre, Department testes de HIV. Folha de S. Paulo. January François-Xavier Bagnoud Center for Berkeley: University of California Press; for International Development; 2003. 18, 2003:C3. Health and Human Rights; 2000. 19 92. Available at: http://www.who.int/3by5/ 27. Kaisernetwork.org. Brazil Launches Working Paper 8. Available at: http:// publications/documents/en/ARTpaper_ Free HIV Testing Campaign. Menlo Park, 36.International Council of AIDS Ser- www.hsph.harvard.edu/fxbcenter/FXBC_ DFID_WHO.pdf. Accessed April 4, Calif: Henry J. Kaiser Family Founda- vice Organizations. The International WP8—Tarantola.pdf. Accessed April 4, 2005. tion; November 3, 2003. Available at: Guidelines on HIV/AIDS and Human 2005. Rights: How Are They Being Used and 17. International HIV Treatment Ac- http://www.kaisernetwork.org/daily_ Applied? To ronto, Canada: International 49. United Nations High Commis- cess Coalition. A Commitment to Action reports/rep_index.cfm?hint=1&DR_ Council of AIDS Service Organizations; sioner for Human Rights. Access to for Expanded Access to HIV/AIDS Treat- ID=20647. Accessed April 4, 2005. 2002. Medication in the Context of Pandemics ment. Geneva, Switzerland: World Health 28.Programa Nacional de DST e Such as HIV/AIDS. Commission on Organization; 2002. Available at: http:// AIDS. O perfil da AIDS no Brasil e 37.Parker R, Aggleton P. HIV and Human Rights Resolution 2001/33. www.who.int/hiv/pub/arv/who_hiv_ metas de governo para o controle da epi- AIDS-related stigma and discrimination: Geneva, Switzerland: United Nations 2002_24.pdf. Accessed April 4, 2005. demia. Brasília, Brazil: National AIDS a conceptual framework and implica- High Commissioner for Human Rights; tions for action. Soc Sci Med. 2003;57: 18.World Health Organization, Joint Program; 2003. Available at: http:// 2001. Available at: http://www.unhchr. 13–24. United Nations Programme on HIV/ www.aids.gov.br/final/biblioteca/metas/ ch/huridocda/huridoca.nsf/framepage/ AIDS. Treating 3 Million by 2005: Mak- metas.pdf. Accessed April 4, 2005. 38. Sabatier R. Blaming Others: Preju- docs+chr?opendocument. Accessed ing It Happen. The WHO Strategy. Ge- 29. Chequer P, Cuchí P, Mazin R, dice, Race and Worldwide AIDS. Lon- April 4, 2005. neva, Switzerland: World Health Orga- Calleja JMG. Access to antiretroviral don, England: The Panos Institute; 50. World Trade Organization. Decla- nization; 2003. Available at: http://www. treatment in Latin American countries 1988. ration on the TRIPS Agreement and Pub- who.int/3by5/publications/documents/ and the Caribbean. AIDS.2002;6(suppl 39. Mann JM. AIDS and human rights: lic Health. Geneva, Switzerland: World en/3by5StrategyMakingItHappen.pdf. 13):S50–S57. where do we go from? Health Hum Trade Organization; 2001. WT/ Accessed April 4, 2005. 30. Antirretrovirales: esfuerzos de Rights.1998;3:143–149. MIN(01)/DEC/2. Available at: http:// www.wto.org/english/thewto_e/minist_ 19.World Bank. World Development América Latina y el Caribe hacia el ac- 40. Mann J, Tarantola D. Responding e/min01_e/mindecl_trips_e.pdf. Ac- Indicators 2003. Washington, DC: World ceso universal [information about access to HIV/AIDS: a historical perspective. cessed April 4, 2005. Bank; 2003. to ARV treatment in the Latin Ameri- Health Hum Rights.1998;2:5–8. 20.Teixeira PR, Vitória MAA, Bar- can region]. Pan American Health Orga- 41. Mann J, Gruskin S, Grodin MA, 51. Correa CM. Implications of the carolo J. The Brazilian experience in nization Web site. Available at: http:// Annas GJ. Health and Human Rights: A Doha Declaration on the TRIPS Agree- providing universal access to antiretrovi- www.paho.org/Spanish/AD/FCH/AI/ Reader. London, England: Routledge; ment and Public Health. Geneva, Switzer- ral therapy. In: Moatti J-P, Coriat B, antiretrovirals_HP.htm. Accessed April 1999. land: World Health Organization; 2002. Souteyrand Y, Barnett T, Dumoulin J, 4, 2005. Available at: http://whqlibdoc.who.int/ 42. Gruskin S, Hendriks A, Tomasevski K. Flori Y-A, eds. Economics of AIDS and 31. Decreto Número 4.830. Diário hq/2002/who_edm_par_2002.3.pdf. Human rights and responses to AIDS. Access to HIV/AIDS Care in Developing Oficial da União. September 5, 2003; Accessed April 4, 2005. Countries, Issues and Challenges. Paris, In: Mann J, Tarantola D, eds. AIDS in 14 0(172):1. Available at: http://www. 52. Joint United Nations Programme France: Agence Nationale de the World II. New York, NY: Oxford aids.gov.br/imprensa/pdf/DECRETO% on HIV/AIDS. Public Report: First Meet- Recherches sur le Sida; 2003:69–88. University Press; 1996: 326–340. 20N%204.830%20DE%2005.09.200 ing of the UNAIDS Global Reference 21. Levi GC, Vitória MA. Fighting 3.pdf. Accessed April 4, 2005. 43. Coutinho A. Pills and promises—a Group on HIV/AIDS and Human Rights, personal account from Uganda. Lancet. against AIDS: the Brazilian experience. 32. Governo cria instrumento que fa- January 23–24, 2003, Geneva, Switzer- 2004;363:1073. AIDS.2002;16:2373–2383. cilita importação de medicamentos land. Geneva, Switzerland: Joint United 22. Oliveira-Cruz V, Kowalski J, genéricos [press release]. Brasília, Brazil: 44. Easley CE, Marks SP, Morgan RE Jr. Nations Programme on HIV/AIDS; McPake B. Viewpoint: the Brazilian National AIDS Program; September 4, The challenge and place of interna- 2003. Available at: http://www.unaids. HIV/AIDS ‘success story’—can others 2003. Available at: http://www.aids. tional human rights in public health. org/html/pub/UNA-docs/HRrefgroup_ do it? Trop Med Int Health.2004;9: gov.br/imprensa/noticiasimpressao.asp? Am J Public Health. 2001;91: 0103mtg_en_pdf.htm. Accessed April 4, 292–297. notcod=49511. Accessed April 4, 2005. 1922–1925. 2005. 23. Médecins Sans Frontières (MSF). 33. Saúde fecha acordo com labo- 45. Farmer P. Political violence and 53. Skidmore T. The Politics of Military

July 2005, Vol 95, No. 7 | American Journal of Public Health Galvão | Peer Reviewed | Health Policy and Ethics | 1115  HEALTH POLICY AND ETHICS 

Rule in Brazil, 1964–85. New York, NY: NACLA Report on the Americas.2002; to AIDS in Brazil. Divulg Saúde Debate. 86. Elbe S. Strategic Implications of Oxford University Press; 1988. 35(5):35–37,40–42. 2003;27:215–227. HIV/AIDS. New York, NY: Oxford Uni- versity Press; 2003. 54. Mattos RA. Os sentidos da integral- 64. Daniel H. Vida Antes da Morte [Life 77. World Bank. Project Information idade: algumas reflexões acerca dos val- Before Death]. Rio de Janeiro, Brazil: Ti- Document (PID), Appraisal Stage: Re- 87.Price-Smith AT. The Health of Na- ores que merecem ser defendidos. In: pografia Jaboti; 1989. gional HIV/AIDS Treatment Acceleration. tions: Infectious Disease, Environmental Pinheiro R, Mattos RA, eds. Os Sentidos 65. Farmer P. Pathologies of Power: Washington, DC: World Bank; 2004. Change, and Their Effects on National da Integralidade na Atenção e no Cuidado Health, Human Rights and the New War Report AB892. Available at: http:// Security and Development. Cambridge, à Saúde. Rio de Janeiro, Brazil: IMS/ on the Poor. Berkeley: University of Cali- www-wds.worldbank.org/servlet/ Mass: MIT Press; 2002. UERJ/ABRASCO; 2001: 39–64. fornia Press; 2003. WDSContentServer/WDSP/IB/2004/ 88. Galvão J. More money for AIDS. 05/24/000104615_20040525112142/ 55. Galvão J. Access to antiretrovirals: Lancet. 2003;361:1569–1570. 66.Parker R. The global HIV/AIDS Rendered/PDF/PID0TAP.pdf. Accessed where South Africa, China, and Brazil pandemic, structural inequalities, and April 4, 2005. 89. International Crisis Group. HIV/ meet. Lancet.2004;363:493. the politics of international health. Am AIDS as a Security Issue. Brussels, Bel- 78.Piot P, Feachem RG, Lee JW, 56. Elias PE, Cohn A. Health reform in J Public Health.2002;92:343–346. gium: International Crisis Group; June 19, Wolfensohn JD. Public health. A global Brazil: lessons to consider. Am J Public 2001. Available at: http://www.crisisweb. 67. Mukherjee JS, Farmer PE, response to AIDS: lessons learned, next Health. 2003;93:44–48. org//library/documents/report_archive/ Niyizonkiza D, et al. Tackling HIV in re- steps. Science. 2004;304:1909–1910. source poor countries. BMJ. 2003;327: A400321_19062001.pdf. Accessed 57. Souza RR. O Sistema Público de 110 4–1106. 79.Parker R. Building the foundations April 4, 2005. Saúde Brasileiro. Brasília, Brazil: Min- for the response to HIV/AIDS in Brazil: 68. Grubb I, Perriëns J, Schwartländer B. 90.World Health Organization. The istry of Health; 2002. Available at: the development of HIV/AIDS policy, A Public Health Approach to Antiretrovi- World Health Report 2004: Changing http://dtr2001.saude.gov.br/editora/ 19 82–1996. Divulg Saúde Debate. ral Treatment: Overcoming Constraints. History. Geneva, Switzerland: World produtos/livros/pdf/02_0784_M.pdf. 2003;27:143—183. Accessed April 4, 2005. Geneva, Switzerland: World Health Or- Health Organization; 2004. ganization; 2003. Available at: http:// 80.Frei Betto condena AIDS ter mais 91. Global HIV Prevention Working 58. Orsi F, Hasenclever L, Fialho B, www.who.int/hiv/pub/prev_care/en/ verba que fome. O Globo. October 14, Group. HIV Prevention in the Era of Ex- Tigre P, Coriat B. Intellectual property PublicHealthApproach_E.pdf. Accessed 2003:11. panded Treatment Access. Global HIV rights, anti-AIDS policy and generic April 4, 2005. Prevention Working Group; 2004. Avail- drugs: lessons from the Brazilian public 81. Instituto Brasileiro de Geografia e able at: http://www.gatesfoundation.org/ health program. In: Moatti J-P, Coriat 69. Laurent C, Diakhaté N, Gueye NF, Estatística. Síntese de indicadores sociais nr/downloads/globalhealth/aids/PWG B, Souteyrand Y, Barnett T, Dumoulin et al. The Senegalese government’s 2003. Rio de Janeiro, Brazil: Instituto 2004Report.pdf. Accessed April 4, 2005. J, Flori Y-A, eds. Economics of AIDS highly active antiretroviral therapy ini- Brasileiro de Geografia e Estatística; and Access to HIV/AIDS Care in Devel- tiative: an 18-month follow-up study. April 13, 2004. Available at: http://www. 92. Brazil and UNAIDS join forces in oping Countries, Issues and Challenges. AIDS.2002;16:1363–1370. ibge.gov.br/home/presidencia/noticias/ the global fight against AIDS [press re- pdf/13042004sintese2003.pdf. Ac- Paris, France: Agence Nationale de 70. Mukherjee JS. HIV-1 care in lease]. Geneva, Switzerland: Joint United cessed April 4, 2005. Recherches sur le Sida; 2003:109–135. resource-poor settings: a view from Nations Programme on HIV/AIDS– Brazil; September 1, 2004. Available at: 59. Ministry of Health. Terapia Anti- Haiti. Lancet.2003;362:994–995. 82. Decosas J. HIV prevention and treatment in South Africa: affordable http://www.unaids.org/NetTools/Misc/ retroviral e Saúde Pública: Um Balanço 71. Orrell C, Bangsberg DR, Badri M, and desirable [published correction ap- DocInfo.aspx?href=http://gva-doc-owl/ da Experiência Brasileira. Brasília, Brazil: Wood R. Adherence is not a barrier to pears in Lancet. 2003;362:84]. Lancet. WEBcontent/Documents/pub/Media/ Ministry of Health; 1999. successful antiretroviral therapy in South 2003;361:1146–1147. Press-Releases02/PR_Brazil_01Sep04_ Africa. AIDS.2003;17:1369–1375. 60.Dispõe sobre a distribuição gra- en.pdf. Accessed April 4, 2005. 83. United Nations Development Pro- tuita de medicamentos aos portadores 72. World Health Organization. Emer- gramme. Thailand’s Response to HIV/ 93. Beaglehole R, Bonita R, Horton R, do HIV e doentes de AIDS. November gency Scale-Up of Antiretroviral Therapy AIDS: Progress and Challenges. Bangkok, Adams O, McKee M. Public health in 13, 1996. Law 9.313. Available at: in Resource-Limited Settings: Technical Thailand: United Nations Development the new era: improving health through http://www.aids.gov.br/assistencia/ and Operational Recommendations to Programme; 2004. Available at: http:// collective action. Lancet. 2004;363: lei9313.htm. Accessed April 4, 2005. Achieve 3 by 5. Geneva, Switzerland: 2084–2086. World Health Organization; 2003. www.undp.or.th/documents/HIV_AIDS_ 61. Negri B. Política Federal de As- Available at: http://www.who.int/3by5/ FullReport_ENG.pdf. Accessed April 4, sistência Farmacêutica, 1990–2002. publications/documents/en/zambia_ 2005. Brasília, Brazil: Ministry of Health; doc_final.pdf. Accessed April 4, 2005. 84. United Nations Population Fund. 2002. Available at: http://dtr2001. Investing in People: National Progress in saude.gov.br/editora/produtos/livros/ 73. XV International AIDS Conference Implementing the ICPD Programme of pdf/03_0049_M.pdf. Accessed April 4, Bangkok Web site. Available at: http:// Action 1994–2004. New York, NY: 2005. www.aids2004.org. Accessed April 4, 2005. United Nations Population Fund; 2004. 62. Bastos FI, Kerrigan D, Malta M, Available at: http://www.unfpa.org/ Carneiro-da-Cunha C, Strathdee SA. 74. Marseille E, Hofmann PB, Kahn JG. upload/lib_pub_file/284_filename_ Treatment for HIV/AIDS in Brazil: HIV prevention before HAART in sub- globalsurvey.pdf. Accessed July 9, 2004. strengths, challenges, and opportunities Saharan Africa. Lancet. 2002;359: 85. Human Rights Watch. Not Enough for operations research. AIDScience. 1851–1856. Graves: The War on Drugs, HIV/AIDS, 2001;1(15). Available at: http://www. 75. World Bank. Confronting AIDS: Pub- and Violations of Human Rights. New aidscience.org/Articles/aidscience012. lic Priorities in a Global Epidemic. New York, NY: Human Rights Watch; 2004. pdf. Accessed April 4, 2005. York, NY: Oxford University Press; 1997. Available at: http://hrw.org/reports/ 63. Passarelli C, Terto V Jr. Good medi- 76. Mattos RA, Terto V Jr, Parker R. 2004/thailand0704/thailand0704.pdf. cine: Brazil’s multifront war on AIDS. World Bank strategies and the response Accessed April 4, 2005.

1116 | Health Policy and Ethics | Peer Reviewed | Galvão American Journal of Public Health | July 2005, Vol 95, No. 7  HEALTH POLICY AND ETHICS 

Antiretroviral Therapy in Resource-Poor Countries: Illusions and Realities

| Moïse Desvarieux, MD, PhD, Roland Landman, MD, Bernard Liautaud, MD, Pierre-Marie Girard, MD, PhD, for the INTREPIDE Initiative in Global Health

The prospects for antiretro- emergence of generic drugs as justice, and curbing of the HIV eral law mandating free drugs viral therapy in resource-poor well as the simplification of treat- epidemic) as much as for the im- through the public health sys- settings have changed recently ment has made care possible in mediate public health goals of re- tem, recorded that US $954 mil- and considerably with the these countries.1 duced morbidity and mortality. lion was spent on providing free availability of generic drugs, However, often overlooked is As a consequence, these larger ARV drugs from 1997 to 2001 the drastic price reduction of the fact that such reductions in population benefits are some- with an estimated savings of brand-name drugs, and the costs, if they allow the implemen- times seen as primary, with the overUS$1 billion, mostly in hos- simplification of treatment. 6 However, such cost reductions, tation of large-scale donor pro- public health objectives consid- pitalization costs. Senegal’s gov- although allowing the imple- grams, have yet to render treat- ered a vehicle toward accom- ernment recently announced mentation of large-scale donor ment economically accessible to plishing these larger goals. But that it will cover costs for all pa- programs, have yet to render or possible for the general popu- are those goals always in har- tients, (initially drug costs, but treatment accessible and pos- lation. Indeed, even with these mony? Or might some of those now expanding to hospitalization sible in the general population. substantial cost reductions, like goals be better attained in other costs), the first African country Successfully providing HIV those negotiated via the United ways? If so, which objectives are to do so. Although the relatively treatment in high-prevalence/ Nations Global Fund,2 the US the most important? modest seroprevalence in Sene- high-caseload countries may president’s initiative,3 and the We review the population gal no doubt renders such an ap- require that we redefine the Clinton Foundation, or even the goals implicit in ARV treatment proach possible, it is not clear problem as a public health advent of generics, treatment re- programs, assess their feasibility, that it will have immediate eco- mass therapy program rather than a multiplication of clini- mains beyond the reach of all and contrast them with the vehi- nomic benefit. Direct savings will cal situations. The public health but the upper classes in numer- cle that is supposed to bring be attained only if AIDS patients 4 goal cannot simply be the re- ous countries. It is indeed the them to fruition—access to ARV would otherwise be hospitalized duction of morbidity and mor- paradox of lower ARV therapy care for the many—before pro- for care, a proposition often ab- tality for those treated but costs that these reductions posing a paradigm shift anchored sent in Africa, where many die must be the reduction in mor- brought with them a cortege of in today’s reality. from their initial opportunistic bidity and mortality for the pressures that must be recog- infections.7 Alternatively, South many, that is, at a population nized with wide-open eyes. In POPULATION OBJECTIVES Africa has recently chosen an level. (Am J Public Health. order for lower ARV therapy OF ARV DRUG THERAPY approach that most directly leads 2005;95:1117–1122. doi:10.2105/ costs to truly usher in the era of PROGRAMS to an economic impact by engag- AJPH.2003.034249) “global treatment” beyond pilot ing large companies and the mil- or research programs, a realistic Objective 1: Maintaining itary in the HIV-related care of THE PROSPECTS FOR discussion of attainable goals of Economic Stability their employees, thus preserving antiretroviral (ARV) therapy in ARV treatment in resource-poor It is often assumed that treat- the workforce and the security Africa and other resource-poor countries is necessary.5 ing HIV patients will automati- apparatus. settings have changed so drasti- The debate is peculiar in that cally result in economic benefits. Of course these observations cally over the last few years that the necessity of making ARV That is not necessarily the case. ignore the intangible economic it is almost embarrassing to real- therapy available in resource- The interplay between effective- impact of the patients’ lost contri- ize that it could have changed poor countries has been justified ness, benefit, and cost varies butions to society and those of much earlier. Prices of drugs for its population benefits across settings. For example, the family members drawn to their from pharmaceutical giants have (namely, maintenance of eco- Brazilian National AIDS Control care. Nevertheless, immediate fallen 10- to 40-fold, and the nomic capacity, distributive Program, emboldened by a fed- economic gains would be better

July 2005, Vol 95, No. 7 | American Journal of Public Health Desvarieux et al. | Peer Reviewed | Health Policy and Ethics | 1117  HEALTH POLICY AND ETHICS 

guaranteed in settings where pa- and this is often overlooked— tries where survival is limited Objective 4: Reducing tients would have received a cer- distributive justice encompasses once AIDS is diagnosed. Curbing Morbidity and Mortality tain level of care leading to pro- the reciprocal obligation be- the epidemic would thus entail At an individual level, the ob- longed life and its attendant tween neighboring countries to treatment of most HIV patients, jectives of reduced morbidity and hospitalization costs. Thus, at a ensure that drug resistance including the asymptomatic ones mortality are naturally the ones population level, economic stabil- emergence is minimized. This with a high viral load, thereby directly sought when ARV treat- ity is likely to be a benefit of un- illustrates the difficulty of having shifting the incidence and sero- ment is initiated. These individ- targeted treatment only if the neighboring countries with and prevalence curve to lesser ual objectives entail an eminently HIV prevalence cuts across social without access to ARV therapy values.5 This naturally presup- clinicobiological approach to care groups to such an extent that a and the disjunction in responsi- poses effective large-scale screen- that originated in resource-rich substantial portion of the produc- bility that ensues. This is also ing and individualized CD4 countries, with the consecrated tive workforce is affected and if true for prevention efforts; for count and viral load assessments mainstays of treatment initiation: the population reached by treat- example, an excellent Population to select those asymptomatic CD4 counts and the viral load. ment is large enough to encom- Services International social people eligible for treatment. It has now become clear that pass a significant portion of the marketing program promoting This approach would be most viral load assessment is not nec- untargeted workforce. Conversely, Kapot Pantè (a condom brand) in similar to that used in resource- essary to initiate treatment,8,9 as a siphoning of resources from Haiti but not in the Dominican rich countries where all at-risk reflected in World Health Orga- other health priorities may in- Republic led to substantial trans- individuals are encouraged to be nization guidelines.9 However, crease economic instability: for border transportation of contested. Even in these countries, the CD4 count guidelines of example, if malaria mortality or doms, draining condoms away however, this strategy is only resource-rich countries remain a morbidity were to increase as an from the intended Haitian popu- marginally successful in identify- prominent goal.10 In an attempt unintended result of increased at- lation, and Population Services ing eligible patients. Thus, al- to replicate resource-rich coun- tention to HIV/AIDS, the impact International recently had to though laudable, this objective is tries’ treatment guidelines, efforts across social classes might be launch a new distribution pro- realistically unattainable in most have been directed toward greater. In the end, the long-term gram in the Dominican Republic. resource-poor countries for the greater availability or affordabil- economic (and social) impact of This could again happen with foreseeable future. ity of CD4 counts, cheaper ways raising generations of orphans ARV therapy, because drugs are Therefore, to summarize the to determine viral loads, or sub- may be the most staggering, al- presently more widely accessible 3 objectives mentioned previ- stitutes for CD4 counts.11,12 beit delayed, draw on resources. in Haiti than in the Dominican ously, the population goal of eco- At a population level, how- Nevertheless, the strict goals of Republic. Thus, the availability nomic stability might be more ever, the goal of reduced morbid- economic benefits might be more of ARV therapy might paradoxi- immediately achieved through ity and mortality can be achieved surely attained with targeted cally further the injustices within targeted treatment of the privi- only if a large number of patients treatment of the critical work- and across resource-poor leged or productive workforce. receive care. In the collective at- force rather than the general countries. The quest for distributive justice tempt to do so, HIV health pro- population; this possibly conflicts might paradoxically lead to a fur- viders have struggled to trans- with the larger public health Objective 3: Curbing the HIV thering of the gap within and pose an individualized, highly goals. Epidemic across resource-poor countries, biological approach to care onto Treatment of severely sympto- and epidemic containment is a massive public health problem. Objective 2: Achieving matic patients, as currently rec- unattainable in most settings. We have ourselves experimented Distributive Justice ommended in resource-poor Thus, these population objectives with these approaches in coun- The issue of distributive jus- countries, is unlikely to affect the should not be seen as primary, tries with varied seroprevalence tice between rich and poor coun- epidemic transmission signifi- because they may indeed conflict rates, for example, in Senegal, tries often obscures the fact that cantly, because most transmis- with the immediate public health Côte d’Ivoire, or Haiti. Indeed, it distributive injustices within a sion occurs via patients with high goal of reduced morbidity and is the difficulties we have en- country may be compounded by viremia who are well enough to mortality for the many; this goal countered in these various sero- the availability of treatments to engage in sexual activities.1 The must stand on its own as a pri- prevalence settings that have led only the privileged segments of latter generally represent the mary objective rather than as a us to realize that current ap- resource-poor countries. Also— larger pool, especially in coun- vehicle to other goals. proaches, based on model trans-

1118 | Health Policy and Ethics | Peer Reviewed | Desvarieux et al. American Journal of Public Health | July 2005, Vol 95, No. 7  HEALTH POLICY AND ETHICS 

position, neglect one singularly ments for CD4 counts,15 ,16 repli- only clinical evaluation.22 Be- population as the unit of analysis. important factor: the actual pa- cating the individual monitoring tween these 2 poles is, of Such a programmatic approach tient load. For example, HIV of resource-rich countries recourse, a middle ground. We implies a small number of spe- seroprevalence rates in the mains illusory. It is not simply submit for discussion a shift cific requirements. United States and in France are a matter of cost. from a clinical to a public health estimated to be around 0.25%, Therefore, addressing the approach to HIV treatment pri- Immediate Planning of leading to a national caseload of global problem of HIV treatment oritizing large-scale programs First- (ARV) and Second-line 800000 (Centers for Disease in high-prevalence/high-caseload rigorously evaluated for their (ARV-plus) Drug Supply Control and Prevention) to (HPHC) countries may require population impact, rather than Unlike tuberculosis therapy, in 95000013 in the United States that health decisionmakers first on a patient-by-patient basis. which programs have been built and 150000 in France.13 As a specifically recognize that the Such an approach presumes that largely on first-line treatments, comparison, an estimated public health goal cannot simply we should not await data estab- with second-line treatments 250000 to 400000 people live be the reduction of morbidity lishing the relative contribution (directly observed therapy with HIV/AIDS in Haiti, a larger and mortality for those treated of CD4 counts or viral loads be- [DOT]-plus) recommended only patient load than in France, but must be the reduction in fore initiating treatment for HIV- in settings with established DOT where the population is nearly 8 morbidity and mortality for the positive patients in resource- programs,23 ARV-plus programs times larger. India, with the many, that is, at a population poor settings. Rather, treatment would need to be planned con- largest number of people living level. Once that goal is clearly would be initiated quickly, fo- currently because, at a population with HIV outside South Africa, stated, the HIV seroprevalence cused on HIV-seropositive symp- level, HIV develops resistance to has 5.1 million seropositive peo- or caseload constitutes the major tomatic patients22 in accordance ARV drugs faster than tuberculo- ple,13 and Nigeria, with a rela- operational factor, necessitating with the goal of a population- sis bacilli do to antibiotics. tively low HIV seroprevalence that we redefine the problem in wide reduction in morbidity and of 5.8%, still yields a caseload the most affected regions as a mortality. Similarly, treatment Program Evaluation in of 7 million, nearly 10 times the public health mass therapy pro- would be monitored on the Concordance with the number of seropositive people in gram rather than simply a multi- basis of improvement in clinical Principles of Mass Therapy the entire United States.13 plication of clinical situations. symptoms and possibly limited We should abandon individ- A 5% to 10% HIV seropreva- Therefore there is a need for a laboratory tests (cell blood ual efficacy in favor of popula- lence in the United States or paradigm shift in delivering, count, liver function tests, and tion efficiency as tenets of pro- France would translate to a 20- monitoring, and assessing success creatinine). The urgency of initi- gram success. We propose the to 40-fold increase in current of ARV therapy programs in ating treatment is reinforced by following criteria for annual or caseloads. Because of this reality, HPHC settings. reports showing lower survival semiannual population-based it is extremely doubtful that cur- rates with delayed treatment ini- evaluation of program success: rent extensive biological monitor- PROPOSAL FOR A NEW tiation among symptomatic pa- (1) death rate (overall and HIV- ing approaches would be either PROGRAM-BASED tients in Africa.21 related); (2) incidence of major used or simply feasible, even STRATEGY Initially, from a pragmatic opportunistic infections, for ex- within the United States or point of view, programs need to ample, tuberculosis; (3) magni- France, given the multifold in- Not so long ago, it was widely be implemented around existing tude of CD4 changes among a crease in labor, personnel, and perceived that ARV treatment centers (generally, but not al- sample of treated individuals; infrastructure that such ap- was impossible in resource-poor ways, urban centers because (4) proportion of patients with proaches require. As an example, countries. After publication of seroprevalence is generally undetectable viral loads among Cohen et al.14 projected that 5.1 our 1997 consensus guidelines higher in urban centers, popula- a sample of treated individuals; million additional patient visits on ARV therapy in Africa17 and tions are more accessible and (5) rates of drug resistance per year would be required to the first studies of efficacy and have more access to care, and among a sample of treated indi- provide routine HIV care in acceptability,18–21 the pendulum tailored training is easier) and viduals; and (6) rates of drug re- western Kenya alone with the swung sharply in the opposite radiate outward. In this program- sistance among a sample of un- current treatment paradigms. direction, with some people suc- matic approach, extensive evalu- treated individuals to estimate Thus, in spite of improvements cessfully introducing ARV drugs ation would move away from the the diffusion in the population. in cost and technical require- in resource-poor countries with individual but would utilize the Sequential data on CD4 counts,

July 2005, Vol 95, No. 7 | American Journal of Public Health Desvarieux et al. | Peer Reviewed | Health Policy and Ethics | 1119  HEALTH POLICY AND ETHICS 

viral loads, and drug resistance, centers in urban sites to ill- for individual monitoring of toxic- fine and integrate their plans for collected on a representative equipped suburban or rural ity that will impact overall public treatment adherence monitoring sample of the population, would areas is an undeniable challenge health. HIV programs should at the time of initiation of ther- thus assess treatment efficiency in HIV care. All personnel therefore be linked to primary apy, and funding agencies should at the population level and in- should be trained in recognizing care,25,27,28 including existing insist on those plans and link form changes in recommended adverse outcomes and among mother–infant programs as well continued funding to their effec- regional or national treatment them those requiring treatment as tuberculosis programs. More tive implementation. Black mar- guidelines. This is somewhat interruption. Clear referral specialized treatment can be only kets for ARV therapies as well as similar to tuberculosis and ma- guidelines should be created at the tertiary care level with counterfeit drugs are major con- laria therapy, in which treatment and taught. training at these levels done cerns in countries with poor ca- guidelines are informed by prev- We must state candidly that in accordingly. The demise of ex- pacity for pharmacological con- alent drug resistance rates in advocating this approach, we rec- ceptionalism can be a good trol, and specific policies must be several countries that do not ac- ognize that treatment may not be thing.26–28 Naturally, improve- in place. From a programmatic tually have the effective capacity as effective for every patient as ments in laboratory techniques standpoint, investing in such ef- for individual drug susceptibility that provided by the biological and logistics (including personnel forts seems more efficient than testing. approach. However, this public training) should be monitored for investing in large-scale individual The annual rates of HIV infec- health approach recognizes real- adaptation of the population- immunologic monitoring. tion among women with first ity. Indeed, we wonder whether based/individual-based ratio of Finally, public health princi- pregnancies also may be incorpo- the difficulties in adapting to the laboratory tests. ples require donor agencies to rated to monitor the continuation reality dictated by high caseloads ensure that drugs are not simply of prevention in a comprehensive are a remnant of the exceptional- A Plan for Adherence delivered to countries and al- program.24 Well-integrated pre- ism that has historically charac- Monitoring lowed to disappear into the local vention programs may benefit terized policy in resource-rich Preliminary data17–19,29 show distribution system. Donor agen- from the availability of ARV ther- countries, wherein HIV was ex- adherence to be high in several cies that engage solely in ARV apy. Like the “combination pre- empted from traditional public African settings. Careful monitor- distribution must guard against vention” advocated by the Gates health practices such as contact ing of clinic visits to replenish the illusion of a policy of treat- Foundation,25 “combination out- tracing and partner notification.26 drug supplies, questioning by ment. Without programs, a nar- comes” should primarily measure To be frank, in the context of re- friendly staff, and validated clini- row policy of drug distribution program success. Utilizing—and source-poor countries, this reluc- cal signs and symptoms might be may simply lead to a policy of adapting—these proposed criteria tance to implement public heath acceptable surrogate markers of drug disappearance masquerad- across countries and regions measures is compounded by the adherence. Regions should be ing as a policy of treatment. Now should allow more direct com- fear of being accused of advocat- able to reasonably tailor adher- that drugs are being made avail- parisons and improve experi- ing a 2-tiered system, lesser for ence monitoring to their reality able to countries, both recipients ence sharing. the poor and disenfranchised. with supervision from funding and donors have new responsi- This fear clouds the reality that agencies and health authorities. bilities. Some would argue that a Simple Schemes even the scaling up of facilities Is universal DOT necessary? In program of drug delivery is bet- for Community-Level and staff will not change the im- spite of the attractiveness of ter than no program at all. They HIV Treatment possibility of meeting the chal- DOT for ARV therapies, it seems might further argue that our po- In a second phase, treatment lenges with the traditional ap- doubtful that in HPHC settings, sition of requiring an effective should move from hospital- and proach. strict DOT will be practical or program of drug treatment is clinic-based to community-based However, we do advocate scal- possible.30 It would entail the equal to advocating that drugs programs with staff at each level ing up. First, 1 central national lifelong mobilization of huge not be delivered at all. This trained for appropriate referral or regional reference laboratory numbers of “accompagnateurs” point of view clouds the debate. dictated by clinical worsening or must be improved for population- if one were to replicate the suc- From a population perspective, if side effects. This approach based evaluation of program successful program of Farmer et it is morally untenable that drugs would differ from that of re- cess (CD4 levels, viral loads, al.22 in central Haiti or those of be withheld from those regions source-rich countries. Moving drug-resistance). Second, general Médecins sans Frontières.31 How- most in need to prevent resist- from well-equipped health care laboratories must be improved ever, programs should clearly de- ance in those regions least in

1120 | Health Policy and Ethics | Peer Reviewed | Desvarieux et al. American Journal of Public Health | July 2005, Vol 95, No. 7  HEALTH POLICY AND ETHICS 

need, it is equally ethically im- individuals and groups like treat HIV as the public health Infectieuses et Tropicales, Centre Hospitalier permissible to abdicate the dual Médecins Sans Frontières, the crisis that it is, with massive pro- Universitaire de Saint-Antoine, Université de Paris, Paris, and the Institut de Médecine et responsibility of ensuring that Clinton Foundation, or the grams evaluated at the popula- d’Épidémiologie Appliquée, Paris. the drugs reach the intended pa- United Nations, as well as the tion level: only strict program Requests for reprints should be sent to tients and of protecting the effi- availability of generic drugs, evaluation can ensure that a pol- Moïse Desvarieux, MD, PhD, Department of Epidemiology, Mailman School of Public cacy of these drugs. An example clearly has to be credited for per- icy of drug disappearance does Health, 722 W 168th St, New York, NY of the potentially deleterious ef- suading the pharmaceutical com- not masquerade as a policy of 10032 (e-mail: [email protected]). fects of unbridled drug delivery panies to lower their prices. treatment; (3) on an individual This article was accepted December 14, 2004. without this assurance is pro- Delaporte’s recent evaluation of level, focus on adherence moni- 34 vided by nevirapine, distributed a fixed-drug combination pill, toring, clinical evaluation, and Contributors largely to pregnant women in regrouping generic ARV drugs toxicity evaluation rather than on M. Desvarieux and P.-M. Girard wrote certain areas; its use has even from different manufacturers, immunologic evaluation; (4) make the original article with substantial been advocated for all women in may be seen as a telling illustra- strategies for preserving the fu- input and editing from R. Landman and B. Liautaud. highly HIV endemic countries.32 tion of the cooperation that is ture a common ethical issue for Even the brief use of nevirapine both possible and needed. How- all neighboring countries, as well Acknowledgments in mother-to-child transmission ever, contrary to popular belief, as donor programs. Financial support for INTREPIDE was prevention leads to increased re- generic drugs are not always The urgency of the need to provided in part by the Institut de sistance to the entire class of cheaper than patented drugs.35 rush HIV treatment to resource- Médecine et d’Epidémiologie Appliquée Paris, France, and the University of Min- drugs, possibly erasing the effi- Therefore, the debate should poor countries underscores a nesota School of Public Health and Divi- cacy of 1 of the 3 major drug not be cast in terms of good ver- true public health emergency, sion of Epidemiology, Minneapolis, Minn. classes available for HIV treat- sus evil, in spite of the obvious and the achievement of public Members of the International Train- 33 ing and Research Program in Infectious ment. Therefore, preserving temptation to do so. health goals requires access to Disease Epidemiology (INTREPIDE) the future for both local popula- the many. However, the under- include Olivier Bouchaud, MD, Anke tions and neighboring countries CONCLUSIONS standable fear of rushing in with Bourgeois, MD, Marc Brodin, MD, Jean-Pierre Coulaud, MD, Eric Dela- should be a primary objective. eyes shut may now lead to the porte, MD, Philippe Deloron, PhD, The argument often advanced pursuit of laudable yet dispropor- Moïse Desvarieux, MD, PhD, Arnaud ARV Therapy Deliverers by those who strongly oppose tional responses to an outsized Fontanet, MD, DrPH, Pierre-Marie Girard, MD, Roland Landman, MD, as Primary Care Deliverers anything less than unfettered de- reality. Only a public health ap- Bernard Larouzé, MD, Jacques Lebras, The dual role of ARV deliv- livery of ARV therapy to high- proach will allow a timely and PhD, Bernard Liautaud, MD, and Sophie erer and primary care deliverer prevalence settings in Africa and proportional impact, with eyes Matheron, MD, from the Université de Paris, the Université de Montpellier, the will reduce the stigmatization of resource-poor countries is that wide open to the dangers and to Institut Pasteur de Paris, the Centre HIV and minimize resources and to do otherwise is a breach of the responsibilities of all health Hospitalo-Universitaire de Martinique. personnel drain. This principle human rights.36 The argument providers and decisionmakers. This article is dedicated to Professor Jean-Pierre Coulaud, who in 1997 also recognizes reality: a public advanced by those advocating chaired the first international meeting health problem affecting a sub- more research and the adapted Guidelines in the Use of Antiretroviral stantial number of the popula- transposition of northern clinical About the Authors Drugs in Africa in Dakar, Senegal, and remains a fervent advocate for, and an tion is a primary care issue. The standards is that we should not Moïse Desvarieux is with the Department of Epidemiology, Mailman School of Pub- inspiration to, many. appellation “HIV doctors” should allow a lower standard of care in lic Health, Columbia University, New York, be discouraged. resource-poor countries. We NY. Roland Landman is with the Service References Of course, easier treatments, note logical and ethical inconsis- de Maladies Infectieuses, Centre Hospital- 1. World Health Organization. Accel- ier Universitaire, Bichat Claude Bernard, with fewer side effects, would tencies that undermine the feasi- erating access to HIV/AIDS care and Université de Paris, Paris, France, and the treatment in developing countries. Anti- make things simpler. We still bility of programs that would em- Institut de Médecine et d’Épidémiologie retroviral Newsletter. June 2001:1–5. need the innovation of pharma- anate from these high-minded Appliquée (IMEA), Paris. Bernard Liau- taud is with the Hôspital de Jour en Mal- 2. The Global Fund to Fight AIDS, ceutical companies to develop notions. We propose a new strat- adies Infectieuses, Centre Hospitalier Uni- Tuberculosis and Malaria Web page. these better treatments, as much egy: (1) realistically recognize the versitaire de Fort-de-France, Fort-de-France, Available at: http://www.theglobalfund. org/en. Accessed April 6, 2005. as we needed generic drugs to diversity of situations and the Martinique, and Groupe Haitien d’Études du Sarcome de Kaposi et des Infections 3. United States Leadership Against render the current debate even truly attainable goals of any pro- Opportunistes, Port-au-Prince, Haiti. Pierre- HIV/AIDS, Tuberculosis, and Malaria possible. The superb advocacy of gram; (2) in the HPHC countries, Marie Girard is with the Service de Maladies Act of 2003. Pub L No. 108-125.

July 2005, Vol 95, No. 7 | American Journal of Public Health Desvarieux et al. | Peer Reviewed | Health Policy and Ethics | 1121  HEALTH POLICY AND ETHICS 

4. HIV/AIDS: not one epidemic but Immune Defic Syndr. 2004;36: in the 21st century. Lancet. 2002;360: many [editorial]. Lancet. 2004;364:1–2. 100 6–1010. 67–72. 5. Rose G. Sick individuals and sick 17.Place of antiretroviral drugs in the 29.Weidle PJ, Malamba S, Mwebaze R, populations. Int J Epidemiol. 2001;30: treatment of HIV-infected people in et al. Assessment of a pilot antiretroviral 427–432 (discussion 433–434). Africa. International AIDS Society. drug therapy program in Uganda: pa- AIDS. 1999;13:IAS1–IAS3. tients’ response, survival, and drug re- 6. Levi GC, Vitoria MA. Fighting sistance. Lancet. 2002;360:34–40. against AIDS: the Brazilian experience. 18. Laurent C, Diakhate N, Gueye NF, AIDS. 2002;16:2373–2383. et al. The Senegalese government’s 30. Liechty CA, Bangsberg DR. Doubts highly active antiretroviral therapy initia- about DOT: antiretroviral therapy for 7. Lucas SB, Hounnou A, Peacock C, tive: an 18-month follow-up study. resource-poor countries. AIDS. 2003; et al. The mortality and pathology of AIDS. 2002;16:1363–1370. 17 :1383–1387. HIV infection in a west African city. AIDS. 19 93;7:1569–1579. 19. Landman R, Schiemann R, Thiam S, 31.Poole C, Chesney MA, Mdani L, et al. Once-a-day highly active antiretro- Dzazela N, Nyatela N, Kasper T. Patient- 8. Yeni PG, Hammer SM, Carpenter CC, viral therapy in treatment-naive HIV-1- centered approaches to adherence to et al. Antiretroviral treatment for adult infected adults in Senegal. AIDS. 2003; highly-active antiretroviral therapy in HIV infection in 2002: updated recom- 17 :1017–1022. resource-poor settings [abstract mendations of the International AIDS ThPeF8195]. In: Program and abstracts Society-USA Panel. JAMA. 2002;288: 20.Agence Nationale de Recherche of the XIV International AIDS Confer- 222–235. sur le Sida. Institut de Médecine et ence; July 7–12, 2002; Barcelona, d’Epidémiologie. Use of antiretroviral 9. Scaling Up Antiretroviral Therapy Spain. drugs in the management of HIV-infected in Resource-Limited Settings: Guidelines persons. Updated Recommendations. 32. Stringer JS, Rouse DJ, Sinkala M, et for a Public Health Approach. Geneva, 2000. Available at: http://www.imea.fr. al. Nevirapine to prevent mother-to-child Switzerland: World Health Organization; Accessed April 20, 2005. transmission of HIV among women of 2002. unknown serostatus. Lancet. 2003;362: 21. Coetzee D, Hildebrand K, Boulle A, 10.Rabkin M, El-Sadr W, Katzenstein 1850–1853. et al. Outcomes after two years of pro- DA, et al. Antiretroviral treatment in re- viding antiretroviral treatment in 33. Jourdain G, Ngo-Giang-Huong N, source-poor settings: clinical research Khayelitsha, South Africa. AIDS. 2004; LeCoeur S, et al. for the Perinatal HIV priorities. Lancet. 2002;360: 18:887–895. Prevention Trial Group. Intrapartum ex- 1503–1505. posure to nevirapine and subsequent 22. Farmer P, Leandre F, Mukherjee J, 11. Mekonnen Y, Dukers NH, Sanders E, maternal responses to nevirapine-based Gupta R, Tarter L, Kim JY. Community- et al. Simple markers for initiating anti- antiretroviral therapy. N Engl J Med. based treatment of advanced HIV dis- retroviral therapy among HIV-infected 2004;351:229–240. ease: introducing DOT-HAART (directly Ethiopians. AIDS. 2003;17:815–819. observed therapy with highly active an- 34. Laurent C, Kouanfack C, Koulla- 12. Didier JM, Kazatchkine MD, De- tiretroviral therapy). Bull World Health Shira S, et al. for the ANRS 1274 study mouchy C, et al. Comparative assess- Organ. 2001;79:1145–1151. group. Effectiveness and safety of a ment of five alternative methods for generic fixed-dose combination of nevi- 23. DOTS-Plus and the Green Light CD4 T-lymphocyte enumeration for im- rapine, stavudine, and lamivudine in Committee. World Health Organization plementation in developing countries HIV-1 infected adults in Cameroon: Web site. Available at: http://www.who. [letter]. J Acquir Immune Defic Syndr. open-label multicentre trial. Lancet. int/tb/dots/dotsplus/management/en. 2001;26:193–195. 2004;364:29–34. Accessed April 19, 2005. 13.UNAIDS 2004 report on the 35. Médecins sans Frontières. Untan- 24. Quinn TC, Wawer MJ, Sewankambo global AIDS epidemic. Available at: gling the web of price reductions: a pric- N, et al. Viral load and heterosexual trans- http://www.unaids.org/bangkok2004/ ing guide for the purchase of antiretrovi- mission of human immunodeficiency report.html. Accessed July 8, 2004. rals for developing countries. Available virus type 1. Rakai Project Study Group. at: http://www.accessmed-msf.org/up- 14 . Cohen J, Kimaiyo S, Winstone N, et N Engl J Med. 2000;342:921–929. al. Addressing the educational void dur- load/ReportsandPublications/ 25. Gayle HD. Curbing the global ing the antiretroviral therapy rollout [let- 91220021653552/Final.pdf. Accessed AIDS epidemic. N Engl J Med. 2003; ter]. AIDS. 2004;18:2105–2106 April 17, 2005. 348:1802–1805. 15. Diagbouga S, Chazallon C, 36. Hogg R, Cahn P, Katabira ET, et al. 26. Bayer R. Public health policy and Kazatchkine M, et al. Successful imple- Time to act: global apathy towards HIV/ the AIDS epidemic. An end to HIV ex- mentation of a low-cost method for AIDS is a crime against humanity. ceptionalism? N Engl J Med. 19 91;324: enumerating CD4 T-lymphocytes in re- Lancet. 2002;360:1710–1711. 1500–1504. source-limited settings: the ANRS 12–26 study. AIDS. 2003;17: 27.Reynolds SJ, Bartlett JG, Quinn TC, 2201–2208. Beyrer C, Bollinger RC. Antiretroviral therapy where resources are limited. N 16. Balakrishnan P, Dunne M, Ku- Engl J Med. 2003;348:1806–1809. marasamy N, et al. An inexpensive, simple and manual method of CD4 T-cell 28. De Cock KM, Mbori-Ngacha D, quantitation in HIV-infected individuals Marum E. Shadow on the continent: for use in developing countries. J Acquir public health and HIV/AIDS in Africa

1122 | Health Policy and Ethics | Peer Reviewed | Desvarieux et al. American Journal of Public Health | July 2005, Vol 95, No. 7  HEALTH POLICY AND ETHICS 

Ethical Goals of Community Consultation in Research

| Neal Dickert, BA, and Jeremy Sugarman, MD, MPH, MA

In response to the traditional larly AIDS and breast cancer been developed for guiding dif- boards, and communities to eval- emphasis on the rights, inter- communities, during the 1980s ferent types of community-based uate community consultation ests, and well-being of individ- lobbied for access to experimen- research.10 , 2 0 , 3 0 Among the processes. We in no way intend ual research subjects, there has tal treatments and for a larger methods for involving the com- to detract from the need to con- been growing attention focused role in the development of re- munity in research, community sider the particularities of any on the importance of involving search agendas, study designs, consultation is particularly com- given research project and com- communities in research de- and drug approval processes.1–4 mon. Formal community consul- munity; careful understanding of velopment and approval. Community consultation is a Research in population genetics tation is required by Food and particular contexts is indispensa- particularly common method of raised awareness about the risks Drug Administration regulations ble for understanding and achiev- involving communities. How- for stigmatization and discrimina- before research is allowed to be ing the goals we recommend. ever, the fundamental ethical tion,5–7 and studies of indigenous conducted without informed goals of community consulta- communities raised concerns consent in emergency settings.31 DISTINGUISHING tion have not been delineated, about threats to communal iden- Similarly, proposed and actual COMMUNITY which makes it difficult for tity and social structure.7–10 Inter- guidelines for research that in- CONSULTATION investigators, sponsors, and national collaborative research volves aboriginal communi- AND CONSENT institutional review boards to includes working with societies ties,27,28,30 population genetics design and evaluate consulta- that have radically different and epidemiologic research,29,32 Community consultation should tion efforts. structures and norms,11–14 and re- international research,12 , 2 9 and not be mistaken for community Community consultation must 25,26,33 be tailored to the communities searchers in emergency settings, HIV-related research rec- consent, although the 2 are not in which it is conducted, but the where obtaining participant or ommend consulting communities mutually exclusive. To consult purposes of consultation—the surrogate consent is impractical, when planning and implement- is “to seek advice or informa- ethical goals it is designed to must conduct community consul- ing research.8 tion.”34 Consulting with a com- achieve—should be universal. tation during project develop- Despite such endorsements, munity includes eliciting feed- We propose 4 ethical goals that ment and approval.15 ,16 Finally, the general ethical goals of com- back, criticism, and suggestions; give investigators, sponsors, informs of community-based re- munity consultation remain unit does not include asking for ap- stitutional review boards, and search, such as participatory ac- clear, which makes designing and proval or permission. Community communities a framework for tion research, include communi- evaluating consultation efforts a consultation is designed to recog- evaluating community consul- ties throughout the research challenge for investigators, spon- nize and accommodate the rele- tation processes. (Am J Public process.17–20 Despite an increas- sors, and institutional review vant particularities of a given Health. 2005;95:1123–1127. doi: 10.2105/AJPH.2004.058933) ing sense of need for community boards. Because of the heteroge- community for a specific project. input, difficult questions persist neity of communities and re- For example, community consul- about how best to involve com- search projects, the methods of tation for HIV-related trials may ALTHOUGH ETHICAL munities as partners in research. community consultation must be include consulting with HIV ad- considerations of human subjects Efforts to expand attention to context specific. However, the vocacy groups, people who are research have historically fo- community perspectives, beyond purposes of community consulta- HIV-infected, and potential par- cused on protecting the rights, representation on institutional tion—the fundamental ethical ticipants.1,35 Conducting genetics interests, and well-being of review boards,21–23 have ranged goals that consultation is de- research in an aboriginal com- individual subjects, growing at- from advocating a principle of signed to achieve—should be uni- munity may necessitate dis- tention has been given to the im- respect for community to estab- versal. We propose a set of gen- cussing studies with existing po- portance of involving communi- lishing guidelines that require eral goals for community litical authorities and community ties in research development and community disclosure, consulta- consultation that will provide a members.9,30 approval. Activists who represent tion, and consent.12 , 2 4–30 Addi- framework for investigators, Rather than soliciting input, “disease communities,” particu- tionally, general principles have sponsors, institutional review community consent involves so-

July 2005, Vol 95, No. 7 | American Journal of Public Health Dickert and Sugarman | Peer Reviewed | Health Policy and Ethics | 1123  HEALTH POLICY AND ETHICS 

liciting approval or permission CHALLENGES when deciding whether to con- ETHICAL GOALS to conduct a study within a OF COMMUNITY duct community consultation in- FOR COMMUNITY community. Community consent CONSULTATION clude the particular community CONSULTATION may occur after community con- under consideration, the nature sultation and does not obviate Potential difficulties exist at of the research, and the likely im- A clear set of ethical goals will the need for individual con- each stage of community consul- pact of the research on that com- help investigators, sponsors, insti- sent.36,37 Rather, the community tation. At the outset, it can be munity. Further analysis is tutional review boards, and regu- decides whether to permit inves- hard to identify communities and needed; however, we hope that lators plan and assess community tigators to solicit participation stakeholders that have legitimate articulating the goals of commu- consultation methods. Addition- from community members. For and relevant interests. Common nity consultation will at least be a ally, such a framework will pro- community consent to be valid, elements exist among concep- helpful step in determining when vide endpoints for measuring the there must be a legitimate politi- tions of community,40 but delin- consultation is warranted. adequacy of consultation meth- cal system in place, with repre- eating and identifying particular The type of community being ods. We propose four ethical sentatives properly empowered communities for consultation can considered for consultation is an goals of community consultation: to make such decisions on be- be challenging. Identifying the important factor when determin- (1) enhanced protection, (2) en- half of the community.37,38 In community at risk for HIV, for ing the way in which community hanced benefits, (3) legitimacy, many aboriginal communities, example, can be problematic, input is solicited. Common solicita- and (4) shared responsibility such legitimate systems exist. because at-risk individuals may tion methods include open public (Table 1). However, disease-based commu- not believe they are a part of forums, meetings with community nities and many social groups any such community.41 Identify- advisory board members, presen- Enhanced Protection typically lack a political struc- ing representatives also can be tations at meetings of religious or Enhancing the protection of ture, which makes community difficult. Helpful procedures for civic organizations, and radio and research participants’ interests consent inappropriate.37,38 identifying representatives have television call-in shows.1,17,44–49 and welfare is grounded in the Although conceptually distinct, been suggested by The National Devising successful methods for researchers’ duty to minimize the line between community con- Institute for General Medical Sci- generating public input can be risks for research subjects. Con- sultation and community consent ence,32 but important conceptual challenging, particularly in com- sultation efforts should be de- is inevitably blurred in practice. and practical challenges remain. munities that lack a well-defined signed and conducted to help It would be disingenuous to For example, no clear representa- structure or are geographically dis- identify risks or hazards for indi- enter into a consulting arrange- tive exists for persons who may parate.37,50 In many cases, multi- viduals and communities and to ment where the consulting party suffer from traumatic brain in- ple modalities of interaction must identify additional protections does not intend, ex ante, to take jury or cardiac arrest.42,43 be employed.44 It can also be diffi- to ensure the safety of research the consultants’ advice. If rele- Closely related to the chal- cult to determine when consulta- participants. vant consultants have strong neg- lenge of identifying communities tion efforts have been sufficient. Some risks, particularly social ative reactions or endorse partic- is deciding when communities Although insufficient consultation risks, may not be apparent at ular modifications, those should be consulted (assuming can be ineffective, requiring overly the outset to investigators, spon- reactions or modifications have they can be identified). In certain extensive consultation may ham- sors, and institutional review significant moral force and war- cases, there are regulatory re- per important work. boards. Members of cancer ad- rant respect and careful consider- quirements for community con- Finally, incorporating consult- vocacy groups, for example, may ation, even though investigators sultation.31 Similarly, when re- ants’ input into research plans serve as important consultants may sometimes justifiably act search poses real risks for social can be challenging. Although it when designing informational contrary to such opinions. Other- stigma to well-defined communi- is undesirable to override or dis- materials or calling attention to wise, community consultation is ties, such as certain genetics stud- miss community objections or concerns about adverse treat- merely symbolic.39 Despite the ies in native communities, the concerns, failure to conduct im- ment effects that may not be ob- clear conceptual distinction be- need for community consultation portant research on the basis of vious to researchers conducting tween consent and consultation, is evident.44 Yet, requiring com- objections by groups who are a cancer trial.51 When research the degree to which consultants’ munity consultation in all re- nonrepresentative or who have is conducted in emergency set- support is necessary represents a search projects is unwarranted. not carefully considered the is- tings, community consultation persistent challenge.15 ,16 , 2 9 Relevant factors to consider sues at hand is also problematic. may generate discussion that

1124 | Health Policy and Ethics | Peer Reviewed | Dickert and Sugarman American Journal of Public Health | July 2005, Vol 95, No. 7  HEALTH POLICY AND ETHICS 

TABLE 1—Ethical Goals of Community Consultation host country researchers and oth- ity of an entire population, com- ers in the host community, the munity consultation seems essen- Ethical Goal Definition areas of infrastructure that need tial for legitimacy. 14 Enhanced protection Enhance protections for subjects and communities by improvement can be identified. The challenges to achieving identifying risks or hazards that were not previously Similarly, a central notion in this goal are well-known. What appreciated and by suggesting or identifying potential community-based participatory counts as a community? Who protections research is that communities counts as a representative? What Enhanced benefits Enhance benefits to participants in the study, the population should be involved in identifying level of community support is for which the research is designed, or the community in research questions and planning needed to legitimize a particular which the study is conducted studies in order to conduct stud- study? These are deep, concep- Legitimacy Confer ethical/political legitimacy by giving those parties ies that benefit the particular tual questions for which we do 20 with an interest or stake in the proposed research the communities involved. In short, not have well-developed an- opportunity to express their views and concerns at a community consultation may en- swers; however, it is important to time when changes can be made to the research protocol hance direct, indirect, and aspira- note that the goal of legitimacy 53 Shared responsibility Consulted communities may bear some degree of moral tional benefits. Investigators are refers to the process of commu- responsibility for the research project and may take on by no means required to provide nity consultation and not the po- some responsibilities for conducting the study all benefits that could conceiv- litical legitimacy of consulted ably be offered to participants or bodies.37 Suggestions or concerns communities, but enhancing ben- that are expressed during com- helps to identify groups who are Enhanced Benefits efits to ensure that research ef- munity consultations are signifi- likely to want to opt out of spe- Enhancing benefits through forts are mutually beneficial is an cant, even when consultants lack cific studies and that suggests community consultation is con- important goal. the authority to provide consent strategies to facilitate the identi- sistent with researchers’ general on behalf of the community. fication of those groups during duty of beneficence toward re- Legitimacy the study. In this sense, commu- search subjects.52 Early HIV re- Community consultation can Shared Responsibility nity consultation may be a par- search illustrates how commu- help to confer ethical and politi- As partners in the research ticularly effective way for inves- nity consultation enhances cal legitimacy on a research proj- process, community members tigators to identify individuals or benefits to individual subjects. ect by engaging in a process in may share responsibility in 2 subgroups with particular needs For example, 1 community advi- which stakeholders (those peo- ways. First, community consult- or vulnerabilities that individu- sory board recommended that a ple, institutions, and groups that ants may assume active roles in als outside the community may clinical trial incorporate referral have an interest in the proposed conducting research. Commu- not recognize. programs for participants to gain research) may express their nity advisory board members, Community consultation also access to available ancillary ser- views and concerns. The oppor- for example, may become in- may enhance nonparticipant pro- vices.35 Based on this recom- tunity to speak has significant volved in recruiting subjects for tections by identifying risks for mendation, investigators chose justificatory power for imposing studies40,48,54 and thus bear community members who are to incorporate such programs research risks, especially when some responsibility for the suc- not enrolled in the study. For ex- into their studies. individuals are unable to provide cess of research efforts. Second, ample, studies of cancer suscepti- Community consultation may consent and surrogate decision by acknowledging the stake that bility that were conducted also enhance benefits to the com- makers are unavailable. In such community members have in among Ashkenazi Jews were op- munity of individuals who have cases, community consultation the conduct of research, and by posed by some community mem- the condition being studied or may be the only chance investi- soliciting their assistance and bers who were concerned that to the larger communities to gators have to assess the likely input through a legitimate pro- research findings might be used which study subjects belong. In preferences of the study popula- cess, community consultation for eugenics or might jeopardize the international setting, a com- tion. Similarly, when a study confers on communities a de- health coverage.5 Although all mon benefit of research involve- poses significant risks for a com- gree of moral responsibility for risks are not preventable, making ment is the improvement of the munity, such as genetics research the research.26 them explicit and minimizing research or health care delivery that could have potentially nega- Shared responsibility is partic- them are essential goals. infrastructure. By consulting with tive implications on the insurabil- ularly evident with cases involv-

July 2005, Vol 95, No. 7 | American Journal of Public Health Dickert and Sugarman | Peer Reviewed | Health Policy and Ethics | 1125  HEALTH POLICY AND ETHICS 

ing HIV advocacy groups, where be measured by observing Johns Hopkins University, Hampton 10.Freeman WL. The Protection of Po- the advocacy groups have taken whether a particular consultation House 351, 624 N Broadway, Baltimore, tential Individual Volunteers and Tribal MD 21205 (e-mail: [email protected]). Communities in Research Involving the on the role of actually conduct- effort identifies additional risks This article was accepted February 15, Indian Health Service (IHS). Available ing studies,1 and with cases in- previously unknown to investiga- 2005. at: http://www.npaihb.org/epi/irb/ volving participatory-action re- tors or whether it proposes new Protections_fr.pdf. Accessed February 4, 2005. search or community-based solutions for minimizing risk. Contributors participatory research, where There are currently few empiri- J. Sugarman originated the idea for this 11. Sugarman J, Popkin B, Fortney J, article; both authors developed ideas Rivera R. International perspectives on communities are involved at cal data on the effectiveness of and reviewed each draft of the article. protecting human research subjects. In: every stage.17–19 It is important consultation strategies.54 By iden- Ethical and Policy Issues in International to clarify that sharing responsibil- tifying the goals of the process, Research: Clinical Trials in Developing Acknowledgments Countries. Bethesda, Md: National ity does not constitute a shifting this framework should facilitate We are extremely grateful for the help- Bioethics Advisory Commission; 2001. of blame or removal of responsi- attempts to assess different types ful comments we received while developing this article. In particular, we 12. National Bioethics Advisory Com- bility from investigators, spon- of consultation efforts in different would like to thank Ezekiel Emanuel, mission. Ethical and Policy Issues in In- sors, and institutional review settings and enhance understand- Christine Grady, Kate MacQueen, Holly ternational Research: Clinical Trials in Developing Countries. Bethesda, Md: boards. On the contrary, commu- ing of which consultation meth- Taylor, and Myron Weisfeldt. We also thank the reviewers for their very National Bioethics Advisory Commis- nity consultation places addi- ods are appropriate in varying thoughtful and instructive suggestions. sion; 2001. tional responsibility on investiga- types of communities and types 13. Macklin R. International research: tors to attend to important of research. Human Participant Protection ethical imperialism or ethical pluralism? Accountability Res. 1999;7:59–83. community concerns. The degree Finally, this framework draws No protocol approval was needed for to which responsibility can be attention to 2 important lingering this project. 14 .Participants in the 2001 Confer- ence on Ethical Aspects of Research in shared is limited by the degree to issues that are beyond the scope References Developing Countries. Enhanced: fair which investigators and sponsors of this article. First, an account is benefits for research in developing coun- 1. Spiers HR. Community consulta- tries. Science. 2001;298:2133–2134. are sensitive to and accommo- needed for determining when in- tion and AIDS clinical trials: part I. IRB. date those concerns. vestigators may justifiably over- 19 91;13(3):7–10. 15. Biros M. Research without consent: current status, 2003. Ann Emerg Med. ride or dismiss community con- 2. Spiers HR. Community consulta- 2003;42:550–564. CONCLUSIONS cerns. Such an account must be tion and AIDS clinical trials: part II. 16.Passamani ER, Weisfeldt ML. Task particularly sensitive to the na- IRB. 19 91;13(4):1–6. force 3: special aspects of research con- 55 3. Spiers HR. Community consulta- As the need for identifying ture of disagreements. Second, ducted in the emergency setting: waiver tion and AIDS clinical trials: part III. of informed consent. J Am Coll Cardiol. and incorporating community further research is needed to de- IRB. 19 91;13(5):3–7. 2000;35:862–880. input into the design, planning, termine what types of studies re- 4. Dresser R. When Science Offers Sal- 17. Quinn SC. Protecting human sub- and conduct of research has be- quire community consultation vation: Patient Advocacy and Research jects: the role of community advisory Ethics. New York, NY: Oxford University come clearer, it is critical to iden- and what types of consultation boards. Am J Public Health. 2004;94: Press; 2001. tify the ethical goals of commu- are needed for particular re- 918–922. 5. Lehrman S. Jewish leaders seek ge- nity consultation. Attention to search projects. In the meantime, 18. Macaulay AC, Commanda LE, netic guidelines. Nature. 19 97;389:322. the 4 ethical goals of enhancing careful attention to the 4 ethical Freeman WL, et al. Participatory research maximises community and lay protection, enhancing benefits, goals we have identified should 6. Foster MW, Eisenbraun AJ, Carter TH. Communal discourse as a supple- involvement. BMJ. 1999;319:774–778. creating legitimacy, and sharing facilitate the proper incorpora- ment to informed consent for genetic re- 19. Lantz PM, Viruell-Fuentes E, Israel search. Nat Genetics. 19 97;17:277–279. responsibility should allow for tion of community views into re- BA, Softley D, Guzman R. Can commu- more effective assessment by search and its oversight. 7. Juengst ET. FACE facts: why nities and academia work together on communities, investigators, insti- human genetics will always provoke public health research? Evaluation re- bioethics. Am J Law, Med, Ethics. 2004; sults from a community-based partici- tutional review boards, and spon- 32:252–266. patory research partnership in Detroit. sors of particular consultation ef- About the Authors J Urban Health. 2001;78:495–506. 8. Weijer C, Goldsand G, Emanuel EJ. The authors are with the Phoebe R. Berman forts. We also hope that these Protecting communities in research: cur- 20. Israel BA, Schulz AJ, Parker EA, Bioethics Institute, Johns Hopkins Univer- rent guidelines and limits of extrapola- Becker AB. Review of community-based goals can be developed into sity, Baltimore, Md. Jeremy Sugarman also tion. Nat Genetics. 1999;23:275–280. research: assessing partnership ap- metrics by which methods of is with the Department of Medicine, Johns proaches to improve public health. Ann Hopkins University. 9. Dodson M. Indigenous peoples and community consultation may be Rev Public Health. 1998;19:173–202. Request for reprints should be sent to the morality of the Human Genome Di- systematically assessed. For ex- Jeremy Sugarman, MD, MPH, MA, versity Project. J Med Ethics. 1999;25: 21.US Department of Health and ample, enhanced protections can Phoebe R. Berman Bioethics Institute, 204–208. Human Services. Protection of Human

1126 | Health Policy and Ethics | Peer Reviewed | Dickert and Sugarman American Journal of Public Health | July 2005, Vol 95, No. 7  HEALTH POLICY AND ETHICS 

Subjects. Title 45, CFR, 46. Washington, ments for emergency research. Title 20, 43. Browder JP. Can community con- 53. King NMP. Defining and describing DC: US Department of Health and Code of Federal Regulations, Section 50.24. sultation substitute for informed consent benefit appropriately in clinical trials. Human Services, National Institutes of Available at: http://www.accessdata.fda. in emergency research: a response. In: Am J Law, Med, Ethics. 2000;28: Health, and Office of Human Research gov/scripts/crdh/cfdocs/cfcfr/cfrsearch. King NMP, Henderson GE, Stein J, eds. 332–343. Protections; 2001. cfm. Accessed April 10, 2005. Beyond Regulations: Ethics in Human 54. Morin SF, Maiorana A, Koester KA, Subjects Research. Chapell Hill, NC: Uni- 22. The National Commission for the 32. Policy for the Responsible Collection, Sheon NM, Richards TA. Community versity of North Carolina Press; 1999. Protection of Human Subjects of Bio- Storage, and Research Use of Samples consultation in HIV prevention re- medical and Behavioral Research. The from Named Populations for the NIGMS 44.Foster MW, Bernsten D, Carter TH. search: a study of community advisory Belmont Report: Ethical Principles and Human Genetic Cell Repository. Available A model agreement for genetic research boards at 6 research sites. J Acquired Guidelines for the Protection of Human at: http://locus.umdnj.edu/nigms/comm/ in socially identifiable populations. Am Immune Defic Syndrome. 2003;33: Subjects. Washington, DC: US Depart- submit/collpolicy.html. Accessed Febru- J Hum Genetics. 1998;63:696–702. 513–520. ment of Health, Education, and Wel- ary 7, 2005. 45. Shah AN, Sugarman J. Protecting 55. Gutmann A, Thompson D. Democ- fare; 1979. 33. MacQueen KM, Sugarman J, on be- research subjects under the waiver of racy and Disagreement. Cambridge, 23. National Bioethics Advisory Com- half of the HPTN Ethics Working Group. informed consent for emergency re- Mass: Belknap Press of Harvard Univer- mission. Ethical and Policy Issues in Re- HIV Prevention Trials Network Ethics search: experiences with efforts to in- sity Press; 1996. search Involving Human Participants. Be- Guidance for Research: HPTN Ethics form the community. Ann Emerg Med. thesda, Md: National Bioethics Advisory Working Group; 2003. Available at: http:// 2003;41:72–78. Commission; 2001. www.hptn.org/web%20documents/ewg/ hptnethicsguidancefinal15april2003.pdf. 46. Santora TA, Cowell V, Trooskin SZ. 24.Weijer C. Protecting communities Accessed April 10, 2005. Working through the public disclosure in research: philosophical and pragmatic process mandated by use of 21 CFR challenges. Camb Q Healthc Ethics. 1999; 34. The American Heritage Collegiate 50.24 (exception to informed consent): 8:501–513. Dictionary. 3rd ed. New York, NY: guidelines for success. J Trauma. 1998; Houghton Mifflin; 1993. 25. Levine C, Dubler NN, Levine RJ. 45:907–913. Building a new consensus: ethical prin- 35. Valdiserri RO, Tama GM, Ho M. 47. Dix ES, Esposito D, Spinosa F, ciples and policies for clinical research The role of community advisory com- Olson N, Chapman S. Implementation of on HIV/AIDS. IRB. 19 91;13:1–17. mittees in clinical trials of anti-HIV community consultation for waiver of agents. IRB. 1988;10:5–7. 26. UNAIDS. Ethical Considerations in informed consent in emergency re- Preventive HIV Vaccine Research: UNAIDS 36. Ijsselmuiden C, Faden R. Research search: one institutional review board’s Guidance Document. Available at: http:// and informed consent in Africa—another experience. J Invest Med. 2004;52: www.unaids.org. Accessed February 5, look. N Engl J Med. 19 92;326: 113–116. 830–833. 2005. 48. Blanchard L. Community assess- 27. Canada Tri-Council. Tri-Council 37.Weijer C, Emanuel EJ. Protecting ment and perceptions: preparations for Policy Statement, Ethical Conduct for Re- communities in research. Science. 2000; HIV vaccine efficacy trials. In: King NMP, search Involving Humans. Available at: 289:1142–1144. Henderson GE, Stein J, eds. Beyond Reg- http://www.pre.ethics.gc.ca/english/pdf/ 38.Weijer C, Miller PB. Protecting ulations: Ethics in Human Subjects Re- tcps%20june2003_e.pdf. Accessed Feb- communities in pharmacogenetic and search. Chapel Hill, NC: University of ruary 7, 2005. pharmacogenomic research. Pharma- North Carolina Press; 1999. 28. National Health and Medical Re- cogenomics J. 2004;4:9–16. 49. Strauss RP, Sengupta S, Quinn SC, search Council. Values and Ethics: Guide- 39. Minkler M, Pies C. Ethical Issues et al. The role of community advisory lines for Ethical Conduct in Aboriginal and Practical Dilemmas in Community boards: involving communities in the in- and Torres Strait Islander Health Re- Organization and Community Participa- formed consent process. Am J Public search. Available at: http://www.health. tion. In: Minkler M, ed. Community Or- Health. 2001;91:1938–1943. gov.au/nhmrc/publications/pdf/e52.pdf. ganization and Community Building for 50. Foster MW, Sharp RR, Freeman WL, Accessed February 7, 2005. Health. 2nd ed. New Brunswick, NJ: Chino M, Bernstein D, Carter TH. The 29. Council for International Organiza- Rutgers University Press; 2005. role of community review in evaluating tions of Medical Sciences. International 40. MacQueen KM, McLellan E, the risks of human genetic variation re- Ethical Guidelines for Biomedical Re- Metzger DS, et al. What is community? search. Am J Hum Genet. 1999;64: search Involving Human Subjects. Ge- An evidence-based definition for partici- 1719–1727. neva, Switzerland: Council for Interna- patory public health. Am J Public Health. 51. Girling DJ. Important issues in tional Organizations of Medical 2001;91:1929–1938. planning and conducting multi-centre Sciences; 2002. 41.Grady C. The Search for an HIV randomised controlled trials in cancer 30. The Australian Institute of Aborig- Vaccine. Bloomington, Ind: Indiana Uni- and publishing their results. Crit Rev inal and Torres Strait Islander Studies. versity Press; 1995. Oncol Hematol. 2000;36:13–25. Guidelines for Ethical Research in Indige- 42. King NMP. Medical research: using 52. National Commission for the Pro- nous Studies. Available at: http://www. a new paradigm where the old might do. tection of Research Risks. The Belmont aiatsis.gov.au/corp/docs/ethicsguidea4. In: King NMP, Henderson GE, Stein J, Report: Ethical Principles and Guidelines pdf. February 7, 2005. eds. Beyond Regulations: Ethics in Human for the Protection of Human Subjects of 31.US Food and Drug Administration. Subjects Research. Chapell Hill, NC: Uni- Research. Washington, DC: Government Exception from informed consent require- versity of North Carolina Press; 1999. Printing Office; 1979.

July 2005, Vol 95, No. 7 | American Journal of Public Health Dickert and Sugarman | Peer Reviewed | Health Policy and Ethics | 1127  PUBLIC HEALTH THEN AND NOW  STERILIZEDSTERILIZED in the Name of Public Health Race, Immigration, and Reproductive Control in Modern California

| Alexandra Minna Stern, PhD In exploring the history of involuntary sterilization in California, I con- nect the approximately 20000 oper- THE YEAR WAS 1979 AND THE was unanimously approved in junction with the family planning ations performed on patients in state place was the state capitol in the State Assembly and Senate, initiatives of the War on Poverty, 2 institutions between 1909 and 1979 to Sacramento, Calif. Assemblyman in committee and on the floor. launched by President Lyndon B. the federally funded procedures car- Art Torres, chairman of the On the surface, this vignette Johnson in 1964. Health Committee, introduced a might seem to encapsulate little For the most part, Madrigal v ried out at a Los Angeles County hos- bill to the legislature to repeal more than the purging of an anti- Quilligan has been understood in pital in the early 1970s. the state’s sterilization law. First quated law, enacted infrequently light of the thousands of un- Highlighting the confluence of fac- passed in the same chambers 70 since the 1950s, from the legisla- wanted sterilizations reported in tors that facilitated widespread ster- years earlier and modified sev- tive annals. Torres, however, had the United States from the late ilization abuse in the early 1970s, eral times over the decades, learned that California’s steriliza- 1960s to the mid-1970s. And I trace prosterilization arguments this statute had sanctioned over tion law was still on the books certainly, the experiences of the predicated on the protection of pub- 20000 nonconsensual steriliza- only when several residents of his Mexican-origin women who suf- lic health. tions on patients in state-run predominantly Latino Los Ange- fered at the scalpels of County This historical overview raises im- homes and hospitals, or one third les district sued the Women’s General physicians mirror those portant questions about the legacy of the more than 60000 such Hospital at the University of of the African American, Puerto of eugenics in contemporary Califor- procedures in the United States Southern California/Los Angeles Rican, and Native American nia and relates the past to recent de- in the 20th century. In a letter to County General Hospital (here- women who came forth with Governor Edmund G. Brown after called County General) for comparable stories during the velopments in health care delivery urging his signature, Torres as- nonconsensual sterilizations in same years. Yet Madrigal v Quilli- and genetic screening. (Am J Public serted that the law was “out- 1975.3 The plaintiffs in this class- gan should also be analyzed lon- Health. 2005;95:1128–1138. doi:10. dated” and that the criteria used action suit, Madrigal v Quilligan, gitudinally, as a concluding link 2105/AJPH.2004.041608) to authorize a sterilization order, were working-class Mexican-ori- in the history of forced steriliza- specifically the clauses referring gin women who had been co- tion in modern California. Just as to a “marked departure from nor- erced into postpartum tubal liga- this case highlights the conflu- mal mentality” and to the genetic tions minutes or hours after ence of factors that facilitated origins of mental disease, had undergoing cesarean deliveries. In sterilization abuse in the early “no meaning in modern medical contrast to the operations carried 1970s, it also illuminates the terminology.”1 Backed by the De- out at state institutions beginning longevity and potency of proster- partment of Developmental Ser- in 1909, these procedures were ilization arguments predicated on vices and the California Associa- financed by federal agencies that the protection of the public’s tion for the Retarded, this bill began to disperse funds in con- health and resources.

1128 | Public Health Then and Now | Peer Reviewed | Stern American Journal of Public Health | July 2005, Vol 95, No. 7  PUBLIC HEALTH THEN AND NOW 

Madrigal v Quilligan demon- if such action would improve his Several years later, the agricul- strates shifts over the past cen- or her “physical, mental, or moral turalist and philanthropist Ezra S. tury in terms of the rationale condition.”6 Gosney, in consultation with the employed to authorize compul- The law was expanded in Eugenics Record Office (located sory sterilizations and the un- 1913 and 1917, when clauses in Cold Spring Harbor, NY), uneven transition from state coer- were added to shield physicians derwrote the Human Betterment cion to patient choice in matters against legal retaliation and to Foundation to foment sterilization pertaining to procreation and re- foreground a eugenic, rather education and legislation. Gosney productive health. To offer his- than penal, rationale for surgery.7 eventually selected Paul Popenoe, torical insight into these complex The 1917 amendment, for exam- a date palm cultivator and social patterns and better comprehend ple, reworded the description of hygienist, to conduct a detailed the fraught politics of reproduc- a diagnosis warranting surgery study of sterilization. After col- tive control, I explore the inter- from “hereditary insanity or in- lecting data and interviewing pa- sections of race, sex, immigra- curable chronic mania or demen- tients and staff at state homes tion, sterilization, and health tia” to a “mental disease which and hospitals, he and Gosney policy by tracing the chronology may have been inherited and is coauthored Sterilization for “Effects of Eugenic Sterilization as Practiced in California,” leaflet and context of involuntary steril- likely to be transmitted to de- Human Betterment: A Summary of 8 disseminated by the Human ization in modern California. scendants.” It also targeted in- Results of 6000 Operations in Cal- Betterment Foundation, Pasadena, I conclude by suggesting some mates afflicted with “various ifornia, 1909–1929, which Calif, from the late 1920s to the of the implications of this history grades of feeblemindedness” and touted the immense value of re- early 1940s. for contemporary public health “perversion or marked depar- programs. tures from normal mentality or from disease of a syphilitic na- JUSTIFYING STERILIZATION: ture.”9 Performed sporadically at FROM DEFECTIVE the beginning, operations began HEREDITY TO to climb in the late 1910s, and OVERPOPULATION by 1921, a total of 2248 people—over 80% of all cases na- When Indiana passed the tionwide—had been sterilized in country’s first sterilization law in California, mostly at the Sonoma 19 07, it was motivated by the and Stockton facilities.10 eugenic family studies of suppos- Home to an extensive eugen- edly defective lineages, such as ics movement that crisscrossed the Jukes and the Kallikaks, that the domains of agriculture, edu- were very much in vogue at the cation, medicine, and govern- turn of the century.4 More ment, California was propitious broadly, such legislation was part terrain for the emergence of a of a wave of Progressive Era far-reaching sterilization regi- public health activism that en- men. Eugenic ideas were es- compassed pure food, vaccina- poused by influential profession- tion, and occupational safety acts. als, such as Stanford University In 1909, driven by the desire to Chancellor David Starr Jordan, apply science to social problems, the Santa Rosa “plant wizard” California passed the third sterili- Luther Burbank, and the Los zation bill in the nation.5 Envi- Angeles politician Dr John R. sioned by F. W. Hatch, the secre- Haynes. In 1924, Charles M. tary of the State Commission in Goethe, a Sacramento business- Lunacy [sic] (renamed the De- man, collaborated with Univer- partment of Institutions in 1921), sity of California zoologist this legislation granted the med- Samuel J. Holmes to found the ical superintendents of asylums Eugenics Section of the San and prisons the authority to Francisco–based Commonwealth “asexualize” a patient or inmate Club of California.

July 2005, Vol 95, No. 7 | American Journal of Public Health Stern | Peer Reviewed | Public Health Then and Now | 1129  PUBLIC HEALTH THEN AND NOW 

productive surgery and rallied One of the reasons for this against communicable diseases, it sterilization crusaders across the longevity was that, from the out- also necessitated “immunizing” United States and Europe.11 This set, California defined steriliza- the hereditarily defective in mission was furthered by the action not as a punishment but as order to prevent the spread of tivities of the Eugenics Society of a prophylactic measure that bad genes. Once seen as integral Northern California, the Califor- could simultaneously defend the to health prophylaxis and as a nia Division of the American Eu- public health, preserve precious cost-saving recourse, sterilization genics Society, and the American fiscal resources, and mitigate the programs intensified at a clipped Institute of Family Relations. In menace of the “unfit” and “fee- pace across the country in the addition to these organizations, bleminded.” California’s pre- 193 0s.15 By 1932, twenty-seven California’s sterilization system science was acknowledged in states had laws on the books and was buoyed by the administration 1927, when the most powerful procedures nationwide reached and involvement of the Depart- judiciary in the land, the US over 3900.16 Not only did opera- ment of Institutions, which man- Supreme Court, ruled affirma- tions increase markedly during aged state homes and hospitals, tively on the constitutionality of this decade, but some states, several of which were run by ar- Virginia’s sterilization statute in such as Georgia and South Car- dent superintendents who de- Buck v Bell, countenancing steril- olina, passed legislation for the vised novel surgical techniques. ization on behalf of the collec- first time.17 tive health of the citizenry.13 In California, at least into the 1950s, compulsory sterilization In California, at least into the 1950s, compulsory sterilization was consistently described as a public health strategy that could was consistently described as a public health strategy that breed out undesirable defects could breed out undesirable defects from the populace from the populace and fortify the “ and fortify the state as a whole. state as a whole. Convinced of its efficacy, sterilization proponents pushed for implementation of the Because of the state’s multi- Shaped by the legal struggles law beyond the walls of state in- faceted eugenics movement and ”over states’ rights to vaccinate stitutions. For example, in his Los the fact that it appreciably out- that had played out in the 19th Angeles Times Sunday magazine paced in absolute terms the century, and drawing from Ja- column “Social Eugenics” (which other 32 states that passed steril- cobson v Massachusetts (1905), ran from 1936 to 1941), Fred ization laws at some point in the which had ruled that maintain- Hogue claimed that “in this coun- 20th century, California stands ing the public health outweighed try we have wiped out the mos- out when compared with the rest individual rights when it came to quito carriers of yellow fever and of the country. California carried smallpox immunization, Justice are in a fair way to extinguish the out more than twice as many Oliver Wendell Holmes wrote in malaria carriers: but the human sterilizations as either of its near- his Buck v Bell opinion: “It is bet- breeders of the hereditary physi- est rivals, Virginia (approxi- ter for the world, if instead of cal and mental unfit are only in mately 8000) and North Car- waiting to execute degenerate exceptional cases placed under olina (approximately 7600). offspring for crime, or to let restraint.”18 To rectify this situa- Furthermore, in many states, them starve for their imbecility, tion, Hogue recommended such as New Jersey and Iowa, society can prevent those who broader intervention and argued sterilization laws were declared are manifestly unfit from contin- that eugenic practices, above all unconstitutional, judged to be uing their kind. The principle that sterilization, were essential to “cruel and unusual punishment” sustains compulsory vaccination is “the protection of the public or in violation of equal protec- broad enough to cover cutting the health” and “the health security tion and due process.12 In con- Fallopian tubes. Three genera- of the citizens of every State.”19 trast, California’s statute— tions of imbeciles are enough Along a similar vein, in the sec- although reworked over the [italics added].”14 ond edition of their popular text- years—remained in effect with- If utilitarian pursuit of the book Applied Eugenics, Popenoe out interruption from initial pas- common good required manda- and colleague Roswell H. Johnson sage until repeal. tory vaccination to inoculate underscored that “if persons

1130 | Public Health Then and Now | Peer Reviewed | STern American Journal of Public Health | July 2005, Vol 95, No. 7  PUBLIC HEALTH THEN AND NOW 

whose offspring will be dysgenic 8% of those sterilized. Without mindedness, or mental defi- are so lacking in intelligence, in the forced repatriations of hun- ciency. A notable percentage of foresight, or in self-control that dreds of Mexicans from state these young patients were typed they do not control themselves, facilities, orchestrated by the De- as masturbators or incest perpe- the state must control them. Ster- portation Office of the Depart- trators if male and as promiscu- ilization is the answer.”20 ment of Institutions, it very likely ous—even nymphomaniacal—or Rooted in this logic and that this figure would have been having borne a child out of wed- shored up by Buck v Bell, steril- higher.26 More striking, at the lock if female.30 izations rose dramatically in Cali- Norwalk State Hospital, in south- As scholars have shown, Cali- fornia in the 1930s, peaking at ern California, where a total of fornia’s sterilization program was 848 in 1939 and 818 in 1941. 380 Mexicans constituted 7.8% propelled by deep-seated preoc- By 1942, over 15000 operations of admissions from 1921 to cupations about gender norms had been performed in the state, 193 0, they were sterilized at and female sexuality.31 Especially most since 1925.21 Even when rates of 11% for females and after the procedure of salpingec- per capita comparisons are made 13% for males.27 tomy became faster and less with states with much smaller In addition, whereas African medically risky in the 1920s, the populations, California’s rates Americans constituted just over sterilization of women and young were always clustered at the top. 1% of California’s population, girls categorized as immoral, Not until the 1940s, when Cali- they accounted for 4% of total loose, or unfit for motherhood in- fornia claimed about 60% of all sterilizations.28 While the age of tensified. This trend is captured operations nationwide, did a few those sterilized varied according by the changing ratio between states, such as Delaware, North to sex, institution, and marital sterilizations carried out at insti- Carolina, and Virginia, begin to status, the bulk were in the 20- tutions for the mentally ill and consistently overtake California to 40-year age bracket, with a those performed at institutions in either per capita or annual mean age of commitment of for the feebleminded. Initially, terms.22 about 30 years for men and 28 most operations took place at the Although, for a variety of rea- years for women; sterilization former, affecting more men than sons, it will be next to impossible typically occurred less than 12 women; by the late 1930s, this ever to determine with any accu- months after admission.29 Fur- pattern reversed itself and the racy the total number of steriliza- thermore, unnamed patient gender ratio approached parity. “Sterilization Operations Performed tions, not to mention the statisti- records from the 1920s docu- Additionally, Popenoe catego- in California Mental Hospitals and cal and demographic trends, ment hundreds of individuals in rized most sterilized women as Institutions for Mental Defectives, to June 30, 1941,” in the Statistical some patterns are discernible.23 their late teens and early 20s homemakers and most men as Report of the Department of In his exhaustive survey of state sterilized for dementia praecox manual laborers, not as white- Institutions of the State of hospitals and homes in the late (schizophrenia), epilepsy, manic collar professionals, indicating California (Sacramento: California 1920s, and in a follow-up study depression, psychosis, feeble- that most of those sterilized were State Printing Office, 1941). about a decade later, Popenoe found that the foreign-born were disproportionately affected, con- stituting 39% of men and 31% of women sterilized.24 Of these, immigrants from Scandinavia, Britain, Italy, Russia, Poland, and Germany were most represented.25 These records also reveal that African Americans and Mexicans were operated on at rates that exceeded their population. Although in the 1920 census they made up about 4% of the state population, Mexican men and Mexican women, respectively, comprised 7% and

July 2005, Vol 95, No. 7 | American Journal of Public Health Stern | Peer Reviewed | Public Health Then and Now | 1131  PUBLIC HEALTH THEN AND NOW 

either working class or lower tion, could not breed out defects; women were reviled in accor- middle class.32 even if viable, they would show dance with long-standing ideas of The final substantial year for results only after thousands of public health protection, along California’s sterilization program years of regulated procreation.36 with more recent claims that was 1951, with 255 operations More and more, eugenicists these fecund female immigrants performed. The following year, traded in “unit characters” for were worsening an already se- the number dropped consider- polygenic inheritance and ge- vere overpopulation problem. ably to 51, undoubtedly because netic predispositions. Accompa- of a revision to the statute insert- nying this realignment was a MADRIGAL V QUILLIGAN ing administrative requirements heightened interest in the manip- for physicians and safeguards for ulation and management of A series of overlapping factors patients.33 This amendment, and human heredity through popula- created the milieu for wide- another 1953 bill, deleted any tion control, which postwar eu- spread sterilization abuse in the references to syphilis (long since genicists and their allies pursued United States from the late understood as microbial, not ge- through groups such as the Pop- 1960s to the mid-1970s. This netic, in etiology) and sexual per- ulation Council, Population Ref- period saw the confluence of the version; instituted more demand- erence Bureau, and Planned gains of mainstream feminism ing processes of notice, hearing, Parenthood. with regard to reproductive and appeal; and removed the On the basis of a revamped ra- rights, an unprecedented federal terms “idiots” and “fools” from tionale of bad parenthood and commitment to family planning the law.34 By turning what had population burden, sterilizations and community health, and the been a mere formality into a increased in the 1950s and popularity of the platform of zero more taxing ordeal, these modifi- 1960s in southern states such as population growth, which was cations deterred many physicians North Carolina and Virginia.37 endorsed by immigration restric- from requesting sterilization or- Concurrently, sterilization often tionists and environmentalists ders.35 Nevertheless, surgeries regained a punitive edge and, and put into practice on the op- continued sporadically at every preponderantly aimed at African erating table by some zealous state institution into the 1970s. American and poor women, physicians. In part, this legal modification began to be wielded by state On the one hand, there was in- reflected a shift in the criteria courts and legislatures as a pun- creased availability of and access employed to sanction reproduc- ishment for bearing illegitimate to birth control, including abor- tive surgery, as an emphasis on children or as extortion to ensure tion. For example, by 1970, parenting skills and welfare de- ongoing receipt of family assis- North Carolina, Virginia, Oregon, pendency began to supplant tance.38 By the 1960s, the pro- and Georgia had passed volun- hereditary fitness and putative tracted history of state steriliza- tary sterilization laws and Wash- innate mental capacity as the tion programs in the United ington, DC and New York had le- determinants of an individual’s States, and the consolidation of a galized abortion.39 Quite simply, social and biological drain on rationale for reproductive sur- more women were using birth society. By this time, many eu- gery that was linked to fears of control, especially after the intra- genicists had conceded that overpopulation, welfare depend- uterine device (IUD) and the earlier attempts to stamp out ency, and illegitimacy, set the birth control pill came on the hereditary traits defined as re- stage for a new era of steriliza- market in the 1960s. Voluntary cessive or latent, including alco- tion abuse. In California, which sterilization rates rose in tandem holism, immorality, and the never explicitly endorsed a puni- so that, in 1973, the same year catchall “feeblemindedness,” had tive model, the state program abortion was decriminalized by been proven futile by the Hardy- was fairly quiescent by the mid- the US Supreme Court in Roe v Weinberg equilibrium principle, 1950s. However, when federal Wade, sterilization was the most which demonstrated that the backing for reproductive surgery used method of birth control by overwhelming tendency of gene began to be distributed in the Americans in the 30- to 44-year frequencies and ratios was to re- late 1960s, the eugenic refrains age bracket.40 On the other main constant from one genera- of previous decades resurfaced. hand, in 1969 the American Col- tion to the next. Thus, targeted The reproductive tendencies of lege of Obstetricians and Gyne- interventions, such as steriliza- working-class Mexican-origin cologists dropped its age-parity

1132 | Public Health Then and Now | Peer Reviewed | Stern American Journal of Public Health | July 2005, Vol 95, No. 7  PUBLIC HEALTH THEN AND NOW 

stipulation, which required that a after some initial hesitation, the ductive and sexual autonomy in woman’s age, multiplied by the OEO incorporated sterilization terms of the right to obtain birth number of her children, equal into its medical armamentarium. control, above all abortion, ele- 12 0 in order to qualify for volun- At the same time, Medicaid was vating its federal legalization to tary sterilization. The following permitted to reimburse up to their utmost goal.46 While many year, it retracted the proviso that 90% for an operation.43 Factor- minority and working-class a woman needed to consult 2 ing in the sterilizations backed by women also clamored for greater doctors and a psychiatrist before Medicaid and the Department of reproductive control, they often surgery.41 Health, Education, and Welfare found themselves combating the Federal funding for birth con- (HEW) before the OEO’s deci- reverse equation, namely, that trol and family planning also rose sion, between the late 1960s and they were destructive overbreed- markedly in the late 1960s, most 1974, when federal guidelines ers whose procreative tendencies decisively with the passage of the were formalized, approximately needed to be managed.47 Given Family Planning Services and 100 000 sterilizations were car- that the family planning model Population Research Act in 1970 ried out annually.44 was underpinned by the princi- and the creation of the Office of In theory, the advent of family ples of population control and Economic Opportunity (OEO). planning resources and reproduc- the ideal of 2 to 3 children per Among its many duties related to tive health clinics could provide couple, a substantial influx of re- coordinating the War on Pov- millions of American women and sources into birth control serv- erty’s programs, the OEO was men with heretofore scarce or ices and the absence of standard- commissioned with introducing nonexistent medical services. ized consent protocols made the contraception and related educa- However, the increasing access environment ripe for coercion. tion programs to millions of un- to contraception overwhelmingly One of the most well-known derserved women. In 1965, benefited middle-class White cases of sterilization abuse was Protest in Los Angeles against coerced sterilizations at the about 450000 women had ac- women.45 Against the injunction that of the Relf sisters, aged 12 Women’s Hospital of the University of cess to family planning projects; to define themselves primarily as and 14, who were sterilized with- Southern California–Los Angeles by 1975, this number had breeders, mainstream feminists out consent in 1973 in Alabama County General Hospital, 1974. With jumped to 3.8 million.42 In 1971, framed their struggle for repro- in OEO-financed operations permission of the Los Angeles Times.

July 2005, Vol 95, No. 7 | American Journal of Public Health Stern | Peer Reviewed | Public Health Then and Now | 1133  PUBLIC HEALTH THEN AND NOW 

overseen by the Montgomery County General, Dr Bernard antipoverty programs, the greed Community Action Committee. Rosenfeld coauthored a report of big international corporations, When the Southern Poverty Law on sterilization abuse across the and the oppression of poor peo- Center sued on their behalf, it nation. At County General, he ple worldwide.57 was revealed that their mother, recorded the following dramatic Madrigal v Quilligan, which ul- who could not read, had unwit- increases during the period from timately pitted 10 sterilized tingly approved the procedures. July 1968 to July 1970: a 742% women against obstetricians at Believing she was authorizing increase in elective hysterec- County General, began in May birth control for her daughters tomies, a 470% increase in elec- 1978. The plaintiffs charged that in the form of Depo-Provera tive tubal ligations, and a 151% their civil and constitutional injections, she signed an “X” on increase in postdelivery tubal lig- rights to bear children had been what was actually a sterilization ations.52 Rosenfeld described a violated, and that between 1971 release.48 situation in which there was “lit- and 1974 they had been victims By the time the Relfs held a tle evidence of informed consent of unwanted operations: coerced press conference in 1973, by the patient,” where doctors into signing consent forms hours African American and Native were “selling” sterilizations “in a or minutes before or after labor, American women from across manner not unlike many other not told that the procedure was the South and Southwest were deceptive marketing practices.”53 irreversible, or simply sterilized coming forth with parallel allega- According to Rosenfeld, County without giving any consent.58 tions.49 When Relf v Weinberger General obstetricians instructed Antonia Hernández and Charles was heard in federal district residents to strong-arm vulnera- Nabarette of the Los Angeles court, Judge Gerhard Gesell con- ble patients into accepting tubal Center for Law and Justice repre- cluded that “an indefinite num- ligations, often packaging the op- sented the plaintiffs, all of whom ber of poor people have been im- eration as a chance to gain were low-income monolingual properly coerced into accepting a needed surgical training.54 Spanish speakers who had emi- sterilization operation under the Cognizant of what was tran- grated to California in their teens threat that variously supported spiring at County General, Mexi- from rural areas in Mexico in welfare benefits would be with- can American women in Los An- search of economic opportunity drawn unless they submitted,” geles began to organize and or to join relatives. and added that “the dividing line investigate, eventually locating Although they varied by age, between family planning and eu- 14 0 women who claimed they occupation, and number of chil- genics is murky.”50 Gesell esti- had been forcibly sterilized in dren, their stories were strikingly mated that over the past several medically unnecessary surger- similar. All of them had been ap- years, 100000 to 150000 low- ies.55 As with Puerto Ricans on proached about sterilization after income women had been steril- the East Coast, the sterilization having been in labor for several ized under federal programs.51 cases galvanized Mexican Ameri- hours and had endured difficult Unlike many of the African can feminists, who distinguished childbirths, eventually performed American women who filed suit themselves both from White by cesarean delivery.59 Their in the South, the plaintiffs in feminists, whose quest for abor- lawyers averred “these women Madrigal v Quilligan were neither tion rights often made them were in such a state of mind that welfare recipients nor on trial for oblivious to reproductive abuse, any consent which they may illegitimacy. Instead, they were and Mexican American national- have signed was not informed,” working-class migrant women ists, who frequently cast birth and that in 3 cases, no consent sterilized in a county hospital control as either superfluous to was obtained.60 Rebecca where obstetric residents were race and class or, more stri- Figueroa was falsely given the pressured to meet a quota of dently, as treason against the per- impression that she was submit- tubal ligations and where the petuation of the ethnic family ting to a reversible procedure. physicians at the top of the chain and nation.56 Mexican American Elena Orozco was told that her of command were partisan to feminists mobilized demonstra- hernia would be repaired only if racially slanted ideas about popu- tions against County General and she agreed to be sterilized, which lation control. In 1973, appalled formed the Committee to Stop she refused repeatedly, “until al- by the unethical behavior he wit- Forced Sterilization, which most the very last minute when nessed during his residency at linked sterilization to federal she was taken to be delivered.”61

1134 | Public Health Then and Now | Peer Reviewed | Stern American Journal of Public Health | July 2005, Vol 95, No. 7  PUBLIC HEALTH THEN AND NOW 

At no point after being admitted voice was marginalized and guidelines, including a 72-hour to County General in 1973 did drowned out against the other, waiting period between consent Guadalupe Acosta sign a consent mostly male, experts heard on and operation, a near morato- form.62 Dolores Madrigal did so the stand. rium on sterilization of persons after a medical assistant told her Hernández and Nabarette younger than 21 years of age, that her husband had already of- waived the option of a jury trial, and a signed statement of con- fered his signature, something placing adjudication in the sole sent preceded by a clear explana- that was patently untrue. Their hands of Judge Jesse Curtis. Al- tion that welfare benefits would accusations were supported by though Curtis acknowledged that not be terminated if the patient the affidavits of 7 additional the women had “suffered severe declined the procedure. These women, one of whom stated that emotional and physical stress be- guidelines officially took effect in a County General doctor told her cause of these operations,” he re- 1974, although persistent viola- after her cesarean delivery that fused to blame County General tions and inconsistencies in hos- “I had too many children” and physicians for what he called “a pitals across the country spurred that “having future children breakdown in communication over 50 organizations to meet in would be dangerous for me.”63 between the patients and the Washington, DC in 1977 to push Despite corroborating testi- doctors.”66 He found “no evi- for stricter enforcement and mony about sterilization abuse at dence of concerted or conspira- oversight by HEW.70 County General, the judge de- torial action” and, furthermore, Madrigal v Quilligan was one cided for the defendants, whom was persuaded by the defen- aspect of the federally funded he determined had acted in good dants’ contentions that they sterilization abuse that unfolded faith and intended no harm. “would not perform the opera- in the United States between the Only one key witness, Karen tion unless they were certain in mid-1960s and mid-1970s. Benker, spoke out against the their own mind that the patient Nonetheless, the language used doctors. Then a medical student understood the nature of the op- to disparage these women, in- and technician, Benker related eration and was requesting the deed to deprive them of their an entrenched system of forced procedure.” 67 Although Curtis human rights, had a much older sterilization based on stereotypes did not sanction neo-Malthusian origin. As early as the 1920s, of Mexicans as hyperbreeders theories, he stated that it was not California eugenicists such as and Mexican women as welfare objectionable for an obstetrician Goethe, Jordan, and Holmes mothers in waiting. She recalled to think that a tubal ligation asseverated that Mexicans were conversations in which Dr Ed- could improve a perceived over- irresponsible breeders who ward James Quilligan, the lead population problem, as long as flooded over the border in defendant and head of Obstetrics the physician did not try to “hordes” and undeservingly and Gynecology at County Gen- “overpower the will of his pa- sapped fiscal resources. In 1935, eral since 1969, stated, “poor mi- tients.”68 Curtis depicted the suit for example, Goethe wrote to nority women in L. A. County as a “clash of cultures,” and, rely- Harry H. Laughlin, superintend- were having too many babies; ing on a simplistic interpretation ent of the Eugenics Record Of- that it was a strain on society; of Mexican culture, suggested fice, “It is this high birthrate that and that it was good that they be that if the plaintiffs had not been makes Mexican peon immigra- sterilized.”64 She also testified naturally inclined toward such tion such a menace. Peons multi- that he boasted about a federal large families, their postpartum ply like rabbits.”71 grant for over $2 billion dollars sterilizations would have never In editorials, pamphlets, and he intended to use to show, in congealed into a legal case. personal correspondence, promi- his words, “how low we can cut Even though the plaintiffs lost, nent eugenicists foregrounded the birth rate of the Negro and Madrigal v Quilligan did have the “Mexican problem” as a dan- Mexican populations in Los An- major consequences for the for- ger to the state’s public and fis- geles County.”65 According to mulation of sterilization stipula- cal health. Moreover, during the Benker, sterilizations were partic- tions—most importantly, securing Great Depression, Popenoe ularly pushed on women with 2 a clause that consent forms be began to reconceptualize this as or more children who underwent bilingual.69 Now under many a “problem” not just of defective cesarean deliveries. Facing the watchful eyes, County General heredity but misguided parent- animosity of the judge, Benker’s began to comply with federal hood. In a 1934 study tracking

July 2005, Vol 95, No. 7 | American Journal of Public Health Stern | Peer Reviewed | Public Health Then and Now | 1135  PUBLIC HEALTH THEN AND NOW 

pregnant mothers be offered Whether in operations in state institutions or federally funded MSAFP (maternal serum alpha- county hospitals, most of those sterilized were the foreign born, fetoprotein) screening to assess the likelihood of Down syn- the working class, and young women deemed “unfit” drome, spina bifida, and neural “ to procreate or parent. tube defects. Rather than making such testing compulsory, this law 504 households that had re- IMPLICATIONS FOR mandates that genetic counselors ceived public aid, many of which CALIFORNIA’S PUBLIC inform patients of the availability were “producing children ” HEALTH PROGRAMS of MSAFP. As studies show, how- steadily at public expense,” ever, given societal pressure to Popenoe and a colleague re- The legacy of involuntary ster- use extant medical technologies ported that of all the groups, ilization lingers in California. It is in order to do the “best” for one’s Mexicans had the largest family no coincidence, for instance, that children, many women accede to size, a mean of 5.20 living chil- the Golden State was home to prenatal testing even if, for lin- dren.72 These kinds of parents, Proposition 187, which was guistic or cultural reasons, the im- however, rarely produced chil- passed by a majority of votes in plications of testing or positive di- dren of “superior quality”; much 1994 and strove to drastically re- agnosis are unclear.75 more common were “eugenically strict health, educational, and so- Focusing on California, 2 med- inferior” offspring. Popenoe rec- cial services to “illegal aliens.” Its ical anthropologists have de- ommended that every new char- intent and rhetoric strongly re- scribed scenarios in which Mexi- ity case be given contraceptive sembled that iterated by Califor- can-origin women are, usually instructions and materials, and nia eugenicists and the Depart- inadvertently, receiving incom- that, “beyond this, sterilization at ment of the Institutions in the plete or distorted information public expense [should] be pro- early 20th century in terms of about genetic screening and its vided for selected patients who who deserved access to health meanings.76 This situation is ex- desire it.”73 services during pregnancy (in acerbated by a dearth of minor- The Madrigal v Quilligan ster- this permutation, denial rather ity and bilingual genetic coun- ilizations were not directed by than the imposition of reproduc- selors trained and prepared to the Department of Institutions, tive control), who was allowed translate complex scientific and but they cannot be extracted to reproduce on American soil, technical information to diverse from the chronology of involun- and who should be deported. patient populations.77 However, tary sterilization in California, Discursively unoriginal, it tar- it is also related to genetic pro- particularly since they occurred geted Mexicans, who were por- fessionals’ desire to distance in Los Angeles, which, after the trayed as infectious hyperbreed- themselves from the coercive dissolution of the Eugenics Sec- ers, alien invaders, and vampires practices associated with eugen- tion of the Commonwealth Club threatening to bankrupt the ics, a psychological technique de- of California in 1935, overtook state. Because of its negation fined as “non-directiveness.” 78 San Francisco as the Pacific of basic rights to an entire class Even if motivated by noble in- West’s eugenic epicenter. Los of individuals, Proposition 187 tentions, attaining such neutrality Angeles was home to some of was swiftly contested in the is not only unrealizable, given the country’s most dynamic eu- courts and ruled unconstitu- that social values are embedded genic organizations, which in- tional in 1998.74 in medical institutions and deci- cluded physicians affiliated with If Proposition 187 demon- sions, but often frustrates pa- the University of Southern Cali- strates the perduring eugenic and tients, especially those from fornia hospital system. Whether fiscal logic of public health pro- newly arrived immigrant groups in operations in state institutions phylaxis, then California’s innova- who expect expert advice from or federally funded county hos- tive prenatal testing program genetic practitioners.79 pitals, most of those sterilized reveals the difficult ethical ques- With California at the fore- were the foreign born, the work- tions raised by contemporary in- front, the demographics of the ing class, and young women stances of public health genetics. United States are changing, and deemed “unfit” to procreate or In 1986, California was first state it is likely that within a century parent. to pass a law requiring that all Whites will no longer constitute

1136 | Public Health Then and Now | Peer Reviewed | Berridge and Loughlin American Journal of Public Health | June 2005, Vol 95, No. 6  PUBLIC HEALTH THEN AND NOW 

the nation’s racial/ethnic major- 1909–1979, Lisa M. Matocq, ed. The American Eugenics Movement and lating sterilization statistics and demo- ity. At the same time, genetic (Sacramento, Calif: Senate Publications; Virginia, 1900–1980,” PhD Disserta- graphic trends: (1) incomplete archival December 2003). tion, University of Virginia, 2000. access, including issues related to pa- and reproductive technologies tient confidentiality; (2) insufficient clar- 3. Author’s interview with Art Torres, 14 . Quoted in Stephen J. Gould, “Car- ity regarding the question of whether or are proliferating and, although November 17, 2003, San Francisco, rie Buck’s Daughter,” Constitutional not the official statistics include numer- not necessarily offering therapy Calif. Commentary 2 (2) (1985): 333; on ous “sent for sterilization only” cases, Buck v. Bell, also see Paul A. Lombardo, or cure, will generate informa- 4. These studies often focused on usually involving young women, not for- “Involuntary Sterilization in Virginia: poor rural White families. See Nicole mally committed to state institutions but tion about the probabilities of From Buck v. Bell to Poe v. Lynchburg,” Hahn Rafter, White Trash: The Eugenic interned for the sole purpose of repro- genetic diseases that, in turn, will Developments in Mental Health Law 3 (3) Family Studies, 1877–1919 (Boston: ductive surgery; (3) exclusion in the of- (1983): 13–21; Lombardo, “Medicine, need to be communicated in a Northeastern University Press, 1988). ficial statistics of sterilizations in state Eugenics, and the Supreme Court: From prisons, 600 of which had been per- culturally sensitive manner. This 5. Several weeks before California, Coercive Sterilization to Reproductive formed in San Quentin alone by 1941; is a great challenge for the 21st the state of Washington passed the sec- Freedom,” The Journal of Contemporary (4) formidable numbers of “voluntary” ond sterilization law in the country. See Health and Law Policy 13 (19 9 6): 1–25; century; as a crucial component sterilizations, primarily of women, who, Harry H. Laughlin, Eugenical Steriliza- also see Lawrence O. Gostin, Public at their own behest or that of relatives of tomorrow’s public and repro- tion in the United States (Chicago: Psy- Health Law: Power, Duty, Restraint or a physician, procured the operation ductive health, it can be ethically chopathic Laboratory of the Municipal (Berkeley: University of California Press, in a private setting. Undoubtedly, some Court, 1922). 2000). enhanced by awareness of the women sought out sterilization as a 6. Cited in Laughlin, Eugenical Steril- ways in which history weighs on 15. See Molly Ladd-Taylor, “Saving Ba- form of permanent birth control, but ization, 17; on Hatch, see Joel Braslow, bies and Sterilizing Mothers: Eugenics the fact that obstetricians affiliated with contemporary biomedicine and Mental Ills and Bodily Cures: Psychiatric and Welfare Politics in the Interwar California eugenics organizations car- ■ Treatment in the First Half of the Twenti- society. United States,” Social Politics 4 (1997): ried out some of these operations raises eth Century (Berkeley: University of Cali- 13 6–153. questions about the extent to which fornia Press, 1997). 16.Reilly, Surgical Solution, 97–101. they were voluntary, and, indeed, how 7. See “Sterilization in California In- to define voluntary or elective at this About the Author stitutions,” Sixth Biennial Report of the 17. See Edward J. Larson, Sex, Race, point in time. The author is with the Center for the His- Department of Institutions for the Year and Science: Eugenics in the Deep South 24. See Gosney and Popenoe, Twenty- tory of Medicine, the Department of Ob- Ending June 30, 1932 (Sacramento: Cal- (Baltimore: Johns Hopkins University Eight Years of Sterilization in California. stetrics and Gynecology, and the Program ifornia State Printing Office [CSPO], Press, 1995). in American Culture, University of Michi- 1932), 146–147. 18.Fred Hogue, “Social Eugenics,” Los 25. “Nationality,” Box 28, Folder 8, Pa- gan, Ann Arbor. pers of Ezra S. Gosney and the Human 8. Laughlin, Eugenical Steriliza- Angeles Times, July 5, 1936, 29. Requests for reprints should be sent to Betterment Foundation (ESGHBF), In- tion,18–19. Alexandra Minna Stern, PhD, Center for 19.Fred Hogue, “Social Eugenics,” Los stitute Archives (IA), California Institute the History of Medicine, 100 Simpson 9. Ibid, 19; also see F. O. Butler, “Ster- Angeles Times Sunday Magazine, March of Technology (CIT). Memorial Institute, 102 Observatory, ilization Procedure and Its Success in 9, 1941, 27. 26.Ibid. Percentages based on 1920 Mail Code 0725, University of Michi- California Institutions,” Third Biennial 20.Paul Popenoe and Roswell Hill census figures. See “Table E-7. White gan, Ann Arbor, MI 48109 (e-mail: Report of the Department of Institutions of Johnson, Applied Eugenics, 2nd ed. Population of Mexican Origin, for the [email protected]). the State of California, Two Years Ending (New York: The MacMillan Company, United States, Regions, Divisions, and This article was accepted August 23, June 30, 1926 (Sacramento: CSPO, 1933), 160–161. States: 1910 to 1930,” available at 2004. 192 6), 92–97. The “sexual perversion” www.census.gov/documents/population, aspect of the law was amended and clar- 21. Statistical Report of the Department accessed May 10, 2004. California’s ified with a 1923 statute that applied to of Institutions of the State of California, total population was 3 264 711, of Acknowledgments those “convicted of carnal abuse of a fe- Year Ending June 30, 1942 (Sacra- which Mexicans constituted 121 176. An earlier version of this article was pre- male under the age of ten years.” mento: CSPO, 1943), 98; California’s sented at the Spirit of 1848 History ses- Compulsory Sterilization Policies; Statisti- 27. “Norwalk Sterilizations,” place of 10.Braslow, Mental Ills, 56. sion at the 131st Annual Meeting of the cal Report of the Department of Institu- birth worksheet for females, Box 30, American Public Health Association; No- 11. Ezra S. Gosney and Paul Popenoe, tions of the State of California, Year End- Folder 12; “Norwalk Sterilizations,” vember 16–19, 2003; San Francisco, Sterilization for Human Betterment: A ing June 30, 1935 (Sacramento: CSPO, place of birth worksheet for males, Box Calif; it also draws from my forthcoming Summary of Results of 6,000 Operations 193 6). 30, Folder 13, ESGHBF, CIT, IA. Num- book Eugenic Nation: Faults and Frontiers in California, 1909–1929 (New York: 22. Figures derived from “US Maps bers and percentages derived from the of Better Breeding in Modern America MacMillan, 1929). This was followed Showing the States Having Sterilization Biennial Reports of the Department of (University of California Press). 9years later by Twenty-Eight Years of Laws in 1910, 1920, 1930, 1940,” Institutions from 1922 to 1930 (Sacra- I am grateful to Anne-Emanuelle Sterilization in California (Pasadena, Publication No. 5 (Princeton: Birthright, mento: CSPO). Birn, Natalia Molina, Howard Markel, 193 8). Inc., nd) in California’s Compulsory Ster- 28.Excerpt of “Nationality,” Box 28, Gabriela Arredondo, Elizabeth Fee, and 12.Philip R. Reilly, The Surgical Solu- ilization Policies; Clarence J. Gamble, Folder 8, ESGHBF, CIT, IA. 3 anonymous reviewers for their inci- tion: A History of Involuntary Steriliza- “Preventive Sterilization in 1948,” Jour- 29. See rough draft of “Twenty-Eight sive comments and suggestions. tion in the United States (Baltimore: nal of the American Medical Association Years of Human Sterilization,” Box 28, Johns Hopkins University Press, 1991), 141 (11) (1949): 773; Gamble, “Steril- Folder 8, ESGHBF, IA, CIT. Endnotes chap. 4. Many of these rulings were de- ization of the Mentally Deficient Under livered in the 1910s and prompted state State Laws,” American Journal of Mental 30. See unnamed patient records in 1. Art Torres to Edmund G. Brown, legislatures to reword and resubmit suc- Deficiency 51 (2) (1946): 164–169. Boxes 39–43, ESGHBF, IA, CIT. Also Jr., September 7, 1979, Legislative His- cessful sterilization statutes. Delaware was the only state that out- see Mike Anton, “Forced Sterilization tory, Assembly Bill 1204, Microfilm 3:3 paced California in per capita terms in Once Seen as a Path to a Better World,” 13. See Paul A. Lombardo, “Three (57), California State Archives (CSA). the 1930s, with a rate ranging between Los Angeles Times, July 16, 2003, A1. Generations, No Imbeciles: New Light about 80 and 100 sterilizations per 2. “Enrolled Bill Report,” August 31, on Buck v. Bell,” New York University 31.For an excellent analysis of gender 100 000 individuals. 1979, Legislative History, Assembly Bill Law Review 60 (1985): 30–62; on ster- and sterilization in California, especially 1204, Microfilm 3:3 (57), CSA; Califor- ilization in Virginia, also see Gregory 23. There are 4 key reasons for the at the Sonoma facility, see Wendy Kline, nia’s Compulsory Sterilization Policies, Michael Dorr, “Segregation’s Science: immense difficulty of accurately calcu- Building a Better Race: Gender, Sexuality,

July 2005, Vol 95, No. 7 | American Journal of Public Health Stern | Peer Reviewed | Public Health Then and Now | 1137  PUBLIC HEALTH THEN AND NOW 

and Eugenics From the Turn of the Cen- 45. See Linda Gordon, Woman’s Body, 59. “Madrigral v. Quilligan,” No. CV Aldama and Naomi Quiñonez (Bloom- tury to the Baby Boom (Berkeley: Univer- Woman’s Right: Birth Control in America, 74–2057-JWC, Report’s Transcript of ington: Indiana University Press, 2002), sity of California Press, 2001). 2nd ed. (New York: Penguin Books, Proceedings, Tuesday, May 30, 1978, 98–112; Dorothy Nelkin and Mark 1990). SA 230–240, Papers of Carlos G. Michaels, “Biological Categories and 32. See Alexandra Minna Stern, “The Vélez-Ibañez (CGVI), Sterilization Border Controls: The Revival of Eugen- Darker Side of the Golden State: Eu- 46. On abortion and mainstream femi- Archive (SA), Item 5, Chicano Studies ics in Anti-Immigration Rhetoric,” Inter- genic Sterilization in California,” in Cali- nism, see Ruth Rosen, The World Split Research Library (CSL), University of national Journal of Sociology and Social fornia’s Compulsory Sterilization Policies. Open: How the Modern Women’s Move- California at Los Angeles (UCLA). Policy 18: (5–6) (1998): 35–63. ment Changed America (New York: 33. “Background Paper” and “Steriliza- Viking, 2000). 60. Ibid, 12. 75. See Nancy Press and Carol H. tion Operations in California State Hos- Browner, “Why Women Say Yes to Pre- pitals, April 26, 1909 through June 30, 47. See Gordon, Woman’s Body, 61. Ibid, 19. natal Diagnosis,” Social Science and 1960,” in California’s Compulsory Steril- Woman’s Right. 62. Ibid, 12. Medicine 45 (7) (1997): 979–989; Bar- ization Policies. 48. See Angela Y. Davis, Women, Race, 63. Affidavit of DG, SA 110, CGVI, bara Katz Rothman, The Tentative Preg- 34. See Legislative History, Senate Bill & Class (New York: Vintage Books, SA, 5, CSL, UCLA. nancy: Prenatal Diagnosis and the Future 750, Microfilm 3:2(4); “Legislative 19 81), chap. 12; Jack Slater, “Steriliza- of Motherhood (New York: Penguin, Memorandum,” April 4, 1953, Legisla- tion: Newest Threat to the Poor,” Ebony 64. “Madrigral v. Quilligan,” No. CV 19 87). tive History, Assembly Bill 2683, Mi- (October 1973): 150–156. 74–2057-JWC, Report’s Transcript of 76. Carol H. Browner, H. Mabel Prelo- crofilm Reel 3:2 (10); Frank F. Tallman Proceedings, Tuesday, May 30, 1978, 49. Ibid; Shapiro, Population Control ran, Maria Christina Casado, Harold N. to Honorable Earl Warren, March 31, SA 230–240, CGVI, SA, 5, CSL, Politics; Nancy Ordover, American Eu- Bass, and Ann P. Walker, “Genetic 1953, Legislative History, Assembly Bill UCLA, p. 802. genics: Race, Queer Anatomy, and the Sci- Counseling Gone Awry: Miscommunica- 2683, Microfilm Reel 3:2(10), CSA. ence of Nationalism (Minneapolis: Uni- 65. Ibid, 797. tion Between Prenatal Genetic Service 35. See “Background Paper,” in Cali- versity of Minnesota Press, 2003). 66. Quoted in “Plaintiffs Lose Suit Providers and Mexican-Origin Clients,” fornia’s Compulsory Sterilization Policies. 50. Quoted in Shapiro, Population Con- Over 10 Sterilizations,” Los Angeles Social Science and Medicine 56 (2003): 36. See Elof Axel Carlson, The Unfit: trol Politics, 5. Times, July 1, 1978, clipping in CGVI, 1933–1946. A History of a Bad Idea (Cold Spring SA, 5, CSL, UCLA; Elena Rebéca 51. Ibid, 5. 77. See Boston Information Solutions, Harbor, NY: Cold Spring Harbor Labo- Gutiérrez, The Racial Politics of Repro- “National Society of Genetic Counselors, ratory Press, 2001); Diane B. Paul, Con- 52. A Health Research Group Study on duction: The Social Construction of Inc. Professional Status Survey 2002,” trolling Human Heredity: 1865 to the Surgical Sterilization: Present Abuses and Mexican-Origin Women’s Fertility, PhD December 2002, available at http:// Present (Atlantic Highlands, NJ: Human- Proposed Regulations (Washington, DC: dissertation, University of Michigan, www.nsgc.org/pdf/PSS_2002_2_22. ities Press, 1995). Health Research Group; 1973), 1. 1999, p. 212. pdf, accessed July 20, 2004. 37. See Johanna Schoen, Choice and 53. Ibid, 2. 67. Quoted in Gutiérrez, “The Racial 78. See Jon Weil, “Psychosocial Ge- Coercion: Birth Control, Sterilization, and 54. Ibid, 7. Politics of Reproduction,” 213; quoted netic Counseling in the Post-Nondirec- Abortion in Public Health and Welfare in Ainsworth, “Mother No More,” 5. tive Era: A Point of View,” Journal of 55. See Diane Ainsworth, “Mother No (Chapel Hill: University of North Car- Genetic Counseling 12 (3) (2003): More,” Reader: Los Angeles’ Free Weekly, 68. Ibid, 208. olina Press, 2005). 19 9–211. January 26, 1979, 4. 69. See Gutiérrez, “Policing Pregnant 38. See Julius Paul, “The Return of 79. See Ilana Mittman, William R. 56. See Virginia Espino, “‘Women Pilgrims,” 393. Punitive Sterilization Proposals: Current Crombleholme, James R. Green, and Sterilized as They Give Birth’: Forced Attacks on Illegitimacy and the AFDC 70.Shapiro, Population Control Politics, Mitchell S. Golbus, “Reproductive Ge- Sterilization and the Chicana Resistance Program,” Law & Society Review 3 (1) 137; Sterilization Abuse: A Task for the netic Counseling to Asian-Pacific and in the 1970s,” in Las Obreras: Chicana (1968): 77–106. Women’s Movement (Chicago Committee Latin American Immigrants,” Journal of Politics of Work and Family, ed. Vicki L. to End Sterilization Abuse, January Genetic Counseling 7 (1) (1998): 49–70. 39. See Schoen, Choice and Coercion, Ruiz and Chon Noreiga (Los Angeles: 1977); Daniel W. Sigelman, Sterilization chap. 2 and 3; Leslie J. Reagan, When UCLA Chicano Studies Research Center Abuse of the Nation’s Poor Under Medic- Abortion Was a Crime: Women, Medi- Publications, 2000), 65–82; also see aid and Other Federal Programs (Wash- cine, and Law in the United States, Vicki L. Ruiz, From Out of the Shadows: ington, DC: Health Research Group, 1867–1973 (Berkeley: University of Mexican Women in Twentieth-Century 19 81). California Press, 1997). America (New York: Oxford University 71. Charles M. Goethe, press release, Press, 1998), chap. 5. 40. See Elena R. Gutiérrez, “Policing March 21, 1935, C-4–6, Papers of ‘Pregnant Pilgrims’: Situating the Steril- 57. Committee to Stop Forced Steril- Harry H. Laughlin (HHL), Special Col- ization Abuse of Mexican-Origin ization, “Stop Forced Sterilization Now!” lections (SC), Truman State University Women in Los Angeles County,” in (Los Angeles, n.d.), 3. (TSU). Women, Health, and Nation: Canada and the United States since 1945, ed. 58. Also see Claudia Dreifus, “Steriliz- 72. Paul Popenoe and Ellen Morton Georgina Feldberg, Molly Ladd-Taylor, ing the Poor,” in Seizing Our Bodies: Williams, “Fecundity of Families Depen- Alison Li, and Kathryn McPherson The Politics of Women’s Health, ed. dent on Public Charity,” American Jour- (Montreal: McGill-Queen’s University C. Dreifus (New York: Vintage Books, nal of Sociology 40 (2) (1934): Press, 2003), 379–403 (citation from 1977), 105–120; Adelaida R. Del 214–220, quote from p. 214, Box 1, p. 381). Castillo, “Sterilization: An Overview,” Folder 6, ESGHBF, IA, CIT. and Carlos G. Vélez-Ibañez, “Se Me 73. Ibid, 220. 41. See Thomas M. Shapiro, Population Acabó La Canción: An Ethnography of Control Politics: Women, Sterilization, Non-Consenting Sterilizations Among 74. See Kent A. Ono and John M. and Reproductive Choice (Philadelphia: Mexican American Women in Los An- Sloop, Shifting Borders: Rhetoric, Immi- Temple University Press, 1985), 87. geles,” in Mexican American Women in gration, and California’s Proposition 187 42. Ibid, 113. the United States: Struggles Past and Pre- (Philadelphia: Temple University Press, sent, ed. Madgalena Mora and Adelaida 2002); Jonathan Xavier Inda, 43. Gutiérrez, “Policing ‘Pregnant Pil- R. Del Castillo (Los Angeles: University “Biopower, Reproduction, and the Mi- grims,’” 381. of California Chicano Studies Research grant Woman’s Body,” in In Decolonial 44.Shapiro, Population Control Poli- Center Publications, Occasional Paper Voices: Chicana and Chicano Cultural tics, 115. No. 2, 1980), 65–70, 71–94. Studies in the 21st Century, ed. Arturo J.

1138 | Public Health Then and Now | Peer Reviewed | Stern American Journal of Public Health | July 2005, Vol 95, No. 7  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

The Limits of Social Capital: Durkheim, Suicide, and Social Cohesion

Recent applications of so- | Howard I. Kushner, PhD, and Claire E. Sterk, PhD cial capital theories to population health often draw on PUBLIC HEALTH SCHOLARSHIP 4(p59) classic sociological theories protective strategy against these lation health.” Challengers for validation of the protec- often cites classic social theorists modernizing forces. However, warn that “an emphasis on social tive features of social cohe- to demonstrate the link between as other studies have shown, cohesion can be used to render sion and social integration. social capital—the collective civic Durkheim’s conceptualization of communities responsible for their Durkheim’s work on suicide value of social networks—and suicide and the interpretation of mortality and morbidity rates: has been cited as evidence population health. Kunitz1 the data were framed by his own a community-level version of that modern life disrupts so- showed that classic theory is biases and by those of his early ‘blaming the victim.’”4(p59) Recent cial cohesion and results in most often cited to authenticate 20th century contemporaries.9,10 research indicates that specific a greater risk of morbidity a current assertion rather than Social capital advocates have mortalities among working class and mortality—including self- to test the validity of a public made their debt to Durkheim ex- populations, even in wealthy destructive behaviors and health maxim. As a result, the plicit.11–16 Although social capital countries, show that increased suicide. We argue that a close work of the same theorist is has a variety of contested defini- social capital is unrelated to im- 4,17,18 4,22 reading of Durkheim’s evi- often used to support contradic- tions, there is general con- proved health. In their evalua- 1,2 23 dence supports the oppo- tory arguments. We examined sensus that the required condition, Muntaner et al. demon- site conclusion and that the the extent to which Durkheim’s tions for social capital include strated that social capital is much incidence of self-destructive claims about the link between so- the existence of community less important than economic and behaviors such as suicide is cial disintegration and suicide networks, civic engagement, social status for predicting infant often greatest among those have lent support to current as- civic identity, reciprocity, and and coronary disease mortality. with high levels of social in- sumptions that social capital is a trust.19 , 2 0 One of the most well Despite ongoing critique, the tegration. A reexamination protective factor in population known works, Putnam’s Bowling number of studies claiming a rela- of Durkheim’s data on fe- health.3,4 Durkheim tied modern Alone,20 identifies social association between social capital and male suicide and suicide urban life to declining birth rates, tions and networks, norms of rec- improved population health in the military suggests that we should be skeptical increasing alienation, and exac- iprocity, and trust as 3 key com- seems undiminished. In part, this about recent studies con- erbated gender role tensions, ponents of social capital. reflects a wider pressure on US necting improved popula- which, he believed, had negative Social capital constructs have public health practitioners to tion health to social capital. health consequences, evidenced had a great impact on recent ex- downplay class in favor of culture. (Am J Public Health. 2005; by increased suicide rates.5–8 aminations of population health, By contrast, studies that examine 95:1139–1143. doi:10.2105/ Durkheim distinguished be- particularly on studies concerned the role of class or institutional so- AJPH.2004.053314) tween egoistic, anomic, altruistic, with health disparities.21 As cial capital have shown that it is a and fatalistic suicide, broad clas- Kawachi et al. argued, citing Put- more powerful predictor of posi- sifications that reflect then-pre- nam, social capital is “the glue tive health outcomes than com- vailing theories of human behav- that holds society together.”14 (p57) munitarian social capital.2,23 ior. Dismissing altruistic and In this context, a growing body Drawing on Durkheim, fatalistic suicide as unimportant, of public health investigators hy- Kawachi et al.14 defined social he viewed egoistic suicide as a pothesize that diminished social capital as synonymous with social consequence of the deterioration capital contributes to an in- cohesion and linked it to health of social and familial bonds and creased risk for an array of ill- outcomes. Kawachi et al. cited linked anomic suicide to disillu- nesses, ranging from chronic Wolf and Bruhn,24 who exam- sionment and disappointment.8 heart disease and diabetes to de- ined the impact of the decline of His claims about suicide among pressive disorders and suicide. social cohesion on the 1600 resi- women and suicide in the mili- Others have challenged this dents of Roseto, Pennsylvania. In tary are emblematic of his asser- view, arguing that social capital the 1950s, death rates in this tion that increasing moderniza- theorists ignore class relations, as- small, close-knit Italian American tion and urbanization led to the suming instead that “social cohe- community were lower than in breakdown of social cohesion. sion rather than political change neighboring communities even He viewed social integration as a is the major determinant of popu- though there was no significant

July 2005, Vol 95, No. 7 | American Journal of Public Health Kushner and Sterk | Peer Reviewed | Critical Concepts for Reaching Populations at Risk | 1139  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

difference in risk factors. How- A review of his discussion on vided a measure of social pathol- in the family and the community, ever, as younger residents began female suicide and military sui- ogy. According to Durkheim, were more immune to suicide exploring employment outside cide shows the enormous dis- anomie and egoism resulted than men.8 Yet Durkheim’s asser- Roseto, social ties weakened. By tance that Durkheim traveled to from the collapse of traditional tion of the immunity of women the mid-1960s, “expensive auto- reconcile his theory with his evi- restraints, and thus their inci- to suicide owed more to his as- mobiles began to appear in the dence. This has not been evident dence could be used as an index sumptions about the socially dis- streets . . . families had joined because most social scientists for social pathology. In his view, integrative impact of urban life country clubs, [and] . . . occa- share 3 of Durkheim’s assump- the rate of anomic suicide mea- and modernity than it did to his sional visits to Atlantic City were tions. The first is the belief that sured alienation, whereas the data. Durkheim asserted that replaced by weekends in Las modernity breeds alienation and rate of egoistic suicide measured “mental illnesses go hand in Vegas and luxury cruises.” 24(p122) egocentrism. The second is the the decline of self-restraint. Al- hand with civilization” and that Wolf and Bruhn tied these be- assumption that women, as most truistic suicide, on the other insanity was more common “in haviors to increases in heart dis- socially integrated in family life, hand, reflected socially sanc- towns than the countryside, ease in Roseto. By contrast, the are the most protected against tioned self-sacrifice.8 Although and in large rather than small examination by Lynch and suicide. Finally, social integration the construct of altruistic suicide towns.”7(p215) Davey Smith25 revealed that the is assumed to be socially protec- makes theoretical sense, such In an 1888 essay entitled original empirical results were tive. The acceptance of these as- acts (heroism) were never re- “Suicide et natalité: étude de statis- weaker than often is claimed by sumptions among scholars can ported as suicides. There could tique morale,” Durkheim linked social capital experts and also partly be explained by their re- be almost no fatalistic suicides low birth rates to increased sui- were open to more plausible alliance on Durkheim’s definition because Durkheim claimed that cide rates.5 “A low birthrate led ternative interpretations. They and typology of suicide. In what “it has so little contemporary im- to the weakening of the fam- pointed out that the original in- follows we reexamine these as- portance and examples are so ily,”5(p462–463) and Durkheim vestigators had rather conserva- sumptions through an explo- hard to find . . . that it seems claimed those areas with the tive preconceptions of what con- ration of 2 issues that Durkheim useless to dwell upon it.”8(p276) least population growth experi- stitutes the “right” way to live addressed—women’s suicide and As a result, subsequent studies enced the highest rates of sui- and what formed “healthy” indi- suicide in the military—and 1 ignored fatalistic suicide.28 cide.5 Because, according to vidual, family, and community issue that he failed to take into Durkheim, the health of society relationships. Others26 have account—attempted suicide. First, DURKHEIM AND THE depended on the density of fami- pointed out that improvements in however, we must review how PUTATIVE IMMUNITY lies, women were expected to be health historically occurred inde- Durkheim constructed his defini- OF WOMEN mothers of many children. By ex- pendently of social capital. tion and typology of suicide. tension, he said, women were “The notion that social cohe- Durkheim’s definition and healthiest and least prone to sui- sion is related to the health of a DEFINITION AND typology of suicide reinforced his cide themselves to the extent population,” Kawachi et al. wrote, TYPOLOGY claim that the breakdown of tra- that they were subsumed in tra- “is hardly new. One-hundred ditional social order was the rea- ditional roles: “Woman is less years ago, Emile Durkheim Durkheim defined suicide as son for an increase in suicide. concerned than man in the civi- demonstrated that suicide rates “death resulting directly or indi- Durkheim pointed to the puta- lizing process,” Durkheim as- were higher in populations that rectly from a positive or negative tive low rates of female suicides, serted in 1893, “she participates were less cohesive”14 (p57) For act of the victim himself, which which he tied to women’s less in it and draws less benefit Durkheim, social cohesion, es- he knows will produce this re- greater social integration. In from it. She thus resembles cer- pecially traditional family life, sult.”8(p44) However, Durkheim’s no case did Durkheim view tain characteristics found in provided the best protection analysis relied on official suicide women’s suicide itself as a cate- primitive cultures.”7(p192) These against self-destructive behav- statistics that were collected gory for systematic analysis.30 presumptions alone go far in ex- ior.27,28 Nevertheless, a reading without regard to his definition. Instead as we demonstrate later, plaining why Durkheim assumed of Durkheim’s evidence sup- For instance, those who sacri- Durkheim’s classificatory system that women were “naturally” im- ports the opposite conclusion, ficed their lives for others were contributed to and sustained an mune to suicide. that is, that the incidence of sui- never recorded in official statis- underreporting of women’s com- Durkheim’s assertion in Le cide is greatest among those tics; those whose deaths resulted pleted suicides. Suicide that “in all the countries most subsumed in social groups. only “indirectly” from their acts Durkheim’s claim that social of the world, women commit sui- Durkheim’s data revealed that generally did not appear in the disintegration led to an increase cide less than men,” was based the highest suicide rates were statistics either.28,29 in suicide, especially among not only on the statistical data found among those who were Durkheim wanted to demon- women, was based on his belief of his predecessors, but also on most socially integrated.29 strate that the suicide rate pro- that women, because of their role their gendered assumptions.9(p471)

1140 | Critical Concepts for Reaching Populations at Risk | Peer Reviewed | Kushner and Sterk American Journal of Public Health | July 2005, Vol 95, No. 7  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

In explaining the immunity of were less “socially integrated.” must be considered suicidal be- found that women living in the women to suicide, Durkheim Even accepting the equivocal havior. Yet suicidologists since most socially integrated societies concluded that “being a more data that women completed sui- Durkheim have relied on statis- had a greater incidence of sui- instinctive creature than man, cide less frequently than men, tics that, by defining only com- cide than men. Johnson40 sug- woman has only to follow her the high rate of attempted suicide pleted suicide as suicide, have ef- gested that women most sub- instincts to find calmness and by women suggested that suici- fectively eliminated the majority merged in the family display peace.”8(p272) dal behavior was a common way of suicidal behavior from their the greatest female suicidal be- Durkheim’s definition of fatal- for women to express their pro- analysis of suicidal behavior. havior. Her views have been ism described the psychological found unhappiness.31–33 The pri- Women attempt suicide at a rate affirmed by recent reports that and social condition of many mary reason that female suicidal approximately 2.3 times greater the highest rates of suicide in women, perhaps the majority of behaviors have been underval- than that of men.29,34,35 Had the world are found among women who inhabit the globe ued is that explanations of the Durkheim included attempted rural Chinese women.41–43 Simi- today. He chose instead to define causes of suicide are almost al- suicides, women rather than men larly, Hasegawa44 found that women in traditional families as ways based on completed sui- would have emerged as the improved population health— socially integrated, despite the cides. Although Durkheim admit- group at greatest risk of self- declining infection rates and fact that, by any measure, most ted that attempted suicide fit destructive behavior. The data rising life expectancy—in Japan women’s lives actually more his definition of suicide as a be- on attempted suicide could have today can be traced to broaden- closely fit his definition of fatal- havior, he excluded it from his been used to demonstrate that ing of access to social resources ism, that is, an excessively regu- typology because attempted women were less content with for Japanese women at the be- lated existence, “with futures piti- suicide fell “short of actual their social roles than were men. ginning of the 20th century. lessly blocked and passions death.”8(p44) Estimates since the Thus, although suicidologists This reinforces the conclusion violently choked by oppressive early 19th century have indi- continue to refine their statisti- of historian Roger Lane,45 who discipline.”8(p276) Durkheim never cated that for every completed cal methods, they rarely have found that contrary to Durkheim’s questioned the supposition that suicide there have been at least questioned the assumption that assumptions, increases in suicide those most subsumed in the fam- 6to8attempts.29,34–36 only completed or successful rates were linked to social inte- ily (women and children) would Reliable data on an ex- suicides should constitute the gration. Lane found that as be most immune to suicide.5 panded definition of suicide database for suicidal behavior. 19th-century Philadelphia ur- Given this paradigm, suicide and were available to Durkheim. Although various ex post facto banized, its suicide rate grew integrative (women’s) behavior— For instance, beginning in explanations have been offered proportionally greater than its what Durkheim labeled fatalism— 1826 (until 1961) the French justifying the exclusion of at- homicide rate. Lane reasoned were opposites. Because social Criminal Justice Ministry pub- tempted suicides from measures that the increasing incidence integration was alleged to be the lished suicide statistics that of suicidal behavior, none of of suicide in late-19th-century cure for suicidal ideation, there made no distinction between these have any logical basis cities served as a barometer of was no way for Durkheim to sup- attempted and successful sui- other than one of convenience— social integration because sui- pose that suicide could be a fe- cides. In the 19th century these that is, completed suicides are cide, unlike homicide, indicated male behavior. The category of were published in the Annales readily available to researchers internalization of social anger.45 fatalistic suicide was constructed d’hygiène, which recorded the as part of national vital statistics Kunitz’s study1 on the effect of mainly for purposes of symmetry incidence of suicide (including, on death rates. In retrospect, it overintegration in the family (as contrasted with egoistic sui- but not separating out at- seems curious that suicide at- among Navajos in the south- cide) and because it would un- tempted suicides) by age and tempts were excluded from all western United States supports dercut his central claims about by sex. Although these statistics considerations of the incidence the views of Johnson and Lane. the role of modern urban life as suffered from the same weak- of suicide just as sophisticated Social relations within extended increasing the incidence of sui- nesses as data on completed statistical methodologies allow- Navajo families, Kunitz found, cide, Durkheim could never seri- suicides, there was no “objec- ing the inclusion of suicide at- often resulted in negative health ously examine the possibility that tive” reason why they could not tempts became available. outcomes, including significantly social integration could result in have been considered.28,37 The high rate of attempted higher rates of depression and suicide. The decision to exclude at- suicide among women alerts us self-destructive behaviors. Data available to Durkheim tempted suicide from considera- to the fact that submersion in reveal what he failed to examine. tion was peculiar because the en- the family provided women with SUICIDE IN THE MILITARY Those most subsumed in tradi- tire enterprise of the sociological no special protection from suici- tional social institutions were at study of suicide was aimed at de- dal behavior.38 Although his evi- The greatest challenge to the as great, if not greater, risk of scribing social behavior. Cer- dence was no more “value free” belief that social integration pro- suicidal behavior than those who tainly, attempting to kill oneself than Durkheim’s, Steinmetz39 vides protection from suicide,

July 2005, Vol 95, No. 7 | American Journal of Public Health Kushner and Sterk | Peer Reviewed | Critical Concepts for Reaching Populations at Risk | 1141  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

however, comes from Durkheim’s tions. The point is not that those scholars who challenge the phylactic impact of social capital own data. Official statistics con- women’s and soldiers’ socializa- relevance of social capital to pop- or social cohesion, public health sistently reported that the highest tion was the same. Rather, ulation health. investigators have been too ac- rates of suicide were in the mili- Durkheim’s description and dis- In Suicide, Durkheim provided cepting of Durkheim’s typology. tary. “It is a general fact in all cussions of military suicide fit a symmetrical typology of sui- Much of the current enthusiasm European countries,” wrote into his category of fatalism more cide in which altruism was con- for social capital as a core con- Durkheim, “that the suicidal apti- clearly than they fit into the cate- trasted with egoism and fatalism cept in suicide prevention rests tude of soldiers is much higher gory of altruism. Durkheim could with anomie.8 The impetus for on unexamined nostalgic and pa- than that of the civilian popula- not admit this because his theory Durkheim’s study, however, was triarchal assumptions, similar to tion of the same age.”8(p228) of the protective role of social in- a concern with what he per- those that informed Durkheim’s Durkheim’s definition of fatalistic tegration rested on his assertion ceived to be a breakdown in Suicide. The lesson here is that suicide as resulting “from exces- that modern urban life (anomy moral order, by which he meant we must remain skeptical about sive regulation,” whose “passions and egoism) were the killers. If what researchers today have current claims that improved [were] violently choked by op- military suicides were catego- labeled social capital. Thus, health outcomes and reduced pressive discipline,”8(p276) seemed rized as fatalistic, Durkheim Durkheim focused on increases mortality will result from in- to describe 19th-century military would have had to question in egoistic and anomic suicides creased submersion in commu- life perfectly. Durkheim’s typolog- his basic assumptions. Because because they provided a statisti- nity activity.4,22 Communities, ical definitions should have led the high rate of military suicide cally viable measure of the de- after all, are heterogeneous, and him to classify military suicide could not be attributed to cline of social capital. In his involvement alone may mean as fatalistic. modernity, Durkheim labeled work, altruistic suicide served less than the meaning that any Durkheim, however, over- it altruistic, which effectively mainly a rhetorical function. Fa- individual brings to an experi- looked the obvious inconsis- eliminated it from consideration. talistic suicide served as a de- ence. The quality of relationships tency that military suicide posed Because altruistic suicides were scriptor for suicides in traditional is always paramount, and partici- for his sociology by arbitrarily socially condoned forms of self- societies, because Durkheim was pation alone does not necessarily classifying military suicide as sacrifice, they were never re- faced with the issue that even in translate into acceptance, trust, “altruistic,” even though recorded as suicides. societies with abundant social or reciprocity. Moreover, the cur- ported military suicides could capital, individuals nevertheless rent enthusiasm for the health not be attributed to self-sacri- CONCLUSIONS killed themselves. But, as we benefits of social capital should fice.8 Given his familiarity with have shown, the data that not serve as an occasion to view suicide statistics, Durkheim Theoretical frameworks are es- Durkheim used was not linked it as a substitute for other forms must have known that those sential for improving population to his definition of what consti- of capital and status. Camouflag- who sacrificed their lives for health, but when adopted uncriti- tuted a suicide or to the typology ing the nostalgia that informs their military comrades in bat- cally they can have unintended he constructed. Moreover, sui- many of these claims with tle were never categorized as consequences. The recent enthu- cide attempts were excluded, metaphors of “social capital,” or suicides in any official statis- siasm for social capital is an ex- even though they fit Durkheim’s “social cohesion” should not con- tics. Indeed, to be reported as ample of a theory whose rhetoric definition. Women’s suicides ceal the traditional assumptions a suicide, a military death is often more liberating than its were made to fit the typology by and antiurban bias that may un- would have to have occurred application. The reason for this is assuming that they resulted from derpin such a project. outside a combat situation. As in part the foundation on which modernity and gender role stress. Although we are persuaded Besnard30(p339) pointed out, this paradigm rests: a theory of Nevertheless, Durkheim can be that significant contributions “The only ‘modern’ example social integration that relies on read as demonstrating that so- have been made by social capital given [by Durkheim] of altruistic Durkheim’s suicide typology. For cial integration can have negative scholars, we fear that a promiscu- suicide is military suicide, Durkheim, suicide rates were a health consequences. ous application of this approach which, nevertheless, could also marker for decreasing social cap- A critical reading of Durkheim’s can be harmful. This may ex- be interpreted in terms of exces- ital. The key conditions for social original text should make re- plain why studies on social capi- sive regulation” that comes from capital–community networks, searchers suspicious of current tal and health have resulted in “very strong social integration.” civic engagement, civic identity, claims that social capital is likely equivocal findings. Even advo- Given his assumption about reciprocity, and trust—appear im- to result in a reduction in mor- cates of a social capital approach the “nature of women” and the portant to health. Hence, numer- bidity and mortality, especially point out that the concept has its prophylactic impact of family life, ous studies have identified a pos- among constituents of communi- limitations. For instance, partici- Durkheim could not acknowl- itive association between social ties with little social and eco- pation in social activities may edge the parallels between sol- capital and population health.22 nomic power. Because it seems to result in engaging in unhealthy diers’ and women’s social situa- Less attention has been given to provide confirmation of the pro- behaviors, and the dynamics sur-

1142 | Critical Concepts for Reaching Populations at Risk | Peer Reviewed | Kushner and Sterk American Journal of Public Health | July 2005, Vol 95, No. 7  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

rounding reciprocity and trust philosophique France l’étranger. 1888;26: 22. Pearce N, Davey Smith G. Is social première moitié du XIXième siècle. Paris, may create power relations that 446–463. capital the key to inequalities in health? France: Hachette; 1984. Am J Public Health. 2003;93:122–129. allow some groups to gain from 6. Durkheim E. De la division du tra- 38. Kushner HI. American Suicide: vail social. 7th ed. Paris, France: Presses social capital while reducing ac- 23. Muntaner C, Lynch JW, Hillemeier M, A Psychocultural Exploration. New Universitaires de France; 1960. et al. Economic inequality, working-class Brunswick, NJ: Rutgers University Press; cess to resources for others. Con- 7. Durkheim E. The Division of Labor power, social capital, and cause-specific 19 91. mortality in wealthy countries. Int J tradictions and concerns as iden- in Society [1893]. Halls WD, trans. Lon- 39. Steinmetz SR. Suicide among prim- Health Serv. 2002;32:629–656. tified in this article warrant don, England: Macmillan; 1984. itive peoples. Am Anthropologist. 1894; continued research on the appli- 8. Durkheim E. Suicide: A Study in 24.Wolf S, Bruhn J. The Power of Clan: 7:55–60. The Influence of Human Relationships on Sociology. Spaulding J, Simpson G, trans. 40. Johnson KK. Durkheim revisited: cation of social capital to popula- Heart Disease. New Brunswick, NJ: Glencoe, Ill: The Free Press; 1951. why do women kill themselves? Suicide tion health as well as continued Transaction Publishers; 1992. 9. Kushner HI. Suicide, gender, and Life-Threatening Behav. 1979;9: public policy. the fear of modernity in nineteenth- 25. Lynch J, Davey Smith G. Rates and 145–153. states: reflections on the health of na- century medical and social thought. J Soc 41.Rosenthal E. Suicides reveal bitter tions. Int J Epidemiol. 2003;32: Hist. 19 93;26:461–490. roots of China’s rural life. New York 663–670. About the Authors 10. Kushner HI. Durkheim and the im- Times. January 24, 1999:1, 1. munity of women to suicide. In: Lester D, 26. Szreter S, Woolcock M. Health by Howard I. Kushner is with the Rollins 42. Suicide blights China’s women. ed. The Centennial of Durkheim’s Le Sui- association? Social capital, social theory, School of Public Health, the Graduate In- “BBC News.” Newscast, November 29, cide. Philadelphia, Pa: The Charles and the political economy of public stitute of Liberal Arts, and the Center for 2002; 9:16 GMT. Available at: http:/ Press; 1994:205–223. health. Int J Epidemiol. 2004;33: the Study of Health, Culture, and Society news.bbc.co.uk/2/hi/asia-pacific/ 650–667. at Emory University, Atlanta, Ga. Claire 11. Berkman LF, Glass T, Brissette I, 2526079.stm. Accessed March 15, E. Sterk is with the Department of Behav- Seeman TE. From social integration to 27. Baudelot C, Establet R. Suicide: 2005. ioral Sciences and Health Education, health: Durkheim in the new millen- l’évolution séculaire d’un fait social. 43. Bezlova A. Women Suicides Re- Rollins School of Public Health, Emory nium. Soc Sci Med. 2000;51:843–857. Economie statistique. 1984;168:59–70. flect Drudgery of Rural Life. IPS. Sep- University. 12. Berkman LF, Glass T. Social inte- 28. Baudelot C, Establet R. Durkheim tember 21, 1998. Available at: http:// Requests for reprints should be sent to gration, social networks, social support, et le suicide. 2nd ed. Paris, France: www.hartford-hwp.com/archives/55/ Howard I. Kushner, PhD, Rollins School of and health. In: Berkman LF, Kawachi I, Presses Universitaires de France; 1986. 353.html. Accessed March 15, 2005. Public Health, 5th floor, Emory University, eds. Social Epidemiology. New York, NY: 1518 Clifton Rd, NE, Atlanta, GA Oxford University Press; 2000: 29. Kushner HI. Women and suicidal 44. Hasegawa T. Japan: historical and 30322 (e-mail: [email protected]). 137–173. behavior: epidemiology, gender, and current dimensions of health and health This article was accepted December lethality in historical perspective. In: equity. In: Evans T, Whitehead M, 24, 2004. 13.Turner B. Social capital, inequality, Canetto SS, Lester D, eds. Women and Diderichsen F, Bhuiya A, Wirth M, eds. and health: the Durkheimian revival. Suicidal Behavior. New York, NY: Challenging Inequities in Health. From Soc Theory Health. 2003;1:4–20. Contributors Springer Publishing; 1995:11–34. Ethics to Action. New York, NY; Oxford 14 .Kawachi I, Kennedy BP, Lochner K. University Press; 2001:90–103. Both authors originated the study and 30. Besnard P. Durkheim et les Long live community: social capital as femmes ou le suicide inachevé. Rev 45. Lane R. Violent Death in the City: jointly conceptualized the ideas for this public health. Am Prospect. November– article. française sociologie. 1973;14:27–61. Suicide, Accident, and Murder in Nine- December 1997; 8(35):56–59. teenth Century Philadelphia. Cambridge, 31. Canetto SS. She died for love and 15.Kawachi I, Kennedy BP. Health Mass: Harvard University Press; 1979. he for glory: gender myths and suicidal Acknowledgments and social cohesion: why care about behavior. Omega. 19 92–1993;26:1-17. Research for this article was supported income inequality. BMJ. 19 97;314: by an Independent Scientist Award 1037–1040. 32. Clifton AK, Lee DE. Gender social- from the National Institute on Drug ization and women’s suicidal behaviors. 16.Kawachi I, Kennedy BP, Lochner K, Abuse (KO2DA0051; C. Sterk, Princi- In: Canetto SS, Lester D, eds. Women Prothrow-Smith D. Social capital, in- pal Investigator). and Suicidal Behavior. New York, NY: come inequality, and mortality. Am J The authors thank Susanna Elliott for Springer Publishing; 1995:61–70. Public Health. 19 97;87:1491–1498. assistance in preparing the article. 17. Durlauf S. Bowling alone: a review 33. Canetto SS. Epidemiology of essay. J Econ Behav Organ. 2002;47: women’s suicidal behavior. In: Canetto SS, References 259–273. Lester D, eds. Women and Suicidal Be- 1. Kunitz S. Social capital and health. havior. New York, NY: Springer Publish- 18. Whitehead M, Diderichsen F. So- Br Med Bull. 2004;69:61–73. ing; 1995:35–57. cial capital and health: tip-toeing through 2. Navarro VA. Critique of social cap- the minefield of evidence. Lancet. 2001; 34. Shneidman ES, Farberow NL. Sta- ital. Int J Health Sciences. 2002;32: 358:165–166. tistical comparisons between attempted 423–432. and committed suicides. In: Farberow NL, 19. Campbell C, Wood R, Kelly M. So- Shneidman ES, eds. The Cry for Help. 3. Van Poppel F, Day LH. A test of cial Capital and Health. London, En- New York, NY: McGraw-Hill; 1961: Durkheim’s theory of suicide—without gland: Health Education Authority; 24–37. committing the ecological fallacy. Am 1999. Sociol Rev. 1996;61:500–507. 35. Maris R. Pathways to Suicide: A 20.Putnam R. Bowling Alone: The Col- Survey of Self-Destructive Behavior. Balti- 4. Muntaner C, Lynch J. Income in- lapse and Revival of American Commu- more, Md: Johns Hopkins University equality and social coercion versus nity. New York, NY: Simon & Schuster; Press; 1981. class relations: a critique of Wilkinson’s 2000. neo-Durkheimian research program. Int 36. Hendin H. Suicide in America. New 21. Hean S, Cowley S, Forbes A, J Health Serv. 1999;29:59–81. York, NY: Norton; 1982. Griffith P, Maben J. The M-C-M cycle 5. Durkheim E. Suicide et natalité: and social capital. Soc Sci Med. 2002; 37. Chevalier L. Classes laborieuses et étude de statistique morale. Rev 56:1061–1072. classes dangereuses à Paris pendant la

July 2005, Vol 95, No. 7 | American Journal of Public Health Kushner and Sterk | Peer Reviewed | Critical Concepts for Reaching Populations at Risk | 1143  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

The Trouble With “MSM” and “WSW”: Erasure of the Sexual-Minority Person in Public Health Discourse

Men who have sex with | Rebecca M. Young, PhD, and Ilan H. Meyer, PhD men (MSM) and women who have sex with women THE BEHAVIORAL CATEGORY (WSW) are purportedly neu- and medical experts had initially tives—or because labels, once tral terms commonly used men who have sex with men has identified gay identity as a risk unleashed, tend to develop a in public health discourse. been used in HIV literature since for HIV/AIDS, a characterization dynamic of their own—the la- However, they are problem- at least 1990. The acronym that stigmatized lesbian, gay, and bels MSM and WSW have ac- atic because they obscure MSM, coined in 1994, signaled bisexual (LGB) populations and complished few if any of the social dimensions of sexu- the crystallization of a new con- confounded efforts to prevent aims that prompted them. Ironi- ality; undermine the self- cept.1,2 MSM and, more recently, HIV infection.4,5 WSW was in- cally, while MSM and WSW labeling of lesbian, gay, and WSW (women who have sex troduced shortly thereafter, in have succeeded in forcing a bisexual people; and do not with women) have since moved the context of controversy over conceptual shift in public health sufficiently describe varia- beyond the HIV literature to the meaning and salience of les- from identity-based to behav- tions in sexual behavior. become established in both re- bian identity in terms of HIV- iorally based notions of sexual- MSM and WSW often imply a lack of lesbian or search and health programming related risk behaviors, including ity, they have not generated gay identity and an absence for sexual-minority people. In sex with men. more complex approaches to of community, networks, part because the terms held Social construction is the sec- sexuality. While the behavioral and relationships in which the promise of reducing AIDS ond perspective driving the focus is useful in specific con- same-gender pairings mean stigma, which has been irra- adoption of MSM and WSW. texts, indiscriminate labeling of more than merely sexual tionally attached to gay men Social construction suggests that MSM and WSW is problematic, behavior. Overuse of the and lesbians, we, the authors, sexualities (like other social cate- as we will argue, on theoretical, terms MSM and WSW adds helped to promulgate these now- gories) are products of social political, ethical, and epidemio- to a history of scientific la- 3 familiar acronyms. But we have processes. A central tenet of so- logical grounds. We are con- beling of sexual minorities become increasingly troubled cial construction is that particular cerned that the ubiquitous use that reflects, and inadver- with the way these terms are sexual practices cannot be inter- of WSW and MSM (1) under- tently advances, heterosex- ist notions. used, especially when they dis- preted as though they carry fixed mines the self-determined sex- Public health profession- place rather than coincide with meanings. Thus, long before the ual identity of members of als should adopt more nu- information about sexual identity. terms MSM and WSW appeared, sexual-minority groups, in par- anced and culturally rele- The 10th anniversary of these social constructionists challenged ticular people of color; (2) de- vant language in discussing terms provides a good occasion the idea that sexualities are cate- flects attention from social di- members of sexual-minority to reflect on their meaning, gorical and rejected the use of mensions of sexuality that are groups. (Am J Public Health. utility, and limitations. sexual identity terms across dif- critical in understanding sexual 2005;95:1144–1149. doi:10. The argument for MSM and ferent cultural and historical con- health; and (3) obscures ele- 2105/AJPH.2004.046714) WSW seems to be driven by the texts. While the epidemiological ments of sexual behavior that convergence of 2 perspectives. perspective aimed to reduce gay are important for public health The first is an epidemiological and lesbian to what is thought of research and intervention. perspective: by using identity- as their necessary core—sexual There are important differ- free terms, epidemiologists behaviors that place individuals ences between MSM and WSW sought to avoid complex social at risk—the social construction- that we do not enumerate here and cultural connotations that, ist critique, with its origin in (e.g., MSM is used more fre- according to a strict biomedical gay and lesbian studies and quently). Because more health view, have little to do with epi- feminist and queer theory, research is conducted on sexual- demiological investigation of dis- seeks to do the opposite: it minority men than on sexual- eases. Accordingly, MSM was in- seeks more textured under- minority women, more of our troduced to reflect the idea that standings of sexuality that do examples are based on the use of behaviors, not identities, place not assume alignments among MSM. Still, we consider both individuals at risk for HIV infec- identity, behavior, and desire. terms because they share some tion, a particularly important Perhaps because of the con- underlying problems and impor- distinction given that scientific flict between these perspec- tant yet distinct social issues.

1144 | Critical Concepts for Reaching Populations at Risk | Peer Reviewed | Young and Meyer American Journal of Public Health | July 2005, Vol 95, No. 7  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

WHAT’S IN A NAME? central identity and that identities countries, “dedicated to achiev- category he implicitly racializes are fixed. In contrast to the no- ing equal rights for lesbian, gay, as White via the contrast with In their naming practices, sci- tions that identities are hierarchi- bisexual and transgendered “injection drug users, inner-city entists reflect the attitudes and cally ordered and that a salient (LGBT) people everywhere.”19 people of color, and persons constructions of their culture and identity could potentially sup- originally from poor countries also advance and promote such press or displace other identities, UNDERMINING in sub-Saharan Africa or the attitudes.6 The history of scien- intersectionality suggests that SELF-DETERMINED Caribbean.”21(p47) He further ex- tific nomenclature referring to identities are mutually constitu- SEXUAL IDENTITIES cludes gay from poor and sug- sexual minorities is a good exam- tive. The central tenets of this ap- gests that “males involved in ple. Richard von Krafft-Ebing’s proach are as follows: (1) no social Gay men, lesbians, queers, prostitution are almost univer- Psychopathia Sexualis introduced group is homogenous, (2) people two-spirited people, and men on sally poor, and it may be their the notion of “antipathic sexual must be located in terms of social the DL prefer to use their own poverty, rather than their sexual instinct,” used interchangeably structures that capture the power identity terms, but many contem- preference, that puts them at risk with “homo-sexual instinct” and relations implied by those struc- porary public health writers pre- of HIV infection. Many men in- referred to as “anomaly,” “abnor- tures, and (3) there are unique, fer the terms MSM and WSW, volved in homosexual prostitu- mal,” “tainted,” “neuropathic,” nonadditive effects of identifying ostensibly because these terms tion, particularly minority adoles- “degenerate,” “inverted,” and “in- with more than one social avoid assumptions about a singu- cents, do not necessarily identify jurious.”7 Later terms, which sim- group.14 Combined with social lar, misleadingly coherent gay as gay.”21(p47) With this juxtaposi- ilarly advanced a disease model, construction theory, an intersec- identity. In practice, however, tion, Farmer seems to suggest included “sexual psychopaths” tional understanding of sexual MSM and WSW often signify not that same-gender behavior and, indeed, “homosexuals.” identity suggests that sexual iden- a neutral stance on the question among poor men of color (espe- Sexual-minority people—like tities, while incorporating diver- of identity but a decided lack of cially youth) is sex work rather other oppressed groups—have sity, can be meaningful, powerful sexual-minority identity. More than sex for pleasure and is de- fought pitched battles over the forces for group affiliation and important, by implication, MSM void of identity and community; right to determine the names by political action.15 and WSW imply absence of com- same-gender behavior among which they will be known in Any term applied generally munity, social networks, and rela- White men is read as synony- public discourse. As stated by obscures important distinctions. tionships in which same-gender mous with gay identity. Epstein: “Power inheres in the However, terms such as sexual pairing is shared and supported. Compare these assumptions ability to name . . . what we call minorities and gay, lesbian, and We are also concerned with the with a recent ethnographic report ourselves has implications for po- LGBT (lesbian, gay, bisexual, ways the terms have been racial- on men at risk for HIV in Dakar, litical practice.”8(p241) In the 20th transgender) have acquired ized. As historian Allan Berube Senegal.22 While many of these century, many sexual-minority global resonance and political observed, “In the United States “men who have sex with men” men and women fiercely rejected and cultural meanings. The diffu- today, the dominant image of the are poor and engage in sex work, homosexual in favor of self-chosen sion of these terms has led not typical gay man is a white man the authors found that they have terms such as homophile, gay, les- to homogenization but to a multi- who is financially better off than indigenous sexual-minority iden- bian, and more recently DL plication of identities, as Western most everyone else.”20(p234) Just as tities that are differentiated and (down low), two-spirited, and and non-Western categories and gay and lesbian are often coded as socially meaningful. Senegalese transgender, among others. Even practices mix and reconfig- “White,” WSW and MSM often sexual-minority identities serve disparaging terms, for example, ure.16–18 Despite their limitations, implicitly refer to people of color, as a basis for social organization, queer, have been reworked, rede- there is ample evidence that the poor people, or racially and ethni- including, but not limited to, sex- fined, and reclaimed. terms gay, lesbian, and LGBT cally diverse groups outside the ual roles. The authors describe It is important to recognize are widely used alongside local perceived mainstream gay and ibbi as men who “tend to adopt that people vary in regard to self- terms, signifying liberationist ide- lesbian communities. feminine mannerism[s] and to be identity labels historically and ologies for sexual-minority peo- To understand how MSM is less dominant in sexual interac- cross-culturally and that, at any ple in many cultures.9–11 These read, it is important to examine tions,”22(p505) whereas yoos are time or place, self-identities vary terms have been used by such how explicit and implicit bound- men who “are generally the in- according to gender, culture, so- organizations as Amnesty Inter- aries are drawn around the cate- sertive partner.” They also stress cial class, ethnicity, and cohort, national, Human Rights Watch, gory gay. Consider, for example, that the categories have “more to among other factors.9–12 Cutting- and the International Lesbian a passage from Paul Farmer in do with social identity and status edge work on identity, such as and Gay Association. The latter which he claims that, in recent than with sexual practices.”22(p506) Crenshaw’s theory of intersec- is a federation of national and years, there have been fewer Despite their careful attention tionality,13 challenges the implica- local community-based groups HIV cases than predicted among to local sexual identities of men tion that individuals have one with representatives from 90 gay men in the United States, a in Senegal, the authors referred

July 2005, Vol 95, No. 7 | American Journal of Public Health Young and Meyer | Peer Reviewed | Critical Concepts for Reaching Populations at Risk | 1145  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

to them in the title and else- because most reported that they men and women as it is of peo- injection drug users relative to where as “men who have sex used an identity term to describe ple of color. Our point is that dis- other injection drug users.3,31 with men.” With this usage, the themselves (53% gay, 12% bisex- course on sexual minorities Many such women are typically rich information on identity is ual, 12% same-gender-loving, should attend to identity. To label thought of as situational WSW, lost, with MSM conveying trans- 12% homosexual). It is an ironic as MSM and WSW people who in reference to same-gender be- actional, decontextualized same- commentary on the pervasive- describe themselves as gay or havior that is purportedly en- gender acts. Ironically, applying ness of men who have sex with lesbian or use another identity gaged in purely for material or MSM in this way universalizes a men and MSM that the authors term is to deny their self-labeling social gain rather than for erotic culturally specific phenomenon of the Senegalese and BMSM and, by extension, their self- or romantic purposes. Such in much the same way that crit- articles provided a nuanced cul- determination. We believe that women are considered the most ics say does the term gay. tural analysis but resorted to a this is ethically indefensible. As appropriate subjects for an The same is true of depictions deliberately anticultural term in stated by Battle et al.: “Debates identity-free label such as WSW. of non-White populations within describing the groups of men about identity are not insignifi- To the contrary, however, Young the United States. For example, they studied. cant since they determine not and colleagues found that female Malebranche et al. studied Black In contrast, Munoz-Laboy24 only the public identity of the injection drug users in New York men, recruited primarily from observed complex sociosexual group, but also help to build and City described complex social Black gay organizations, to assess identities and meanings among solidify feelings of pride, empow- situations that included emotion- the impact of a social environ- US Latino men, noting that some erment, and political purpose ally and erotically invested rela- ment characterized by prejudice adopt and others reject sexuality among group members. Thus, tionships with their female part- on health services provided to as an important feature of their the politics of identity are an es- ners; most identified themselves these men.23 In qualitative inter- identity. We concur with the au- sential component of the politics as gay, a term often preferred by views, the respondents talked thor’s conclusion that “Latino of recognition and distribution.”11 these women over lesbian.32 about the challenges of being MSM is far from being a homo- To understand the consistently Black and gay men. They re- geneous sexual category, and, OBSCURING THE SOCIAL elevated rates of HIV among ported that conflicts among and as a framework, [the category MEANING OF SEXUALITY these women, it is important to displacement from communities MSM] is insufficient to capture consider precisely the sort of are important aspects of their the multidimensional aspects of We agree that sexual identity shared social experiences that lives. Regarding health services, Latino male bisexuality.”24(p75) is not sufficient for understand- WSW obscures. In fact, we be- one respondent poignantly Is MSM auseful term for de- ing the epidemiology and preven- lieve that WSW may thwart un- said: “When I go to a physi- scribing groups that eschew tion of HIV/AIDS or other health derstanding of risk for HIV cian’s office, and when I iden- prominent LGB categories? Much problems, but it is far from irrele- among these women because it tify myself as a gay person, part has been made of the fact that vant. In modern social studies of focuses attention on purely be- of that is looking for accept- men on the DL lead secret lives sexuality, distinctions have been havioral factors. But HIV risk ance from them, because I and do not consider themselves made among sexual identities, among WSW injectors cannot haven’t gotten it from my family, gay.25,26 But DL is not a behav- desires, and behaviors.29,30 In- be explained as a direct result of you know?”23(p100) ioral category that can be con- deed, understanding that these woman-to-woman sexual behav- In this context, where commu- veyed as MSM.AsFrank Leon dimensions of sexuality do not ior. The pattern only makes sense nities and identities explain the Roberts has put it, “DL is . . . always travel together in predict- when we understand that WSW substance of the concerns raised about performing a new identity able ways was one impetus for injectors also share a sexual- by the article, especially in a and embracing a hip-hop sensi- introducing the terms MSM and minority status that involves ex- qualitative piece that purports to bility [italics added].”27 DL func- WSW. But WSW and MSM can posure to discrimination and ex- explore meanings, it is important tions not as a nonidentity but as obscure critical inquiry into the clusion, relationship patterns, and to be vigilant in regard to named an alternative sexual identity and social meaning of sexuality. subcultural norms. These insights identities and communities. Yet, community denoting same-gender Thinking in flat behavioral terms can be extended to other health here too the authors referred to interest, masculine gender roles may lead us to ignore affiliation disparities, such as smoking or their respondents as BMSM distinct from the feminized sissy networks and communities that obesity, that also differ between (Black men who have sex with or faggot, Black racial/ethnic are important sources of informa- sexual-minority and other women men). This seems especially identity, and a dissociation from tion, norms, and values and that but that are not connected di- amiss because so many of the both White and Black middle- provide resources for health pro- rectly to, and cannot be ex- respondents belonged to Black class gay cultures.26–28 motion efforts. plained by, sexual behavior. gay organizations—for example, Certainly not all individuals A striking case in point is the For men, too, the narrow focus the explicitly named “New York experience sexual identity as study of elevated HIV rates on sexual behavior reflected in State Black Gay Network”—and salient. This is as true of White among sexual-minority female the use of MSM clouds under-

1146 | Critical Concepts for Reaching Populations at Risk | Peer Reviewed | Young and Meyer American Journal of Public Health | July 2005, Vol 95, No. 7  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

standing of HIV (and other specific sexual practices between HIV prevention strategies.”24(p58) discussion among colleagues. In health concerns). We use as one women.35,36 While important This led Munoz-Laboy24(p59) to our analysis of current usages of example the EXPLORE project, knowledge has been gained advocate that “we . . . move be- MSM and WSW, we hope it is an HIV prevention trial aimed about the risk of sexually trans- yond MSM ” so that we gain a clear that the task of naming is at MSM.33,34 A majority of the mitted infections (including HIV) more nuanced understanding of challenging and is thoroughly en- 4295 men who participated in among sexual-minority women sexuality. With such attention to gulfed in the substantive context EXPLORE were apparently re- from analyses of data collected identity and labels, for example, of the text. It is the intellectual cruited from predominantly gay for other purposes, specific same- Black men who identify as gay, responsibility of writers to disen- venues, and many enrolled in the gender practices are rarely as- DL, or who claim no sexual iden- tangle meanings within contexts study for altruistic reasons such sessed and reported.31,37 Studies tity would require different HIV and to carefully choose the terms as “helping stop the AIDS epi- on the specific sexual practices prevention approaches. that best fit their purpose. De- demic” and “giving to their com- of women are critical if models Similarly, “top” and “bottom,” spite our reluctance to offer solu- munity.”33(p928) The authors sug- are to be built that combine bio- to denote sexual roles, and “bare- tions to the labeling question, our gested that attention to sexual logical plausibility with empirical back,” to denote sex without con- article reveals at least 4 princi- self-identification, affiliation with information on associations be- doms, are part of a sex culture ples: (1) we view MSM and the gay communities, and per- tween sexual practices and inci- and connote meanings as well as WSW as lowest denominator ceived community norms are im- dence rates of sexually transmit- behaviors that are associated terms that tell us little about risks portant factors in HIV prevented infections. Without such data, with HIV risk and are relevant to for HIV/AIDS or any other dis- tion. Despite this recommendation, public health professionals can- HIV prevention.39,40 These terms ease; (2) at the most general the authors referred to respon- not provide sexual-minority and others could be more useful level, we prefer terms, such as dents as MSM and provided no women with meaningful harm than MSM in public health re- sexual minorities, that allow for information on self-identification reduction information. search and intervention in that sociocultural and political con- or sociosexual affiliations. This Similarly, gay and bisexual they reveal more nuanced infor- texts; (3) in more specific con- exemplifies a missed opportu- men organize sexual behavior in mation about sexuality, identity, texts, we prefer local terms that nity for public health research a variety of ways that MSM does and risk for HIV infection. respect the self-identifications of to more fully describe sociocul- not convey. As noted by Ayala: the populations in question; and tural factors related to HIV CONCLUSIONS (4) when relevant to the research prevention. HIV prevention has become question, we would report the synonymous with condom use and condom use has become MSM and WSW have become full range of identity terms repre- BEHAVIORAL TERMS solely about anal sex. . . . This ubiquitous terms in public health sented in samples, and, when dis- THAT SAY LITTLE focus on anal sex and condom discourse but have failed to live cussing individuals, we would use is reductionist; it narrows ABOUT BEHAVIOR the sexual possibilities for gay up to their promise. We do not use the terms they use. and bisexual men of color. It advocate the demise of MSM It has also been suggested to Purportedly, one of the great- also limits HIV prevention mes- and WSW, but we believe that, us that some investigators might sages in the media, as well as est advantages of WSW and the individual- and group-level a decade after their introduction, prefer MSM or WSW because MSM is that unlike lesbian and interventions conducted by the terms have become institution- the terms allow important but gay, they are anchored in con- [community-based organiza- alized and risk inattentive usage. potentially controversial research tions]. Within this rubric there crete behaviors that are more rel- is little room for discussing, un- Readers of an earlier version to fly under the radar of social evant than identity terms to epi- derstanding, or promoting other of this article—having been con- conservatives who want to block demiological investigations. MSM sexual options and choices vinced by arguments against research on sexual minorities. apart from anal sex and con- and WSW have often been un- dom use.38(p8) MSM and WSW—were frus- We disagree. First, the strategy derstood as stand-ins for pre- trated that we did not provide a does not work. Many grants on sumed risk behaviors. With this Ignoring identity in HIV pre- list of acceptable terms and us- the infamous list of the National usage, researchers ignore the im- vention efforts can be perilous, ages. We continue to balk at Institutes of Health–funded stud- portant task of describing actual because sexual identities may that task. We believe that the so- ies that were targeted by the Tra- sexual behaviors, even though provide important clues for pub- lution resides not in discovering ditional Values Coalition as a this information has greater rele- lic health prevention efforts. In better terminology but in adopt- “waste of tax-payer’s money”41 vance to public health. his study of Latino sexuality, ing a more critical and reflective used WSW or MSM, not the For example, reports about the Munoz-Laboy noted: “The prob- stance in selecting the appropri- identity terms that are suppos- risks of sexually transmitted in- lem with the MSM category is ate terms for particular popula- edly risky. Second, even if this fections involved in woman-to- that many men do not identify tions and contexts. strategy did keep our work from woman sex typically fail to pro- with the label, which leads to Rather than offer a menu of being attacked, it is perilous be- vide any information regarding their increased alienation from terms, our aim here is to open a cause the terms we use are not

July 2005, Vol 95, No. 7 | American Journal of Public Health Young and Meyer | Peer Reviewed | Critical Concepts for Reaching Populations at Risk | 1147  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

merely a matter of semantics but lumbia University, New York, NY. Ilan H. tity: the limits of social constructionism. 22. Niang CI, Tapsoba P, Weiss E, et al. are referents for important con- Meyer is with the Department of So- In Stein E, ed. Forms of Desire: Sexual ‘It’s raining stones’: stigma, violence and ciomedical Sciences, Mailman School of Orientation and the Social Constructionist HIV vulnerability among men who structs. Using inappropriate con- Public Health, Columbia University. Controversy. New York, NY: Routledge; have sex with men in Dakar, Senegal. structs can compromise the in- Requests for reprints should be sent 19 92:239–293. Cult Health Sexuality. 2003;5:499–512. tegrity of our work. to Ilan H. Meyer, PhD, Columbia Uni- 9. Altman D. Global Sex. Chicago, Ill: 23. Malebranche DJ, Peterson JL, versity, Mailman School of Public University of Chicago Press; 2001. Fullilove RE, Stackhouse RW. Race and We recognize that MSM and Health, Department of Sociomedical 10.Adam BD, Duyvendak JW, Krowel sexual identity: perceptions about med- WSW may also have strategic Sciences, 722 W 168th St, 9th Floor, A. The Global Emergence of Gay and ical culture and healthcare among black New York, NY 10032 (e-mail: im15@ appeal for international work, Lesbian Politics: National Imprints of a men who have sex with men. J Natl columbia.edu). Worldwide Movement. Philadelphia, Pa: Med Assoc. 2004;96:97–107. especially in instances in which This article was accepted December Temple University Press; 1999. 24. Munoz-Laboy M. Beyond “MSM”: “gay and lesbian” work may be 24, 2004. 11. Battle J, Cohen CJ, Warren D, sexual desire among bisexually-active blocked. Here, too, we caution Fergerson G, Audam S. Say It Loud, I’m Latino men in New York City. Sexuali- that this strategy is risky because Contributors Black and I’m Proud: Black Pride Survey ties. 2004;7:55–80. it may reinforce the position of The authors conceived and wrote the 2000. Washington, DC: Policy Institute 25. Vargas A. HIV-positive, without a article jointly. of the National Gay and Lesbian Task clue: black men’s hidden sex lives im- local conservatives who portray Force; 2002. periling female partners. Washington minority sexualities as Western, Acknowledgments 12.Ponse B. The social construction of Post. August 4, 2003:B1. foreign, and corrupt. Contempo- Work on this article was supported in part identity and its meanings within the les- 26. Denizet-Lewis B. Double lives rary work on sexuality and by the National Institute on Drug Abuse bian subculture. In: Nardi PM, Schnei- on the down-low. New York Times Mag- der BE, eds. Social Perspectives in Les- azine. August 3, 2003:28. human rights shows that local (grant R03 DA14399-01); (Rebecca M. Young, principal investigator). bian and Gay Studies: A Reader. New 27.Roberts FL. Beyond the down-low: struggles over the meaning and We thank Juan Battle, Sherry Deren, York, NY: Routledge; 1998:246–260. the East Bay’s black scene provides a legitimacy of particular sexual and Sally Cooper for their insightful 13.Crenshaw K. Mapping the margins: safe space for all. Available at: http:// forms are often cloaked in the comments on an earlier version of this intersectionality, identity politics, and www.youthoutlook.org/stories/2003/ article. We also thank the article’s editor, violence against women of color. In: 07/15. Accessed August 13, 2003. language of “tradition” versus Michael Ross, and the anonymous re- Crenshaw KN, Gotanda GP, Thomas K, 28.Boykin K. Anatomy of a media “modern corruption” or “local” viewers for helpful and challenging com- eds. Critical Race Theory. New York, NY: frenzy. Available at: http://content.gay. versus “Western” sexual norms ments throughout the review and revi- New Press; 1996:357–383. com/channels/news/boykin/boykin_36. sion process. and behaviors.42 In such a con- 14 . Stewart AJ, McDermott C. Gender html. Accessed August 13, 2003. psychology. Annu Rev Psychol. 2004;55: 29.Vance CS. Social construction the- text, MSM and WSW may inad- References 519–544. ory: problems in the history of sexuality. vertently undermine local strug- 1. Glick M, Muzyka BC, Salkin LM, 15. Glenn NE. The social construction In: van Kooten Niekerk A, van der gles for sexual rights. Lurie D. Necrotizing ulcerative periodon- and institutionalization of gender and Meer T, eds. Homosexuality, Which race. In: Marx Ferree M, Lorber J, Homosexuality. Amsterdam, the Nether- We have argued that use of titis: a marker for immune deterioration and a predictor for the diagnosis of Hess BB, eds. Revisioning Gender. lands: An Dekker/Schorer; 1989: reductive labels is unethical AIDS. J Periodontol. 1994;65:393–397. Walnut Creek, Calif: AltaMira Press; 13–34. 2000:3–43. because it denies the right of 2. Dowsett GW. Reaching men who 30. Laumann EO, Gagnon JH, Michael identity to members of sexual- have sex with men in Australia—an 16.Donham DL. Freeing South Africa: RT, Michaels S. The Social Organization the “modernization” of male-male sexu- of Sexuality: Sexual Practices in the minority groups whose margin- overview of AIDS education, community intervention and community attach- ality in Soweto. Cult Anthropol. 1998; United States. Chicago, Ill: University alization and mistreatment in ment strategies. Aust J Soc Issues. 1980; 13:3–21. of Chicago Press; 1994. medical settings have been 25:186–198. 17.Brooks P, Bocahut L. Woubi Cheri 31.Friedman SR, Ompad D, Maslow C, amply documented and to 3. Young RM, Friedman SR, Case P, [videotape; French with English subti- et al. HIV prevalence, social marginal- Asencio MW, Clatts M. Women injec- tles]. France/Ivory Coast; 1998. ization, risk behaviors, and high-risk whom we have the responsibil- sexual and injection networks among 43,44 tion drug users who have sex with 18.Swarr AL, Nagar R. Dismantling ity of heightened sensitivity. women exhibit increased HIV infection assumptions: interrogating “lesbian” young women injectors who have sex Sensitivity in discussing the and risk behaviors. J Drug Issues. 2000; struggles for identity and survival in with women. Am J Public Health. 2003; 93:902–906. health needs of members of 30:499–524. India and South Africa. Signs. 2004;29: 491–516. 32. Young RM, Case P, Friedman SR. sexual-minority groups requires 4. Treichler P. How to Have Theory in an Epidemic: Cultural Chronicles of AIDS. 19. International Lesbian and Gay As- Exploring an HIV paradox: an ethnog- neither avoiding identity simply Durham, NC: Duke University Press; sociation. About ILGA: working to raphy of sexual minority women injec- because it is complex nor re- 1999. achieve equal rights on a global scale tors. J Lesbian Stud. In press. since 1978. Available at: http://www. 33. Koblin BA, Chesney MA, Husnik MJ, treating to yet another generic 5. Herek GM, Glunt EK. An epidemic ilga.org/index.asp. Accessed April 30, et al. High-risk behaviors among men of stigma—public reactions to AIDS. Am term. Instead, researchers 2004. who have sex with men in 6 US cities: Psychol. 1988;43:886–891. should aim for a deeper under- 20. Berube A. How gay stays white baseline data from the EXPLORE study. 6. Gergen KJ. Psychological science in and what kind of white it stays. In: Ras- Am J Public Health. 2003;93:926–932. standing of variations in the a postmodern context. Am Psychol. mussen BB, Klinenberg E, Nexica IJ, 34. Chesney MA, Koblin BA, Barresi MJ, 2001;56:803–813. meaning of sexual identity and Wray M, eds. The Making and Unmak- et al. An individually tailored interven- community. 7. von Krafft-Ebing R. Psychopathia ing of Whiteness. Durham, NC: Duke tion for HIV prevention: baseline data Sexualis, With Especial Reference to the University Press; 2001:234. from the EXPLORE study. Am J Public Antipathic Sexual Instinct: A Medico- 21.Farmer P. Pathologies of Power: Health. 2003;93:933–938. About the Authors Forensic Study. Brooklyn, NY: Physicians Health, Human Rights, and the New War 35. Bauer GR, Welles SL. Beyond as- Rebecca Young is with the Department of and Surgeons Book Co; 1930. on the Poor. Berkeley, Calif: University sumptions of negligible risk: sexually Women’s Studies, Barnard College, Co- 8. Epstein S. Gay politics, ethnic iden- of California Press; 2003. transmitted diseases and women who

1148 | Critical Concepts for Reaching Populations at Risk | Peer Reviewed | Young and Meyer American Journal of Public Health | July 2005, Vol 95, No. 7  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

have sex with women. Am J Public lence and risk behaviors. Am J Public in gay men in New York City. J Psychol 42. Narayan U. Undoing the “package Health. 2001;91:1282–1286. Health. 19 95;85:1366–1371. Hum Sexuality. 2000;12(3):43–61. picture” of cultures. Signs. 2000;25: 36.Fethers K, Marks C, Mindel A, 40.Yep GA, Lovaas KE, Pagnosis AV. 10 83–1086. 38.Ayala G. Reconceiving the whole: Estcourt CS. Sexually transmitted infec- The case of “riding bareback”: sexual 43. O’Hanlan KA, Cabaj RP, Schatz B, a framework for HIV prevention and tions and risk behaviors in women who practices and the paradoxes of identity Lock J, Nemrow P. A review of the gay men of color. Paper presented at: have sex with women. Sex Transm Infec- in the era of AIDS. J Homosex. 2002; medical consequences of homophobia MSM of Color Summit, May 2003, tions. 2000;76:345–349. 42(4):1–14. with suggestions for resolution. J Gay Los Angeles, Calif. 37. Bevier PJ, Chiasson MA, Heffernan 41.Brainard J. NIH begins review of Lesbian Med Assoc. 19 97;1:25–39. RT, Castro KG. Women at a sexually 39.Wegesin DJ, Meyer-Bahlburg HFL. studies that were questioned at a con- 44. Lewin S, Meyer IH. Torture, ill- transmitted disease clinic who reported Top/bottom self-label, anal sex prac- gressional hearing. Chronicle of Higher treatment and sexual identity. Lancet. same-sex contact: their HIV seropreva- tices, HIV risk and gender role identity Education. November 7, 2003:24. 2001;358:1899–1900.

July 2005, Vol 95, No. 7 | American Journal of Public Health Hart et al. | Peer Reviewed | Critical Concepts for Reaching Populations at Risk | 1149  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

Rural Definitions for Health Policy and Research

The term “rural” suggests | L. Gary Hart, PhD, Eric H. Larson, PhD, and Denise M. Lishner, MSW many things to many people, such as agricultural landscapes, isolation, small towns, THE UNITED STATES HAS periences. The term suggests pas- ately to the situation at hand can and low population density. evolved from a rural agricultural toral landscapes, unique demo- we discern differences in health However, defining “rural” society to a society dominated by graphic structures and settlement care concerns and outcomes for health policy and research its urban population. Depending patterns, isolation, low popula- across rural areas and between purposes requires research- on which definition is used, tion density, extractive economic rural and urban locales. The defi- ers and policy analysts to roughly 20% of the US popula- activities, and distinct sociocultu- nition of rurality used for one specify which aspects of ru- tion resides within rural areas. ral milieus. But these aspects of purpose may be inappropriate or rality are most relevant to the 1 topic at hand and then select Approximately three fourths of rurality fail to completely define inadequate for another. an appropriate definition. the nation’s counties are rural, as “rural.” For example, rural cul- 1 Rural and urban taxonomies is 75% of its landmass. While the tures can exist in urban places. WHEN IS RURAL NOT SO often do not discuss impor- rural population is in the minor- Only a small fraction of the rural RURAL? tant demographic, cultural, ity, it is the size of France’s total population is involved in farm- and economic differences (rural and urban) population. ing, and towns range from tens Rural and urban taxonomies, across rural places—differ- As important as the rural pop- of thousands to a handful of resi- researchers, policy analysts, and ences that have major im- ulation and its resources are to dents. The proximity of rural legislation generally view all rural plications for policy and the nation, there is considerable areas to urban cores and services areas as uniform in character. research. Factors such as ge- confusion as to exactly what rural may range from a few miles to However, there are, in fact, huge ographic scale and region means and where rural popula- hundreds of miles. Generations variations in the demography, also must be considered. Several useful rural tax- tions reside. We will discuss of rural sociologists, demogra- economics, culture, and environ- onomies are discussed and defining rural and why it is im- phers, and geographers have mental characteristics of different 2,3 compared in this article. Care- portant to do so in the context of struggled with these concepts. rural places. Large rural towns ful attention to the definition health care policy and research. Despite the theoretical limita- that are not too distant from of “rural” is required for ef- tions of the concept of rurality, it larger metropolitan areas often fectively targeting policy and WHAT DOES RURAL LOOK is very useful as a practical ana- have more in common with met- research aimed at improving LIKE? lytic and policy tool. Common ropolitan areas than they do with the health of rural Americans. definitions of rurality are the remote and isolated small towns. (Am J Public Health. 2005;95: Although many policymakers, basis for many policy decisions, By treating these diverse types of 1149–1155. doi:10.2105/AJPH. researchers, and policy analysts including criteria for the alloca- rural cities and towns and the 2004.042432) would prefer one standardized, tion of the nation’s limited re- problems they confront similarly, all-purpose definition, “rural” is sources. It is important to specify policy analysts may fail to iden- a multifaceted concept about which aspects of rurality are rele- tify each site’s distinct health care which there is no universal vant to the phenomenon being concerns and effective methods agreement. Defining rurality can examined and then use a defini- for resolving those problems. Ac- be elusive and frequently relies tion that captures those elements. cess to medical specialists and on stereotypes and personal ex- Only by defining “rural” appropri- surgical services is a case in point.

The absence of certain services that residents of counties with gender.16 While there are many problems at the local level. As a in a small place is expected. The larger numbers of workers who common threads between urban result, policies may fail to include lack of such services in a larger commute out of the county and clinical medicine and its rural appropriate intrarural targeting. rural place might be construed who travel more than 30 minutes cousin, there are many substan- Rural and urban taxonomies as a critical provider shortage. each way to reach their care pro- tive differences.7,8,17,18 have usually been developed A small rural town’s population viders received substantially lower based on population size, density, base may only support 1 or 2 levels of health resources. Access HOW DO ACCURATE proximity, degree of urbaniza- generalist physicians and a nurse to proximate services for care DEFINITIONS HELP? tion, adjacency and relationship practitioner or physician assistant. often makes the difference be- to a metropolitan area, principal A larger rural town, whose geo- tween life and death.10 The federal government defines economic activity, economic and graphic service area may include The environment in which “rural” in a variety of ways. The trade relationships, and work the small town, may serve as a re- rural physicians and other provid- Office of Management and Bud- commutes. An appropriate gional center for accessing spe- ers practice also differs enor- get’s (OMB) definition of metro- rural and urban taxonomy cialists and surgeons. Health plan- mously both across rural areas politan and nonmetropolitan pop- should (1) measure something ning, recruitment and retention, and between rural and urban ulations and the Census Bureau’s explicit and meaningful; (2) be and identifying and optimizing areas.11–13 Physicians who practice definition of rural and urban fail replicable; (3) be derived from the supply and mix of providers in smaller and more remote rural to identify the same populations available, high-quality data; (4) be are going to be different in each towns practice in a medical care as rural. When the 2 definitions quantifiable and not subjective, and place.1,4 delivery system characterized by were cross-tabulated for the 2000 (5) have on-the-ground validity. financially vulnerable medical or- census, 72% of the population To some extent, all definitions HOW IS RURAL ganizations, small populations, was classified as both metropolitan will either underbound or over- DIFFERENT? long distances to specialists and (OMB definition) and urban (Cen- bound rurality. Some large coun- tertiary hospitals, longer practice sus Bureau definition), while 10% ties, for example, have large On average, rural populations hours, lack of collegial support, was classified as nonmetropolitan cities and less densely settled have relatively more elderly peo- limited access to advanced tech- and rural (Figure 1). However, areas that may be considered ple and children, higher unem- nologies, and relatively high fixed nearly 18% of the nation’s popula- rural in terms of economic activi- ployment and underemployment costs per delivered service. This tion was divided between the 2 ties, landscape, and service level. rates, and lower population den- milieu creates especially difficult taxonomies: 11% were metropoli- However, because of the pres- sity with higher percentages of circumstances for rural providers tan but rural, and 7% were non- ence of a large urban core the poor, uninsured, and underin- and populations.14 Rural physician metropolitan but urban. Depend- entire county is often considered sured residents. Rural populations practice concerns—patient pri- ing on how the categories are urban. In this case, “rural” is are more vulnerable than their vacy, clinical adaptations in the combined, the rural population being underbounded—areas that urban counterparts to economic absence of nearby specialists, gen- can vary from 10% to 28% of the might reasonably be called rural downturns because of their con- eralist scarcities, quality assurance nation’s total (i.e., a population of are actually being classified as centrated economic specializa- programs, compliance with the 29–79 million). Research findings urban. At the same time, “urban” tion. Other unique circumstances Health Insurance Portability and and policies may appear to con- is being overbounded. A certain include longer travel distances to— Accountability Act of 1996 regu- flict when those findings and poli- amount of overbounding and un- and higher costs associated with— lations, and continuing medical cies are based on different rural derbounding is inherent to any needed health care services; disec- education—are different from definitions and populations. The definition of rurality; the re- onomies of scale; high rates of those of their large city contempo- use of noncongruent definitions searcher must simply be aware fixed overhead per-patient rev- raries, differences that have a po- of rural may result in markedly of this problem when evaluating enue; fewer health care providers tential impact on health out- different conclusions and policy data across the rural and urban and a greater emphasis on gener- comes. For example, studies have implications. dimension.1(p15) alists; health care facilities with shown substantial differences be- Another problem associated Because numerous taxonomies limited scopes of service; econom- tween rural and urban physicians with defining “rural” is that con- have been used to categorize the ically fragile hospitals with high in clinical prenatal and intra- ventional definitions use a single rural/urban continuum, we ex- closure rates; greater dependency partum practice styles for similar rural classification and thereby amined the 4 that are most often on Medicare and Medicaid reim- low-risk patients, without appar- fail to differentiate categories of applied (Table 1). bursement; higher rates of chronic ent differences in outcome,15 and rurality. Rural areas are not ho- diseases; and different clinical that physician attitudes regarding mogeneous across the nation, OMB Metropolitan and practice behaviors, practice physician-assisted suicide vary and aggregating rural areas of Nonmetropolitan Taxonomy arrangements, and reimbursement dramatically by rural or urban differing sizes and levels of re- The federal government most levels.5–8 Hong and Kindig9 found practice location and practitioner moteness may obscure emerging frequently uses the county-based

1150 | Critical Concepts for Reaching Populations at Risk | Peer Reviewed | Hart et al. American Journal of Public Health | July 2005, Vol 95, No. 7  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

the largest urban settlement within the county. To qualify as adjacent to a metropolitan county, a nonmetropolitan county must share a boundary with a metropolitan county and must meet a minimum work commuting threshold.1 The UICs’ use of the size of the largest town in a county is as a taxonomic criterion. The largest town, as used for health care purposes, is associated with the likelihood of local availability of hospitals, clinics, and specialty services. While the codes are often used for research, they are infrequently used in federal and state policies. In 2003, the UICs were updated in accordance with 2000 census data.

Census Bureau Rural and Urban Taxonomy

Note. “Metropolitan” and “nonmetropolitan” are Office of Management and Budget terminology; “urban” and “rural” are Census Bureau The Census Bureau partitions terminology. urban areas into urbanized areas a n=202000000. and urban clusters. The same bn=30000000. cn=20000000. census tract–based criteria are dn=29000000. used for both; however, the urbanized areas have cores with FIGURE 1—Comparison of metropolitan and nonmetropolitan classifications with urban and rural populations of 50 000 or more, classifications, by proportion of 2000 US population (N=281421898): 2000 Census Bureau data. and the urban clusters have cores with populations that range from 2500 to 49999. All OMB metropolitan and non- was measured by commuting to urban cores. The metropolitan other areas are designated as metropolitan classifications as work. The United States has and nonmetropolitan taxonomy rural. The nation has more than policy tools. These county-based 1090 metropolitan counties and was most recently updated in 65 000 census tracts that are definitions are the foundation for 2052 nonmetropolitan counties 2003 in accordance with the made up of blocks and block other, more detailed taxonomies (674 micropolitan and 1378 non- 2000 census data. groups. In 2000, 59 million resi- and are used when determining core) that have (according to dents—21% of the US popula- eligibility and reimbursement lev- 2002 census data) 239 million US Department of Agriculture tion—were deemed rural by the els for more than 30 federal pro- metropolitan and 49 million non- Economic Research Service Census Bureau taxonomy. The grams, including Medicare reim- metropolitan residents, of whom Urban Influence Codes Census Bureau’s rural and urban bursement levels, the Medicare 29 million lived in micropolitan The Urban Influence Codes taxonomy is the source of much Incentive Payment program, and counties and 20 million lived in (UIC) taxonomy is a county-based of the available demographic programs designed to ameliorate noncore counties. Micropolitan definition that builds on the OMB and economic data. A weakness provider shortages in rural areas.4 counties are those nonmetropoli- metropolitan and nonmetropoli- of this system with regard to Metropolitan areas were de- tan counties with a rural cluster tan dichotomy. Counties are clas- health care policy is the paucity fined in 2003 as central counties with a population of 10000 or sified into 9 groups: 2 metropoli- of health-related data at the cen- with 1 or more urbanized areas more. Noncore counties are the tan and 7 nonmetropolitan. The sus tract level. The Census Bu- (cities with a population greater residual. The most significant nonmetropolitan counties are reau and others often aggregate than or equal to 50000) and problem with this taxonomy is grouped according to their adja- urban clusters with urbanized- outlying counties that are eco- that county boundaries both cency and nonadjacency to met- area data. Depending on the nomically tied to the core, which overbound and underbound their ropolitan counties and the size of purpose at hand, this may be

July 2005, Vol 95, No. 7 | American Journal of Public Health Hart et al. | Peer Reviewed | Critical Concepts for Reaching Populations at Risk | 1151  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

TABLE 1—Comparison of Commonly Applied Rural Taxonomiesa

Geographical Unit Characteristics Strengths Weaknesses

OMB Metropolitan/ County This OMB definition is used extensively in federal Useful for a general definition of rural status. Substantial underbounding of rurality in Nonmetropolitan policy. Significant 2003 update with 2000 The methodology and county assignments many large metropolitan counties. census data. Counties are assigned as were significantly changed in 2003. The large size of counties often metropolitan (n=1090) or nonmetropolitan Underlying geographic unit (county) is obscures intracounty differences. (n=2052). Nonmetropolitan counties are very stable over time. now designated as micropolitan or noncore based on the presence of an urban cluster (areas with a population less than 50000 but greater than 2500 people) with a population of 10000 or more. In 2000, 17.4% of the US population resided in nonmetropolitan counties. ERS Urban Influence County Metropolitan counties are grouped into 2 Use of largest city criterion, which differentiates Overbounding and underbounding occurs Codes categories based on size of urban counties with several small towns from just as in OMB metropolitan and population. Nonmetropolitan counties those with 1 or 2 large towns for grouping nonmetropolitan categories. Does not are grouped into 7 categories based on nonmetropolitan counties, may be better differentiate metropolitan counties size of largest city in county and adjacency/ than RUCA method for suggesting level as well as does RUCA.The large size nonadjacency to a metropolitan county. of locally available services.Adjacency of counties often obscures intracounty Updated with significant changes associated criteria may be suggestive of degree of differences. with the OMB metropolitan/nonmetropolitan economic integration with metropolitan definition changes in 2003 on the basis of county. 2000 census data. Census Bureau Rural Census tract Census Bureau definition of rurality based on Helps reduce problems of underbounding Data other than census data are infrequently and Urban census tracts. Rural census tracts are and overbounding associated with collected by census tract. Difficult to those outside of urbanized areas and county-based terminology. apply to health data that are often clusters with populations of 2500 or collected at the county or zip code greater that do not have substantial area levels. Lack of familiarity of most commuting data users with census tract geography In 2000, 21% of the US population lived and terminology. Not stable across in rural areas. census years—there were substantial changes for the 2000 census. RUCA Census tract Multitiered taxonomy developed by University Use of work commuting data strongly Difficult to apply to health data of Washington and the Economic Research differentiates rural areas according to that are often collected at the Service, with funding from the Federal their economic integration with urban county or zip code area levels. Office of Rural Health Policy and the areas and other rural areas.Very sensitive Will not be stable over time. Economic Research Service. Uses census to demographic change.The structure of Complex structure of codes not commuting data to classify census tracts the codes allows for many levels of easy to master for casual users. on the basis of geography and work generalization—from 2 groups (rural and commuting flows between places. urban) to 33. Updated in spring 2005 with 2000 census data. RUCA–zip code US Postal Multitiered system developed by University of Use of commuting data strongly differentiates Complex structure of codes is not easy approximation Service zip Washington, with funding from the Office rural areas according to their economic to master for casual users.The code areas of Rural Health Policy. Census work integration with urban areas and other rural underlying geographic unit is commuting data are used to classify census areas.Very sensitive to demographic subject to some change by the tracts on the basis of geography and work change.The structure of the codes allows US Postal Service across time. commuting activity between places. for many levels of generalization—from 2 Approximates the census tract RUCA codes groups (rural/urban) to 33. Use of the zip for 2000 zip codes. code unit makes them useful with a wide Updated in spring 2005 with 2000 census data variety of data collected at that level, and 2004 zip codes. including health data.

Source. aAdapted from Larson and Hart.1 Note. OMB=Office of Management and Budget; RUCA=Rural–Urban Commuting Area.

1152 | Critical Concepts for Reaching Populations at Risk | Peer Reviewed | Hart et al. American Journal of Public Health | July 2005, Vol 95, No. 7  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

misleading for rural health plemented by a zip code–based for research purposes, because it have important policy conse- policymakers. For example, a version. There are more than has a very different meaning for quences. For example, research- town with a population of 3000 30000 zip code areas. demographers and geogra- ers and analysts who examine in a very remote area is consid- RUCAs range from the core phers.20 There also are many data across years, when different ered urban under the Census areas of urbanized areas to iso- rural and urban definitions de- definitions were in place, need to Bureau definition, but that same lated small rural places, where veloped by the states for various be aware of these changes and town is often nonmetropolitan the population is less than 2500 geographic scales. For an intro- adjust result analyses and inter- under the OMB definition. and where there is no meaning- duction to older rural and urban pretations accordingly. (For more ful work commuting to urban- taxonomies, see Hewitt.21 Other detailed information about meth- Rural/Urban Commuting-Area ized areas. While the zip code taxonomies that lend themselves odological changes in the Census, Taxonomy version of the RUCAs is slightly to use with the rural and urban see the US Census Bureau and A recently developed taxon- less precise than the census tract taxonomies include the new ERS Web sites at http://www. omy uses census tract–level de- version, the RUCA zip codes county-based amenity index.22 census.gov and http://www.ers. mographic and work-commuting have an advantage in the health Other schemes regionalize usda.gov, and see Slifkin, data to define 33 categories of field because they can be used the nation or individual states for Randolph, and Ricketts.24–26 ) rural and urban census tracts.19 with zip code health-related data. diverse uses, for example, ambu- While it is beyond the scope of The Rural–Urban Commuting The RUCAs are widely used for latory care utilization via the this article to describe all of the Areas (RUCAs) were developed policy and research purposes national Primary Care Service concerns associated with new and are maintained by the Uni- (e.g., by the Centers for Medicare Areas.23 The federal government methodologies, 2 are most versity of Washington Rural and Medicaid Services and many has used taxonomies and mea- noteworthy. Health Research Center and the researchers). RUCAs can identify sures to allocate resources to USDA Economic Research Ser- the rural portions of metropolitan rural and urban areas. In these 1. There is some confusion about vice (ERS), with the support of counties and the urban portions schemes, factors such as physi- a new OMB metropolitan and the federal Health Resource of nonmetropolitan counties. cian-to-population ratios, infant nonmetropolitan taxonomy and Service Administration’s RUCAs are flexible and can be mortality rate, poverty, and resi- category: micropolitan—an Office of Rural Health Policy grouped in many ways to suit par- dent age are used to rate geo- urban cluster with a popula- and the ERS. (For more infor- ticular analytic or policy purposes. graphic units (combinations of tion that ranges from 10000 mation about RUCAs, see For example, there is a tool that counties, census tracts, facilities, to 49999. While some of the http://www.fammed.washington. provides the road mileage and the populations, etc.) and to delin- designation criteria have un- edu/wwamirhrc and http:// travel time along the fastest route eate those places and populations dergone subtle changes, the www.ers.usda.gov.) between each zip code area and most in need of federal health micropolitan counties have The RUCA categories are the nearest edge of a core in an care resources. These methods historically been designated as based on the size of settlements urbanized area and the closest (e.g., Health Professional Short- nonmetropolitan. Unfortu- and towns as delineated by the large rural city. When this tool is age Areas) have significant flaws, nately, the term micropolitan Census Bureau and the functional used with the RUCA codes, users and efforts are being made to has led some to think of these relationships between places as can identify highly isolated “fron- substantially revise them. counties as being urban or measured by tract-level work- tier” areas—counties with 6 or metropolitan in nature. How- commuting data. For example, a fewer persons per square mile—in How Have OMB and Census ever, changing the terminol- small town where the majority of a more precise manner than with Bureau Methodologies ogy does not make these commuting is to a large city is previous definitions. The RUCA Changed After the 2000 counties any more or any less distinguished from a similarly taxonomy was updated in the Census? rural and urban than they sized town where there is com- spring of 2005. Despite the common assump- were before—historically rural, muting connectivity primarily to tion that Census Bureau and albeit larger rural towns/cities. other small towns. Because 33 Other Taxonomies OMB methodologies change little Micropolitan counties could categories can be unwieldy, the Common taxonomies that between decennial censuses, just as well have been titled codes were designed to be aggre- have been designed for related about a quarter of the census macrorural or large rural gated in various ways that high- purposes include (1) ERS’s tract boundaries changed be- counties. A general problem light different aspects of connec- Rural–Urban Continuum Codes, tween the 1990 and 2000 cen- with the creation of so many tivity, rural and urban settlement, (2) ERS’s Economic Typology of suses, and the number of coun- taxonomies is that they take and isolation, aspects that facili- Nonmetropolitan Counties, and ties designated as metropolitan on a life of their own and are tate better program intervention (3) frontier areas, which is a by OMB in 2003 based on often used without consider- targeting. The census tract ver- crude measure at best. The term 2000 census data increased by ing the suitability or meaning sion of the RUCAs has been sup- “frontier” is a problematic term 27%. Many of these changes of the category.27

July 2005, Vol 95, No. 7 | American Journal of Public Health Hart et al. | Peer Reviewed | Critical Concepts for Reaching Populations at Risk | 1153  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

2. The 2003 update of the met- the demographic and economic policymakers and legislators 46.4 per 100000 population. ropolitan and nonmetropolitan heterogeneity that often exists often do not understand rural When we examined these same taxonomy resulted in a net within counties, can weaken the variability and diversity or the data with the census tract version gain of 253 counties in the meaningfulness of policy analy- methods for making these dis- of the RUCA taxonomy, we metropolitan ranks (now ses. Both the strengths and weak- tinctions. Third, self-interest often found a much lower ratio of 1090 counties). This also re- nesses of any given definition are prevails, wherein people advo- 38.5 per 100000 population— sulted in a net loss of 7.3 mil- strongly rooted in the underlying cate greater selectivity and more 17% lower. For resource alloca- lion residents who would have geographic unit used in the defi- effective targeting as long as they tion purposes, where money is been counted, in accordance nition.1 As already noted, some do not lose anything in the pro- spent is clearly influenced by with the older definition, as degree of overbounding and un- cess—regardless of what they how that locale is defined. A re- nonmetropolitan. Whether or derbounding is inherent in any may also gain. Finally, in some cent study of acute myocardial in- not the rural population is definition of rurality. It is impor- cases the availability of data at farction that used zip code–based shrinking depends on how the tant to consider which way the different geographic scales dic- RUCAs10 found substantial rural question is asked. As rural “error” goes when evaluating data tates the geographic unit that is and urban and intrarural differ- counties and cities grow, they and policy.1 The more mixing of used in policies. ences in the use of needed initial are designated as being met- diverse groups within units of anal- hospital services, where a previ- ropolitan and urban. Thus, be- ysis, the more difficult it is to show WHY SHOULD WE CARE? ous county-based study found lit- cause these rural populations real differences between groups. tle difference.33 (For a compre- are no longer counted as non- Rural data from federal surveil- Definitions of rural are the hensive explanation of the policy metropolitan or rural, it ap- lance systems and surveys are ex- basis for targeting resources to consequences of rural definitions, pears that the nation’s rural tremely limited,18 and funds for underserved rural populations. see Hewitt.21) Health care re- population is shrinking or rural surveys are scarce, both of If the only outcome of clarifying searchers focus great attention stagnant. However, according which impede rural health re- the definition were an improved and time on statistical methodolo- to Dr Calvin Beale (senior de- search and policy analysis. Better mechanism for funneling health gies; however, geographical meth- mographer, ERS), if the 1993 rural health research methods care to where it is needed most, odologies are often neglected.34 nonmetropolitan definition is and tools are needed to produce the clarification would be well Expert geographic consultation held constant, the overall pop- meaningful findings. Substantial worth the effort. Because there should be sought when determin- ulation change between 1990 progress has been made recently are 50 to 60 million rural resi- ing the most appropriate geo- and 2000 shows an 11% in- in data procurement and methods dents in the nation, decisions graphical unit and rural definition crease compared with a 13% because of focused funding from about resource use have signifi- to use in a given analysis. increase for the nation (writ- the Health Resource and Service cant ramifications in terms of ten communication, March Administration’s Office of Rural the dollars spent and the well- CONCLUSIONS 2004). Nevertheless, some Health Policy, the Bureau of being of rural populations. Inap- rural areas are experiencing Health Professions, and the Bu- propriate definitions may bias Deciding which rural defini- population loss.28 reau of Primary Health Care. To research findings and policy tion to apply to an area depends maximize the utility of these new analyses and may result in dif- on the purpose at hand, the GEOGRAPHIC SCALE AND methods, they must be widely ferent conclusions than those availability of data, and the ap- DATA AVAILABILITY disseminated to state offices of that are based on another unit propriate and available taxon- rural health, primary care officers, of analysis (often called the omy. There is no perfect rural Another problem associated and researchers and analysts.29–31 modifiable unit problem).32 The definition that meets all pur- with rural health research in- more we aggregate different poses. Researchers must be de- volves the geographical level of WHY ARE DIFFERENTIATED types of rural areas, the less we liberate and insightful when available data. Typical units used LEVELS OF RURALITY NOT can pinpoint localized health defining rural and when apply- for the collection of health and GENERALLY USED? care and delivery problems at ing the appropriate definition demographic data include states, the state, region, county, town, and its associated taxonomy to counties, municipalities, census Federal and state policies tend or zip code levels. program targeting, intervention, tracts, and zip codes. The county to treat rural areas as a single en- We examined the 2000 Amer- and research. It is recommended is a convenient and frequently tity for several reasons. First, the ican Medical Association Master- that researchers familiarize used unit of analysis, and many political process often requires file data on the nation’s physi- themselves with various rural health-related data are collected that a significant coalition be cian distribution and found that definitions and geographic meth- at this level. However, the large formed to pass rural-related legis- the most remote UIC subgroup odologies and then carefully geographic size of counties, and lation, and it is more expedient of counties had a generalist weigh the pros and cons of the failure to distinguish between to lump than to divide. Second, physician–to-population ratio of available definitions. Defining

1154 | Critical Concepts for Reaching Populations at Risk | Peer Reviewed | Hart et al. American Journal of Public Health | July 2005, Vol 95, No. 7  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

rural and urban must be a meth- tion and Health Care Challenges in Rural 19. Morrill R, Cronmartie J, Hart LG. Medicare in Rural America. Washington, odological priority at the start and Inner-City Areas. Washington, DC: Metropolitan, urban, and rural commut- DC: Medicare Payment Advisory Com- Government Printing Office; 1998. ing areas: toward a better depiction of mission; 2001. of any project that examines 5. Hassenger EW, Hobbs DJ. Rural the US settlement system. Urban Geog- 34. Thomas R. Geomedical Systems: In- health-related concerns associ- society—the environment of rural health raphy. 19 99;20:727–748. tervention and Control. London, UK: ated with the rural and urban care. In: Straub LA, Walzer N, eds. 20.Popper DE, Lang RE, Popper FJ. Routledge; 1992. dimension. Grappling early and Rural Health Care: Innovation in a From maps to myth: the census, Turner, Changing Environment. Westport, Conn: and the idea of the frontier. J Am Comp systematically with the problems Praeger; 1992:178–190. Cult. 2000;Spring:91–102. of defining rurality will signifi- 6. Ricketts TC III. The rural patient. 21. Hewitt M. Defining “Rural” Areas: cantly enhance the validity and In: Geyman JP, Norris TE, Hart LG, eds. Impact on Health Care Policy and Re- the utility of health research Textbook of Rural Medicine. New York, search. Washington, DC: US Govern- NY: McGraw-Hill; 2001:15–26. ment Printing Office; 1989. US Office work, which is essential in rural- 7. Geyman JP, Norris TE, Hart LG. of Technology Assessment Staff Paper. focused health research. Textbook of Rural Medicine. New York, 22. McGranahan DA. Natural Ameni- NY: McGraw-Hill; 2001. ties Drive Rural Population Change. 8. Loue S, Quill BE. Handbook of Washington, DC: Economic Research Service, US Department of Agriculture; About the Authors Rural Health. New York, NY: Kluwer Academic/Plenum Publishers; 2001. 1999. Agricultural Economic Report The authors are with the WWAMI Rural No. 781. Health Research Center, Department of 9. Hong W, Kindig DA. The relation- 23. Goodman DC, Mick SS, Bott D, et al. Family Medicine, University of Washing- ship between commuting patterns and Primary care service areas: a new tool ton, Seattle, Wash. health resources in nonmetropolitan for the evaluation of primary care ser- Requests for reprints should be sent to counties of the United States. Med Care. vices. Health Serv Res. 2003;38(1 Pt 1): L. Gary Hart, PhD, WWAMI Rural 19 92;30:1154–1158. 287–309. Health Research Center, Department of 10. Baldwin L-M, MacLehose RF, Family Medicine, University of Washing- Hart LG, Beaver SK, Every N, Chan L. 24. Slifkin RT, Randolph R, Ricketts TC. ton, Box 354982, Seattle, WA, Quality of care for acute myocardial in- The changing metropolitan designation 98195–4982 (e-mail: garyhart@ farction in rural and urban US hospitals. process and rural America. J Rural u.washington.edu). J Rural Health. 2004;20:99–108. Health. 2004;20:1–6. This article was accepted September 9, 11. Ricketts TC III, Johnson-Webb KD, 25. US Census Bureau. Census Bu- 2004. Randolph RK. Populations and places in reau Home Page. Available at: http:// rural America. In: Ricketts TC III, ed. www.census.gov. Accessed June 25, Contributors Rural Health in the United States. New 2004. L.G. Hart originated the study and was York, NY: Oxford University Press; 26. Economic Research Service. Eco- the principal writer. E.H. Larson helped 1999. nomic Research Service—USDA. Avail- develop the study and wrote and edited 12.Fuguitt GV, Brown DL, Beale CL. able at: http://www.ers.usda.gov. Ac- portions of the article. D.M. Lishner ed- Rural and Small Town America. New cessed June 25, 2004. ited the article. York, NY: Russell Sage Foundation; 27.Ratcliffe MR. Creating metropoli- 1989. tan and micropolitan statistical areas. Human Participant Protection 13.Hart JF. The Rural Landscape. Balti- Paper presented at: American Sociologi- No protocol approval was needed for more, MD: John Hopkins University cal Association Annual Meeting; August this study. Press; 1998. 19, 2002; Chicago, Ill. 14 .Rosenblatt RA. The health of rural 28. McGranahan DA, Beale CL. Un- people and the communities and envi- derstanding rural population loss. Rural References Am. 2002;17:2–11. 1. Larson EH, Hart LG. Rural health ronments in which they live. In: Gey- workforce methods and analysis. In: man JP, Norris TE, Hart LG, eds. Text- 29. Ricketts TC III, Savitz LA, Gesler Larson EH, Johnson KE, Norris TE, et al., book of Rural Medicine. New York, NY: WM, Osborne DN. Geographic Methods eds. State of the Health Workforce in McGraw-Hill; 2001:3–14. for Health Service Research. New York, Rural America: State Profiles and Com- 15. Hart LG, Dobi.e., SA, Baldwin LM, NY: University Press of America; 1994. parisons. Seattle, Wash: WWAMI Rural Pirani MJ, Fordyce M, Rosenblatt RA. 30. Hart LG, Taylor P. The emergence Health Research Center, University of Rural and urban differences in physi- of federal rural policy in the United Washington; 2003:15–22. cian resource use for low-risk obstetrics. States. In: Geyman JP, Norris TE, Hart LG, 2. Miller MK, Luloff AE. Who is Health Serv Res. 1996;31:429–452. eds. Textbook of Rural Medicine. New rural? A typological approach to the 16. Hart LG, Norris TE, Lishner DM. York, NY: McGraw-Hill; 2001:73–90. examination of rurality. Rural Sociol. Attitudes of family physicians in Wash- 31. Hart LG, Salsberg E, Phillips DM, 19 81;46:608–625. ington State toward physician-assisted Lishner DM. Rural health care providers 3. Miller KM, Farmer FL, Clarke LL. suicide. J Rural Health. 2003;19: in the United States. J Rural Health. Rural populations and their health. In: 461–469. 2002;18(suppl):211–232. Beaulieu JE, Berry DE, eds. Rural 17.Yawn BP, Bushy A, Yawn RA. Ex- 32. Openshaw S, Taylor PJ. The modi- Health Services: A Management Perspec- ploring Rural Medicine: Current Issues fiable areal unit problem. In: Wrigley N, tive. Ann Arbor, Mich: AUPHA Press/ and Concepts. Thousand Oaks, Calif: Bennet RJ, eds. Quantitative Geography: Health Administration Press; 1994: Sage Publications; 1994. a British View. London, UK: Routledge 3–26. 18. Ricketts TC III, ed. Rural Health in & Kegan Paul; 1981:60–69. 4. Council on Graduate Medical Edu- the United States. New York, NY: Oxford 33. Medicare Payment Advisory Com- cation. Tenth Report: Physician Distribu- University Press; 1999. mission. Report to the Congress:

July 2005, Vol 95, No. 7 | American Journal of Public Health Hart et al. | Peer Reviewed | Critical Concepts for Reaching Populations at Risk | 1155  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

The Right to Health Under International Law and Its Relevance to the United States

In recent years, there have | Alicia Ely Yamin, JD, MPH been considerable developments in international law with respect to the norma- 7 THE RIGHT TO HEALTH HAS ity. The Declaration sets forth The language of progressive re- tive definition of the right to evolved rapidly under interna- the right to a “standard of living alization and maximal available health, which includes both tional law, and its normative adequate for the health and well- resources, which suggests differ- health care and healthy con- clarification has significant conditions. These norms offer a being of himself and his family, ent standards for different coun- framework that shifts the ceptual and practical implica- including . . . medical care and tries, does not easily jibe with the analysis of issues such as tions for health policy. The ... the right to security in the absoluteness with which people disparities in treatment from framework that international event of . . . sickness, disability in the United States generally questions of quality of care human rights offers with respect ... or other lack of livelihood think about rights. Yet in practice, to matters of social justice. to health shifts the analysis of in circumstances beyond his con- due process and other civil rights Building on work in social issues such as disparities in trol.”7(article 25) The Declaration may vary just as much. Indeed, epidemiology, a rights para- treatment in the United States does not define the components the egregious disparities among digm explicitly links health from questions of quality of of a right to health; however, countries, and in particular be- with laws, policies, and prac- care to fundamental matters of they both include and transcend tween the global north and south, tices that sustain a func- democracy and social justice, as medical care. suggest not the irrelevance of tional democracy and fo- well as suggesting avenues for cuses on accountability. In The Cold War polarized coun- defining a right to health but the United States, framing accountability. tries’ positions on human rights. rather the need to situate state a well-documented prob- In 1966, instead of the indissolu- obligations within a global politi- lem such as health dispari- THE RIGHT TO HEALTH ble whole reflected in the Decla- cal economy in which interna- ties as a “rights violation” UNDER INTERNATIONAL ration, twin covenants on civil and tional institutions and third-party attaches shame and blame LAW political rights and economic, so- states often exercise inordinate to governmental neglect. cial, and cultural rights were influence over developing coun- Further, international law Under international law, there promulgated.8 The right to tries’ economies and policies. The offers standards for evalu- is a right not merely to health health was included in the In- right to health demands, as do all ating governmental con- care but to the much broader ternational Covenant on Eco- human rights, “international assis- duct as well as mecha- concept of health. Because rights nomic, Social and Cultural tance and cooperation.”1,9(article2)–11 nisms for establishing some degree of accountability. must be realized inherently Rights (ICESCR). Article 12 of The reference to a “highest (Am J Public Health. 2005; within the social sphere, this for- the ICESCR explicitly sets out a attainable standard” of health, 95:1156–1161. doi:10.2105/ mulation immediately suggests right to health and defines steps taken from the World Health Or- AJPH.2004.055111) that determinants of health and that states should take to “realize ganization constitution,12 builds ill health are not purely biologi- progressively” “to the maximum in a reasonableness standard.10–13 cal or “natural” but are also fac- available resources” the “highest That is, the state has a role to tors of societal relations.1,2 Thus, attainable standard of health,” in- play in leveling the social playing a rights perspective is entirely cluding “the reduction of the still- field with respect to health; how- compatible with work in epide- birth-rate and of infant mortality ever, there are factors that are miology that has established so- and for the healthy development beyond the state’s control.1,10,11 cial determinants as fundamental of the child”; “the improvement Furthermore, the highest attain- causes of disease.3–6 of all aspects of environmental able standard will necessarily The first notion of a right to and industrial hygiene”; “the pre- evolve over time, in response to health under international law is vention, treatment and control of medical inventions, as well as found in the 1948 Universal De- epidemic, endemic, occupational demographic, epidemiological, claration of Human Rights (here- and other diseases”; and “the cre- and economic shifts. after called Declaration), which ation of conditions which would In addition to the ICESCR, a was unanimously proclaimed by assure to all medical service and wide array of international and the UN General Assembly as a medical attention in the event of regional treaties recognizes common standard for all human- sickness.”9(article 12(2)) health as a rights issue, and these

1156 | Critical Concepts for Reaching Populations at Risk | Peer Reviewed | Yamin American Journal of Public Health | July 2005, Vol 95, No. 7  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

reflect a broad consensus on the affecting their well-being.1,19 Pub- countries legislate various aspects universal health care coverage content of the norms.14–19 A re- lic health has a long tradition of of the right to health.11 Further, and some kind of legal recogni- view of the international instru- recognizing that participation is recent clarification of normative tion of a right to care.39,40 ments and interpretive docu- integral to health promotion.21 obligations has permitted greater It was not predestined that ments makes it clear that the Further, analyses of the impor- attention to be devoted to poten- the United States should have di- right to health as it is enshrined tance of structural determinants tial violations of the right to verged so greatly from the rest of in international law extends of health and political economic health by treaty-monitoring com- the developed world in its atti- well beyond health care to in- context are increasingly committees in their “concluding ob- tudes about access to health care. clude basic preconditions for mon.3–6,22–25 Framing health as servations”—or judgments on Even after various reform pro- health, such as potable water a right adds to the growing litera- states’ compliance—as well as en- posals had floundered, in 1944, and adequate sanitation and ture in social epidemiology that forcement by quasi-judicial inter- before Eleanor Roosevelt and nutrition.1,10,11,13–19 links health with social justice; national institutions and national her colleagues included health it does this by first making ex- courts in specific cases.27–32 and medical care in the Declara- CLARIFYING CONTENT plicit the link between health Domestic courts and regional tion, President Franklin Delano AND STATES’ OBLIGATIONS and the construction of a func- bodies that have addressed the Roosevelt argued that every tional democracy. That is, health- question have generally agreed American was entitled to, inter Since the end of the Cold War, related resource distribution, evi- on what minimal standards gov- alia, “the right to adequate med- the interdependence and indivisi- dence of discrimination and ernments can be required to ical care and the opportunity bility of economic, social, and disparities, and the like are ana- meet. First, states have an obliga- to achieve and enjoy good cultural rights and civil and polit- lyzed not just in terms of their tion not to adopt retrogressive health.”41(ix) Cold War sentiments ical rights has been broadly ac- impact on health status but also measures. For example, if a state undoubtedly played a role in de- cepted.20 Further, there is now their relation to laws, policies, and administers a program to provide feating an important proposal by widespread agreement that the practices that limit popular partici- antiretroviral drugs, backsliding President Harry Truman for na- right to health entails both nega- pation in decisionmaking and, in because of budgetary difficulties tional health insurance, just as tive freedoms (e.g., from noncon- turn, the establishment of a gen- is impermissible.33,34 Second, they distorted international sensual medical treatment and uinely democratic society.2,6,26 health policies and programs human rights law.42 A wide vari- experimentation) and positive Second, failure to respect, pro- must not be discriminatory.33–35 ety of factors have been cited for freedoms or entitlements (e.g., ac- tect, or fulfill responsibilities re- Third, states must undertake ef- the continuing failure of the cess to care).1 Under international lating to health are construed not forts to regulate the conduct of United States to embrace a na- law, states that are party to a va- only in terms of ensuing social third parties that are interfering tional health program, including riety of different treaties assume or economic problems, but also with the right to health, such a lack of class identification and tripartite obligations: (1) to respect explicitly in terms of the account- as environmental polluters.36 a weak labor movement, nega- the right to health by refraining ability of the state and, to a cer- Fourth, governments can be re- tive attitudes about government, from direct violations, such as tain extent, other actors, under quired to develop national poli- and political-structural explana- systemic discrimination within national and international cies and plans of action to re- tions relating to the US constitu- the health system; (2) to protect law.1,6,10–12,26 Thus, a human spond to health concerns.37 tional system.43–45 the right from interference by rights framework simultaneously In this context, the discourse third parties, through such mea- acknowledges health as inher- RELEVANCE TO THE of rights can reconfigure public sures as environmental regulation ently political—intimately bound UNITED STATES expectations and commitments. of third parties; and (3) to fulfill up with social context, ideologies, For instance, when access to the right by adopting deliberate and power structures—and re- The relevance of consensually health care is construed as a mat- measures aimed at achieving uni- moves health policy decisions agreed-upon international norms ter of right, it is not dependent versal access to care, as well as to from being matters of pure politi- to domestic debates on health on good behavior. Even if there preconditions for health.1,10 Thus, cal discretion by placing them policy would be self-evident in is a widespread belief that ill it is wrong to think of the right to squarely into the domain of law. most of the world. However, his- health is often the result of poor health in terms of a package of As with all international torically the United States has personal choices, just as this soci- services, even a package extend- human rights, implementation been uniquely averse to accept- ety provides defense counsel to ing beyond medical care. and enforcement of the right to ing international human rights criminal defendants, who ar- Realization of the right to health critically depend on leg- standards and conforming na- guably may have exercised poor health further implies providing islative and judicial action at the tional laws to meet them.38 The choices, so too would the state individuals and communities national level. More than 70 na- United States is also the only in- have an obligation to ensure ac- with an authentic voice in deci- tional constitutions recognize dustrialized country in the world cess to health facilities, goods, sions defining, determining, and the right to health, and far more that does not provide a plan for and services.1 Further, once

July 2005, Vol 95, No. 7 | American Journal of Public Health Yamin | Peer Reviewed | Critical Concepts for Reaching Populations at Risk | 1157  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

health is framed as a right, the tions for the screening and treat- For example, the implementa- tions.55–58 A national study by contours of debates about the ment of persons coming to the tion of the right to education is the Department of Health and role of the state and markets emergency department of any surely deeply flawed in the Human Services’ Agency for shift. Despite discontent with in- hospital participating in the fed- United States, and experiments Healthcare Research and Quality efficiencies and poor quality of eral Medicare program.47 Federal with privatizing aspects of educa- determined that “racial, ethnic, care in many industrialized coun- courts have also been active in tion through charter schools and and socioeconomic disparities tries, the government’s obliga- ensuring that the conditions and voucher systems may hold ap- are national problems that affect tions with respect to health care treatment of patients in psychiat- peal to some in the face of gross health care at all points in the are well entrenched in society ric hospitals comply with consti- inefficiencies.51 Nonetheless, in process, at all sites of care, and as well as law in most of the de- tutional standards.48 Also, certain contrast with notions about for all medical conditions—in fact, veloped world.39,40 states have been less reluctant health care, the notion that disparities are pervasive in our The United States has under- than the federal government to everyone has a right to basic ed- health care system.”58 taken international legal obliga- impose an affirmative obligation ucation as a prerequisite for sus- A human rights analysis of tions relating to the right to to provide services to under- taining a democratic society is this situation determines first health. The United States is a served populations.49 now well ingrained in American the normative obligation and party to the International Con- Moreover, constitutional law culture, as well as in law.52 Yet then the violation. As a party to vention on the Elimination of all can and does change to accom- constitutional recognition of the the Race Convention, the US Forms of Racial Discrimination modate changes in public percep- idea that the federal government government has undertaken not (Race Convention), binding itself tions and political philosophy. As has a role to play in adopting just a moral but a legal obliga- to take measures to eliminate ra- Archibald Cox argued, accepting proactive measures to ensure tion “to prohibit and to elimi- cial disparities in public health that government is “not merely some degree of equality in basic nate racial discrimination in all and health care.14 In other cases, about policy but also has affirma- education is only 50 years old, its forms and to guarantee the the president has signed treaties tive obligations to satisfy basic which suggests that conscious- right of everyone, without dis- signaling the government’s intent necessities of citizens [including ness of health as a right might tinction as to race, colour, or na- to be bound by the provisions in medical attention] is the next be susceptible to a similarly dra- tional or ethnic origin, to equal- the future, but the Senate has not great challenge of North Ameri- matic shift. ity before the law, notably in the given its “advice and consent” for can constitutionalism.”50(p118–119) In the immediate term, well- enjoyment of . . . the rights to ratification. Nevertheless, as a sig- Invoking legal realism, Cass Sun- established standards relating to public health [and] medical natory to the ICESCR, the Con- stein goes further, suggesting that nondiscrimination and equal care.”14(article5(e)(iv)) Under the vention on the Elimination of All “with a modest shift in person- protection are extremely relevant Race Convention, the govern- Forms of Discrimination Against nel” on the Supreme Court, eco- to creating accountability for as- ment undertakes not just to Women, the Convention on the nomic and social rights, including pects of the right to health in the sanction incidents of discrimina- Rights of the Child, and others, health, “could well be included United States.53 For example, tion but to affirmatively eradi- the United States is bound not in our constitutional understand- once a state has taken steps to cate racial discrimination in all to contravene object or purpose ings, and certainly in the nation’s implement health rights, such as its forms.14 Further, even under of those treaties, an obligation constitutive commitments, which through Medicaid, courts are devolution or decentralization that becomes relevant in, inter is where they belong.”41(p108) obligated to ensure that it is schemes, the ultimate accounta- alia, assessing US trade and aid Even if—especially if—such a done in a nondiscriminatory bility for state and local law and policies to the extent that these “shift in personnel” is not immedi- manner, which affords judicial policy resides with the federal have health impacts.46 ately forthcoming, human rights protection.54 government under international Further, despite the notori- as enshrined in international law law. Thus, when state or local ously thin legal grounding for any offer a powerful alternative dis- APPLICATION OF A RIGHT governments fail to eliminate right to health care in domestic course to the prevailing market- TO HEALTH IN THE health disparities, the federal law, discrete aspects of health— oriented one through which to UNITED STATES government cannot divest itself including health care—are al- understand and mobilize public of final responsibility.59 ready construed in terms of judi- concern regarding issues such Major reviews of the more On the domestic level, Title cially protected rights in the United as disparities in treatment and than 1000 studies done recently VI of the Civil Rights Act of States. For example, in addition access to care in the United on health disparities in the United 19 64 prohibits discrimination to entitlements to health coverage States.10 , 41 Public consciousness States have found consistent, in all health care activities re- for defined population groups at can precede and encourage legal credible, and robust evidence of ceiving federal funding, which the federal level, the Emergency recognition, which in turn rein- differences based on race and virtually all do in one form or Medical Treatment and Active forces public awareness of con- ethnicity in diagnostic procedures another.53 Title VI, together with Labor Act imposes some obliga- cerns in terms of rights.41 as well as therapeutic interven- its regulations, arguably pro-

1158 | Critical Concepts for Reaching Populations at Risk | Peer Reviewed | Yamin American Journal of Public Health | July 2005, Vol 95, No. 7  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

hibits both intentional and dis- requires recognizing that discrim- Explicitly applying the dis- CONCLUSIONS proportionate adverse impact ination travels along various axes course of rights to a well-known discrimination.17 ,60, 61,62(p953),63 of identity.56 As CERD has rec- social problem such as disparities In recent years, international Under international law it is ognized, discrimination can have in treatment signals that certain law has developed swiftly with clear that discrimination under both racial and gendered dimen- people and social groups have respect to the normative defini- international law need not be in- sions, doubly disempowering been aggrieved by the govern- tion of the right to health, which tentional; it need only have the women of color and affecting ment’s inaction and failure to includes both health care and effect of impairing or nullifying their health in ways different regulate others’ actions. In the healthy conditions. Having a the enjoyment of rights to consti- from those affecting men.66 Simi- recommendations in his 1994 right to health also implies hav- tute a violation.1,14,64 larly, as Krieger writes, “Since report, the Special Rapporteur ing a right to participate in deci- In its review of the United the global expansion of Euro- forcefully stated that it was in- sions affecting one’s health and States’ country report in 2001, pean power and economies in cumbent on the government to therefore links health issues with the UN committee that monitors the mid-15th century and contin- ensure that both public and pri- active social citizenship. Among compliance with the Race Con- gent territorial conquest and in- vate actors recognize that “when other things, national recognition vention (CERD) specifically tercontinental slave trade, people persons from ethnic minorities of a right to health creates a role noted its concern with respect to have lived in a world of racial- aspire to equal treatment, they for the state in ensuring equality “persistent disparities in the en- ized class relations and class- are not asking for favours, but of access to care and the precon- joyment of, in particular, . . . ac- contingent race relations. It logi- seeking to enjoy the rights guar- ditions for health and demands cess to public and private health cally follows that racial/ethnic anteed by the United States the elimination of systemic dis- care” and recommended that the inequalities are shaped and fos- Constitution in their daily crimination. United States “take all appropri- tered by class inequalities and lives.”68(¶112[2]) Further, CERD In the United States, framing ate measures, including [affirma- vice versa.”25(p197) has specifically called for protec- an otherwise acknowledged tive] measures . . . to ensure Discrimination affects multi- tions of the rights, including problem such as disparities in [these rights].”65(¶398) The ple social determinants of health health, of migrant workers and treatment as a “rights violation” CERD’s concluding observations in the United States, as well as other noncitizens as well.69,70 suggests that the situation could make it clear that a right-to- treatment, and minorities are far Although statements by the be different and that the govern- health framework goes beyond more likely to lack access to Special Rapporteur and the ment bears responsibility. The both medical and ethical and care than Whites; this demon- CERD do not have the same language of rights attaches quality-of-care issues to focus on strates that discrimination within practical effect as domestic shame and blame to govern- state accountability. As illus- the health care system must be statutes and court decisions in mental neglect. Further, the in- trated by a 2003 Physicians for understood and addressed the United States, no amount of ternational norms relating to a Human Rights report, a rights ap- within the broader society, not cynical dismissals can change the right to health offer standards proach to racial disparities in just as a health issue but as a fact that they reflect on the US for evaluating governmental treatment includes such issues as democracy issue.67 For example, government’s compliance with conduct and mechanisms for provider education and service in his report on his site visit to legal obligations under interna- establishing some degree of delivery but emphasizes govern- the United States in 1994, the tional law.10 , 5 9 Additionally, these accountability. mental accountability for redress, UN Special Rapporteur on Con- statements are relevant to efforts as well as for improved collec- temporary Forms of Racism, Ra- to create moral and political action, analysis, and dissemination cial Discrimination, Xenophobia countability. That is, they can be About the Author The author is with the Harvard School of of appropriately disaggregated and Related Intolerance (Special instrumental in mobilizing aware- Public Health and is a human rights attor- data that permits the detection of Rapporteur) noted not only the ness among the general public of ney who at the time of writing was work- disparities and potential discrimi- manifold consequences of rac- health as fundamental to creating ing with nongovernmental organizations in Latin America. nation; it also includes the cre- ism and racial discrimination in a genuinely inclusive society in Requests for reprints should be sent to ation of effective enforcement the field of health but also re- the United States. Thus, the Alicia Ely Yamin, Law and Public Health mechanisms, such as a Health ferred to the responsibility of rights framework helps to change Program, Department of Health Policy and Management, 677 Huntington Ave., Section within the Civil Rights the US government for “socio- the bounds of discussions about 4th floor, Boston, MA 02115 (e-mail: Division of the Department of logical inertia, structural obsta- what needs to be done from how [email protected]). Justice.56 cles and individual resistance to improve service delivery to This article was accepted January 30, 2005. Data collection and analysis as hindering the emergence of a what laws, policies, and political well as policymaking concerned truly integrated society based actions are necessary to promote with creating a society in which on the equal dignity of the and protect the peoples’ dignity References 1. UN Committee on Economic, Social diverse individuals of all racial members of the American in all spheres of life on a nondis- and Cultural Rights. General Comment groups are on an equal footing nation.”68(¶112) criminatory basis. 14: The Right to the Highest Attainable

July 2005, Vol 95, No. 7 | American Journal of Public Health Yamin | Peer Reviewed | Critical Concepts for Reaching Populations at Risk | 1159  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

Standard of Health. Geneva, Switzerland: 15. Convention on the Rights of the OpenDocument. Accessed April 26, 38. Henkin L. Foreign Affairs and the United Nations: 2000. UN Document E/ Child. New York, NY: United Nations; 2005. United States Constitution. 2nd ed. New C.12/2000/4. Available at: http://www. 1989. UN document A/44/736. Avail- 28.UN Committee on the Rights of York, NY: Oxford University Press; unhchr.ch/tbs/doc.nsf/(symbol)/E.C.12. able at: http://www.unhchr.ch/html/ the Child. General Comment No. 3: HIV/ 1996:170–230. 2000.4.En?OpenDocument. Accessed menu3/b/k2crc.htm. Accessed April AIDS and the Rights of the Child. Geneva, 39. Light DW. Universal health care: April 26, 2005. 26, 2005. Switzerland; 2003. UN document CRC/ lessons from the British experience. Am 2. Yamin AE. Transformative combi- 16. Convention on the Elimination of All GC/2003/1. J Public Health. 2003;93:25–30. nations: women’s health and human Forms of Discrimination Against Women. 29.UN Committee on the Rights of 40.Brown L. Comparing health sys- rights. J Am Womens Assoc. 19 97;52: New York, NY: United Nations; 1979. the Child. General Comment No. 4: Ado- tems in four countries: lessons for the 16 9–173. UN document A/34/36. lescent Health and Development in the United States. Am J Public Health. 2003: 3. Link BG, Phelan J. Social condi- 17. Additional Protocol to the American Context of the Rights of the Child. Geneva, 93:52–56. tions as fundamental causes of disease. Convention on Human Rights in the Area Switzerland; 2003. UN document CRC/ 41. Sunstein C. The Second Bill of J Health Soc Behav. 19 95;Spec No: of Economic, Social and Cultural Rights GC/2003/4. Available at: http://www. Rights: FDR’s Unfinished Revolution and 80–94. (Protocol of San Salvador). Washington, unhchr.ch/tbs/doc.nsf/(symbol)/CRC. Why We Need It More Than Ever. New DC: Organization of American States; GC.2003.4.En?OpenDocument. Ac- 4. Marmot M, Wilkinson RG, eds. So- York, NY: Basic Books; 2004. cial Determinants of Health. London, En- 1988. cessed April 26, 2005. 42. Starr P. The Social Transformation gland: Oxford University Press; 1999. 18. African [Banjul] Charter on Human 30. Report of the International Confer- of American Medicine. New York, NY: and Peoples’ Rights. Banjul, Gambia: Or- ence on Population and Development. 5. Berkman L, Kawachi I. A historical Basic Books; 1982:281–283. framework for social epidemiology. In: ganization of African Unity (African Cairo, Egypt: United Nations; 1994. 43. Vladeck B. Universal health insur- Berkman L, Kawachi I, eds. Social Epi- Union); 1982. OAU document CAB/ UN document A/Conf.171/13. Available ance in the United States: reflections on demiology. New York, NY: Oxford Uni- LEG/67/3 rev 5. at: http://www.un.org/popin/icpd/ conference/offeng/poa.html. Accessed the past, the present, and the future [ed- versity Press; 2000:3–12. 19. Convention Concerning Indigenous April 26, 2005. itorial]. Am J Public Health. 2003;93: and Tribal Peoples in Independent Coun- 6. Marks S. The new partnership of 16–19. health and human rights. Hum Rights tries (Convention 169). Geneva, Switzer- 31. Interim Report of Paul Hunt, Special 44.Klein R. Lessons for (and from) Dialogue. 2001;2:21–22. Available at: land: International Labour Organization; Rapporteur of the Commission on Human America. Am J Public Health. 2003;93: http://www.cceia.org/viewMedia.php/ 1989. Rights. New York, NY: United Nations; 2003. Available at: http://ods-dds-ny. 61–63. prmTemplateID/8/prmID/650. Ac- 20.Vienna Declaration and Pro- un.org/doc/UNDOC/GEN/G03/109/ cessed March 13, 2005. gramme of Action, Report of the World 45. Navarro V. Policy without politics: 79/PDF/G0310979.pdf. Accessed the limits of social engineering. Am J Pub- 7. Universal Declaration of Human Conference on Human Rights. Vienna, March 13, 2005. lic Health. 2003;93:64–66. Rights. United Nations General Assembly Austria: United Nations; 1993. UN doc- Resolution 217 A (III). New York, NY: ument A/CONF.157/23. 32. Concluding Observations of the 46.Vienna Convention on the Law of Committee on Economic, Social and Cul- United Nations; 1948. 21. Report of the International Confer- Treaties. Article 18. 1155 UNTS 331, tural Rights: Ecuador. Geneva, Switzer- ence on Primary Health Care [Declara- 340; 8 ILM. 679, 692 (1969). 8. Craven M. The International land: United Nations; 2004. Available tion of Alma-Ata]. Geneva, Switzerland: Covenant on Economic, Social and Cul- at: http://www.unhchr.ch/tbs/doc.nsf/ 47. Emergency Medical Treatment and World Health Organization; 1978. tural Rights: A Perspective on Its Develop- (Symbol)/E.C.12.1.Add.100.En. Ac- Labor Act, 42 USCA §1395dd(e)(2) ment. Oxford, England: Clarendon Press; 22. Krieger N. Theories for social epi- cessed March 6, 2005. (1994); Interim Final Rule with Com- 19 95. demiology in the 21st century: an ment period, 59 Federal Register 32086 33. Cortez v Instituto Venezolano de ecosocial perspective. Int J Epidemiol. (1994) (codified at 59 CFR §32086). 9. International Covenant on Eco- Seguros Sociales (Supreme Court of 2001;30:668–677. nomic, Social and Cultural Rights. New Venezuela, Constitutional Chamber 48. Wyatt v Stickney, 344 F Supp 373 York, NY: United Nations; 1966. UN 23. Navarro V, Shi L. The political 2001). Available at: http://www.tsj.gov. (MD Ala 1972). document A/6316. Available at: http:// context of social inequalities in health. ve/decisiones/scon/Mayo/881- 49. State Health Facilities Association v www.unhchr.ch/html/menu3/b/a_cescr. Int J Health Serv. 2001;31:1–21. 080502-00-0995.htm. Accessed Axelrod, 568 NYS2d 1, 569 NE2d 860 htm. Accessed April 26, 2005. 24. Geronimous A. Addressing struc- March 6, 2005. (NY 1991). 10. Kinney E. The international right tural influences on the health of popula- 34. Jofre Mendoza v Ministerio de Salud. 50. Cox A. The Role of the Supreme to health: what does this mean for our tions. Am J Public Health. 2000;90: Constitutional Tribunal of Ecuador, Court in American Government. New nation and our world? Indiana Law Rev. 867–872. Third Court, Decision No, 0749-2003- York, NY: Oxford University Press; 2001;34:1457–1475. 25. Krieger N. Does racism harm RA (2004). 1977:118–119. 11.Toebes B. The Right to Health as a health? Did child abuse exist before 35. Alejandro Moreno Alvarez v 51. Report of the Special Rapporteur on Right in International Law. Oxford, En- 19 62? On explicit questions, critical sci- Estado Colombiano. SU.819/99 (Con- the Right to Education, Katarina Toma- gland: Intersentia/Hart; 1999. ence, and current controversies: an eco- stitutional Court of Colombia, 1999). sevski, Mission to the United States of 12.World Health Organization. Consti- social perspective. Am J Public Health. Available at: http://bib.minjusticia.gov. America. New York, NY: United Nations; tution. Geneva, Switzerland: World 2003;93:194–199. co/jurisprudencia/CorteConstitucional/ 2001. Health Organization; 1946. 1999/Tutela/su819-99.htm. Accessed 26. Krieger N, Gruskin S. Frameworks 52. Brown v Board of Education I, 347 March 18, 2005. 13. Chapman A. Core obligations re- matter: ecosocial and health and human US 483 (1954). lated to the right to health. In: Chap- rights perspectives on disparities among 36. Social and Economic Rights Action 53. Title VI, Civil Rights Act of 1964. man A, Russell S, eds. Core Obligations: women’s health—the case of tuberculo- Center v Nigeria. Communication 155/96 As amended, 42 USC §2000d (1994). Building a Framework for Economic, So- sis. J Am Womens Assoc. 2001;56: (African Commission on Human & Peo- cial and Cultural Rights. New York, NY: 137–142. ples’ Rights, Oct. 2001) ¶ 67. Available 54. Linton v Tennessee Commissioner of Intersentia; 2002:85–216. 27.UN Committee on the Elimination at: http://www.achpr.org/DECISIONS_ Health and Environment, 779 F Supp 14 . International Convention on the of Discrimination Against Women. Gen- 30th_Session-_Oct.2001_eng.pdf. Ac- 925 (MD Tenn 1990); affirmed 65 F3d Elimination of All Forms of Racial Dis- eral Recommendation No. 24: Women cessed March 13, 2005. 508 (6th Cir, 1995). crimination. New York, NY: United Na- and Health. Geneva, Switzerland; 1999. 37. Minister of Health v Treatment Ac- 55. Smedley BD, Stith A, Nelson AR, tions; 1966. UN document A/6014. UN document CEDAW/C/1999/I/ tion Campaign. CCT 8/02 (Constitu- eds, Unequal Treatment: Confronting Ra- Available at: http://www.unhchr.ch/ WG.II/WP.2/Rev.1. Available at: http:// tional Court of South Africa, 2002). cial and Ethnic Disparities in Health html/menu3/b/d_icerd.htm. Accessed www.unhchr.ch/tbs/doc.nsf/(symbol)/ Available at: http://www.tac.org.za Care. Washington, DC: National Acade- April 26, 2005. CEDAW+General+recom.+24.En? (PDF file). Accessed April 28, 2005. mies Press; 2002.

1160 | Critical Concepts for Reaching Populations at Risk | Peer Reviewed | Yamin American Journal of Public Health | July 2005, Vol 95, No. 7  CRITICAL CONCEPTS FOR REACHING POPULATIONS AT RISK 

56. Panel on Racial and Ethnic Dispar- land: United Nations; 2005. UN docu- ities Convened by Physicians for ment E/CN.4/1995/78/Add.1. Human Rights. The Right to Equal 69.UN Committee on the Elimination Treatment. Boston, Mass: Physicians of All Forms of Racial Discrimination. for Human Rights; 2003. Thematic Discussion on Non-Citizens and 57. Lantz PM, House JS, Lepkowski JM, Racial Discrimination. Geneva, Switzer- Williams DR, Mero RP, Chen J. Socio- land: United Nations; 2004. UN docu- economic factors, health behaviors, and ment CERD/C/SR.1624. mortality: results from a nationally representative prospective study of US 70.UN Committee on the Elimination adults. JAMA. 1998;279:1703–1708. of All Forms of Racial Discrimination. General Recommendation XXX: Discrimi- 58. Agency for Healthcare Research nation Against Non-Citizens. Geneva, and Quality. The National Healthcare Switzerland: United Nations; 2004. UN Disparities Report. Washington, DC: US document CERD/C/64/Misc.11/rev.3. Department of Health and Human Ser- vices; 2003. Available at: http://www. ahrq.gov/qual/nhdr03/nhdrsum03. htmFindings. Accessed March 6, 2005. 59. McDougall G. Toward a meaningful international regime: the domestic relevance of international efforts to eliminate all forms of racial discrimination. Howard Law J. 19 97;40:571. 60. Alexander v Choate, 469 US 287, 293 (1985). 61. Alexander v Sandoval. 532 US 275, 281–282 (2001). 62. Watson SD. Reinvigorating Title VI: defending health care discrimination—it shouldn’t be so easy. Fordham Law Rev. 1990;58:939–978. 63. Furrow BR, Greaney TL, John- son SH,Stoltzfus Jost T, Schwartz RL. Health Law. 2nd ed. Saint Paul, Minn: Westlaw; 2000. 64. Committee on Economic, Social and Cultural Rights. General Comment No. 3: the Nature of States Parties’ Oblig- ations. Geneva, Switzerland: United Na- tions; 1990. UN document E/1991/23, annex III. 65. Concluding Observations of the Committee on the Elimination of Racial Discrimination: United States of America. Geneva, Switzerland: United Nations: 2001. UN document A/56/18. 66.UN Committee on the Elimination of All Forms of Racial Discrimination. General Recommendation XXV: Gender- Related Dimensions of Racial Discrimina- tion. Geneva, Switzerland: United Na- tions. 2000. UN document CERD/A/ 55/18, annex V. 67. Institute of Medicine. Coverage Matters: Insurance and Health Care. Washington, DC: National Academies Press; 2001. 68. Implementation of the Programme of Action for the Second Decade to Combat Racism and Racial Discrimination, Report of Mr. Maurice Glele-Ahanhanzo, Special Rapporteur on Contemporary Forms of Racism, Racial Discrimination, Xenopho- bia and Related Intolerance on His Mis- sion to the United States of America from 9 to 22 October 1994. Geneva, Switzer-

July 2005, Vol 95, No. 7 | American Journal of Public Health Yamin | Peer Reviewed | Critical Concepts for Reaching Populations at Risk | 1161  PUBLIC HEALTH MATTERS 

A Critical Analysis of the Brazilian Response to HIV/AIDS: Lessons Learned for Controlling and Mitigating the Epidemic in Developing Countries

| Alan Berkman, MD, Jonathan Garcia, BA, Miguel Muñoz-Laboy, DrPH, Vera Paiva, PhD, and Richard Parker, PhD

incidence rates of HIV are much lower than The Brazilian National AIDS Program is widely recognized as the leading ex- projected a decade ago, and mortality rates ample of an integrated HIV/AIDS prevention, care, and treatment program in a developing country. We critically analyze the Brazilian experience, distinguishing have fallen by 50% and inpatient hospitaliza- those elements that are unique to Brazil from the programmatic and policy de- tion days by 70% to 80% over the past 7 4 cisions that can aid the development of similar programs in other low- and middle- years. While implementation of this program income and developing countries. required the commitment of significant re- Among the critical issues that are discussed are human rights and solidarity, sources, it is now estimated that by 2001 an the interface of politics and public health, sexuality and culture, the integration investment of US$232 million resulted in a of prevention and treatment, the transition from an epidemic rooted among men total savings of US$1.1 billion.5 who have sex with men to one that increasingly affects women, and special pre- We do not believe that the Brazilian NAP vention and treatment programs for injection drug users. (Am J Public Health. can serve as a “model” that can be uncritically 2005;95:1162–1172. doi:10.2105/AJPH.2004.054593) implemented in other countries; in fact, the most basic lesson from the Brazilian experi- For those concerned about the HIV/AIDS rate enhanced care and treatment for those ence may well be that there is no homoge- pandemic, we are living through the best of infected with HIV. The challenge to develop neous HIV/AIDS epidemic nor a prepackaged times and the worse of times. Since the 13th such a program in the context of poorly de- approach to dealing with it. The way in which International AIDS Conference in Durban, veloped health systems is profound, and there a nation responds to the social, political, eco- South Africa, there has been growing interna- is an understandable and urgent need for di- nomic, and human stress (and distress) caused tional attention to the scope and nature of the rection. “Best practice” strategies have been by HIV/AIDS will be shaped by that coun- catastrophe, increased political will in a num- one answer; however, while inspiring, they try’s unique history, culture, governmental in- ber of countries, and a substantial, albeit in- are often small-scale projects that focus on a stitutions, and economic resources and the di- sufficient, increase in available resources. At single element of a comprehensive plan (treat- verse social forces and institutions that get the same time, the epidemic continues to ment, care, prevention) with limited heuristic lumped together as “civil society.” However, grow, reversing decades of development in a value for those charged with formulating an we believe there is value in looking at Brazil number of African countries and promoting integrated national plan.3 There is also a as a case study, briefly examining the unique the very economic and social conditions that temptation to decontextualize such programs Brazilian context and then focusing on specific facilitate its spread to yet another generation and mechanically transplant them to radically policy decisions that may be helpful to those of young people. different settings. Yet, the need to learn from grappling with their own national realities. A consensus formed at the Durban Con- others’ experiences so that mistakes can be ference was that a strategic approach to the minimized and scarce resources allocated cor- THE BRAZILIAN CONTEXT HIV epidemic must integrate prevention with rectly remains critical. care, treatment, and mitigation. This was an With this environment in mind, we present As a consequence of the deep inequalities implicit rejection of the dominant interna- a critical analysis of the development of the and regional differences that exist in Brazilian tional paradigm that poor and developing Brazilian National AIDS Program (NAP), a society, the spread of HIV infection has been countries must focus only on prevention. Be- widely recognized, leading example of the complex, characterized by a number of di- cause the demand for treatment has become feasibility and effectiveness of an integrated verse patterns in different regions of the coun- such a contentious topic, advocates, policy- approach to the epidemic in the setting of a try.6,7 In spite of regional differences, however, makers, and researchers have focused special middle-income country characterized by the Brazilian epidemic is currently character- attention on Brazil’s successful program for significant levels of social inequality. Even ized by 3 major, interrelated, epidemiological providing universal access to free antiretrovi- though United Nations indices of human de- trends that are evident in all regions of the ral therapy.1,2 velopment have consistently placed Brazil country, which are described by Brazilian re- Many national governments are now devel- around 70th place, the impact of the Brazil- searchers as (1) heterosexualization, (2) femi- oping new, strategic AIDS plans that incorpo- ian response to AIDS has been impressive: nization, and (3) pauperization.8,9

1162 | Public Health Matters | Peer Reviewed | Berkman et al. American Journal of Public Health | July 2005, Vol 95, No. 7  PUBLIC HEALTH MATTERS 

Homosexual Heterosexual Injection drug use Blood transfusion Mother-to-child 100

66.1 60 entage rc

Pe 40

1.0 20 12.5 2.4 0.1 0 1980 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Year Source. National AIDS Program, Brazilian Ministry of Health. Note. Notified cases up to December 31, 2003. FIGURE 1—Percentage of AIDS cases by type of transmission and year of diagnosis: Brazil, 1980–2003.

Although the epidemic began in Brazil in mounted.2 Fourteen years later, this scenario in turn, prove to be central to the Brazilian the early 1980s primarily through sexual has yet to materialize. On the contrary, an es- response to HIV/AIDS) were “citizenship” transmission between men, heterosexual timated 600000 people in Brazil are infected and “solidarity.” Citizenship defined the rela- transmission has gradually become the major with HIV and 362364 have AIDS.6 Inci- tionship between the Brazilian people and the mode of HIV infection (Figure 1).6 Increased dence rates of HIV infection are much lower state (through its democratic institutions); soli- heterosexual transmission has resulted in sub- than projected a decade ago (Figure 4), and darity, and respect for human rights, defined stantial growth of HIV infection and AIDS mortality rates have fallen by roughly 50% the relationship among the people.10 ,13 ,14 cases among women, and the male-to-female (Figure 5). Inpatient hospitalization days have In asserting their rights as citizens in the ratio of reported cases has shifted from 23.5 been significantly reduced, resulting in lower new constitution of 1988, Brazilians were de- to 1 in 1985 to 1.7 to 1 in 2002 (Figure 2).6 hospital expenses owing to the investment in manding that the city, state, and national ad- When level of education is used as a proxy treatment access.1,2,4 ministrations enter into a dialog with civil for socioeconomic status, the increasing pro- Aspects of the Brazilian response to HIV/ society about the future of the country.15–17 portion of cases among people with lower AIDS have been described and analyzed by This redemocratization movement built politi- education levels indicates a trend of pauper- a number of the outstanding activists, social cal parties, trade unions, and nongovernmen- ization in the epidemic (Figure 3).6 These pat- scientists, and public health officials who tal organizations (NGOs) throughout the terns are important in revealing the key chal- helped shape that response. There is wide- country in the 1980s, culminating in a de- lenges that must still be overcome to control spread agreement among these analysts that mand for elections for a new and free Con- the epidemic. the Brazilian mobilization against HIV must gress. Democratic elections were initially held Nonetheless, the effectiveness of Brazil’s be viewed in the context of the larger social only at the municipal and state levels. The response to HIV/AIDS has been demon- mobilization of Brazilians confronting the mil- negotiation and promulgation of the new strated through Brazil’s historical epidemio- itary dictatorship and demanding democracy “democratic” constitution, passed in 1988, logical profile, with a clear trend toward the and a return to civilian rule.10–12 included the reinstitution of free national stabilization of the epidemic over time.8 In elections as of 1990. 1990, the World Bank predicted that within Citizenship, Solidarity, and One strong player in this national mobi- 10 years there would be 1.2 million people Social Mobilization lization for democracy was the “sanitary re- infected with HIV in Brazil unless an effec- Two key concepts that underlay the social form movement,” a loose affiliation of health tive, nationally based intervention was mobilization for democracy (and that would, care workers, collective health academics,18

July 2005, Vol 95, No. 7 | American Journal of Public Health Berkman et al. | Peer Reviewed | Public Health Matters | 1163  PUBLIC HEALTH MATTERS 

25 23.5

14.9 15

10 8.6

6.2 5.9

le-to-Female Ratio of Reported AIDS Cases le-to-Female 5.3 4.7 5 4 Ma 3.6 3.3 2.8 2.5 2.2 2 1.9 1.8 1.7 1.7

0 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

Year

Source. National AIDS Program, Brazilian Ministry of Health. FIGURE 2—Gender ratio (male to female) of notified AIDS cases: Brazil, 1985–2002.

100

University education Primary education Secondary education Illiterate 60

20 AIDS Cases Among Those Aged 19 Years and Older, % and Older, 19 Years Aged Among Those AIDS Cases 0 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Year Source. National AIDS Program, Brazilian Ministry of Health. FIGURE 3—Percentage of AIDS cases among those aged 19 years and older, by level of education: Brazil, 1985–2002.

trade unions, Catholic and Christian human right to be guaranteed by the consti- ment were appointed to senior positions in churches, and new political parties, who de- tution. The sanitary reform movement19 , 2 0 the health department. São Paulo was the manded a public health system responsive to was particularly strong in São Paulo state and epicenter of the AIDS epidemic, and the São and controlled by the public and who de- city, and when opposition parties won the Paulo State Health Department led the re- fended the right to health as a fundamental first state elections, members of that move- sponse to the emergence of the first reported

1164 | Public Health Matters | Peer Reviewed | Berkman et al. American Journal of Public Health | July 2005, Vol 95, No. 7  PUBLIC HEALTH MATTERS 

35000 25 Notified cases Incidence 30000 Number of Cases per 100 000 habitants 20

25000

es 15

20000

Cas of of

b er b 15000

10 Num

10000

5 5000

0 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Year Source. National AIDS Program, Brazilian Ministry of Health. Note. Notified cases up to December 31, 2003. FIGURE 4—Number of AIDS cases and incidence rate, by year of diagnosis: Brazil, 1992–2003.

16 Male 14 Female Both 12

6 per 100 000 Inhabitants

te 4 Ra 2

0 1985 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

Year of Diagnosis

Source. National AIDS Program, Brazilian Ministry of Health. FIGURE 5—AIDS mortality rate, by gender: Brazil, 1984–2002.

cases of AIDS (in 1983). It later became the gle for democracy, was the basis for a sense standing negative impact on the health of model for the National Unitary Health Sys- of social solidarity across many traditional so- the Brazilian population.23,24 Despite these tem, typically referred to by its acronym in cietal divisions.10 This should not be idealized very real differences in power and prestige, Brazilian Portuguese, SUS (Sistema Único de or romanticized: Brazil was and is a nation however, social solidarity built up out of Saúde).12 , 21, 2 2 with great disparities of wealth, a long history common suffering and the struggle for This mobilization process, in which many of social discrimination based on skin color, democracy and citizenship became a counter- diverse social movements made up of Brazil- and oppressive gender relationships, all of vailing force to the stigma surrounding the ian citizens came together in a common strug- which had (and continue to have) a long- emergence of HIV.14 , 2 4,25

July 2005, Vol 95, No. 7 | American Journal of Public Health Berkman et al. | Peer Reviewed | Public Health Matters | 1165  PUBLIC HEALTH MATTERS 

What were the factors that effectively mit- important lessons for other countries in the International Monetary Fund and the World igated the worst aspects of the stigma sur- Brazilian experience. The issue of political Bank as a solution to problems such as infla- rounding both HIV and homosexuality? A leadership is often put forward as critical to tion and the debt crisis. These policies, often critical number of gay men and human an effective response to HIV. While that may called “the Washington Consensus,” encour- rights activists, as well as men and women be true in Uganda and certain other fre- aged foreign capital investment in finance and infected or affected by HIV, openly con- quently cited examples, political leadership is industry and prioritized fighting inflation fronted the stigma, demanding that the not necessarily synonymous with governmen- through currency devaluation and restricted rights of people living with AIDS be re- tal leadership. The situation in Brazil (and this governmental expenditures on social services. spected by the government and by their fel- is true of many other countries) was that lead- Financing for the public health system was low citizens. These developments were par- ership emerged from civil society.13 This is slashed, and the privatization of health ser- ticularly important in São Paulo, where not to downplay the critical role of govern- vices grew rapidly. It was in this context that opposition to the military regime had been ment in confronting HIV; it was the some- the movement for redemocratization in deeply rooted and where opposition political times tense dialog between civil society and Brazil made public health and the human parties had come to power as soon as demo- the government in Brazil that resulted in an right to health central demands on govern- cratic elections had been restored. In 1983, effective national response. One only has to ment (see the section “Health Care as a in response to demands from gay activists, examine the painful situation in South Africa Human Right”).5,33 the São Paulo State Secretariat of Health over the past several years to understand the South Africa and most other sub-Saharan founded the first governmental AIDS pro- impact of a government that is unresponsive African countries have a much different polit- gram in the country. In 1985, an alliance of or too slow to answer and collaborate with ical history. The progressive colonization of gay men, human rights activists, and health civil society initiatives.30,31 the continent by European powers was for- professionals came together to form GAPA malized in 1885. Colonial governments were (the AIDS Prevention and Support Group), Big State, Little State primarily charged with maximizing extraction the first nongovernmental AIDS service or- Attempts to take lessons from the Brazilian of raw materials and profits for the colonizing ganization, which became an important experience and use them in developing na- country; health care was largely limited to model for similar organizations in cities tional AIDS programs in sub-Saharan Africa those interventions necessary to control epi- around the country.25–28 Similarly, in Rio de must take into account the relative strength demics that might affect Europeans and to Janeiro (like São Paulo, an important center of the Brazilian public health care system. Its do the minimum necessary to maintain a sta- for political opposition), researchers, health strength is not solely a function of Brazil’s ble work force. This policy resulted in a professionals, and activists came together to economic standing as a middle-income coun- stunted public health care system centered in form ABIA (the Brazilian Interdisciplinary try; South Africa’s per capita gross national large cities with the greatest European popu- AIDS Association) in 1986, and the Grupo product is also considered middle income by lations, and a health system for African work- Pela Vidda (the Group for Life), the first self- international standards. Brazil and South ers in the extractive industries that was under identified HIV-positive advocacy group in Africa share similarly high levels of economic the control of mining companies. Colonial the country, was founded in 1989. polarization—both have a GINI Index of 59.32 governments (with the exception of some Throughout the late 1980s and early In Brazil, as in any other country, political de- coastal West African countries) reserved ad- 1990s, a vibrant rebirth of civil society29 led cisions as well as economic resources shape ministrative and professional positions in the to the formation of NGOs (described in Brazil the health care system. health care system for Europeans and limited as ONGs/AIDS or AIDS NGOs) in other key The SUS has unquestionably been a quali- access to higher education for Africans. Per- cities and states around the country. Working tative advance in the history of public health haps the most extreme, but not unrepresenta- together with progressive state and municipal in Brazil.21 Its core principles of integrality tive, example was the Belgian Congo, which health departments, they would pressure the (prevention and treatment), public accounta- had a total of 8 university graduates at the federal government to create a national AIDS bility, and public funding distinguish it from time of independence in 1960. program. These factors combined to create an early versions of governmental health systems Political decolonization in most of Africa early response to HIV that was based on soli- and make it a proper vehicle for comprehen- occurred during the period 1960 to 1970 darity and inclusion rather than stigma and sive management of HIV. While recognizing and was often accompanied by the emigration exclusion, which in turn provided the founda- the unique aspects of the SUS, it is equally of the European administrators and physi- tion for the later development of the national important to recognize that it emerged from cians responsible for the health care system. response to AIDS, as discussed under the sec- a long tradition of advocacy for governmental A number of newly independent countries tion heading “Culture.”13 , 2 4,25 responsibility for the health of the nation, al- made attempts to develop primary health The political crisis of military rule that pre- beit a tradition frequently marred by ineffi- care systems in the decade after independ- cipitated the social mobilization of large num- ciency, waste, and corruption.22 ence, but such efforts were often handicapped bers of Brazilians cannot be artificially recre- This social pact was challenged by the em- by insufficient funds and human resources. ated in other countries. Yet there may be brace of the macroeconomic policies of the In other countries, the functions of the state

1166 | Public Health Matters | Peer Reviewed | Berkman et al. American Journal of Public Health | July 2005, Vol 95, No. 7  PUBLIC HEALTH MATTERS 

apparatus were never reoriented to serve the AIDS advocacy groups developed legal aid values their lives, and adds to the stigma that needs of the citizenry.34–36 programs and brought a series of successful may accompany illness.42,43 This lesson was Attempts to strengthen public health sys- class action suits focused on specific program- learned during the development of a plan for tems during this period met strong opposition matic issues (e.g., free viral-resistance testing, Hansen’s disease (leprosy) in São Paulo years from the International Monetary Fund and an expanded drug formulary) that have oper- before the first case of AIDS in Brazil.12 the World Bank. Rather than promote public ationalized the constitutional right to health. Social control refers to the direct role that health, structural adjustment programs forced These law suits, in turn, created a public civil society plays in setting the priorities for governments to cut spending on health care venue where PLWHA can assert their rights the SUS. Public health councils with elected and institute users’ fees in the public system. as Brazilian citizens and function as protago- community representatives exist at all levels To make a reasonable salary, professionals in nists in their own struggle for life.29,39 of the SUS: municipal, state, and federal.44 the public health system often sought work Many countries recognize health care as a Planning is the responsibility of the federal in the private health care system. The weak human right, but in relatively few instances and state levels, while implementation is done health systems that now plague efforts to con- have legal strategies been as fruitful as in through the municipalities. Over 120000 trol HIV in sub-Saharan Africa must be seen Brazil. The Treatment Action Campaign and people serve on these health councils, setting as the product of both colonial history and the AIDS Law Project in South Africa have local programmatic and budgetary priorities the “small government” model promoted by pursued a similar strategy in the South Afri- within the overall national health plan. the Washington consensus.33,37 can courts with some success. Within Latin Every 4 years, there is a structured debate The AIDS epidemic may force African America and the Caribbean, a number of or- at the local and state levels about national governments to make public health a priority. ganizations representing PLWHA have de- health planning; the SUS uses input from this African heads of state meeting in Abuja, manded antiretroviral treatment in suits filed debate to present a plan to a national health Nigeria, in 2001 pledged to increase spend- against their respective governments before conference. This system of controle social or ing on health to 15% of their national budg- the InterAmerican Court of Human Rights. “social control” (as it is described by health ets. Not one has yet achieved that goal. Corre- This court has ruled in favor of the plaintiffs, activists and government officials alike) is still spondingly, promises by the United States and but it has no direct authority to force govern- in the process of being constructed and can its major European allies to eliminate debt re- ments to comply with its orders.40 What still be vulnerable to changing political priori- payment and increase development aid to seems to distinguish the Brazilian situation ties, as was the case during the Collor govern- 0.7% of gross national product have not from that of many other countries is that the ment in the early 1990s. Nonetheless, this been implemented.38 Brazilian government acts in a timely and ap- process has been extended steadily over the The lesson that one can reasonably draw propriate manner to implement court rulings. course of the past decade, and it starkly con- from the Brazilian experience is that govern- Health care as a fundamental right has trasts with the bureaucratic nature of public ments must acknowledge that health care is been operationalized in the SUS. The SUS health in many other countries.15 Involvement as much a central responsibility as national was founded on and developed from 4 key of PLWHA and other sectors of society is still defense and that international agencies can- principles: (1) universal access, (2) integral contentious or only given lip service in many not merely lament weak health care systems care, (3) social control, and (4) public funding. countries; however, such involvement existed but should take steps to change those macro- Integral care was a core concept of the san- in the state of São Paulo and in other regions economic policies that hamstring governmen- itary reform movement in Brazil before the of Brazil from the very beginning of the AIDS tal efforts to strengthen those systems. debate about the need for linking treatment epidemic, eventually becoming the model for and prevention emerged within the interna- the NAP and the proactive response to HIV/ Health Care as a Human Right tional AIDS movement.41 Integrality recog- AIDS within the SUS. Health care is recognized in the Brazilian nizes that the governmental responsibility to constitution as a fundamental right of all citi- health is not limited to the basic prevention Centralization vs Decentralization zens and a fundamental responsibility of the measures (e.g., vaccines) found in maternal The balance between centralized functions government. This status as a fundamental and child health programs. It asserts that pre- such as planning, standards, and budgeting, right creates an obligation on the part of the vention must be integrated with care and and decentralized functions, primarily imple- government to take all reasonable steps to ac- treatment. The right to health extends to mentation, is a problem all national health tualize that right. those already ill and in need of treatment, systems confront. In most countries, the min- The Brazilian constitution created both a and there is recognition that having people istry of health initiates programs, issuing di- moral and a legal basis for the demand for access the health system will improve the rectives to state or provincial health depart- comprehensive treatment for people living whole range of public health initiatives. Inte- ments responsible for regional planning. with HIV/AIDS (PLWHA). However, it must grality also is based on a commitment to the These regional ministries then direct local be recognized that, at least until the mid- human rights of those afflicted: a prevention- health departments to implement the pro- 1990s, the government itself rarely took the only approach to health violates those rights grams. Financing, unfortunately, often does initiative to expand services for PLWHA.26 and the dignity of those in need of care, de- not follow the same direction as the directives.

July 2005, Vol 95, No. 7 | American Journal of Public Health Berkman et al. | Peer Reviewed | Public Health Matters | 1167  PUBLIC HEALTH MATTERS 

The SUS and, particularly, the NAP have a on behalf of PLWHA.29 Projects have been institutions, such as the Catholic Church and different dynamic. As discussed in the section funded for lesbian organizations, independent the Brazilian state apparatus.47,48 This same “Health Care as a Human Right,” members of the relatively low epidemiological risk of principle of solidarity has clearly resonated of the sanitary reform movement were ap- HIV infection in this population, precisely in response to the plight of PLWHA. pointed to a number of large municipal and because strengthening sexual rights has been Just as moral principles of solidarity in state health departments in the late 1970s understood as central to a broader effective Brazilian culture have been central to the and began to reorganize public health along response to the epidemic. Even the annual foundation of a national response to HIV democratic principles. Dialog, responsiveness, Gay Pride Parade in São Paulo, which has and AIDS, sexuality and sexual expression and cooperation characterized the relation- grown in recent years to draw up to 1 million are also an integral part of Brazilian culture ship between the health department and civil people from all over the country, has received and have facilitated the development of an society groups. In 1983, when the first cases regular financial support from the Brazilian effective response to the epidemic.10,45,49 of AIDS were reported in São Paulo, the Ministry of Health. In short, the battle against Certainly there is more than 1 discourse Brazilian government’s response to demands stigma and discrimination has been under- about sexuality in Brazil—some sectors of the from gay rights groups was rapid and positive, stood as central to the response to HIV and religious community may make moral judg- and strong links were forged between the AIDS, and it has been waged consistently ments, just as some in the medical profes- health department and NGOs.13 , 2 0 The pro- through the development of partnerships be- sions may reduce sexuality to decontextual- gram that emerged combined prevention, tween government and civil society.45 ized risk behaviors—but there is a capacity treatment, surveillance, and support for The experience in most other countries dif- for HIV prevention programs to address sex- human rights. While treatment and surveil- fers from that of Brazil. Centralization is dom- uality more openly than in most other coun- lance remained governmental functions, inant in most health ministries, and it is not tries.14 It is notable that condom sales and NGOs increasingly took the lead in the pre- uncommon for regional and municipal de- distribution have risen dramatically in the vention of HIV and the promotion of human partments to be responsible for implementing general population, and there are data that rights. The São Paulo AIDS program became programs without receiving funding to deliver suggest that condom use among HIV-positive the model for other states and ultimately the services. It is less common for govern- people has increased as well.50 Openness helped shape the NAP.13 ments to welcome the input and involvement about sexuality and the diversity of gender The Brazilian response to AIDS thus of NGOs, although a nominal NGO presence and sexual identities have helped to break emerged from the bottom up. It has been is required by almost all international funding down the stigma surrounding both homosex- characterized by an active collaboration be- agencies. Even fewer governments accept uality and HIV. tween government and NGOs, as well as by their responsibility to promote and defend the Nowhere is the importance of sexual cul- mobilization of activist political support and human rights of PLWHA; on the contrary, ture in Brazil as clear as in the ways in which commitment within the machinery of the governments often contribute to civil and prevention programs have been able to ad- state itself, particularly on the part of local human rights abuses through criminalization dress sexuality, focusing on condom promo- service providers in the public health system. of risk behaviors (sodomy laws, drug laws, tion while also combating stigma and discrim- While the dynamic between centralization prostitution) and punitive policies toward ination. The public service announcements and decentralization within the NAP has PLWHA in prisons. sponsored by the NAP have been among the fluctuated over time, there remains room for most explicit of any governmental informa- local initiatives, and the alliance with NGOs Culture tion campaign in the world. Condom use has remains strong. The commitment to human rights and the been promoted relentlessly, female as well as Equally important, through a succession of early emphasis placed on solidarity as central male condoms have been widely distributed different presidential administrations, is that to the response to HIV/AIDS in Brazil, while by the Brazilian government, and studies of the Brazilian AIDS Program has managed to articulated as a response to the military au- sexual behavior have demonstrated significant sustain a consistent commitment to strength- thoritarian regime and social inequality, also increases in the adoption of condom use ening previously marginalized communities, is clearly deeply rooted in a long-standing across population groups (especially among to defending their rights, and to articulating emphasis on solidarity in Brazilian culture. young people).50 Public information cam- respect for diversity as a key component of Principles of solidarity and reciprocity have paigns also have focused on the need to com- official government policy. Organizations rep- long been understood as central to the moral bat stigma and support sexual diversity, with resenting sex workers; drug users; gay and economy of the poor in Brazilian society.46 1 recent campaign focusing on the need for lesbian, bisexual, and transgender popula- Solidarity among family members and neigh- parents to accept and support children who are tions; PLWHA; and other groups affected by bors is a key element of the survival strate- homosexual. These mass media approaches the epidemic have received significant fund- gies traditionally employed by poor people have been accompanied by significant levels ing from the government.25 Support has been with little access to services and social welfare of government support for community-based provided for more than a decade now for benefits in Brazil. These same principles have prevention programs among men who have legal aid work carried out by NGOs working been extremely important to critical societal sex with men, sex workers, young people, and

1168 | Public Health Matters | Peer Reviewed | Berkman et al. American Journal of Public Health | July 2005, Vol 95, No. 7  PUBLIC HEALTH MATTERS 

other populations perceived to be at elevated of HIV infections in the country were esti- access and stay in care, and active input from risk of HIV infection. mated to be the result of needle sharing and the target group itself helped create an effec- Just as Brazil has confronted the interna- related sexual transmission.57 tive program.1 tional community around issues of treatment In Brazil, as elsewhere, the initial response access, it has also resisted international pres- of the public health system was constrained PROGRAMMATIC LESSONS sure with regard to prevention programs. by criminal justice authorities who sought to While the Brazilian NAP has acknowledged interpret the issue as the province of the jus- While national and local context will funda- that reducing the number of sexual partners tice system rather than the public health sys- mentally shape a country’s response to its can decrease an individual’s risk of infection, tem. Early attempts to implement needle ex- AIDS epidemic, there are programmatic ele- it has also recognized that many people, es- change programs in the city of Santos and the ments that public health planners must address pecially women, are not always able to con- state of São Paulo were met with extreme rein all countries. A critical analysis of the Brazil- trol the multiple relations of their primary sistance, including threats to imprison public ian approach, both its strengths and weak- partners. The NAP has therefore been firm health officials promoting needle exchange nesses, may give insights helpful to others. in putting condom use at the center of its programs. By the early 1990s, however, a program.51 This position has caused tension process of negotiation had begun that in- Prevention with some international agencies, such as volved representatives of the Ministry of The SUS, for all its accomplishments, has USAID, which came close to closing its Health and the Ministry of Justice, with not been the primary vehicle for HIV pre- AIDS prevention activities in Brazil because behind-the-scenes support from a number of vention efforts in Brazil. From the earliest of the Brazilian refusal to adopt USAID’s United Nations agencies. The result was the days of the epidemic, civil society organiza- “ABC” (Abstinence, Be faithful, Condoms) establishment of a task force to develop a tions (CSOs), in alliance with local govern- prevention strategy, a strategy that explicitly national policy to respond to HIV and injec- mental AIDS programs, have led the devel- prioritizes both abstinence and fidelity over tion drug use. As part of the more general opment and implementation of most and above an emphasis on promoting con- program of prevention initiatives developed prevention programs. The national govern- dom use.52–54 for support from the World Bank, a set of ment was slow to respond throughout the Culture cannot be reduced to “best prac- pilot needle exchange and harm reduction 19 8 0s, and CSOs, primarily the newly tices” and transferred from one social reality programs were established and implemented formed AIDS NGOs, emerged as the most to another. In many countries, AIDS preven- in key cities across the country.1 vocal and active critics of the federal govern- tion efforts have been blocked by societal and In 1998, the state of São Paulo passed leg- ment’s HIV policies. It was only as redemoc- governmental leaders claiming that discussion islation authorizing the health department to ratization proceeded and key personnel from of sexuality is antithetical to traditional cul- buy and distribute sterile needles and sy- some of the progressive state and municipal ture. This position assumes that culture is ringes. The success of this publicly sponsored public health departments were brought into static, unresponsive to changing conditions or program led to similar legislation in other the Federal Health Ministry that collabora- focused intervention. The Brazilian experi- states, culminating in modifications to the tions at the national level developed.58 The ence, as well as that of Uganda, Senegal, and Brazilian Law on Drugs that authorized the lessons from the local initiatives based on a number of other countries, disproves that Ministry of Health to implement national nondiscrimination and solidarity began to generalization.55,56 harm reduction programs. Input from injec- shape the NAP.13 tion drug users has helped shape these pro- Perhaps the most crucial development in Harm Reduction grams. Whether as a direct result of these HIV control efforts in Brazil emerged from Finally, building on many of the same prin- policies and programs or not, the percentage the prolonged (1992–1994) negotiation be- ciples discussed earlier, Brazil’s response to of AIDS cases linked to injection drug use tween Brazil and the World Bank over the injection drug users provides another key ex- had declined to 11% in 2003.6 terms of a large loan to help finance its re- ample of the important ways in which the It was a longer and more difficult process sponse to the HIV epidemic. The successful government’s approach has differed from the to build a public and governmental consensus negotiation of the US$160 million loan re- responses of other governments and yet has that drug use should be addressed as a public quired active collaboration across many gov- still achieved positive results. Injection drug health rather than criminal justice problem. ernmental ministries, active participation of use was limited in Brazil prior to the 1980s; As with other aspects of the NAP, certain CSOs through the NAP, and support from a however, as international drug control efforts themes and processes underlie that change: wide range of political parties; it also rein the highland Andean region intensified local initiatives at the municipal and state quired US$90 million in matching funds over the course of that decade, Brazil’s largely levels shaped the national program, a strong from the Brazilian Treasury. (Throughout this uncontrolled border became an attractive emphasis on human rights was the context process, Brazil refused to conform to the route for drug trafficking. Subsequently, rates for reaching out to an extremely marginalized World Bank demands that it halt the free of HIV infection linked to injection drug use population, free and universal HIV treatment distribution of azidothymidine, or AZT, a began to rise. By the mid-1990s, almost 30% was an incentive for injection drug users to program it had started several years before.)

July 2005, Vol 95, No. 7 | American Journal of Public Health Berkman et al. | Peer Reviewed | Public Health Matters | 1169  PUBLIC HEALTH MATTERS 

The total program of US$250 million over 5 in many other countries. However, the pre- Universal access. That Brazil’s treatment years financed a large-scale control effort ca- vention program has not succeeded in stop- program is free has received considerable at- pable of a major impact throughout much of ping increasing rates of HIV infection among tention, but less publicized is the fact that it the country.25,59 the poorest strata of society, particularly poor is universal. Universal distribution, in contrast There are important lessons to be drawn women. HIV prevention programs have not to free medication solely in the public health from the experience of the first World Bank yet been integrated into other aspects of sector, created many more points of access to loan, as well as 2 subsequent loans. women’s health programs, such as family treatment and allowed more rapid scale-up. Broad-based political support. Key individu- planning, treatment of sexually transmitted Quite intentionally, it also eliminated the fi- als in government committed to an aggressive infections, and routine gynecological care. nancial incentive for such corrupt practices response to HIV were politically adept Programs for the prevention of mother-to- as theft from central supplies or resale of enough to use the loan negotiation process to child transmission have inadequate coverage medication by individuals. Free and universal build support for the HIV control program for pregnant women despite the availability distribution became a proactive solution to across a wide range of governmental and of testing and treatment.60 Perhaps most im- the potential development of a domestic black nongovernmental sectors. Control of HIV be- portant, at least in terms of long-term sustain- market for antiretrovirals. came a national priority, even if implementa- ability, while the overall NAP addresses pre- Local manufacture. The Brazilian program tion efforts largely remained within the Min- vention, care, and treatment, the SUS has of universal, free access is financially viable in istry of Health. continued to view its primary responsibility large measure because of Brazil’s capacity for Adequate funding. A large-scale prevention as care and treatment. Prevention efforts have local manufacture of pharmaceuticals. Local program capable of a major impact on the still not been fully integrated with care and manufacture, particularly but not exclusively epidemic requires equally large-scale funding. treatment at the programmatic level in the of generics, creates systemic downward pres- As in many other areas of AIDS program- SUS, and there is still not an effective inter- sure on patented drug prices and, impor- ming, half-hearted and inadequately funded face between the SUS and the CSOs involved tantly, avoids the currency fluctuations that programs are destined to fail. in prevention efforts. make it extremely difficult for importing Human resources. Not only financial re- countries to project drug costs effectively. sources, but also human resources have been Treatment The domestic pharmaceutical manufacturing essential. There was a successful training pro- The Brazilian program of free, universal ac- capacity strengthens the government’s hand gram for human resources in and out of gov- cess to antiretroviral treatment has had a drain its negotiations with the multinational ernment, which had begun even before the matic impact on morbidity and mortality from pharmaceutical companies by enabling the World Bank loans but was intensified and ex- AIDS in Brazil and has gained considerable government to issue a compulsory license if panded dramatically after the loans. There is international recognition for its efforts. Since companies abuse their patent monopoly by an emphasis on health educators and a partic- the Durban International AIDS Conference, pricing the drug out of the range of the ular focus on peer educators, who serve as a the Brazilian government has offered free Brazilian market. natural link between the most vulnerable technical assistance to other countries devel- Capacity to use complex therapies. The communities and the health system. oping similar programs. The following sec- Brazilian program is proof that health care CSO involvement. CSOs were involved tions provide a few lessons that may be of systems outside the wealthy countries can throughout the process and helped shape a value to other countries. effectively use complex therapies such as anti- prevention program that funded a wide Integration of treatment and prevention. The retroviral treatment. Although less tangible range of NGO-led initiatives. While there is integration of care and treatment was funda- than the manufacture of generic drugs, using some concern that governmental funding mental to the Brazilian program even before complex therapies is as important a lesson for of NGOs compromises their willingness and the development of effective antiretroviral other countries with weak health care sys- capacity to criticize the government, there is treatment. When AZT became available in tems. The Brazilian treatment program was little question that it has made it possible to the late 1980s, the state of São Paulo made initiated as a vertical program guided by the reach many of the most vulnerable people small quantities available at no cost. The NAP with its own administration, staff, logisti- in Brazil.43 promise of treatment gave an incentive for cal systems, and budget. This has resulted in HIV control efforts in Brazil are decentral- more at-risk individuals to be tested and gave ongoing difficulties in creating horizontal ized and multifaceted, but there are some doctors an incentive to report AIDS cases, linkages within the SUS, but realistically it significant generalizations, both positive and thus improving surveillance and prevention was the only way to rapidly establish and negative, about the prevention program. For programs. The success of the São Paulo free scale up the program. One of the major chal- example, the mass media (press, radio, televi- drug program led to its adoption by other lenges in the last 4 to 5 years has been to sion) has played a positive role in control ef- states and ultimately by the federal govern- decentralize this program within the SUS and forts. Generally, stigma and stereotyping have ment. While AZT monotherapy was of lim- at the state and municipal levels—a challenge been avoided, and there is an openness about ited value, it did create the principle that that is accentuated owing to the continental sexuality and condom use that is not present PLWHA have the right to free treatment.25,41 size of the country.61

1170 | Public Health Matters | Peer Reviewed | Berkman et al. American Journal of Public Health | July 2005, Vol 95, No. 7  PUBLIC HEALTH MATTERS 

Creating international alliances. The grams in Brazil seek to emulate and that through the International Core of the HIV Center for Brazilian government has acted proactively other countries look to for inspiration as they Clinical and Behavioral Studies at the New York State Psychiatric Institute and Columbia University (sup- and strategically to protect the NAP from seek to develop their own unique responses ported by center grant P30 MH43520 from the Na- international pressures. As mentioned previ- to the challenges posed by the epidemic. tional Institute of Mental Health; principal investigator ously, Brazil firmly resisted World Bank de- and center director, A.A. Ehrhardt; international core director, R. Parker). mands that it drop its free distribution of Conclusions AZT as a condition of the first loan agree- Controlling the HIV/AIDS pandemic will References ment; it subsequently resisted threats from likely be the greatest challenge to public health 1. National Coordination for STD and AIDS. The the United States to challenge Brazil’s in the 21st century. HIV is a minuscule bit of Brazilian Response to HIV/AIDS. Brasília, Brazil: Min- generic manufacture of some antiretrovirals RNA, but this viral event causes a profoundly istry of Health; 2000. before the World Trade Organization. The human phenomenon. Modifying intimate ex- 2. Parker R, Passarrelli CA, Terto V, Pimenta C, Berkman A, Muñoz-Laboy M. The Brazilian response NAP has also resisted pressure to change its periences, changing established social rela- to HIV/AIDS: assessing its transferability. Divulgação open approach to the prevention of sexual tionships, and challenging global inequalities em Saúde para Debate. 2003;27:140–142. transmission. Brazil has allied itself with are all part of the response to HIV. 3. Ministry of Health of Brazil, National Coordina- other developing and poor countries to cre- Brazil has done all of these things with some tion for STD and AIDS. Drugs and AIDS. The Brazilian Response to HIV/AIDS: Best Practices. Brasilia, Brazil: ate a global consensus more favorable to success; insights into the process can hopefully Ministry of Health; 2000:116–130. health initiatives; these countries have led be of some value to all of us grappling with 4. World Health Organization. The World Health efforts to challenge the restrictive interpreta- these concerns. There is a final lesson from Report 2004: changing history. Available at: http:// tion of the TRIPS (Trade Related Aspects of Brazil that is worthy of notice: the NAP has be- www.who.int/whr/2004/en/report04_en.pdf. Ac- cessed October 7, 2004. Intellectual Property) agreement, succeeded come a source of national pride for the Brazil- 5. Antonio de Ávila Vitória M. The experience of in having the United Nations Human Rights ian people. It is “owned” by the government, providing universal access to ARV drugs in Brazil. Di- Commission declare access to treatment part civil society, the media, and, most importantly, vulgação em Saúde para Debate. 2003;27:247–264. of the human right to health, and helped people living with HIV. Solidarity and pride, it 6. Coordenação Nacional de DST/AIDS. Boletim forge a bloc of nations that made the right seems, may be the most effective counter to Epidemiológico AIDS, Ano XVIII. No. 01–01a–26a to treatment a prominent part of the Con- stigma. To control HIV, we must first admit semanas epidemiológicas. Brasília, Brazil: Ministério da Saúde; January–June 2004. sensus Statement from the UN General As- that the problem belongs to all of us. 7. Epidemiological Fact Sheets on HIV/AIDS and sembly Special Session on AIDS. Sexually-Transmitted Infections, 2004, Update: Brazil. Like Brazil’s prevention record, Brazil’s suc- Geneva, Switzerland: Joint United Nations Programme About the Authors cess in integrating care and treatment into a on HIV/AIDS; 2004. The authors are with the Department of Sociomedical Sci- 8. Barreira D. Números e Tendências Atuais da Epi- unified approach to the control and mitigation ences and the Center for Gender, Sexuality and Health, demia do HIV e AIDS. Brasília, Brazil: Ministério da Mailman School of Public Health, Columbia University, of the HIV/AIDS epidemic is thus impressive. Saúde; July 2004. and the International Core of the HIV Center for Clinical No one, not even the greatest supporters of and Behavioral Studies, New York State Psychiatric Insti- 9. Parker R, Rochel de Camargo K Jr. Pobreza e the Brazilian program, would suggest that tute and Columbia University, New York, NY. Alan Berk- HIV/AIDS: aspectos antropológicos e sociológicos. these achievements have come easily or that man is also with the Department of Epidemiology, Mail- Cadernos de Saúde Pública. 2000;16(suppl 1):89–102. man School of Public Health, Columbia University. Vera there is not still much important work to be 10. Daniel H, Parker R. Sexuality, Politics and AIDS in Paiva is also with NEPAIDS, the Nucleus for AIDS Pre- Brazil: In Another World? London, England: Falmer done to strengthen existing programs and to vention Studies, Institute of Psychology, University of São Press; 1993. ensure their sustainability over time. Political Paulo, São Paulo, Brazil. Vera Paiva and Richard Parker are also with the Brazilian Interdisciplinary AIDS Associ- 11.Parker R. Introdução. In: Parker R, ed. Políticas, tensions that sometimes threaten to disrupt ation, Rio de Janeiro, Brazil. Instituições e AIDS: Enfrentando a Epidemia no Brasil. service provision are still all too common, Requests for reprints should be sent to Richard Parker, Rio de Janeiro, Brazil: ABIA; 1997:7–15. particularly when different political parties or PhD, Department of Sociomedical Sciences, Mailman School 12.Teixeira PR. Políticas públicas em AIDS. In: of Public Health, Columbia University, 600 West 168th St, Parker R, ed. Políticas, Instituições e AIDS: Enfrentando factions control municipal, state, and national New York, NY 10032 (e-mail: [email protected]). a Epidemia no Brasil. Rio de Janeiro, Brazil: ABIA; health programs in Brazil’s federalist system This article was accepted January 3, 2005. 19 97:43–68. 24 of government. Logistics in relation to the 13. Galvão J. AIDS e activismo: o surgimento e a distribution of medications is still often un- Contributors construção de novas formas de solidariedade. In: Parker even, sometimes requiring aggressive inter- A. Berkman and R. Parker developed the initial draft R, Bastos C, Galvão J, Stalin Pedrosa J, eds. A AIDS No Brasil. Rio de Janeiro, Brazil: ABIA; 1994:341–350. ventions on the part of CSOs as well as the of this article. All other authors helped to further refine the ideas and contributed to drafts of the article. 58 14 .Paiva V. Beyond magic solutions: prevention of NAP. At least thus far, however, these chal- HIV and AIDS as a process of psychosocial emancipa- lenges have been met with consistent success, tion. Divulgação em Saúde para Debate. 2003;27: Acknowledgments 192–2003. in large part through partnership and collabo- This article draws on data collected with support from ration between the Brazilian government and the National Science Foundation (grant 1025–0440; 15.Weyland K. Social movements and the state: the politics of health reform in Brazil. World Dev. 19 95; civil society. Brazil’s response to HIV and principal investigator, R. Parker) and analyses developed with support from the Ford Foundation (grant 23(10):1699–1712. AIDS at every level has increasingly emerged BCS-9910339; principal investigator, R. Parker). Addi- 16.Weyland K. Obstacles to social reform in Brazil’s as a model program that other health pro- tional support for writing and analysis was provided new democracy. Comp Polit. 1996;29(1):1–22.

July 2005, Vol 95, No. 7 | American Journal of Public Health Berkman et al. | Peer Reviewed | Public Health Matters | 1171  PUBLIC HEALTH MATTERS 

17.Weyland K. The Brazilian state in the new 37.Deng L, Kostner M, Young C. Democratization and HIV decline and behavioral risk avoidance in Uganda. democracy. J Interamerican Stud World Aff. 19 97;39(4): Structural Adjustment in Africa in the 1990s. Madison: Science. 2004;304:714–718. 63–94. University of Wisconsin Press; 1991. 56. Rosenfield A, Figdor E. Where is the M in MTCT? 18.Waitzkin H, Iriart C, Estrada A, Lamadrid S. So- 38. Smith M. False hope or new start? The Global The broader issues in mother-to-child transmission of cial medicine in Latin America: productivity and dan- Fund to Fight HIV/AIDS, TB and Malaria. Oxfam HIV. Am J Public Health. 2001;91:703–704. gers facing the major national groups. Lancet. 2001; Briefing Paper, 2002. Available at: http://www.oxfam. 358:315–323. org/eng/pdfs/pp0206_false_hope_or_new_start.pdf. 57. Castilho EA, Chequer P. Epidemiologia do HIV/ Accessed May 12, 2005. AIDS no Brasil. In: Parker R, ed. Políticas, Instituições e 19.Paiva V, Ayres JR, Buchalla CM, Hearst N. Build- AIDS: Enfrentando a Epidemia no Brasil. Rio de Janeiro, ing partnerships to respond to HIV/AIDS. AIDS. 39.Raupp Rios R. Legal responses to the HIV/AIDS Brazil: ABIA; 1997:17–42. 2002;16(suppl 3):76–83. epidemic in Brazil. Divulgação em Saúde para Debate. 2003;27:228–238. 58. Passarelli CA, Parker R, Pimenta C, Terto V Jr. 20. Arretche M. Decentralização das Políticas Sociais AIDS e Desenvolvimento: Interfaces e Políticas Públicas. 40. Latin American and Caribbean Council of AIDS no Estado de São Paulo. São Paulo, Brazil: Edições Rio de Janeiro, Brazil: ABIA; 2003. Fundap; 1998. Service Organizations (LACCASO). Report on Access to Comprehensive Care, Antiretroviral Treatment (ARVs) and 59. World Bank. Acordo de Empréstimo (Projeto de 21. Monteiro de Andrade LO. SUS Passo a Passo: Human Rights of People Living With HIV/AIDS in Latin Controle da AIDS e das DST) entre a República Federa- Normas, Gestão e Financiamento. São Paulo, Brazil: America. Washington, DC: Inter-American Commission tiva do Brasil e o Banco Mundial, 16 March 1994. Editora Hucitec Ltda; 2001. on Human Rights; October 2002. Brasília, Brazil: Brazilian Ministry of Health; 1994. 22. Elias PE, Cohn A. Health reform in Brazil: lessons 41.Pinheiro R, Araújo de Mattos R. Os Sentidos da 60. Barbosa RM, Di Giacomo do Lago T. AIDS e dire- to consider. Am J Public Health. 2003;93:44–48. Integralidade: Na Atenção e no Cuidado à Saúde. Rio de itos reprodutivos: para além da transmissão vertical. In: 23. Bastos FI. A Feminização da Epidemia de AIDS Janeiro, Brazil: UERJ; 2001. Parker R, ed. Políticas, instituições e AIDS: Enfrentando no Brasil: Determinantes Estruturais e Alternativas de 42. Pinheiro R, Araújo de Mattos R. Construção da a Epidemia no Brasil. Rio de Janeiro, Brazil: ABIA; Enfrentamento. Rio de Janeiro, Brazil: ABIA; 2001. Integralidade: Cotidiano, Saberes e Práticas em Saúde. 19 97:163–175. Coleção ABIA: Saúde Sexual e Reprodutiva publica- Rio de Janeiro, Brazil: UERJ; 2003. tion no. 3. 61.Westphal MF. Gestão de Serviços De Saúde: Decen- 43. Passarelli CA, Terto V Jr. Non-governmental or- tralização, Municipalização do SUS. São Paulo, Brazil: 24.Parker R. Building the foundations for the re- ganizations and access to anti-retroviral treatments in Editora da Universidade de São Paulo; 2001. sponse to HIV/AIDS in Brazil: the development of Brazil. Divulgação em Saúde para Debate. 2003;27: HIV/AIDS policy, 1982–1996. Divulgação em Saúde 252–264. para Debate. 2003;27:143–183. 44. Arretche M. Graus de decentralização na munici- 25. Galvão J. AIDS no Brasil. São Paulo, Brazil: palização do atendimento básico. In: Arretche M, ed. Editora 34; 2000. Estado Federativo e Políticas Sociais: Determinantes de 26. Galvão J. As respostas das organizações não- Decentralização. Rio de Janeiro, Brazil: Revan; 2000: governamentais brasileiras frente à epidemia de HIV/ 197–239. AIDS. In: Parker R, ed. Políticas, Instituições e AIDS: 45. Parker R. Na Contramão da AIDS: Sexualidade, Enfrentando a Epidemia no Brasil. Rio de Janeiro, Intervenção, Politica. Rio de Janeiro, Brazil: ABIA; 2000. Brazil: ABIA; 1997:67–108. 46. Zaluar AM. Exclusion and public policies: theoret- 27. Paiva V. Em tempos de AIDS. São Paulo, Brazil: ical dilemmas and political alternatives. Revista Summus; 1992. Basileira de Ciências Sociais. 2000;1:25–42. 28.Parker R. Políticas, Instituições e AIDS: Enfrentando 47. Zaluar AM. Condomínio do Diabo. São Paulo, a Epidemia no Brasil. Rio de Janeiro, Brazil: Zahar/ABIA; Brazil: Editora Brasiliense; 1996. 19 97. 48.Fonseca C. Mãe é uma só? Reflexões em torno de 29.Ventura M. Strategies to promote and guarantee alguns casos brasileiros. Psicologia USP. 2002;13(2): the rights of people living with HIV/AIDS. Divulgação 49–68. em Saúde para Debate. 2003;27:239–246. 49.Parker R. Bodies, Pleasures and Passions: Sexual 30. Heywood M. Current developments: preventing Culture in Contemporary Brazil. Boston, Mass: Beacon mother-to-child transmission in South Africa. S Afr Press; 1991. J Health. 2003;19:278–315. 50. Berquó E. Comportamento Sexual da População 31. Barnett T, Whiteside A. AIDS in the Twenty-First Brasileira e Percepções do HIV/AIDS. Brasília, Brazil: Century: Disease and Globalization. New York, NY: Pal- Ministério da Saúde; 2000. grave Macmillian; 2002. 51. Project Appraisal Document on a Proposed Loan 32. World Bank. World development indicators, GINI in the Amount of US $165 Million. Washington, DC: Index. 2004. Available at: http://www.worldbank.org/ World Bank; 1998. Report 18338-BR. data/wdi2004. Accessed October 10, 2004. 52. US–Brazil Joint Venture on HIV/AIDS in Luso- 33. Araújo de Mattos R, Terto V Jr, Parker R. World phone Africa. Available at: http://www.whitehouse. Bank strategies and the response to AIDS in Brazil. gov/news/releases/2003/06/20030620–14.html. Divulgação em Saúde para Debate. 2003;27:215–246. Accessed June 20, 2003. 34. Mamdani M. Citizen and Subject: Contemporary 53. Girard F. Global Implications of US Domestic and Africa and the Legacy of Late Colonialism. Princeton, International Policies on Sexuality. New York, NY: Inter- NJ: Princeton University Press; 1996. national Working Group on Sexuality and Social Policy; 35. Young C. The African Colonial State in Comparative June 2004. Working paper no. 1. Perspective. New Haven, Conn: Yale University Press; 54.Agência nacional de AIDS. Estados Unidos 1994. cancelam grande programa de combate à AIDS no 36. Amin S. Underdevelopment and dependence in Brasil. Available at: http://www.agenciaaids.com.br. Ac- black Africa: origins and contemporary forms. J Modern cessed September 2003. Afr Stud. 1972;10(4):503–524. 55. Stoneburner RL, Low-Beer D. Population-level

1172 | Public Health Matters | Peer Reviewed | Berkman et al. American Journal of Public Health | July 2005, Vol 95, No. 7  PUBLIC HEALTH MATTERS 

Which Patients First? Setting Priorities for Antiretroviral Therapy Where Resources Are Limited

| Laura J. McGough, PhD, Steven J. Reynolds, MD, MPH, Thomas C. Quinn, MD, MS, and Jonathan M. Zenilman, MD

these suggestions, however, it is important for The availability of limited funds from international agencies for the purchase donors to recognize patient selection as part of antiretroviral (ARV) treatment in developing countries presents challenges, especially in prioritizing who should receive therapy. Public input and the proof the process of introducing or expanding tection of human rights are crucial in making treatment programs equitable and access to ARV therapy and long-term strate- accountable. By examining historical precedents of resource allocation, we aim gic planning. to provoke and inform debate about current ARV programs. Decisionmaking occurs in specific historical Through a critical review of the published literature, we evaluate 4 precedents and cultural contexts: each recipient country for key lessons: the discovery of insulin for diabetes in 1922, the release of pen- faces a unique set of challenges in deciding icillin for civilian use in 1943, the development of chronic hemodialysis programs how to allocate resources. Because of these in 1961, and current allocation of liver transplants. We then describe current widely disparate contexts, both the Global rationing mechanisms for ARVs. (Am J Public Health. 2005;95:1173–1180. Fund and PEPFAR have given recipient coun- doi:10.2105/AJPH.2004.052738) tries the responsibility for patient selection. We do not disagree with the decision to as- sign responsibility for patient selection to re- Currently, at least 5.8 million people in devel- programs. Open, public discussions about cipient countries. Instead, we recognize the oping countries urgently need antiretroviral patient selection—or rationing—have largely complex responsibilities and burdens that (ARV) therapy, but only approximately 12% been avoided in recipient countries. It is pos- countries face in expanding access to ARV are receiving it.1 Although the number of sible that government officials fear the poten- treatment. Patient selection is a difficult issue people receiving ARV therapy worldwide in- tially divisive consequences of open discus- involving competing demands between eq- creased from 440000 in June 2004 to ap- sions about who receives access to a life- uity, urgency, efficiency, and other factors. proximately 700000 in December 2004, saving medication available in only limited In order to facilitate discussion about pa- ARV therapy is still reaching only a fraction quantities.4 However, avoiding a decision tient selection in recipient countries, we have of the people who need it, especially in coun- about rationing does not mean that decisions critically analyzed the history of patient selec- tries such as India (4%), the Russian Federa- are not made. “Passive decisions”—that is, tion for scarce medical resources in the US tion (3%), and many sub-Saharan African limiting access to patients who have already context. Our goal is not to provide a one-size- countries (Ethiopia, 5%; South Africa, 7%).2 tested HIV positive or live near a clinic site— fits-all model based on experiences unique to The delivery of ARV treatment in developing favor those with economic, political, or social the United States, but rather to use these his- countries through programs such as the Global power.5 Trust and social capital (trust demon- torical experiences as an entry into the com- Fund to Fight AIDS, Tuberculosis and Ma- strated by mutual reciprocity and cooperation plex issues regarding resource allocation. laria; President Bush’s US Emergency Plan among members of society) have been identi- Policymakers in other countries may want to for AIDS Relief (PEPFAR); and other private fied as key elements in the successful imple- evaluate their own histories of patient selec- sector or nongovernmental organization pro- mentation of major public health intervention and scarce resource allocation, since cur- grams presents tremendous challenges and tions. Resources that benefit only 1 group, or rent reactions to ARV patient selection will opportunities, especially in prioritizing who 1 subset of patients, have the potential to dis- be shaped partially by each country’s histori- should receive therapy. The World Health Or- rupt community relations, promote conflict, cal experiences. ganization’s (WHO’s) goal of treating 3 mil- and undermine public health programs. lion people by 2005 would reach only half of Because the limited ability to provide care METHODOLOGY those who urgently need medication3 and an raises ethical issues about patient selection, even smaller fraction of those who could po- Dr Peter Piot, executive director of the Joint We evaluated the following 4 historical tentially benefit. United Nations Programme on HIV/AIDS, precedents of rationing of a scarce medical has called for the establishment of national 6 resource: the discovery of insulin for diabetes WHY IS PATIENT SELECTION A ethics panels in recipient countries, while in 1922, the release of penicillin for civilian PROBLEM? WHO has commissioned a series of back- use in 1943, the development of chronic he- ground papers and a guidance document on 7 modialysis programs in 1961, and current ap- Developing countries now face the chal- ethics, equity, and access to ARV therapy. In proaches to allocating liver transplants. Each lenge of implementing these ARV delivery order for beneficiary countries to implement

July 2005, Vol 95, No. 7 | American Journal of Public Health McGough et al. | Peer Reviewed | Public Health Matters | 1173  PUBLIC HEALTH MATTERS 

of these cases involved the availability of new cians within Canada and from the United on penicillin’s efficacy did not exist, but treat- therapies for formerly fatal diseases. States flooded Banting’s laboratory with re- ment of patients with rare diseases could po- The limiting factor for both insulin and quests for insulin. Emotional, political, and tentially advance knowledge. One patient suf- penicillin was the ability to manufacture large personal appeals often influenced who re- fering from what was regarded as a rare form quantities of the medication. With hemodialy- ceived treatment; the patient of a doctor who of blood disease was offered penicillin to eval- sis, however, the constraint was limitations of served in the Army with Banting, as well as uate the drug’s efficacy against this infection, financial resources to pay for the therapy, the politically well-connected daughter of the thereby possibly advancing the war effort.12 similar to the constraints related to delivery US secretary of state, received insulin while Unlike researchers’ rationing of insulin, the of ARV therapies. Pharmaceutical companies others were denied treatment.9 COC’s decisionmaking about which patients are able to manufacture sufficient quantities received penicillin remained uninfluenced by of ARVs; the limitation is the amount of THE PROCESS IS AS IMPORTANT criteria such as a patient’s political ties, media money available for these medications in ad- AS THE CRITERIA attention, or number of letters written on be- dition to the limitations of infrastructure and half of the patient. The potential for clinical health personnel to treat patients in develop- The therapeutic value of penicillin against efficacy overrode patients’ political, financial, ing countries. Both diabetes and kidney dis- a variety of infections was recognized in the or social status. Despite this apparent success, ease are chronic illnesses which were not period 1941 to 1943, when war provoked the committee nonetheless endured a barrage “cured” by insulin and hemodialysis, similar renewed research into the drug’s efficacy. of criticism from the media and the public, to the case with HIV/AIDS, where ARV ther- Initially, the drug was limited to military use. which often viewed the committee’s chairman, apies do not provide a cure but a way to Determined to avoid the problems connected Dr Chester Keefer, who served as chief of manage the disease. with insulin rationing, a panel of experts de- medicine at Boston University, as coldhearted We critically review the published, peer- veloped medical criteria to handle rationing. and unresponsive. reviewed literature on these medical discover- Rationing for the military was handled sepa- The public became angry with the COC ies, specifically concentrating on identifying rately from civilian rationing according to 2 and Dr Keefer for 2 reasons. First, the public the approach to resource allocation decisions sets of criteria: (1) the efficacy of the drug regarded rationing as a political process from and the key determinants of those decisions. against the disease in question and (2) the which it had been excluded. Although the We then compare these case studies with the speed of treatment in returning soldiers to COC was a civilian committee independent situation of AIDS and ARV therapy. active duty. Since penicillin was effective of both the military and the government, it against gonorrhea and patients recovered was perceived by the public as being either DECISIONS CANNOT BE AVOIDED quickly, gonorrhea patients in military ser- the army or the federal government, which vice received priority. As production in- had unnecessarily intruded into civilian life.13 Between 1921 and 1922, Fredrick Banting creased, it was released to the civilian sector Many members of the public disagreed with and Charles Best discovered insulin as treat- under strict control.10 the criteria established, since civilian needs ment for diabetes mellitus and learned how From January 1942 to May 1944, the came second to military needs. When Norris to produce it from livestock pancreas. With Committee on Chemotherapeutic and Other Higgins, a 29-year-old physician from Con- this discovery, diabetes was transformed from Agents (COC) handled the rationing of peni- necticut, became sick and then died in the a fatal illness into a chronic, manageable dis- cillin to civilians. The COC was responsible fall of 1943, family and friends criticized the ease. As scientists and researchers, Banting for investigating the clinical uses of penicillin, committee for having denied him penicillin. and Best never anticipated that they would especially the application of the drug for In a letter written October 22, 1943, former be in the position of allocating medications. wound infections. Officially independent of Congressman William Fitzgerald expressed Consequently, they developed no concrete government, the committee was composed his frustrations that a worthy young man plan until a crisis developed. Flooded by re- of leading academic physicians responsible would die while soldiers who contracted quests for the limited quantities of the new for developing clinical guidelines for penicillin syphilis or gonorrhea had access to penicillin: drug, Banting himself decided which termi- use. Civilian demand for penicillin skyrock- “I think it is a crime that the Health Depart- nally ill patients would receive insulin.8 eted during the summer of 1943, when the ment in Washington [sic] refused to release With neither experience in health policy public became aware of the drug’s usefulness. any of this drug for his benefit, and then I nor any reliable precedents for guidance, The COC’s allocation decisions were made read in the paper this drug is available for Banting resorted to subjective criteria in de- according to severity and type of infection, men who have been careless in their lives ciding whom to treat. One third of the insulin with priority given to acute illnesses caused and have contracted a dreadful disease. . . .”14 went to his own private practice, another one by sulfonamide-resistant streptococci, gono- Although Fitzgerald misunderstood which third to his clinic, and the remaining one third cocci, and staphylococci, which had proven agency was responsible for rationing penicil- to Toronto General Hospital and the Hospital responsive to penicillin in clinical trials.11 lin, he did understand that the public had for Sick Children. Because the discovery was Research value was also part of the criteria exercised no authority in establishing crite- well publicized, patients’ families and physi- for patient selection. For rare diseases, data ria for the allocation of penicillin. Not all

1174 | Public Health Matters | Peer Reviewed | McGough et al. American Journal of Public Health | July 2005, Vol 95, No. 7  PUBLIC HEALTH MATTERS 

members of the public agreed that every feasible for chronic, end-stage kidney disease mittee then chose according to 2 other sets of soldier’s right to medication preceded the when Belding Scribner and his colleagues at criteria: who had the largest number of de- right of any civilian. the University of Washington developed a pendents requiring state financial assistance The second reason for dissatisfaction with shunt that allowed patients to be easily if that patient could not work and who “de- 18 the COC was that the public did not always treated 3 times a week. (Although other served” treatment because of their social understand the criteria for patient selection. medical centers in the United States soon worth. “Social worth” criteria included a pa- Seriously ill patients who suffered from a va- began offering chronic hemodialysis, we tient’s marital status, occupation, income, edu- riety of diseases, such as cancer, hoped that focus on Seattle’s experience because it was cation, emotional stability, and future poten- this new wonder drug would cure them. The the first and most widely publicized example tial, along with the patient’s gender and age.21 COC denied access to penicillin on clinical of priority setting for dialysis.) It quickly be- As guardians of a public resource, the grounds, but the media and the public did not came apparent to Seattle’s Artificial Kidney committee evaluated patients’ eligibility to re- clearly understand why access was granted in Center that, given the scarcity of the dialysis ceive medication according to their value to some cases but not others. Media attention machines and trained personnel, choices society as a whole: their contributions as focused on dramatic cases of sick young chil- would need to be made about which pa- workers, family members, and volunteers dren who had been denied penicillin and tients would have access to this life-long and versus the loss to society if they died, espe- followed their cases with attention-grabbing expensive treatment. While the doctors in cially in leaving dependent children. The headlines, such as “7 hours to live” in the charge were willing to determine which kid- committee’s first selection was a 33-year-old case of a 2-year-old girl.15 ney patients qualified medically for dialysis, electrician with 7 dependents who would re- The highly centralized structure and scien- they soon decided that choosing “who shall quire state financial assistance if he were un- tific focus of the COC alienated civilians, who live” among these patients was a task better able to work. Generally speaking, preferred had already been asked to make sacrifices left to others, and in 1961 a 9-member com- candidates for dialysis were those who were during wartime.16 This model of expert cen- mittee composed of 7 lay people and 2 phy- successful at work but possessed limited tralized decisionmaking remained politically sicians was appointed to recommend pa- monetary savings, had numerous children, 19 unacceptable after the war, as the case of the tients for treatment. and were actively involved in church and drug cortisone, discovered in 1949, demon- The physicians had already made a few community affairs. The importance placed strates.17 Experts from the “disease founda- “medical decisions” about who should receive upon the male head of household’s responsi- tions” (which focus on particular diseases, treatment. They excluded children as candi- bilities to his family created gender inequal- such as arthritis) successfully challenged the dates on the grounds that it would be a mis- ity in the selection of patients and was authority of a committee at the National take to accept children and reject heads of weighted heavily in favor of males: a car Academy of Sciences to control the allocation households who might be supporting several salesman (male), a physicist (male), an engi- 20 of scarce supplies of corisone.17(p420) The COC children. The logic was utilitarian: the great- neer (male), and a homemaker (female) were failed to realize that rationing of a scarce est number of children, they reasoned, could among the earliest patients. The committee’s medical resource is an inherently political benefit from allocation of resources to heads decisionmaking process reflected the values process. Despite the development of clinical of households rather than directly to children. and biases of White, middle-class Protestant criteria, nonmedical criteria were part of the Although the decision to exclude children American society: the committee chose pa- rationing process simply by giving priority to was presented as a medical one, in fact it in- tients most like themselves.22 the military and by including speed in return- volved a moral decision about overall benefit After a critical article by reporter Shana ing soldiers to active duty as part of the crite- to society. Even these rules still left several Alexander appeared in Life magazine in ria. Even during a critical national emergency candidates for just one slot, and hence the 19 62, the committee received widespread such as World War II, it was difficult to sus- necessity for the 9-member committee to de- criticism for its reliance on social criteria in tain public confidence in rationing decisions cide who would receive treatment. selecting patients.18 The subsequent contro- because the public had no voice in how those The committee members were aware of versy induced changes in the process. The decisions were made. Ensuring that the deci- the impact of disease not just on health and committee no longer assessed the “social sionmaking body for patient selection is ac- mortality but also on the social and economic worth” of candidates, but nonetheless relied cepted as politically legitimate and account- lives of households and the surrounding com- on interviews with patients and patients’ fam- able to the public is as important as agreeing munity. Their decisions reflected their under- ily members; psychological assessments; and on a system of criteria for patient selection. standing of disease as social and economic detailed employment, educational, and per- phenomena, and public money for treatment sonal histories, all of which reinforced the MEDICAL CRITERIA ARE NOT as a kind of collective good for the benefit of committee’s bias in favor of middle-class “NEUTRAL” society as a whole. Their first requirement Americans who conformed to the commit- was therefore residence within the state of tee’s ideals about proper behavior and norms. In 1960, hemodialysis, which had been Washington (since the program was sup- Furthermore, the composition of the commit- limited to treating acute renal failure, became ported by the state’s tax revenues). The com- tee changed so that it no longer included

July 2005, Vol 95, No. 7 | American Journal of Public Health McGough et al. | Peer Reviewed | Public Health Matters | 1175  PUBLIC HEALTH MATTERS 

members from the surrounding community. to therapy. The reaction was 2-fold. First, the years of survival.26 Medical criteria are not Two physicians, along with the center’s ad- new committee tried to make its moral completely “value-free” but often incorporate ministrative director and 2 board members, choices sound as medical—and therefore decisions about whether to give priority to served on the new committee, which relied morally neutral—as possible. Evaluating a the most urgent cases, to those with the best heavily on reports by experts in psychology candidate as “psychologically unfit” made the chance of survival, or to those with the best and social work. decisionmaking process sound like a techni- chances of quality and length of life outwards. The emphasis changed from judging a pa- cal, scientific decision, even if it actually in- Liver transplant allocation, for example, tient’s overall contribution to society to judg- volved disguising what amounted to moral moved from a system that prioritized the po- ing a patient’s “psychological suitability” for choices as medical ones. tential for clinical efficacy to a system that dialysis, a demanding treatment that required Second, to prevent difficult choices about prioritized urgency, partly because of public patients’ commitment to their own care. The which patients to treat, the US Congress ap- pressure over the underlying values embed- committee was trying to ensure that scarce proved legislation in 1972 to provide funding ded in different medical scoring systems. resources would not be wasted on patients for all patients with end-stage renal disease Before 1998, liver transplant patients re- who were unmotivated or unlikely to succeed who could not afford to pay for dialysis them- ceived organs according to their score on the in therapy, an understandable concern. But selves. This legislation dramatically increased Child–Turcotte–Pugh system27 and their po- the committee’s definition of psychological resources available for kidney dialysis but, by sition on a waiting list. Allowing patients to suitability reveals how difficult it is to disen- removing selection criteria, led to a higher receive liver transplants on a first-come-first- tangle factors that contribute to a patient’s rate of poor health outcomes since sicker pa- served basis (if they met clinical criteria) pro- ability to benefit from medical therapy con- tients were now being treated.25 Ironically, duced a conflict between the principle of trasted with social prejudices about desirable the legislation introduced a new kind of in- fairness and the problem of urgency. Patients and undesirable behavior that may or may equality: between patients with end-stage with more severe disease would not receive not affect patient survival. renal disease and patients with other chronic, a transplant if a less sick person were “ahead Forexample, one candidate for dialysis, an fatal conditions not covered by national in line.” unemployed 22-year-old truck driver (re- legislation. Furthermore, the legislation com- This method of establishing priorities was ferred to here as Mr A), received critical re- promised the goal of using scarce resources especially vulnerable to the criticism that it marks from committee members because his efficiently—sick patients not only received systematically discriminated against those common-law wife was obese and apparently dialysis but a range of support services, de- with less access to the health care system, critical of medical authorities.23 Subsequent spite their slim chances of long-term survival such as the poor and minorities. Opponents discussion of Mr A’s case focused on the man- and recovery. of the system feared that the poor and mi- agement problems his common-law wife norities were more likely to have clinically ad- would present. Mrs A clearly failed to meet BALANCING COMPETING DEMANDS vanced stages of disease when they were the committee’s definition of a suitable wife, FOR ACCESS placed on waiting lists. In addition to fears because she was obese, defensive about her about systematic (even if unintended) discrim- weight, and emotional. It is not clear, however, The trend toward relying on “neutral” med- ination about waiting lists, the public also whether these traits would have impeded her ical criteria for patient selection is especially voiced concerns about the use of predictors ability to support her husband’s dialysis pro- evident in decisionmaking about organ trans- of clinical efficacy in selecting patients for gram. Under the earlier system, 1 committee plants. By using seemingly neutral criteria, transplants. One 1996 study showed that the member voted for a candidate with strong policymakers can avoid fundamental conflicts public valued giving “everyone a chance” and church involvement because the committee over ethical choices in patient selection, since allocating livers according to a first-come-first- member thought religious belief was a reliable an appeal to medical criteria masks the ethi- served basis, rather than linking organ alloca- indicator of commitment to undergo a difficult cal choices that are often involved. Defining tion to patient prognosis so that those most therapeutic regimen.24 In practice, it was diffi- the “most good” is itself a controversial pro- likely to survive according to clinical criteria cult to isolate traits that predicted potential for cess involving value judgments about the would receive priority.28 clinical success from socially desirable, but relative importance of urgency, chances for In response to these 2 different criticisms, clinically irrelevant, factors. survival, or quality and quantity of life. In in 1998 the US Department of Health and The Seattle experience illustrates the diffi- choosing to do the most “good” with a scarce Human Services adopted a policy that priori- culty decisionmakers face in trying to make resource, should priority be given to a patient tizes urgency rather than potential for clinical efficient use of scarce resources, especially who faces death more immediately, or to a efficacy and time on waiting lists. The new when the therapy is lifelong and difficult. patient who is not as sick but stands a better system stratifies patients according to their Even before the civil rights movement’s em- chance of survival if given treatment? “Doing score on the Model for End-Stage Liver Dis- phasis on human rights and equality, Ameri- good” with a limited resource could mean not ease, with priority given to the more seriously cans were uncomfortable with defining the only averting death but factoring in the qual- ill patients.29 Even medical criteria often in- “social worth” of patients to determine access ity of life after treatment, as well as expected corporate a decision to prioritize clinical effi-

1176 | Public Health Matters | Peer Reviewed | McGough et al. American Journal of Public Health | July 2005, Vol 95, No. 7  PUBLIC HEALTH MATTERS 

cacy versus urgency; public debate about and TABLE 1—Number of Patients Receiving Furthermore, the weak health infrastruc- input into these 2 often-competing values is Antiretroviral Therapy in 4 Countries ture, limited availability of trained medical necessary. Furthermore, seemingly fair crite- personnel to deliver and monitor ARV ther- ria such as waiting lists can also incorporate Country No. of Patients apy, and widespread prevalence of HIV pres- significant inequalities, if certain groups have Ethiopia ent serious challenges not faced by the previ- limited access to health care and therefore HIV positivea 1500000 ous historical examples. The feasibility of cannot get their names entered on waiting Therapy urgently neededb 211 000 ARV treatment programs has been such a lists as quickly as others. Receiving therapyc 10 000–13 000 serious concern that issues of equity and Uganda fairness have received comparatively less ACCESS TO ANTIRETROVIRAL HIV positivea 530 000 attention. That said, treatment programs THERAPY Therapy urgently neededb 114 000 nonetheless face the problem of having Receiving therapyc 40 000–50 000 more patients medically eligible for ARV With the rapid expansion of ARV therapy India therapy than resources permit, and hence under new programs such as the Global Fund HIV positivea 5100000 have developed a variety of exclusion crite- 34 and PEPFAR—especially, but not only, in sub- Therapy urgently neededb 770 000 ria to ration treatment. Saharan Africa—more patients are now or will Receiving therapyc 20 000–36 000 be receiving ARV therapy. These programs Malawi POSSIBLE ARV SELECTION have established treatment targets that are HIV positivea 900 000 FACTORS widely regarded as ambitious and challeng- Therapy urgently neededb 140 000 ing, but will nonetheless reach only about Receiving therapyc 10 000–12 000 In the sections that follow, I explain the 50% of those urgently needing ARV current or proposed patient selection criteria a 1 therapy.30 However, if the Global Fund does For adults and children. at the national and local levels. bFor ages 15–49.31 not receive adequate funding, even these cFor adults and children.31 targets will be difficult to reach, making the Source. “3 by 5” Progress Report December 2004 Selection at the Clinical Stage issue of patient selection more pressing. As and UNAIDS/WHO Epidemiological Fact Sheet–2004 Selecting patients at the clinical stage in- Update (see endnotes 1 and 31). treatment programs are being implemented volves (1) prioritization of WHO-defined over the next 1 to 3 years, it will be neces- stage IV patients whose prognosis suggests sary to focus initial efforts in limited areas. that they are urgently in need of ARV thera- Step-by-step implementation means that pain light of the historical experiences described pies or (2) prioritization of WHO-defined tient selection will be the reality for the next here. Several important differences exist be- stage III patients who are more likely to several years, even if ideal targets are met. tween those historical examples and the cur- survive, and to survive free of long-term On a country-by-country basis, the percent- rent ARV programs, however. First, the ARV complications, than stage IV patients. age of patients receiving ARV therapy out of treatment programs take place within the con- Prioritization of stage IV patients has been the estimated population in urgent need of text of global development, international the explicit recommendation of at least 2 ARV therapy varies tremendously. In sub- health, and trade regulations, unlike the previ- studies35 and has been the de facto experi- Saharan Africa, for example, where coverage ous examples of rationing within the United ence of pilot projects providing ARVs, simply of the population that needs medication aver- States. Some of the decisionmaking occurs because most patients seeking care have been ages 8%, the figure ranges from 1% in the by donors, such as the decision to select 15 critically ill patients—desperate enough for ac- Central African Republic to a high of 50% in specific countries as recipients of PEPFAR cess to ARVs to overcome the stigma of being Botswana.31 Table 1 shows the numbers of funds.33 Recipient countries are more con- identified as AIDS patients.36 The decision is patients receiving ARV therapy in 4 coun- strained in the kinds of decisions they can reminiscent of the choices outlined in the ex- tries, the first 2 eligible for PEPFAR funds make than if these programs were initiated ample of liver transplants: how to balance the (Ethiopia and Uganda) and the following 2 in- and funded by themselves; they cannot shift conflicting values of fair chances at access eligible (India and Malawi) and hence depen- their own national health budget from HIV/ versus best outcomes. dent on the Global Fund and other donors. AIDS to other diseases, but must use both The argument in favor of prioritizing Although many governments, such as that of Global Funds money and PEPFAR money as stage IV patients combines aspects of fair Uganda, have announced plans to provide additions to their current spending on HIV/ chances and best outcomes. Stage IV patients free ARV therapy to all patients,32 the reality AIDS. These treatment programs exist within face imminent death and do not have the lux- for the present and the next several years is the wider political economy of international ury of waiting for treatment programs to that medications will be rationed. relations, where health care in general and begin, while presumably stage III patients Several rationing mechanisms exist at the HIV/AIDS in particular are only part of a may survive until treatment programs are international, national, and local levels. It is complex relationship between developing fully operational. Selection of stage IV pa- worth evaluating these rationing mechanisms countries, donors, and international agencies. tients still offers the possibility that stage III

July 2005, Vol 95, No. 7 | American Journal of Public Health McGough et al. | Peer Reviewed | Public Health Matters | 1177  PUBLIC HEALTH MATTERS 

patients will be treated in the future, espe- patients according to perceived psychological, sion of HIV involves a limited-duration course cially as they progress to clinical stage IV. emotional, and social strengths introduced of medications to prevent primary infection. This scenario depends on the continuing ex- significant, unnecessary social biases into the But the issue of providing lifelong ARV ther- pansion of ARV therapy; otherwise, if funds selection process. It is not clear how a “family apy to children is complex and is similar to remain fixed, stage III patients may never support network” is defined or its existence the decisions initially faced in Seattle regard- have access to treatment. Selection of stage IV determined, again leaving a lot of room for ing dialysis. The Seattle committee excluded patients also produces “good outcomes” in that discrimination. Second, no evidence yet exists sick children from treatment on the grounds it maximizes the number of deaths averted that shows a clear link between social factors that more children would benefit by provid- and may reduce the stigma associated with such as socioeconomic status and an individ- ing treatment to their parents. HIV/AIDS mortality. ual patient’s ability to adhere to ARV therapy. More than virtually any other disease, Like the problem of liver transplants, how- However, because the issue of adherence AIDS has a serious economic and social im- ever, the definition of “best outcomes” is not is so important to the success of ARV treat- pact because its victims are working-age and as simple as averting deaths. What about ment, a number of pilot projects have devel- reproductive-age adults.43 In a recent study, long-term chances of survival and overall oped ways of trying to ensure that patients Ugandans apparently unwittingly share the quality of life? The problem with prioritizing have the commitment to adhere to therapy. same concerns as the Seattle dialysis policy- stage IV patients is that they require more in- Medecins Sans Frontières, for example, re- makers in prioritizing the health care needs tensive and specialized care, which both costs quires that patients enroll in support proof income-producing adults over those of more and requires medical expertise not grams, attend clinics regularly for a period children. Stakeholders (community leaders) readily available. Patients with low CD4 cell of 3 months prior to receiving ARV therapies, in rural and urban Uganda ranked lifelong counts at the initiation of highly active anti- and be open about their HIV status.40 Other provision of ARV therapy to adults above retroviral therapy (HAART) are less likely to programs require a treatment support partner short-term prevention of mother-to-child survive for 5 years37 and may develop (or or patient commitment to safe sex practices.41 transmission of HIV so that income-generat- have already developed) long-term disabilities The advantage of these approaches is that ing adults could care for children.44 If gov- that make a full recovery difficult to achieve. they depend on an individual’s actual behav- ernments are trying to mitigate the social ior, rather than on predictions of behavior and economic impact of the AIDS epidemic Adherence Criteria based on factors beyond the patient’s control. by providing treatment, it will be difficult to Whether to include estimations of a pa- It is important to recognize that even these sort out which groups should be the primary tient’s ability to adhere to therapy as 1 of the criteria can be influenced by unconscious bi- beneficiaries of treatment programs. The po- criteria for patient selection is an important ases and potentially unnecessary social crite- tential for gender discrimination exists, be- question to consider when determining who ria (the availability of a treatment support cause male heads of households contribute will receive necessary treatment. partner, for example). significantly to household wealth. Accounting for the crucial importance of adherence in the success of HAART, it is Age Occupation potentially useful to screen patients for their Whether to prioritize certain age groups, Prioritizing patients with certain occupa- ability and willingness to adhere to a de- such as children or people with dependents, tions is another possible criterion that needs manding medical regimen. The problem, is another factor to consider in determining to be considered. however, is that it is difficult to predict accu- access to ARV therapies. The major argument for favoring a certain rately which individuals will adhere to ARV The government of India announced its occupation is that the skills possessed by that therapy, as evidence from the United States decision to offer as of April 2004 free ARV group are vital to the survival of the society has shown.38 One of the principal reasons for therapy to all new parents who are HIV- as a whole, such as the need for health care nonadherence in Senegal and Botswana has positive, all children aged younger than 15 workers to protect the population as a whole, been the cost of drugs; treatment interrup- years, and eventually all patients with AIDS the need for teachers to train the next generations occurred because of financial difficul- in the 6 states with the highest prevalence tion, and the need for soldiers to defend soci- ties, which presumably would not have oc- of HIV/AIDS.42 ety against attack. However, as the experience curred had the drugs been provided free of Prioritizing children younger than 15 years with penicillin during World War II suggests, charge.39 Once cost is removed as an obsta- is meant to save both the next generation and giving priority to a certain group can create cle, socioeconomic status is not a reliable “innocent victims,” since most of these chil- considerable conflict. Although American so- predictor of adherence. dren become infected from their mothers ciety was fairly united behind the military The situation is similar to the decision rather than from sexual intercourse. It is criti- during World War II, even in these circum- faced in the case of kidney dialysis, since this cally important to distinguish between the stances the public did not always agree that chronic disease necessitated patients’ willing- provision of ARV therapy to HIV-positive chil- the military deserved priority in access to ness to adhere to a diet and rigorous treat- dren and prevention of mother-to-child trans- medications, especially if the soldiers had ac- ment regimen. In practice, however, selecting mission. Preventing mother-to-child transmis- quired gonorrhea or syphilis through “im-

1178 | Public Health Matters | Peer Reviewed | McGough et al. American Journal of Public Health | July 2005, Vol 95, No. 7  PUBLIC HEALTH MATTERS 

moral means.” Some members of the public Patient selection is a potentially divisive issue We acknowledge Harry Marks, Nancy Kass, Randall Packard, Khalil Ghanem, Chris Beyrer, and the Journal’s even questioned the legitimacy of the deci- that could aggravate tensions between ethnic anonymous reviewers for their comments on earlier 48 sionmaking body. For ARV therapy, policy- groups, religious groups, and social classes. drafts of this article. makers should therefore anticipate conflict if To avoid systematic discrimination—or even only 1 occupational group is given priority of the perception that certain individuals or Endnotes access to medications. groups are favored at the expense of others— 1. World Health Organization, “ ‘3 by 5’ Progress Report December 2004,” 2005, available at http:// Widespread public input into the selection broad public participation in the process of www.who.int/3by5/progressreport05/en, accessed Jan- criteria will not eliminate conflict, but it will establishing criteria for patient selection uary 27, 2005. at least give the public some voice and lend should be encouraged. It is also important to 2. Ibid. political legitimacy to the selection criteria. In operate within a human rights framework to 3. Ibid; “Treating 3 Million by 2005: Making It a paper commissioned by WHO, Norman prevent the systematic discrimination in ac- Happen: The WHO Strategy,” available at http://www. Daniels described the central features of a cess to treatment according to such factors unaids.org, accessed April 1, 2005. fair process for distributing ARV therapy; as sex, ethnicity, and sexual orientation. 4. R.B. Cheek, “Playing God With HIV: Rationing HIV Treatment in Southern Africa,” African Security WHO has also issued guidelines on ethical Countries will no longer face the tough de- Review, 2001, available at http://www.iss.co.za/PUBS/ and equitable access to care.45 One of the cisions regarding which patients receive ARV ASR/10-No4/Cheek.html, accessed May 23, 2005. significant advantages of this process is that it therapy first if prevention efforts are success- 5. L. Rosen, J.L. Simon, A. Collier, and I.M. Sanne, allows for local communities to have a voice ful, developing countries’ productivity and in- “Hard Choices: Rationing Antiretroviral Therapy for HIV/AIDS in Africa,” Lancet 365 (2005): 354–356. in patient selection. Each country will have comes expand as a result of greater interna- 6. P. Piot, “Defeating HIV/AIDS: Africa Is Changing the flexibility to adopt policies relevant to its tional investment, and the costs of treatment Gear,” speech presented at closing ceremony of the own situations. continue to decline. Meanwhile, lessons International Conference on AIDS and STD in Africa, Other methods of selecting patients are not learned from earlier experiences of patient Nairobi, Kenya, September 26, 2003, available at http://www.unaids.org/Unaids/EN/About+UNAIDS+ carefully thought-out selection criteria but selection can make the process more efficient, executive+director/unaids+executive+director/unaids+ passive systems that make treatment available equitable, accountable, and legitimate in the executive+directors+speeches+.asp?StartRow=20, to the small numbers of people who have eyes of the public. At a minimum, we hope accessed May 23, 2005. been tested (partly owing to the stigma sur- policymakers are aware that patient selection 7. R. Macklin, “Ethics and Equity in Access to HIV Treatment—3 by 5 Initiative,” background paper for rounding HIV/AIDS) and to those living is not a technical problem best solved by ex- the Consultation on Equitable Access to Treatment near clinic sites who request treatment.46 perts. Open, public debate will not resolve all and Care for HIV/AIDS, World Health Organization, The advantage of this system is that it is cost- conflict, but it will be an important part of the 2004, available at http://www.who.int/ethics/en/ background-macklin3.pdf, accessed March 29, 2005; effective since it requires no investments in process of building local ownership of AIDS N. Daniels, “How to Achieve Fair Distribution of ARTs infrastructure. The major problem is that it treatment programs, a critical element for in 3 by 5: Fair Process and Legitimacy in Patient Selec- does not provide a fair chance of access for their sustained success. tion,” background paper for the Consultation on Equi- table Access to Treatment and Care for HIV/AIDS, all, because those who have not been tested World Health Organization, 2004, available at http:// are more likely to live in rural areas, experi- www.who.int/ethics/en/background-daniels3.pdf (ac- About the Authors ence poverty, and have inadequate access to cessed March 29, 2005); Pro-Poor Health Policy Team, At the time this article was written, all authors were with “‘3 by 5,’ Priority in Treatment, and the Poor,” back- health care. the Department of Infectious Diseases, Johns Hopkins Uni- ground paper for the Consultation on Equitable Access Furthermore, socially marginalized groups versity, Baltimore, Md. Laura J. McGough is also with the to Treatment and Care for HIV/AIDS, World Health also have the most difficulty in mitigating the Department of the History of Medicine, Johns Hopkins Organization, 2004, available at http://www.who.int/ University. Steven J. Reynolds and Thomas C. Quinn are 47 ethics/en/background-pro-poor3.pdf (accessed March effects of stigma. Relying on a system that also with the National Institute of Allergy and Infectious 29, 2005); World Health Organization, “Guidance on selects patients according to access to testing Diseases, National Institutes of Health, Bethesda, Md. Ethics and Equitable Access to HIV Treatment and sites and absence of stigma exacerbates exist- Requests for reprints should be sent to Jonathan M. Care,” 2004, available at http://www.who.int/ethics/ Zenilman, MD, Infectious Diseases Division, Johns Hopkins en/ethics_equity_HIV_e.pdf (accessed March 29, ing social inequalities, although it is a feasible Bayview Medical Center, 4940 Eastern Ave, B-3 North, 2005). Baltimore, MD 21224 (e-mail: [email protected]). and cost-effective method of delivering ther- 8. For historical background on the discovery of in- This article was accepted January 30, 2005. apy. This system bears some similarity to the sulin and its treatment of diabetes, see M. Bliss, The experience of insulin rationing, where no care- Discovery of Insulin (Chicago: University of Chicago Contributors Press, 1982); C. Feudtner, Bittersweet: Diabetes, Insulin, fully thought-out criteria were in place and ac- and the Transformation of Illness (Chapel Hill: University L. J. McGough conceived the study and led the re- of North Carolina Press, 2003). cess was restricted to certain clinics and hospi- search and writing. S. J. Reynolds and T.C. Quinn of- tals. Furthermore, this system was especially fered technical advice and analysis. J. M. Zenilman su- 9. Ibid, 135,144,146,151. vulnerable to unfairness, as personal and po- pervised the entire process of analysis and article 10.For the history of penicillin rationing during preparation. All authors helped to conceptualize ideas, litical connections to the physicians and re- World War II, see D.P. Adams, “The Greatest Good to interpret findings, and revise drafts of the article. the Greatest Number”: Penicillin Rationing on the Ameri- searchers brought access to medications. can Home Front, 1940–1945 (New York: Peter Lang, The success of ARV programs rests at least Acknowledgments 19 91). partially on their public acceptance as fair Research for this article was supported by a fellowship 11. Ibid, 70. programs accountable to local communities. from the Association of Teachers of Preventive Medicine. 12.From the records of the Committee on Medical

July 2005, Vol 95, No. 7 | American Journal of Public Health McGough et al. | Peer Reviewed | Public Health Matters | 1179  PUBLIC HEALTH MATTERS 

Research General Correspondence Files, quoted in who.int/GlobalAtlas/PDFFactory/HIV/index.asp (ac- 45. Daniels, “How to Achieve Fair Distribution”; Adams, “The Greatest Good to the Greatest Number,” 71. cessed March 29, 2005). World Health Organization, “Guidance on Ethics and 13.Adams, “The Greatest Good to the Greatest Num- 32. C. Wendo, “Uganda Begins Distributing Free Anti- Equitable Access.” ber,” 75. retrovirals,” Lancet 363 (2004): 2062. 46. Stepping Back From the Edge: The Pursuit of Anti- 14 . Quoted in Adams, “The Greatest Good to the Great- 33. The countries are Botswana, Côte d’Ivoire, retroviral Therapy in Botswana, South Africa, and est Number,” 86. Ethiopia, Kenya, Mozambique, Namibia, Nigeria, Uganda (Geneva: UNAIDS, April 2004). 15. Ibid, 74. Rwanda, South Africa, Tanzania, Uganda, Zambia, 47. R. Parker and P. Aggleton, “HIV and AIDS-Re- Guyana, Haiti, and Vietnam. lated Stigma and Discrimination,” Social Science and 16. M. Leff, “The Politics of Sacrifice on the American Medicine 57 (2003): 13–24. Home Front in World War II,” Journal of American His- 34. M. Rabkin, B. Tonwe-Gold, and W. El-Sadr, “Non- tory 77 (1991): 1296–1318. Medical Eligibility Criteria for Pilot HIV/AIDS Treat- 48. Cheek, “Playing God With HIV.” ment Programs in Resource-Poor Settings: The Colum- 17. H.M. Marks, “Cortisone, 1949: A Year in the Polit- bia University MTCT-Plus Experience,” oral ical Life of a Drug,” Bulletin of the History of Medicine presentation, XV International AIDS Conference, July 66 (1992): 419–439. 14 , 2004, Bangkok, Thailand, abstract WeOrB1278. 18.For a historical background of hemodialysis, see 35. Christopher Kenyon, Jolene Skordis, Andrew Albert Jonsen, The Birth of Bioethics (Oxford: Oxford Boulle, and Karrisha Pillay, “The ART of Rationing— University Press, 1998); S. Alexander “They Decide The Need for a New Approach to Rationing Health In- Who Lives, Who Dies,” Life 53 (November 9, 1962): terventions,” South African Medical Journal 93 (2003): 102–125; R.C. Fox and J.P. Swazey, The Courage to 56–60; M. Badri, L.G. Bekker, C. Orrell, J. Pitt, F. Cil- Fail: A Social View of Organ Transplants and Dialysis, liers, and R. Wood, “Initiating Highly Active Antiretro- 2nd ed (Chicago: University of Chicago Press, 1978). viral Therapy in sub-Saharan Africa: An Assessment 19. The lay members were a minister, a lawyer, a of the Revised World Health Organization Scaling-Up homemaker, a banker, a labor leader, a surgeon (who Guidelines,” AIDS 18 (2004): 1159–1168. was not an expert in kidney disease), and a state gov- 36. Louisana Lush and Ernest Darkoh, “HIV and ernment official. See Alexander, “They Decide Who Health Systems: Botswana,” paper presented at 6th In- Lives, Who Dies.” ternational Conference on Healthcare Resource Alloca- 20. Ibid, 106. tion for HIV/AIDS: Healthcare Systems in Transition, 21. Quoted in Fox and Swazey, The Courage to Fail, 232. October 13–15, 2003, Washington, DC. 22. Alexander, “They Decide Who Lives, Who Dies,” 37.Ard van Sighem, Mark van de Wiel, Azra Ghani, et Case Studies in 10 6; and R.C. Fox and J.P. Swazey, The Courage to al., “Mortality and Progression to AIDS After Starting Fail: A Social View of Organ Transplants and Dialysis Highly Active Antiretroviral Therapy,” AIDS 17 Public Health (Chicago: University of Chicago Press, 1974), (2003): 2227–2236. 245–246. 38. R. Murri, A. Antinori, A. Ammassari, et al., “Physi- Ethics 23. Fox and Swazey, The Courage to Fail, 241. cian Estimates of Adherence and the Patient–Physician Relationship as a Setting to Improve Adherence to An- By Steven S. Coughlin, PhD, Colin L. 24. Quoted in Alexander, “They Decide Who Lives, tiretroviral Therapy,” Journal of Acquired Immune Defi- Soskolne, PhD, and Kenneth W. Who Dies,” 123. ciency Syndrome 31 (2002): S158–S162. Goodman, PhD 25. R.W. Evans and C.R. Blagg, “Lessons Learned 39. I. Lanièce, M. Ciss, A. Desclaux, et al., “Adherence From the End-Stage Renal Disease Experience: Their to HAART and Its Principal Determinants in a Cohort Implications for Heart Transplantation,” in Organ Sub- uitable for classroom discussions and of Senegalese Adults,” AIDS 17 supplement 3 (2003): professional workshops, this book of stitution Technology: Ethical, Legal, and Public Policy Is- S103–S108; S. Weiser, W. Wolfe, D. Bangsberg, et al., S edited public health case studies illus- sues, ed. D. Mathieu (Boulder, Colo: Westview Press, “Barriers to Antiretroviral Adherence for Patients Liv- 1988), 175–197. ing With HIV Infection and AIDS in Botswana,” Jour- trates the ethical concerns and problems 26.D.W. Brock, “Ethical Issues in Recipient Selection nal of Acquired Immune Deficiency Syndrome 34 in public health research and practice. for Organ Transplantation,” in Organ Substitution Tech- (2003): 281–288. The sixteen chapters cover privacy and nology, 86–99. 40.T. Kasper, D. Coetzee, F. Louis, A. Boulle, and confidentiality protection, informed con- 27.For the Child-Turcotte-Pugh classification, see P.S. K Hilderbrand, “Demystifying Antiretroviral Therapy in sent, ethics of randomized trials, the insti- Kamath, R.H. Wiesner, M. Malinchoc, et al., “A Model Resource-Poor Settings,” available at http://www.msf. tutional review board system, scientific to Predict Survival in Patients With End-Stage Liver org/content/page.cfm?articleid=3EC42CE5-ADDB- misconduct, conflicting interests, cross- Disease,” Hepatology 33 (2001): 464–470. 4384-BC25F9F03313DC04, accessed May 23, 2005. cultural research, genetic discrimination, 28.P.A. Ubel and G. Loewenstein, “Distributing 41.Wendo, “Uganda Begins Distributing Free and other topics. An instructor’s guide is Scarce Livers: The Moral Reasoning of the General Antiretrovirals.” also provided at the end. Public,” Social Science and Medicine 42 (1996): 42. A. Waldman, “India Plans Free AIDS Therapy, 1049–1055. ISBN 0-87553-232-2 But Effort Hinges on Price Accord With Drug Makers,” 1997 ❚ 182 pages ❚ softcover 29. A.J Muir, L.L. Sanders, M.A. Heneghan, P.C. Kuo, W.E. New York Times, December 1, 2003. Available at Wilkinson, and D. Provenzale, “An Examination of Fac- http://www.aegis.com/news/ads/2003/AD032491. $26.00 APHA Members tors Predicting Prioritization for Liver Transplantation,” html, accessed May 23, 2005. $37.00 Nonmembers plus shipping and handling Liver Transplantation 8 (2002): 957–961; P.A. Ubel, 43. C. Bell, S. Devarajan, and H. Gersbach, “Thinking R.M. Arnold, and A.L. Caplan, “Rationing Failure: The About the Long-Run Economic Costs of AIDS,” in The ORDER TODAY! Ethical Lessons of the Retransplantation of Scarce Vital Macroeconomics of HIV/AIDS, ed. M. Haacker (Wash- Organs,” Journal of the American Medical Association ington: International Monetary Fund, 2004), 96–133. American Public Health Association 270 (1993): 2469–2474. Publication Sales 44. L. Kapiriri, B. Robbestad, and O.F. Norheim, “The 30.“Treating 3 Million by 2005,” 31. Web: www.apha.org Relationship Between Prevention of Mother to Child E-mail: [email protected] 31.UNAIDS/WHO epidemiological fact sheet—2004 Transmission of HIV and Stakeholder Decision Making Tel: 888-320-APHA update; country-by-country search engine with HIV in Uganda: Implications for Health Policy,” Health Pol- FAX: 888-361-APHA ET02J2 prevalence and incidence data available at http://www. icy 66 (2003): 199–213.

1180 | Public Health Matters | Peer Reviewed | McGough et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

Adolescent Birth Rates, and, if they have common roots, whether cents in 2000 came from the US National these roots might lie in relative or in absolute Vital Statistics System.16 Data are presented Total Homicides, and deprivation. as births per 1000 females aged 15 to 19 years. Homicide rates per 100 000 persons Income Inequality In Rich METHODS came from the Federal Bureau of Investiga- tion’s annual report on crime statistics for Countries International Comparisons the United States for 1999.17 Data were un- Countries were included if they were available for Wyoming and the District of | Kate E. Pickett, PhD, Jessica Mookherjee, MSc, among the 50 countries with the highest Columbia. From a national report on juve- and Richard G. Wilkinson, MMedSci gross national income per capita by purchas- nile offenders and victims, we also extracted ing power parity in 2002, had populations of juvenile homicide arrest rates per 100 000 more than 3 million, and had data on income juveniles (aged 10–17 years) for 46 states Income inequality has been asso- inequality. The eligible countries were Aus- in 1997.18 ciated with both homicides and tralia, Austria, Belgium, Canada, Denmark, births to adolescents in the United Finland, France, Germany, Greece, Ireland, Statistical Methods States and with homicides interna- Israel, Italy, Japan, The Netherlands, New We first estimated simple correlations be- tionally. We found that adolescent birth rates and general homicide Zealand, Norway, Portugal, Singapore, Slove- tween homicide rates and adolescent birth rates were closely correlated with nia, Spain, Sweden, Switzerland, the United rates both within the United States and inter- each other internationally (r = 0.95) Kingdom, and the United States. nationally. We then measured the indepen- and within the United States (r = Data on income inequality came from the dent effects of income inequality and per ca- 0.74) and with inequality inter- United Nations Development Program Human pita (absolute) income on homicide rates and nationally and within the United Development Indicators; dates for each coun- adolescent birth rates internationally and States. These results, coupled with try vary slightly from country to country but within the United States. Within the United no association with absolute in- are within the period 1992–1998.10 Income States only, this analysis was repeated for ju- come, suggested that violence and inequality was measured as the ratio of the venile homicide rates. We also examined the births to adolescents may reflect total annual income received by the richest effect of income inequality on each outcome gender-differentiated responses to 20% of the population to the total average while controlling for the other outcome. low social status and could be reduced by reducing income inequal- annual income received by the poorest 20% ity. (Am J Public Health. 2005;95: of the population. Gross national income per RESULTS 1181–1183. doi:10.2105/AJPH.2004. capita by purchasing power parity was mea- 056721) sured in US dollars and obtained from the Figure 1 shows a plot of adolescent birth World Bank World Development Indicators rates and homicide rates in the United States. database.11 Data on births per 1000 women The Pearson correlation coefficient between Violence and births to adolescents seem to aged 15 to 19 years for 1998 came from adolescent birth rates and homicide rates stand out as gender-differentiated markers of UNICEF.12 Adolescent birth rates were un- was 0.95 (P<.001) internationally and 0.74 the corrosive effects of poverty among young available for Israel, Singapore, and Slovenia, (P<.001) across the United States. The inde- people.1–3 Although adolescent births and and these countries were excluded from our pendent effects of income inequality and per levels of violence are strongly associated with study. Data on homicides came from the capita income on adolescent births and homi- poverty within developed countries, national United Nations’ Survey on Crime Trends and cides are shown in Table 1. The partial corre- rates of both violence and adolescent births the Operations of Criminal Justice Systems.13 lation coefficients for income inequality and are nevertheless higher in several wealthy Homicide data were period averages of both outcomes in both settings were statisti- countries compared with poor countries. In available rates per 100000 persons for cally significant (P<.01) and ranged from other words, homicides and adolescent preg- 1990–2000. 0.51 to 0.73. Figure 2 shows a plot of in- nancies appear to be associated with relative come inequality and adolescent birth rates rather than absolute poverty. Indeed, the de- US Comparisons for 21 developed countries. gree of income distribution within a society Data on income inequality for the 50 The effect of per capita income differed in has been linked to homicide rates within and United States and the District of Columbia the 2 comparisons. Internationally, higher per outside the United States (see, for example, were obtained from the US Census Bureau.14 capita income was associated with higher rates Hsieh and Pugh,4 Wilkinson et al.,5 Daly et Income inequality was measured as the Gini of adolescent births and homicides (P<.001), al.,6 and Fajnzylber et al.7), but only within coefficient based on family income for 1999. whereas in the United States, higher per ca- the United States for adolescent births.8,9 Per capita income in 1999 was obtained pita income was associated with lower adoles- We decided to investigate how much these from the US Census Bureau Census 2000 cent birth rates (P<.001) and was not signifi- 2 social problems were related to each other Summary File 3.15 Data on births to adoles- cantly related to homicide rates. The Pearson

July 2005, Vol 95, No. 7 | American Journal of Public Health Pickett et al. | Peer Reviewed | Research and Practice | 1181  RESEARCH AND PRACTICE 

correlation coefficient for juvenile homicides and income inequality was 0.31 (P=.035), and for juvenile homicides and per capita income it was effectively zero (0.001; P=.994) (data not shown). The international associations between each outcome and income inequality were removed by controlling for the other. In the United States, an attenuated correlation remained between income inequality and adolescent births.

DISCUSSION

Our findings suggested that the links between deprivation and both violence and adolescent births reflect the destructive psychosocial and behavioral effects of inequality. As Luker19 put it, it is “the discouraged among FIGURE 1—Homicide rates and rates of adolescent births in 49 states (indicated by postal the disadvantaged” who become adolescent codes): United States, Federal Bureau of Investigation, 1999, and US National Vital mothers. Gilligan20 and others pointed out Statistics System, 2000. how often violence among young men is trig- gered by humiliation and disrespect. Success- ful programs for preventing adolescent pregnancy and violence have often focused on personal development, attempting to undo the psychosocial costs of low social status.1,21 But these patterns demand a common explanation. Our study suggests that levels of relative deprivation may be an underlying cause. Interestingly, the decline in US homicide and adolescent birth rates since the 1990s was accompanied by declining unemployment and improved relative income among the poorest individuals.17 , 2 2

About the Authors Kate E. Pickett is with the Department of Health Sciences, University of York, York, England. Jessica Mookherjee is with the Department of Anthropology, University College London, London, England. Richard G. Wilkinson is with the Division of Epidemiology and Public Health, Univer- FIGURE 2—Income inequality and rates of adolescent births among 21 of the richest sity of Nottingham Medical School, Nottingham, England. Requests for reprints should be sent to Kate E. Pickett, developed countries: United Nations, 1990–2000, and UNICEF,1998. PhD, Department of Health Sciences, University of York, Seebohm Rowntree Building—Room A/TB/220, Hesling- ton, York YO10 5DD UK (e-mail: [email protected]). TABLE 1—Independent Effects of Absolute Income and Income Inequality on Homicide This brief was accepted January 22, 2005. Rates and Births to Adolescents Among Rich Countries and Within the United States Contributors Among Rich Countries Within the United States All authors contributed to the origination and design of Adolescent Births Homicides Adolescent Births Homicides the study and to the writing and revising of the brief. K.E. Pickett conducted the statistical analysis. rPrP r P RP Income inequality 0.73 <.001 0.71 <.01 0.72 <.001 0.51 <.001 Human Participant Protection Per capita income 0.75 <.001 0.78 <.001 –0.55 <.001 –0.17 .245 No protocol approval was needed for this ecological study.

1182 | Research and Practice | Peer Reviewed | Pickett et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

References age Pregnancy. Cambridge, Mass: Harvard University 1. Health Development Agency. Teenage Pregnancy Press; 1996. and Parenthood: A Review of Reviews. London, England: 20. Gilligan J. Violence: Our Deadly Epidemic and Its Health Development Agency, National Health Service; Causes. New York, NY: GP Putnam; 1996. 2003. Evidence Briefing. 21. Center for the Study and Prevention of Violence. 2. The Alan Guttmacher Institute. Teen Sex and Preg- Blueprints Model Program Descriptions, FS-BP02. Boul- nancy: Facts in Brief. New York, NY: The Alan der: University of Colorado; 2004. Available at: http:// Guttmacher Institute; 1999. www.colorado.edu/cspv/blueprints/model/overview. 3. Wilson M, Daly M. Competitiveness, risk-taking html. Accessed April 13, 2005. and violence: the young male syndrome. Ethol Socio- biol. 19 85;6:59–73. 22. The Alan Guttmacher Institute. The Guttmacher Report on Public Policy. New York, NY: The Alan 4. Hsieh C-C, Pugh MD. Poverty, income inequality, Guttmacher Institute; 1998. and violent crime: a meta-analysis of recent aggregate data studies. Crim Justice Rev. 19 93;18:182–202. 5. Wilkinson RG, Kawachi I, Kennedy BP. Mortality, the social environment, crime and violence. Sociol Health Illn. 1998;20:578–597. 6. Daly M, Wilson M, Vasdev S. Income inequality and homicide rates in Canada and the United States. Can J Criminol. 2001;43:219–236. 7. Fajnzylber P, Lederman D, Loayza N. Inequality and violent crime. J Law Econ. 2002;45:1–40. 8. Gold R, Kawachi I, Kennedy BP, Lynch JW, Connell FA. Ecological analysis of teen birth rates: association with community income and income inequality. Matern Child Health J. 2001;5:161–167. 9. Gold R, Kennedy B, Connell F, Kawachi I. Teen births, income inequality, and social capital: developing an understanding of the causal pathway. Health Place. 2002;8(2):77–83. 10. United Nations Development Program. Human Development Report. New York, NY: Oxford University Press; 2003. 11.World Development Indicators [database online]. Washington, DC: World Bank; 2004. Available at: http://www.worldbank.org/data/wdi2004. Accessed December, 2004. 12. A League Table of Teenage Births in Rich Nations: Innocenti Report Card. Florence, Italy: UNICEF Inno- centi Research Centre; 2001. Report No. 3. 13. United Nations Crime and Justice Information Network. Survey on Crime Trends and the Operations of Criminal Justice Systems (Fifth, Sixth, Seventh). New York, NY: United Nations; 2000. 14 .Table S4: Gini ratios by state: 1969, 1979, 1989, 1999. Washington, DC: Income Statistics Branch/ Housing and Household Economic Statistics Division, US Census Bureau; 2004. Available at: http://www. census.gov/hhes/income/histinc/state/state4.html. Ac- cessed April 13, 2005. 15. Census 2000 Summary File 3. Washington, DC: US Census Bureau; 2003. 16.Ventura SJ, Mathews TJ, Hamilton BE. Teenage births in the United States: state trends, 1991–2000, an update. Natl Vital Stat Rep. May 30, 2002;50(9): 1–4. 17. Crime in the United States. Washington, DC: Fed- eral Bureau of Investigation; 1990–2000. 18. Office of Juvenile Justice and Delinquency Preven- tion. Juvenile Offenders and Victims: 1999 National Re- port. Pittsburgh, Pa: National Center for Juvenile Jus- tice; 1999. 19.Luker K. Dubious Conception: The Politics of Teen-

July 2005, Vol 95, No. 7 | American Journal of Public Health Pickett et al. | Peer Reviewed | Research and Practice | 1183  RESEARCH AND PRACTICE 

Quality of Parental Consent in a Ugandan Malaria Study

| Christine Pace, BA, Ambrose Talisuna, PhD, MBchB, MSc, David Wendler, PhD, Faustin Maiso, MBchB, MSc, Fred Wabwire-Mangen, MD, Nathan Bakyaita, MD, Edith Okiria, PhD, Elizabeth S. Garrett-Mayer, PhD, Ezekiel Emanuel, MD, PhD, and Christine Grady, PhD, RN

There is concern that illiteracy and limited Objectives. We surveyed Ugandan parents who enrolled their children in a education, lack of familiarity with research, randomized pediatric malaria treatment trial to evaluate the parents’ levels of and limited access to health care can jeopard- understanding about the treatment trial and the quality of the parents’ consents ize the ability of study participants, especially to allow their children to participate in the study. those in developing countries, to provide fully Methods. We conducted 347 interviews immediately following enrollment at informed and voluntary consent.1–4 Despite 4 Ugandan sites. this concern, few studies have been con- Results. A majority (78%) of the parents, most of whom where mothers (86%) ducted to examine the quality of informed had at most a primary school education. Of the participating mothers, a sub- consent in the aforementioned settings.5–10 stantial percentage reported that they remembered being told about the study’s purpose (77%), the required number of visits (88%), the risks involved (61%), In developing countries, where children are treatment allocation (84%), and their ability to discontinue their children’s par- involved in vital research on diseases such as ticipation (64%). In addition, most reported knowing the trial’s purpose (80%) and malaria and dysentery, even fewer studies the required number of visits (78%); however, only 18% could name possible have evaluated the quality of informed paren- side effects from the drugs being administered, and only 19% knew that children tal consent for pediatric trial participants un- would not all be administered identical treatments. Ninety-four percent reported able to provide their own consent. This aspect that they made the enrollment decision themselves, but 58% said they felt pres- of informed consent also is understudied in sure to participate because of their child’s illness, and 15% said they felt some type developed countries. of pressure to participate from others; 41% reported knowing that they did not At 4 separate sites in Uganda, we inter- have to participate. viewed parents and guardians whose children Conclusions. The consent Ugandan parents provided to allow their children were participating in a randomized study of to participate in the malaria study was of mixed quality. Parents understood many of the study details, but they were not very aware of the risks involved or of ran- antimalarial treatments. Immediately after domization. Many parents felt that they could not have refused to participate be- parents consented to enroll their child in the cause their child was sick and they either did not know or did not believe that their antimalarial study, we interviewed those par- child would receive treatment outside of the study. Our results indicate that fur- ents about their motivations for enrolling ther debate is needed about informed consent in treatment studies of emergent their child in the study, their experiences with illnesses in children. (Am J Public Health. 2005;95:1184–1189. doi:10.2105/AJPH. the informed consent process, their compre- 2004.053082) hension of the trial, and the extent to which the enrollment decision was voluntary. lance Project, a partnership between the of blood for future research), the voluntary METHODS Makerere University Institute of Public Health; nature of participation, and the participants’ the Makerere University–University of Califor- freedom to withdraw from the study. A Design and Setting nia, San Francisco Malaria Research Collabo- shorter consent form provided further de- Children aged 6 months or older were ration; the Ugandan Ministry of Health, and tails on collection and storage of blood for enrolled in a randomized, single-blind treat- the University of California, Berkeley. future research. ment study that assessed the efficacy of different oral antimalaria drug regimens. Because Informed Consent Process Participants of increasing Plasmodium falciparum resist- Parents or guardians of children who met Our study involved the use of a conve- ance to chloraquine (CQ) and sulfadoxine the study’s eligibility criteria received copies nience sample. All parents or guardians en- pyrimethamine (SP), the standard first-line of the trial consent forms in both English rolling a 6-month-old to 12-year-old child treatments for malaria in Uganda, this study and the local language. Parents participated in the malaria study between October 2002 compared the efficacy of CQ and SP with that in a discussion with a nurse-coordinator or and March 2003 were invited to partici- of amodiaquine plus SP and, at 1 site, with investigator during which the consent form pate. In-person interviews were conducted that of amodiaquine plus artesunate. The was reviewed. The form, about 5 pages and immediately after consent was provided for trial, conducted, at 3 rural sites (Mubende, 12 00 words long, contained information on the malaria study. Respondents provided Kyenjojo, and Kanungu) and 1 periurban site the study’s purpose, its risks and benefits, its verbal informed consent to complete in- ( Jinja), was led by the Uganda Malaria Surveil- procedures (e.g., randomization and storage person interviews.

1184 | Research and Practice | Peer Reviewed | Pace et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

Survey Development and Administration Statistical Analysis TABLE 1—Characteristics of Study The survey instrument, developed by the We used descriptive statistics to summarize Respondents authors in conjunction with the National the data. In the case of categorical variables, Opinion Research Center (NORC) at the Uni- we calculated and compared proportions in Percentage χ2 versity of Chicago, was based on relevant each category via tests. Seven variables Female 86 domains from the research literature and the (gender, marital status, head of household, Parent of child in study 92 11–13 malaria study consent form. The final in- education, previous research participation of Head of household 19 strument (available from the authors on re- other children, site, and pressure because of Education quest) consisted of 60 questions that focused child’s sickness) were examined in regard to None 20 on 6 domains. With the exception of ques- associations with (1) knowledge about treat- Some primary school 40 tions gathering data on sociodemographic ment allocation and side effects of the malaria Primary school 18 characteristics, the question formats used drugs and (2) the voluntariness of the enroll- Some secondary school 14 α were multiple choice (17 questions), yes or ment decision. The statistical significance ( ) Secondary school 4 no (22 questions), or open ended (9 ques- level was set at .05. Any college/university 4 tions). The data from these interviews pro- Rural dweller 71 vided (1) parents’ motivations for enrolling RESULTS Electricity in house 10 their children in the malaria study; (2) par- Homeowner 70 ents’ experiences with the informed consent Overall, 85% of respondents identified Previous enrollment of 1 or more 25 process; (3) parents’ comprehension of study themselves as the participating child’s children in a research study details, including procedures, possible side mother, 7% as the child’s father, and 19% as effects, and randomization; (4) the extent to the head of their household. Most owned the which the decision to participate was volun- house they lived in (70%), and these homes signment (71%) (Table 2). Seventy-nine per- tary; and (5) sociodemographic characteris- were primarily dwellings with tin roofs (76%) cent of the respondents reported that they re- tics. Data on a sixth domain, attitudes regard- but no electricity (90%). Other respondent ceived all of the information they wanted; ing samples stored for future research, are sociodemographic characteristics are shown 67% reported being given a chance to ask reported elsewhere.14 in Table 1. questions, of which two thirds (45% of re- The instrument was translated into the spondents overall) actually asked questions. Experience With the Informed Consent Luganda, Rukiga, and Rutooro languages by Process Comprehension of Study Information professional translators in Uganda, who then Most respondents remembered being told Eighty percent of the respondents correctly back-translated the instrument into English. by study personnel about the study purpose identified the study purpose as determining It was then tested among the 8 Ugandan (77%), study risks (61%), number of visits which malaria drugs are most effective for interviewers to determine its comprehensi- (88%), treatment assignment (84%), and the children; 20% chose finding the cheapest bility. The instrument was pretested with ability to discontinue participation (64%). drugs, collecting data for the Ugandan gov- parents who had been research participants Similarly, most reported that they felt in- ernment, or making money for the research at the site where the Luganda version of the formed about the study risks (65%), the num- team. Most knew they would have to bring survey was used. The in-person interview ber of visits (84%), the ability to quit (67%), their child to the clinic 7 times for the study format allowed participation regardless of the study purpose (67%), and treatment as- (78%), that drugs would be administered literacy level. Specially trained Ugandan personnel, fluent TABLE 2—Respondent’s Experience With the Informed Consent Process in both English and the language spoken at their assigned site, conducted the interviews. Reported Being Demonstrated Interviewers were paid and supervised sepa- Told About Topic Felt Informed Knowledge of rately from the malaria study and clinical Topic by Study Staff, % About Topic, % Specific Details, % staff. Neither the malaria study nor clinical Purpose of the study 77 67 80a staff had access to completed questionnaires, Risks of the study 61 65 18–45b which were sent to NORC for data entry. No. of clinic visits 88 84 78 Of the 353 individuals invited to partici- Way treatments are assigned 84 71 19c pate, 347 completed interviews, 5 terminated Option of quitting 64 67 65 their interviews before completion, and 1 in- a dividual refused to be interviewed (response Recognized that the reason the study was being conducted was to find the best treatment for children with malaria. b18% were able to name 1 or more side effects; 45% recognized 1 or more in multiple-choice questions (Table 3). rate: 98%). The mean duration of the inter- cAlthough 88% knew that all children would receive treatment for malaria, only 19% knew that not all children would receive views was 33 minutes, and the range was 7 the same drugs (Table 3). to 152 minutes.

July 2005, Vol 95, No. 7 | American Journal of Public Health Pace et al. | Peer Reviewed | Research and Practice | 1185  RESEARCH AND PRACTICE 

ment) believed that treatment would be de- TABLE 3—Comprehension of Study TABLE 4—Decisionmaking and Volition cided on the basis of what the doctors Details thought was best for each child. Percentage Percentage Univariate analyses indicated that feeling Made decision to join her- or himself 94 pressure because of the child’s sickness (P< Identified no. of required clinic visits 78 Knew she or he could have refused 41 .001) and, to a lesser extent, having com- Identified treatment administration 79 if so desired pleted more years of education (P=.042) as oral Degree of pressure felt from others to join were significantly associated with understand- Knew child’s blood would be taken 98 the study ing that not all children would receive the Knew child’s blood would be used 52 A lot of pressure 12 same malaria treatment. Feeling pressure be- for future research A moderate amount of pressure 2 cause of the child’s sickness was also signifi- Named 1 or more side effects 18 A little pressure 1 cantly associated with recognizing possible Accurately identified the following No pressure 85 side effects (P<.001 for each side effect). symptoms as possible side effects Felt pressure to join because child is sick 58a Knowledge that not all children would receive Fainting 20 Knew she or he could quit the study 65 the same malaria treatment was significantly Skin rash or itching 32 Knew she or he could quit at any time 17b lower at 1 of the rural sites than at the other Knew the following symptoms were sites (P<.001). a53% specified that they felt “a lot of pressure” not possible side effects because of their child’s illness. b Red eyes 65 Decisionmaking and Volition Asked only of the 25% of those who knew they could Hearing problems 70 quit at all. The primary reason most respondents gave Knew that all children would be treated, 19 for enrolling their child in the study was to but not with the same drug obtain malaria treatment. Several respondents Knew that doctors would determine 7a mentioned specifically that they enrolled their source), or a combination of these sources, treatment allocation according child because previous treatment attempts created the pressure. A smaller percentage of to chance had failed. A small number of parents (7%) respondents reported feeling pressure from a This question was asked only of the 19% who knew reported that their primary motivation was other doctors or nurses (5%) or health center that not all children would receive the same drug; the opportunity to learn about their child’s staff (4%). 39% of that group knew that treatment would be allocated by chance. sickness. One respondent said, “In this study, Just 41% of the respondents reported they they teach a lot of things related to malaria.” could have refused to enroll their child in the Another said that she enrolled her child “in malaria study, and 86% of this group (36% order to enable researchers to get the best of the respondents overall) stated that it orally (79%), and that the investigators would treatment for future kids,” and still another would have been at least moderately easy to take blood samples to study (98%). Only said that “when the doctor explained, I saw refuse. Overall, 25% of the respondents 52% knew that these samples would be used the importance of this study.” Others com- (56% of those who reported that they could for future research. In addition, only 18% mented that malaria was a major problem have refused) reported that if they had re- could name 1 or more side effects of the for their region (e.g., “In this village we have fused to participate, their child would still study drugs when asked an open-ended been suffering a lot”). have received malaria treatment. Of those question. When asked about treatment side Almost all of the respondents (94%) re- who stated that they could not have refused effects in a question involving a yes-or-no ported that they personally made the decision to participate, nearly all reported the reason format, 20% knew that fainting was possible, to enroll their child, although 22% indicated as being their child needed malaria treatment. and 32% knew that their children could ex- that another person helped them with the No one attributed it to pressure from others. perience a skin rash or itching (Table 3). decision (Table 4). Most commonly, help Sixty-five percent of the respondents knew While most of the respondents (88%) came from the malaria study team (13% that they could quit the study, but only a knew that all children taking part in the study overall), other doctors and nurses (13%), or quarter of these individuals (17% of respon- would receive malaria treatment, only 19% the health center (11%) rather than from dents overall) knew that they could quit at responded that not all children would receive spouses (6%) or family and friends (8%). any time, instead of when the treatment had the same treatment. Seven percent of the re- Many of the respondents (58%) felt pres- been completed or the doctor said they could spondents (39% of those who knew that not sure to join the study because their child was (Table 4). Fifty-three percent of respondents all children would receive the same treat- sick, yet most (85%) reported no pressure thought that their child would still receive ment) knew that treatment assignments from other people (Table 4). Among the 47 malaria treatment if they quit the study. would be determined according to chance, respondents who indicated feeling pressure Those who felt pressure from others to and 10% (59% of those who knew that not from others, family and friends, the study join the trial were more likely to feel pressure all children would receive the same treat- team, or respondents’ spouses (6% from each because of their child’s sickness (P<.001)

1186 | Research and Practice | Peer Reviewed | Pace et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

and to belong to a household headed by the clinic visits, administration of oral treatment, pation was the time commitment and number child’s other parent (P=.049). Female respon- and blood to be drawn. However, their level of clinic visits rather than concerns about dents were more likely than male respon- of understanding about the possible side ef- drug side effects. Because our respondents dents to feel pressure because of their child’s fects of study medications was substantially understood the nature of their time commit- sickness (P=.002), and they also were more lower. Only 61% of the respondents recalled ment far better than the risks involved with likely to know that they could quit the study being told about risks, and many fewer (18%) the study, their decision may have been (P=.05). Those with more education were could accurately name 1 or more side effects based on information they found most salient. more likely to know that they could quit at of the study drugs. While most (84%) re- A common worry among commentators any time (P=.008). membered being told how malaria treatments focusing on informed consent is that women The site variable showed a significant as- would be assigned, 69% incorrectly thought in Africa, and Uganda in particular, do not sociation with respondents’ belief that they that all children would receive the same treat- make independent decisions for themselves could have refused to join the study (all Ps ment. Of the small number who knew not all or their families because of accepted gender <.001). Significant variation by site was children would receive the same treatment, roles in their societies.21–23 Yet, in our study, noted in (1) the amount of pressure re- many fewer knew that treatment assignment nearly all of the predominantly female respondents felt because of their child’s illness, would be “based on chance, like pulling a spondents themselves made the decision to (2) previous research experience, (3) educa- number out of a hat,” although assignment enroll their child, with only 8 of 347 respon- tion, and (4) beliefs regarding whether the was explicitly described in this way in the dents reporting that the enrollment decision child could be treated if not enrolled in the consent form. was made by their spouse. Furthermore, study. However, there was no consistent vari- Previous informed consent studies con- among respondents who reported that al- ation by site. At 1 of the sites, many respon- ducted in both developed and developing though they decided about enrollment them- dents felt pressure from others and from their countries have revealed people’s understand- selves, they received some help with the deci- child’s sickness and did not know they could ing of randomization and of side effects often sion, spouses were the least common source quit if they wanted, whereas many partici- is poor.8,15–20 Such poor understanding may of help. Most respondents felt no pressure pants at another site did not know that they not be surprising in the context of an active from anyone to enroll their child in the study, could have refused to join the study or when controlled trial in which risks associated with and, among those few who did feel pressure, they could quit. different study arms are similar and, most im- spouses were infrequently the source. In addi- portant, similar to the risks of treatment out- tion, women knew more often than men that DISCUSSION side the trial. Furthermore, parents were prob- they could quit the study. ably aware that malaria treatment, even with However, similar to our data on compre- To our knowledge, this is the largest study side effects, is less risky than not treating ma- hension, our data on the extent to which re- to date of the quality of informed consent in laria. These children were ill, and their par- spondents’ enrollment decisions were volun- a developing country. Our data show that a ents were informed that all children would re- tary reveal a complex picture. More than half parent or guardian’s informed consent for a ceive treatment. Knowledge of the particular of the respondents felt pressure to join the child enrolled in this malaria treatment study drugs’ side effects or randomization may study because of their child’s sickness, and was of uneven quality. Most respondents had therefore have played no or only a minor role most cited their child’s need for treatment as a high level of comprehension of some of the in parents’ decisions to enroll their children, their main reason for enrolling. This may be study procedures, and most respondents re- or it may at least have played understandably surprising because the malaria study consent ported having made an autonomous decision less of a role than it might play in decisions form emphasized that children who did not to enroll, with few feeling any pressure from about enrolling in placebo-controlled or high- participate would still receive malaria treat- others. Conversely, respondent’s comprehen- risk trials or studies of nonurgent interven- ment from the clinic, and in theory this treat- sion of risks and randomization was relatively tions. ment was also free. Yet, fewer than half of poor, and responses regarding availability of Indeed, it may be difficult for investigators the respondents indicated that they could malaria treatment outside of the study and to impress such details upon parents who are have refused to join the study, and only the possibility of discontinuing participation worried, first and foremost, about treating 56% of these individuals reported that if raise questions about the extent to which re- their children’s sickness. From an ethical they had refused, their child could still have spondents’ decisions were voluntary. This standpoint, these details may not be critical received treatment. mixed portrait illustrates the complexity of for parents in making an informed decision To the extent that these responses reflect both obtaining informed consent and assess- about participation. Focus group discussions parents’ failure to grasp what they were told ing its quality in the context of an active con- (conducted in a parallel study [E. Okiria, un- (or investigators’ failure to thoroughly explain trolled trial of treatment for an emergent published data, May 2003]) with mothers alternatives to participation), this finding has condition in children. whose children were already participating in worrisome implications for the parents’ con- Respondents understood what the study the malaria study suggest, in fact, that for sent. Possibly, however, drug access was more required of their children, such as number of many parents the greatest burden of partici- limited among nonparticipants owing to dif-

July 2005, Vol 95, No. 7 | American Journal of Public Health Pace et al. | Peer Reviewed | Research and Practice | 1187  RESEARCH AND PRACTICE 

ferent funding sources. If so, respondents’ an- region. Finally, although the same written helped recruit and train interviewers, conducted paral- swers may say more about their unfortunate consent materials and procedures were used lel focus groups with parents of children in the malaria study, assisted with analysis of open-ended questions, health care situation than about problems at all of the sites, we did not observe the con- and reviewed the article. E.S. Garrett-Mayer was re- with informed consent. Although parents may sent process itself. Variation in communica- sponsible for the statistical analysis in collaboration have felt that their options were constrained tion styles and clarity of disclosure may have with C. Pace, E. Emanuel, and C. Grady. C. Pace and C. Grady prepared the article. All of the authors con- by their children’s sickness and limited access influenced site-specific differences. tributed substantially to the editing of the article. to treatment, it does not necessarily follow In conclusion, Ugandan parents or that enrollment choices were not voluntary.11 guardians who consented to enroll their chil- Acknowledgments Indeed, limited treatment options are one re- dren in a malaria study had mixed compre- This project was funded through the Department of minder of the importance, in this trial and hension of the study details and felt some Clinical Bioethics at the National Institutes of Health. We are grateful to the Uganda Malaria Surveillance others, of taking into account other critical pressure to enroll their children because their Project (USMP), particularly Kristin Banek and the re- aspects of research ethics even before anyone children were sick and needed treatment that search staff at the 4 sites; Michael Reynolds of the Na- is asked to provide informed consent, includ- the parents were not confident they could tional Opinions Research Center (NORC); our interviewers at 4 sites; and all of the study participants. ing whether short- and long-term benefits of otherwise obtain. Although these data raise Note. The opinions expressed here are those of the research are appropriately responsive to the questions about the quality of informed con- authors and do not reflect the policies and positions of needs of a community with few health care sent, they must be interpreted in the context the National Institutes of Health, the US Public Health 24 Service, or the US Department of Health and Human resources. of an active controlled trial of an emergent Services. In previous informed consent studies, par- intervention, in a setting where it may not al- ticipants were interviewed months or years ways have been possible to obtain treatment Human Participant Protection after the informed consent process,6 so recall outside the trial. Further debate is needed on This study was reviewed by Uganda Malaria Surveil- rather than comprehension was assessed. what is ethically necessary for informed con- lance Project investigators and approved by the institutional review boards of the National Institute for Al- Conversely, immediately after parents or sent in such a context. Such debate can pave lergy and Infectious Diseases, the NORC, and the guardians gave consent for their children to the way for future studies involving thought- Ugandan National Council for Science and Technology. be enrolled in the malaria study, we inter- ful assessments of these aspects of informed Participants provided verbal informed consent. viewed those parents and guardians to assess consent and evaluations of strategies designed their knowledge, experience, and perspectives to improve the quality of consent in diverse References 1. Angell M. Investigators’ responsibilities for human concomitant with their decision to allow their trial settings. subjects in developing countries. N Engl J Med. 2000; children to participate. Nonetheless, there are 342:967–969. limitations associated with our data. The ur- 2. Annas G, Grodin M. Human rights and maternal- gent nature of the malaria treatment trial is a About the Authors fetal HIV transmission prevention trials in Africa. Am At the time of this study, Christine Pace was with the De- J Public Health. 19 9 8;88:560–563. crucial context for interpreting our results and partment of Clinical Bioethics, Warren G. Magnuson Clin- 3. French H. AIDS research in Africa: juggling risks ical Center, National Institutes of Health, Bethesda, Md. may limit the generalizability of our findings and hopes. New York Times. October 9, 1997:A1. to different types of trials. Furthermore, al- David Wendler, Ezekiel Emanuel, and Christine Grady are with the Department of Clinical Bioethics, Warren G. 4. LaFraniere S, Flaherty M, Stephens J. The though 4 different sites in Uganda were in- Magnuson Clinical Center, National Institutes of Health. dilemma: submit or suffer. Washington Post. December volved, these sites may not represent all of Ambrose Talisuna and Faustin Maiso are with the Uganda 19, 2000:A1. Uganda or other African settings. Malaria Surveillance Project, Ministry of Health, Kampala, 5. Fitzgerald D, Marotte C, Verdier R, Johnson W, Uganda. Fred Wabwire-Mangen and Nathan Bakyaita are Pape J. Comprehension during informed consent in a Conducting the study at 4 different sites with the Uganda Malaria Surveillance Project, Makerere less-developed country. Lancet. 2002;360:1301–1302. provided us with a broad sample of respon- University, Kampala. Edith Okiria is with the Depart- 6. Joubert G, Steinberg H, Ryst E, Chikobvu P. Con- ment of Women and Gender Studies, Makerere Univer- dents but also resulted in site variability for sent for participation in the Bloemfontein Vitamin A sity. Elizabeth S. Garrett-Mayer is with the Department of Trial: how informed and voluntary? Am J Public Health. which we could not entirely account. For ex- Oncology, Division of Biostatistics, Johns Hopkins School 2003;93:582–584. ample, more people at 2 of the sites than at of Medicine, Baltimore, Md. the other 2 sites reported pressure to join the Requests for reprints should be sent to Christine Grady, 7. Karim Q, Karim S, Coovadia H, Susser M. In- PhD, RN, Section on Human Subjects, Department of formed consent for HIV testing in a South African study because their child was sick, whereas, Clinical Bioethics, National Institutes of Health, Bldg 10, hospital: is it truly informed and truly voluntary? Am at a third site, more people responded that Room 1C118, Bethesda, MD 20892-1156 (e-mail: J Public Health. 19 9 8;88:637–640. their child could not be treated if they refused [email protected]). 8. Leach A, Hilton S, Greenwood B, et al. An evalua- This article was accepted August 25, 2004. to join. Site variations were not statistically as- tion of the informed consent procedure used during a trial of a Haemophilus influenzae type B conjugate vac- sociated with differences in respondents’ char- cine undertaken in The Gambia, West Africa. Soc Sci acteristics, but they may have been associated Contributors Med. 1999;48:139–148. C. Pace, A. Talisuna, D. Wendler, E. Emanuel, and with differences in the informed consent pro- C. Grady designed and planned the study and devel- 9. Lynoe N, Hyder Z, Chowdhury M, Ekstrom L. Ob- cess or access to health services, or they may oped the interview script. F. Maiso monitored the day- taining informed consent in Bangladesh. N Engl J Med. 2001;344:460–461. even have been influenced by the use of dif- to-day operations of the study in Uganda and coordinated data entry. F. Wabwire-Mangen and N. Bakyaita 10.Pitisuttithum P, Migasena S, Laothai A, ferent interviewers or different translations to reviewed the proposed design and identified sites and Suntharasamai P, Kumpong C, Vanichseni S. Risk be- accommodate the languages spoken in each procedures for implementation in Uganda. E. Okiria haviors and comprehension among intravenous drug

1188 | Research and Practice | Peer Reviewed | Pace et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

users volunteered for an HIV vaccine trial. J Med Assoc Thai. 19 97;80:47–50. 11.Faden RR, Beauchamp TL, King NM. A History Community-Based Public Health: and Theory of Informed Consent. New York, NY: Oxford University Press Inc; 1986. A Partnership Model 12. Levine R. Ethics and Regulation of Clinical Re- Edited by Thomas A. Bruce, MD, and Steven Uranga search. New Haven, Conn: Yale University Press; 1988. McKane, DMD 13. Applebaum P, Lidz C, Meisel A. Informed Consent. New York, NY: Oxford University Press Inc; 1987. Published by APHA and the W.K. Kellogg Foundation 14 .Wendler D, Pace C, Talisuna A, Maiso F, Grady C, Emanuel E. Research on stored biological samples: the eveloping meaningful partnerships with the views of Ugandans. IRB Ethics Hum Res. 2005;27(2): communities they serve is crucial to the success 2–5. D of institutions, nonprofit organizations and corpora- 15. Bergler J, Pennington A, Metcalfe M, Freis E. In- tions. This book contributes to a wider understanding formed consent: how much does the patient under- ISBN 0-87553-184-9 stand? Clin Pharmacol Ther. 19 8 0;27:435–439. 2000 ❚ 129 pages ❚ softcover of how to initiate and sustain viable partnerships and 16. Howard JM, DeMets D. How informed is in- $17.00 APHA Members improve community life in the process. Community- formed consent? The BHAT experience. Controlled Clin $22.00 Nonmembers Based Public Health: A Partnership Model focuses on Trials. 19 81;2:287–303. plus shipping and handling public health practice in communities, the education 17. Estey E, Wilkin G, Dossetor J. Are research sub- and training of public health professionals at colleges jects able to retain the information they are given during the consent process? Health Law Rev. 1994;3: and universities, and public health research and 37–41. scholarly practice within academic institutions. 18. Hietanen P, Aro A, Holli K, Absetz P. Information and communication in the context of a trial. Eur J Can- cer. 2000;36:2096–2104. ORDER TODAY! 19. Joffe S, Cook E, Cleary P, Clark J, Weeks J. Quality American Public Health Association of informed consent in cancer clinical trials: a cross- Publication Sales sectional survey. Lancet. 2001;358:1772–1777. Web: www.apha.org E-mail: [email protected] 20.Van Stuijvenberg M, deVos S, Tjiang GCH, Tel: 888-320-APHA Steyerberg EW, Derksen-Lubsen G, Moll HA. Informed FAX: 888-361-APHA KL02J2 consent, parental awareness and reasons for participating in a randomized controlled study. Arch Dis Child. 1998;79:12 0–125. 21. The Ethics of Research Related to Healthcare in De- veloping Countries. London, England: Nuffield Council on Bioethics; 2002. 22. Ethical and Policy Issues in International Research: Clinical Trials in Developing Countries. Bethesda, Md: National Bioethics Advisory Commission; 2001. 23. Loue S, Okello D, Kawuma M. Research bioethics in the Ugandan context: a program summary. J Law Med Ethics. 19 96;24:47–53. 24.Participants in the 2001 Conference on Ethical Aspects of Research in Developing Countries. Fair benefits for research in developing countries. Science. 2002;299:2133–2134.

July 2005, Vol 95, No. 7 | American Journal of Public Health Pace et al. | Peer Reviewed | Research and Practice | 1189  RESEARCH AND PRACTICE 

Maternal Risk Factors for Fetal Alcohol Syndrome in the Western Cape Province of South Africa: A Population-Based Study

| Philip A. May, PhD, J. Phillip Gossage, PhD, Lesley E. Brooke, BA(Hons), Cudore L. Snell, DSW, Anna-Susan Marais, RN, Loretta S. Hendricks, Julie A. Croxford, RN, BA(Hons), and Denis L. Viljoen, MD

The search for specific maternal risk factors Objectives. We defined risk factors for fetal alcohol syndrome (FAS) in a region for fetal alcohol syndrome (FAS) has been on- with the highest documented prevalence of FAS in the world. going for more than 2 decades via prenatal Methods. We compared mothers of 53 first-grade students with FAS (cases) with 1–6 7–10 clinic and epidemiological studies. 116 randomly selected mothers of first-grade students without FAS (controls). Population-based research is particularly help- Results. Differences between case and control mothers in our study population ful in identifying traits of the very highest-risk existed regarding socioeconomic status, religiosity, education, gravidity, parity, mothers—those who have borne children and marital status. Mothers of children with FAS came from alcohol-abusing fam- with full-blown FAS—and in designing pre- ilies in which heavy drinking was almost universal; control mothers drank little vention strategies.10–14 to no alcohol. Current and past alcohol use by case mothers was characterized FAS has been associated with heavy, by heavy binge drinking on weekends, with no reduction of use during pregnancy in 87% of the mothers. Twenty percent of control mothers drank during episodic (binge) drinking that produces high pregnancy, a rate that declined to 12.7% by the third trimester. The percentage blood alcohol concentration (BAC); ad- who smoked during pregnancy was higher for case mothers than for control vanced maternal age; high gravidity and par- mothers (75.5% vs 30.3%), but the number of cigarettes smoked was low among ity; unstable marital status; cigarette use; case mothers. The incidence of FAS in offspring of relatively young women 5,15–18 and use of other drugs. In the United (28 years) was not explained by early drinking onset or years of drinking (mean, States, higher FAS rates are reported among 7.6 years among case mothers). In addition to traditional FAS risk factors, case Black and American Indian women, low– mothers were smaller in height, weight, head circumference, and body mass socioeconomic status (SES) groups, people index, all anthropomorphic measures that indicate poor nutrition and second- with high scores on various alcohol abuse generation fetal alcohol exposure. assessment tools, and women with alcoholic Conclusions. Preventive interventions are needed to address maternal risk fac- male partners.19–24 Studies of mothers of tors for FAS. (Am J Public Health. 2005;95:1190–1199. doi:10.2105/AJPH.2003. 037093) children with fetal alcohol spectrum disorder (FASD; referred to by the Institute of Medi- cine18 as FAS, partial FAS, alcohol-related FAS is associated with low SES among sub- ences in susceptibility.10,42–44 Some alcohol- birth defects, and alcohol-related neurode- populations23,33 in developed and developing abusing families appear to escape many velopmental deficits) point to a dose– countries.9,34 In South Africa,8,35 mothers of symptoms of FASD.7,33,45 Families with 1 or response effect. The probability of anom- children with FAS were of lower SES than more children with FASD experience serious alies such as microcephaly, craniofacial de- were control mothers. In 1 US study that physical and mental problems that pose a fects, and behavioral problems depends on compared women of differing SES who con- challenge to all types of service providers.46 the level of alcohol exposure as modified by sumed 12 drinks daily, the rate of FAS was Because maternal risk for FASD involves an certain maternal characteristics, such as 45 times greater in women of low SES than interaction of biological, familial, historical, those on which this article reports.5,25–28 in women of middle and upper SES.34 social, and psychological factors,46 research The rate of FAS in US children is 0.05 to In the United States, England, and Canada, and prevention foci are interdisciplinary.14 2.0 per 1000 births.29 All levels of FASD 20%–32% of pregnant women drink, and in In the general literature on alcohol abuse, affect, at minimum, 1% of the birth popula- some European countries the rate is higher, maternal risk factors for FASD include smok- tion.30 The highest rates of FAS in the world exceeding 50%.13–18,36–39 In the Western ing; abusing drugs; cohabiting with an alco- have been reported in the Republic of South Cape Province, 34% of urban women and holic male partner; sexual dysfunction; having Africa. The rate of full-blown FAS alone has 46%–51% of rural women drink during alcohol-abusing parents; initiating drinking at been reported to be 46 cases per 1000 pregnancy.40,41 Maternal drinking during an early age; and having low self-efficacy, births in the Western Cape Province.8 Cur- pregnancy varies among and within popula- poor life goals, and few interests.3,19,48–55 Pro- rent research is documenting even higher tions throughout the world.33 tective factors identified as providing strong rates in Western Cape Province 31 and high That alcohol abuse and FAS cluster in normative or cultural support for abstinence rates elsewhere in South Africa.32 families implies both social and genetic influ- or light drinking include high education; reli-

1190 | Research and Practice | Peer Reviewed | May et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

giosity; and unique social, psychological, bio- of the 12 public schools of this area were Mothers taking the questionnaire were asked logical, and genetic traits.9,35,56–58 Neverthe- screened for height, weight, and head cir- to recall behavior and conditions before, dur- less, many risk factors for FASD are not well cumference. Children at or below the 10th ing, and after gestation of the index child. understood, and their explication is vital for percentile on height and weight and head During our study, community residents prevention efforts.14,59,60 circumference charts (n = 300) were ad- had little knowledge about FAS and there- vanced to a second tier of the diagnostic fore little stigma regarding maternal drinking. Background of the Region process. Two teams of dysmorphologists, Mothers did not know whether their children We describe a study9 of maternal risk for who were blinded to each child’s medical had FAS at time of interview, because diag- FASD in a town and its rural areas (popula- history and to one another’s findings, con- noses had not been finalized. Nevertheless, to tion=45225; 22% of the area is rural) in the ducted a physical examination and mea- establish rapport, nonthreatening questions Western Cape Province of South Africa. Most sured all features of FAS, recording findings were asked regarding birth and childhood, inhabitants are “Colored,” defined as racially on a quantified dysmorphology checklist in occupation, education, diet, reproduction, mixed individuals of African, European, and which a high score indicates more FASD and general health. Alcohol consumption Asian descent. The town is similar to many features.8 Ninety-three children who exhib- responses are more accurate in such a others in this agricultural and wine-producing ited physical symptoms of FAS that were format, especially in the context of dietary region. Heavy, episodic drinking has been the less consistent or severe were administered questions.73 Respondents were first asked norm among laborers for generations. For psychological and life skills tests72 to assess about the drinking habits of their relatives several centuries, alcohol was provided daily development.19 Next, the biological mothers and friends. The context, quantity, and fre- to farmworkers as partial payment for work, a of these children were located and inter- quency of the mother’s current drinking were system known as the “dop” system, after the viewed about maternal risks. In a formal then explored by means of a 1-week, day-by- Afrikaans word for drink. Though this system case conference on each child, findings of day log. Drinks were measured in standard of payment was formally outlawed by multi- all tests/examinations were reviewed, and ethanol units: 340 mL of beer, 120 mL of ple statutes years ago, its effects persist. Local final diagnoses were made. Sixty-four chil- wine, or 44 mL of distilled spirits (5%, 11%, people who are forced to tolerate low pay, dren were diagnosed with FAS. Maternal in- and 43% ethanol, respectively). Respondents limited opportunity, and humble living condi- terviews were completed for 53 of the 64 were shown pictures of standard containers tions value alcohol as a favored commodity. mothers of children with FAS who were of local brands. Questions on current drink- Frequent binge drinking, defined as 3 or alive and who could be contacted. These in- ing became benchmarks for reconstructing more drinks per episode of drinking, is com- terviews provided the data for our study. maternal drinking during pregnancy, and mon. South Africa researchers have docu- All Sub-A children, with the exception of for aiding in memory recall to accurately mented high levels of alcohol abuse among the 93 children suspected to have FAS, were measure the amount of alcohol consumed, male workers of the region.61–64 Although no eligible for selection as control children. when the alcohol was consumed, and over formal dop system survives, drinking heavily One hundred forty-six were selected with a what duration of time the alcohol was con- in groups on weekends and holidays remains random-number table; their mothers were sumed. These questions were asked using the a common form of recreation. Commercially potential controls. Of these mothers, 30 were timeline-follow back method, a method that produced beer and wine are cheap, readily not included for the following reasons: 15 associates drinking with particular events, available, and consumed by a population that, (10.3%) mothers could not be located or con- such as illness, holidays, and celebrations, to although poor, can allocate enough money to tacted, the children of 12 (8.2%) mothers pinpoint the amount of drinking that occurred obtain and consume substantial quantities were in foster or adoptive placement, and 3 during each stage of pregnancy and during over short periods of time. This pattern re- (2.1%) mothers refused participation. The any celebrations or events that occurred sults in high BAC values, placing fetuses at final control sample contained 116 mothers. while the woman was pregnant.74,75 risk for FASD.65–69 We refer to this pattern as The development of the children selected as Smoking was explored more directly, be- the “dop legacy.” control children was assessed in exactly the cause smoking purchases and practices were Maternal drinking was identified as a seri- same manner as described for case children; more easily remembered and reported, be- ous health problem in Western Cape Province none had major anomalies. All mothers were cause, unlike drinking, which, in this culture, in the mid-1990s.70,71 Research confirmed administered identical questionnaires and re- occurs in groups where drinks are often high rates of FAS.8,9,72 We describe risk fac- ceived incentive gift baskets of food staples. shared, cigarettes are not shared; an individ- tors for FAS to improve FAS prevention ef- The questionnaire was developed specifi- ual must take time from any activities to con- forts in this and similar communities. cally for the Western Cape Province popula- struct a hand-rolled cigarette for smoking. tion by adapting items and techniques from Respondents were asked about tobacco con- METHODS studies in various US ethnic populations. sumed currently and during pregnancy. One After pilot testing and use with more than hand-rolled cigarette in South Africa was Beginning in 1999, all children in the 100 women in a previous South Africa found in pilot field trials to contain 1 g of to- Sub-A (first grade) public school classrooms study,9 this version contained 240 items. bacco; prerolled cigarettes were rare and

July 2005, Vol 95, No. 7 | American Journal of Public Health May et al. | Peer Reviewed | Research and Practice | 1191  RESEARCH AND PRACTICE 

TABLE 1—Demographic, Socioeconomic, and Reproductive Characteristics of Mothers of were counted directly. Questions also as- Children With Fetal Alcohol Syndrome (FAS) (n=54) and Randomly Selected Control sessed other drugs used. Body mass index Mothers (n=116): South Africa, Wave II, 1999–2001 (BMI) was calculated by means of the following metric formula76: weight in kilograms/ Mothers of Children With FAS Control Mothers P (OR) (height in meters2). Age at interview, y, mean (SD) 35.9 (6.3) 34.5 (6.4) NSa Interviews were conducted in Afrikaans, Residence during index pregnancy, % the primary regional language. Over 95% of Rural 66.0 20.9 the participants were Colored; the remainder Urban 34.0 79.1 <.001b (7.36) were Blacks and Whites. We used Epi Info a Educational attainment, y mean (SD) 5.0 (3.2) 8.7 (2.6) <.001 (US Centers for Disease Control and Preven- Frequency of church attendance, % tion, Atlanta, GA) for analyses to compare Never 14.3 4.9 groups with 2-tailed statistical tests (P<.05), Not very often (<1 times per month) 49.0 17.5 and to calculate odds ratios (ORs). Pearson Often (1–2 times per month) 14.3 37.9 correlation coefficients were used. We com- Very often (1 time per week) 22.4 39.8 <.001b Frequency of praying, % pared the characteristics of mothers of chil- Never 5.6 0.9 dren with FAS (case mothers) with the char- Not very often (<1 time per week) 14.8 9.8 acteristics of randomly selected mothers with Often (2–3 times per week) 53.7 22.3 normal children (control mothers) from the Very often (1 time per day) 25.9 67.0 <.001b same schools. Religiosity index score,c mean (SD) 3.0 (0.9) 3.5 (0.7) <.001a Currently employed, % 69.8 66.7 NSb RESULTS Usual occupation, % Factory worker 9.6 17.7 In this study, 54 of 64 mothers of children Farmworker 34.6 12.4 with FAS were located alive, and 1 declined Office worker 11.5 16.8 an interview (1.9%), yielding 83% participa- Housewife 3.8 8.0 Domestic (housekeeper, servant) 15.4 8.8 tion. After comparing characteristics of the Other 0.0 12.4 children of the 53 case mothers interviewed Usually does not work 25.0 23.9 <.003b with the characteristics of children of the 11 Employment status, % case mothers not interviewed, 1 significant Full–time 42.3 53.2 difference was found. The head circumfer- Part–time 11.5 6.3 ence percentile of children of mothers who Seasonal 15.4 9.0 were not interviewed was significantly larger d b Unemployed 30.8 31.5 NS (14.9 vs 6.1) than that of children of mothers Weekly income when working, mean (SD) who were interviewed. Height, weight, verbal Rands 105.5 (101.9) 252.9 (339.6) .002a and nonverbal ability, behavioral test scores, US$ 17.58 42.15 and dysmorphology scores were not signifi- Reproductive variables, mean (SD) Gravidity 3.5 (1.4) 2.8 (1.1) <.000a cantly different for FAS children of mothers Parity, pre- and full term 3.3 (1.4) 2.7 (1.1) .002a who were interviewed versus FAS children of Miscarriages 0.2 (0.5) 0.2 (0.5) NSa mothers who were not interviewed. For the Still births 0.0 (0.0) 0.0 (0.2) NSa FAS children of mothers who were inter- Living children 3.0 (1.2) 2.5 (1.0) .002a viewed versus FAS children of mothers who Age at birth of index child, y 28.0 (6.4) 26.1 (6.3) NSa were not interviewed, there was also 1 differ- a Birth order of index child 2.8 (1.4) 2.1 (1.2) <.001 ence: dysmorphology scores for children of Marital status during pregnancy with index child, % the interviewed mothers were higher than Married 29.6 40.9 scores for children of the noninterviewed Unmarried, living with partner 40.7 10.9 mothers (2.5 vs 1.4, P =.034). Otherwise, Separated/divorced/widowed 1.9 1.8 FAS children were similar on all physical and Single 27.8 46.4 <.001b behavioral measures. Note. NS=nonsignificant.The control group was used as referent. As can be seen in Table 1, 3 of the social a t test. and demographic variables were not signifi- bχ2 test. cCombined values for frequency of church attendance and prayer. cantly different for case mothers versus con- dCombined categories of unemployed, not employed because of disability, and not employed and not looking for work. trol mothers: age at interview, current employment, and full- or part-time employment.

1192 | Research and Practice | Peer Reviewed | May et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

However, case mothers were more likely than and control children approached significance when control group drinkers (21%), women control mothers to be rural residents during (P =.07). who were currently drinking at the time of index pregnancies (OR=7.36), to be em- The quantity of alcohol consumed by fa- the interview or who had consumed alcohol ployed on farms as their usual occupation thers, mothers, brothers, and sisters of case in the week prior to the interview, were con- (35% vs 12%), and to have lower incomes. mothers was significantly greater than for sidered (13.4 vs 11.9 years). By the time of Educational attainment of control mothers control mothers; fathers of case mothers interview, 30.2% of case mothers had quit was 74% higher than that of case mothers drank more than fathers of control mothers drinking, but a significant difference in cur- (8.7 vs 5 years). Religious practices were (Table 2). More than 95% of the fathers of rent drinking remained between control scored significantly lower among case moth- case children drank during the course of the mothers and case mothers. ers (frequency of church attendance, praying, study, consuming a reported 81 drinks per Case mothers who drank at the time of the and mean religiosity index). month during the index pregnancies. interview consumed 15.2 drinks per week Case mothers had greater measurements There was no difference in the age at (2.8 times the consumption rate for control than control mothers for gravidity, parity, liv- which women began drinking (Table 3). How- mothers who drank); 96% binged in the ing children, birth order of index child, and ever, total past years of drinking at time of in- week before the interview. About 90% of all cohabiting when not married. Miscarriage terview differed significantly (13.4 years for alcohol was consumed on weekends by both and stillbirth rates did not differ. The differ- case mothers vs 3.7 years for control moth- groups. The standard for case mothers was ence in maternal age at birth of case children ers, OR=8.14) and differed nonsignificantly much higher than for control mothers (13.1–11.2 drinks/week vs 2.9–4.7 drinks/ week). Of case mothers, 39.6% drank more TABLE 2—Reported Drinking Habits (No. Drinks per Month) of Family and Friends of than the group mean of 12.6 drinks per week, Mothers of Children With Fetal Alcohol Syndrome (FAS) (n=54) and Randomly Selected and 24.5% drank 18 or more drinks per Control Mothers (n=116): South Africa, Wave II, 1999–2001 week. Only 1.8% of control mothers who Mothers of Children With FAS Control Mothers t Test drank consumed 12.6 or more drinks per Total Drinkers Total Drinkers Total Drinkers week, and 6.4% consumed 6 or more drinks Sample Only Sample Only Sample Only per week. Because most drinks were consumed on weekends, average daily consump- Father tion by case mothers on drinking days was No. 48 44 79 55 <.001 .012 7.6 drinks if consumption took place over 2 Mean (SD) 63.3 (57.1) 69.0 (56.2) 31.7 (35.7) 45.5 (34.6) days but 5.1 drinks if consumption took place Mother over 3 days. At the time of interview, average No. 50 31 91 35 .027 NS daily consumption of the upper 25% of case Mean (SD) 28.7 (36.9) 46.2 (37.3) 15.7 (30.7) 40.7 (38.0) mothers was 9.0–24.2 drinks if consumption First brother took place over 2 drinking days and 6.0–16.1 No. 44 39 82 50 .034 NS drinks if consumption took place over 3 days. Mean (SD) 51.8 (70.8) 58.4 (72.7) 28.8 (48.7) 47.2 (55.1) During pregnancy, case mothers were sig- Second through sixth brothers nificantly more likely than control mothers No. 31 26 37 24 NS NS to drink during all trimesters. Case mothers Mean (SD) 24.0 (25.7) 33.6 (26.9) 22.0 (29.6) 36.0 (29.9) drank at least as much in the months before First sister pregnancy (90.3%) and in all trimesters as No. 37 24 69 21 .002 NS they did at the time of the interview. Mean (SD) 23.7 (32.8) 36.5 (34.5) 7.9 (18.8) 26.0 (26.6) Beer was the most consumed and favored Second through sixth sisters beverage for both case and control mothers No. 27 22 41 15 NS NS (59% vs 71%), followed by wine (45% vs Mean (SD) 23.0 (20.8) 36.8 (33.7) 11.4 (24.9) 34.8 (33.2) 20%) and spirits (5% vs 6.5%). Four percent Woman’s best friend of the case mothers reported having had a No. 32 13 64 13 .015 NS problem with alcohol abuse (compared with Mean (SD) 12.3 (24.6) 30.4 (31.1) 3.2 (11.7) 15.6 (22.5) 2% of the control mothers); 2% of case moth- Father of index child during ers had received treatment. index pregnancy Smoking was common among both groups. No. 49 47 96 70 <.001 .002 No significant differences were found in age Mean (SD) 81.1 (81.7) 84.6 (81.7) 34.6 (45.9) 47.5 (47.8) at which smoking commenced. Current Note. NS=nonsignificant. smoking was 66% for case mothers and 30% for control mothers; however, quantity

July 2005, Vol 95, No. 7 | American Journal of Public Health May et al. | Peer Reviewed | Research and Practice | 1193  RESEARCH AND PRACTICE 

TABLE 3—Drinking and Smoking Behaviors of Mothers of Children With Fetal Alcohol Syndrome (FAS) (n=54) and Randomly Selected Control Mothers (n=116): South Africa, Wave II, 1999–2001

Mothers of Children With FAS Control Mothers P (OR) Total Sample Drinkers Onlyc Total Sample Drinkers Only Total Drinkers (n=53) (n=35) (n=109) (n=19) Sample Only

Drinking behavior Age first drank alcohol, y, mean (SD) 19.4 (4.5) 19.6 (4.6) NSa Age began drinking regularly, y, mean (SD) 20.4 (4.3) 20.3 (3.7) NSa No. years of drinking alcohol, mean (SD) 13.4 (7.3) 3.7 (5.7) <.001a Current drinker, % 69.8c 21.1 <.001b (8.14) Current alcohol consumption (no. drinks per week), mean (SD) 12.6 (13.1) 15.2 (11.2) 1.0 (2.9) 5.4 (4.7) <.001a <.001a Binge drinking (3+ drinks) 1 or more days during past week, % 69.6 96.4 5.5 31.3 <.001b (39.31) <.000b (59.40) Current alcohol consumption on weekends (i.e., Friday, Saturday, 11.1 (11.1) 13.6 (8.9) 0.9 (2.7) 5.0 (4.2) <.001a <.000a Sunday), mean no. drinks (SD) Proportion of alcohol consumed during weekends, % 88.1 89.5 90.0 92.6 Before pregnancy, % Drank about the same (vs current use) 61.5 19.4 Drank less (vs current use) 1.9 2.2 Drank more (vs current use) 28.8 9.7 Did not drink 7.7 67.7 Stopped during this period 0.0 1.1 Drank during index pregnancy, % 92.4d 19.5 During first trimester of pregnancy, % Drank about the same (vs current use) 54.7 11.7 Drank less (vs current use) 5.7 3.6 Drank more (vs current use) 32.1 4.5 Did not drink 7.5 78.4 Stopped during this period 0.0 1.8 <.001b During second trimester of pregnancy, % Drank about the same (vs current use) 52.8 8.1 Drank less (vs current use) 5.7 2.7 Drank more (vs current use) 34.0 2.7 Did not drink 7.5 80.2 Stopped during this period 0.0 6.3 <.001b During third trimester of pregnancy, % Drank about the same (vs current use) 54.7 5.5 Drank less (vs current use) 1.9 5.5 Drank more (vs current use) 32.1 1.8 Did not drink 7.5 85.5 Stopped during this period 3.8 1.8 <.001b Beverage of choice, %e Beer 58.5 19.6 <.001a (5.76) Wine 45.3 5.4 <.001a (14.62) Spirits 5.7 1.8 NSa Combination 1.9 0.9 NSa Ever had a problem with alcohol abuse, % 4.0 1.9 NSb Ever received treatment for alcohol abuse, % 1.9 0.0 NSb Smoking behavior Age first used tobacco, y, mean (SD) 18.3 (3.5) 18.6 (3.8) NSa Age began smoking regularly, y, mean (SD) 18.5 (3.7) 19.3 (4.3) NSa

Continued

1194 | Research and Practice | Peer Reviewed | May et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

TABLE 3—Continued

Current smoker, % 66.0 30.1 <.001b (4.52) Current tobacco consumption, g/wk, mean (SD) 27.5 (32.0) 38.2 (32.4) 9.3 (17.1) 27.9 (19.1) <.001a NSa Before pregnancy , % Smoked about the same (vs current use) 61.5 15.7 Smoked less (vs current use) 7.7 10.2 Smoked more (vs current use) 5.8 3.7 Did not smoke 25.0 70.4 Stopped during this trimester 0.0 0.0 <.001b Smoked during index pregnancy, % 75.5 30.3 <.001b (7.08) During first trimester of pregnancy, % Smoked about the same (vs current use) 58.5 14.8 Smoked less (vs current use) 13.2 10.2 Smoked more (vs current use) 5.7 2.8 Did not smoke 22.6 71.3 Stopped during this trimester 0.0 0.9 <.001b During second trimester of pregnancy, % Smoked about the same (vs current use) 62.3 13.5 Smoked less (vs current use) 9.4 9.0 Smoked more (vs current use) 5.7 3.6 Did not smoke 22.6 72.1 Stopped during this trimester 0.0 1.8 <.001b During third trimester of pregnancy, % Smoked about the same (vs current use) 64.7 12.6 Smoked less (vs current use) 7.8 9.0 Smoked more (vs current use) 3.9 3.6 Did not smoke 23.5 73.0 Stopped during this trimester 0.0 1.8 <.001b Current drinker and smoker, % 73.6 11.2 <.001b (22.07)

Note. Sample sizes for drinking behavior were mothers of children with FAS, total sample, n=53; mothers of children with FAS, drinkers only, n=35; control mothers, total sample, n=109; control mothers, drinkers only, n=19. Sample sizes for smoking behavior were: mothers of children with FAS, total sample, n=52; mothers of children with FAS. smokers only, n=34; control mothers, total sample, n=99; control mothers, smokers only, n=33. at test. bχ2 test. cAlthough 37 of the 53 women interviewed reported that they were current drinkers, only 35 had consumed alcohol during the past week.Therefore, the data on current quantity and frequency are based on the responses of the 35 women who reported current drinking. dFour women whose children were diagnosed with FAS did not admit to drinking during the index pregnancy.When this inconsistency was revealed during the diagnostic case conferences, these women’s children were reassessed. In 2003, 2 dysmorphologists confirmed diagnosis of FAS in each of the 4 cases, as did the doctors evaluating the results of psychological tests. Institute of Medicine criteria allow for a diagnosis of FAS without confirmation of maternal drinking. ePercentage of beverages reported as favorites and reported as consumed exceeded 100%, because some mothers reported 2 favorites.

consumed by smokers in the 2 groups was General physical measurements (Table 4) (–.29), weight (–.24), and height (–.23). In not significantly different (38 vs 28 g/week). revealed that case mothers were signifi- other words, the smaller mothers appear to Rural women were more likely than town- cantly smaller than control mothers on be more likely to produce children with FAS dwelling women to smoke. Most women height, weight, head circumference, and than do the larger mothers. “rolled their own,” and this practice, along BMI. Head circumference and weight were with rural women’s low income, tends to especially reduced for mothers of FAS chil- DISCUSSION limit quantity of use. During pregnancy, dren, as indicated by tests of significance. 76.5%–77.4% of case mothers smoked. The significant negative correlation coeffi- Limitations and Strengths of the Study After we combined current drinking and cients indicate that lower values on the In a previous maternal risk study in South smoking, 73.6% of case mothers reported mother’s physical measures were associated Africa,9 35 of 46 (76%) mothers of children both smoking and drinking (compared with with higher dysmorphology scores of their with FAS were located alive (13% had died), 11.3% of control mothers). children: occipitofrontal circumference and 100% of those who were located agreed

July 2005, Vol 95, No. 7 | American Journal of Public Health May et al. | Peer Reviewed | Research and Practice | 1195  RESEARCH AND PRACTICE 

TABLE 4—Anthropometric Measures of Mothers of Children With Fetal Alcohol Syndrome much greater for case mothers than for con- (FAS) (n=54) and Randomly Selected Control Mothers (n=116): South Africa, Wave II, trol mothers in all comparisons. 1999–2001 Control mothers were more likely to have been abstainers or light drinkers compared Mean (SD) with case mothers, who showed significantly Correlation with Child’s heavier drinking patterns and reported drink- Mothers of Control Dysmorphology ing at the same level (53%–55%) or higher Variable Children With FAS Mothers P a Score (Total Sample) during pregnancy (32%–34%) compared Height, cm 154.3 (6.6) 157.1 (7.6) .034 –.23 with current drinking levels. As noted previ- Weight, kg 59.6 (14.2) 67.9 (15.2) .002 –.24 ously,8 South Africa case mothers often de- Occipitofrontal circumference, cm 53.8 (0.8) 54.7 (1.5) .001 –.29 scribed stressful life events as causes of heavy Body mass index 24.9 (5.5) 27.4 (5.9) .019 –.17 maternal drinking during pregnancy. A higher risk for FAS clearly exists among those of the at test. lowest SES.8,9

Dop Legacy to an interview. Such high numbers located fusal (20%) has been much lower than in US Most of the alcohol consumed was ob- and participating are unparalleled in the FAS studies, and candid reporting seems likely in tained commercially. Only 5% of the women literature. However, this study of mothers of this population thus far. Because all case in this study reported having received alcohol confirmed FAS children has limitations. Data mothers in our study bore a child with FAS, through the dop system (i.e., as payment for were collected retrospectively for the previous we describe only the very-highest-risk moth- labor) in their lifetimes (14% of case mothers 7years. Accuracy of recall may be a problem ers, as only a few other studies have done.9,10 vs 1% of control mothers). Of the case moth- despite our efforts to reconstruct accurate pat- This study’s detailed data on drinking among ers, 2% reported having received dop during terns from reports of current daily drinking mothers of children with FAS are unique. the index pregnancy, and 0.7% reported hav- and timeline-follow back methods. The study Most studies of maternal drinking during ing received dop at time of interview. Because population also had limited formal education, pregnancy are among lower-risk women. Ma- of these low rates of actual historical and which may have affected the quality of self- ternal risk is relative and variable between contemporary contact with the dop system reported alcohol consumption data. Further- and among populations,58 underscoring the among the study population, the contempo- more, the modal drinking pattern was binge importance of examining control groups from rary drinking pattern is better characterized drinking, which could have resulted in mem- the same population. Risk for FAS births in a as a dop legacy than as a systematic issuance ory loss. However, retrospective reports of al- single population may not provide accurate of alcohol to laborers, as stated in our intro- cohol consumption have been found to be as measures of generalized or absolute risk (e.g., duction. Contrary to popular misconception, accurate as77 or even more accurate than pre- thresholds). Our findings may be most rele- beer, not wine, is the beverage of choice and natal clinic data, or at least to produce reports vant for comparisons with populations of abuse. of higher drinking levels.78,79 In addition, re- other developing nations. cent literature supports the use of day-by-day Maternal Age, Nutrition, and reporting and reconstruction of drinking histo- Identified Risk of FAS Anthropomorphic Considerations ries,80,81 including details such as BACs.82 No All of the women studied belonged to a In a previous study in this community,9 the retrospective methods, however, are believed population of a modernizing society charac- mean age of mothers at birth of a child with to be as accurate as daily reporting systems to terized by generally low SES. However, com- FAS was low (26.7±7.6 years); this age was collect data on drinking outside prenatal set- pared with control mothers, case mothers had also relatively low in this sample (28.0 ±6.4 tings.80 Maternal drinking during pregnancy is even fewer social resources, such as educa- years). In both studies, the difference between a highly sensitive issue that affects validity. tion, income, or spirituality. As in other stud- case mothers and control mothers was not We believe that the questions, sequence, em- ies, risk for FAS was associated with higher significant (P =.07, 2-tailed) despite the larger pathic style, and follow-back methods used in gravidity and parity and thus later-born chil- sample in our study. Maternal age at birth of our study produced more accurate data (espe- dren. Case mothers were more likely to be FAS children was lower than that previously cially for heavy drinking) than have been col- unmarried and to live with a male partner observed in populations in developed coun- lected in prenatal settings.78,79 and had extended families, sexual partners, tries,7,25,33 in which a significant difference is In 2 waves of research in South Africa, and friends who drank heavily. As evidenced always reported between FAS case mothers, only 1 of 90 mothers of children with FAS from the control group drinking reported in case mothers, and control mothers.10 ,7 9 This contacted has refused an interview. The pro- interviews, frequent binge drinking was nor- lack of significance is unique in the literature portion of women with FAS children not in- mative among 50% of men and less than and is not explained by early age at onset of terviewed owing to death, migration, and re- 20% of women. Alcohol consumption was drinking or drug use in this community.

1196 | Research and Practice | Peer Reviewed | May et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

Rather, it is substantially explained by dura- provement and proven techniques of alcohol of the project, overseeing the South African medical, re- tion, degree, and regularity of binge drinking treatment and birth control. New treatment search, and programmatic operations. during pregnancy, with some other unique and prevention emphasizing outreach in the cofactors, such as nutrition, body size, and highest-risk populations of the Western Cape Acknowledgments This project was funded by the National Institute on general SES. Province can benefit from this research.13 ,19 Alcohol Abuse and Alcoholism, the National Institutes of Nutrition and maternal body size may par- Prevention has been undertaken by health of- Health National Center on Minority Health and Health tially explain the low maternal age at birth of ficials of Western Cape Province, but more re- Disparities, and the Foundation for Alcohol Related Research (grants R01AA09440 and R01AA11685). children with FAS, the high rates of FAS in sources for these efforts are needed. Despite Sincere thanks are due to Mayor Herman Baily, the this population, and the severity of FAS. Poor efforts to establish initiatives in FAS preven- Western Cape Community Town Council, Cecil Driver, nutrition (lifelong and current), genetic influ- tion,84 more awareness and activity are and the principals of the 12 elementary schools where the study was initiated. Furthermore, Chris Shaw, Car- ences, and multiple generations of fetal alco- needed. Impediments to FAS prevention in olyn Tullett, Maggie September, Dickie Naude, and oth- hol exposure likely contribute to the high rate South Africa are similar to those in the ers from the Foundation for Alcohol Related Research, of FAS. Case mothers were, on average, sig- United States: salaries for full-time workers University of Cape Town, and the community contributed greatly to the administration, professionalism, nificantly smaller on all physical measures: are lacking, and committed individuals cannot and vital energy of the project. Many individuals from height, weight, head circumference, and BMI. effectively transfer time and energy from the schools, the public health organizations, and the Maternal physical traits were negatively asso- other commitments to sustain FAS preven- local community, including farm owners and operators, were indispensable in the research process. ciated with their children’s dysmorphology tion activities.12 Integrating alcohol use The clinical dysmorphology team consisted of 85 scores. Smaller, lighter mothers who engage into prenatal screening (with HIV and Kenneth Lyons Jones, MD, Denis L. Viljoen, MD, in binge drinking may be less able to elimi- tuberculosis) could be a partial solution.86 Luther K. Robinson, MD, H. Eugene Hoyme, MD, Nathaniel Khaole, MD, Kwade “Kojo” Asante, MD, nate alcohol via first-pass metabolism, allow- In Western Cape Province, the rates of all Richard Findley, MD, and Barbara Quinton, MD. ing more alcohol to enter the placenta and 3of these problems are high, and each affects Psychological testing of children was overseen by cause more fetal damage.83 Conversely, heav- 4.5%–8% of the population.87,88 Because our Colleen Adnams, MD, and conducted by Andrea Hay and Ansie Kitching. Faye Calhoun, DPA, and Kenneth ier mothers have more adipose tissue to research indicates that FAS is increasing in R. Warren, PhD, of the National Institute on Alcohol 31 which alcohol can be distributed, thereby pro- the study community, prevention is needed Abuse and Alcoholism were responsible for initiating all tecting the fetus. Undernutrition and frequent there and elsewhere in Western Cape Prov- US collaboration on FAS in South Africa. hunger during pregnancy were reported by ince and South Africa.32 more case mothers than control mothers Human Participant Protection (11.5% vs 4.6%). Finally, the findings of Protocols and consent forms used were approved by About the Authors the University of New Mexico, the National Institutes of smaller average head circumference among Health Office of Protection From Research Risks, the Philip A. May and J. Phillip Gossage are with The Univer- case mothers and of heavier drinking among ethics committee of the University of Cape Town, and a sity of New Mexico Center on Alcoholism, Substance local, single-site assurance (oversight) committee. All maternal grandmothers of FAS children raise Abuse and Addictions, Albuquerque. Lesley E. Brooke, mothers participating in the study provided active con- Anna-Susan Marais, and Loretta S. Hendricks are with questions about intergenerational prenatal al- sent to participate, and each mother or legal guardian the University of Cape Town and the Foundation for Alco- cohol exposure and damage.10 Some mothers provided active consent for her child’s participation in hol Related Research, Cape Town Republic of South , FAS screening. of FAS children appear to have FASD them- Africa. Cudore L. Snell is with the Department of Social selves; their alcohol abuse may originate in Work, Howard University, Washington, DC. Julie A. Croxford is with the Department of Psychiatry and Behav- part from behavioral traits associated with References ioral Neurosciences, Wayne State University, Johannes- 1. Bingol N, Schuster C, Fuchs M, et al. The influ- FASD (e.g., impulsivity, poor judgment). bury, Republic of South Africa. Denis L. Viljoen is with ence of socioeconomic factors on the occurrence of the Department of Human Genetics, Faculty of Health Sci- fetal alcohol syndrome. Adv Alcohol Subst Abuse. 19 87; Protective Factors ences, University of Witwatersrand, National Health Lab- 6:105–118. oratory Service, and the Foundation for Alcohol Related 2. Day NL, Cottreau CM, Richardson GA. The epi- Potential protective mechanisms with pre- Research, Johannesbury. demiology of alcohol, marijuana, and cocaine use Requests for reprints should be sent to Philip A. May, ventive implications for this population have among women of childbearing age and pregnant PhD, The University of New Mexico, Center on Alco- been identified. Key protective factors were women. Clin Obstet Gynecol. 19 93;36:232–245. holism, Substance Abuse and Addictions, 2650 Yale Blvd low gravidity and parity, larger body size, SE, Albuquerque, NM 87106 (e-mail: [email protected]). 3. Day NL, Robles N, Richardson G, et al. The ef- higher educational attainment and income, This article was accepted May 9, 2004. fects of prenatal alcohol use on the growth of children at three years of age. Alcohol Clin Exp Res. 19 91;15: religiosity, nondrinking male partner, and ad- 67–71. equate nutrition. Contributors 4. Day NL, Zuo Y, Richardson GA, Goldschmidt L, P. A. May was the principal investigator and epidemiolo- Larkby CA, Cornelius MD. Prenatal alcohol use and Prevention gist of the study and led the writing. J.P. Gossage was offspring size at 10 years of age. Alcohol Clin Exp Res. the data manager and data analyst for the entire project. 1999;23:863–869. Public health education on the dangers of L.E. Brooke was the program coordinator of the field re- 5. Ernhart CB, Sokol RJ, Martier S, et al. Alcohol maternal binge drinking is needed locally. search in the target community and oversaw clinical as- teratogenicity in the human: a detailed assessment of sessment and interviews. C.L. Snell, L.S. Hendricks, and Prevention is needed in the community, par- specificity, critical period, and threshold. Am J Obstet A.S. Marais translated the questionnaire into Afrikaans Gynecol. 19 87;156:33–39. ticularly in rural areas. Many risk factors iden- and, with J. Croxford, completed all of the interviews of tified are amenable to change via social im- case and control mothers. D.L. Viljoen was co-investigator 6. Sokol RJ, Ager J, Martier S, et al. Significant deter-

July 2005, Vol 95, No. 7 | American Journal of Public Health May et al. | Peer Reviewed | Research and Practice | 1197  RESEARCH AND PRACTICE 

minants of susceptibility to alcohol teratogenicity. Ann 23. Abel EL. An update on incidence of FAS: FAS is 41.Croxford J, Viljoen D. Alcohol consumption by N Y Acad Sci. 19 86;477:87–102. not an equal opportunity birth defect. Neurotoxicol pregnant women in the Western Cape. S Afr Med J. Teratol. 19 95;17:437–334. 1999;89:962–965. 7. May A, Hymbaugh KJ, Aase JM, Samet JM. Epide- miology of fetal alcohol syndrome among American In- 24. Abel EL, Hannigan JH. Maternal risk factors in 42. Kendler KS, Heath AC, Neale MC, et al. A dians of the Southwest. Soc Biol. 19 83;30:374–387. fetal alcohol syndrome: provocative and permissive in- population-based twin study of alcoholism in women. fluences. Neurotoxicol Teratol. 19 95;17:445–465. JAMA. 19 92;268:1877–1882. 8. May PA, Brooke L, Gossage JP, et al. Epidemiol- ogy of fetal alcohol syndrome in a South African com- 25. Jacobson JL, Jacobson SW, Sokol RJ. Increased 43. McCaul ME, Turkkan JS, Svikis DS, Bigelow GE. munity in the Western Cape Province. Am J Public vulnerability to alcohol-related birth defects in the off- Familial density of alcoholism: effects on psychophysio- Health. 2000;90:1905–1912. spring of mothers over 30. Alcohol Clin Exp Res. 1996; logical responses to ethanol. Alcohol. 19 91;8:219–222. 20:356–363. 9. Viljoen D, Croxford J, Gossage JP, Kodituwakku PW, 44. Schuckit MA, Smith TL. An 8-year follow-up of May PA. Characteristics of mothers of children with 26.Kodituwakku PW, Kalberg W, May PA. The ef- 450 sons of alcoholic and control subjects. Arch Gen fetal alcohol syndrome in the Western Cape Province fects of prenatal alcohol exposure on executive func- Psychiatry. 1996;53:202–210. of South Africa: a case–control study. J Stud Alcohol. tioning. Alcohol Res Health. 2001;25:192–198. 45. Pierog S, Chandavsu O, Wexler I. The fetal alco- 2002;63:6–17. 27. Mattson SN, Goodman AM, Caine C, Delis DC, hol syndrome: some maternal characteristics. Int J 10. Kvigne VL, Leonardson GR, Borzelleca J, Brock E, Riley EP. Executive functioning in children with heavy Gynecol Obstet. 1979;16:412–415. Neff-Smith M, Welty TK. Characteristics of mothers prenatal alcohol exposure. Alcohol Clin Exp Res. 1999; 46. Davis A, Lipson A. A challenge in managing a who have children with fetal alcohol syndrome or 23:1808–1815. family with fetal alcohol syndrome. Clin Pediatr. 1984; some characteristics of fetal alcohol syndrome. J Am 28. Mattson SN, Riley EP. Implicit and explicit mem- 23:304. Board Fam Pract. 2003;16:296–303. ory functioning in children with heavy prenatal alcohol 47. Gomberg ESL. Women and alcohol: use and 11. May PA, Hymbaugh KJ. A pilot project on fetal exposure. J Int Neuropsychol Soc. 1999;5:462–471. abuse. Nerv Ment Dis. 19 93;181:211–219. alcohol syndrome among American Indians. Alcohol 29. May PA, Gossage JP. Estimating the prevalence of 48.Wilsnack SC. Women at high risk for alcohol Health Res World. 19 82;7:3–9. fetal alcohol syndrome: a summary. Alcohol Res Health. abuse. The Counselor. 1989;7(1):16–17, 29. 12. May PA, Hymbaugh KJ. A macro-level fetal alco- 2001;25:159–167. 49.Wilsnack SC, Klassen AD, Schur BE, Wilsnack RW. hol syndrome prevention program for Native Ameri- 30. Sampson PD, Streissguth AP, Bookstein FL, Little RE, Predicting onset and chronicity of women’s problem cans and Alaska Natives: description and evaluation. Clarren SK, Dehane P. Incidence of fetal alcohol syn- drinking: a five-year longitudinal analysis. Am J Public J Stud Alcohol. 1989;50:508–518. drome and prevalence of alcohol-related neurodevelop- Health. 19 91;81:305–318. 13. May PA. A multiple-level, comprehensive ap- mental disorder. Teratology. 19 97;56:317–326. 50. Wilsnack SC. Sexuality and women’s drinking. proach to the prevention of fetal alcohol syndrome 31. May P, Gossage J, Brooke L, Croxford J, Viljoen D. Alcohol Health Res World. 19 91;15:147–150. (FAS) and other alcohol-related birth defects (ARBD). An epidemiological analysis of a second wave of data Int J Addict. 19 95;30:1549–1602. from children with fetal alcohol syndrome and controls 51.Wilsnack SC, Beckman LJ. Alcohol Problems in Women: Antecedents, Consequences, and Intervention. 14 . May PA. Research issues in the prevention of fetal in the Western Cape, South Africa [abstract 208]. New York: Guilford Press; 1984. alcohol syndrome and alcohol-related birth defects. In: Alcohol Clin Exp Res. 2000;24:41A. Howard J, Martin S, Mail P, Hilton M, Taylor E, eds. 32. Viljoen DL, Craig P, Hymbaugh K, Boyle C, 52. Shore ER, Batt S. Contextual factors related to the Women and Alcohol: Issues for Prevention Research. Be- Blount S. FAS in South Africa, 2001. MMWR Morb drinking behaviors of American business and profes- thesda, Md: National Institutes of Health; 1996: Mortal Wkly Rep. 2003;52:660–662. sional women. Br J Addict. 19 91;8 6:171–176. 93–131. 33. Abel EL. Fetal Alcohol Abuse Syndrome. New York, 53. Shore ER, Pieri SA. Drinking behaviors of women 15. Sokol RKJ, Miller SI, Reed G. Alcohol abuse dur- NY: Plenum Press; 1998. in four occupational groups. Women Health. 19 92;19: ing pregnancy: an epidemiologic study. Alcohol Clin 55–64. 34. Bingol N, Schuster C, Fuchs J, et al. Adv Alcohol Exp Res. 1980;4:135–145. Subst Abuse. 19 87;6:105–118. 54. Schlesinger S, Susman M, Koenigsberg J. Self-esteem 16. Serdula M, Williamson DF, Kendrick JS, et al. and purpose of life: a comparative study of women al- 35. Morse BA, Weiner L. Rehabilitation approaches for Trends in alcohol consumption by pregnant women: coholics. J Alcohol Drug Educ. 1990;36:127–141. fetal alcohol syndrome. In: Spohr HL, Steinhausen HC, 19 85 through 1988. JAMA. 19 91;265:876–879. eds. Alcohol, Pregnancy, and the Developing Child. 55. Baily S. Women with alcohol problems: a psycho- 17.Waterson EJ, Murray-Lyon IM. Drinking and New York, NY: Cambridge University Press; 1996: social perspective. Drug Alcohol Rev. 1990;9:125–131. smoking patterns among women attending an antenatal 249–268. 56. Blume SB. Chemical dependence in women: im- clinic, II: during pregnancy. Alcohol Alcohol. 1989;24: 36. Day NL, Cottreau CM, Richardson GA. The epi- portant issues. Am J Drug Alcohol Abuse. 1990;16: 163–173. demiology of alcohol, marijuana, and cocaine use 297–307. 18. Godel JC, Pabst HF, Hodges PE, et al. Smoking among women of childbearing age and pregnant 57.Viljoen DL, Carr LG, Foroud TM, Brooke L, and caffeine and alcohol intake during pregnancy in a women. Clin Obstet Gynecol. 19 93;36:232–245. Ramsay M, Li TK. Alcohol dehydrogenase-2*2 allele is northern population: effect on fetal growth. Can Med 37. Centers for Disease Control and Prevention. Alco- associated with decreased prevalence of fetal alcohol Assoc J. 19 92;147:181–187. hol consumption among pregnant and child-bearing syndrome in the mixed-ancestry population of the 19. Stratton K, Howe C, Battaglia F, eds. Fetal Alcohol aged women: United States, 1991 and 1995. MMWR Western Cape Province, South Africa. Alcohol Clin Exp Syndrome Diagnosis, Epidemiology, Prevention, and Morb Mortal Wkly Rep. 19 97;46:346–350. Res. 2001;25:1719–1722. Treatment. Washington, DC: National Academy Press; 38. Bonati M, Fellin G. Changes in smoking and 58. May PA, Gossage JP, White-Country M, et al. Al- 1996. drinking behavior before and during pregnancy in Ital- cohol consumption and other maternal risk factors for 20. Sokol RJ, Ager J, Matier S, et al. Significant deter- ian mothers: implications for public health intervention. fetal alcohol syndrome among three distinct samples of minants of susceptibility to alcohol teratogenicity. Ann Int J Epidemiol. 19 91;20:927–932. women before, during, and after pregnancy: the risk is N Y Acad Sci. 19 8 6;4 47:87–102. relative. Am J Med Gen. 2004;127C:10–20. 39.Primatesta P, DelCorno G, Bonazzi MC, Waters WE. 21. Abel EL, Sokol RJ. Maternal and fetal characteris- Alcohol and pregnancy: an international comparison. 59. Hanna EZ, Faden VB, Harford TC. Marriage: tics affecting alcohol’s teratogenicity. Neurobehav Toxicol J Public Health Med. 19 93;15:69–76. does it protect young women from alcoholism? J Subst Teratol. 1986;8:329–334. Abuse. 19 93;5:1–14. 40.Croxford JA, Viljoen D. Prospective analysis of 22. Darrow SL, Russell M, Cooper ML, et al. Sociode- alcohol ingestion in 636 pregnant women in rural and 60. Schimidt C, Klee L, Ames G. Review and analysis mographic correlates of alcohol consumption among urban areas of the Western Cape. Cape Town, South of literature on indicators of women’s drinking prob- African-American and white women. Women Health. Africa: University of Cape Town, Department of lems. Br J Addict. 19 9 0;85:179–192. 19 92;18:35–50. Human Genetics; 1998. 61. London L, Meyer J, Well V, Taylor T, Thompson MR,

1198 | Research and Practice | Peer Reviewed | May et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

Mibuli S. An Investigation into the Neurological and Neu- year reliability of self-reported alcohol consumption. robehavioral Effects of Long-Term Agrochemical Exposure J Stud Alcohol. 1990;51:68–76. and Deciduous Fruit Farm Workers in the Western Cape, 79. Jacobson JL, Jacobson SW, Sokol RJ. Increased South Africa [masters thesis]. Cape Town, South Africa: vulnerability to alcohol-related birth defects in the off- University of Cape Town; 1995. spring of mothers over 30. Alcohol Clin Exp Res. 1996; 62. London L. Alcohol consumption amongst South 20:359–363. African farm workers: a challenge for post-apartheid 80. Searles JS, Helzer JE, Rose GL, et al. Concurrent health sector transformation. Drug Alcohol Depend. and retrospective reports of alcohol consumption 2000;59:199–206. across 30, 90, and 366 days: interactive voice re- 63. Crome IB, Glass Y. The Dop system: a manifesta- sponse compared with the timeline follow back. J Stud tion of social exclusion. A personal commentary on Alcohol. 2002;63:352–362. “alcohol consumption amongst South African workers: a post-apartheid challenge, by L. London 1999.” Drug 81. Gruenewald PJ, Russell M, Light J, et al. One Alcohol Depend. 2000;59:207–208. drink to a lifetime of drinking: temporal structures of drinking patterns. Alcohol Clin Exp Res. 2002;26: 64.Parry CDH, Bennetts AL. Alcohol Policy and Public 916–925. Health in South Africa. New York, NY: Oxford Univer- sity Press; 1998. 82. Carey KB, Hustad JTP. Are retrospectively recon- structed blood alcohol concentrations accurate? Prelim- 65. Pascoe JM, Kokotailo PK, Broekhuizen FF. Corre- inary results from a field study. J Stud Alcohol. 2002; lates of multigravida women’s binge drinking during 63:762–766. pregnancy: a longitudinal study. Arch Pediatr Adolesc Med. 19 95;149:1325–1329. 83. Khaole NK, Li TK. Protective alcohol dehydrogenase genotypes for FAS and blood alcohol profiles 66. Maier SE, West JR. Drinking patterns and alcohol- among mothers of FAS children. Paper presented at: related birth defects. Alcohol Res Health. 2001;25: the Annual Meeting of the Research Society on Alco- 16 8–174. holism, June 24–29, 2000, Denver, Colo. 67.Pierce DR, West JR. Blood alcohol concentration: 84. Fetal Alcohol Syndrome: South Africa. A Progress a critical factor for producing fetal alcohol syndrome. Report on the 1997 Pilot Study, Information Exchange, Alcohol. 1986;3:269–272. and Prevention Workshops. Rockville, Md: National In- 68. Goodlett CR, Horn K. Mechanisms of alcohol- stitute on Alcohol Abuse and Alcoholism; 1998. induced damage to the developing nervous system. 85. Bad Heart Bull LB, Kvigne VL, Leonardson GL, Alcohol Res Health. 2001;25:175–184. Lacuia L, Welty TK. Validation of a self-administered 69. Mattson SN, Schoenfeld AM, Riley EP. Terato- questionnaire for prenatal alcohol use in Northern genic effects of alcohol on brain and behavior. Alcohol Plains Indian women. Am J Prevent Med. 1999;16: Res Health. 2001;25:185–191. 240–243. 70.Parry CDH. Alcohol problems in developing coun- 86. Li C, Olsen Y, Kvigne V, Welty T. Implementation tries: challenges for the new millennium. Suchtmedizin of substance use screening in prenatal clinics. S D J in Forschung und Praxis. 2000;2:216–220. Med. 1999;52:59–64. 71. Meyers B, Parry CH. Alcohol use in South Africa, 87.Groenewald P. Annual Report 2000: Boland Over- 2001: Fact Sheet 6. Tygerberg, South Africa: Medical berg Region. Worcester, South Africa: Dept of Informa- Research Council of South Africa. tion Management; 2002. 72. Adnams CM, Kodituwakku P, Hay A, Molteno CD, 88.Republic of South Africa, Western Cape Province. Viljoen D, May PA. Patterns of cognitive-motor devel- Health Status Report. Cape Town, South Africa: Western opment in children with fetal alcohol syndrome from Cape Department of Health; 2001. a community in South Africa. Alcohol Clin Exp Res. 2001;25:557–562. 73. King AC. Enhancing the self-report of alcohol consumption in the community: two questionnaire formats. Am J Public Health. 19 9 4;8 4:294–296. 74. Sobell LC, Sobell MB, Leo GI, et al. Reliability of a timeline method: assessing normal drinkers’ reports of recent drinking and a comparative evaluation across several populations. Br J Addict. 1988;83:393–402. 75. Sobell LC, Agrwal S, Annis H, et al. Cross-cultural evaluation of two drinking assessment instruments: alcohol timeline followback and inventory of drinking situations. Subst Use Misuse. 2001;36:313–331. 76. Molarius A, Seidell JC, Sans S, Tuomilehto J, Kuulasmaa K. Educational level, relative body weight, and changes in their association over 10 years: an international perspective from the WHO MONICA Proj- ect. Am J Public Health. 2000;90:1260–1268. 77. Robles N, Day NL. Recall of alcohol consumption during pregnancy. J Stud Alcohol. 1990;51:403–407. 78. Czarnecki DM, Russell M, Cooper ML, et al. Five-

July 2005, Vol 95, No. 7 | American Journal of Public Health May et al. | Peer Reviewed | Research and Practice | 1199  RESEARCH AND PRACTICE 

Cancer Prevention Among Working Class, Multiethnic Adults: Results of the Healthy Directions–Health Centers Study

| Karen M. Emmons, PhD, Ann M. Stoddard, ScD, Robert Fletcher, MD, MSc, Caitlin Gutheil, MS, Elizabeth Gonzalez Suarez, MA, Rebecca Lobb, MPH, Jane Weeks, MD, MSc, and Judy Anne Bigby, MD

Disparities in cancer morbidity and mortality Objectives. We analyzed outcomes from a study that examined social-contextual rates by race/ethnicity and socioeconomic po- factors in cancer prevention interventions for working class, multiethnic populations. sition have been well documented and are a Methods. Ten community health centers were randomized to intervention or 1,2 key research priority. Across health behav- to control. Patients who resided in low-income, multiethnic neighborhoods were iors, patterns of risk by socioeconomic posi- eligible; the intervention targeted fruit and vegetable consumption, red meat tion remain constant: persons of higher socio- consumption, multivitamin intake, and physical activity. Outcomes were mea- economic position engage in fewer high-risk sured at 8 months. behaviors than do persons of lower socioeco- Results. The intervention led to significant increases in fruit and vegetable con- nomic position.3–5 Similar patterns are found sumption and multivitamin intake and reductions in red meat consumption; no by race/ethnicity.3,6–8 change was found in physical activity levels. The intervention effect was not changed when contextual variables that may function as confounders or effect The health care system is an important modifiers (e.g., gender, education, race/ethnicity, respondent and parents’ coun- channel for reducing behavioral risk factors try of birth, and poverty status) were included in the analyses. among diverse populations.9–11 Brief physician Conclusions. The intervention led to significant improvements in health counseling has been found to be effective with behaviors among a working class, multiethnic population, regardless of race/ 12–15 diet and tobacco use, although evidence ethnicity and socioeconomic status. Interventions that respond to the social con- is currently inconclusive regarding the effect text of working class individuals across racial/ethnic categories hold promise for of provider counseling on physical activity.16 ,17 improving cancer-related risk behaviors. (Am J Public Health. 2005;95:1200–1205. The rates of physicians providing behavior doi:10.2105/AJPH.2004.038695) change counseling are quite low.18 ,19 Adjuncts to brief provider counseling are effective,17 used a common conceptual model and inter- tice style, and other systems factors that might 21 and provider interventions may be more effec- vention paradigm and focused on common have affected outcomes. Patients who resided tive if patients are referred to other programs primary outcomes: fruit and vegetable con- in low-income, multiethnic neighborhoods were that provide ongoing social support.10 , 2 0 sumption, red meat consumption, multivitamin identified and approached for participation. We present the outcome results of Healthy intake, and physical activity. The design of the Directions–Health Centers, an intervention interventions was largely based on social epi- Setting study designed to reduce cancer risk factors demiological findings related to disparities in This study was conducted in collaboration among working class, multiethnic populations health behaviors and health outcomes. Healthy with Harvard Vanguard Medical Associates, a seen in community health centers. This study Directions was designed to take into account health care delivery system composed of 14 was part of the Harvard Cancer Prevention elements of the social context that are critical multispecialty medical group practices that Program Project, the theme of which was to components of an ecological approach to serve more than 270000 patients. The 10 create cancer prevention interventions that are health behavior change. We present the out- health centers that were invited to participate effective with working class, multiethnic popu- comes of the 8-month follow-up of the Healthy took part in the study. lations. This program project was designed to Directions–Health Centers project. develop and test behavioral interventions for Sample multiple cancer risk factors in working class METHODS Patient eligibility criteria included (1) living and ethnically diverse groups through 2 inter- in an eligible neighborhood (discussed later in vention channels: health centers and worksites. Study Design this paragraph), (2) being 18 to 75 years of The 2 intervention projects (Healthy Healthy Directions–Health Centers was a age, (3) having a well-care or follow-up visit Directions–Health Centers and Healthy randomized controlled trial in which the health scheduled with a participating provider, Directions–Small Businesses) were randomized center was the unit of randomization and inter- (4) being able to speak and read either controlled trials that used the organization vention. Ten health centers were paired on English or Spanish, (5) not having cancer at (e.g., health center or small business) as the membership size and randomized within pairs the time of enrollment, (6) not being em- unit of randomization and intervention and the to the intervention condition or usual care; the ployed by the participating health centers, individual as the unit of analysis. Both projects health centers shared practice guidelines, prac- and (7) not being employed at a worksite

1200 | Research and Practice | Peer Reviewed | Emmons et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

participating in the companion study. Eligible Intervention Components session was designed to be approximately 20 neighborhoods were defined according to The intervention is described in detail else- minutes in length, and the follow-up telephone census definitions to be census block groups where25 and is summarized here. The inter- calls were approximately 10 minutes in length. that were predominantly working class (66% vention used a social-contextual approach tar- The intervention was developed with substan- or more of the employed persons are in work- geting multiple levels of influence on behaviors, tial input from a community advisory board, ing class occupational groups) or impover- with special attention to low literacy skills and which ensured that the intervention was de- ished (20% or more of the population lives the shared and unique features of culture signed and implemented in accordance with below the poverty level) or with low levels of across racial/ethnic groups.21,25 We drew community priorities and expectations. education (25% or more of the adult popula- heavily on the social epidemiology literature tion has not completed high school). to broaden the intervention beyond the stan- Measures All providers practicing in the internal dard behavioral and psychological theoretical Health behaviors. The National Cancer Insti- medicine departments of these centers were literature and to consider societal factors that tute’s 5 A Day for Better Health Program approached for permission to recruit from may influence health behavior in the inter- screener was used to measure servings of fruit among their patient pools. Provider partici- vention design. and vegetables consumed per day.28 Responses pation averaged 83% across sites (range = The intervention activities and materials were recoded to equivalent servings per day 50%–100%); a total of 97 clinicians partici- included strategies, images, messages, and and summed to obtain total fruit and vegeta- pated, with no differences in clinician partici- vocabulary that were inclusive and nonstereo- ble servings per day. We then computed a di- pation between the 2 study conditions.22 All typing for the multicultural audiences and chotomous measure: 5 or more servings per participating providers received brief train- that used tactics to reach participants with day or fewer than 5 servings per day. ing (1 hour) during a regular staff meeting limited literacy skills (e.g., plain language, sim- Red meat consumption was assessed with about the study design, intervention mes- ple graphics, stories). Project messages explic- an abbreviated form of the semiquantitative sages (intervention sites only), and an imple- itly acknowledged that health behavior is in- Food Frequency Questionnaire.28 The re- mentation plan. fluenced by context. sponses were recoded to equivalent servings Patients who were scheduled for appoint- Only those participants who were patients per week and summed to obtain total serv- ments with the participating providers and in the health centers randomized to the inter- ings of red meat per week. The totals were were in the eligible age range were identified vention condition and who consented to be in dichotomized to 3 or fewer servings or more through the automated central appointment the study received the intervention, which than 3 servings per week. system. Geocoding was used to determine consisted of (1) study endorsement from the Physical activity was assessed based on the whether a potential participant lived in an participant’s clinician at a scheduled routine- questionnaire used in the Nurses’ Health eligible neighborhood, assessed at the census care visit, including provision of a tailored pre- Study,29 adapted to include specific activities block group level.23,24 Eligible patients resid- scription for the recommended health behav- that were common in our target population. ing in the target census block groups with an ior changes; (2) an initial in-person counseling The questionnaire asked how often in the last appointment no sooner than 2 weeks after session with a health adviser; (3) 4 follow-up 4 weeks respondents engaged in each activity, identification received a letter in the mail de- telephone counseling sessions with the health on average. Responses were recoded to equiv- scribing the study and giving them the oppor- adviser; (4) 6 sets of tailored materials written alent minutes per week and summed to obtain tunity to request no further contact. Patients for low-literacy audiences that targeted social total minutes of physical activity per week. were then contacted by telephone where eli- contextual factors (e.g., family composition, Walking was included in the total only if the gibility was confirmed, an invitation to partici- available social support and networks, occupa- walking pace was reported to be “normal/ pate was extended, and the baseline survey tional status, neighborhood safety concerns); average” or faster. The total was collapsed to was completed. and (5) links to relevant local activities. Health 150 minutes (2.5 hours) per week or more Study staff attempted to recruit 8963 po- advisers were college-educated individuals versus less than 150 minutes per week. tentially eligible candidates. Of these at- with substantial community experience, had We asked respondents how many days, on tempts, 2547 (28%) were unreachable, 867 diverse racial/ethnic and socioeconomic back- average, they took a multivitamin. Responses (10%) were ineligible, 3330 (37%) refused, grounds, and were bilingual in Spanish. were coded as daily if subjects reported tak- and 2219 (25%; 40% of those reached and The counseling used motivational interviewing a multivitamin 6 or 7 days per week. eligible) were enrolled. The cohort recruited ing as a strategy to contextualize the interven- Sociodemographic characteristics. Respon- at baseline was contacted by telephone after tion and to enhance understanding of the fac- dents were asked their date of birth, gender, the intervention period to complete a follow- tors that influenced a patient’s motivation and and highest level of education completed. up survey. Of the 2219 who completed the ability to change.26,27 This approach was par- They were asked to identify all the racial baseline survey (n=1088 intervention condi- ticularly helpful when addressing concerns rel- and ethnic groups to which they belonged. tion; n=1131 control condition), 1954 (88%) evant to this multiethnic population while min- We coded participants who reported being completed the follow-up survey. Follow-up re- imizing assumptions about factors related to of Hispanic or Latino origin in the Hispanic sponse rate was equivalent across conditions. participants’ health behaviors. The in-person group regardless of any other ethnic groups

July 2005, Vol 95, No. 7 | American Journal of Public Health Emmons et al. | Peer Reviewed | Research and Practice | 1201  RESEARCH AND PRACTICE 

mentioned. Those who reported only 1 eth- ness of the intervention, we analyzed each GLIMMIX macro to the SAS statistical soft- nic group were categorized in that group; re- outcome measure separately and in the binary ware (SAS Institute Inc, Cary, NC). This spondents who selected more than 1 group scale. We computed a repeated-measures, macro uses iteratively reweighted likelihoods (not including Hispanic) were classified as mixed-model logistic regression analysis with to fit a logistic regression model in which the mixed heritage. the intervention group and survey (baseline subjects are clustered in the random effect.33 Household income was assessed in 5 cate- or follow-up) as fixed effects and the health To explore analyses of the target subgroups gories (<$10000 per year to $50000 or center as a random effect.31 The participant and to control for factors that may have been more). We combined the responses to this was included as a repeated random effect unbalanced despite randomization, we added item with the number of people supported by within the randomization unit in the interven- covariates to the logistic regression analysis. the income and compared it with the federal tion condition. This method incorporates Per-subject incremental costs were calculated poverty guidelines.30 On the basis of this infor- cases with missing data, as long as the data by summing the costs for each group and di- mation, subjects were classified as (1) below are missing at random.32 The test of hypothe- viding by the number of subjects randomized the poverty guideline, (2) above the poverty ses of no difference in change in behavior to that group. guideline but below 185% of the guideline, or between intervention and control conditions (3) above 185% of the poverty guideline. was examined by the interaction effect of in- RESULTS We combined information about the par- tervention group by survey. We used the ticipants’ and their parents’ birthplaces into coefficients from the linear logistic regression Demographics a 3-category measure: (1) participant born model to compute the adjusted percentages. Table 1 presents the demographic character- outside the United States; (2) participant To carry out the analyses, we used the istics of the sample by intervention condition. born in the United States, but 1 or more parents born outside the United States; or TABLE 1—Frequency of Selected Characteristics at Baseline, by Randomization Group, (3) participant and both parents born in the With P Value for Test of Effectiveness of Randomization (N=2219) United States. We combined information about the respondent’s employment status Control Intervention and job title into a 3-category job-status Frequency Adjusted,a % Frequency Adjusted,a % P b variable: (1) employed in a blue-collar job, (2) employed in a job that is not blue collar, Gender .08 or (3) unemployed or retired. Male 328 29.1 419 39.5 Costs. A process tracking system was de- Female 800 70.9 669 61.5 veloped to collect detailed data on all costs Education .61 ≥ associated with the intervention delivery. BA degree 448 40.6 394 38.2 Staff and health adviser time were valued 185 902 81.6 912 85.3 number of subjects; the proportion of the <185 207 18.4 157 14.7 survey required for generating the tailored Birth country .95 intervention reports was derived by pro- Both parents and subjects born in United States 687 60.4 649 60.0 rating the costs by the percentage of the Other 442 39.9 433 40.0 survey devoted to the collection of infor- Occupational class e mation for tailoring (42% of survey ques- Unemployed 124 11.0 122 11.2 .86 f tions). This is likely an overestimate of Working class 522 46.1 463 42.5 .14 costs because some of this information Professional/manager 485 42.9 503 46.3 would be required for intervention evalua- Age, adjusted mean 47.8 50.8 .18 tion irrespective of tailoring. aPercentages adjusted for clustering of respondents in health centers. bP value for test of equality of group percentages, controlling for clustering of respondents in health centers. c Data Analysis P value for test of equality of percent White, non-Hispanic vs percent Black, non-Hispanic and “other.” dP value for test of equality of percent Black, non-Hispanic vs percent “other.” In all analyses, we controlled for the clus- eP value for test of equality of percent unemployed vs percentage employed. tering of respondents in the randomization fP value for test of equality of percent working class vs percentage professional/manager. unit—health centers. To assess the effective-

1202 | Research and Practice | Peer Reviewed | Emmons et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

TABLE 2—Adjusted Percentages of Participants With Each Health Behavior at Baseline and meat intake, and physical activity—represented Follow-Up, by Intervention Group (N=1954) as continuous variables (Table 3)—receiving similar results to those found with the dichoto- Control Intervention mous measure. Overall, a 0.3 serving-per-day a Survey n % n % P average increase in fruit and vegetable con- Servings of fruit and vegetables, ≥5/d sumption occurred in the intervention group, Baseline 982 14.8 953 13.9 and a 0.6 servings-per-week average decrease Follow-up 979 11.0 959 17.2 of red meat consumption was reported. Change –3.8 +3.3 .005 Servings of red meat, ≤3/wk Covariate Analyses Baseline 976 53.2 962 48.8 We investigated each of the following co- Follow-up 978 53.0 960 60.6 variates as potential confounders or effect Change –0.2 +11.8 <.001 modifiers: gender, education, race/ethnicity, Physical activity, ≥2.5 h/wk respondent’s and parents’ birth country, and Baseline 916 64.8 900 66.6 poverty status. No confounding effect of any Follow-up 939 63.2 919 66.4 variable was found. Controlling for these vari- Change –1.6 –0.2 .51 ables did not change the intervention effect Multivitamins, ≥6 d/wk for any of the behavioral outcomes. No signif- Baseline 985 36.6 965 39.2 icant effect modification occurred for any Follow-up 984 44.3 959 68.6 variable considered: none of the 3-way inter- Change +7.7 +29.4 <.001 action effects was statistically significant.

a P value is for the test of the intervention×survey interaction (control vs intervention). Intention-to-Treat Analysis To evaluate the effect of attrition on outcomes, we computed an intention-to-treat No significant differences in the demographic meat consumption to 3 or fewer servings per analysis by assuming that those study partici- characteristics evaluated were seen between week, compared with no change in the con- pants who did not complete the follow-up participants in the intervention and partici- trol participants, and 29% of the intervention survey did not change any of the outcome be- pants in the control health centers (Table 1). participants began daily multivitamin intake, haviors from baseline. When we included the compared with 8% of the control participants. nonrespondents in the analysis, the results Treatment Outcomes We also evaluated the effect of the inter- were almost identical to those computed for We evaluated the effect of the intervention vention on fruit and vegetable intake, red the respondents with complete data only. on health behavior by examining whether participants met the criterion for having the risk TABLE 3—Adjusted Means for Health Behaviors at Baseline and Follow-Up, by Intervention factor (e.g., if they ate fewer than 5 servings of Groupa (N=1954) fruits and vegetables per day, they were considered at risk for that variable). Table 2 pre- Control Intervention sents the percentages of participants who re- Survey Mean SE Mean SE ported the target values of the outcome Servings of fruit and vegetables per day variables according to the risk factor criteria. Baseline 3.19 0.062 3.28 0.062 Significantly greater change was found Follow-up 3.13 0.064 3.57 0.064 among participants in intervention health Change –0.04 +0.29 centers in fruit and vegetable consumption Servings of red meat per week (P=.005), red meat consumption (P<.001), Baseline 3.89 0.16 3.75 0.17 and multivitamin intake (P<.001). No signifi- Follow-up 3.97 0.17 3.14 0.17 cant differences were found in physical activ- Change +0.08 –0.61 ity. These data indicate that 3.3% of the in- Hours of physical activity per week tervention participants increased fruit and Baseline 4.93 0.16 4.80 0.16 vegetable consumption to at least 5 servings Follow-up 4.91 0.16 4.77 0.17 per day, whereas 3.8% of the control partici- Change –0.02 –0.03 pants decreased consumption to below this a level over the study period. Twelve percent Adjusted for clustering of patients in health centers. of the intervention participants reduced red

July 2005, Vol 95, No. 7 | American Journal of Public Health Emmons et al. | Peer Reviewed | Research and Practice | 1203  RESEARCH AND PRACTICE 

Costs At baseline, the sample had a relatively low vention are greater to the extent that the The total intervention delivery cost per pa- level of fruit and vegetable consumption. In- same behaviors underlie these other diseases. tient in the intervention arm was $168. Staff tervention participants increased their fruit Several limitations should be noted. Re- time related to intervention delivery was 67% and vegetable consumption by almost one sponse rates ideally would have been higher; of these costs. Nonresearch costs were not in- third of a serving, whereas control participants only 40% of the eligible patients participated, curred for patients in the control arm; there- reported a decrease in consumption. This although participation did require a substan- fore, the incremental cost of the intervention level of change is similar to that found in tial effort on the patient’s part. When partici- was$16 8 per patient. worksite interventions,38,39 although it is pation in the intervention study was consid- somewhat less than the effects found with ered as an eligibility criterion, the response DISCUSSION health care interventions in health mainte- rate was considerably higher (62%). The study nance organization populations,20,40,41 likely sample consisted predominantly of women and We have shown that a tailored intervention because of sample differences. The current older participants (mean age=49). The results that incorporated aspects of the social context study also yielded a significant reduction in are only generalizable to similar populations. led to significant improvements in behavioral red meat consumption compared with a slight The intervention was implemented by health risk factors for cancer among low-income pa- increase in red meat intake in the control educators and would be feasible only in rela- tients receiving care in a health care system group. Few interventions have targeted red tively large practices with substantial integra- that serves multiethnic populations. Effects meat consumption, so it is not possible to com- tion and structure, not in small groups with were largest for multivitamin use and sub- pare this finding with those in other studies. less extensive resources. However, these stantial for both red meat and fruit and vege- Physical activity did not change as a result kinds of practices constitute much of US table consumption, whereas physical activity of the intervention. The physical activity lev- health care and are increasing in prevalence. did not change. Effectiveness was comparable els of the patients in this study were higher We recognize that this study included only across subgroups defined by ethnicity and in- than expected by other estimates.42 We con- short-term outcomes and that we did not pro- come, which is highly significant given the ducted a validation study in which we com- vide evidence of long-term success. Future substantial evidence in the literature that pared self-reported minutes of physical activ- analyses also should address cost-effectiveness. health behavior change interventions are ity with those measured on an actigraph However, the available data suggest that the often less effective for underserved popula- activity monitor. The estimates of total activ- Healthy Directions–Health Centers interventions. The intervention was embedded in the ity (moderate plus vigorous) were very similar tion is feasible in many health care settings health care system, but we used clinicians between the 2 methods. Within the context of and, if as effective in those settings, has the only to endorse the behavior changes being a randomized controlled trial, the fact that the potential to produce changes in health-related recommended; we then added several other baseline physical activity levels were rela- behaviors that are large enough to be of pub- intervention strategies to maximize the effect tively high in both conditions did not influ- lic health and clinical importance. Because of the provider interaction. The intervention ence our ability to examine the between- the Healthy Directions–Health Centers inter- was well received by patients and providers group differences in outcomes. However, vention was equally effective across individu- and cost only $168 per patient. future work will need to evaluate physical ac- als regardless of race/ethnicity and socioeco- The intervention produced the largest ef- tivity in similar populations in urban settings nomic status, this study provides a step fect on multivitamin use, with almost 70% of to learn more about specific sources of physi- toward addressing the increasing disparities the intervention participants taking a daily cal activity in lower-income groups and how in cancer risk. multivitamin at follow-up. The perceived rela- to best capture these activities from an assess- tive “ease” of pill consumption should not un- ment perspective. dermine consideration of the importance of This study had several strengths. Health About the Authors this outcome, because multivitamin use is centers were the unit of randomization and Karen M. Emmons is with the Dana-Farber Cancer Institute strongly related to disease outcomes. Long- analysis. The intervention targeted several and the Harvard School of Public Health, Boston, Mass. At the time of the study, Anne M. Stoddard was with the Uni- term use of folate-containing multivitamins risk factors concurrently, increasing its poten- versity of Massachusetts, Amherst. Robert Fletcher is with has been associated with a 75% reduction in tial effects on cancer incidence and efficiency. Harvard Medical School, Boston, Mass. Caitlin Gutheil and colorectal cancer mortality.34 The change in The sample size was large enough to detect Elizabeth Gonzalez Suarez are with the Dana-Farber Can- cer Institute. Rebecca Lobb is with Harvard Medical School, multivitamin use seen in our study, if sus- clinically important changes in health-related Division of Ambulatory Care and Prevention. Jane Weeks is tained, thus would be associated with as behaviors and to examine differences among with the Dana-Farber Cancer Institute and Harvard Medical much as a 22% reduction in colorectal can- subgroups defined by income and race/ School. Judy Ann Bigby is with the Brigham and Women’s Hospital and Harvard Medical School. cer incidence. Folic acid supplements also ethnicity. The cancer-related behaviors that Requests for reprints should be sent to Karen M. Em- have been associated with lower rates of were the targets of this study are also related mons, PhD, Harvard School of Public Health and Dana- breast cancer35 and reduction in morbidity to other diseases, which, taken together, ac- Farber Cancer Institute, Center for Community-Based Research, 44 Binney St, Boston, MA 02115 (e-mail: related to cardiovascular disease, osteoporo- count for a burden of suffering even greater [email protected]). sis, and birth defects.36,37 than for cancer alone. The effects of the inter- This article was accepted July 1, 2004.

1204 | Research and Practice | Peer Reviewed | Emmons et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

Contributors activity among employed adults—United States, 1990. 25. Emmons KM, Stoddard AM, Gutheil C, Suarez K.M. Emmons originated the study, served as principal MMWR Morb Mortal Wkly Rep. 2000;49:420–424. EG, Lobb R, Fletcher R. Cancer prevention for working investigator, supervised all aspects of study implementa- 8. Crespo CJ, Smit E, Andersen RE, Carter-Pokras O, class, multi-ethnic populations through health centers: tion, and oversaw article preparation. A.M. Stoddard Ainsworth BE. Race/ethnicity, social class and their rela- the Healthy Directions Study. Cancer Causes Control. served as study biostatistician and oversaw completion of tion to physical inactivity during leisure time: results from 2003;14:727–737. the data analysis. R. Fletcher served as a co-investigator the Third National Health and Nutrition Examination 26. Emmons KM, Rollnick S. Motivational interview- within the participating health care system and provided Survey, 1988–1994. Am J Prev Med. 2000;18:46–53. ing in health care settings: opportunities and limita- key input on study design and implementation related to 9. Chakravarthy M, Joyner M, Booth F. An obligation tions. Am J Prev Med. 2001;20:68–74. the managed care environment. C. Gutheil was project for primary care physicians to prescribe physical activ- 27.Resnicow K, DiIorio C, Soet JE, et al. Motivational director at Dana-Farber Cancer Institute and oversaw im- ity to sedentary patients to reduce the risk of chronic interviewing in medical and public health settings. In: plementation of all aspects of the study design and evalu- health conditions. Mayo Clin Proc. 2002;77:165–173. Miller W, Rollnick S, eds. Motivational Interviewing. 2nd ation. E.G. Suarez assisted with study implementation 10. Increasing physical activity: a report on recom- ed. New York, NY: Guilford Press; 2002:251–269. and supervision of project management staff and the mendations of the Task Force on Community Preven- 28. Subar AF, Heimendinger J, Patterson BH, health advisers. R. Lobb oversaw intervention delivery tive Services. MMWR Morb Mortal Wkly Rep. 2001; Krebs-Smith SM, Pivonka E, Kessler R. Fruit and vege- within the health care system and was responsible for 50(RR-18):1–16. table intake in the United States: the baseline survey of provider training and on-site supervision of the health the Five A Day for Better Health Program. Am J Health advisers. J. Weeks conducted the cost analyses. J.A. 11. Institute of Medicine. Speaking of Health: Assessing Promot. 19 95;9:352–360. Bigby participated in the interpretation of the findings. Health Communication Strategies for Diverse Populations. All authors were involved in article preparation. Washington, DC: National Academy Press; 2002. 29.Wolf AM, Hunter DJ, Colditz GA, et al. Repro- 12.Pelto G, Santo I, Goncalves H, Victora C, Martines J, ducibility and validity of a self-administered physical Habicht JP. Nutrition counseling training changes physi- activity questionnaire. Int J Epidemiol. 1994;23: Acknowledgments cian behavior and improves caregiver knowledge ac- 991–999. This research was supported by the National Institutes quisition. J Nutr. 2004;134:357–362. 30. The 2001 HHS Poverty Guidelines. Washington, of Health (grant 5 P01 CA75308) and by Liberty Mu- 13. Beresford SA, Curry SJ, Kristal AR, Lazovich D, DC: US Dept of Health and Human Services; 2001. tual, National Grid, and the Patterson Fellowship Fund Feng Z, Wagner EH. A dietary intervention in primary 31. Murray DM, Hannan PJ, Wolfinger RD, Baker WL, (to the Dana-Farber Cancer Institute). care practice: the Eating Patterns Study. Am J Public Dwyer JH. Analysis of data from group-randomized tri- The authors would like to thank the numerous staff Health. 19 97;87:610–616. als with repeat observations on the same groups. Stat members who contributed to this study, including Med. 1998;17:1581–1600. Elizabeth Alvarez, Jamie Baron, Tracy Liwen Chen, 14 . Goldstein MG, Niaura R, Willey C, et al. An aca- Martha Fay, Simone Pinheiro, Kathleen Scafidi, Tatyana demic detailing intervention to disseminate physician- 32. Littell R, Milliken G, George A, Stroup W, Wolfin- Pinchuk, and George Moseley. In addition, this work delivered smoking cessation counseling: smoking cessa- ger R. SAS Systems for Mixed Models. Cary, NC: SAS could not have been done without the participation of tion outcomes of the Physicians Counseling Smokers Institute Inc; 1996. the internal medicine departments of Harvard Van- Project. Prev Med. 2003;36:185–196. 33. Wolfinger R, O’Connell M. Generalized linear guard Medical Associates. 15. Rigotti NA, Munafo MR, Murphy MF, Stead LF. models: a pseudo-likelihood approach. J Stat Computa- Interventions for smoking cessation in hospitalised pation Simulation. 19 93;48:233–243. tients. Cochrane Database Syst Rev. 2003;(1):CD001837. Human Participant Protection 34. Giovannucci E, Stampfer MJ, Colditz G, et al. Mul- 16. Eden KB, Orleans CT, Mulrow CD, Pender NJ, tivitamin use, folate, and colon cancer in women in the The research presented here was approved by the insti- Teutsch SM. Does counseling by clinicians improve Nurses’ Health Study. Ann Intern Med. 19 9 8;129: tutional review boards for Harvard School of Public physical activity? A summary of the evidence for the 517–524. Health and Harvard Medical School’s Division of Am- U.S. Preventive Services Task Force. Ann Intern Med. 35. Rohan TE, Jain MG, Howe GR, Miller AB. Dietary bulatory Care and Prevention. 2002;137:208–215. folate consumption and breast cancer risk. J Natl Can- 17.Green BB, McAfee T, Hindmarsh M, Madsen L, cer Inst. 2000;92:266–269. References Caplow M, Buist D. Effectiveness of telephone support 36.Fairfield KM, Fletcher RH. Vitamins for chronic 1. National Cancer Institute. The Nation’s Investment in increasing physical activity levels in primary care pa- disease prevention in adults: scientific review. JAMA. in Cancer Research: A Plan and Budget Proposal for Fis- tients. Am J Prev Med. 2002;22:177–183. 2002;287:3116–3126. cal Year 2004. Washington, DC: National Cancer Insti- 18.Van Weel C. Dietary advice in family medicine. 37. Fletcher RH, Fairfield KM. Vitamins for chronic tute; October 2003. Am J Clin Nutr. 2003;77(4 suppl):1008S–1010S. disease prevention in adults: clinical applications. 2. Lantz PM, House JS, Lepkowski JM, Williams DR, 19. Stafford R, Farhat JH, Misra B, Schoenfeld DA. JAMA. 2002;287:3127–3129. Mero RP, Chen J. Socioeconomic factors, health behav- National patterns of physician activities related to obe- 38. Beresford SA, Thompson B, Feng Z, Christianson A, iors, and mortality: results from a nationally representa- sity management. Arch Fam Med. 2000;9:631–638. McLerran D, Patrick DL. Seattle 5 a Day worksite pro- tive prospective study of US adults. JAMA. 19 9 8;279: 20. Delichatsios HK, Hunt MK, Lobb R, Emmons K, gram to increase fruit and vegetable consumption. Prev 1703–1708. Gillman MW. EatSmart: efficacy of a multifaceted pre- Med. 2001;32:230–238. 3. Healthy People 2010: Understanding and Improving ventive nutrition intervention in clinical practice. Prev 39. Sorensen G, Stoddard A, Peterson K, et al. In- Health. Washington, DC: US Dept of Health and Med. 2001;33(2 Pt 1):91–98. creasing fruit and vegetable consumption through Human Services; 2001. 21. Sorensen G, Emmons K, Hunt MK, et al. Model worksites and families in the treatwell 5-a-day study. 4. US Department of Health and Human Services. for incorporating the social context in health behavior Am J Public Health. 1999;89:54–60. Physical Activity and Health: A Report of the Surgeon interventions: applications for cancer prevention for 40. Kristal AR, Curry SJ, Shattuck AL, Feng Z, Li S. General. Atlanta, Ga: Centers for Disease Control and working-class, multiethnic populations. Prev Med. A randomized trial of a tailored, self-help dietary inter- Prevention, National Center for Chronic Disease Pre- 2003;37:188–197. vention: the Puget Sound Eating Patterns study. Prev vention and Health Promotion; 1996. 22. Lobb R, Gonzalez Suarez E, Fay ME, et al. Imple- Med. 2000;31:380–389. 5. Enn C, Goldman J, Cook A. Trends in food and mentation of a cancer prevention program for working 41. Stevens VJ, Glasgow RE, Toobert DJ, et al. Ran- nutrient intakes by adults: NFCS 1977–1978, CSFII class, multiethnic populations. Prev Med. 2004;38: domized trial of a brief dietary intervention to decrease 1989–1991, and CDFII 1994–1995. Fam Econ Nutr 766–776. consumption of fat and increase consumption of fruits Rev. 19 97;10:2–15. 23. Krieger N, Williams DR, Moss NE. Measuring so- and vegetables. Am J Health Promot. 2002;16: 6. Smith E. Estimates of animal and plant protein cial class in US public health research: concepts, meth- 12 9–134. intake in U.S. adults: results from the Third National odologies, and guidelines. Annu Rev Public Health. 42. Centers for Disease Control and Prevention. Prev- Health and Nutrition Examination Survey, 1988–1991. 19 97;18:341–378. alence of physical activity, including lifestyle activities J Am Diet Assoc. 1999;99:813–820. 24. Census Use Study: Health Information System II. among adults—United States, 2000–2001. MMWR 7. Prevalence of leisure-time and occupational physical Washington, DC: US Bureau of Census; 1971. Morb Mortal Wkly Rep. 2003;52:764–769.

July 2005, Vol 95, No. 7 | American Journal of Public Health Emmons et al. | Peer Reviewed | Research and Practice | 1205  RESEARCH AND PRACTICE 

Work Factors and Occupational Class Disparities in Sickness Absence: Findings From the GAZEL Cohort Study

| Maria Melchior, ScD, Nancy Krieger, PhD, Ichiro Kawachi, MD, PhD, Lisa F. Berkman, PhD, Isabelle Niedhammer, PhD, and Marcel Goldberg, MD, PhD

Among working populations, occupational Objectives. To estimate the contribution of stress-related and physical work hazards and job stress may contribute to oc- factors to occupational class disparities in sickness absence from work. 1 cupational class disparities in health. Specifi- Methods. Our sample consisted of 8847 men and 2886 women participating in cally, adverse work conditions may influence the French GAZEL cohort study. Occupational class and medically certified sick- the risk of musculoskeletal disorders, psychi- ness absence data (1995–2001) were obtained from the participants’ employer. atric symptoms, and injury that occur fre- Work characteristics (physical and stress-related) were self-reported. We calcu- quently among middle-aged populations and lated rate ratios with Poisson regression models; fractions of sickness absence may constitute some of the leading reasons attributable to work factors were estimated with the Miettinen formula. for taking sick leave.2,3 There is some evi- Results. Sickness absence was distributed along an occupational gradient. dence that job stress contributes to occupa- Work characteristics accounted for 19% (women) and 21% (men) of all absences. Physical work conditions accounted for 42% and 13% of absences for muscu- tional class differences in both health and loskeletal reasons, and work stress accounted for 48% and 40% of psychiatric sickness-related absence from work (sickness absences. Overall, about 20% of the occupational class gradient in sickness ab- absence)4,5; however, little research has exam- sence could have been associated with deleterious work conditions. ined the contribution of other work condi- Conclusion. Work conditions contribute to sickness absence, particularly among tions. To date, only 2 studies have investi- manual workers and clerks. Policies that decrease ergonomic constraints and gated the joint contribution of physical and work stress also could reduce the burden of ill health and sickness absence among psychosocial work characteristics to occupa- the lowest strata of working populations. (Am J Public Health. 2005;95:1206–1212. tional class health disparities; however, both doi:10.2105/AJPH.2004.048835) were cross-sectional and used self-reported health as an outcome.6,7 In a previous analysis of the GAZEL co- EDF-GDF employees hold a civil servant– exposure data, which left a sample of 8847 hort study, we showed that job stress is most like status that entails job stability and oppor- men and 2886 women. The study population prevalent among manual workers and office tunities for occupational mobility. Typically, was healthier than GAZEL participants who clerks and predicts the occurrence of sickness employees are hired when they are in their did not complete the 1995 questionnaire.8 absence.8 In this study, we examined the 20s and stay with the company until retire- Person-time of follow-up with regard to contribution of both stress-related and physi- ment (usually around 55 years of age). Re- the outcome—sickness absence—was accrued cal work exposures to the occupational class tirees’ pensions are paid by the company. Be- from the date of completion of the 1995 gradient in overall and cause-specific sick cause of these characteristics, study follow-up GAZEL survey until the date of retirement, leave. is very thorough: since baseline (1989), less death, withdrawal from EDF-GDF or the than 1% of participants were lost to follow-up GAZEL cohort, or December 31, 2001 METHODS (39 left the company, and 19 withdrew from (whichever occurred first). the study). Study Population GAZEL participants are followed with an Measures The GAZEL cohort study began in 1989, annual mailed survey, which is usually com- Occupational class in 1995, which was ob- when 44 922 employees of France’s national pleted by 75% of the cohort.9 Additionally, tained from company records, was coded in gas and electricity company, Electricité de participants’ records are linked to validated accordance with the French national job clas- France-Gaz de France (EDF-GDF), were occupational and health data collected by the sification: manual workers, clerks, foremen/ asked to participate in a long-term observa- company, including medically certified sick- technicians, administrative associate profes- tional study. Forty-five percent of those ness absence. sionals, engineers, and managers. There eligible—14752 men and 5317 women— In this study, we analyzed data from GAZEL were no female manual workers in our accepted. At baseline, men were aged 40 to participants who responded to the 1995 survey study population. 50, and women were aged 35 to 50 years (11183 men and 4095 women; 75% of the Other explanatory variables were collected old. Women represented only 20% of original cohort), which included measures of from GAZEL surveys. Physical work character- company employees and therefore were job stress. We excluded respondents who had istics, including postural constraints (7 items), oversampled.9 retired (n=2304) or who had incomplete work occupational hazards (5 items), night work

1206 | Research and Practice | Peer Reviewed | Melchior et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

TABLE 1—Work Characteristics and Sickness Absence, by Gender and Participants’ Occupational Class: the GAZEL Cohort Study, 1995–2001

Manager Engineer Associate ProfessionalForeman/Technician Clerk Manual Worker P RR (95% CI)a

Men (n=8307) Percentage of study population 17.7 23.1 12.7 36.7 2.7 7.1 . . . Postural constraints, mean (SD) 0.15 (0.43) 0.22 (0.53) 0.30 (0.65) 0.78 (0.81) 0.57 (0.81) 1.68 (1.29) <.0001 1.60 (1.47–1.73) Hazardous work conditions, mean (SD) 0.53 (1.04) 1.33 (1.73) 0.53 (1.05) 2.29 (1.87) 0.87 (1.29) 3.04 (1.55) <.0001 1.28 (1.18–1.38) Night work, % 22.3 42.2 18.5 53.2 21.5 68.5 <.0001 1.10 (1.02–1.19) Work outdoors, % <.0001 Sometimes 34.2 47.9 30.4 52.9 28.3 16.1 1.05 (0.97–1.15) 50% of time 1.7 3.6 13.5 29.6 19.3 80.4 1.59 (1.44–1.75) Customer contact, % 24.5 21.7 50.8 38.0 69.0 77.0 <.0001 1.19 (1.10–1.29) Low decision latitude, % 14.9 17.7 32.1 39.9 73.9 67.2 <.0001 1.64 (1.54–1.75) High psychological demands, % 52.3 52.8 44.8 38.7 41.7 32.1 <.0001 0.90 (0.83–0.96) Low social support at work, % 36.0 31.3 38.9 32.9 49.2 38.2 <.0001 1.28 (1.20–1.36) Sick leave taken 1 or more times, % 56.8 52.7 71.4 72.7 87.4 80.5 <.0001 . . . Women (n=2886) Percentage of study population 7.4 5.1 58.3 10.9 18.3 . . . Postural constraints, mean (SD) 0.04 (0.20) 0.12 (0.45) 0.16 (0.44) 0.28 (0.59) 0.20 (0.44) . . . <.0001 1.39 (1.24–1.56) Hazardous work conditions, mean (SD) 0.07 (0.28) 0.29 (0.87) 0.15 (0.48) 0.34 (0.82) 0.14 (0.43) . . . <.0001 1.11 (0.97–1.27) Night work, % 5.6 12.2 2.4 4.4 0.8 . . . <.0001 0.83 (0.62–1.11) Work outdoors, % <.0001 Sometimes 7.5 9.8 8.3 14.3 3.0 . . . 0.97 (0.82–1.15) 50% of time ...... Customer contact, % 21.7 14.9 35.4 27.4 56.8 . . . <.0001 1.24 (1.13–1.36) Low decision latitude, % 18.9 20.9 51.5 53.1 74.4 . . . <.0001 1.34 (1.23–1.45) High psychological demands, % 56.1 61.5 40.7 29.6 40.3 . . . <.0001 1.17 (1.05–1.31) Low social support at work, % 53.3 52.7 52.7 48.7 61.7 . . . <.0001 1.06 (0.97–1.15) Sick leave taken 1 or more times, % 70.7 75.6 86.8 85.3 92.0 . . . <.0001 . . .

Note.CI=confidence interval.aAge-adjusted rate ratios of all-cause sickness-related absences from work. For postural constraints and hazardous work conditions, participants who reported any level of exposure were compared to the nonexposed.

(yes or no), and outdoor work activities factory factorial validity and adequate inter- for women), marital status (married/living (never, sometimes, >50% of the time) were nal consistency reliability, with Cronbach α with a partner, single, or divorced), current measured in 1990 (Table 1). These were the coefficients of 0.65 for decision latitude, 0.69 smoking (none vs at least 1 cigarette per only measures of physical work exposures for psychological demands, and 0.52 for day), alcohol consumption in drinks per available to us. For postural constraints and work social support (our measure of work so- week (none, light [1–13 for men, 1–6 for occupational hazards, we summed all the cial support showed lower reliability than in women], intermediate [14–27 for men, relevant items into summary scales, with other studies because it included fewer 7–20 for women], or heavy [≥ 28 for men, ranges of 0 to 4 and 0 to 7, respectively. The items).12 After verifying that the association ≥ 21 for women), body mass index (under- questionnaire also included a question about between quartiles of psychosocial work fac- weight [< 20], normal [20–24.9], over- customer contact (yes or no), which we con- tors and sickness absence was graded (data weight [25–29.9], and obese [≥ 30 kg/m2]), sidered to be potentially stressful. not shown), we dichotomized each scale at its and the occurrence of stressful life events Job stress measures, which were based median value.13 To examine the contribution (divorce or separation, partner’s death, death on the work of Karasek and Johnson,10 ,11 of work factors to the occupational gradient, of another family member, partner’s hospi- were obtained in 1995: control over the we modeled job stress factors as continuous talization, or partner’s unemployment) dur- content and the execution of work-related variables.14 ing the preceding 12 months (0, 1, 2, or tasks (6 items); psychological demands, evalu- Demographic and behavioral characteris- more). The number of dependents, as a ated work load, and time pressures (5 items); tics measured in 1995 included age (45–49, covariate, did not predict the occurrence of and social support received from colleagues 50–54, and 55–56 years for men, and sickness absence and was not included in (5 items). Each summary scale showed satis- 42–44, 45–49, 50–54, and 55–56 years our analysis.

July 2005, Vol 95, No. 7 | American Journal of Public Health Melchior et al. | Peer Reviewed | Research and Practice | 1207  RESEARCH AND PRACTICE 

Sickness Absence Data work factors to the occupational gradient was person-years of observation, and women ex- Sickness absence data (date, length, and measured by fitting a linear term for occupa- perienced 95; corresponding median num- medically certified diagnosis) were obtained tional class and comparing Models 4 and 1 bers of sick leave days were 8 and 36, re- from EDF-GDF company records. In princi- (we show Models 1 and 4; more detailed data spectively. Most absences (58%) lasted 7 days ple, the medical cause of all occurrences of are available upon request). Additionally, we or less, 26% lasted 8 to 21 days, and 17% sickness absence is verified by company phy- estimated the fraction of sickness absence at- lasted more than 21 days. Respiratory illness sicians and is recorded in accordance with a tributable to work factors with the Miettinen (14%) and musculoskeletal disorders (14%) classification derived from the International formula (attributable fraction=[RR–1]/ were the leading causes of absence, followed Classification of Diseases (ICD; Ninth Edition RR[no. exposed cases/no. cases]).17 For each by psychiatric reasons (7%) and injury (6%). for all nonpsychiatric diagnoses; for psychiat- work exposure statistically significant in Manual workers and office clerks more fre- ric disease, Ninth Edition until January 1, Model 4, the attributable fraction was calcu- quently reported job stress and physical work 19 97, and 10th Edition thereafter).14 ,15 In our lated controlling only for adjustment vari- exposures (Table 1). Male office clerks were data, the underlying diagnosis was missing for ables; the contribution of all work factors was more likely to be unmarried, to smoke ciga- 35% of the absences, 85% of which lasted calculated controlling for all work factors and rettes, and to abstain from alcohol; manual less than 8 days. We studied sick leave due to adjustment variables. workers reported the highest alcohol con- all medical causes and the 4 most common We verified the consistency of the results sumption and were more likely to be obese. medical causes: respiratory illness (ICD 9 in a subsample restricted to participants who Among women, clerks were most likely to be 460–519), disorders of the musculoskeletal worked in the same occupation in both 1990 divorced and obese, but smoking and alcohol system (ICD 9 710–739), psychiatric dis- and 1995 (n=8830) by using different consumption were more frequent among ease (ICD 9 299–305; ICD 10 F10–F99), lengths of sickness absence (≤7 days, 8–21 managers (data not shown). and injury (work and nonwork related, EDF- days, > 21 days) as an outcome. We tested As expected, sickness absence occurred GDF’s own classification). Absences due to reverse causation by repeating the analyses along an occupational gradient (Figure 1) and less common or unknown medical reasons in a “healthy” subsample composed of partici- was associated with work factors (Table 2). were classified as “other.” In secondary pants who did not experience sickness ab- After we adjusted for age, demographic char- analyses, we examined the occurrence of sence during the year preceding the begin- acteristics, and health behaviors, manual work- short (≤ 7 days), intermediate (8–21 days), ning of follow-up (n=7632). ers and clerks had rates of absence that were 3 and long ( > 21 days) absences. All analyses were conducted separately for times higher than the rates of managers men and women with SAS, version 8.2 (SAS (Table3, Model 1), regardless of the length of Statistical Analysis Institute Inc, Cary, NC); log-linear Poisson re- absence (data not shown). Occupational class Sickness absence occurrence was ex- gression models were fitted with the PROC disparities were greatest for absences owing to pressed in rates of absences over 100 per- GENMOD procedure.18 musculoskeletal causes (men and women) and son-years of observation. Rate ratios of sick- injury (men only). Work factors contributed to ness absence were computed with log-linear RESULTS these occupational class differences (Table3, Poisson regression models. Sickness absence Model 4): physical exposures among men who during any calendar year was correlated with The study population included 8847 men had musculoskeletal- and injury-related sick- future absences (by a factor of 0.5), which and 2886 women, on average, aged 50 years ness absences (particularly if the injury oc- could have led us to overestimate the vari- (range=47–56) and 48 years (range= curred at work), and work stress among men ance of the Poisson regression parameters. 42–56), respectively (Table 1). Men were and women who had musculoskeletal- and Therefore, all standard errors were adjusted most likely to work as technicians (37%), en- psychiatric-related sickness absence. with a scale parameter equivalent to the gineers (23%), or managers (18%); a majority Adjustment for all work factors reduced the residual deviance divided by the number of of women worked in administrative jobs occupational class gradient in all-cause sick- degrees of freedom.16 (58% in intermediate-grade positions, 18% as ness absence by 16% for men and 25% for The contribution of work factors to the oc- clerks). The proportion of manual workers women; musculoskeletal absences were re- cupational class gradient in sickness absence was lower than among the French population duced by 27% and 25%, and psychiatric ab- was evaluated in 4 steps. First, with managers of the same age. Eighty percent of partici- sences were reduced by 10% and 40%. Work as the reference category, we calculated rate pants worked in the same occupation in both factors’ contribution was greatest with regard ratios across occupational groups and adjusted 19 95 and 1990, when physical work charac- to men and women’s gradient in absences of for age, demographic characteristics, and teristics were measured. 8 to 21 days (21% and 27%, respectively) health behaviors (Model 1). Next, we succes- Over the 6 years of follow-up (mean=4.8, and was less for short and long absences (13 sively added physical work factors (Model 2) SD=2.0 for men; mean=5.8, SD=1.72 for days and 26 days; 9% and 20%, respec- and job stress (Model 3) before including all women), there were 18818 absences among tively). Associations between occupational work and adjustment variables in a single sta- men and 15803 female absences. On aver- class and work factors and sickness absence(s) tistical model (Model 4). The contribution of age, men experienced 47 absences per 100 were weaker among healthier participants

1208 | Research and Practice | Peer Reviewed | Melchior et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

TABLE 2—Occupational Class Gradients in Sickness Absence (1995–2001) and the Contribution of Work Factors: the GAZEL Cohort Study, 1995–2001

Total Sickness Respiratory Musculoskeletal Psychiatric Other Absence Illness Cause Diagnosis Injury Diagnosis RR (95% CI) RR (95% CI) RR (95% CI) RR (95% CI) RR (95% CI) RR (95% CI)

Men No. absences 18 818 absences 2721 absences 2945 absences 1015 absences 1339 absences 10 798 absences Model 1: adjusted for age, demographics, health behaviors Manager 1.0 1.0 1.0 1.0 1.0 1.0 Engineer 0.95 (0.85, 1.08) 1.14 (0.90, 1.44) 1.05 (0.79, 1.39) 0.67 (0.41, 1.10) 1.22 (0.89, 1.68) 0.91 (0.80, 1.04) Administrative associate professional 1.89 (1.68, 2.13) 2.74 (2.19, 3.42) 2.20 (1.67, 2.90) 2.82 (1.88, 4.22) 2.20 (1.60, 3.02) 1.59 (1.39, 1.82) Foreman/Technician 1.97 (1.78, 2.18) 2.92 (2.40, 3.56) 2.84 (2.24, 3.59) 1.63 (1.11, 2.38) 3.07 (2.35, 4.01) 1.62 (1.44, 1.81) Clerk 2.88 (2.46, 3.38) 3.61 (2.67, 4.88) 3.91 (2.76, 5.53) 3.17 (1.80, 5.58) 3.02 (1.93, 4.70) 2.55 (2.12, 3.06) Manual worker 3.06 (2.70, 3.45) 4.21 (3.33, 0.31) 5.83 (4.49, 7.57) 2.60 (1.63, 4.16) 4.58 (3.35, 6.27) 2.31 (2.00, 2.67) Model 4: Model 1+ work factors Manager 1.0 1.0 1.0 1.0 1.0 1.0 Engineer 0.96 (0.85, 1.08) 1.14 (0.89, 1.43) 1.04 (0.79, 1.38) 0.68 (0.42, 1.10) 1.21 (0.88, 1.67) 0.91 (0.81, 1.04) Administrative associate professional 1.79 (1.59, 2.01) 2.58 (2.06, 3.23) 1.99 (1.51, 2.61) 2.54 (1.70, 3.79) 2.08 (1.51, 2.87) 1.51 (1.32, 1.73) Foreman/Technician 1.82 (1.63, 2.03) 2.70 (2.18, 3.35) 2.26 (1.75, 2.91) 1.57 (1.03, 2.37) 2.73 (2.03, 3.68) 1.52 (1.34, 1.72) Clerk 2.42 (2.05, 2.85) 2.93 (2.13, 4.03) 2.84 (1.97, 4.09) 2.03 (1.11, 3.70) 2.52 (1.58, 4.03) 2.09 (1.73, 2.54) Manual worker 2.59 (2.23, 3.01) 3.66 (2.75, 4.87) 3.53 (2.56, 4.85) 2.28 (1.28, 4.05) 3.51 (2.38, 5.16) 2.03 (1.73, 2.54) Women No. absences 15 803 absences 2242 absences 1970 absences 1564 absences 754 absences 9273 absences Model 1: adjusted for age, demographics, health behaviors Manager 1.0 1.0 1.0 1.0 1.0 1.0 Engineer 1.20 (0.88, 1.63) 1.16 (0.65, 2.10) 1.99 (1.02, 3.89) 0.72 (0.29, 1.77) 1.25 (0.63, 2.47) 1.16 (0.83-, .61) Administrative associate professional 1.99 (1.61, 2.46) 2.56 (1.71, 3.84) 2.64 (1.57, 4.45) 2.01 (1.18, 3.41) 1.63 (1.01, 2.63) 1.83 (1.46, 2.31) Foreman/Technician 2.07 (1.63, 2.62) 2.44 (1.56, 3.80) 2.97 (1.69, 5.22) 1.73 (0.95, 3.17) 1.78 (1.03, 3.05) 1.96 (1.52, 2.53) Clerk 2.76 (2.21, 3.44) 3.71 (2.45, 5.61) 4.15 (2.44, 7.07) 3.14 (1.82, 5.40) 2.33 (1.41, 3.82) 2.39 (1.88, 3.03) Model 4: Model 1+ work factors Manager 1.0 1.0 1.0 1.0 1.0 1.0 Engineer 1.18 (0.88, 1.60) 1.16 (0.64, 2.08) 1.92 (1.00, 3.68) 0.73 (0.30, 1.74) 1.28 (0.65, 2.50) 1.15 (0.83, 1.60) Administrative associate professional 1.80 (1.46, 2.23) 2.42 (1.61, 3.62) 2.26 (1.35, 3.76) 1.70 (1.00, 2.86) 1.53 (0.95, 2.47) 1.69 (1.34, 2.13) Foreman/Technician 1.88 (1.48, 0.39) 2.36 (1.51, 3.69) 2.40 (1.37, 4.17) 1.54 (0.85, 2.79) 1.69 (0.98, 2.92) 1.82 (1.41, 2.36) Clerk 2.30 (1.84, 2.89) 3.22 (2.10, 4.93) 3.20 (1.87, 5.46) 2.17 (1.24, 3.78) 2.04 (1.21, 3.42) 2.03 (1.58, 2.60)

Note. RR=relative ratio; CI=confidence interval. aBased on 39647 person-years of observation. bBased on 20509 person-years of observation.

than among the full sample. All-cause absen- sences and for 15% of their injury-related tional class gradient in absenteeism were re- teeism rates among manual workers and absences. Work stress contributed to lated to adverse work conditions. Adding to clerks were 2 times higher than among man- musculoskeletal-related absences (15% for previous research,19 we found that occupa- agers, which is 30% lower than among the low decision latitude) and psychiatric-related tional gradients in cause-specific sickness ab- entire study population, and work factors’ con- absences (a lack of social support among sences were associated with physical and psy- tribution to men and women’s occupational men; low decision latitude among women). chosocial work exposures. Occupational class gradients were 35% and 70% lower. differences in sickness absences due to injury Across occupational groups, 21% of men’s DISCUSSION were associated with physical work exposures, and 19% of women’s all-cause sickness ab- and work stress contributed to psychiatric sences were attributable to work factors Main Results sickness absence gradients. Sickness absence (Table 3). Postural constraints accounted for Overall, 19% to 21% of all-cause sickness due to musculoskeletal reasons reflected both 24% of men’s musculoskeletal-related ab- absences and 16% to 25% of the occupa- physical and stress-related work exposures.

July 2005, Vol 95, No. 7 | American Journal of Public Health Melchior et al. | Peer Reviewed | Research and Practice | 1209  RESEARCH AND PRACTICE 

causes and injury, which is plausible. Further- 140 133 Managers more, sickness absence data were obtained Engineers from company records, which limited the in- 120 Associate fluence of reporting bias. Nonetheless, it is professionals possible that individuals who complain about Technicians their work conditions are also likely to take 96 100 Clerks 93 90 sick leave for reasons other than health. Fu- Manual workers 86 ture research should test the role of more ob- 80 jective measures of work characteristics. A second limitation is that physical work exposures were obtained 5 years before the 56 57 60 53 beginning of follow-up, which may have re- 45 sulted in measurement error. About 80% of 40 participants held the same job in both 1990 29 and 1995, and results among this subgroup

Absences per 100 Person-Years Absences 27 were comparable to the full sample, which is 20 reassuring. However, it is possible that levels of exposure change over time, even within oc-

0 cupational groups, which may have reduced MenWomen the precision of our estimates. Gender Third, to calculate attributable fractions, we dichotomized work-stress variables at their FIGURE 1—Occupational class and sickness absence, by gender, in the GAZEL cohort study, median value. Although this is the standard 1995–2001. method used in the field, we recognize that it makes comparisons with other studies difficult. To our knowledge, work stress in GAZEL 22 TABLE 3—Population Fraction of Sickness Absence Attributable to Work Factors, by Cause: was as frequent as in other studies, and our the GAZEL Cohort Study, 1995–2001 results are valid among other populations. However, identifying meaningful work-stress Musculoskeletal Psychiatric exposure thresholds, which would make com- All Causes, % Reasons, % Reasons, % Injury, % parisons across study populations more Men Women Men Women Men Women Men Women straightforward, is an important goal for fu- Postural constraints 10 4 24 9 a 6154 ture research. Occupational hazards 6 1 12 3 ...... 6 . . . Finally, we studied employees of a large Night work ...... <1 ...... public sector company who were healthier 23 Outdoor work activities . . . 6 1 . . . <1 . . . 2 than the general population of France, and Customer contact . . . 4 . . . 2 . . . 9 . . . 9 we probably underestimated both occupa- Low decision latitude 9 9 15 15 13 19 7 . . . tional class differences in sickness absence High psychological demands <1 4 ...... 11 7 . . . 4 and the role of work factors. At the same Low social support at work 8 2 10 . . . 24 5 . . . 10 time, GAZEL participants’ jobs are not at All work factorsb 21 19 46 27 42 43 26 26 stake if they take sick leave, and they may be less reluctant to be absent from work when ill a The exposure was not significantly associated with the outcome in a model that adjusted for age, marital status, than men and women who are in more unsta- occupational class, cigarette smoking, alcohol consumption, body mass index, and life events. 24 b The fraction of sickness absence attributable to all work factors was estimated in models that included all work exposures ble job situations. Other large cohorts, such and that were adjusted for age, marital status, occupational class, cigarette smoking, alcohol consumption, body mass index, as the Whitehall II study, faced similar and life events. issues.4 In our study, sickness absence rates were somewhat lower than among the White- hall II cohort, but not by very much (47 and Limitations tivity. However, our measures of psychosocial 95 absences per 100 person-years for Our study has several limitations. First, work characteristics have previously been val- GAZEL men and women, respectively, com- work exposures were self-reported and may idated,8,20,21 and postural constraints and oc- pared with 70 and 120 absences, respec- have reflected personality factors for which cupational hazards predicted the occurrence tively, in the Whitehall II study). Therefore, we could not control, such as negative affec- of sickness absence owing to musculoskeletal our findings can be compared with previous

1210 | Research and Practice | Peer Reviewed | Melchior et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

reports from that cohort.4 Yet, more broadly, planations.25 Low job control and insufficient within a broader social, political, and eco- it is important to recognize that sickness ab- support from colleagues and supervisors may nomic context, and sickness absence also re- sence patterns vary across work sectors, directly undermine psychological well-being flects the generosity of sick leave provisions workplaces, time periods, countries, and study and thus increase the risk for depression5 and macroeconomic trends (e.g., downsizing populations, and their association with occu- while simultaneously affecting health behav- of firms and contingent job insecurity).32,33 pational factors needs to be studied among iors (e.g., cigarette smoking, alcohol con- The population effects of these macrolevel other working populations. sumption, and behaviors that result in being factors, which we did not take into account overweight).26,27 because we restricted our study to middle- Gender Differences Manual workers and clerks are simultane- aged employees of a single company with Patterns and causes of sickness absence and ously exposed to a variety of deleterious work high levels of job security, deserve further re- distributions of jobs and work factors vary by factors, some of which can be interrelated. search attention. gender. In our study, men were more likely to For example, occupational hazards may cause work as manual workers, report numerous stress. In our study, work variables showed at Health Selection into Occupational Groups postural constraints, and seek sickness absence most modest collinearity, and we chose to in- We hypothesized that higher sickness due to musculoskeletal problems; women were clude them jointly in our statistical models. absence rates among manual workers and more likely to work in white collar jobs, suffer We assumed that the effects of physical and clerks would partly reflect detrimental work from work stress, and show susceptibility to stress-related factors were additive; however, conditions. Yet, individuals are not randomly psychiatric sickness absence. Even within we acknowledge that other approaches are selected into occupations, and health influ- broad occupational groups, job titles differed possible (e.g., multiplicative models). ences occupational attainment. In France, between men and women. For instance, fe- An important question when interpreting salaried workers undergo a medical examina- male managers held jobs with lower responsi- our results is whether the association between tion before starting a new job, and the bility levels than men, and although female work factors and sickness absence can be GAZEL cohort did not include individuals clerks often worked as receptionists, male considered causal. Indeed, it may be that indi- who were severely ill when recruited by clerks were in positions with less frequent cus- viduals whose work conditions are the worst EDF-GDF. However, less healthy individuals tomer contact (e.g., administrative staff). It is and who are employed in subordinate jobs were probably less likely to be promoted34: important to keep these gender specificities in are also exposed to nonwork situations associ- in our study, 39% of male office clerks previ- mind when studying occupational gradients in ated with sickness absence (e.g., comorbidity ously held a manual occupation and may both sickness absence and health. or a lack of personal social support). More have switched to an office-based job for broadly, thought needs to be given to the health reasons. Furthermore, occupational Work Factors and Sickness Absence complex associations between work expo- class and work characteristics were less The global contribution of work factors to sures, nonwork characteristics, and sickness strongly associated with sickness absence the occupational gradient in sickness absence absence, including their patterning along oc- among participants who had not experienced reflects several mechanisms, some of which cupational class lines. sickness absence during the preceding year, are occupation- and gender-specific. Postural which, along with findings from other popu- constraints encompass a wide array of expo- Sickness Absence as a Measure of Health lations, suggests that both health-related se- sures that increase the risk for disorders of or Sickness absence reflects not only health lection and social causation contribute to oc- injury to the back, neck, shoulder, and upper but also attitudes toward health and work, as cupational class health disparities.34–36 limbs.2,25 Some of these are characteristic of well as job authority. We were concerned that Additional longitudinal studies are needed to manual occupations (e.g., carrying heavy managers may miss work for health reasons disentangle the role of these 2 processes. loads, experiencing vibrations), while others without requesting sick leave, particularly for also occur in office jobs (e.g., standing or sit- short periods of time. Reassuringly, in our Conclusions ting in uncomfortable positions over extended study occupational gradients were similar In the GAZEL study, work conditions ac- periods of time). To gain a better understand- across different durations of sickness absence, count for approximately 20% of occupational ing of occupational gradients in specific dis- which suggests that our measure was ade- class differences in sickness absence. Our eases of the musculoskeletal system, studies quate across occupational groups. study did not include extremely disadvan- that use more precise postural exposures are Despite potential biases that may have af- taged workers, and our results probably unnecessary. fected reports of sickness absence, this indica- derestimate the effect of work on health- The health effects of work stress may be tor bears public health relevance because it related absenteeism among the general due to direct and indirect mechanisms. The reflects individuals’ general physical, psycho- population. Policies that decrease postural specific pathways of musculoskeletal problems logical, and social well-being8,28–30 and col- constraints and job stress have the potential are not yet well understood, but increased lective workplace factors (e.g., it is lower in to improve the health of men and women muscle tension and the inability to take neces- workplaces that have equitable policies).31 who work in manual and clerical jobs and to sary breaks from work are the most likely ex- Organizations and employees are embedded reduce the burden of sickness absence, partic-

July 2005, Vol 95, No. 7 | American Journal of Public Health Melchior et al. | Peer Reviewed | Research and Practice | 1211  RESEARCH AND PRACTICE 

ularly when associated with musculoskeletal social class differences in depression and well-being. 22. Karasek R, Brisson C, Kawakami N, Houtman I, problems and psychiatric reasons. Soc Psychiatry Psychiatr Epidemiol. 19 9 8;33:1–9. Bongers P, Amick B. The Job Content Questionnaire 6. Schrijvers CT, van der Mheen HD, Stronks K, (JCQ): an instrument for internationally comparative Mackenbach JP. Socioeconomic inequalities in health in assessments of psychosocial job characteristics. J Occup Health Psychol. 1998;3:322–355. About the Authors the working population: the contribution of working conditions. Int J Epidemiol. 1998;27:1011–1018. 23. Goldberg M, Chastang J-F, Leclerc A, Zins M, Maria Melchior, Isabelle Niedhammer, and Marcel Gold- Bonenfant S, Bugel I. Socioeconomic, demographic, oc- berg are with the National Institute of Health and Medical 7. Borrell C, Muntaner C, Benach J, Artazcoz L. So- cupational and health factors associated with participa- Research, Saint-Maurice, France. Nancy Krieger, Ichiro cial class and self-reported health status among men tion in a long-term epidemiological survey. A prospec- Kawachi, and Lisa F. Berkman are with the Department of and women: what is the role of work organisation, tive study of the French Gazel cohort and its target Society, Human Development and Health, Harvard School household material standards, and household labour? population. Am J Epidemiol. 2001;154:373–384. of Public Health, Boston, Mass. Soc Sci Med. 2004;58:1869–1887. Reprint requests should be sent to Maria Melchior, 8. Melchior M, Niedhammer I, Berkman LF, Gold- 24.Aronsson G, Gustafsson K, Dallner M. Sick but yet INSERM, U687-IFR69, HNSM, 14, rue du Val d’Osne, berg M. Psychosocial work factors, social relations, and at work. An empirical study of sickness presenteeism. 94415 Saint-Maurice, France (e-mail: maria.melchior@ sickness absence: a 6-year prospective study of the J Epidemiol Community Health. 2000;54:502–509. st-maurice.inserm.fr). GAZEL cohort. J Epidemiol Community Health. 2003; 25. Leclerc A, Chastang J-F, Niedhammer I, Landre M-F, This article was accepted October 12, 2004. 57:285–293. Roquelaure Y. Study Group on Repetitive Work. Inci- 9. Goldberg M, Leclerc A, Chastang J, Morcet J, dence of shoulder pain in repetitive work. Occup Envi- Contributors Marne M, Luce D. Mise en place d’une cohorte ron Med. 2004;61:39–44. épidémiologique à Electricité de France—Gaz de M. Melchior designed the study, performed the analy- 26. Sorensen G, Stoddard A, Hammond SK, Hebert JR, France: principales caractéristiques de l’échantillon. ses, and wrote the article. N. Krieger, I. Kawachi, L.F. Avrunin JS, Ockene JK. Double jeopardy: workplace Rev Epidemiol Sante Publique. 1990;38:378–380. Berkman, and M. Goldberg helped design the study hazards and behavioral risks for craftspersons and la- and interpret the findings. I. Niedhammer originated 10. Karasek R, Theorell T. Healthy Work: Stress, Pro- borers. Am J Health Promotion. 1996;10:355–363. ductivity and the Reconstruction of Working Life. New the study and assisted with study design. 27. Brisson C, Larocque B, Moisan J, Vezina M, Dage- York, NY: Basic Books; 1990. nais G. Psychosocial factors at work, smoking, seden- Acknowledgments 11. Johnson J, Hall E, Theorell T. Combined effects of tary behavior, and body mass index: a prevalence job strain and social isolation on cardiovascular disease The authors wish to thank EDF-GDF and the men and study among 6995 white collar workers. J Occup Envi- morbidity and mortality in a random sample of the women who participate in the GAZEL study. We partic- ron Med. 2000;42:40–46. Swedish male working population. Scand J Work Envi- ularly thank the Service des Etudes Médicales, Service ron Health. 19 8 9;15:271–279. 28. Kivimäki M, Head J, Ferrie JE, Shipley MJ, Général de Médecine de Contrôle, who collected the Vahtera J, Marmot MG. Sickness absence as a global 12. Stansfeld SA, Bosma H, Hemingway H, sickness absence data. We are grateful to the GAZEL measure of health: evidence from mortality in the Marmot MG. Psychosocial work characteristics and cohort study team, particularly Sébastien Bonenfant, Whitehall II prospective cohort study. Br Med J. 2003; social support as predictors of SF-36 health function- who is responsible for data management. Additionally, 327:1–6. we thank Alice Guégen for statistical advice, and David ing: the Whitehall II study. Psychosom Med. 1998;60: 29. Kristensen TS. Sickness absence and work strain Ellwood, Annette Leclerc, Doris V. Báez-Feliciano, and 247–255. among Danish slaughterhouse workers: an analysis of an anonymous reviewer for their comments on previ- 13. Landsbergis P, Theorell T. Measurement of psy- absence from work regarded as coping behaviour. Soc ous versions of the article. chosocial workplace exposure variables. Occup Med. Sci Med. 19 91;32:15–27. 2000;15:163–188. 30.Rael E, Stansfeld S, Shipley M, Head J, Feeney A, Human Participant Protection 14 . International Classification of Diseases. Manual of the International Statistical Classification of Diseases, In- Marmot M. Sickness absence in the Whitehall II study, The GAZEL cohort received the approval of France’s juries and Causes of Death. Geneva, Switzerland: World London: the role of social support and material prob- national ethics committee (Commission Nationale Infor- Health Organization; 1977. lems. J Epidemiol Community Health. 19 95;49:474–481. matique et Liberte); our study was conducted with the 31. Kivimäki M, Elovainio M, Vahtera J, Ferrie JE. Or- approval of the human subjects committee of the Har- 15. International Classification of Diseases: 10th Revi- ganisational justice and health of employees: prospec- vard School of Public Health. sion. Geneva, Switzerland: World Health Organization; 19 92. tive cohort study. Occup Environ Med. 2003;60:27–33. References 16. Niedhammer I, Bugel I, Goldberg M, Leclerc A, 32. Kivimäki M, Vahtera J, Ferrie J, Hemingway H, Guegen A. Psychosocial factors at work and sickness Pentti J. Organisational downsizing and musculoskeletal 1. Goldberg M, Melchior M, Leclerc A, Lert F. absence in the Gazel cohort: a prospective study. Occup problems in employees: a prospective study. Occup En- Epidémiologie et déterminants sociaux des inégalités Environ Med. 1998;55:735–741. viron Med. 2001;58:811–817. de santé. Rev Epidemiol Sante Publique. 2003;51: 381–401. 17. Hanley J. A heuristic approach to the formulas for 33. Knutsson A, Goine H. Occupation and unemploy- population attributable fraction. J Epidemiol Community ment rates as predictors of long-term sickness absence 2. Hoogendoorn WE, Bongers PM, de Vet HC, Health. 2001;55:508–514. in two Swedish counties. Soc Sci Med. 1998;47:25–31. Ariens GA, van Mechelen W, Bouter LM. High physi- 18.SAS Institute. SAS/STAT Software: Changes and 34. Ribet C, Zins M, Guéguen A, et al. Occupational cal work load and low job satisfaction increase the risk Enhancements Through Release 6.12. Cary, NC: SAS In- mobility and cardiovascular risk factors in working of sickness absence due to low back pain: results of a stitute; 1997. men: selection, causality, or both? Results from the prospective cohort study. Occup Environ Med. 2002; GAZEL study. J Epidemiol Community Health. 2003; 59:323–328. 19. North F, Syme SL, Feeney A, Head J, Shipley MJ, Marmot MG. Explaining socioeconomic differences in 57:901–906. 3. Paterniti S, Niedhammer I, Lang T, Consoli SM. sickness absence: the Whitehall II study. Br Med J. 35. Power C, Matthews S, Manor O. Inequalities in Psychosocial factors at work, personality traits and de- 19 93;306:361–365. self-rated health in the 1958 birth cohort: lifetime so- pressive symptoms. Longitudinal results from the 20. Niedhammer I. Psychometric properties of the cial circumstances or social mobility? Br Med J. 1996; GAZEL Study. Br J Psychiatry. 2002;181:111–117. French version of the Karasek Job Content Question- 313:449–453. 4. North F, Syme SL, Feeney A, Shipley M, naire: a study of the scales of decision latitude, psycho- 36. Hallqvist J, Lynch JW, Bartley M, Lang T, Blane D. Marmot MG. Psychosocial work environment and logical demands, social support and physical demands. Can we disentangle life course processes of accumula- sickness absence among British civil servants: the Int Arch Occup Environ Health. 2002;75:129–144. tion, critical period and social mobility? An analysis of Whitehall II study. Am J Public Health. 19 9 6;8 6: 21. Césard M, Dussert F. Le travail ouvrier sous con- disadvantaged socio-economic positions and myocardial 332–340. trainte. In: INSEE, ed. Données Sociales. Paris, France: infarction in the Stockholm Heart Epidemiology Pro- 5. Stansfeld SA, Head J, Marmot MG. Explaining INSEE; 1993:202–211. gram (SHEEP). Soc Sci Med. 2004; 8:1555–1562.

1212 | Research and Practice | Peer Reviewed | Melchior et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

Injuries at Work in the US Adult Population: Contributions to the Total Injury Burden

| Gordon S. Smith, MB, ChB, MPH, Helen M. Wellman, MS, Gary S. Sorock, PhD, Margaret Warner, PhD, Theodore K. Courtney MS, CSP, Glenn S. Pransky, MD, MoccH, and Lois A. Fingerhut, MA

In 1997, an estimated 34.4 million medically Objectives. We estimated the contribution of nonfatal work-related injuries on treated injuries and poisonings occurred in the injury burden among working-age adults (aged 18–64 years) in the United the United States—a rate of 12.9 episodes per States. 1 100 persons. Injuries to working-age adults, Methods. We used the 1997–1999 National Health Interview Survey (NHIS) to which represent an important part of this in- estimate injury rates and proportions of work-related vs non–work-related injuries. jury burden, are often overlooked despite Results. An estimated 19.4 million medically treated injuries occurred annu- having a larger economic impact than injuries ally to working-age adults (11.7 episodes per 100 persons; 95% confidence in- to other age groups.2 Until recently, compara- terval [CI]=11.3, 12.1); 29%, or 5.5 million (4.5 per 100 persons; 95% CI=4.2, 4.7), ble national data that distinguished work- occurred at work and varied by gender, age, and race/ethnicity. Among employed related from non–work-related injuries were persons, 38% of injuries occurred at work, and among employed men aged 55–64 years, 49% of injuries occurred at work. not available on the incidence and types of Conclusions. Injuries at work comprise a substantial part of the injury burden, injuries to working-age adults. Moreover, accounting for nearly half of all injuries in some age groups. The NHIS provides existing workplace-based reporting systems an important source of population-based data with which to determine the work may substantially underestimate occupational relatedness of injuries. Study estimates of days away from work after injury were 3–23 injuries. When adjustment is made for 1.8 times higher than the Bureau of Labor Statistics (BLS) workplace-based esti- underreporting, annual estimates of nonfatal mates and 1.4 times as high as BLS estimates for private industry. The prominence work-related injuries in the United States of occupational injuries among injuries to working-age adults reinforces the need range from 6 million to 13 million.3,5 No com- to examine workplace conditions in efforts to reduce the societal impact of injuries. prehensive national data exist on nonfatal (Am J Public Health. 2005;95:1213–1219. doi:10.2105/AJPH.2004.049338) work-related injuries or even work-related injury hospitalizations.12–14 sample of the noninstitutionalized US civilian a computer-assisted personal interview. Data The National Health Interview Survey population.1,28,29 For each sampled house- from 1997 to 1999, the first 3 years of the (NHIS) is one of the most important tools hold, in-person interviews are conducted with redesigned survey, were combined to de- for monitoring the health of the US popula- a resident adult who provides information on velop national estimates of nonfatal work- tion.24–27 It recently has been revised to all household members. Information collected related and non–work-related injuries. The improve the quality of data collected on in- includes demographic and personal character- unit of analysis for the study was the injury juries and includes work relatedness.1 This istics, health status, and use of health re- episode; an injured person may have multi- revision now provides an opportunity to ex- sources. Most analyses in this article were de- ple episodes. amine all injuries to adults in the US popu- rived from data reported by the resident adult Injuries were defined as work related or lation and to assess those occurring at work, on behalf of all members of the family. How- at work if the response “Working at a paid independent of workplace reporting. We ever, to obtain the most accurate information job” was selected in answer to the question, used data from the redesigned NHIS for on employment, occupation, industry, and “What were you doing when the injury hap- 19 97–1999 to examine the contribution of other items, 1 adult in each household is ran- pened?” Employed persons were defined as nonfatal injuries at work to the total injury domly selected as the “sample” adult from those who reported employment in a job or burden of working-age adults (aged 18–64 whom data are gathered directly (data avail- business during the week before the inter- years). able as a separate sample adult file). view (regardless of whether they worked Beginning in 1997, more detailed informa- that week). It is possible that employment METHODS tion on injuries was collected by the NHIS.1 status changed between the time of the in- Details concerning all medically treated injury, which could be up to 3 months prior The NHIS is a nationwide survey con- juries (including those injuries about which to the interview, and the week before the ducted by the National Center for Health advice was received) during the past 3 interview. Such a status change could result Statistics with trained interviewers from the months to any member of the household, in some misclassification of employment sta- Census Bureau. The survey provides health including injuries related to any paid work, tus, but we assumed such changes to be information about a nationally representative were obtained from the respondent through minimal. We excluded all poisonings from

July 2005, Vol 95, No. 7 | American Journal of Public Health Smith et al. | Peer Reviewed | Research and Practice | 1213  RESEARCH AND PRACTICE 

our analyses, because the NHIS asked sepa- TABLE 1—Number, Percentage, and Rate of All Injury Episodes and Work-Related Injury rate questions for injuries and for poison- Episodes Among the Working-Age Population (Aged 18–64 years), by Age, Gender, and ings and did not ascertain the work related- Race/Ethnicity: National Health Interview Survey, 1997–1999 ness of poisonings during 1997–1999. Musculoskeletal disorders are also not in- All Injuries Annually Work-Related Injuries Annually cluded by definition. No. in Rate per 100 No. in Rate per 100 Employed 1000s (%) Populationa (95% CI) 1000s (%) Populationb (95% CI) Study Population The working-age population was defined as Age, y persons aged 18 to 64 years. People younger 18–24 3807 (19.6) 14.9 (13.8, 16.1) 1088 (19.6) 6.6 (5.7, 7.5) than 18 years and people 65 years and older 25–34 4858 (25.1) 12.5 (11.8, 13.3) 1561 (28.2) 5.0 (4.4, 5.5) were excluded because of the small numbers 35–44 5191 (26.8) 11.7 (11.0, 12.5) 1401 (25.3) 3.8 (3.4, 4.2) of work-related injury episodes reported for 45–54 3576 (18.4) 10.4 (9.6, 11.1) 1069 (19.3) 3.9 (3.4, 4.4) these age groups (n=19 and 25, respectively 55–64 1961 (10.1) 8.8 (8.0, 9.6) 426 (7.7) 3.4 (2.7, 4.0) [unweighted]). A total of 113614 households Gender were surveyed by the NHIS during the 3 year- Male 11 189 (57.7) 13.8 (13.2, 14.4) 3863 (69.7) 5.8 (5.4, 6.2) period, with an adult representative providing Female 8204 (42.3) 9.8 (9.3, 10.2) 1683 (30.3) 2.9 (2.6, 3.2) information for 298388 household mem- Race/ethnicity bers; 99357 sample adults were interviewed White non-Hispanic 15 320 (79.0) 12.8 (12.3, 13.3) 4392 (79.2) 4.7 (4.4, 5.0) directly. The overall survey response rate for Black non-Hispanic 2030 (10.5) 10.4 (9.5, 11.4) 542 (9.8) 4.0 (3.3, 4.7) all ages was 90.3% in 1997, 88.2% in 1998, Hispanic 1438 (7.4) 7.9 (7.2, 8.7) 486 (8.8) 3.8 (3.3, 4.4) and 86.1% in 1999.29,30 Other 605 (3.1) 8.0 (6.5, 9.5) 126 (2.3) 2.3 (1.5, 3.2) Total 19 393 (100.0) 11.7 (11.3, 12.1) 5546 (100.0) 4.5 (4.2, 4.7) Data Analysis The descriptive analysis was performed Note. CI=confidence interval. aAll people in the population aged 18 to 64 years regardless of employment (n=165 million). with SAS version 8.0 (SAS Institute Inc, bPeople who were reported to be employed at a job or business in the week before the interview (n=124 million). Cary, NC). To derive national estimates, sample weights that accounted for the complex sample design of the survey were as- age group was tested by fitting a linear regres- nually among the estimated 124 million peo- signed by the National Center for Health sion model weighted inversely to the variance ple who were employed at the time of the in- Statistics for each respondent on the basis of of each rate. Tests of differences between esti- terview, a rate of 4.5 per 100 employed per- the number and composition of households; mates discussed in the text were conducted sons. For men, the overall injury rate was these weights included adjustment for non- with these methods (results not shown). 40% higher than the rate for women, but the response.29 Weights also were provided for work-related injury rate was double that of the sample adults. The denominator popula- RESULTS women. Overall injury rates decreased with tion used for calculating injury rates was ob- age. The trend, however, was driven by the tained directly from the survey. Survey respondents reported a total of rate among men, because no trend by age The distribution of injuries was examined 4925 injury episodes (unweighted) among was found among the rates for women (Fig- for all injury episodes by various characteris- adults aged 18 to 64 years from 1997 to ure 1). When stratified by gender, the rates tics and by work relatedness. Standard errors 1999. When weighted, these responses for men were highest among those aged 18 to and 95% confidence intervals were calcu- formed the basis of our national estimates. 24 years (Figure 1), both for all injuries (19.8 lated with SUDAAN software31 to account for There were 1386 persons in the survey who per 100) and for work-related injuries (9.2 the complex, multistage sample design used were injured at work, resulting in 1422 work- per 100), and declined significantly by age. in the survey. Estimates with relative standard related injury episodes (2.4% sustained 2 epi- Non-Hispanic White adults had the highest errors greater than 30% were considered un- sodes during the 3-month period, and 2 peo- injury rates, followed by non-Hispanic Black reliable. To identify contrasts between sub- ple sustained 3 or more episodes). adults and Hispanic adults (Table 1). When populations that were both meaningful and the data were limited to work-related injuries, significant, differences in injury rates and pro- Injury Estimates and Rates the order of the rates was the same, but the portions were tested with a 2-sided z test at The civilian, noninstitutionalized adult US differences among racial/ethnic groups were the 0.05 level of significance. Because multi- population aged 18 to 64 years (approxi- smaller and nonsignificant. When rates were ple, simultaneous tests were performed, the mately 165 million) sustained an estimated stratified by age and gender, differences in Bonferroni method was used to provide a 19.4 million injury episodes annually, a rate work-related injury rates by race/ethnicity more conservative threshold for identifying of 11.7 per 100 persons (Table 1). About 5.5 persisted and were greatest among younger significant differences. The trend in rates by million at-work injury episodes occurred an- men (data not shown).

1214 | Research and Practice | Peer Reviewed | Smith et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

mated 3.6 million work-related injury episodes resulted in at least part of a day off work. Of all injury episodes among the working-age population, 46.1% resulted in loss of at least part of a day of work, and 38.1% involved loss of 1 or more days (1–5 days lost, 21.6%; 6 or more days lost, 16.5%). Of all work- related injury episodes, 65.6% resulted in at least part of a day of lost work, and 51.0% involved loss of 1 or more days (1–5 days lost, 27.5%; 6 or more days lost, 23.6%) (data not shown). To compare NHIS data with data from Bu- reau of Labor Statistics (BLS) surveys,6,7,15–17 which cover only private industry, we conducted separate analyses of the NHIS sample adult file; industry data were gathered only from the sampled adults. On the basis of the data from the sample adult file, 50.4% of the FIGURE 1—Annual rates of total injuries and work-related injuries by gender and age: work-related injury episodes resulted in 1 or National Health Interview Survey, 1997–1999. more days off work. Of the estimated 3.01 million annual occupational injuries resulting in at least 1 day off work reported by sample Proportion of All Injuries That Were accounted for the highest percentages of adults, 2.36 million (78.4%) were reported Work Related work-related injuries. Only 17% of fractures by employees in private industry; the remain- Among all working-age persons, 28.6% of and 18% of superficial injuries were work der were reported by government employees, injuries occurred while people were working related. Overall, 11.3% of injury episodes farm workers, or the self-employed (not in- (Table 2). Annually, the employed population (9.9% of work-related injuries and 11.8% of cluded in BLS surveys). sustained 14.8 million injuries (data not non–work-related injuries) were coded as na- shown), of which 37.5% occurred at work ture unspecified (denoting that the respondent DISCUSSION (Table 2). Among all working-age men, provided insufficient description of the injury 34.5% of injury episodes occurred at work, to enable assignment of an International Clas- Nonfatal injuries to working-age adults are compared with only 20.5% of such episodes sification of Diseases, Ninth Revision, Clinical common, annually resulting in 12 episodes among working-age women, with higher pro- Modification [ICD-9-CM]32 medical diagnosis of medically treated injuries per 100 adults portions among employed people, 42.3% and code). aged 18 to 64 years. The redesigned NHIS 29.7%, respectively. When data were strati- The upper- and lower-body extremities ac- enabled us to determine that work-related fied by the age of the injured person, the per- counted for more than 60% of work-related injuries are also common (4.5 episodes per centage of episodes that were work related and non–work-related injuries (Table 3). The 100 employed working-age persons) and are ranged between 22% and 32%; when data upper extremities were involved in the largest a significant part of the total injury burden; were limited to employed persons, the percentage of work-related injuries, whereas such injuries accounted for almost 30% of figure was between 33% and 41%. When for non–work-related injuries, the regions injuries to working-age adults, a figure that data were stratified by age and gender, the most affected were the lower extremities. rose to almost 50% for 1 employed age proportion of injuries that were work related Over a third of the upper-extremity injuries group. Our ability to assess each injury re- was consistently higher for employed men (36%) were work related, compared with less corded in the NHIS for work relatedness than for employed women. than a quarter of the lower-extremity injuries overcomes many of the shortcomings inher- (23%). Fractures occurring at work were also ent in estimates of work-related injuries from Nature of Injury and Body Region more likely to affect an upper extremity, and other sources. Comparable data sources con- For both work-related and non–work- non–work-related fractures were more likely taining information on both work-related related nonfatal injuries, sprains and strains to affect a lower extremity (data not shown). and non–work-related injuries have not were most common, followed by open been available; this lack of information has wounds and fractures (Table 3). Crushing Time Lost From Work prevented an examination of the contribu- injuries, the presence of foreign bodies in an Among working-age adults, an estimated tion of work-related injuries to the total in- organ or body cavity, amputations, and burns annual 8.9 million injury episodes and an esti- jury burden.

July 2005, Vol 95, No. 7 | American Journal of Public Health Smith et al. | Peer Reviewed | Research and Practice | 1215  RESEARCH AND PRACTICE 

TABLE 2—Percentage of Injuries That Are Work Related Among Working-Age Adults the total NHIS work-related injury count for (Aged 18–64 Years), by Age, Race/Ethnicity, and Gender: National Health Interview Survey, the United States (3.01 million) was 1.8 times 1997–1999 higher than the BLS. This ratio is a function of both incomplete coverage of the US work- Work-Related Injuries, % force by the BLS and underreporting of Whole Working-Age Employed Working-Age work-related injuries in the BLS. a b Population (95% CI) Population (95% CI) Other population-based sources of Age, y occupational-injury data rely heavily on the 18–24 28.6 (25.5, 31.7) 41.2 (36.8, 45.6) recording of work relatedness in the medical 25–34 32.1 (29.2, 35.0) 38.9 (35.4, 42.4) record and may examine only 1 level of 35–44 27.0 (24.5, 29.5) 33.4 (30.4, 36.4) medical care. Surveys of occupational injuries 45–54 29.9 (26.6, 33.2) 37.9(33.8, 42.0) with emergency department records, for ex- 5,33–35 55–64 21.7 (17.8, 25.6) 38.0 (31.7, 44.3) ample, exclude medical care provided Gender and age in private physician offices and occupational Male 34.5 (32.5, 36.5) 42.3 (40.0, 44.6) health clinics. One 1988 study found that 18–24 31.7 (27.6, 35.8) 44.6 (39.0, 50.1) only 34% of all occupational injuries were 13 25–34 39.9 (36.0, 43.8) 43.8 (39.5, 48.1) treated in emergency departments. The re- 35–44 31.4 (28.1, 34.7) 37.1 (33.0, 41.1) porting of work relatedness in clinical record 5,36 45–54 37.0 (32.3, 41.7) 43.4 (37.8, 49.0) systems is also known to be inconsistent. 55–64 30.9 (24.4, 37.4) 49.2 (39.6, 58.7) Many data sources, such as hospital dis- Female 20.5 (18.5, 22.5) 29.7 (27.1, 32.3) charge databases, do not systematically rec- 18–24 22.4 (17.3, 27.5) 34.0 (26.5, 41.5) ord work relatedness. Health care providers 25–34 20.6 (16.9, 24.3) 29.4 (24.1, 34.7) can play an important role in improving 35–44 20.6 (17.3, 23.9) 27.5 (23.2, 31.8) these data sources for occupational injury 45–54 22.9 (18.6, 27.2) 31.5 (25.6, 37.3) surveillance by documenting work related- 55–64 13.7 (9.4, 18.0) 26.2 (18.3, 34.1) ness in medical charts. Race/ethnicity It is important to determine the work relat- White non-Hispanic 28.7 (27.1, 30.3) 36.9 (35.0, 38.8) edness of all injuries to adults, because the Black non-Hispanic 26.7(22.2, 31.2) 39.8 (33.2, 46.4) type of injury usually provides little indica- Hispanic 33.8 (30.1, 37.5) 45.1 (40.2, 50.0) tion of whether it occurred at work. The Other non-Hispanic 20.8 (14.5, 27.1) 27.9 (19.8, 36.0) characteristics of work-related and non– Total 28.6 (27.2, 30.0) 37.5 (35.7, 39.3) work-related injuries are similar, with little variation in distribution (Table 3). Although a Everyone in the population aged 18–64 years regardless of employment status (n=165 million). crushing injuries, the presence of foreign bod- bPeople who were reported to be employed at a job or business in the week before the interview (n=124 million). ies, and amputations are more likely to occur at work, many such episodes occur in nonwork settings. Documenting and reporting Work-related injuries are often regarded as of all injuries occurred among workers not work relatedness is just as important for in- very different from non–work-related injuries covered by the BLS survey. Direct compar- juries as it is for occupational diseases.37 in terms of etiology, surveillance, and preven- isons of our study with data from the BLS are Everyone involved in treating and preventing tion.12 For instance, separate workplace-based possible only for cases involving lost work injuries should also evaluate work hazards, data sources, such as BLS surveys or worker’s days, because these 2 data sources use differ- because many of the injuries occur at work, compensation data, are used to study injuries ent definitions of injury. Our NHIS-based especially among working-age men. at work.6,7,15–17 These sources are vulnerable estimate of occupational injury episodes oc- Several earlier studies used NHIS data to to a variety of underreporting effects.5–7,9,10,18,19 curring in US private industry that result in 1 examine injuries but did not compare all in- For example, the BLS Survey of Occupational or more days off work (2.36 million) was 1.4 juries with those occurring on the job. Warner Injuries and Illnesses is the primary source of times higher than the BLS’s 1998 estimate of and colleagues1 reported that during the first national nonfatal work-related injury data, but 1.649 million injuries that result in 1 or more year of the revised NHIS, 53% of injuries to it covers only private industry. It excludes days off work (i.e., excluding cases involving all age groups occurred among adults aged 22 self-employed persons, government workers, only restricted-duty days).16 This estimate to 64 years, but they provided limited data on and most workers on farms and in the infor- confirms findings of previous studies that doc- occupational injuries. Two recent studies ex- mal workforce (e.g., domestic help, undocu- umented underreporting among the industry amined sports and recreation injuries but did mented immigrants).15 ,16 In our study, 21.6% groups covered by the BLS.3–23 In addition, not consider whether any of these may have

1216 | Research and Practice | Peer Reviewed | Smith et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

TABLE 3—Nature of Injury and Affected Body Region in Work-Related and Non–Work- enhanced recall accuracy by defining injuries Related Injuries Among the Working-Age Population (Aged 18–64 Years): National Health as those that were severe enough to require 47 Interview Survey, 1997–1999 medical attention. It is possible that, in our study, home injuries were better recalled than Work-Related Injuries Non–Work-Related Injuries % of All Injuries That Were work-related injuries because the interview Annually, No. in 1000s (%) Annually, No. in 1000s (%) Work Related (95% CI) took place in the home. However, we found Nature of injury no studies documenting such an “interview Sprains and strains 1892 (34.3) 4536 (32.8) 29.4 (26.9, 31.9) context” effect. Open wounds 1170 (21.2) 2354 (17.0) 33.2 (29.9, 36.5) However, the NHIS definition of injury re- Fractures 498 (9.0) 2393 (17.3) 17.2 (14.3, 20.1) quires that medical attention be sought, and Contusions 360 (6.5) 1197 (8.6) 23.1 (18.6, 27.6) people injured at work may be more likely to Dislocations 270 (4.9) 485 (3.5) 35.7 (28.4, 43.0) seek medical attention than people injured at Injury to internal organs 120 (2.2) 354 (2.6) 25.3 (16.5, 34.1) home for such reasons as improved access, Burns 183 (3.3) 251 (1.8) 42.1 (31.7, 52.5) compensation, and legal liability. This defini- Superficial injuries 62 (1.1) 277 (2.0) 18.3 (9.9, 26.7) tion could result in the reporting of more Foreign bodies 164 (3.0) 97 (0.7) 62.9 (49.6, 76.2) work-related injuries than home injuries, but Crushing injuries 138 (2.5) 77 (0.6) 64.2 (50.1, 78.3) we were not able to address this issue in our Nerves 50 (0.9) 91 (0.7) 35.6 (20.5, 50.7) study. Similarly, differences among recent im- Amputation of limbs 57 (1.0) 43 (0.3) 56.9 (37.5, 76.3) migrants in access to care or immigrants’ re- Other specifieda ...b (. . .b) 61 (0.4) 35.9 (15.9, 55.9) luctance to identify an injury as work related Unspecified 548 (9.9) 1632 (11.8) 25.1 (21.2, 29.0) may also explain some of the observed varia- Body region tion by race/ethnicity. Other studies have in Lower extremity 1401 (25.3) 4716 (34.1) 22.9 (20.7, 25.1) fact found higher rates of occupational in- Upper extremity 2036 (36.7) 3641 (26.3) 35.9 (33.4, 38.4) juries among Hispanics and Blacks, but stud- Spine and back 949 (17.1) 2140 (15.5) 30.7 (27.4, 34.0) ies across a broad range of industries have Face and neck 307 (5.5) 955 (6.9) 24.3 (19.4, 29.2) been restricted to evaluation of fatalities.50,51 Skull and brain 154 (2.8) 455 (3.3) 25.3 (17.9, 32.7) A detailed examination of racial/ethnic differ- Thorax 71 (1.3) 499 (3.6) 12.5 (7.0, 18.0) ences would require careful adjustment for Abdomen and pelvis . . .b (. . .b) 54 (0.4) 38.3 (16.3, 60.3) factors such as differences in occupational Other body regions 430 (7.8) 1291 (9.3) 25.0 (20.5, 29.5) exposures and employment patterns. Total 5546 (100.0) 13 847 (100.0) 28.6 (27.2, 30.0) Poisonings were excluded from our analyses because the activity question (asking what a“Other specified” includes injuries to blood vessels, toxic effects, external causes, late effects of injuries, early complications of trauma, and medical/surgical complications. the person was doing when the injury oc- bEstimates are unreliable, with relative standard error of the estimate greater than 30%. curred), which we used to define work relatedness, was not asked of those who had been poisoned until 2000. Poisonings account for been work related.30,38 Prior to revision of the ity and impairment in the workplace.27,42,43 less than 0.6% of all occupational injuries NHIS, most analyses considered only broad The 1988 Occupational Health Supplement (from national emergency department visit classes of injury based on 4 combinations of to the NHIS included questions on work- data52), and their exclusion is thus unlikely to place and activity (i.e., home, motor vehicle, related injury, but lack of detail on the in- have significantly altered our study findings. work, and other place), but these classes were juries, small sample size, and problems associ- Owing to other questionnaire wording and not mutually exclusive. ated with using a 1-year recall period survey design modifications, we did not com- The revised NHIS now collects data on restricted the supplement’s utility in the study bine the 1997–1999 estimates with more both cause and place of injury.1,39 One study of workplace injuries.44,45 recent estimates because of concerns about that used NHIS data from 1983–1987 re- comparability.41 The activity question used to ported higher rates of occupational injuries Study Limitations define work relatedness allowed up to 2 re- than in our study, as did the BLS for that Any study relying on data from respon- sponses. For instance, a respondent could period (BLS rates declined in subsequent dents may suffer from recall bias, which can choose “working for a paid job” and “driving years).16 Comparisons with NHIS data from result in underestimation or overestimation of or riding in a motor vehicle” as the activity before 1997 are not valid because of the dif- injury rates.44,46–49 A 3-month recall period being engaged in when the injury occurred. ferent survey instruments used.40,41 Other was used to increase the number of episodes However, very few injury episodes had more studies have used special annual supplements of injury reported in our study.1 Because re- than 1 activity listed, which may have re- to the NHIS to examine the effects of disabil- call improves as injury severity increases, we sulted in underrepresentation of work relat-

July 2005, Vol 95, No. 7 | American Journal of Public Health Smith et al. | Peer Reviewed | Research and Practice | 1217  RESEARCH AND PRACTICE 

edness, especially for motor vehicle injuries. Contributors tional injury prevention: do they work? Inj Prev. 2001; Information on employment, including occu- G.S. Smith conceived the study, supervised all aspects 7(suppl 1):i3–i10. of its implementation, and led the writing of the article. pation and industry, was asked only of the 13. Centers for Disease Control and Prevention. Sur- H.M. Wellman conducted the analyses, and M. Warner veillance for nonfatal occupational injuries treated in sample adult, limiting our ability to conduct assisted in calculating sample variances. All authors hospital emergency departments—United States, 1996. detailed comparisons by occupation and in- helped to conceptualize ideas, interpret findings, and MMWR Morb Mortal Wkly Rep. 1998:47;302–306. review and revise drafts of the article. dustry. Asking direct questions about work re- 14 .Welch LS, Hunting K. Injury surveillance in con- latedness (e.g., whether the injury occurred struction: what is an “injury,” anyway? Am J Ind Med. 2003:44:191–196. during the course of paid work, or in what Acknowledgments This study was supported in part by the Liberty Mutual 15.BLS Handbook of Methods. Chapter 9, Occupa- industry and at what occupation the injured Research Institute for Safety’s Visiting Scholars program. tional Safety and Health Statistics. US Dept of Labor, person was working) would help to reduce We acknowledge the helpful comments on earlier Bureau of Labor Statistics, Washington, DC. Bulletin the limitations of many health data systems. drafts of the article provided by Dr David A. Lombardi 2490, April 1997. Available at: www.bls.gov/opub/ and Barbara S. Webster, as well as Patti Boelsen for edi- hom. Accessed April 20, 2005. torial assistance. Conclusions 16. Occupational Injuries and Illnesses: Counts, Rates, and Characteristics, 1998. Washington, DC: US Dept of Injuries to adults are an important public Labor, Bureau of Labor Statistics; July 2001. Bulletin Human Participant Protection health problem that affects not only the per- 2538. The Liberty Mutual Research Institute for Safety institu- 17. Courtney TK, Webster BS. Disabling occupational son sustaining the injury but also other tional review committee for the protection of human morbidity in the United States. An alternative way of household members dependent on the in- subjects approved the study. seeing the Bureau of Labor Statistics’ data. J Occup En- jured adult for support. Injuries on the job viron Med. 1999;41:60–69. are a significant part of this injury burden, References 18. Shannon HS, Lowe GS. How many injured work- comprising almost 30% of all medically 1. Warner M, Barnes P, Fingerhut L. Injury and ers do not file claims for workers’ compensation benefits? Am J Ind Med. 2002;42:467–473. treated injuries to adults aged 18 to 64 years. poisoning episodes and conditions; National Health Interview Survey, 1997. Vital Health Stat 10. 2000; For employed men, 42% of all injuries were 19. Oleinick A, Guire KE, Hawthorne VM, et al. Cur- No. 202. rent methods of estimating severity for occupational work related (49% among men aged 54–64 2. Boden LI, Galizzi M. Economic consequences of injuries and illnesses: data from the 1986 Michigan years). The NHIS is an important new source workplace injuries and illnesses: lost earnings and ben- Comprehensive Compensable Injury and Illness Data- of data with which to determine the work re- efit adequacy. Am J Ind Med. 1999;36:487–503. base. Am J Ind Med. 19 93;23:231–252. latedness of injuries and confirms previous 3. Leigh JP, Markowitz SB, Fahs M, et al. Occupa- 20. Smith GS, Veazie MA. Principles of prevention: the public health approach to preventing injuries in the studies in documenting serious underreport- tional injury and illness in the United States: estimates of costs, morbidity, and mortality. Arch Intern Med. workplace. In: ILO Encyclopedia of Occupational Health ing by traditional workplace-based reporting 19 97;157:1557–1568. and Safety. Geneva, Switzerland: International Labor systems.3–11 Population-based data, such as Office; 1998:56.26–56.30. Available at http://www. 4. Cormack RM, Chang YF, Smith GS. Estimating ilo.org/encyclopedia. Accessed April 20, 2005. those obtained from the NHIS, provide a deaths from industrial injury by capture-recapture: a 21. Tracking Occupational Injuries, Illnesses, and Haz- cautionary tale. Int J Epidemiol. 2000;29:1053–1059. model for improving occupational injury sur- ards: The NIOSH Surveillance Strategic Plan. Cincinnati, veillance and support proposals for community- 5. McCaig LF, Burt CW, Stussman BJ. A comparison Ohio: US Dept of Health and Human Services, Public of work-related injury visits and other injury visits to oriented approaches that look at work-related Health Service, Centers for Disease Control and Pre- emergency departments in the United States, vention, National Institute for Occupational Safety and 20–22 and non–work-related injuries together. 19 95–1996. J Occup Environ Med. 1998;40:870–875. Health; January 2001. NIOSH publication 2001-118. The fact that such a large proportion of in- 6. Azaroff LS, Levenstein C, Wegman DH. Occupa- Available at: http://www.cdc.gov/niosh/2001-118.html. juries to working-age adults are occupational tional injury and illness surveillance: conceptual filters Accessed June 13, 2003. in nature also reinforces the need to examine explain underreporting. Am J Public Health. 2002;92: 22. National Institute for Occupational Safety and 14 21–1429. Health. Traumatic Occupational Injury Research Needs workplace conditions in efforts to reduce the and Priorities: A Report by the NORA Traumatic Injury 7. Murphy PL, Sorock GS, Courtney TK, Webster BS, Team. Cincinnati, Ohio: US Dept of Health and Human impact of injuries on society. Leamon TB. Injury and illness in the American work- Services, Centers for Disease Control and Prevention; place: a comparison of data sources. Am J Ind Med. 1998. DHHS publication (NIOSH) 98–134. 1996:30:130–141. 23. Smith GS. Injury prevention: blurring the distinc- 8. Webb GR, Redman S, Wilkinson C, Sanson-Fisher tions between home and work. Inj Prev. 2003;9:3–5. About the Authors RW. Filtering effects in reporting work injuries. Accid Gordon S. Smith is with the Liberty Mutual Research In- Anal Prev. 1989;21:115–123. 24. Narayan KM, Boyle JP, Thompson TJ, Sorensen SW, stitute for Safety, Hopkinton, Massachusetts, and the Williamson DF. Lifetime risk for diabetes mellitus in Johns Hopkins Bloomberg School of Public Health, Balti- 9. Glazner JE, Borgerding J, Lowery JT, et al. Con- the United States. JAMA. 2003;290:1884–1890. struction injury rates may exceed national estimates: more, Maryland. Helen M. Wellman, Theodore K. Court- 25. Weller WE, Minkovitz CS, Anderson GF. Utiliza- ney, and Glenn S. Pransky are with the Liberty Mutual Re- evidence from the construction of Denver International Airport. Am J Ind Med. 19 98;34:105–112. tion of medical and health-related services among search Institute for Safety. Gary S. Sorock is with the Johns school-age children and adolescents with special health Hopkins Bloomberg School of Public Health. Margaret 10. Leigh JP, Marcin JP, Miller TR. An estimate of the care needs (1994 National Health Interview Survey on Warner and Lois A. Fingerhut are with the National Cen- US Government’s undercount of non-fatal occupational Disability [NHIS-D] Baseline Data). Pediatrics. 2003; ter for Health Statistics, Hyattsville, Maryland. injuries. J Occup Environ Med. 2004;46:10–18. 112(3 Pt 1):593–603. Request for reprint requests should be sent to Gordon S. 11.Pollack ES, Keimig DG. Counting Injuries and Ill- Smith, MB, ChB, MPH, Liberty Mutual Research Institute 26.Weil E, Wachterman M, McCarthy EP, et al. Obe- nesses in the Workplace: Proposals for a Better System. for Safety, Hopkinton, MA 01748 (e-mail: gordon. sity among adults with disabling conditions. JAMA. Washington, DC: National Academy Press; 1987. [email protected]). 2002;288:1265–1268. This article was accepted October 8, 2004. 12. Smith GS. Public health approaches to occupa- 27. Zwerling C, Whitten PS, Davis CS, Sprince NL.

1218 | Research and Practice | Peer Reviewed | Smith et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

Occupational injuries among workers with disabilities: 1994 to 1995. J Occup Environ Med. 2002;44: the National Health Interview Survey, 1985–1994 358–364. [published erratum appears in JAMA. 1998;279: 43. Zwerling C, Whitten PS, Sprince NL, et al. Work- 1350]. JAMA. 19 97;278:2163–2166. Second place accommodations for people with disabilities: Na- 28. Botman SL, Moore TF, Moriarity CL, et al. Design tional Health Interview Survey Disability Supplement, Edition and estimation for the National Health Interview Sur- 1994–1995. J Occup Environ Med. 2003;45:517–525. vey, 1995–2004. Vital Health Stat 2. 2000;130: 44. Landen DD, Hendricks S. Effect of recall on re- 14–19. porting of at-work injuries. Public Health Rep. 19 95; 29. National Center for Health Statistics. National 110:350–354. Health Interview Survey (NHIS), 1997, 1998, 1999: 45. Landen DD, Hendricks SA. Estimates from the NHIS survey description and dataset documentation. National Health Interview Survey on occupational in- Available at: http://www.cdc.gov/nchs/nhis.htm. Ac- jury among older workers in the United States. Scand cessed September 9, 2003. J Work Environ Health. 19 92;18(suppl 2):18–20. 30. Ni H, Barnes P, Hardy AM. Recreational injury Chronic Disease 46. Zwerling C, Sprince NL, Wallace RB, Davis CS, and its relation to socioeconomic status among school Whitten PS, Heeringa SG. Effect of recall period on aged children in the US. Inj Prev. 2002;8:60–65. Epidemiology and the reporting of occupational injuries among older 31. Shah H, Barnwell B, Bieler G. SUDAAN User’s workers in the Health and Retirement Study. Am J Ind Control Manual, Release 7.0. Research Triangle Park, NC: Re- Med. 19 95;28:583–590. Edited by Ross C. Brownson, PhD, search Triangle Institute, 1996. 47. Harel Y, Overpeck MD, Jones DH, et al. The ef- Patrick Remington, MD, MPH, and 32. Official ICD-9-CM Guidelines for Coding and Re- fects of recall on estimating annual nonfatal injury James R. Davis porting. Hyattsville, Md: National Center for Health rates for children and adolescents. Am J Public Health. Statistics; 1997. 1994;84:599–605. With this book, you’ll learn to: ❚ 33. Jackson LL. Non-fatal occupational injuries and ill- 48. Jenkins P, Earle-Richardson G, Slingerland DT, Locate critical background informa- nesses treated in hospital emergency departments in May J. Time dependent memory decay. Am J Ind Med. tion for developing appropriate the United States. Inj Prev. 2001;7(suppl 1):i21–i26. 2002;41:98–101. interventions ❚ Enhance your technical capacity for 34. Worker Health Chartbook, 2000 Nonfatal Injury. 49. Mock C, Acheampong F, Adjei S, Koepsell T. The delivering effective programs Cincinnati, Ohio: National Institute for Occupational effect of recall on estimation of incidence rates for in- ❚ Improve your knowledge about the Safety and Health; May 2002. DHHS (NIOSH) publi- jury in Ghana. Int J Epidemiol. 1999;28:750–755. cation 2002-119. Available at: www.cdc.gov/NIOSH. methods used in chronic disease 50. Richardson D, Loomis D, Bena J, Bailer AJ. Unin- Accessed January 3, 2003. epidemiology tentional fatal occupational injury rates in the US by ❚ Identify diseases and risk factors 35. Centers for Disease Control and Prevention. Na- race and Hispanic ethnicity: a comparison of the South ❚ Examine the underlying biological tional estimates of nonfatal injuries treated in hospital to the rest of the country. Am J Public Health. 2004; or physiological processes of disease emergency departments—United States, 2000. MMWR 94:1756–1761. ❚ Learn about high risk populations, Morb Mortal Wkly Rep. 2001;50:340–346. 51. Loomis D, Richardson D. Race and the risk of geographic variations, and trends 36. Fingar AR, Hopkins RS, Nelson, M. Work-related fatal injury at work. Am J Public Health. 1998;88: ❚ Plan, organize, and address preven- injuries in Athens County 1982–1986: a comparison 40–41. tion and control methods of emergency department and workers’ compensation 52. National Institute for Occupational Safety and data. J Occup Med. 19 92;34:779–787. ISBN 0-87553-237-3 Health. Work-Related Injury Statistics Query System 1998 ❚ 546 pages ❚ softcover 37.Freund E, Seligman PJ, Chorba TL, Safford SK, (Work-RISQS). Available at: http://www2a.cdc.gov/ $32.00 APHA Members Drachman JG, Hull HF. Mandatory reporting of occu- risqs/default.asp. Accessed June 13, 2003. $45.00 Non-members pational diseases by clinicians. JAMA. 1989;262: Plus shipping and handling 3041–3044. 38. Conn JM, Annest JL, Gilchrist J. Sports and recre- ORDER TODAY! ation related injury episodes in the US population, American Public Health Association 19 97–99. Inj Prev. 2003;9:117–123. Publication Sales 39. Smith GS, Sorock GS, Wellman HM, Courtney TK, Web: www.apha.org E-mail: [email protected] Pransky GS. Blurring the distinctions between home Tel: 888-320-APHA and work: similarities and differences in injury circum- FAX: 888-361-APHA stances. Inj Prev. In press. CHRN04J5 40.Wagener D, Winn D. Injuries in working populations: black-white differences. Am J Public Health. 19 91; 81:1408–1414. 41.Warner M, Heinen MA, Barnes P, Fingerhut L. Collection of injury and poisoning data in the National Health Interview Survey, 1997–2000. Paper presented at: 130th Annual Meeting of the American Public Health Association; November 12, 2002; Philadelphia, Pa. Available at: http://apha.confex.com/apha/130am/ techprogram/paper_46858.htm. Accessed February 11, 2004. 42. Zwerling C, Whitten PS, Sprince NL, et al. Work- force participation by persons with disabilities: the Na- tional Health Interview Survey Disability Supplement,

July 2005, Vol 95, No. 7 | American Journal of Public Health Smith et al. | Peer Reviewed | Research and Practice | 1219  RESEARCH AND PRACTICE 

Staffing and Worker Injury in Nursing Homes

| Alison M. Trinkoff, ScD, Meg Johantgen, PhD, Carles Muntaner, MD, PhD, and Rong Le, MS

The health care industry is one of the most Objectives. We examined the relationship between nursing home staffing lev- dangerous industries, ranking with construc- els and worker injury rates in 445 nursing homes in 3 states. tion, trucking, and meatpacking in nonfatal Methods. We obtained First Reports of Injury and workers’ compensation data 1 injury rates. Because of the growing elderly from 3 states (Ohio, West Virginia, and Maryland) for the year 2000. We then population, nursing homes have become linked these data to Medicare’s Online Survey, Certification and Reporting system major care providers to the elderly within to obtain nursing home staffing details and organizational descriptors. We used the health care industry.2 About 1.5 million ordinary least squares and log-transformed regression models to examine the as- elderly and disabled Americans reside in sociation between worker injury rate and nursing home staffing and organizational nursing homes, and nursing assistants pro- characteristics. vide the majority of their care.3 According Results. Total nursing hours per resident day were significantly associated with worker injury rates in nursing homes after we adjusted for organizational char- to the Bureau of Labor Statistics, the rate acteristics and state dummy variables (P=.0004). of worker injuries within nursing and per- Conclusions. Our findings suggest that nursing home staffing levels have an sonal care facilities is second among all important impact on worker health. These findings were supported for multiple 4 industries. Nursing homes are among the facilities across different states; therefore, policies and resources that increase top 10 industries for musculoskeletal prob- staffing levels in nursing homes are warranted. (Am J Public Health. 2005;95: lems, which is the major cause of worker 1220–1225. doi:10.2105/AJPH.2004.045070) absenteeism, workers’ compensation claims, and worker injury and illness.1,4–7 Higher rates of musculoskeletal injury have been care by the Institute of Medicine noted that adjusted for resident acuity, profit status, reported among nursing home workers there is empirical evidence that shows back nursing home size, and availability of nurse compared with rates among workers in injuries among nurses are associated with aide training. other occupations.8,9 staffing levels.17 Although the extent of Nursing home employees working in direct- worker injuries among resident care staff in METHODS care facilities perform many physically taxing nursing homes has been documented,18–20 activities, such as lifting heavy loads, working there have been few studies about the associ- Study Design in awkward postures, and transferring resi- ation between injuries and staffing. Our descriptive correlational design used dents.6,10–14 Additionally, manipulating the The occurrence of these injuries has im- administrative data to examine the association technology that supports patient care is physi- portant implications for staff retention. Owen between worker injuries and the organiza- cally straining. The increased worker injury and Garg21 found that 20% of nurses who tional characteristics of nursing homes in rates likely result from increased exposure to reported they had back pain said they had Medicare-approved facilities in 3 states: Ohio, hazardous conditions and diminished recov- made at least 1 job change in order to de- West Virginia, and Maryland. We used nurs- ery time between exposures.15 crease the number of nursing home resi- ing homes listed in Medicare’s Online Survey, Worker injuries in health care institutions dents that had to be lifted and transferred. Certification and Reporting (OSCAR) Year associated with staffing levels and skill mix Turnover among unlicensed personnel was 2000 database as the sampling frame for our have been previously examined. Because even higher,22 with 23% annual turnover re- study. health care institutions have been required ported among nursing assistants in one facil- to perform more efficiently, the resultant ity.23 In a statewide survey of nursing assis- Data Sources and Measures changes are lower staffing levels and higher tants, 30% reported they planned to quit We obtained staffing and organizational patient loads, both of which have been shown their jobs.24 descriptors from the OSCAR database. Our to increase worker injury. In a study of 12 We used an ecological design that was variables included number of beds, special hospitals in the Minneapolis–St Paul, Min- based on administrative data to examine the services, RN and other personnel staffing, nesota, area that used data from 1990 to association between staffing rates and worker type of nursing home ownership, and resident 1994, Shogren and Calkins16 found that injuries. To do this, we analyzed the associa- acuity. These data are routinely collected by when registered nurse (RN) positions were tion between staffing variables (total nursing the Centers for Medicare and Medicaid Ser- decreased by 9%, work-related illnesses and care hours per resident day) and adverse vices (CMS) to support the survey and certifi- injuries among nurses increased by 65%. worker outcomes (reported worker injuries) cation function and to monitor deficiencies Areview of the impact of staffing on health at the institutional level. Analyses were also and quality of care in US nursing homes that

1220 | Research and Practice | Peer Reviewed | Trinkoff et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

receive Medicare or Medicaid funds. Because 2-week period and multiplying by 70 work sented injuries to RNs, LPNs, and aides. To the OSCAR data are continually updated by hours for the period. We divided the total facilitate analysis and linkage of the data- overwriting the previous data, we purchased staffing hours by the total number of resi- bases, we assigned the CMS 6-digit provider historical data and documentation from the dents and then by 14 days in the reporting number for each facility to each worker in- Cowles Research Group.25 OSCAR data for period. In accordance with Harrington et jury record. Because facility names in the in- the 3 states in our sample were extracted al.,29 we included all full-time, part-time, and jury database were written as text with abbre- from this large database. contract positions for RNs; directors of nurs- viations, common names, and corporation We used First Report of Injury (FROI) ing were excluded. We included all LPNs and names, direct linkage to the OSCAR name databases for 2 of the states, Ohio and West licensed vocational nurses, and, for nursing was not always possible. In such cases, we Virginia, to measure worker injuries. This is aide staffing, we included all certified nursing used the CMS Nursing Home Compare data- believed to be the best source of injury re- assistants, nursing assistants in training, and base and other sources to match the nursing ports because the process of filing workers’ medication aides. home with its address. compensation claims has many systematic biases that can lead to suboptimal ascertain- File Construction Data Analysis ment of injury.26 On the other hand, because We applied exclusion criteria to remove Statistical analyses were performed with workers’ compensation claims tend to be filed nursing homes from the database based on SAS, version 8.2 (SAS Institute Inc, Cary, for only the most severe injuries,27 we felt it the recommendations by Harrington et al29: NC). We used descriptive statistics to exam- was also important to include worker injury (1) too small—those with fewer than 15 beds, ine the association between organizational data from these claims to examine the study (2) hospital based, (3) no RN hours—having characteristics and facility by state. We used question. Therefore, we also used workers’ 60 or more beds but no RN hours, (4) extra multivariate regression to identify the inde- compensation claims data from Maryland to RN hours—more than 12 RN hours per resi- pendent effect of these organizational char- calculate worker injuries. dent day, (5) few nursing staff hours—less acteristics—particularly staffing—on worker FROI and workers’ compensation data than 0.5 total nursing hours per resident day, injury rate. Because linked-facility sample were obtained from state agencies. The FROI and (6) excess nursing hours—more than 12 sizes for West Virginia and Maryland were data are comparable to Occupational Safety total nursing hours per resident day. We ex- small, we combined nursing home data from and Health Administration OSHA-200 log cluded skilled nursing facilities that had ex- all 3 states into 1 file (n = 445 linked facili- data, but they are obtainable at the state cess nursing hours to remove those facilities ties) to eliminate concerns about adequacy level for some states. Although injury data that function as acute care step-down facili- of power for these analyses. We included were obtained for individual workers, we ties and therefore do not reflect the staffing state dummy variables in regression models aggregated injury data to the organizational patterns for long-term care providers. For ex- because of systematic differences across the level for the analyses. All reported injuries ample, in West Virginia, after we applied states. Before we analyzed the association were included, regardless of type, although each of the exclusion criteria the original between worker injury rate and nursing the overwhelming majority of injuries were sample of 133 facilities decreased to 129 home characteristics (acuity index, total resi- musculoskeletal in origin (predominantly after deleting small facilities, to 103 after dents, percentage of Medicaid, location, back injuries). deleting hospital-based facilities, and to 102 profit status, aide training, and nursing hours Worker injury rates by skilled nursing fa- after deleting facilities with excess nursing per resident day), we screened the data for cility were calculated with formulas for injury staff hours; only 77% of the original facilities normality, missing values, outliers, and mul- incidence from the Bureau of Labor Statis- remained. ticollinearity. Acuity was measured with the tics’ Occupational Safety and Health Defini- Worker injury data required considerable Acuindex, which was developed as part of tions.28 To produce an overall rate, we aggre- cleanup. We culled the data received from the work on the CMS Minimum Data Set. gated the total number of nonfatal injuries the states to extract injuries that occurred in The Acuindex takes into account the propor- among RNs, licensed practical nurses (LPNs), nursing homes during 2000. For the data- tion of residents with activities of daily living and aides for each facility and divided the bases that included a Standard Industrial dependencies and the proportion requiring aggregate by the sum of the full-time equiva- Code (http://www.osha.gov/pls/imis/sic_ special treatments (e.g., suctioning, par- lents (FTE) for these 3 employee categories. manual.html), nursing homes were identified enteral feeding). Because this measure re- Multiplying the rates by 100 allows reporting with an 805 code (skilled nursing, intermedi- flects resident care burden, it also could in- per 100 FTE. ate nursing, and nursing and personal care fluence worker injury rates across facilities. Staffing variables were created with coding facilities). Upon review of these codes, we de- Therefore, we included acuity in our analy- rules designed by Harrington et al.29 FTE termined that some facilities were not nursing sis to control for variation in case mix. We data were reported for a 14-day period, and homes (assisted living, temporary staffing defined facilities with aide training as those we used the coding rules to convert staffing agency, system corporate office) and deleted facilities with an approved Nurse Aide Train- data to staffing hours per resident day by tak- them. Through further analysis and recoding, ing and Competency Evaluation Program. ing the total nursing staff FTEs reported for a we retained only those records that repre- Among predictors, the percentage of Medic-

July 2005, Vol 95, No. 7 | American Journal of Public Health Trinkoff et al. | Peer Reviewed | Research and Practice | 1221  RESEARCH AND PRACTICE 

TABLE 1—Nursing Home Characteristics in Ohio, Maryland, and West Virginia: Online by state for the linked sample (Tables 1 and 3). Survey, Certification, and Reporting System Database, 2000 Once again, there was little or no change in average staffing (nursing hours per resident Ohio Maryland West Virginia day) by state, before and after linkage. As a (n=778) (n=196) (n=102) result, we proceeded with the linked sample Nurse Staffing, mean (SD) for our analysis. RN hours per resident day 0.57 (0.27) 0.54 (0.40) 0.35 (0.17) Table 3 shows injury rates for the 3 states. LPN hours per resident day 0.76 (0.30) 0.62 (0.29) 0.80 (0.52) We found that overall injury rates in Mary- Aide hours per resident day 2.16 (0.59) 2.30 (0.66) 2.14 (0.62) land were the lowest, which was expected be- Total nursing hours per resident day 3.49 (0.80) 3.46 (0.94) 3.29 (0.88) cause the injury rates were calculated from Proportion RN hours out of total nursing hours per 0.16 (0.07) 0.15 (0.08) 0.11 (0.05) workers’ compensation claims rather than resident day from FROI data. However, the almost total Acuity index, mean (SD) 10.33 (1.22) 11.05 (1.35) 10.74 (1.08) lack of claims filed by nurse aides in Mary- Total residents, mean (SD) 87.76 (46.43) 113.29 (57.35) 89.48 (35.32) land was unexpected. We also found other Percentage residents enrolled in Medicaid, mean (SD) 68.12 (19.01) 59.68 (27.23) 75.84 (13.76) differences in rates by state, which likely re- Location, number (%) flected differences in injury reporting and Urban 584 (75.1) 174 (88.8) 35 (34.3) coding. Rural 194 (24.9) 22 (11.2) 67 (65.7) The results of the ordinary least squares re- Ownership, number (%) gression showed that total nurse hours per Profit 606 (77.9) 127 (64.8) 82 (80.4) resident day was significantly associated with Nonprofit/government 172 (22.1) 69 (35.2) 20 (19.6) worker injuries after we adjusted for acuity, Aide training available, number (%) 289 (37.2) 73 (37.2) 65 (63.7) profit status, aide training, total residents, and state (P=.0004). Our analysis showed that Note. RN=registered nurse; LPN=licensed practical nurse. 25% of the variance in worker injury was explained by the model (Table 4). For each additional hour increase in nursing care, injuries aid residents was highly correlated with the and location differed. Maryland facilities had were predicted to decrease by 2.4 per 100 acuity index in all 3 states. The percentage more residents on average, a lower propor- FTEs. The number of total residents also had of Medicaid was then dropped from further tion of for-profit facilities, and the lowest per- a significant negative effect: as size increased, multivariate analysis. Additionally, because centage of Medicaid residents. In West Vir- worker injuries decreased. To examine this injury rates were highly skewed, we modeled ginia, two thirds of the facilities were in rural further, we stratified nursing homes by num- the log of total injuries. areas, whereas in Ohio and Maryland, at least ber of residents and found that injury rates three fourths were in urban areas. The pro- were lower in homes where there were more RESULTS portion of facilities with aide training was residents, although staffing did not vary. Be- twice as high in West Virginia compared with cause of the apparent underreporting of in- We examined the characteristics of nursing Ohio and Maryland. jury rates among nurse aides in Maryland, homes in the 3 states for the entire sample Many worker injury records were listed we reran the regression models and excluded (Table 1). Ohio had the most nursing homes under a corporation and could not be linked Maryland. The results were the same (data (n=778) of the 3 states sampled, followed by to a specific facility; therefore, we used only not shown). Maryland (n=196) and West Virginia (n= the subset of nursing homes that could be Because the injury rates were skewed, we 102). Staffing levels were less distinct across linked in our analyses. To assess the impact regressed log-transformed injury rates on un- the states, with all 3 states having total nurs- of this reduced sample on the variables of transformed predictors. For the log-transformed ing hours per resident day (sum of RN, LPN, interest, we compared the characteristics of rates, all relationships and parameter esti- and nursing aide staffing) that ranged from an the skilled nursing facilities included in the mates that were significant in the untrans- average of 3.3 to 3.5 hours. However, West OSCAR sample with the sample that re- formed model remained significant, although Virginia had a far lower average number of mained after we linked the OSCAR and the elimination of positive skewness under- RN hours per resident day (0.35) compared worker injury data. As shown in Table 2, standably increased the predictability of the with Maryland (0.54) and Ohio (0.57), which there were no large differences in the aver- model to 38% of total variance (Table 4). indicated a lower skill mix. Our analysis of age staffing levels for the linked sample com- Each additional hour of nursing care de- the overall nursing home characteristics by pared with the OSCAR sample. The organi- creased the injury rate by nearly 16%. Thus, state showed that while acuity was similar zational characteristics also were similar. We for every unit increase in staffing (total hours across states, total residents, percentage of then compared the staffing means by state of nursing care), worker injury rates de- Medicaid residents, profit status, aide training, for the total OSCAR sample with the means creased by 2 per 100 FTEs.

1222 | Research and Practice | Peer Reviewed | Trinkoff et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

TABLE 2—Characteristics of Skilled Nursing Facilities Included in Total Sample vs. Linked because of turnover, floating among health Facilities: Online Survey, Certification, and Reporting (OSCAR) System Database, 2000 care workers, and multiple workers caring for a single resident. Examining injuries to Total OSCAR Sample Sample After Linking OSCAR all resident care workers who worked in for 3 States (n=1076) and Injury Data (n=445) long-term care facilities in 3 states (a total of Nurse staffing, mean (SD) more than 400 facilities) enhanced the gen- RN hours per resident day 0.54 (0.30) 0.54 (0.27) eralizability of the findings. LPN hours per resident day 0.74 (0.33) 0.76 (0.38) The consistency of the association between Aide hours per resident day 2.18 (0.60) 2.23 (0.61) staffing and injury across states and facilities Total nursing hours per resident day 3.46 (0.83) 3.52 (0.85) is noteworthy and supports the credibility of Proportion RN hours out of total nursing hours per 0.16 (0.07) 0.15 (0.06) the findings, although there are limitations to resident day our study. The ecological design did not allow Acuity index, mean (SD) 10.50 (1.26) 10.59 (1.28) us to make inferences about individual work- Total residents, mean (SD) 92.57 (48.66) 96.47 (51.82) ers.32 Missing data from the injury databases Percentage residents enrolled in Medicaid, mean (SD) 67.32 (20.78) 70.21 (19.17) hindered our ability to link across databases, Location, number (%) and the presence of missing data in certain Urban 793 (73.7) 333 (74.8) fields (e.g., occupation) also reduced the com- Rural 283 (26.3) 112 (25.2) pleteness of the data analysis. Despite these Ownership, number (%) limitations, comparison of descriptors from Profit 815 (75.7) 345 (77.5) nursing homes in the original OSCAR sam- Nonprofit/government 261 (24.3) 100 (22.5) pling frame with those in the linked frame Aide training available, number (%) 427 (39.7) 176 (39.6) showed surprisingly few differences. As expected, state variables were highly Note. RN=registered nurse; LPN=licensed practical nurse. significant, which underscores the importance of adjusting for them in a combined model. DISCUSSION proportion of variance in worker injury was A minimal number of injuries were reported explained by staffing. These data support by nurse aides in Maryland. Although the We combined facility-level data from nurs- smaller studies of single nursing homes and exact reason for this is unknown, the injured ing homes in 3 states and found that worker hospitals, which have also shown this associ- aides in Maryland most likely did not file injury rates were strongly associated with ation.17 , 3 0 Using an ecological framework workers’ compensation claims, probably staffing levels. Findings were consistent allowed us to circumvent some of the limita- owing to a lack of awareness; posting the law across the 3 states despite differences in tions of individual-level research on working in the workplace is not required. Also, the in- data collection, injury classifications, and conditions.31 Individual-level associations jury definitions can reduce the likelihood of reporting procedures. Additionally, a sizable may be difficult if not impossible to ascertain filing claims, e.g., back injuries in Maryland must have an acute onset to be claimable. Furthermore, claims in Maryland must be TABLE 3—Mean Nursing Home Staffing and Injury Rates for Linked Sample, by State: filed and signed by the injured employee—a Online Survey, Certification and Reporting System Database, 2000 provider or other party cannot initiate the Ohio (n=323) Maryland (n=76) West Virginia (n=45) claim—which may serve as a disincentive to Mean (SD) Mean (SD) Mean (SD) file among those who have insecure jobs. Nurse staffing Profit status and acuity were not signifi- RN hours per resident day 0.57 (0.27) 0.52 (0.29) 0.35 (0.19) cantly associated with worker injury when LPN hours per resident day 0.79 (0.33) 0.60 (0.30) 0.77 (0.68) state, size, and staffing were controlled. On 22 Aide hours per resident day 2.24 (0.66) 2.29 (0.39) 2.09 (0.47) the other hand, Banuszak-Holl and Hines Total nursing hours per resident day 3.60 (0.80) 3.41 (0.70) 3.20 (0.82) found that nursing turnover, a factor corre- Injuries (per 100 FTE) lated with injuries, was higher among for- RN 4.45 (8.63) 2.67 (8.06) 24.21 (22.78) profit nursing homes, which also tended to 33 LPN 5.55 (11.54) 18.21 (20.42) 10.66 (17.92) have lower staffing ratios. This was also Aide 16.62 (21.35) 0.15 (0.59) 45.01 (96.31) true for our sample. The lack of impact of aide training was unexpected, because train- Total 11.60 (11.94) 3.09 (2.46) 26.83 (18.50) ing has been associated with lower injury Note. RN=registered nurse; LPN=licensed practical nurse; FTE=full-time equivalent. rates,17 although we did not take into account the impact of staffing in these studies. Adjust-

July 2005, Vol 95, No. 7 | American Journal of Public Health Trinkoff et al. | Peer Reviewed | Research and Practice | 1223  RESEARCH AND PRACTICE 

TABLE 4—Ordinary Least Squares Regression (OLS) and Log-Transformed Models of Worker workers, it is imperative that we promote the Injury per 100 FTE on Staffing (nursing hours PRD), For-Profit ownership, Aide Training, health of this essential group of care provid- Acuity, Total Residents and State: 2000 (N=445) ers; they will be increasingly needed to care for an aging population. OLS Model Log-Transformed Model Worker injury/100FTE Worker Injury/100 FTE β SE(β) P (β) β SE(β) P (β) About the Authors Alison M. Trinkoff, Carles Muntaner, and Rong Le are Nursing hours PRD –2.39 0.67 0.0004 –0.16 0.047 0.001 with the Department of Family and Community Health, and Meg Johantgen is with the Department of Organiza- For-profit ownership 0.24 1.41 0.862 0.06 0.10 0.571 tional Systems and Adult Health, University of Maryland Aide training –0.16 1.18 0.892 –0.01 0.08 0.933 School of Nursing, Baltimore, Md. Carlos Muntaner is also Acuity (Acuindex) –0.01 0.45 0.986 –0.03 0.03 0.385 with the Center for Addiction and Mental Health, Univer- sity of Toronto, Toronto, Ontario, and the Institute of Work Total residents –0.03 0.01 0.003 –0.01 0.00 <.0001 and Health, Toronto. West Virginia (reference=Maryland) 21.82 2.27 <.0001 2.04 0.16 <.0001 Requests for reprints should be sent to Alison M. Ohio (reference=Maryland) 7.77 1.59 <.0001 1.03 0.11 <.0001 Trinkoff, ScD, University of Maryland School of Nursing, 655 W Lombard St, Rm 625, Baltimore, MD 21201 Note. FTE=full time equivalent; PRD=per resident day; SE=standard error. (email: [email protected]). OLS model R2 =0.25; log-transformed model R2 =0.38. This article was accepted September 9, 2004.

Contributors A.M. Trinkoff originated the study, supervised its imple- ing for differences in resident acuity removed FROI data. For example, workers may seek mentation, and led the writing. M. Johantgen created the database and directed the data analysis. C. Muntaner case mix as a potential source of confounding, injury care from their regular health provider assisted with the study, analysis of findings, and article which was important because nursing homes and fail to mention that the injury is work- preparation. R. Le assisted with the study and com- with more dependent residents may have related.26 Despite such limitations, FROIs are pleted the analyses. All authors originated ideas, interpreted findings, and reviewed drafts of the article. higher rates of worker injury. It is also possi- generally a more complete source of poten- ble that such homes have more assistive tially claimable injuries to health care workers Acknowledgments equipment that reduces injury risk to work- than workers’ compensation data.27 Ideally, Support was provided by the Agency for Healthcare Re- 22 ers. The current approaches to nursing the hours worked would exclude paid non- search and Quality (grant number R01 HS11990). home staffing are often made on the basis of work time, although we had no way to re- staff-to-resident ratio or hours per resident move this from our analysis. However, this Human Participant Protection day, with no accounting for differences in time is minimal among nurses, who often skip The project was reviewed by the institutional review acuity. This is reflected in our data, wherein breaks and lunches and perform uncompen- board of the University of Maryland and was deter- 36,37 mined to be exempt from the institutional review the acuity index from the OSCAR database sated overtime because of short staffing. board approval process according to DHHS 45 CFR was not correlated with staffing (r =0.03). Because injury data from the 3 states were 46.101.b (4). Ongoing research is being conducted to ex- treated similarly in our analysis, these distinc- amine the association between acuity and tions should not affect the ability to associate References staffing in nursing homes. injuries with staffing. 1. Bureau of Labor Statistics. Workplace Injuries and Illness in 1998. Available at: http://www.bls.gov/iif/ The OSCAR data also have limitations. Despite our successful attempt at using dif- oshwc/osh/os/osnr0009.pdf. Accessed October 10, The Centers for Medicare and Medicaid Ser- ferent worker injury databases from multiple 2000. vices performs edit checks on the OSCAR states in this analysis, there should be stan- 2. Kinsella K, Velkoff VA. An Aging World: 2001. data to identify errors. Straker34 compared dardization of both reported data and defini- Available at: http://www.census.gov/prod/2001pubs/ p95-01-1.pdf. Accessed July 9, 2002. 19 95 OSCAR data with data from the Ohio tions of worker injury.38 Outcomes data re- 3. American Association of Retired Persons (AARP). Department of Health to examine consistency ported at the facility level should be available Across the States: Profiles of Long-Term Care Systems. in several variables, including staffing. Staffing even when facilities manage injuries at the Washington, DC: Long-Term Care Policy Institute; correlations per patient day were 0.61, al- corporate level to allow for analysis of staffing 2000. though self-reports did not typically cover the and related outcomes. The National Quality 4. Bureau of Labor Statistics. Lost-Worktime Injuries same period reported as the OSCAR assess- Forum now recommends that staffing and and Illnesses: Characteristics and Resulting Time Away from Work, 1998. Available at: ftp://ftp.bls.gov/pub/ ment. Another study examined actual payroll skill mix be examined as performance mea- news.release/History/osh2.04202000.news. Accessed and found correlations less than 0.5 between sures for evaluating health care quality.39 October 10, 2000. the data reported in both the OSCAR and the Our study has shown that the impact of 5. Bureau of Labor Statistics. Lost-Worktime Injuries payroll,35 although these analyses had strict staffing is also important for worker health. and Illnesses: Characteristics and Resulting Time Away from Work, 1999. Available at: ftp://ftp.bls.gov/pub/ exclusion criteria. By improving staffing levels in nursing homes, news.release/history/osh2.03282001.news. Accessed As for the worker injury data, some in- both workers and residents will benefit. With May 17, 2001. juries will be missed even with the use of the impending shortage of long-term care 6. Collins JW, Owen BD. NIOSH research initiatives

1224 | Research and Practice | Peer Reviewed | Trinkoff et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

to prevent back injuries to nursing assistants, aides, and 23. Remsburg RE, Armacost KA, Bennett RG. Im- orderlies in nursing home. Am J Ind Med. 1996;29: proving nursing assistant turnover and stability rates 421–424. in a long-term care facility. Geriatr Nurs. 1999;20: 7. Personick ME. Nursing home aides experience in- 203–208. crease in serious injuries. Mon Labor Rev. 1990;113: 24.Parsons SK, Simmons WP, Penn K, Furlough M. 30–37. Determinants of satisfaction and turnover among nurs- 8. Meyer J, Muntaner C. Injuries in home health care ing assistants. The results of a statewide survey. J Geron- workers: An analysis of occupational morbidity from a tol Nurs. 2003;29:51–58. state compensation database. Am J Ind Med. 1999;35: 25. Cowles Research Group. Available at: http://www. 295–301. longtermcareinfo.com/crg/. Accessed May 17, 2001. 9. SHARP: Safety and Health Assessment and Re- 26. Boden LI. Workers’ compensation in the United search for Prevention Program. Work-Related Disorders States: high costs, low benefits. Annu Rev Public Health. of the Back and Upper Extremity in Washington State, 19 95;16:189–218. 1989–1996. Olympia, Wash: Washington State De- 27.Wickizer TM, Franklin G, Plaeger-Brockway R, partment of Labor and Industries; 1998. Technical Re- Mootz RD. Improving the quality of workers’ compen- Local Public Health port No.: 40-1-1997. sation health care delivery: the Washington state occu- Practice: 10. Allen A. On-the-job injury: a costly problem. J Post pational health services project. Milbank Q. 2001;79: Anesth Nurs.1990;5:367–368. 5–33. Trends & Models 11. Brulin C, Gerdle B, Granlund B, Hoog J, Knutson A, 28. Bureau of Labor Statistics. Occupational Safety and Sundelin G. Physical and psychological work-related By Glen P.Mays, PhD, MPH; Health Definitions. Available at: http://stats.bls.gov/iif/ C. Arden Miller, MD; and risk factors associated with musculoskeletal symptoms oshdef.htm. Accessed June 29, 2004. among home care personnel. Scand J Caring Sci. 1998; Paul K. Halverson, DrPH, MHSA 12:10 4–110. 29. Harrington C, Carrillo H, Thollaug S, Summers P, Wellin V. Nursing Facilities, Staffing, Residents, and Fa- 12. Marras WS, Davis KG, Kirking BC, Bertsche PK. cility Deficiencies, 1993 Through 1999. Available at: his book describes the varied spec- A comprehensive analysis of low-back disorder risk http://www.cms.hhs.gov/medicaid/services/nursfac99. trum of work done at the local public and spinal loading during the transferring and reposi- T pdf. Accessed October 12, 2000. health level, and how practitioners take tioning of patients using different techniques. Ergonom- the lead in social justice today. The wide ics. 1999;42:904–926. 30. Clarke SP, Sloane DM, Aiken LH. Effects of hospital staffing and organizational climate on needlestick array of public health department ap- 13. Sosnowitz B, Hriceniak J. Neonatal intensive care injuries to nurses. Am J Public Health. 2002;92: proaches, such as budgeting, staffing, ser- units can be hazardous to nurses’ health. J Perinatol. 1115 – 1119 . vices, involvement in personal health ser- 1998;8:253–257. vices, and their relationships with states 31.Petrisek AC, Mor V. Hospice in nursing homes. 14 .Trinkoff AM, Storr CL, Lipscomb JA. Physically is disclosed. Gerontologist. 1999;39:279–290. demanding work and inadequate sleep, pain medica- This book is an incredible resource tion use, and absenteeism in registered nurses. J Occup 32. Robinson WS. Ecological correlations and the for: local public health officers, adminis- Environ Med. 2001;43:355–363. behavior of individuals. Am Sociol Rev. 1950;15: trators, and state and local health plan- 351–357. 15. Bongers PM, De Winter CR, Kompier MAJ, Hilde- ners for use in their own local public brandt VH. Psychosocial factors at work and muscu- 33. Harrington C, Woolhandler S, Mullan J, Carrillo H, health practice. loskeletal disease. Scand J Work Environ Health. 19 93; Himmelstein DU. Does investor-ownership of nursing 19:2 97–312. homes compromise the quality of care. Int J Health ISBN 0-87553-243-8 ❚ ❚ 16. Shogren E, Calkins A. Findings of the Minnesota Serv. 2002;32:315–325. 2000 281 pages softcover nurses Association Research Project on Occupational Injury/ 34. Straker, J. Reliability of OSCAR Occupancy, Census $20.95 APHA Members Illness in Minnesota Between 1990–1994. St. Paul, and Staff Data: A Comparison with the Ohio Department $29.95 Nonmembers Minn: Minnesota Nurses Association;1997. of Health Annual Survey of Long-Term Care Facilities. plus shipping and handling 17.Wunderlich GS, Sloan FA, Davis CK, eds. Nursing Miami, Fla: Scripps Gerontology Center, Miami Univer- Staff in Hospitals and Nursing Homes. Washington, DC: sity; 1999. Technical Report Series 3-01. ORDER TODAY! National Academy Press; 1996. 35. Centers for Medicare and Medicaid Services. Re- American Public Health Association 18.Evidence-Based Practice Center (Oregon Health port to Congress: Appropriateness of Minimum Nurse Publication Sales and Science University). The Effect of Healthcare Working Staffing Ratios in Nursing Homes Phase II Final Report; Web: www.apha.org Conditions on Patient Safety. Evidence Report/Technology 2001 Dec. Contract No.: 500-95-0062-T.O.3. Available E-mail: [email protected] at: http://www.cms.hhs.gov/medicaid/reports/ Tel: 888-320-APHA Assessment No. 74. Rockville, Md: Agency for Health- FAX: 888-361-APHA care Research and Quality; 2003. AHRQ Publication rp1201home.asp. Accessed July 9, 2002. No. 03-E024. Contract number 290-97-0018. 36. Shulman B. The Betrayal of Work: How Low-Wage LP01J7 19. Muntaner C, Lynch J, Oates G. The social class Jobs Fail 30 Million Americans. New York, NY: WW determinants of income inequality and social cohesion. Norton & Co Inc; 2003. Int J Health Serv. 1999;29:699–732. 37. Gass TE, Vladeck BC. Nobody’s Home: Candid Re- 20.Myers D, Silverstein B, Nelson NA. Predictors of flections of a Nursing Home Aide. New York, NY: Cor- shoulder and back injuries in nursing home workers: a nell University Press; 2004. prospective study. Am J Ind Med. 2002;41:466–476. 38. Johantgen M, Trinkoff A, Gray-Siracusa K, 21.Owen BD, Garg A. Patient handling tasks per- Muntaner C, Nielsen K. Using state administrative data ceived to be most stressful by nursing assistants. In: to study nonfatal worker injuries: challenges and op- Mital A, editor. Advances in Industrial Ergonomics and portunities. J Safety Res. 2004;35:309–315. Safety 1. London, UK: Taylor & Francis; 1989: 39. National Quality Forum. National Voluntary Con- 775–781. sensus Standards for Nursing-Sensitive Performance Mea- 22. Banaszak-Holl J, Hines MA. Factors associated surement. Available at: http://www.qualityforum.org/ with nursing staff turnover. Gerontologist. 1996;36: nursing_sensitive_performance_measurement.html. 512–517. Accessed February 4, 2004.

July 2005, Vol 95, No. 7 | American Journal of Public Health Trinkoff et al. | Peer Reviewed | Research and Practice | 1225  RESEARCH AND PRACTICE 

Occupational Injury and Absence From Work Among African American, Hispanic, and Non-Hispanic White Workers in the National Longitudinal Survey of Youth

| Larkin L. Strong, MPH, and Frederick J. Zimmerman, PhD

Occupational injuries and illnesses represent a Objectives. We examined how race and ethnicity influence injury and illness risk significant source of morbidity and mortality and number of days of work missed as a result of injury or illness. in the United States. The Bureau of Labor Methods. We fit logistic regression and negative binomial regression models Statistics reported more than 5 million work- using generalized estimating equations with data from 1988 to 2000 on currently related injuries and illnesses in private indus- employed African American, Hispanic, and non-Hispanic White participants in try workplaces in 2001, resulting in an the National Longitudinal Survey of Youth. annual rate of 5.7 cases per 100 full-time Results. Occupational factors—having a blue-collar occupation, working full- workers.1 Historically, racial/ethnic minority time, having longer tenure, working 1 job versus 2, and working the late shift— workers have exhibited higher rates of work- were associated with increased odds of an occupational injury or illness. Al- related morbidity than have non-Hispanic though racial/ethnic minority workers were no more likely than Whites to report 2–4 an occupational injury or illness, they reported missing more days of work. Afri- White workers. can American and Hispanic men missed significantly more days of work than The extent to which such inequalities per- non-Hispanic White men, and African American women missed significantly sist today is less clear. Over time, notable more days of work than non-Hispanic White women. trends such as improved workplace safety, de- Conclusions. Factors associated with occupational health are multifaceted and creased rates of fatal and nonfatal occupa- complex. Our findings suggest that race/ethnicity influences the duration of work tional injuries and illnesses, the expansion of absence owing to injury or illness both indirectly (by influencing workers’ occu- affirmative action, and an increasingly diversi- pational characteristics) and directly (by acting independently of occupational fied workforce are likely to have influenced factors). (Am J Public Health. 2005;95:1226–1232. doi:10.2105/AJPH.2004.044396) both the types of occupations held by individuals of different races and ethnicities and occupational injury rates.1,5 The studies investi- proportion of non-Hispanic White male systems,4,7 emergency department hospital gating this topic have yielded inconsistent workers are found in the managerial and records,10 ,12 and employer records,6 all of findings; some suggest that minority workers professional specialties.14 which depend upon injured workers having are at greater risk of occupational injuries The extent to which factors other than sought medical care or filed for workers’ than their non-Hispanic White counter- occupation may also contribute to racial/ compensation. As occupational injuries and parts,4,6,7 whereas others have found compa- ethnic differences in work-related injuries illnesses are highly underreported,2 data rable rates of occupational injuries between and illnesses has not been adequately ex- drawn from these sources may not be fully these groups.8–10 plored in the public health literature. Myriad representative of all injured workers. Third, Racial and ethnic differences in occupa- factors in addition to occupation are be- the majority of studies have examined differ- tional injury and illness rates are often at- lieved to influence the risk of occupational ences between African American and White tributed to the greater likelihood of minority injury, among them education, age, work populations without distinguishing or includ- workers being employed in more hazardous experience, gender, shift worked, and union ing Hispanics.4,8,18 Finally, few studies have occupations.3,11,12 Minority workers, in turn, membership.9,15–17 utilized data from the past 15 years; injury are employed in hazardous occupations A number of methodological issues limits rates and job opportunities have changed largely because of lower educational attain- the conclusions of previous studies examin- since the late 1980s, so it is important to ment, greater difficulty finding work, and ing the relationship between race/ethnicity reexamine this issue in a current context. higher rates of poverty compared with non- and risk of occupational injury and illness. Additionally, few studies have considered Hispanic Whites, as well the effects of lin- First, the use of multiple data sources in the outcomes of occupational injuries and ill- gering racism.2,3,13 The occupational cate- some studies prevented linking demographic, nesses, such as the duration of absence from gories in which the greatest proportion of occupational, and injury data at the individ- work, a variable likely influenced by a num- African American and Hispanic men are ual level,4,7,10 limiting the inferences that can ber of factors. Increased absence from work is employed include machine operators, fabri- be drawn. Second, previous studies obtained costly to both society and the injured worker cators, and laborers, whereas the greatest data from state workers’ compensation claims and is associated with the worker’s reduced fu-

1226 | Research and Practice | Peer Reviewed | Strong and Zimmerman American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

ture participation in the workforce.19 , 2 0 Examin- (2) “Did the [injury or illness] cause you to age group comprises a large proportion of ing this outcome may prove useful in further miss one or more scheduled days of work, employed adults as well as injured or ill work- elucidating the relationship between occupa- not counting the day of the incident?” and ers.22 Data from 1991 were excluded owing tional health and race/ethnicity. We used (3) “Not counting the day of the incident, how to missing injury data for that year. data from a nationally representative sample many days was this?” Although individuals We performed χ2 tests and t tests to assess to (1) describe the occupational characteris- may have been injured more than once be- bivariate associations between race/ethnicity tics of non-Hispanic White, African American, tween interviews, data for this study pertain and demographic characteristics, occupational and Hispanic workers; (2) examine the rela- only to the most recent injury or illness for characteristics, and the mean number of tionship between race/ethnicity and self- each survey wave. workdays missed by injured respondents. reported injuries and illnesses that caused The independent variable of interest was Unadjusted injury rates per 100 FTE workers respondents to miss work; and (3) assess race/ethnicity, which was based on respon- were calculated for each demographic and oc- the relationship between race/ethnicity and dents’ self-identification with a primary ethnic cupational category taking into account sam- the number of workdays missed owing to the origin. Non-Hispanic White workers served as pling weights and the average number of jobs injury or illness. the reference group. Other independent and held by respondents in each category. control variables included age, marital status, To examine the association between the in- METHODS education level, gender, region of residence in dependent variables and injury status, we fit 1988, and, in each survey year, respondents’ multivariate logistic regression models that Data Source occupation, industry, job tenure, membership incorporated sample weights. We used nega- The source of our data is the National Lon- in a collective bargaining agreement, work tive binomial regression to examine the asso- gitudinal Survey of Youth (NLSY), administered shift, hourly rate of pay, full-time status, hav- ciation between the independent variables by the US Bureau of Labor Statistics. This sam- ing a second job, being paid hourly or salary, and the number of workdays missed among ple of 12686 youth was originally surveyed in time in the workforce, and a variable repre- injured respondents. Negative binomial re- 1979, when participants were 14 to 21 years senting each of the survey waves. gression assumes a Poisson-like distribution old, and participants have been resurveyed Respondents’ industry and occupation were and is used for nonnegative count data; how- every 1 to 2 years thereafter. The initial cohort classified according to 3-digit census codes.21 ever, unlike Poisson, the variance of the de- constituted a nationally representative sample Manufacturers and laborers served as the ref- pendent variable is assumed to be larger than of the US population, with African Americans erence groups for industry and occupation, the mean.23 We used generalized estimating and Hispanics oversampled. Survey weights respectively. Age represented respondents’ equations to account for possible correlations are available for drawing nationally representa- age at the first interview in 1979. Tenure was between repeated observations using the tive inferences.21 Attrition has been low; follow- coded as the number of weeks respondents “cluster(id)” option in Stata Statistical Software up rates ranged from 85% to 90% by 1998.21 had worked at a particular job. Respondents 7.0 (Stata Corp, College Station, Tex), which The sample is currently highly representative were considered a member of a collective specifies an independent working correlation of employed US adults in their 40s, not count- bargaining unit if they reported being in a structure and robust variance estimates. Inde- ing recent immigrants. We used data from union or if their wages were set by a collec- pendent and control variables were included 1988 through 2000, a total of 10 survey tive bargaining agreement. Hourly rate of pay based on a priori considerations of variables waves. was based on respondents’ self-report. The believed to be associated with injury status. shift during which respondents most often All analyses were stratified by gender. Variables worked was coded as “day,” “evening,” For each regression, we analyzed age, edu- Dependent variables included respondents’ “night,” or “other” (irregular, rotating, or split cation, tenure, hourly rate of pay, survey self-reported injury status since their prior in- shift), with the day shift as the reference year, and time in the workforce as continuous terview, and, among those injured, the num- group. Time in the workforce represented the variables. Survey year controlled for secular ber of workdays missed owing to an occupa- number of full-time equivalent (FTE) years trends and the maturation of the cohort over tional injury or illness. NLSY data managers worked since the last interview for each job. time. Respondents’ age at first interview con- coded the injuries and illnesses according to All occupational variables were collected for trolled for relative differences in age and a modified version of the International Classi- a second job where relevant and were consid- avoided collinearity with the survey year. We fication of Diseases, Ninth Revision.21 Here, an ered new observations. included age squared and education squared injury or illness represents an incident that to control for nonlinearity between the de- caused respondents to miss work for at least Statistical Analysis pendent variables and age and education. 1 day besides the day the injury occurred. In the data analysis we included individu- Tenure and hourly rate of pay were log- NLSY respondents were asked: (1) “Since als between the ages of 29 and 43 who re- transformed, as this was the most appropriate [date of last interview], have you had an in- ported currently working; who were African form for both variables. We analyzed mem- cident at any job we previously discussed American, Hispanic, or non-Hispanic White; bership in a collective bargaining agreement, that resulted in an injury or illness to you?” and who completed at least 3 interviews. This full-time status, having a second job, and

July 2005, Vol 95, No. 7 | American Journal of Public Health Strong and Zimmerman | Peer Reviewed | Research and Practice | 1227  RESEARCH AND PRACTICE 

being paid hourly or salary as dichotomous Compared with non-Hispanic White and Several occupational characteristics were variables. Region of residence, marital status, African American men, a significantly larger predictive of reporting a lost-worktime injury shift worked, industry, and occupation cate- proportion of Hispanic men reported an in- or illness. Overall, occupation was a much gories were represented by dummy variables. jury or illness that resulted in lost work-time stronger predictor than industry. Compared All analyses were performed with Stata Statis- (3.2% and 3.2% vs 4.7%, respectively). His- with laborers, male and female managers and tical Software 7.0. panic men who were injured also missed officials, sales, and professional/technical more days of work, on average, than other workers exhibited significantly lower odds of RESULTS injured men. Among women, African Ameri- reporting an injury or illness. Working the cans were significantly more likely to report a evening shift increased odds of a lost-work- Table 1 presents the demographic and oc- lost-worktime injury or illness compared with time injury or illness by nearly 150% in cupational characteristics reported in the Hispanic and non-Hispanic White women women, yet shift did not significantly affect 18 871 (52.8%) interviews with men and (3.0% vs. 2.6% and 2.2%, respectively), and men’s odds of an injury or illness. Surpris- 16 839 (47.2%) interviews with women. Ap- the mean number of workdays missed was ingly, longer tenure was associated with being proximately 57% of the interviews were with similarly high for African American and His- injured and with more missed work; for ex- non-Hispanic Whites, 15.1% were with His- panic women. ample, a 10% increase in tenure was associ- panics, and 27.9% were with African Ameri- Table 2 shows unadjusted lost-worktime in- ated with a 2.4% increase in the number of cans. These interviews represented 3267 men jury and illness rates per 100 FTE workers by workdays missed among women (to facilitate and 3027 women. In general, Hispanics and select demographic and occupational charac- interpretation, incidence rate ratios were con- African Americans reported significantly fewer teristics. African Americans exhibited the verted to the percentage change in the num- years of education than non-Hispanic Whites. highest rate of lost-worktime injuries and ill- ber of missed workdays associated with a We observed notable differences in occu- nesses among both men and women, followed 10% change in the exposure). Working full- pational characteristics by race/ethnicity. by Hispanics among men and non-Hispanic time significantly increased the odds of re- Non-Hispanic White men were employed Whites among women. In general, injury and porting an injury or illness for all respon- most often as craftsmen, managers/officials, illness rates fell with increasing years of edu- dents. Both male and female respondents and professional/technical workers, whereas cation and varied considerably by occupation, who had 2 jobs had significantly lower odds the 3 most commonly reported occupations with machine operators and laborers having of reporting an injury or illness than respon- among Hispanic and African American men the highest rates. Among men, the injury and dents reporting 1 job. were craftsmen, machine operators, and ser- illness rate was substantially higher for night- Considerably fewer occupational character- vice workers. Non-Hispanic White, Hispanic, time shift work than for other shifts, whereas, istics were significant predictors of the num- and African American women were em- for women, evening and nighttime shift work ber of workdays missed by injured respon- ployed most often as professional/technical exhibited similarly high rates. dents. Among men, managers and officials, workers, clerical workers, and service work- Table 3 shows the adjusted odds ratios service workers, and professional/technical ers, respectively. We also saw significant dif- (ORs) of reporting a lost-worktime injury or workers missed significantly fewer days of ferences in the shift worked, with African illness and adjusted incidence rate ratios work than did laborers. In addition, men who Americans more likely to work evening and (IRRs) of the number of missed workdays worked irregular or rotating shifts missed night shifts relative to other workers. Among among respondents. IRRs were obtained by 66% more days than men working the day both men and women, significantly more Af- exponentiating the coefficients produced by shift. Occupation was not significantly associ- rican Americans reported an hourly rate of negative binomial models. For both men and ated with the number of workdays missed pay in the lowest quintile compared with women, the odds of reporting an injury or by women. Rather, working the night shift, workers of other ethnicities. Racial/ethnic illness did not vary by race/ethnicity. How- longer tenure, and being a member of a col- differences in tenure were most pronounced ever, among injured male workers, Hispanics lective bargaining unit significantly increased among men in the shortest and longest tenure missed 64% more workdays than non-Hispanic the duration of work missed by women. categories; African American men were more Whites (approximately 18 more days at the likely than other men to report having sample mean), and African Americans missed DISCUSSION worked less than 6 months at their current 66% more workdays than non-Hispanic job and less likely to report having worked Whites (approximately 19 more days at the We analyzed the occupational characteris- more than 10 years. Non-Hispanic White sample mean). Among injured female work- tics and occupational health experiences of workers were the least likely to report being ers, African Americans missed 110% more currently employed non-Hispanic White, Af- a member of a collective bargaining unit. The workdays than non-Hispanic Whites (approxi- rican American, and Hispanic individuals majority of respondents reported working mately 26 more days at the sample mean). originally sampled for the NLSY in 1979. full-time, yet non-Hispanic White women No significant differences were observed be- Our analysis addresses several gaps in the lit- were significantly less likely than other tween Hispanic and non-Hispanic White fe- erature related to occupational injury and ill- women to report this. male workers. ness and race/ethnicity. First, we present esti-

1228 | Research and Practice | Peer Reviewed | Strong and Zimmerman American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

TABLE 1—Demographic and Work Characteristics of Current Workers, by Gender and Race/Ethnicity: National Longitudinal Survey of Youth, 1988–2000

Men, % Women, % Non-Hispanic Hispanic African American Non-Hispanic Hispanic African American White (n=10969) (n=2933) (n=4969) White (n=9408) (n=2454) (n=4977)

Education Some high school 8.7** 21.6 9.0 4.3** 12.6 6.0 High school graduate 40.9** 42.5 52.0 40.5** 36.1 37.9 Some college 20.7** 24.3 22.8 24.8** 33.4 37.4 College graduate 16.7** 7.0 10.8 15.8** 7.8 11.8 Postgraduate 13.0** 4.5 5.3 14.6** 10.1 7.1 Mean age in 1988, y 27.5 27.7 27.6 27.7 27.6 27.7 Marital status Married/living as married 69.5** 65.9 48.9 68.3** 63.7 38.7 Separated/widowed/divorced 12.6** 14.1 17.9 17.8** 20.7 29.3 Never married 17.9** 20.0 33.2 13.9** 15.6 31.9 Industry Agriculture 4.5** 6.4 2.7 1.8** 1.1 0.2 Mining/construction 13.0** 10.0 10.1 1.8** 0.8 0.5 Manufacturing 21.7 22.2 20.7 12.4** 10.1 12.8 Transportation/communications/utilities 8.4** 10.5 12.2 4.2** 5.9 4.6 Wholesale/retail trade 16.1 16.3 15.1 17.1** 13.4 12.6 Finance/insurance/real estate 4.7** 3.7 3.3 8.0** 8.1 6.6 Business/repair services 8.6** 9.3 10.4 6.3 5.1 6.3 Personal services 1.6** 2.0 2.8 5.3** 6.9 6.7 Entertainment/recreational services 1.8** 1.4 1.0 1.5** 0.5 1.3 Public administration 6.6 6.7 7.5 7.3** 7.9 11.2 Professional services 13.1** 11.4 14.1 34.2** 40.4 37.3 Occupation Manager/official 18.6** 12.8 8.5 14.2** 9.7 9.1 Sales 4.9** 3.0 3.5 4.8** 4.0 3.2 Clerical 5.3** 7.3 8.1 24.5** 32.4 27.8 Craftsman 20.0** 20.8 15.2 2.6** 1.6 1.8 Machine operator 12.8** 18.5 20.7 5.7** 7.1 10.9 Laborer 6.8** 10.0 12.8 1.8* 1.0 1.4 Farmer/manager/laborer 1.7** 2.5 0.2 0.5** 0.5 0 Service 10.9** 15.1 19.9 20.8** 20.9 29.0 Professional/technical 19.0** 10.0 11.3 25.2** 22.7 16.8 Sustained a lost-worktime injury or illness 3.2** 4.7 3.2 2.2** 2.6 3.0 Mean number of days missed due to injury or illness 28.1 38.0 30.9 24.0 45.5 45.2a Shift worked Day 70.5** 70.8 63.8 71.7** 78.9 69.5 Evening 5.3** 7.2 10.7 5.5** 4.6 8.6 Night 3.9** 5.7 6.7 2.8** 2.6 6.1 Other 20.3 16.4 18.9 20.0 14.0 15.8 Hourly rate of pay 1st quintile 11.8** 18.0 24.6 24.1** 24.5 31.7 2nd quintile 13.5** 18.1 22.0 20.1** 21.4 23.8 3rd quintile 19.5** 22.4 21.3 20.2** 21.6 18.6 4th quintile 24.2** 21.0 17.2 18.2** 18.8 16.0 5th quintile 31.0** 20.4 15.0 17.3** 13.8 10.0

Continued

July 2005, Vol 95, No. 7 | American Journal of Public Health Strong and Zimmerman | Peer Reviewed | Research and Practice | 1229  RESEARCH AND PRACTICE 

TABLE 1—Continued

Tenure <6 months 10.3** 11.5 14.4 10.9* 11.1 12.3 6 mo–23 mo 20.5** 22.4 25.9 24.8 24.5 24.4 24 mo–59 mo 24.9 24.1 26.1 26.3 26.9 24.9 60 mo–119 mo 23.8** 22.3 20.7 21.2 23.4 21.6 ≥10 y 20.5** 19.7 12.9 16.7** 14.1 16.8 Member of collective bargaining unit 19.0** 24.8 25.2 15.8** 18.8 24.3 Full-time employee 88.1** 87.8 85.8 66.7** 76.6 79.7 Have 2 jobs 8.7* 8.0 9.8 9.5** 7.5 9.4

Note.The sample size refers to the number of observations rather than to individual respondents. aP<.01 for 2-sample t test comparing mean number of days missed due to injury among female African American and non-Hispanic White workers. *P<.05 for χ2 tests comparing differences by race/ethnicity; **P<.01 for χ2 tests comparing differences by race/ethnicity.

mates using recent data from a nationally conceivable that such factors influenced their An interesting finding is that demographic representative data source, using self- likelihood of reporting an injury or illness in characteristics, such as age and education, reported occupational injury or illness seri- ways that differed from White workers. were not associated with reporting an occupa- ous enough to result in missed work as the Occupational characteristics proved to be tional injury or illness. The restricted age outcome variable. Second, our large sample the most important contributors to respon- range in our sample (29 to 43 years) likely enables us to control for secular trends in dents’ risk of an occupational injury or ill- reduced potential differences owing to age. work safety over this period. Third, our anal- ness. Such factors, principally occupation and Education, however, varied considerably. Pre- ysis controls for occupation and industry, as work experience, are recognized in the litera- vious studies have found education to be an well as shift, tenure, hourly rate of pay, and ture as important variables potentially medi- important determinant of occupational injury collective bargaining membership—all vari- ating the relationship between race/ethnicity even after control for occupation and work ables that have frequently been omitted from and risk of occupational injury.4,9,11,13,15 We experience.8,9,13 We observed decreasing past analyses. observed that having a white-collar occupa- injury and illness rates with increasing educa- Consistent with other studies, African tion and working 2 jobs significantly reduced tion; however, this relationship was not appar- American and Hispanic respondents were both male and female respondents’ odds of ent in our multivariate analyses, suggesting more likely than non-Hispanic Whites to be being injured, whereas working full-time and that differences in education may be reflected less educated, work in higher-risk occupa- having longer tenure significantly increased in occupational characteristics. tions, receive a lower wage rate, work later those odds. The finding that full-time status Race/ethnicity was a significant predictor shifts, and be members of a collective bar- and working 2 jobs acted in opposite direc- of the number of workdays missed in injured gaining unit.4,11,13,14 In general, we did not find tions suggests that on-the-job exposure time respondents. Among men, both Hispanic and race/ethnicity to be a significant predictor of plays a pivotal role in predicting occupational African American respondents missed signifi- reporting a lost-worktime injury or illness. Al- injuries and illnesses. These results may have cantly more days than non-Hispanic Whites, though unadjusted injury rates per 100 FTE important implications for the length of shifts and African American women missed signifi- workers were higher for African American in hazardous occupations. Additionally, possi- cantly more days than non-Hispanic White men and women and Hispanic men com- ble alternative explanations involving differ- women. This finding is supported by Johnson pared with non-Hispanic Whites, the adjusted ent work or benefits patterns for part-time and Ondrich,24 who found that, following an odds of reporting an injury or illness were not workers will need to be further explored in injury, Blacks and women were absent from significantly different. future research. work longer than White men. The implica- Reporting a lost-worktime injury or illness Evening shift work proved to be a source tions of this finding are not entirely clear, as a may have been influenced by a number of of occupational injury and illness for women combination of demographic, work-related, factors, such as perceptions of what constitutes but not men. Of 2 international studies inves- biomedical, economic, and psychosocial fac- an injury, financial pressure to remain at tigating the relationship between shift and oc- tors is believed to influence how long injured work, level of job autonomy, and workplace cupational injury, 1 observed similar injury workers are absent from work.19 , 2 0,25 discrimination against taking time off or filing rates on all shifts,16 and the other found night Although results of studies investigating for workers’ compensation. As minority work- shift work to significantly increase one’s risk this issue are varied, evidence suggests that ers in this study were more likely than Whites of an occupational injury.17 In our study, the older age, female gender, lower socioeco- to be less educated, work in blue-collar occu- evening shift increased women’s odds of in- nomic status, more severe injuries, poor base- pations, and receive a lower wage rate, it is jury by 150%. line health status, physically demanding work,

1230 | Research and Practice | Peer Reviewed | Strong and Zimmerman American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

TABLE 2—Unadjusted Lost-Worktime TABLE 3—ORs for Reporting a Lost-Worktime Injury or Illness Among All Respondents and Injury/Illness Rate, per 100 Full-Time IRRs of the Number of Days of Work Missed Among Injured Respondents Equivalent Workers Men Women Men Women OR (95% CI) IRR (95% CI) OR (95% CI) IRR (95% CI) Race/ethnicity n=18871 n=644 n=16839 n=413 Non-Hispanic White 2.7 2.5 Race/ethnicity Hispanic 4.0 2.3 African American 4.8 4.1 Non-Hispanic White Reference Reference Reference Reference Education Hispanic 1.13 (0.84, 1.53) 1.64 (1.02, 2.64) 0.81 (0.54, 1.21) 1.50 (0.83, 2.72) Some high school 4.8 5.3 African American 0.83 (0.64, 1.09) 1.66 (1.03, 2.68) 1.04 (0.77, 1.43) 2.10 (1.37, 3.22) High school graduate 4.2 2.9 Age, y 3.03 (0.89, 10.3) 1.04 (0.15, 7.48) 1.58 (0.32, 7.85) 0.58 (0.07, 5.17) Some college 2.9 3.1 Education, y 1.29 (0.94, 1.77) 0.99 (0.60, 1.63) 0.76 (0.50, 1.16) 0.91 (0.63, 1.31) College graduate 1.0 1.4 Occupation Postgraduate 2.9 3.1 Industry Laborer Reference Reference Reference Reference Agriculture 2.8 0.4 Manager/official 0.41 (0.25, 0.66) 0.44 (0.20, 0.99) 0.44 (0.21, 0.92) 1.27 (0.47, 3.41) Mining/construction 5.0 1.8 Sales 0.35 (0.15, 0.84) 0.87 (0.35, 2.11) 0.34 (0.13, 0.86) 2.64 (0.68, 10.32) Manufacturing 3.0 3.5 Clerical 0.66 (0.38, 1.15) 1.00 (0.43, 2.36) 0.24 (0.12, 0.45) 0.82 (0.31, 2.21) Transportation/communications/ 2.8 4.0 Craftsman 0.91 (0.65, 1.29) 0.89 (0.54, 1.45) 0.57 (0.24, 1.33) 0.58 (0.19, 1.78) utilities Machine operator 1.06 (0.73, 1.54) 1.30 (0.74, 2.30) 0.73 (0.38, 1.41) 2.41 (0.86, 6.74) Wholesale/retail trade 4.0 2.7 Finance/insurance/real estate 1.0 2.3 Service 0.79 (0.51, 1.22) 0.48 (0.24, 0.97) 0.55 (0.29, 1.04) 1.42 (0.54, 3.73) Business/repair services 2.0 2.2 Professional/technical 0.37 (0.21, 0.64) 0.24 (0.09, 0.62) 0.40 (0.20, 0.81) 0.85 (0.29, 2.48) Personal services 2.3 1.7 Shift worked Entertainment/recreational services 2.4 1.0 Day Reference Reference Reference Reference Public administration 3.1 1.7 Evening 0.70 (0.45, 1.09) 1.09 (0.49, 2.42) 2.49 (1.70, 3.66) 0.85 (0.49, 1.47) Professional services 1.8 2.8 Occupation Night 1.17 (0.75, 1.83) 0.65 (0.33, 1.26) 1.61 (0.97, 2.66) 2.28 (1.20, 4.31) Manager/official 0.9 2.1 Other 1.17 (0.88, 1.55) 1.66 (1.02, 2.68) 1.17 (0.79, 1.74) 0.63 (0.38, 1.06) Sales 0.7 2.8 Tenure (log) 1.19 (1.11, 1.30) 1.05 (0.89, 1.24) 1.22 (1.11, 1.35) 1.27 (1.02, 1.57) Clerical 2.5 1.5 Hourly rate of pay (log) 0.71 (0.58, 0.88) 1.18 (0.71, 1.98) 0.88 (0.69, 1.13) 1.35 (0.87, 2.09) Craftsman 4.1 3.9 Member of collective 1.24 (0.95, 1.60) 1.39 (0.92, 2.11) 1.31 (0.94, 1.82) 1.60 (1.04, 2.47) Machine operator 4.5 5.7 bargaining unit Laborer 7.3 4.5 Farmer/manager/laborer 4.2 3.5 Full-time employee 2.20 (1.25, 3.88) 0.82 (0.40, 1.66) 2.67 (1.71, 4.15) 0.91 0.54, 1.53) Service 0.8 2.4 Have 2 jobs 0.41 (0.19, 0.90) 0.87 (0.14, 5.35) 0.31 (0.13, 0.77) 1.57 (0.71, 3.49) Shift worked Wald statistic 308 81 182 267 Day 2.7 2.2 P value <.001 <.001 <.001 <.001 Evening 2.4 6.9 Night 9.1 6.4 Note.OR=odds ratio; IRR=incidence rate ratio; CI=confidence interval. In addition to the variables listed above, these Other 3.0 2.3 models were also adjusted for age squared, education squared, industry category, survey year, marital status, region of Tenure residence, and exposure time. <6 months 3.1 1.3 6 mo–23 mo 2.7 3.0 24 mo–59 mo 2.6 3.1 poor relations with coworkers, low job satis- ence on the number of days missed that is 60 mo–119 mo 4.0 2.7 faction, low job autonomy, and greater com- independent of these variables. Our results ≥10 y 2.6 2.4 pensation for lost wages are likely to prolong may indicate that the longer duration of work Member of collective bargaining unit 19 , 2 0,25 Yes 5.0 3.6 absence from work. None of these stud- missed by minority workers represents more No 2.5 2.5 ies examined the impact of race or ethnicity, severe injuries and illnesses, differential wage Full-time employee so it is difficult to determine the extent to replacement rates, poorer overall perceptions Yes 2.9 2.9 which the relationship between race/ethnicity of work, or greater psychosocial issues re- No 3.3 2.1 and duration of work absence may be con- lated to work, perhaps owing to discrimina- Have 2 jobs founded by these factors. tion, compared with non-Hispanic White Yes 3.2 1.3 No 2.9 2.6 In our study, the observation that signifi- workers. However, we cannot validate these cant differences by race/ethnicity persisted speculations. Note. Injury rates were calculated taking into account even after accounting for demographics, oc- Several limitations of this study should be population weights and the average number of jobs held by respondents in each category. cupation, and specific work-related factors noted. Because our data are based on self- suggests that race/ethnicity exerts an influ- report, our estimates of occupational injuries

July 2005, Vol 95, No. 7 | American Journal of Public Health Strong and Zimmerman | Peer Reviewed | Research and Practice | 1231  RESEARCH AND PRACTICE 

and illnesses and the duration of missed work Cancer Prevention Research Program, Fred Hutchinson nority workers and communities. Occup Med. 1999;14: may not be accurate. However, interviews Cancer Research Center, Seattle. Frederick J. Zimmerman 495–517. is with the Child Health Institute and the Department of 12. Anderson JT, Hunting KL, Welch LS. Injury and with workers would seem to provide a strong Health Services, School of Public Health and Community employment patterns among Hispanic construction opportunity for investigating racial and ethnic Medicine, University of Washington. workers. J Occup Environ Med. 2000;42:176–186. differences in occupational injury and illness Request for reprints should be sent to Larkin L. Strong, PO Box 19024, 1100 Fairview Ave N, M3-B232, Seattle, 13.Robinson JC. Racial inequality and the probability because these data are not reliant on report- WA 98109 (e-mail: [email protected]). of occupation-related injury or illness. Milbank Mem ing systems, claims data, or physician records. This article was accepted August 5, 2004. Fund Q Health Soc. 1984;62:567–590. Another limitation stems from the lack of data 14 . Gabor M, Houlder D, Carpio M, eds. Report on the on the number of injuries and illnesses sus- Contributors American Workforce. 5th ed. Washington, DC: Bureau tained between survey waves. Unfortunately L.L. Strong originated the study, conducted the analy- of Labor Statistics, US Department of Labor; 2001. ses, and led the writing of the article. F.J. Zimmerman the NLSY does not collect such data, so our 15. Leigh JP. Individual and job characteristics as pre- oversaw all aspects of the study and assisted with the dictors of industrial accidents. Accid Anal Prev. 1986; findings reflect information related only to the analyses and writing. Both authors interpreted the find- 18:2 0 9–216. most recent injury or illness. However, in ap- ings and reviewed drafts of the article. 16. Oginski A, Oginska H, Pokorski J, Kmita W, proximately 97% of the interviews, respondents Gozdziela R. Internal and external factors influencing reported that the most recent work-related Acknowledgments time-related injury risk in continuous shift work. Int J injury or illness was the most severe one they This research was supported by the National Institute Occup Saf Ergon. 2000;6:405–421. for Occupational Safety and Health (training grant T42 17. Smith L, Folkard S, Poole CJ. Increased injuries on had experienced since the previous interview, CCT010418). night shift. Lancet. 1994;344:1137–1139. indicating that for many people the most recent injury or illness was their only one. 18. Loomis D, Richardson D. Race and the risk of Human Participant Protection fatal injury at work. Am J Public Health. 1998;88: This study was exempted from review by the University 40–44. of Washington institutional review board because it Conclusions 19.Turner JA, Franklin G, Turk DC. Predictors of used only anonymous, preexisting data. Controlling for a variety of covariates chronic disability in injured workers: a systematic liter- and using a recent, large data set, we found ature synthesis. Am J Ind Med. 2000;38:707–722. References that racial and ethnic disparities exist in 20.Krause N, Frank JW, Dasinger LK, Sullivan TJ, 1. Bureau of Labor Statistics. Workplace Injuries and Sinclair SJ. Determinants of duration of disability and unadjusted occupational injury and illness Illnesses in 2001 [news release]. Washington, DC: return-to-work after work-related injury and illness: rates but that these disappeared when job Bureau of Labor Statistics, US Department of Labor; challenges for future research. Am J Ind Med. 2001;40: 2002. Available at: http://www.bls.gov/iif/oshwc/osh/ 464–484. and individual attributes were controlled. os/osnr0016.pdf. Accessed April 11, 2005. The social determinants of occupational in- 21. Center for Human Resource Research. NLSY79 2. Friedman-Jimenez G. Occupational disease among jury and illness in these data operate through User’s Guide 1999. Columbus: The Ohio State Univer- minority workers: a common and preventable public sity; 1999. racial differences in shift work, tenure, and health problem. AAOHN J. 1989;37:64–70,84–86. 22. Bureau of Labor Statistics. Lost-Worktime Injuries 3. Murray LR. Sick and tired of being sick and tired: occupation. Such results suggest that the poor and Illnesses: Characteristics and Resulting Days Away scientific evidence, methods, and research implications labor market outcomes for African Ameri- from Work, 2001 [news release]. Washington, DC: for racial and ethnic disparities in occupational health. Bureau of Labor Statistics, US Department of Labor; cans and Hispanics are not limited to wages, Am J Public Health. 2003;93:221–226. 2003. Available at: http://www.bls.gov/iif/oshwc/osh/ but rather extend to job safety features as 4. Robinson JC. Trends in racial inequality and expo- case/osnr0017.pdf. Accessed April 11, 2005. well. Moreover, differences in the duration sure to work-related hazards, 1968–1986. AAOHN J. 23. Cameron AC, Trivedi PK. Regression Analysis of 1989;37:56–63. of work absence suggest that the outcomes Count Data. New York, NY: Cambridge University of occupational injuries and illnesses vary by 5. Stout NA, Linn HI. Occupational injury prevention Press; 1998. Econometric Society Monographs Series; research: progress and priorities. Inj Prev. 2002;8 No. 30. race/ethnicity, even after accounting for oc- (suppl 4):iv9–iv14. cupational characteristics. 24. Johnson WG, Ondrich J. The duration of post- 6. Simpson CL, Severson RK. Risk of injury in Afri- injury absences from work. Rev Econ Stat. 1990;72: Although myriad factors are associated with can American hospital workers. J Occup Environ Med. 578–586. occupational health, the results of our study 2000;42:1035–1040. 25. Cheadle A, Franklin G, Wolfhagen C, et al. Fac- strongly suggest that race and ethnicity act both 7. Robinson JC. Exposure to occupational hazards tors influencing the duration of work-related disability: through occupational characteristics (e.g., occu- among Hispanics, Blacks, and non-Hispanic Whites in a population-based study of Washington State workers’ California. Am J Public Health. 1989;79:629–630. pation, shift work, tenure) and independently of compensation. Am J Public Health. 1994;84:19 0–196. 8. Wagener DK, Winn DW. Injuries in working pop- them. As long as work-related factors continue ulations: Black–White differences. Am J Public Health. to be racially inflected, racial equality in the 19 91;81:1408–1414. workplace demands improved workplace safety 9. Oh JH, Shin EH. Inequalities in nonfatal work in- efforts around dangerous occupations. jury: the significance of race, human capital, and occupations. Soc Sci Med. 2003;57:2173–2182. 10. Chen GX, Layne LA. Where African-American About the Authors women work and the nonfatal work-related injuries they experienced in the U.S. in 1996, compared to women of Larkin L. Strong is a doctoral candidate in the Department other races. Am J Ind Med. 1999;36(suppl 1):34–36. of Health Services, School of Public Health and Community Medicine, University of Washington, Seattle, and is with the 11.Frumkin H, Walker ED, Friedman-Jimenez G. Mi-

1232 | Research and Practice | Peer Reviewed | Strong and Zimmerman American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

Predictors of Work-Related Repetitive Strain Injuries in a Population Cohort

| Donald C. Cole, MD, MSc, Selahadin Ibrahim, MSc, and Harry S. Shannon, PhD

Repetitive strain injury (RSI) and cumulative Objectives. We assessed predictors of work-related repetitive strain injuries trauma disorder (CTD) are 2 of several terms using data from 4 waves of the Canadian National Population Health Survey. used to describe a group of activity-related Methods. Participants were 2806 working adults who completed an abbrevi- soft-tissue injuries that include tendonitis, ated version of the Job Content Questionnaire in 1994–1995 and did not expe- forearm myalgia, and nerve entrapment syn- rience repetitive strain injuries prior to 2000–2001. Potential previous wave pre- dromes, among other conditions.1 The area dictors of work-related repetitive strain injuries were modeled via multivariate affected by RSI and CTD may be only the logistic regression. upper limbs, may include the neck and upper Results. Female gender (odds ratio [OR] = 1.98; 95% confidence interval back,2 or may encompass the lower back and [CI]=1.24, 3.18), some college or university education (OR=1.98; 95% CI=1.06, lower limbs as well. RSIs and CTDs represent 3.70), job insecurity (OR=1.76; 95% CI=1.07, 2.91), high physical exertion levels (OR = 2.00; 95% CI = 1.29, 3.12), and high levels of psychological demands an important burden arising from both sport-3 (OR=1.61; 95% CI=1.02, 2.52) were all positively associated with work-related and work-related4 activity, the latter generat- repetitive strain injuries, whereas working less than 30 hours per week exhibited ing considerable societal and employer costs a negative association with such injuries (OR=0.2; 95% CI=0.1, 0.7). 5 through workers’ compensation claims. Conclusions. Modifiable job characteristics are important predictors of work- Performing biomechanical/physical tasks, related repetitive strain injuries. (Am J Public Health. 2005;95:1233–1237. organization of work associated with tasks, doi:10.2105/AJPH.2004.048777) and psychosocial stressors at work are among the causes of work-related RSI and CTD. These diverse causes have lead many work those residing in institutions and in some re- of the NPHS (the wave closest in time to the and health researchers to prefer the term mote areas, were excluded.14 In 1994–1995, 2000–2001 wave). Data included gender, work-related musculoskeletal disorders.6–8 the household response rate was 88.7%. age (consolidated into 4 groups), education Other researchers have studied the work- Within each household, 1 adult was asked (aggregated into 3 groups), and household related exposures that contribute to RSI and detailed questions, and the response rate was income, categorized on the basis of respon- CTD. A systematic review9 conducted in 96.1% at this individual level (i.e., 17626 dents’ household income after adjustment for 19 97 indicated a preponderance of cross- individuals, each from a different household). household size according to low income cut- sectional studies focusing on work-related ex- Of the 1994–1995 participants, 17276 off criteria (Table 1).17 posures among specific populations.10 A few (98.0%) were eligible for reinterviews in Comorbidity variables included chronic longitudinal general population studies in 1996–1997. Among those individuals, conditions and activity limitations. Respon- which exposures are assessed in advance of 93.6% responded in 19 9 6–1997, 88.9% dents were asked whether they had long- outcomes have been carried out.11,12 In our responded in 1998–1999, and 84.8% re- term conditions (those having persisted or study, inclusion of job characteristic questions sponded in 2000–2001.15 expected to persist 6 months or more) diag- in the first wave of a Canadian national longi- nosed by a health professional. Two of the tudinal survey and questions about RSI in Population possible responses were “arthritis” and subsequent waves allowed us to analyze pre- We focused on respondents who, in “back problems excluding arthritis.” We dictors of work-related RSI. 1994–1995, were aged 18 through 64 years, refer to the latter simply as “back prob- had paid employment (including those who lems.” The questions on activity limitations METHODS were self-employed), and responded to an ab- tapped long-term physical or mental condi- breviated set of items derived from the Job tions, or health problems that limited the Surveys Content Questionnaire (JCQ).16 Exclusions kind or amount of activity in which the re- We used data from the Canadian National and listwise deletion of relevant missing vari- spondent could engage, as well as long-term Population Health Survey (NPHS).13 In this ables resulted in an analysis sample of 2806. disabilities or handicaps. A derived variable survey, a complex stratified, multistage sample was used to capture an affirmative response design was used to randomly select approxi- Measures to any of these questions. To assess the mately 20000 households; people living on Independent variables. We obtained sociode- contribution of such chronic conditions Indian reserves and military bases, along with mographic data from the 1998–1999 wave and activity limitations to RSIs reported in

July 2005, Vol 95, No. 7 | American Journal of Public Health Cole et al. | Peer Reviewed | Research and Practice | 1233  RESEARCH AND PRACTICE 

TABLE 1—Reports of New Work-Related Repetitive Strain Injury (RSI) in 2000–2001, by Respondents were asked whether they had Sociodemographic, Comorbidity, and Lifestyle Variables: Canadian National Population worked for pay in the past 12 months; up to Health Survey 6 jobs were recorded in each wave. Using the Canadian National Occupational Classifica- New Work-Related tion,18 we grouped respondents’ main jobs No. of Participantsa RSI Reported, No. (%) Pb into the following categories: unskilled, semi- Sociodemographic variables (1998–1999) skilled with secondary training, semiskilled Gender .001 with college or university training, and skilled/ Female 1178 85 (7.2) supervisor/semiprofessional/professional/ Male 1628 70 (4.3) management. Type of employment in Age, y .675 1998–1999 was dichotomized into full time 18–34 721 39 (5.4) (30 hours or more worked per week) and 35–44 1029 61 (5.9) part time (less than 30 hours worked per 45–54 754 43 (5.7) week). Responses to the JCQ item focusing >54 302 12 (4.1) on job insecurity were dichotomized into high Marital status .225 and low. A “high” level of insecurity was de- Marriedc 2035 106 (5.2) fined as respondents’ agreement or strong Other 771 49 (6.4) Household income .066 agreement that they had an insecure job. Low 77 4 (5.5) Job characteristic variables from the Middle/high 2612 139 (5.3) 1994–1995 wave, assessed via JCQ items, Missing variable 117 12 (10.4) were rated on a 5-point Likert scale (“strongly 19 Education .002 agree” to “strongly disagree”). Physical exer- Secondary school or less 773 33 (4.3) tion, measured via a single item, was dichoto- Some college/university 778 62 (8.0) mized into high and low. A “high” level of College/university 1255 60 (4.8) exertion was defined as agreeing or strongly Comorbidity variables (1994–1995, 1996–1997, or 1998–1999) agreeing that one’s level of physical exertion Back problem .000 at work was high. Decision latitude or “con- Yes 640 53 (8.3) trol” was measured with a 5-item scale (learn No 2166 102 (4.7) new things, job requires high level of skill, Arthritis .069 freedom to decide how to do the job, work Yes 303 24 (7.8) not repetitious, and a lot to say about what No 2503 131 (5.3) happens in the job). Psychological demands Activity limitation .002 were measured with a 2-item scale (hectic Yes 503 42 (8.4) No 2303 113 (4.9) work and conflicting demands). Social support Lifestyle variables (1998–1999) at work was measured via a 3-item scale (su- Leisure time physical activity .781 pervisor helpful in getting work done, Engage 2608 145 (5.57) coworkers helpful in getting work done, peo- Do not engage 198 10 (5.1) ple you work with not hostile). We dichoto- Daily smoker .117 mized responses for psychological demands Yes 644 44 (6.8) (top third of distribution vs other) as well as No 2162 112 (5.2) for decision latitude and social support (bottom third of distribution vs other). aAdjusted survey weights adding up to a sample size of 2806 were used. bχ2 test for difference in proportions. Dependent variable. Data on RSIs were cIncludes common-law marriages. gathered via a question asking respondents whether they had experienced “injuries caused by overuse or repeating the same 2000–2001, we created a combined vari- tivity and smoking. Participation in leisure movement frequently in the previous year able comprising any report of chronic condi- time physical activity was based on at least 1 (for example, carpal tunnel syndrome, tennis tion or activity limitation across 3 previous positive response indicating that respondents elbow, or tendonitis) . . . which were serious waves (1994–1995, 1996–1997, 1998– had engaged in “physical activities not re- enough to limit your normal activities.” Those 1999). lated to work” in the past 3 months. Smoking who responded “yes” were considered to Data on 2 lifestyle variables were available was dichotomized into daily smoking and have an RSI and then were asked about its from the 1998–1999 wave: leisure time ac- less frequent/no smoking. cause (“Was this injury the result of doing

1234 | Research and Practice | Peer Reviewed | Cole et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

something [in various settings, including] at TABLE 2—Reports of New Work-Related Repetitive Strain Injury (RSI) in 2000–2001, by work?”). As our dichotomous dependent vari- Occupation, Availability of Work Variables, and Job Characteristics From Previous Waves: able, work-related RSI was determined on the Canadian National Population Health Survey basis of participants’ reports of incurring an RSI in 2000–2001 as a result of engaging in New Work-Related RSI in No. of Participantsa 2000–2001, No.a (Row %) Pb a task at work. To ensure incident rather than prevalent cases, we excluded those who re- Occupation (1998–1999) .365 ported RSIs attributed to any setting in Management, skilled/university 744 35 (4.7) 1996–1997 or 1998–1999. Semiskilled/secondary, postsecondary 1799 108 (6.0) Unskilled 263 12 (4.9) Analysis Work availability variables As recommended by Statistics Canada, ad- Employment status (1998–1999) .024 justed survey weights were used in all analy- Full time 2504 147 (5.9) ses. Proportions of work-related RSI in Part time 302 8 (2.7) 2000–2001 across previous wave variables Job insecurity (1994–1995) .000 were calculated and compared via a χ2 test. High 560 49 (8.7) Univariate logistic regression analyses were Low 2246 106 (4.7) then conducted for each previous wave vari- Job characteristics (1994–1995) able as a predictor of incident RSI in Decision latitude .084 2000–2001. High 1868 93 (5.0) Because of the possibility of the NPHS mul- Low 938 62 (6.6) tistage sample design resulting in correlated Psychological demands .001 observations, the usual methods of regression High 1035 78 (7.5) would lead to underestimations of the stan- Low 1771 78 (4.4) dard errors of the estimated coefficients.20 To Social support at work .155 allow for this sampling design effect, we used High 1744 88 (5.1) bootstrap weights provided by Statistics Can- Low 1062 67 (6.3) ada.21 All independent variables were entered Physical exertion .000 into an initial multivariate logistic model. Vari- High 1194 90 (7.6) ables that were not statistically significant at Low 1612 65 (4.0) the .05 level (P>.05) were removed sequen- aAdjusted survey weights adding up to a sample size of 2806 participants were used. tially; their omission did not substantially bχ2 test of presence of RSI by row categories for each variable. alter (10% change)22 the estimated coefficients for the main predictors of interest (work availability and job characteristic variables). or lower leg (n=10; 6.6%), and (7) neck or mained significant predictors (Table 3). Life- Hosmer–Lemeshow and deviance goodness- other (n=10; 6.5%). Interestingly, participants style variables remained unimportant in terms of-fit statistics were used in assessing models.23 with some college or university education were of subsequent work-related RSI. Both work All statistical analyses were performed with more likely than participants in other educa- availability variables remained predictors, with the SAS version 8.1 software package (SAS tion groups to experience an RSI (Table 1). Al- high job insecurity elevating risk of experienc- Institute Inc, Cary, NC).24 though comorbidity was infrequent, reports of ing an RSI (odds ratio [OR]=1.76; 95% confi- new work-related RSIs in 2000–2001 were dence interval [CI] = 1.07, 2.91) and part-time RESULTS more prevalent among those with an activity work decreasing risk (OR=0.33; 95% CI= limitation and those with a back problem re- 0.13, 0.88). Finally, high levels of both psycho- Overall, 155 of the cohort participants ported in previous waves. logical demands (OR=1.61; 95% CI=1.02, (5.5%) reported a new work-related RSI in A smaller proportion of part-time workers 2.52) and physical exertion (OR=2.00; 95% 2000–2001. Significantly different percent- (less than 30 hours per week) than full-time CI=1.29, 3.12) remained important predictors ages of women and men experienced such in- workers reported a work-related RSI (Table 2). of development of future work-related RSI. juries (7.2% vs 4.3%: P=.0011). The primary High levels of job insecurity, psychological parts of the body affected were (1) wrist or demands, and physical exertion were all asso- DISCUSSION hand (n=57; 36.8%), (2) shoulder or upper ciated with greater proportions of subsequent arm (n=31; 20.0%), (3) elbow or lower arm work-related RSIs. Our analyses show that sociodemographic, (n=23; 14.8%), (4) lower back (n=17; In multivariate analyses, the demographic work availability, and job factors predict new 11.0%), (5) upper back (n=7; 4.5 %), (6) knee variables gender and educational level re- occurrences of work-related RSIs in a popula-

July 2005, Vol 95, No. 7 | American Journal of Public Health Cole et al. | Peer Reviewed | Research and Practice | 1235  RESEARCH AND PRACTICE 

TABLE 3—Predictive Factors for Occurrence of New Work-Related RSI in 2000–2001 in education as high relative to others employed Final Multivariate Logistic Regression Model: Canadian National Population Health Survey in the same work were more likely to report back pain.30 Estimated Odds Ratio Our finding in the bivariate analyses—that (95% Confidence Interval) those reporting back problems or activity lim- Female vs male (reference) 1.98 (1.24, 3.18)a itations in previous waves had a greater risk Marriedb vs other (reference) 0.80 (0.48, 1.34) of developing new work-related RSIs in Age, y 2000–2001—extends the finding that a pre- 18–34 (reference) 1.00 vious back injury is an important risk factor 35–44 1.24 (0.68, 2.26) for sustaining a subsequent back injury8 to 45–54 1.28 (0.66, 2.48) the broader outcome of work-related RSI. Co- >54 1.10 (0.49, 2.48) hort analyses examining extension and recur- Education rence of musculoskeletal injuries in working Secondary school or less (reference) 1.00 populations31 must control for previous injury Some college/university 1.98 (1.06, 3.70)a history. College/university 1.21 (0.66,2.24) Back problemc vs none (reference) 1.53 (0.92, 2.55) Limitations Activity limitation vs none (reference) 1.58 (0.88, 2.55) Unfortunately, the NPHS data set did not Leisure time physical activity, yes vs no (reference) 0.88 (0.30, 2.58) include questions on RSI in 1994–1995 Daily smoker, yes vs no (reference) 1.20 (0.69, 2.09) or questions on job characteristics after Occupation 1994–1995, and thus we were not able to Unskilled 0.94 (0.40, 2.19) explicitly model ongoing exposure–symptom Semiskilled 1.25 (0.75, 2.07) relationships across waves. Furthermore, self- Professional/semiprofessional/skilled/supervisor (reference) 1.00 reports of RSIs occurring at work may either Part time vs full time (reference) 0.33 (0.13, 0.88)a overrepresent or underrepresent “work- High job insecurity vs low (reference) 1.76 (1.07, 2.91)a related conditions” categorized according to High psychological demands vs low (reference) 1.61 (1.02, 2.52)a the World Health Organization definition,32 High physical exertion vs low (reference) 2.00 (1.29, 3.12)a which encompasses both work-caused and work-aggravated conditions. Because RSIs in- Note. Hosmer–Lemeshow goodness-of-fit χ2 =10.4, df=8,P=.2355; deviance χ2 =870, df=1667, P=1.000. a Confidence interval does not include 1. volve a sufficient number of causal factors, as b Includes common-law marriages. exemplified by the significant predictors in c Refers to nonarthritic back pain. our study, specific attribution is exceedingly difficult even with additional work and clinical information.1 In addition, the NPHS in- tion cohort involving broad coverage of occu- work, have been characterized as involving a volved the use of an abbreviated set of JCQ pations and an excellent response rate.19 high risk for CTD or work-related RSI, even items, impairing traditional measurement These findings extend those of cross-sectional though recognition of workers’ compensation properties.25 However, accounting for the analyses of the NPHS population25 and that of claims has been proportionately higher heterogeneous nature of the items, particu- its successor, the Canadian Community Health among men in some jurisdictions.29 Interest- larly those making up the decision latitude Survey.26 Also, they are consistent with similar ingly, participants with some formal college and psychological demands scales, we research showing the importance of job-related or university education were more likely than would argue that the scale variables should physical risk factors,6,9 job psychosocial fac- those in other education groups to report ex- be considered composites, in which case trators,7,27 and their combination11,12 , 2 8 in regard periencing work-related RSIs. Such individu- ditional psychometric performance applied to work-related RSIs, CTDs, and work-related als may be more aware of the relationship to latent variables is less crucial.33 musculoskeletal disorders among working between work, demanding conditions, and populations. The protective nature of part- having an RSI, and therefore may be more Implications for Prevention time (relative to full-time) employment is most likely to attribute their RSI to a work-related Our findings add to the existing empirical likely attributable to decreased exposures to activity. Alternatively, they may be more con- evidence of the role of both physical and psy- such risk factors and greater time for rest of cerned than individuals in other education chosocial work factors in the onset of RSI, the affected body areas. groups that their current jobs do not match CTD, and work-related musculoskeletal disor- Women were more likely than men to de- their job expectations, as observed in the ders.34 Such evidence should lead workplaces velop a new work-related RSI. Women’s jobs, Ontario Universities Back Pain Study, in and governments to consider the wide range particularly in micro-assembly and office which individuals who perceived their level of of preventive measures documented by re-

1236 | Research and Practice | Peer Reviewed | Cole et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

searchers into and practitioners of ergonom- 2. RSI. The Hague, the Netherlands: Health Council 21.Yeo D, Mantel H, Liu TP. Bootstrap variance esti- ics.35 Manuals have already been prepared to of the Netherlands; 2000. Publication 2000/22E. mation for the National Population Health Survey. Paper presented at: Annual Meeting of the American guide workplaces in the implementation of 3. Almeida SA, Williams KM, Shaffer RA, et al. Epi- demiological patterns of musculoskeletal injuries and Statistical Association, August 1999, Baltimore, Md. ergonomic programs that can reduce the physical training. Med Sci Sports Exerc. 1999;31: 22. Greenland S. Modeling and variable selection in physical demands of work.36 1176 – 1182. epidemiologic analysis. Am J Public Health. 1989;79: Similarly, there is considerable evidence 4. Beaton DE, Cole DC, Manno M, et al. Describing 340–349. from the organizational behavior and indus- the burden of upper extremity musculoskeletal disor- 23. Hosmer DW, Lemeshow S. Applied Logistic Re- ders in newspaper workers: what difference do case gression. New York, NY: John Wiley & Sons Inc; 1989. trial psychology literature that work reorgani- definitions make? J Occup Rehabil. 2000;10:39–53. 24. SAS/STAT User’s Guide, Version 6. 4th ed. Cary, 37 zation can reduce psychological demands. 5. Association of Workers’ Compensation Boards of NC: SAS Institute Inc; 2000. Yet, even among those with RSIs or CTDs, Canada. Canadian work injuries and diseases. Available 25. Cole DC, Ibrahim SA, Shannon HS, et al. Work at: http://www.awcbc.org/english/NWISP_Stats.htm. secondary prevention activities designed to correlates of back problems and activity restriction due Accessed April 4, 2005. to musculoskeletal disorders in the Canadian National correct risky conditions may be very restricted 6. Kuorinka I, Forcier, L, eds. Work-Related Muscu- Population Health Survey (NPHS): 1994/95 data. 38 in terms of coverage. Etiological research loskeletal Disorders (WMSDs): A Reference Book for Pre- Occup Environ Med. 2001;58:728–734. findings, such as those presented here, must vention. London, England: Taylor & Francis; 1995. 26. Tjepkema M. Repetitive strain injury. Health Rep. be complemented with rigorous evidence on 7. Houtman ILD, Bongers PM, Smulders P, et al. Psy- 2003;14(4):11–30. chosocial stressors at work and musculoskeletal prob- the effectiveness of workplace and regulatory 27.Papageorgiou AC, Macfarlane GJ, Thomas E, et al. lems. Scand J Work Environ Health. 1994;20:139–145. Psychosocial factors in the workplace: do they predict interventions to persuade company and union 8. Frank JW, Kerr MS, Brooker AS, et al. Disability new episodes of low back pain? Evidence from the officials and government policymakers to resulting from occupational low back pain part I: What South Manchester Back Pain Study. Spine. 19 97;22: do we know about primary prevention? A review of reduce the burden of work-related RSI, CTD, 1137–1142. the scientific evidence on prevention before disability 28.Warren N, Dillon C, Morse T, et al. Biomechani- and work-related musculoskeletal disorders. begins. Spine. 1996;21:2908–2917. cal, psychosocial and organizational risk factors for 9. Bernard BP, ed. Musculoskeletal Disorders and WRMSD: population-based estimates from the Con- Workplace Factors: A Critical Review of Epidemiologic necticut Upper-Extremity Surveillance Project (CUSP). Evidence for Work-Related Musculoskeletal Disorders of J Occup Health Psychol. 2000;5:164–181. About the Authors the Neck, Upper Extremity and Low Back. Cincinnati, 29. Chung J, Cole DC, Clarke J. Women, work and Donald C. Cole and Selahadin Ibrahim are with the Insti- Ohio: National Institute for Occupational Safety and injury. In: Sullivan TJ, ed. Injury and the New World of tute for Work & Health and the Department of Public Health; 1997. Work. Vancouver, British Columbia, Canada: University Health Sciences, Faculty of Medicine, University of 10.Polanyi MFD, Cole DC, Beaton DE, et al. Upper of British Columbia Press; 2000:69–90. Toronto, Toronto, Ontario. Harry S. Shannon is with the limb work-related musculoskeletal disorders among Program in Occupational Health and Environmental Med- 30.Kerr MS, Frank JW, Shannon HS, et al. Bio- newspaper employees: cross-sectional survey results. mechanical and psychosocial risk factors for low back icine, Faculty of Health Sciences, McMaster University, Am J Ind Med. 19 97;32:620–628. Hamilton, Ontario, and the Institute for Work & Health. pain at work. Am J Public Health. 2001;91:1069–1075. Requests for reprints should be sent to Donald C. Cole, 11. Macfarlane GJ, Hunt IM, Silman AJ. Role of 31. McGorry RW, Webster BS, Snook SH, et al. The MD,MSc, Institute for Work & Health, 481 University mechanical and psychosocial factors in the onset of relations between pain intensity, disability and the Ave, Suite 800, Toronto, Ontario M5G 2E9, Canada forearm pain: prospective population based study. episodic nature of chronic and recurrent low back pain. (e-mail: [email protected]). BMJ. 2000;321:676–679. Spine. 2000;25:834–841. 12.Feville H, Jensen C, Burr H. Risk factors for neck- This article was accepted December 11, 2004. 32. Identification and Control of Work-Related Diseases. shoulder and wrist-hand symptoms in a 5-year follow-up Geneva, Switzerland: World Health Organization; study of 3,990 employees in Denmark. Int Arch Occup Contributors 19 85. WHO Technical Report Series 714. Environ Health. 2002;75:243–251. All of the authors interpreted the findings and reviewed 33. Bollen K, Lennox R. Conventional wisdom on 13. National Population Health Survey, 1994–95. Ot- multiple drafts of the article. D.C. Cole led the writing, measurement: a structural equation perspective. Psy- tawa, Ontario, Canada: Statistics Canada; 1995. with the participation of S. Ibrahim and H.S. Shannon. chol Bull. 19 91;111:305–314. 14 .Tambay JL, Catlin G. Sample design of the Na- 34. Institute of Medicine. Musculoskeletal Disorders and tional Population Health Survey. Health Rep. 19 95;7(1): Acknowledgments the Workplace: Low Back and Upper Extremities. Wash- 29–38. Initial work on the Canadian National Population ington, DC: National Academy Press; 2001:301–329. 15. National Population Health Survey, Cycle 4 (2000/ Health Survey (NPHS) was funded by the National 35. Muggleton JM, Allen R, Chappell PH. Hand and 01), Health Component, Longitudinal Documentation. Health Research Development Program (grant 6606- arm injuries associated with repetitive manual work in Ottawa, Ontario, Canada: Statistics Canada; 2002. 6406). Our analysis was sponsored by the Institute for industry: a review of disorders, risk factors and preven- Work & Health, an independent, not-for-profit research 16. Karasek R, Brisson C, Kawakami N, et al. The Job tive measures. Ergonomics. 1999;42:714–739. Content Questionnaire (JCQ): an instrument for interna- organization that receives support from the Ontario 36. Cohen AL, Gjessing CC, Fine LJ, et al. Elements of tionally comparative assessments of psychosocial job Workplace Safety and Insurance Board. Ergonomics Programs: A Primer Based on Workplace Eval- characteristics. J Occup Health Psychol. 1998;3:322–355. uations of Musculoskeletal Disorders. Cincinnati, Ohio: Na- Human Participant Protection 17. Statistics Canada. Re-Basing Low Income Cut-Offs tional Institute for Occupational Safety and Health; 1997. to 1978. Ottawa, Ontario, Canada: Minister of Supplies We used the NPHS data sets with previous approval of 37.Parker SK, Jackson PR, Sprigg CA, et al. Organiza- and Services; 1983. Statistics Canada for the proposed analyses, in keeping tional Interventions to Reduce the Impact of Poor Work with Canadian federally mandated health research re- 18. National Occupational Classification: NOC Training Design. Norwich, England: Health and Safety Execu- quirements for such data. NPHS participants consented Tutorial. Ottawa, Ontario, Canada: Human Resources tive; 1998. to share their survey responses with third-party re- Development Canada; 2001. 38.Keogh JP, Gucer PW, Gordon JL, et al. Patterns searchers through Statistics Canada. 19.Wilkins K, Beaudet MP. Work stress and health. and predictors of employer risk-reduction activities Health Rep. 1998;10(3):47–62. (ERRAs) in response to a work-related upper extremity References 20. Lee ES, Forthofer RN, Lorimer RJ. Analyzing cumulative trauma disorder (UECTD): reports from 1. Yassi A. Repetitive strain injuries. Lancet. 19 97; Complex Survey Data. Beverly Hills, Calif: Sage Publica- workers’ compensation claimants. Am J Ind Med. 2000; 349:943–947. tions; 1989. 38:489–497.

July 2005, Vol 95, No. 7 | American Journal of Public Health Cole et al. | Peer Reviewed | Research and Practice | 1237  RESEARCH AND PRACTICE 

Outcome Evaluation of a Public Health Approach to Suicide Prevention in an American Indian Tribal Nation

| Philip A. May, PhD, Patricia Serna, MSW, Lance Hurt, PhD, and Lemyra M. DeBruyn, PhD

Recently, a number of scholars and officials Objectives. We evaluated the efficacy of 15 years of a public health–oriented have called for population-based public suicidal-behavior prevention program among youths living on an American health strategies for preventing suicide in en- Indian reservation. tire communities and racial/ethnic groups. Methods. All suicides, suicide attempts, and suicidal gestures were monitored. Noting that approaches to suicide prevention Age-specific analyses over time were used to assess outcomes. are often narrowly based on psychiatric and Results. Both descriptive and linear regression analyses indicated that a sub- individual dynamics, the surgeon general,1 stantial drop occurred in suicidal gestures and attempts. Suicide deaths neither the US Public Health Service,2 and the Insti- declined significantly nor increased, although the total number of self-destructive tute of Medicine3 have voiced the need for a acts declined by 73% (P=.001). new, comprehensive understanding of suici- Conclusions. Data from this community-based approach document a remarkable downward trend—measured by both magnitude and temporal trends in the dal behavior on which to base suicide preven- specifically targeted age cohorts—in suicidal acts. The sequential decrease in age- tion efforts. Three recent publications explain specific rates of suicide attempts and gestures is indicative of the program’s suc- the rationale for a population-based approach cess. (Am J Public Health. 2005;95:1238–1244. doi:10.2105/AJPH.2004.040410) to preventing suicidal behavior,3 calling suicide prevention a national imperative3 for public health intervention.4,5 and homicide were the second and third program comprised 284 and 291 youths American Indian/Alaska Native (AIAN) sui- leading causes of death, respectively, among aged 10 to 14 years, 261 and 293 aged 15 cide studies and prevention programs have American Indians/Alaska Natives aged to 19 years, and 224 and 245 aged 20 to frequently been approached from a public 15–24 years, with unintentional injury and 24 years23 (IHS Office of Program Planning health perspective,6–22 but detailed evaluation adverse effects of trauma combined being and Evaluation, unpublished program data, of such programs among American Indians/ the leading cause of death.25,26 In the same July 23, 2003). Eighty percent of those aged Alaska Natives has been quite rare in practice time period, suicide rates among American 16 years or older were unemployed, with and in the literature.18 We describe an out- Indians/Alaska Natives aged 15–24 years some seasonal employment from firefighting, come evaluation of a suicide prevention pro- were consistently 2 times higher than those hunting and fishing guiding, and agriculture gram among the Western Athabaskan Tribal among all American 15- to 24-year-olds.3,25 (P. Serna, unpublished data, 1991). Nation (a pseudonym used to protect the iden- However, suicide rates vary from tribe to In 1989, the IHS provided initial funding tity of this tribe—one of several Athabaskan tribe,20,21 with higher rates in Western for a small model Adolescent Suicide Preven- tribes in the southwestern United States—and states.25 They also vary over time in individ- tion Project. The IHS designed and imple- its reservation) of New Mexico. This program ual AIAN communities.15 ,22–23 mented the project in collaboration with tribal is part of an ongoing effort to evaluate AIAN In 1988, the annual rate of suicide and officials in 1990 and funded it through 1994. suicide trends, potential causes of these trends, suicide attempts combined for the Western During that time, it was the only active proj- and the efficacy of prevention programs in Athabaskan Tribal Nation was 15 times ect in a New Mexico community specifically New Mexico19,23,24 (P. Serna, unpublished higher than the US rate and 5 times higher designed for suicide intervention and preven- data, 1991; Western Athabaskan Tribal Na- than rates for other New Mexico American tion (P. Serna, unpublished data, 1991; West- tion, unpublished data, 2003). Indians/Alaska Natives.23 From 1957 through ern Athabaskan Tribal Nation, unpublished An increase in suicidal activity among 2000, suicide death rates on this reservation data, 2003; IHS Office of Program Planning AIAN adolescents and young adults on this fluctuated in 6- to 8-year cycles.20–23 The and Evaluation, unpublished program data, reservation in 1988 prompted the tribal Western Athabaskan Tribal Nation is rural July 23, 2003; K. Gaylord, MA, injury pre- council and community and the Indian Health and isolated, with 90% of the population liv- vention manager, Office of Injury Prevention, Service (IHS) to work together to establish an ing on the reservation, primarily in a single New Mexico State Department of Health, adolescent suicide prevention program on the town. In 1990 and 2000, the total popula- oral communication, July 8, 2003). With reservation in 1990.17 Chronic fluctuations in tion served by the IHS on this reservation grants from various sources, the project de- suicidal behavior, including clusters and up- numbered 2639 and 3047, respectively. In veloped into a more broadly focused pro- surges, are not unusual in some AIAN com- 1990 and 2000, respectively, the population gram; it is now the Department of Behavioral munities. Between 1981 and 2000, suicide targeted by the suicidal behavior prevention Health of the Western Athabaskan Tribal

1238 | Research and Practice | Peer Reviewed | May et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

Nation, with an annual budget of more than notes were subsequently compiled into a in cars or outdoors) to avoid feelings of dis- $1 million. single document and distributed throughout comfort that might arise in visiting the mental The program focused primarily on 10- to the community. This document formed the health program offices. 19-year-olds, the population in this commu- foundation for program components and the Staff growth was substantial and constant nity identified by prior research as most at development and implementation of the throughout the evaluation period. Before risk for developing symptoms of self-destruc- Western Athabaskan Tribal Nation’s judicial 19 8 9, the IHS employed 1 full-time mental tion.16 , 2 0 , 21 Five years into the project, specific and program policies, such as a domestic health technician and contracted for the ser- re-education and awareness-raising activities violence code and a domestic violence and vices of 1 master’s-level counselor and 1 for the 20- to 24-year-old age group were child abuse prevention program. Interest- doctoral-level (PhD) psychologist 1 day per added. The goals of the program were to re- ingly, suicide itself was not one of the most week. As part of the Adolescent Suicide Pre- duce the incidence of adolescent suicides and important problems identified by the com- vention Project, the tribe hired 1 full-time suicide attempts and to increase community munity. Instead, community members stated master’s-level social worker and increased education and awareness about suicide and that, because many of the problems identi- the psychologist’s time to 3 days per week in related behavioral issues, such as child abuse fied could lead to suicide, suicide could not 19 8 9. In 1992, staff increased to 5 full-time and neglect, family violence, trauma, and al- be addressed in isolation. To prevent sui- positions (82% clinical). By 1996, funding cohol and substance abuse (P. Serna, unpub- cide, underlying issues of alcoholism, do- for child abuse, fetal alcohol syndrome, and lished data, 1991). All levels of prevention mestic violence, child abuse, and unemploy- domestic violence prevention activities and a (universal, selective, and indicated)34 were ment must also be confronted. Community contract with the IHS for mental health ser- targeted in program activities. Key compo- meetings were held in conjunction with a vices increased the full-time support staff to nents were based on the following program community mobilization project,35 spon- 12. In 1997, the tribe entered into another goals and objectives: (1) identifying suicide sored by the IHS in collaboration with Rut- contract with the Bureau of Indian Affairs risk factors specific to the Western gers University, and continued to be held for additional clinical social workers for Athabaskan Tribal Nation that may be appli- on a regular basis long after the mobiliza- child and adult welfare and financial ser- cable to other AIAN communities; (2) identi- tion project was completed. vices. In 2000, the behavioral health pro- fying specific individuals and families at high The model Adolescent Suicide Prevention gram had a staff of 33, including support risk for suicide, violence, and mental health Project had the following integrated program personnel. In 2001, the tribe merged the problems; (3) identifying and implementing components: surveillance through constant substance abuse program with the mental prevention activities to target high-risk indi- data and information gathering; screening/ health and social services functions to form viduals, families, and groups; (4) providing clinical interventions with extensive outreach the Department of Behavioral Health De- direct mental health services to high-risk indi- in conventional institutions (health clinics, partment of the Western Athabaskan Tribal viduals, families, and groups; and (5) imple- schools, and social welfare programs) and un- Nation, and project staff increased to 57 po- menting a communitywide systems approach conventional settings (outdoor venues where sitions, including the director, the clinical di- to enhance community knowledge and troubled youths and alcohol abusers fre- rector, a psychiatrist 3 days per week, and awareness (P. Serna, unpublished data, 1991). quently congregate, community functions 21 clinical positions. The department en- A broad, community-wide systems suicide such as traditional and modern dances); social compasses 19 positions for an inpatient so- prevention model was developed4,5 that so- services (child and adult welfare activities); cial detoxification program; 5 positions for licited active involvement from key constitu- school-based prevention programs on topics maintenance/transportation services; and 9 encies—tribal leadership, health care provid- including general life skills development14 ; positions for support services such as ac- ers, parents, elders, youths, and clients—in its and community education for adults and countants, medical records personnel, recep- design and implementation. The community youths on general topics (e.g., parenting) tionists, and secretaries. Most of these people planning process35 had a broad effect on and specific topics (e.g., the nature of self- also assist in public education and other pre- other public institutions of the reservation, destructive behaviors). Neighborhood volun- vention activities. such as those involved with judicial, social teers of various ages were chosen as “natural service, and health program policies. helpers” to engage in peer training, personal METHODS More than 50 interactive community and program advocacy, referral of clients for workgroup sessions were held to examine professional mental health services, and pro- Project data (i.e., clinical record reviews, the following questions: What are the prob- vision of counseling to people who preferred lists of prevention activities, biannual reports) lems and issues in the community? What are to seek help and assistance from knowledge- were maintained for recordkeeping, and sys- the barriers to resolving these problems? able and trusted laypersons in less formal tematic analyses of these data were con- What can be done to solve problems and settings.27–29 Professional mental health staff ducted. Program evaluation has been integral overcome barriers? worked as a team with the natural helpers, in shaping public health initiatives; 3 external At each meeting, all statements and com- and these staff often provided services in un- evaluators have produced 3 formal evalua- ments were recorded, and these transcribed conventional settings in the community (e.g., tions of the prevention program.23,24,31,32

July 2005, Vol 95, No. 7 | American Journal of Public Health May et al. | Peer Reviewed | Research and Practice | 1239  RESEARCH AND PRACTICE 

The following definitions have been used TABLE 1—Number (and Proportion Female) of Suicidal Gestures, Attempts, and throughout the operation of the prevention Completions Among All Ages Combined, by 2-Year Period: Western Athabaskan Tribal program. Suicide attempts are defined as life- Nation, 1988–2002 threatening, self-inflicted injuries that in the absence of medical intervention would result Total Self- Destructive Acts in death. Suicide gestures are behaviors that are Individual Suicidal Behaviors (Including Suicide not in themselves physically life-threatening Gestures Attempts Completions Completions) but that require intervention because of their No. Mean No. No. Mean No. No. Mean No. Mean No. self-destructive intent. Suicide completions are Years (% Female) per Year (% Female) per Year (% Female) per Year No. per Year acts of intentional self-inflicted injury that result in death. 1988–1989 30 (66.7) 15.0 39 (51.3) 19.5 3 (0.0) 1.5 72 36 Each nonfatal suicidal act is classified 1990–1991 29 (55.2) 14.5 17 (29.4) 8.5 2 (0.0) 1.0 48 24 through staff consultation in case conferences. 1992–1993 20 (55.0) 10.0 24 (41.7) 12.0 3 (0.0) 1.5 47 23.5 For example, the behavior of a client who 1994–1995 24 (41.7) 12.0 11 (45.5) 5.5 3 (0.0) 1.5 38 19 talks about killing himself, engages in behav- 1996–1997 18 (44.4) 9.0 11 (36.4) 5.5 4 (0.0) 2.0 33 16.5 ior leading to his arrest and detention in jail, 1998–1999 8 (62.5) 4.0 24 (58.3) 12.0 2 (0.0) 1.0 34 17 and in his cell removes his socks and ties 2000–2001 8 (37.5) 4.0 10 (50.0) 5.0 2 (0.0) 1.0 20 10 them around his neck—but then neither at- 2002 8 (75.0) 8.0 4 (100.0) 4.0 2 (50.0) 2.0 14 14 taches the socks to a fixed object nor tightens Total 145 (54.5) 9.7 140 (47.9) 9.3 21 (4.8) 1.4 306 20.4 them with any tool or implement—would be classified as a suicidal gesture. By contrast, the behavior of a client who ingests overdose TABLE 2—Annual Number of Suicidal Gestures and Attempts (Combined), by Year and Age quantities of 1 or more medications and then Group: Western Athabaskan Tribal Nation, 1988–2002 drives to an isolated spot on the reservation would be classified as a suicide attempt, even Annual Suicidal Gestures and Attempts Combineda if he leaves the empty bottle where it can be Age Group All Ages noticed or found by others. More serious at- Years 11–18 y 19–24 y ≥25 y Combined tempts could involve a firearm or attempted hanging, which, because the client flinched 1988–1989 29 (14.5) 21 (10.5) 22 (11) 72 (36) when the trigger was pulled or the neck re- 1990–1991 3 (1.5) 21 (10.5) 24 (12) 48 (24) straint was faulty, he survived. The profes- 1992–1993 7 (3.5) 10 (5.0) 30 (15) 47 (23.5) sional staff conducts case reviews, held as 1994–1995 8 (4.0) 9 (4.5) 21 (10.5) 38 (19) roundtable discussions on a weekly basis, 1996–1997 10 (5.0) 6 (3.0) 17 (8.5) 33 (17.5) with staff discussing issues of trauma or con- 1998–1999 9 (4.5) 9 (4.5) 16 (8.0) 34 (17) cerns about clients. These peer reviews assist 2000–2001 3 (1.5) 5 (2.5) 12 (6.0) 20 (10) in coordination of services, development of 2002 3 (1.5) 1 (0.5) 10 (5.0) 14 (14) treatment plans, referral for additional ser- Total (1988–2002) 72 (4.8) 82 (5.5) 152 (10.1) 306 (20.4) vices, and final diagnosis. The program’s di- aNumbers are for indicated 2-year period and yearly average (in parentheses). rector and psychologist have been with the program since its inception, providing continuity of definition and direction. Data on completed suicides are recorded by the local Chicago, Ill). Where linear regression was ap- RESULTS IHS clinic and verified by the State Office of plied to the data in Table 3, the dependent Medical Investigation, which investigates all variables were annual frequencies of gestures, A steady reduction in suicidal gestures sudden and unexplained deaths. attempts, and completions separately and of and attempts was recorded throughout the Descriptive analyses for 2 years before gestures and attempts combined among se- course of the program (Tables 1 and 2). program implementation and for 13 years of lected age groups and among all age groups Annual averages dropped from 15 gestures the program used frequencies, percentages, combined. The frequencies of the specific and 19.5 attempts before the program began and rates. Graphic analyses in Figures 1 and self-destructive behaviors are presented in (1988–1989) to 14.5 gestures and 8.5 2 were performed with Microsoft Excel Tables 1 and 2 as 2-year averages and as attempts during the first 2 years of the (Microsoft Corp, Redmond, Wash), and simple means per year to ease the reading and inter- program (1990–1991); annual averages linear regression analyses in Table 3 were pretation of trends. The independent variable dropped further to lows of 4.0 gestures (dur- performed with SPSS Version 10 (SPSS Inc, was time in years. ing 1998–2001) and 4.0 attempts (during

1240 | Research and Practice | Peer Reviewed | May et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

youths aged into this group or were influenced by universal prevention (e.g., education, community awareness, social change). Frequencies of combined gestures and attempts in this group were highest in 1992 and 1993 (the third and fourth years of the program), at 15 per year, dropping to 5.0 by 2002, as illustrated in Figure 2. Thus, suicidal behavior rates remained highest in the nontargeted age group of 25 years and older. Regression analyses showed that the clear- est drop in frequency of occurrence was for suicidal gestures (R 2 =.595) (Table 3). A steep, significant downward slope (β=–.790, P =.000) in suicidal gestures was seen over time, with 59.5% of the variance explained by the existence of the program. Suicide at- FIGURE 1—Suicide gestures, attempts, and completions among members of the Western tempts were also significantly lowered (R 2 = Athabaskan Tribal Nation. .322, β=–.609, P =.016). Suicide completions were unaffected. Young adults aged 19 to 24 years had the greatest drop in com- 2002). No decrease or increase occurred in Because the targeted age group was bined gestures and attempts (R 2 =.533, β= suicide completions, which remained at 1 to youths aged 11 to 18 years, the most imme- –.765, P =.001). The frequency of gestures 2 per year throughout the period 1988 to diate effect was expected in this group, and and attempts also declined significantly 2002. Overall, the annual mean number of with a lasting effect on individuals and fami- among youths aged 11 to 18 (R 2 =.211, β= total self-destructive acts dropped from 36 to lies, one would expect a gradual change in –.517, P =.048), but not among adults aged 14 between 1988 and 2002, an absolute suicidal behavior among older age groups as 25 years and older. For all age groups com- drop of 61.1%. Figure 1 shows the trajectory these youths age. In 1988 to 1989, before bined, rates of self-destructive behavior of decline in gestures and attempts. the program began, the annual frequency showed significant effects from the program Fifty-five percent of gestures, 48% of at- of gestures and attempts for 11- to 18-year- regardless of whether suicide completions tempts, and 5% of completions occurred olds was 14.5; this number decreased to were included in the analysis (P =.000). The among women. Suicidal gestures and attempts 1.5 during the first 2 years of the program regression analysis indicates that the fre- were common among both sexes, but suicide (1990–1991), rose briefly to 5.0 in 1996 to quency of total self-destructive acts declined deaths occurred almost exclusively among 19 97, but returned to 1.5 during the period by 73% (R 2 =.729; β=–.865) over the en- men (95%). 2000 to 2002, an absolute reduction of tire span of the program. Alcohol involvement and family issues 89.7% (Tables 1 and 2). The decrease in linked to all 3 types of self-destructive behav- annual frequency of gestures and attempts DISCUSSION ior are not presented in the tables. Two thirds occurred later among 19- to 24-year-olds, (66.7%) of all self-destructive acts on the who were not intensively targeted, but as The consistency and timing of reduction in Western Athabaskan Tribal Nation reserva- targeted youths aged, a positive program ef- suicidal behavior correlated with the develop- tion were alcohol-related, similar to the pro- fect was observed. The annual frequency of ment and delivery of prevention program in- portion for suicide completions among all gestures and attempts of 10.5 remained un- terventions in this small AIAN community, New Mexico American Indians (69%) but changed during the first 2 years of the pro- indicating that prevention efforts were likely lower than the proportion for suicide comple- gram but subsequently declined to 2.5 (in successful in reducing suicidal behavior. We tions among all Western Athabaskans in New 2000–2001) and to 0.5 (in 2002); these believe that this success is owed to the pro- Mexico (83%).19 Most suicidal behavior declines represent absolute reductions of gram’s public health approach, which in- (95.4%) occurred among people who had a 76.2% and 90.2%, respectively. cluded mental health services integrated into significant family history of trauma (e.g., fam- As expected, because young adults aged a comprehensive continuum of services that ily disruption, violent death of relatives, abuse 25 years and older were not specifically tar- target other behavioral health topics at all lev- or neglect); 86.3% of self-destructive individ- geted by the prevention program, the slowest els of prevention.34 Lessons learned from this uals experienced significant individual trauma. and smallest declines in total self-destructive program are as follows. First, a suicide pre- Most suicidal acts (68%) occurred among un- acts was in this age group. Any program ef- vention program should not focus on a lim- employed persons (aged 16 years and older). fect could only have come later, as targeted ited range of self-destructive behaviors;

July 2005, Vol 95, No. 7 | American Journal of Public Health May et al. | Peer Reviewed | Research and Practice | 1241  RESEARCH AND PRACTICE 

FIGURE 2—Age-specific total suicide gestures and suicide attempts among members of the Western Athabaskan Tribal Nation.

We believe that the temporal association TABLE 3—Results of Simple Independent Linear Regression Analysis Estimating Effects between program delivery and reduction in Over Time of a Model Adolescent Suicide Prevention Program: Western Athabaskan Tribal suicidal behavior is suggestive of a causal as- Nation, 1988–2002 sociation. However, one can hypothesize al- Variable Adjusted R2 SE Df β tP ternative explanations for this trend. One explanation is that suicidal youths might have Individual suicidal behaviors left the reservation, thus reducing the number Gestures .595 3.01 14 –.790 –4.65 .000 of those who were high risk and independently Attempts .332 4.54 14 –.609 –2.77 .016 lowering the frequency of self-destructive be- Completions .051 0.84 14 –.156 –.057 .578 (NS) haviors during the study period. Although Suicidal gestures and attempts combined, by age group this explanation may account for trends 11–18 y .211 4.03 14 –.517 –2.18 .048 among other rural populations, such a trend 19–24 y .553 2.50 14 –.765 –4.28 .001 did not occur in the Western Athabaskan ≥25 y .185 3.48 14 –.493 –2.05 .062 (NS) Tribal Nation. At any given time, 90% of the All ages combined .746 4.03 14 –.874 –6.48 .000 enrolled population of the tribe is on the Total self-destructive acts (including suicide .729 4.28 14 –.865 –6.21 .000 reservation. Approximately 54% of the popu- completions) lation was younger than 21 years. This iso- Note. NS=not significant. lated tribal nation is neither geographically nor socially mobile. A second possible explanation is that rapid improvement in social and rather, it must include an emphasis on root development should be based on continuous economic conditions might have affected the conditions and an array of social, psychologi- evaluation and feedback from community population,36 reducing suicidal behavior. cal, and developmental issues.1–5 Second, and program staff. Therefore, objectives and However, the tribe’s hotel and casino, devel- community involvement from the beginning activities may shift while goals remain the oped in the early 1990s, are small and have is critical in developing strategies with which same. Constant vigilance is necessary for been among the least successful in the South- to address issues identified in a culturally, en- maintaining a public health–focused suicide west, primarily because of the isolation of the vironmentally, and clinically appropriate man- prevention program, as well as program and community from any substantial population ner. Third, flexibility in program development staff development, community relations, re- center. No substantial growth has occurred and implementation is essential, and program source development, and administration. in the economic sector. Therefore, the most

1242 | Research and Practice | Peer Reviewed | May et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

likely explanation for the observed reduction before, during, and after the program. For ex- equally successful in adopting a public health in suicidal behavior is the presence and example, using ethnographic studies elsewhere approach to suicide prevention. If federal and pansion of the behavioral health program and in the Southwest, Levy and Kunitz analyzed other funds become available to other tribes, the comprehensive mental health, suicide pre- the ethnic meanings of suicidal behavior the applicability of this model intervention vention, and clinical services it provides. among the Hopi and uncovered an associa- can be tested among larger tribal communi- Although we observed a reduction in sui- tion between completed suicide and other ties. The program may also serve as a model cide gestures and attempts, suicide deaths re- self-destructive behavior and marriages un- for non-AIAN communities seeking to adopt mained the same for all ages combined. How- approved by traditional values and cultural a public health approach to preventing self- ever, most suicides in the community occur leaders.33 Uncovering similar cultural and so- destructive behaviors. among people in their late 20s and in their cial factors among the Western Athabaskan 30s. Furthermore, there is some evidence people studied here may explain the exact that suicide completions may have decreased underlying themes or mechanisms that al- About the Authors 23 Philip A. May is with the Departments of Sociology and for the target population, although the num- lowed the program to work effectively. Family and Community Medicine, University of New Mex- bers are too small for statistical inference. This AIAN community is a small and iso- ico, Albuquerque. Patricia Serna and Lance Hurt are with Because most suicide completions among lated one, which allows researchers relatively the Western Athabaskan Behavioral Health Services Pro- gram. At the time of the study, Lemyra M. DeBruyn was New Mexico American Indians have histori- easy access to all community members. The with the Division of Violence Prevention, Centers for Dis- cally occurred among tribal members youn- longevity and growth of the evaluated pro- ease Control and Prevention, Atlanta, Ga. ger than 35 years,8,22,23 have frequently in- gram is uncommon in the American Indian Requests for reprints should be sent to Philip A. May, 19 PhD, Center on Alcoholism, Substance Abuse, and Addic- volved alcohol, have been by firearms or context. Typically, suicide prevention initiatives tions, University of New Mexico, 2650 Yale Blvd, Albu- hanging,22,23 and have been characterized as are not sustained because of limited financial querque, NM 87106 (e-mail: [email protected]). impulsive,37 one would expect suicide com- resources, demands on public health staff time This article was accepted October 2, 2004. Note. The opinions expressed by the authors are theirs pletions to be more resistant to programmatic and energy, changes in personnel, and the and do not necessarily represent those of the Indian Health intervention and prevention than gestures waning of interest once a suicidal cluster sub- Service, the Western Athabaskan Tribal Nation, or the and attempts. A potentially useful conceptu- sides. Recently, the US Commission on Civil Centers for Disease Control and Prevention. alization in this regard, appearing in both the Rights reported that federal resources directed general and the AIAN-specific literature,37–39 to AIAN communities have not been sufficient Contributors is the classification of suicidal behavior as ei- to address these communities’ basic and very P. A. May was the principal investigator and codesigner 40 of the prevention program evaluation. P. Serna directed ther chronic or acute. Chronic suicidal behav- urgent needs. The commission cited health the prevention program, instituted the program compo- ior, in which the individual progresses care needs, as well as the need to develop the nents and the evaluation plan, and oversaw the routine steadily through a pattern of self-destructive tribal funding and infrastructure without which collection of all data during the project period described in this report. P. Serna, L. Hurt, and L. DeBruyn thoughts and acts of increasing severity from public health services cannot be delivered. The cowrote the program description portion. L. Hurt was gestures to attempts to eventual suicide, is suicide prevention program described here is the clinical director of the prevention program, and more common in the general population unique in that its development and funding L. DeBruyn was with the Indian Health Service and subsequently the Centers for Disease Control (particularly among older people) than is have been increased and maintained over an and Prevention and provided direct consultation to acute ideation and suicidal behavior, in extended period of time. Importantly, there the program. which no obvious progression precedes the was continuity of focus and emphasis on sui- impulsive act. It appears from our evaluation cide prevention by key staff members who Acknowledgments of the suicide prevention program in the remained with the program throughout the This program evaluation research was supported in part Western Athabaskan Tribal Nation that this period evaluated. The components of this ef- by the US Indian Health Service (grant 242-95-0022) and the Centers for Disease Control and Prevention, as program may be most successful in changing fort can be adapted to larger tribal groups, be- well as by allocations from the Western Athabaskan the path of chronic suicidal behavior but may cause they include sound public health theory Tribal Nation. have minimal or no effect on acute suicidal and practice, cultural relevance, and integra- We are grateful to the Western Athabaskan Tribal Council for their support of this program. We also behavior. However, no increase in suicide tion into the tribal infrastructure. Specifically, thank all who have contributed to this project and the completions occurred on the reservation dur- these components include local surveillance research and who have created a successful program, ing the period of the program. In the past, and outreach, careful record keeping and on- including Debra Hurt, Ricardo Gonzales, PhD, Michael Biernoff, MD, Nancy Van Winkle, PhD, the tribe in our study registered significant going evaluation, a strong behavioral health Everett Vigil, Denton Garcia, and all the staff of West- increases in the rate of suicide in 6- to 8-year base, school and community-based education, ern Athabaskan Behavioral Health Program. cycles.20,21,23 Whether the program pre- integration of violence and substance abuse JoAnn Pappalardo of the Indian Health Service provided population data for the article. David Buckley vented such a cyclical upswing in suicide prevention services, a team approach, and con- assisted in the preparation of the graphics and data completion requires further analysis with stant vigilance. analysis and in the corrections for various drafts of the larger numbers. Ethnographic analysis is also On the basis of the positive results reported article. Phyllis Trujillo coordinated the schedules and communicated between the various authors for the en- needed to describe program functioning and here, we believe that other AIAN communi- tire length of this prevention project and typed parts of how suicide was viewed in the community ties can develop similar programs and be the article and the tables for the article.

July 2005, Vol 95, No. 7 | American Journal of Public Health May et al. | Peer Reviewed | Research and Practice | 1243  RESEARCH AND PRACTICE 

Human Participant Protection 18. Middlebrook DL, LeMaster PL, Beals J, Novins DK, 35. Chavis DM, Speer P, Reznick I, Zippay A. Build- Institutional review of all research protocols was carried Manson SM. Suicide prevention in American Indian ing community capacity to address alcohol and drug out and approved by the Indian Health Service Albu- and Alaska Native communities: a critical review of abuse. In: Davis RC, Lurgio AJ, Rosenbaum DP, eds. querque Area and National Boards. In addition, the programs. Suicide Life Threat Behav. 2001;31(suppl): Drugs and the Community. Springfield, Ill: Charles C Western Athabaskan Tribal Council and Tribal Health 132–139. Thomas; 1993:251–284. Board approved the original version of this article. 19. May PA, Van Winkle NW, Williams M, McFeeley PJ, 36. Costello EJ, Compton S, Keeler G, Angold A. Rela- DeBruyn LM, Serna P. Alcohol and suicide death tionships between poverty and psychopathology: a nat- References among American Indians of New Mexico: 1980–1998. ural experiment. JAMA. 2003;290:2023–2029. Suicide Life Threat Behav. 2002;32:240–255. 1. The Surgeon General’s Call to Action to Prevent Sui- 37. Berlin IN. Prevention of adolescent suicide among cide. Washington, DC: US Public Health Service; 1999. 20.Van Winkle NW, May PA. Native American sui- some Native American tribes. In: Feinstein SC, ed. Ado- cide in New Mexico, 1957–1979: a comparative study. 2. National Strategy for Suicide Prevention: Goals and lescent Psychiatry: Developmental and Clinical Studies. Hum Organ. 19 8 6;45:296–309. Objective for Action. Washington, DC: US Public Health Chicago, Ill: University of Chicago Press; 1985:77–93. 21.Van Winkle NW, May PA. An update on Native Service; 2002. 38.Kehoe JP, Abott AP. Suicide and attempted sui- American suicide in New Mexico, 1980–1987. Hum 3. Goldsmith SK, Pellmar TC, Kleinman AM, Bunney cide in the Yukon territories. Can Psychiatr Assoc J. Organ. 19 93;52:304–315. WE, eds. Reducing Suicide: A National Imperative. 1975;20:15–23. Washington, DC: National Academy Press; 2002. 22. May PA, Van Winkle NW. Durkheim’s suicide theory and its applicability to contemporary American 39. Berlin IN. Suicide among American Indian adoles- 4. Knox KL, Conwell Y, Caine ED. If suicide is a Indians and Alaska Natives. In: Lester D, ed. Emile cents: an overview. Suicide Life Threat Behav. 19 87;17: public health problem, what are we doing to prevent Durkheim: Le Suicide 100 Years Later. Philadelphia, 218–232. it? Am J Public Health. 2004;94:37–45. Pa: Charles Press; 1994:296–318. 40.US Commission on Civil Rights. A quiet crisis: 5. Sanddal ND, Sanddal TL, Berman AL, Silverman 23. Van Winkle NW, Williams M. Evaluation of the federal funding and unmet needs in Indian country. MM. A general systems approach to suicide prevention: National Model Adolescent Suicide Prevention Project: A July 2003. Available at: http://www.usccr.gov. Ac- lessons learned from cardiac prevention and control. Comparison of Suicide Rates Among New Mexico Ameri- cessed July 17, 2003. Suicide Life Threat Behav. 2003;33:341–352. can Indian Tribes, 1980–1998. Tulsa, Okla: Oklahoma 6. Dizmang LH, Watson J, May PA, Bopp J. Adoles- State University School of Osteopathic Medicine; 2001. cent suicide at an Indian reservation. Am J Orthopsychi- 24. DeBruyn LM, May PA, O’Brien M. Suicide Inter- atry. 1974;44:43–49. vention and Prevention: Evaluation of Community-Based 7. Shore JH, Bopp JE, Waller TR, Dawes JW. A sui- Programs in Three American Indian Communities. At- cide prevention center on an Indian reservation. Am lanta, Ga: Division of Injury Control, US Centers for J Psychiatry. 2001;128:91–96. Disease Control and Prevention; 1997. 8. May PA. Suicide and self-destruction among 25. Web-based Injury Statistics Query and Reporting American Indian youths. Am Indian Alsk Native Ment System (WISQARS) Web site. Atlanta, Ga: National Health Res. 19 87;1:52–69. Center for Injury Prevention and Control, Centers for Disease Control and Prevention. Available at: http:// 9. Tower M. A suicide epidemic in an American In- www.cdc.gov/ncipc/wisqars. Accessed October 24, dian community. Am Indian Alsk Native Ment Health 2002. Res. 1989;3:34–44. 26. Indian Health Service. Trends in Indian Health, 10. Bechtold DW. Indian adolescent suicide: clinical 1998–1999. Washington, DC: US Dept of Health and and developmental considerations. Am Indian Alsk Na- Human Services; 2001. tive Ment Health Res Monogr Ser. 19 9 4;4:71–80. 27.Forbes NE. The state’s role in suicide prevention 11. Bee-Gates D, Howard-Pitney B, LaFramboise TD, programs for Alaska Native youth. Am Indian Alsk Na- Rowe W. Help-seeking behavior among Native Ameri- tive Ment Health Res Monogr Ser. 19 9 4;4:235–249. can Indian high school students. Prof Psychol Res Pr. 28.Fox J, Manitonabi D, Ward JA. An Indian commu- 1996;27:495–499. nity with a high suicide rate: five years after. Can J Psy- 12. Claymore BJ. A public health approach to suicide chiatry. 1984;29:425–427. attempts on a Sioux reservation. Am Indian Alsk Native 29.Rogers DD. Community crisis intervention in sui- Ment Health Res. 19 8 8;1:19–24. cide epidemics. Arctic Med Res. 19 91;50(suppl): 13. LaFromboise TD. American Indian Life Skills De- 276–280. velopment Curriculum. Madison, Wis: University of Wis- 30.Ward JA. Preventive implications of a Native men- consin Press; 1996. tal health program: focus on suicide and violent death. 14 . LaFromboise TD, Howard-Pitney B. The Zuni life J Prev Psychiatry. 1984;2:371–385. skills development curriculum: description and evalua- 31. Centers for Disease Control and Prevention. Sui- tion of a suicide prevention program. J Couns Psychol. cide prevention evaluation in a Western Athabaskan 19 95;42:479–486. American Indian tribe: New Mexico, 1988–1997. 15. Thompson JW, Walker RD. Adolescent suicide MMWR Morbid Mortal Wkly Rep. 1998;47:257–261. among American Indians and Alaska Natives. Psychiatr 32. May PA, Del Vecchio A. National Model Adoles- Ann. 1990;20:12 8–133. cent Suicide Prevention Project: Evaluation Report to the 16. May PA. Suicide and suicide attempts among Western Athabaskan Tribe. Albuquerque, NM: Center American Indians and Alaska Natives: a bibliography. on Alcoholism, Substance Abuse and Addictions, Uni- Omega. 1990;21:199–214. versity of New Mexico; 1994. 17. DeBruyn LM, Hymbaugh K, Simpson D, Wilkins B, 33. Levy JE, Kunitz SJ. A suicide prevention program Nelson S. When communities are in crisis: planning for for Hopi youth. Soc Sci Med. 19 87;25:931–940. response to suicides and suicide attempts among Amer- 34. Institute of Medicine. Reducing Risks for Mental ican Indian tribes. Am Indian Alsk Native Ment Health Disorders: Frontiers for Preventive Intervention Research. Res Monogr Ser. 1994;4:223–234. Washington, DC: National Academy Press; 1994.

1244 | Research and Practice | Peer Reviewed | May et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

Childhood Socioeconomic Position, Educational Attainment, and Adult Cardiovascular Risk Factors: The Aberdeen Children of the 1950s Cohort Study

| Debbie A Lawlor, PhD, G. David Batty, PhD, Susan M.B. Morton, PhD, Heather Clark, MSc, Sally Macintyre, PhD, and David A. Leon, PhD

Studies have revealed associations between Objectives. We assessed the associations of childhood socioeconomic posi- low childhood socioeconomic position or status tion with cardiovascular disease risk factors (smoking, binge alcohol drinking, (SEP) and cardiovascular disease (CVD) and and being overweight) and examined the roles of educational attainment and CVD risk factors in later life that are indepen- cognitive functioning in these associations. dent of adult SEP.1–15 However, most of these Methods. Data were derived from a cohort study involving 7184 individuals studies have involved the use of adults’ reports who were born in Aberdeen, Scotland, between 1950 and 1956; had detailed of childhood SEP, which may be incomplete records on perinatal characteristics, childhood anthropometry, and cognitive and inaccurate.16 Before developing effective functioning; and responded to a mailed questionnaire when they were aged 45 policy interventions to abolish the link be- to 52 years. tween low childhood SEP and CVD, it is nec- Results. Strong graded associations existed between social class at birth and smoking, binge drinking, and being overweight. Adjustment for educational at- essary to understand the causal pathways containment completely attenuated these associations. However, after control for necting them. Low childhood socioeconomic adult social class, adult income and other potential confounding or mediating position is associated with low cognitive func- factors, some association remained. 17 tioning and low educational attainment. Conclusions. Educational attainment is an important mediating factor in the Thus, it is plausible that cognitive function and relation between socioeconomic adversity in childhood and smoking, binge drink- educational attainment are important interme- ing, and being overweight in adulthood. (Am J Public Health. 2005;95:1245–1251. diaries in the association between childhood doi:10.2105/AJPH.2004.041129) SEP and adult risk factors, factors that may in turn lead to CVD. The aims of the present study were to assess the associations of child- Assessment of Childhood Assessment of Cognitive Functioning hood SEP with CVD risk factors (smoking, Socioeconomic Position and Educational Attainment binge alcohol drinking, and being overweight) Childhood SEP was assessed at 2 time Throughout the 1950s in Aberdeen, cogni- and to examine the role of educational attain- points. Social class at birth was based on the tive function was routinely tested through the ment and cognitive function in these associa- occupation of the study participant’s father; administration of IQ tests to children at the tions in a cohort born in the city of Aberdeen, these data were obtained from Aberdeen Ma- ages of 7, 9, and 11 years. These IQ scores Scotland, between 1950 and 1956. ternity and Neonatal Databank obstetric rec- were abstracted for the study participants, and ords. Six categories were included: profes- results of tests taken after December 1962 METHODS sional, managerial, skilled nonmanual, skilled and up until 1964 also were obtained as they manual, semiskilled, and unskilled manual. became available. In this study, we used the Study Participants These categories can be collapsed into the 2 results from the tests administered when the The Aberdeen Children of the 1950s Study general categories of nonmanual (profes- participants were aged 7 and 11 years. The involved a cohort of 12150 children who sional, managerial, and nonmanual/skilled tests conducted at 7 years provided an indica- were born in Aberdeen, Scotland, between nonmanual) and manual (manual/skilled tion of the children’s functioning before they 1950 and 1956 and who took part in a child manual, semiskilled, and unskilled manual). had completed a significant amount of formal development survey in the 1960s.18–20 Com- At the time of the childhood survey (1962), education, thus reflecting early postnatal expo- prehensive information was abstracted from father’s occupation was reported by each sures and family environment. Conversely, the the Aberdeen Maternity and Neonatal Data- child and was similarly coded. All analyses tests conducted at age 11 reflected cognitive bank regarding course of mother’s preg- were conducted with these 2 measures as ex- functioning at the end of the children’s pri- nancy and children’s physical characteristics posure variables; however, none of the re- mary school education, and thus scores were at birth.19 In 1999, this cohort study was sults differed substantively between the 2 influenced by the educational process and reinitiated, and between 2000 and 2002 measures, and thus we present results only peer relationships as well as earlier exposures. surviving cohort members were mailed a for associations with social class at the time At 7 years of age the participants were ad- health questionnaire.18 of birth. ministered the Moray House Picture Intelli-

July 2005, Vol 95, No. 7 | American Journal of Public Health Lawlor et al. | Peer Reviewed | Research and Practice | 1245  RESEARCH AND PRACTICE 

gence Tests.18 All participants were adminis- were more likely to be female, to have been from 0 (highest SEP) to 1 (lowest SEP) repre- tered the test within 6 months of their 7th categorized as affluent in terms of childhood sented the entire population, participants in birthday. The tests given at age 11, also con- SEP, and to have had high cognitive function this group would be allocated a score of 0.05 ducted within 6 months of participants’ 11th scores as children.18 , 21 (0.1/2); if 20% of the respondents were in birthday, included a battery of Moray House Participants were asked about their most re- the managerial category, this group would be tests: 2 tests of verbal reasoning and 1 each cent occupation. Occupations were categorized allocated a score of 0.20 (0.1 + 0.2/2); and of arithmetic and English. Because the mean into the same 6 categories used for childhood so on. The index of inequality was then ob- verbal reasoning score was highly correlated social class. Also, participants were asked to in- tained by regressing the outcome on each of with the arithmetic and English scores (Pear- dicate their personal gross annual income (no these SEP scores. The virtue of this technique son correlation coefficients of 0.86 and 0.89, income, less than £2000, £2000–£5999, is that it is directly interpretable as compar- respectively, P<.001 for both) and all 3 £6000–£9999, £10000–£14999, ing, in each case, the highest (0) and lowest showed similar associations with exposures £15000–£19999, £20000–£29999, (1) SEP indicators assigned.23 and outcomes in this study, mean verbal rea- £30000–£39999, £40000 or more); A series of multiple logistic regression soning scores at 11 years were used as the weekly equivalent amounts were provided models was used to assess the associations of measure of cognitive functioning at that age. for each category. social class at birth with the risk factors ex- In the mailed health questionnaire, partici- Respondents were asked about their own amined. In these models, participants’ age, pants were asked to report the age at which smoking behavior and whether their parents cognitive functioning at ages 7 and 11, age at they left secondary education and to indicate had smoked when they were children. Binge leaving secondary school, intrauterine growth their educational or vocational qualifications. drinking was defined as consumption of 4 or z score, and childhood height and body mass A list was provided that included an option more alcoholic drinks in 1 episode at least index were all entered as continuous vari- of “no formal qualification” and then a hier- once a week. ables. Birth order, family size, adult social archy of seven formal United Kingdom edu- In addition, participants were asked to rec- class, adult income, parental smoking, and cational qualifications from leaving certificate ord their weight and height. Of the 7007 re- educational qualifications were all entered as (lower level of qualification by those leaving spondents who provided an estimate of their categorical variables. These analyses were school at a minimum school-leaving age— weight, 6092 (87%) reported that they had repeated with the inclusion of RII scores (as aged 15 years for this cohort) through a uni- used a scale. There was a tendency for those just detailed) for social class at birth and each versity degree. who had not used a scale to report weights measure of adult SEP and education. Likeli- lower than their weights (74.5 kg vs 75.5 kg; hood ratio tests were used to assess interac- Assessment of Other Childhood P=.07). Therefore, all analyses that included tions. All analyses were conducted with Stata Characteristics body mass index, obesity, or weight were ad- version 8.0 (Stata Corp, College Station, Tex). Birthweight and gestational age data were justed by including a dummy variable for abstracted from obstetric records at the time scale nonuse in the multivariate models. RESULTS of the 1962 survey.18 Participants’ intrauter- Overweight was defined as a body mass ine growth rate was estimated by calculating index of 25 kg/m2 or more.22 Table 1 presents the distributions of CVD standardized z (standard deviation) scores for risk factors among the participants and corre- sex and gestational age (in weeks). Height Statistical Analysis sponding distributions for women and men and weight at school entry were measured Means for continuous variables, and preva- aged 45 to 54 years who took part in the directly, and these data were linked to the lences for dichotomous variables, are presented Scottish Health Survey of 1998.24 Men were childhood survey data participants in 1962. in regard to participant characteristics accord- more likely than women to engage in binge Age- and sex-standardized z scores based on ing to social class at birth. We fit regression drinking and to be overweight, but smoking 3-month age categories were derived for models to assess linear trends across these prevalence rates were similar among men height and weight. categories by entering social class at birth as a and women. Our study participants were less Adult Characteristics score in these models and conducting an F test. likely to be smokers and to be overweight At the time of the questionnaire mailing Relative indexes of inequality (RIIs)23 were than Scottish Health Survey respondents. (2000), 291 (2.4%) of the 12150 original estimated for the associations of social class No evidence was found of any interactions cohort members had emigrated (outside the at birth, educational attainment, adult social between gender and social class at birth in United Kingdom), 62 (0.5%) were in the class, and adult income with childhood and any of the associations (P>.4 for all associa- armed forces or institutionalized, 479 (4.0%) adult height. For each SEP exposure, a score tions), and thus we present all results for had died, and 36 could not be reached for was assigned to each category on the basis women and men combined. The associations other reasons. The remaining 11282 partici- of the midpoint of the proportion of the popu- of adult income with both current smoking pants were mailed a follow-up health ques- lation in that category. For example, if 10% behavior and being overweight differed ac- tionnaire, and 7183 (63.7%) responded. In of the respondents were assigned to the pro- cording to gender (P<.01 for both). The comparison with nonresponders, responders fessional social class category and a score odds ratio (OR) for current smoking among

1246 | Research and Practice | Peer Reviewed | Lawlor et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

TABLE 1—Prevalence Rates of Adult Cardiovascular Risk Factors Among Aberdeen Children of the 1950s Study Participants Compared With Results From the Scottish Health Survey, 1998

Women Men Scottish Scottish Aberdeen Study Health Survey Aberdeen Study Health Survey Total: Aberdeen Study Group No. % (95% CI) % (95% CI) No. % (95% CI) % (95% CI) No. % (95% CI)

Current smokers 3698 27.3 (25.9, 28.8) 34 (31, 37) 3393 27.2 (25.7, 28.7) 40 (37, 44) 7091 27.3 (26.2, 28.3) Ever smokers 3698 50.0 (48.4, 51.6) 54 (50, 58) 3393 54.6 (52.9, 56.2) 60 (56, 63) 7091 52.2 (51.0, 53.3) Binge drinkers 3419 18.3 (17.0, 19.6) . . .a 3252 36.3 (34.7, 38.0) . . .a 6671 27.1 (26.0, 28.1) Overweight individuals 3625 50.3 (48.6, 51.9) 63 (59, 67) 3365 64.7 (63.0, 66.3) 75 (72, 78) 6990 57.2 (56.0, 58.4)

Note.CI=confidence interval. aNot available.

low-income versus higher income women tainment in any of the associations (P >.2 means of adult characteristics according to was1.95 (95% confidence interval [CI]= for all associations). whether the participant’s father was unem- 1.65, 2.31); the corresponding OR among Table 2 presents the adult characteristics ployed at the time of the participant’s birth men was 2.73 (95% CI=2.31, 3.23). Odds of the participants according to their social are also shown in Table 2. In general, results ratios for being overweight were 1.17 (95% class at birth. The response proportion de- for this unemployed category were similar to CI=1.02, 1.35) among women and 0.73 creased linearly with decreasing SEP. All 3 those for the manual social class category. (95% CI=0.61, 0.86) among men. An inter- CVD risk factors showed a graded associa- Because those who are unemployed are likely action term was incorporated into all of the tion across the social class distribution, with to be a heterogeneous group and cannot be logistic regression models that included both the most unfavorable level of each seen in grouped according to an occupational classifi- gender and income as covariates. No evi- adults who were born in the lowest-status cation, all further results, including the trend dence of interactions between gender and in- socioeconomic groups. Adult social class, tests presented in Table 2, exclude partici- come for other outcomes was found, nor was income, and educational attainment also pants whose fathers were unemployed. there evidence of interactions between gen- showed strong incremental associations with Table 3 shows the RII coefficients of child- der and adult social class or educational at- social class at birth. The prevalence and hood and adult height for all SEP indicators.

TABLE 2—Adult Cardiovascular Disease Risk Factors and Other Adult Characteristics of Participants, by Social Class at Time of Birth: Aberdeen Children of the 1950s Study

Social Class at Birth Professional/Managerial Skilled Nonmanual Skilled Manual Semiskilled Unskilled Manual Unemployed Characteristic (n=789) (n=870) (n=3154) (n=975) (n=1048) (n=348) P for Trenda

Questionnaire response rate, % 73.3 69.8 64.0 61.9 57.7 55.0 <.001 Current smoking, % (95% CI) 19.8 (17.1, 22.7) 20.2 (17.6, 23.0) 26.6 (25.1, 28.2) 33.6 (30.7, 36.7) 33.4 (30.6, 36.4) 31.5 (26.8, 36.6) <.001 History of smoking, % (95% CI) 45.1 (41.6, 48.6) 44.2 (40.9, 47.6) 51.5 (49.8, 53.3) 58.5 (55.3, 61.6) 60.0 (57.0, 62.9) 53.5 (48.2, 58.8) <.001 Binge drinking, % (95% CI) 23.3 (20.4, 26.5) 24.2 (21.4, 27.3) 26.9 (25.3, 28.5) 32.0 (29.0, 35.2) 28.8 (26.0, 31.7) 25.1 (20.6, 30.2) <.001 Overweight or obese (BMI≥25 kg/m2), 48.5 (45.0, 52.0) 54.9 (51.5, 58.2) 58.1 (56.4, 59.9) 58.1 (54.9, 61.2) 60.2 (57.1, 63.2) 62.0 (56.7, 67.1) <.001 % (95% CI) BMI, kg/m2, mean ±SD 25.4 ±4.1 26.2 ±4.7 26.6 ±4.7 26.9 ±5.0 27.1 ±5.4 27.4 ±5.4 <.001 Age at leaving secondary school, y, mean ±SD 16.9 ±1.08 16.4 ±1.16 15.9 ±1.12 15.6 ±0.96 15.5 ±0.86 15.9 ±1.12 <.001 University education, % (95% CI) 49.2 (45.7, 52.7) 28.9 (25.9, 32.1) 17.2 (15.9, 18.6) 9.8 (8.0, 11.9) 7.7 (6.2, 9.6) 19.2 (15.3, 24.0) <.001 No formal qualifications or education only up 5.4 (4.0, 7.3) 14.5 (12.3, 17.1) 23.6 (22.1, 25.2) 36.1 (33.0, 39.3) 40.0 (37.0, 43.1) 28.7 (24.0, 33.9) <.001 to grades 2–5, % (95% CI) Manual social class as adult, % (95% CI) 12.7 (10.5, 15.2) 21.7 (19.1, 24.6) 32.3 (30.7, 34.0) 42.8 (39.7, 46.0) 48.6 (45.5, 51.7) 34.1 (29.3, 39.4) <.001 Low income as adult, % (95% CI) 31.1 (28.7, 35.2) 37.0 (33.8, 40.4) 45.6 (43.8, 47.4) 53.8 (50.6, 57.0) 56.8 (53.7, 59.8) 51.4 (46.0, 56.7) <.001

Note.CI=confidence interval; BMI=body mass index. a Trend tests across occupational social class categories not including unemployed individuals.

July 2005, Vol 95, No. 7 | American Journal of Public Health Lawlor et al. | Peer Reviewed | Research and Practice | 1247  RESEARCH AND PRACTICE 

TABLE 3—Relative Index of Inequality (RII) in Childhood and Adult Height, by Social Class at Birth and Adult Indicators of Socioeconomic Position: Aberdeen Children of the 1950s Study

Childhood Height for Age and Gender z Score Childhood Height Adult Height RII Coefficienta (95% CI) P for Linear Trend RII Coefficienta (95% CI) P for Linear Trend RII Coefficienta (95% CI) P for Linear Trend

Social class at birth –0.64 (–0.72, –0.59) <.001 –0.08 (–0.09, –0.07) <.001 –0.06 (–0.07, –0.05) <.001 Educational attainment –0.47 (–0.55, –0.36) <.001 –0.04 (–0.05, –0.03) <.001 –0.03 (–0.04, –0.02) <.001 Social class in adulthood –0.54 (–0.63, –0.46) <.001 –0.05 (–0.06, –0.04) <.001 –0.05 (–0.06, –0.04) <.001 Adult income –0.45 (–0.54, –0.37) <.001 –0.04 (–0.05, –0.03) <.001 –0.15 (–0.16, –0.14) <.001

Note.CI=confidence interval. a The RII is the gradient of the slope in height between the most and least affluent for each measure in the study population, based on an allocated score ranging from 0 (most affluent) to 1 (least affluent) that takes into account the fact that each measure has different categories with different proportions of the population in each category.

TABLE 4—Associations Between Social Class at Birth and Adult Behavioral Risk Factors for Cardiovascular Disease After Adjustment for Potential Confounding and Mediating Variables: Aberdeen Children of the 1950s Study

OR (95% CI) for Manual vs Nonmanual Social Class at Birth, by Risk Factor Model Variable(s) Current Smoking Ever Smoking Binge Drinking Overweight

1 Age and gender only 1.65 (1.43, 1.89) 1.49 (1.33, 1.67) 1.27 (1.11, 1.45) 1.33 (1.17, 1.49) 2 Adult social class 1.30 (1.13, 1.50) 1.26 (1.12, 1.42) 1.17 (1.01, 1.34) 1.27 (1.12, 1.43) 3 Adult income 1.43 (1.24, 1.65) 1.38 (1.23, 1.55) 1.21 (1.05, 1.39) 1.36 (1.21, 1.52) 4 Family size 1.46 (1.27, 1.68) 1.35 (1.20, 1.51) 1.23 (1.07, 1.40) 1.32 (1.17, 1.48) 5 Birth order 1.48 (1.28, 1.70) 1.30 (1.14, 1.47) 1.20 (1.05, 1.39) 1.30 (1.16, 1.46) 6 Birthweight z score 1.62 (1.40, 1.87) 1.48 (1.32, 1.66) 1.32 (1.15, 1.52) 1.36 (1.20, 1.53) 7 Childhood height z score 1.56 (1.35, 1.79) 1.46 (1.30, 1.64) 1.28 (1.11, 1.48) 1.36 (1.21, 1.54) 8 Childhood body mass index z score 1.63 (1.42, 1.87) 1.49 (1.33, 1.67) 1.29 (1.12, 1.48) 1.33 (1.18, 1.50) 9 Cognitive function at 7 y 1.38 (1.19, 1.60) 1.34 (1.19, 1.51) 1.20 (1.04, 1.39) 1.23 (1.08, 1.39) 10 Cognitive function at 11 y 1.28 (1.10, 1.48) 1.24 (1.10, 1.41) 1.14 (0.98, 1.32) 1.17 (1.03, 1.32) 11 Parental smoking 1.54 (1.34, 1.76) 1.42 (1.26, 1.59) 1.22 (1.06, 1.40) 1.29 (1.15, 1.45) 12 Educational attainment 1.05 (0.90, 1.22) 1.02 (0.90, 1.15) 1.03 (0.89, 1.20) 1.15 (1.02, 1.31) 13 All except educational attainment 1.20 (1.03, 1.40) 1.17 (1.02, 1.34) 1.17 (1.01, 1.38) 1.20 (1.05, 1.38) 14 All 1.01 (0.85, 1.20) 1.01 (0.88, 1.17) 1.05 (0.89, 1.24) 1.15 (1.00, 1.33)

Note.OR=odds ratio; CI=confidence interval.All models were adjusted for age and gender.An interaction term between income and gender was included for the current smoking and overweight variables.

In the case of each of these indicators, there tors after adjustment for potential confound- of the associations to a smaller degree than was a linear incremental decrease in height ing and mediating variables. After adjustment did the effect of educational attainment and in childhood and adulthood with decreasing for gender and age, manual (vs nonmanual) did not remove the independent effects of SEP. Social class at birth was more strongly social class status at birth was associated with childhood social class on these outcomes. related to childhood growth and height than increased odds of current or ever smoking in Cognitive function at both ages 7 (model 9) education or adult income (both P values adulthood, binge drinking, and being over- and 11 (model 10) years attenuated the asso- for differences in regression coefficients, weight. Of the potential mediating factors, ad- ciations, although some effects remained. estimated with z scores, were less than .01). justment for educational attainment had the Family size, birth order, birthweight for gesta- Adult income exhibited the strongest inverse most profound effect, completely eliminating tional age, childhood height and body mass association with adult height (all P values for the association between childhood manual so- index, and parental smoking had little effect the difference between this regression coeffi- cial class status and smoking and binge drink- on the associations between childhood social cient and the other 3 coefficients were less ing and attenuating the association between class and adult behavioral risk factors. than .001). manual social class status and being over- To further illustrate the impact of educa- Table 4 shows the association of social weight (model 12). Adult social class (model 2) tion, we compared (1) the OR for manual (vs class at birth with adult behavioral risk fac- and income (model 3) lowered the magnitude nonmanual) social class at birth for each risk

1248 | Research and Practice | Peer Reviewed | Lawlor et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

factor after adjusting for all of the remaining importantly attenuated the association be- flect true disposable income, because individ- potential confounding and mediating factors tween childhood SEP and being overweight uals with spouses will vary widely in regard (model 13) and (2) the ORs obtained when in adulthood. to household income. This may explain the educational attainment was subsequently interactions between gender and income that added to the models (model 14). In the case Study Limitations we observed in the case of some of the out- of all outcomes, there was evidence that Although the majority of eligible survivors comes. To address this issue, we included an some association remained when all of the co- (64%) responded to the questionnaire mailed interaction term between gender and income variates other than education were included. in 2000, important differences were found in the multivariate models. With the additional inclusion of education, between responders and nonresponders. Adult body mass index was determined some of the association with being overweight Prevalence rates of smoking and overweight from self-reported weight and height, which remained, but the associations with all other were higher among Scottish Health Survey have been shown to be strongly correlated outcomes were eliminated. respondents aged 45 to 54 years than with direct measurements.25–27 However, de- When RII scores were estimated for among respondents in our cohort (Table 1).24 spite these strong correlations, indicative that childhood SEP, education, adult income, These differences may reflect socioeconomic self-reports and direct measurements result and social class in these models, the pat- differentials between the 2 groups of respon- in similar height and weight values, individu- terns of the associations and effects of po- dents, real differences in smoking and over- als who are obese tend to underestimate their tential mediating factors were similar to those weight rates between Aberdeen and the rest body mass index. 25–27 If this systematic mis- shown in Table 4. For example, after adjust- of Scotland, or both.24 Differences between reporting of weight was similar across social ment for all covariates other than education responders and nonresponders in our study class groups in our study, it would have (equivalent to model 13 in Table 4), the RII would have resulted in exaggerated associa- tended to dilute rather than exaggerate the for current smoking associated with social tions being observed between SEP at birth magnitude of the associations observed. Any class at birth was 1.35 (95% CI=1.03, 1.72); and adult behaviors only if there were no variations in misreporting according to social after additional adjustment for educational associations among nonresponders or these class could have biased our results in either attainment (equivalent to model 14), this associations were in the opposite direction direction. We found that participants who RII decreased to 1.19 (95% CI = 0.92, (i.e., nonresponders in more affluent family indicated not using scales to estimate their 1.54). Results were similar for binge drink- circumstances at the time of their birth weight reported slightly lower weights on ing (corresponding RIIs of 1.23 [95% CI = were more likely to smoke, binge drink, and average, and this difference did not vary ac- 1.01, 1.55] and 1.04 [95% CI = 0.79, 1.35]) be overweight later in life) from those cording to childhood or adult social class. and being overweight (corresponding RIIs of among responders. This result provided some evidence that 1.37 [95% CI = 1.08, 1.72] and 1.19 [95% Similarly, for the role of education in this misreporting of weight did not differ accord- CI=0.93, 1.49]). association to be exaggerated by response ing to social class. bias, one would have to assume that among Data on parental smoking were not collected DISCUSSION nonresponders the association between low during the original survey, so we relied on in- SEP at birth and low educational attainment formation obtained from the 2000–2002 Consistent with previous work,4,5,9,12,13 the or the association between low educational questionnaire. If there were no differences results of this study show that low SEP at attainment and adult risk behaviors was non- in misreporting with respect to social class birth is associated with adverse behavioral existent or in the opposite direction from that at birth, our results may have underesti- CVD risk factors (smoking, binge drinking, among responders. While we cannot rule out mated the effects of parental smoking on and being overweight) independent of adult these possibilities, they seem unlikely. the association between social class at social class and income. These associations A strength of this study is the use of a mea- birth and adult CVD risk factors. This may are sometimes regarded as commonplace, sure of childhood SEP assessed at study initi- in part explain why adjustment for adult but they are in fact remarkable. Simply on ation rather than one retrospectively reported smoking had very little effect on the associ- the basis of knowledge of the occupation in adulthood. However, a weakness is that we ation between social class at birth and of a participant’s father at the time of the par- included only 1 measure of childhood SEP: smoking in adulthood. ticipant’s birth, one could have predicted father’s social class. Such a single measure is The participants in this cohort are too whether this individual would be likely to unlikely to encompass the entire spectrum of young to have experienced a sufficient num- smoke, engage in binge drinking, and be childhood social circumstances, and its effect ber of CVD events to allow determination of overweight 40 years after he or she took part on adult risk factors—and, therefore, our the pathways between childhood SEP and in the initial survey in 1962. results—may have led to underestimations adult disease events. However, by continuing The present associations between SEP at of the true magnitude of this association. to conduct follow-ups with these individuals, birth and smoking and binge drinking were Our measure of adult income was based we will be able to obtain important informa- largely explained by educational attainment. on individuals rather than households. tion on the roles of cognitive functioning, edu- Adjustment for educational attainment also Among women in particular, this may not re- cational attainment, and CVD risk factors in

July 2005, Vol 95, No. 7 | American Journal of Public Health Lawlor et al. | Peer Reviewed | Research and Practice | 1249  RESEARCH AND PRACTICE 

the associations between childhood SEP and Kingdom that survey participants may be Contributors risk of CVD in adulthood. reluctant to provide information on income D.A. Lawlor and D.A. Leon developed the idea for this article. D.A. Lawlor undertook the statistical analysis, and that, when they do, the information is wrote the first draft of the article, and coordinated the Public Health Implications inaccurate33; however, this observation has writing of the article. H. Clark managed the study. All of Our results concerning CVD risk factors been disputed.34 the authors were involved in the writing of the article. are consistent with a systematic review of Our results suggest that, rather than mate- previous studies indicating that childhood so- rial resources, other factors related to higher Acknowledgments The Aberdeen Children of the 1950s Study was funded cioeconomic adversity is associated with ad- educational achievement most likely explain as a component project of the Medical Research Coun- verse behavioral risk factors both at young the association between childhood SEP and cil (grant G0828205). Debbie A. Lawlor was funded ages28 and with other studies showing simi- behavioral risk factors in adulthood. The by a United Kingdom Department of Health Career Sci- entist Award, and G. David Batty was supported by a lar associations with adverse risk factors in sociocultural characteristics of those at higher fellowship from the University of Copenhagen. adulthood.4,5,7,9,11,12 Previous studies also educational levels, for example self-confidence We are very grateful to Raymond Illsley for pro- have demonstrated associations of educa- and ability to access and understand health viding us with the data from the Aberdeen Child 29–31 Development Survey and for his advice about the tional attainment with CVD risk factors. promotional materials, may be relevant. More- study. Graeme Ford played a crucial role in identify- The importance of our results resides in over, people’s behaviors with respect to ing individual cohort members and in helping us demonstrating the role of educational attain- tobacco and alcohol consumption, diet, and initiate the process of revitalizing the cohort. Doris Campbell, George Davey Smith, Marion Hall, Bianca ment as a mediator in the association be- physical activity, which will affect their body de Stavola, David Godden, Di Kuh, Glyn Lewis, and tween childhood SEP and adult behavioral mass index, are likely to be influenced by Viveca Östberg collaborated with the authors to re- risk factors. Given that cigarette smoking, their peers. Educational experiences will de- vitalize the cohort. Margaret Beveridge assisted in study management. binge drinking, and being overweight are termine one’s peers at the sensitive life course We also thank staff at the Information and Statistics known to increase CVD risk, our results sug- periods (late adolescence and early adulthood) Division (Edinburgh), the General Regional Office gest that a pathway leading from childhood during which these behaviors tend to be (Scotland), and the National Health Service Central Register (Southport, England) for their substantial con- socioeconomic adversity to low educational adopted. tributions and John Lemon, who undertook the linkage attainment and adverse CVD risk factors In conclusion, we have shown that child- to the Aberdeen Maternity and Neonatal Databank. may explain in part the association between hood SEP is associated with adult CVD risk Finally, we thank the study participants who responded to a mailed questionnaire 40 years after the original childhood SEP and CVD. factors. It is notable that the father’s occupa- survey was completed. The role of educational attainment in the tion when participants were born could have Note. The views expressed in this article are those association between childhood SEP and predicted which participants were most of the authors and not necessarily those of any funding body. adult behaviors could be explained via a likely to be smokers and indulge in other number of pathways. Educational attainment risky behaviors in adulthood. Educational Human Participant Protection will itself be influenced by childhood SEP, attainment appears to largely explain these The revitalization of the Aberdeen Children of the and its effect on the association between so- associations. Our findings suggest that pro- 1950s Study cohort was approved by the Scottish cial class at birth and adult behaviors may grams aimed at improving educational at- multicenter research ethics committee and local research ethics committees, along with the Scottish Pri- indicate its value as a measure of childhood tainment may be important in enhancing vacy Advisory Committee. Participants responding to SEP. The association between social class at health behaviors and therefore reducing the questionnaire provided informed consent to be in- birth and childhood growth and height was CVD risk. volved in the study. stronger than that between education and References childhood growth, suggesting that our mea- About the Authors 1. Notkola V, Punsar S, Karvonen MJ, Haapakoski J. sure of social class at birth was a better Socio-economic conditions in childhood and mortality Debbie A. Lawlor is with the Department of Social Medi- and morbidity caused by coronary heart disease in indicator of childhood socioeconomic cir- cine, University of Bristol, Bristol, England. At the time of adulthood in rural Finland. Soc Sci Med. 19 85;21: cumstances than educational attainment. this study, G. David Batty was with the Department of So- 517–523. Educational attainment is associated with cial Medicine, Institute of Public Health, University of Co- penhagen, Copenhagen, Denmark. Susan M.B. Morton is 2. Kaplan GA, Salonen JT. Socioeconomic conditions adult occupation and income, and thus it with the School of Population Health, University of Auck- in childhood and ischaemic heart disease during mid- may reflect the availability of material re- land, Auckland, New Zealand. Heather Clark is with the dle age. BMJ. 1990;301:1121–1123. sources, which are thought to be important Dugald Baird Centre for Research on Women’s Health, 3. Gliksman MD, Kawachi I, Hunter D, et al. Child- University of Aberdeen, Aberdeen, Scotland. Sally Macin- determinants of health outcomes.32 hood socioeconomic status and risk of cardiovascular tyre is with the MRC Social and Public Health Sciences disease in middle aged US women: a prospective study. We found that neither adult social class Unit, University of Glasgow, Glasgow, Scotland. David A. J Epidemiol Community Health. 19 95;49:10–15. nor income fully explained the associations Leon is with the Department of Epidemiology and Popula- tion Health, London School of Hygiene and Tropical Medi- 4. Blane D, Hart CL, Davey Smith G, Gillis CR, between childhood social class and adult risk cine, London, England. Hole DJ, Hawthorne VM. Association of cardiovascular factors, whereas education did explain these Requests for reprints should be sent to Debbie A. disease risk factors with socioeconomic position during childhood and during adulthood. BMJ. 1996;313: associations. Income is arguably the best Lawlor, Department of Social Medicine, University of Bristol, Canynge Hall, Whiteladies Road, Bristol BS7 14 3 4–1438. single indicator of material living standards, 8QA, United Kingdom (e-mail: [email protected]). 5. Wannamethee SG, Whincup PH, Shaper G, but there is some evidence in the United This article was accepted July 13, 2004. Walker M. Influence of fathers’ social class on cardio-

1250 | Research and Practice | Peer Reviewed | Lawlor et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

vascular disease in middle-aged men. Lancet. 19 9 6; ical and Epidemiologic Study. Baltimore, Md: Williams & 348:1259–1263. Wilkins; 1970. 6. Davey Smith G, Hart C, Blane D, Hole D. Adverse 21. Batty GD, Clark H, Morton SMB, Macintyre S, socioeconomic conditions in childhood and cause spe- Leon DA. Intelligence in childhood and mortality, mi- cific adult mortality: prospective observational study. gration, questionnaire response rate, and self-reported BMJ. 1998;316:1631–1635. morbidity and risk factor levels in adulthood—preliminary 7. Frankel S, Davey Smith G, Gunnell D. Childhood findings from the Aberdeen ‘Children of the 1950s’ socioeconomic position and adult cardiovascular mor- Study. J Epidemiol Community Health. 2002;56 tality: the Boyd Orr cohort. Am J Epidemiol. 1999;150: (suppl 2):A1. 10 81–1084. 22. Definition, Diagnosis and Classification of Diabetes 8. Davey Smith G, McCarron P, Okasha M, McEwen J. Mellitus and Its Complications: Report of a WHO Consul- Social circumstances in childhood and cardiovascular tation. Part 1: Diagnosis and Classification of Diabetes disease mortality: prospective observational study of Mellitus. Geneva, Switzerland: World Health Organiza- Glasgow University students. J Epidemiol Community tion; 1999. Health. 2001;55:340–341. 23. Mackenbach JP, Kunst AE. Measuring the magni- 9. Heslop P, Davey Smith G, Macleod J, Hart C. The tude of socio-economic inequalities in health: an over- socioeconomic position of women, risk factors and view of available measures illustrated with two exam- mortality. Soc Sci Med. 2001;53:477–485. ples from Europe. Soc Sci Med. 19 97;44:757–771. 10.Wamala SP, Lynch J, Kaplan GA. Women’s expo- 24. Shaw A, McMunn A, Field J. The Scottish Health sure to early and later life socioeconomic disadvantage Survey 1998. Edinburgh, Scotland: Scottish Executive and coronary heart disease risk: the Stockholm Female Dept of Health; 2000. Coronary Risk Study. Int J Epidemiol. 2001;30: 25. Rowland ML. Self-reported weight and height. 275–284. Am J Clin Nutr. 19 9 0;52:1125–1133. 11. Lawlor DA, Ebrahim S, Davey Smith G. Socioeco- 26. Stevens J, Keil JE, Waid LR, Gazes PC. Accuracy nomic position in childhood and adulthood and insulin of current, 4-year, and 28-year self-reported body resistance: cross sectional survey using data from the weight in an elderly population. Am J Epidemiol. 1990; British Women’s Heart and Health Study. BMJ. 2002; 132:1156–1163. 325:805. 27. Lawlor DA, Taylor M, Bedford C, Ebrahim S. 12.Poulton R, Caspi A, Milne BJ, et al. Association Agreement between measured and self-reported between children’s experience of socioeconomic disad- weight in older women: results from the British vantage and adult health: a life-course study. Lancet. Women’s Heart and Health Study. Age Ageing. 2001; 2002;360:1640–1645. 31:169–174. 13. Lawlor DA, Davey Smith G, Ebrahim S. The association of childhood socioeconomic position with coro- 28. Batty D, Leon D. Socio-economic position and nary heart disease risk in post-menopausal women: coronary heart disease risk factors in children and findings from the British Women’s Heart and Health young people. Eur J Public Health. 2002;12:263–272. Study. Am J Public Health. In press. 29. Matthews KA, Kelsey SF, Meilahn EN, Kuller LH, 14 . Claussen B, Davey Smith G, Thelle D. Impact of Wing RR. Educational attainment and behavioral and childhood and adulthood socioeconomic position on biologic risk factors for coronary heart disease in middle- cause specific mortality: the Oslo Mortality Study. aged women. Am J Epidemiol. 19 8 9;129:1132–1144. J Epidemiol Community Health. 2003;57:40–45. 30. Cirera L, Tormo MJ, Chirlaque MD, Navarro C. 15. Hardy R, Kuh D, Langenberg C, Wadsworth ME. Cardiovascular risk factors and educational attainment Birthweight, childhood social class, and change in adult in southern Spain: a study of a random sample of blood pressure in the 1946 British birth cohort. Lancet. 3091 adults. Eur J Epidemiol. 1998;14:755–763. 2003;362:1178–1183. 31. Kubzansky LD, Berkman LF, Glass TA, Seeman TE. 16. Looker ED. Accuracy of proxy reports of parental Is educational attainment associated with shared deter- status characteristics. Sociol Educ. 1989;62:257–276. minants of health in the elderly? Findings from the MacArthur Studies of Successful Aging. Psychosom 17. Oldman D, Bytheway B, Horobin G. Family struc- Med. 1998;60:578–585. ture and educational achievement. J Biosoc Sci Suppl. 1971;3:81–91. 32. Lynch JW, Davey Smith G, Kaplan GA, House JS. Income inequality and mortality: importance to health 18. Batty GD, Morton SMB, Campbell D, et al. The of individual income, psychosocial environment or ma- Aberdeen Children of the 1950s Cohort Study: back- terial conditions. BMJ. 2000;320:1200–1204. ground, methods, and follow-up information on a new resource for the study of life-course and intergenera- 33. Turrell G. Income non-reporting: implications for tional effects on health. Paediatr Perinatal Epidemiol. health inequalities research. J Epidemiol Community 2004;18:221–239. Health. 2000;54:207–214. 19. Illsley R, Wilson F. Longitudinal studies in Ab- 34. Dorling D. Who’s afraid of income inequality? erdeen, Scotland: C. The Aberdeen Child Development Environ Plann. 19 99;31:571–574. Survey. In: Mednick S, Baert A, Bachmann B, eds. Pro- spective Longitudinal Research: An Empirical Basis for the Primary Prevention of Psychosocial Disorders. Oxford, England: Oxford University Press Inc; 1981:66–68. 20. Birch HG, Richardson SA, Baird D, Horobin G, Illsley R. Mental Subnormality in the Community: A Clin-

July 2005, Vol 95, No. 7 | American Journal of Public Health Lawlor et al. | Peer Reviewed | Research and Practice | 1251  RESEARCH AND PRACTICE 

The Health of Poor Women Under Welfare Reform

| George A. Kaplan, PhD, Kristine Siefert, PhD, Nalini Ranjit, PhD, Trivellore E. Raghunathan, PhD, Elizabeth A. Young, PhD, Diem Tran, MPH, Sandra Danziger, PhD, Susan Hudson, RN, MSN, MS, John W. Lynch, PhD, and Richard Tolman, PhD

Despite previous research indicating that Objectives. We compared the health of single mothers affected by welfare re- people who are poorer have worse health, form with the health of a nationally representative sample of women to document and a recent exponential increase in such re- the prevalence of poor health as single mothers experience the effects of welfare 1,2 search, relatively little attention has been reform. given to the health of one of the poorest seg- Methods. We compared risk factors and measures of health among women ments of the population—single mothers re- randomly sampled from the welfare rolls with similar data from a nationally rep- ceiving welfare. Because these women experi- resentative sample of women. ence chronic exposure to economic, social, Results. Women in our welfare recipient sample had higher rates of elevated ≥ and environmental stresses and have few re- glycosylated hemoglobin ( 6%; prevalence ratio [PR]=4.87; 95% confidence in- ≥ ≥ sources with which to reduce these stresses, terval [CI]=2.69, 7.04), hypertension (systole 140 or diastole 90; PR=2.36; 95% CI=1.47, 3.24), high body mass index (≥ 30; PR=1.78; 95% CI=1.49, 2.08), and we have every reason to believe that their high-density lipoprotein cholesterol (≤35 mg/dL; PR=1.91; 95% CI=1.17, 2.65); health is at risk. The introduction of the Per- lower peak expiratory flow; and less physical functioning. Current smoking rates sonal Responsibility and Work Opportunity were higher (PR = 1.85; 95% CI = 1.50, 2.19) and smoking cessation rates were Reconciliation Act in 1996, which dramati- lower (PR=0.62; 95% CI=0.37, 0.86) than in the national sample. cally altered cash assistance for poor families Conclusions. Current and former welfare recipients bear a substantial burden with children, may have led to changes in the of illness. Further studies are necessary to interpret our findings of worsened health of that population. health in the wake of welfare reform. (Am J Public Health. 2005;95:1252–1258. Welfare reform has been touted as a great doi:10.2105/AJPH.2004.037804) success by its proponents. Welfare caseloads more than halved between 1996 and 2000.3 were based on self-reported measures of Michigan in February of 1997.18 To partici- This reduction has been attributed to the physical health or self-reported medical diag- pate in the study, women were required to introduction of a new program called Tempo- noses, either of which may cause many inter- have been county residents enrolled in wel- rary Aid to Needy Families (TANF) and its as- pretive problems. fare in February 1997, to be single mothers sociated “welfare-to-work” provisions, as well In our study, we estimated the prevalence of with children, to be US citizens aged 18–54 as to a period of rapid economic expansion.4,5 measured health problems, disease markers, years, and to report a racial identity of White Still, the economic and social status of many and important risk factors, as well as of self- or African American. The response rate was current and former welfare recipients and reported medical conditions, in a population- 86.2% among the first wave (753 of 874). their children, who all are part of this “natural based sample of poor mothers who were re- The original data collection was conducted experiment,” remains deeply at risk.6,7 Al- ceiving cash assistance immediately following from September through December of though the proportion of poor women who implementation of TANF, and we compared 19 97, with a second wave of data collection are working has dramatically increased, many these prevalence levels with levels in a con- during the fall of 1998, a third wave during such women remain below the poverty level temporary, nationally representative sample the fall of 1999, and a fourth wave during and in need of government assistance.8 Typi- of women matched for age and race. We also the fall of 2001. The number of respon- cally, the new jobs held by these women in- compared health indicators in the welfare dents and response rates among the last 3 volve constantly changing work schedules, sample with indicators in a comparable na- waves were 693 (92%), 632 (92%), and less than full-time work opportunities, few or tionally based sample surveyed before wel- 577 (91%). no fringe benefits, and long commutes.9–11 fare reform was enacted. Face-to-face interviews of approximately Employment gains often are constrained by 60–90 minutes’ duration were conducted in low levels of skill and lack of prior work ex- METHODS the respondent’s home and gathered informa- perience, as well as by child care and trans- tion on employment histories, income from portation needs.12 We know very little about Study Sample various sources, barriers to work, child- and the health of these women, let alone the ef- The participants in this ongoing panel family-related stressors, trauma, neighbor- fects of welfare reform on their health. The study came from the Women’s Employment hood circumstances and living conditions, existing evidence generally suggests a popu- Study (WES), a random sample (n=753) of mental health, self-report of health, self-report lation that is not as healthy as the general all single mothers with children who were re- of diagnoses of a variety of health conditions, population.13–17 Previous studies, however, ceiving cash benefits in an urban county in smoking behavior, and physical functioning.

1252 | Research and Practice | Peer Reviewed | Kaplan et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

In June 2000, after completion of the third equal to or greater than 0.8 was classified as 2 groups and is the appropriate parameter wave of WES data collection, a health supple- a risk factor. Physical functioning was as- with which to express a cross-sectional comment (WES-HS) was administered to the sur- sessed with the SF-36 subscale21 and comparison of 2 groups. vey respondents. Of the 632 eligible respon- pared with national standards scored by the For peak expiratory flow, comparable infor- dents, 299 completed the survey. After those RAND method.22 Peak expiratory flow (high- mation was not available in NHANES, so who had moved out of the county (n=19) or est of 3 expirations) was compared with na- comparisons were made with age- and height- could not be located (n=35) were excluded, tional norms.23 adjusted norms. For comparisons of physical the response rate was calculated as 52% function, we used the national standards for (299 of 578). Demographics, self-reported Self-Reported Measures of Health women aged 20–54 years reported in the health status, and physical limitations of the WES respondents were asked whether Medical Outcomes Study.21,22 299 respondents were quite similar to data they had ever been diagnosed by a doctor as We used SUDAAN software (Research Tri- for the full Wave 3 sample, although respon- having hypertension, diabetes, breathing angle Institute, Research Triangle Park, NC) dents were slightly younger (mean age=30.0 problems, arthritis/rheumatism, or bone prob- to compute standard errors, and corrected for vs 31.4 years, P<.05). A comparison of the lems. The WES questions were less specific spatial correlations in NHANES 1999–2000 Wave 1 sample with the Wave 3 sample and about particular health conditions than were data with a jackknife “leave-one-out” proce- with the WES-HS sample did not indicate the National Health and Nutrition Examina- dure with 52 replicate weights24 that accom- any important differences. tion Survey (NHANES) questions, and panied the data. For analyses that used NHANES questions that were comparable to NHANES III, design-consistent estimates of Measures the WES items were combined. Thus, a mea- variance were obtained with a Taylor-series Blood pressure, body measurements, and sure equivalent to a WES physician diagnosis linearization procedure implemented with peak expiratory flow were recorded. Blood of breathing problems was obtained by com- Stata (Stata Corp, College Station, Tex). All pressure and pulse were obtained with an auto- bining information from NHANES responses estimates were derived with weights provided mated oscillographic device (Omron HEM-737; to 3 separate questions about asthma, chronic by the National Center for Health Statistics to Omron Healthcare Inc, Kyoto, Japan). Read- bronchitis, and emphysema. The WES survey account for unequal probabilities of selection ings with this device have been found to be question, “Has a doctor or health professional and nonresponse. highly comparable to those obtained with ever told you that you had arthritis, rheuma- mercury devices.19 Three readings, 1 minute tism, or bone problems?” was approximated RESULTS apart, were obtained after a 5-minute rest. by combining responses to 2 questions from Hip and waist circumferences, weight in light NHANES that asked about arthritis and Table 1 presents the age and racial distribu- clothing, and height without shoes were ob- osteoporosis. The global measure of self-rated tions of the WES-HS sample and the NHANES tained with standard techniques. Approxi- health (excellent, very good, good, fair, or 1999–2000 sample and NHANES III respon- mately 30 mL of blood were drawn, either in poor) was identically worded in the NHANES dents who reported receipt of welfare. a laboratory (57%) or at the home of the par- and WES surveys. Table 2 presents the age-specific preva- ticipant (43%). Glycosylated hemoglobin, total lence of risk factors and disease. WES-HS cholesterol, high-density lipoprotein choles- Analyses women were 1.35 times more likely than terol (HDLC), and high-sensitivity C-reactive We compared data from WES-HS respon- women in NHANES 1999–2000 to have protein were assayed as milligrams per dents with data from women of the same ever smoked (63% vs 47%; 95% confi- deciliter. Peak expiratory flow was measured age and race in NHANES 1999–2000 or dence interval [CI] = 1.18, 1.52) and 1.85 3 times with a full-range peak flowmeter NHANES III. Measures for waist–hip ratio times more likely to be current smokers (Zoey Personal Best Peak Flow Meter Stan- and HDLC were taken from the second phase (51% vs 27%; 95% CI = 1.50, 2.19), an indi- dard Range; Zoey L.P., San Antonio, Texas). (1991–1994) of NHANES III, because this cation of lower levels of smoking cessation phase could provide the most recent available within the WES-HS population. Black Clinical Indicators measures. For analyses that compared the women were generally less likely than Hypertension (systolic blood pressure WES-HS women with the pre-TANF welfare White women to be current smokers or to ≥ 14 0 or diastolic pressure ≥ 90), high choles- population in NHANES, we used both phases have ever smoked. terol ( ≥ 240 mg/dL), low HDLC (< 35 mg/ of NHANES III (1988–1994) to ensure ade- With regard to anthropometric measure- dL), high C-reactive protein (≥1 mg /dL), quate numbers. All comparisons with ments, BMI was 16% greater in the WES-HS and high glycosylated hemoglobin (≥6%) NHANES were standardized by age to the sample (32.6) than in the NHANES popula- were measured. Obesity was defined as a 2000 census with the direct method. The ration (28.0). Women in the WES-HS were also body mass index (BMI; weight in kilograms tios of age- and race-standardized preva- 1.8 times more likely than those in NHANES divided by height in meters squared) of 30.0 lences (i.e., the prevalence ratios) were used to be obese (56% vs 32%; prevalence ratio or more, and overweight was defined as a to compare the 2 samples. The prevalence [PR]=1.78; 95% CI=1.49, 2.08). Obesity BMI of 26.0 or more.20 A waist–hip ratio ratio is simply the ratio of prevalence in the levels among White WES-HS participants

July 2005, Vol 95, No. 7 | American Journal of Public Health Kaplan et al. | Peer Reviewed | Research and Practice | 1253  RESEARCH AND PRACTICE 

TABLE 1—Age and Race Distributions TABLE 2—Prevalence of Risk Factors and Health Status Indicators Among Women in the for Women in the Health Supplement to Health Supplement to the Women’s Employment Study (WES-HS) 2000–2001 and in the the Women’s Employment Study National Health and Nutrition Examination Survey (NHANES) 1999–2000a (WES-HS) 2000–2001, and the Age-Standardized Prevalence Ratio National Health and Nutrition Prevalence Rate (95% CI) (WES-HS/NHANES) (95% CI) Examination Survey (NHANES) 1999–2000, and for Women Reporting All women Welfare Receipt in NHANES III for NHANES (n=973) WES-HS (n=299) 1998–1994 Smoking status Ever 0.47 (0.43, 0.50) 0.63 (0.57, 0.69) 1.35 (1.18, 1.52) NHANES III Past 0.19 (0.16, 0.22) 0.12 (0.08, 0.16) 0.62 (0.37, 0.86) WES-HS, NHANES, Welfare Current 0.27 (0.24, 0.31) 0.51 (0.44, 0.57) 1.85 (1.50, 2.19) % % Population, Obesity and overweight (n=299) (n=973) % (n=359) BMI, mean 28.0 (27.4, 28.7) 32.6 (31.5, 33.7) 1.2 (1.1, 1.2) Age, y Obese (BMI ≥30) 0.32 (0.28, 0.35) 0.56 (0.50, 0.63) 1.78 (1.49, 2.08) 20–29 35 26 46 Overweight (BMI ≥25) 0.57 (0.53, 0.61) 0.79 (0.74, 0.84) 1.38 (1.24, 1.51) 30–39 43 31 35 Waist–hip ratio >0.8 0.74 (0.72, 0.76) 0.72 (0.66, 0.78) 0.97 (0.89, 1.05) 40–56 21 43 19 Hypertension (systolic ≥140 0.09 (0.07, 0.12) 0.22 (0.16, 0.27) 2.36 (1.47, 3.24) Race or diastolic ≥90) White 47 84 55 High cholesterol (≥240 mg/dL) 0.14 (0.10, 0.18) 0.11 (0.06, 0.15) 0.75 (0.38, 1.12) Black 53 16 45 Low HDL (≤35 mg/dL) 0.06 (0.05, 0.07) 0.11 (0.07, 0.15) 1.91 (1.17, 2.65) Age, y, by race C-reactive protein (≥1 mg/dL) 0.14 (0.11, 0.16) 0.18 (0.13, 0.23) 1.30 (0.86, 1.74) White Glycosylated hemoglobin ≥6% 0.05 (0.03, 0.06) 0.22 (0.17, 0.28) 4.87 (2.69, 7.04) 20–29 32 26 48 Doctor’s diagnosis of 30–39 46 30 39 Hypertension 0.17 (0.14, 0.19) 0.19 (0.14, 0.25) 1.14 (0.78, 1.50) 40–56 22 44 13 Diabetes 0.03 (0.02, 0.04) 0.08 (0.04, 0.11) 2.95 (1.17, 4.73) Black Breathing problems 0.20 (0.17, 0.23) 0.25 (0.19, 0.31) 1.27 (0.92, 1.62) 20–29 38 25 43 Arthritis 0.16 (0.13, 0.19) 0.37 (0.31, 0.44) 2.37 (1.76, 2.98) 30–39 42 37 31 Self-reported health poor/fair 0.14 (0.11, 0.17) 0.39 (0.33, 0.46) 2.79 (2.10, 3.49) White women 40–56 21 37 26 NHANES (n=650) WES-HS (n=140) Smoking status (55%) were nearly twice those among White Ever 0.49 (0.45, 0.53) 0.67 (0.58, 0.76) 1.37 (1.15, 1.59) NHANES participants (28%). By contrast, a Past 0.21 (0.17, 0.24) 0.10 (0.05, 0.16) 0.50 (0.23, 0.78) much smaller difference was found between Current 0.28 (0.24, 0.33) 0.57 (0.48, 0.66) 2.00 (1.55, 2.45) Black WES-HS and Black NHANES partici- Obesity and overweight pants (58% vs 49%; PR=1.18; 95% CI= BMI, mean 27.4 (26.7, 28.1) 31.8 (30.3, 33.4) 1.2 (1.1, 1.2) 0.95, 1.41). Waist–hip ratios did not vary be- Obese (BMI ≥30) 0.28 (0.24, 0.33) 0.55 (0.45, 0.64) 1.93 (1.48, 2.37) tween the 2 populations. Overweight (BMI ≥25) 0.54 (0.49, 0.58) 0.75 (0.66, 0.83) 1.39 (1.19, 1.59) Hypertension was 2.4 times more likely in Waist–hip ratio >0.8 0.72 (0.70, 0.74) 0.69 (0.60, 0.77) 0.95 (0.83, 1.08) the WES-HS sample (95% CI=1.47, 3.24), Hypertension (systolic ≥140 0.08 (0.05, 0.10) 0.19 (0.11, 0.26) 2.45 (1.10, 3.80 with White women mainly accounting for the or diastolic ≥90) difference. Prevalence of elevated levels of High cholesterol (>240 mg/dL) 0.14 (0.10, 0.18) 0.13 (0.06, 0.20) 0.91 (0.35, 1.47) total cholesterol (≥240 mg/dL) was similar Low HDL (≤35 mg/dL) 0.06 (0.05, 0.06) 0.19 (0.12, 0.27) 3.45 (2.01, 4.90) in the 2 populations (11% vs 14%; 95% CI= C-reactive protein (≥1 mg/dL) 0.14 (0.11, 0.16) 0.14 (0.08, 0.21) 1.04 (0.51, 1.57) 0.38, 1.12), but the prevalence of low levels Glycosylated hemoglobin ≥6% 0.03 (0.01, 0.05) 0.03 (0.00, 0.06) 0.93 (0.22, 2.08) of HDLC was significantly higher in the Doctor’s diagnosis of WES-HS sample (PR=1.91; 95% CI=1.17, Hypertension 0.15 (0.12, 0.18) 0.18 (0.10, 0.25) 1.17 (0.64, 1.70) 2.65) relative to the NHANES sample. White Diabetes 0.02 (0.01, 0.03) 0.12 (0.05, 0.18) 6.99 (1.19, 12.79) women in the WES-HS sample were 3 times Breathing problems 0.20 (0.16, 0.23) 0.31 (0.22, 0.40) 1.57 (1.04, 2.09) as likely as White women nationally to have Arthritis 0.16 (0.12, 0.19) 0.46 (0.37, 0.56) 2.98 (2.08, 3.88) Self-reported health poor/fair 0.12 (0.09, 0.15) 0.38 (0.29, 0.47) 3.06 (2.01, 4.10) low levels of HDLC (PR=3.45; 95% CI= 2.01, 4.90). No statistically significant differ- Continued ences were found in C-reactive protein levels

1254 | Research and Practice | Peer Reviewed | Kaplan et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

TABLE 2—Continued scribe their health as poor or fair (PR=2.79; 95% CI=2.10, 3.49), and the WES-HS–to– Black women NHANES ratio for White women was almost NHANES (n=323) WES-HS (n=159) twice as large as that for Black women (3.1 vs Smoking status 1.7). Overall, the WES-HS–to–NHANES ra- Ever 0.34 (0.29, 0.39) 0.59 (0.51, 0.67) 1.75 (1.38, 2.12) Past 0.12 (0.08, 0.15) 0.14 (0.07, 0.20) 1.18 (0.48, 1.87) tios of poor-health self-report and medical di- Current 0.22 (0.18, 0.27) 0.45 (0.36, 0.54) 2.03 (1.44, 2.62) agnoses were much larger for White women Obesity and overweight than for Black women. BMI, mean 31.7 (30.7, 32.7) 33.4 (31.8, 35.0) 1.1 (1.0, 1.1) More than half of women in the WES-HS Obese (BMI ≥30) 0.49 (0.43, 0.55) 0.58 (0.49, 0.67) 1.18 (0.95, 1.41) sample aged 40 to 56 years old had compro- Overweight (BMI ≥25) 0.76 (0.71, 0.81) 0.83 (0.77, 0.89) 1.09 (0.98, 1.20) mised peak expiratory flow (<80% of age- Waist–hip ratio>0.8 0.81 (0.80, 0.83) 0.75 (0.68, 0.83) 0.93 (0.84, 1.02) specific norms) (Table 3). However, because Hypertension (systolic ≥140 0.20 (0.14, 0.25) 0.25 (0.17, 0.33) 1.28 (0.74, 1.81) the peak expiratory flow norms are based on or diastolic ≥90) 19 83 data, caution in interpreting this result High cholesterol (≥240 mg/dL) 0.14 (0.07, 0.20) 0.08 (0.03, 0.14) 0.62 (0.10, 1.13) is warranted. Women older than 24 years in Low HDL (≤35 mg/dL) 0.07 (0.06, 0.08) 0.04 (0.01, 0.08) 0.63 (0.09, 1.17) the WES-HS sample also had significantly lower C-Reactive protein (≥1 mg/dL) 0.16 (0.11, 0.21) 0.22 (0.14, 0.30) 1.41 (0.72, 2.09) levels of physical functioning on the SF-36 sub- Glycosylated hemoglobin ≥6% 0.13 (0.09, 0.17) 0.40 (0.32, 0.49) 3.14 (1.94, 4.34) scale compared with national norms. Doctor’s diagnosis of Hypertension 0.27 (0.22, 0.33) 0.21 (0.13, 0.29) 0.77 (0.44, 1.10) Comparison of WES-HS and NHANES Diabetes 0.08 (0.04, 0.11) 0.04 (0.01, 0.07) 0.53 (0.03, 1.02) Welfare Populations Breathing problems 0.20 (0.15, 0.25) 0.20 (0.12, 0.28) 1.01 (0.55, 1.47) Table 4 shows characteristics of the WES- Arthritis 0.17 (0.13, 0.21) 0.29 (0.21, 0.38) 1.72 (1.08, 2.36 HS population and those of women in Self-reported health poor/fair 0.24 (0.19, 0.28) 0.41 (0.32, 0.49) 1.72 (1.23, 2.21) NHANES who reported receiving welfare. Note. CI=confidence interval; BMI=body mass index; HDL=high-density lipoprotein cholesterol. Compared with the women in NHANES who aFor the waist–hip ratio and low HDL measures, the most recent available data were from the second phase of NHANES III reported having received Aid to Families with (1991–1994).The sample sizes of all women,White women, and Black women for these 2 measures were n=2891, n=1740, Dependent Children benefits in the past and n=1151, respectively. month, women in WES-HS were more likely to have ever smoked (63% vs 49%; PR= overall; however, Black women in the WES-HS tis did Black women in the WES-HS have a 1.29; 95% CI=1.02, 1.55) or to be current sample tended to have (nonsignificantly) higher prevalence of reported diagnoses than smokers (51% vs 40%; PR=1.27; 95% CI= higher levels compared with Black women in their NHANES counterparts. White women in 0.95, 1.60), to have higher BMIs (32.6 vs NHANES (22% vs 16%; PR=1.41; 95% the WES-HS survey were almost 3 times as 29.2; PR=1.1; 95% CI=1.05, 1.18), to be CI=0.72, 2.09). Twenty-two percent of likely as White women in NHANES to de- obese (42% vs 56%; PR=1.33; 95% CI= WES-HS participants had glycosylated hemoglobin levels greater than 6%, almost 5 times TABLE 3—Physical Function and Peak Expiratory Flow Among Women in the Health Supplement the prevalence observed in the NHANES to the Women’s Employment Study (WES-HS) and National Reference Populations population (PR=4.87; 95% CI=2.69, 7.04). This elevation was a product of higher levels Peak Expiratory Flow, %a among Black women in the WES-HS, of Physical Functioning (SF-36) Normalc (>80% Moderate Problem Severe Problem whom 40% had elevated levels, compared Age, y US Normb WES-HS P of US Norm) (60%–80% of US Norm) (<60% of US Norm) with 13% of Black women in the NHANES 18–24 90.2 82.4 sample (PR=3.14; 95% CI=1.94, 4.34). 24–34 89.1 83.3 <.05 Similar differences in burden of illness were 34–44 88.1 74.7 <.05 found for the self-reported measures of health. 44–54 82.9 60.2 <.05 A self-reported physician diagnosis of diabetes 20–29 67 29 4 was nearly 3 times more common in the 30–39 53 34 13 WES-HS population than in the NHANES 40–56 48 44 8 population (8% vs 3%; PR=2.95; 95% CI= 1.17, 4.73). In general, White women in the Note. SF-36=Medical Outcomes Study 36-item short-form health survey.21 a WES-HS were much more likely than the Numbers are percentages of women in the WES-HS who exhibited the indicated peak expiratory flows. bNorms from baseline Medical Outcomes Study22 (n=2471), by age. NHANES population to report a physician di- cNorms based on height and age.23 agnosis of diabetes. Only in the case of arthri-

July 2005, Vol 95, No. 7 | American Journal of Public Health Kaplan et al. | Peer Reviewed | Research and Practice | 1255  RESEARCH AND PRACTICE 

TABLE 4—Prevalence of Risk Factors and Health Status Indicators Among Women in the Health welfare reform. Our definition of hyperten- Supplement to the Women’s Employment Study (WES-HS) 2000–2001 and in the Welfare sion was based solely on 3 measurements of Subpopulation of the National Health and Nutrition Examination Survey (NHANES) III, blood pressure and did not include informa- 1988–1994 tion on use of antihypertensive drugs. The decision to exclude this information was made Age-Standardized Prevalence (95% CI) on the basis of previous findings suggesting NHANES WES-HS Prevalence Ratio that socioeconomic status and race can influ- (n=359) (n=299) (WES-HS/NHANES) (95% CI) ence the likelihood of a physician diagnosis.25 Smoking status Because the NHANES sample was wealth- Ever 0.49 (0.40, 0.57) 0.63 (0.57, 0.69) 1.29 (1.02, 1.55) ier than the WES-HS sample, it is likely that a Past 0.09 (0.03, 0.15) 0.12 (0.08, 0.16) 1.32 (0.30, 2.34) larger proportion of NHANES participants Current 0.40 (0.31, 0.48) 0.51 (0.44, 0.57) 1.27 (0.95, 1.60) with hypertension than of WES-HS partici- Obesity and overweight pants with hypertension would have had ac- BMI, mean 29.2 (27.84, 30.62) 32.6 (31.49, 33.74) 1.1 (1.05, 1.18) cess to medical care that would lead to their Obese (BMI ≥30) 0.42 (0.34, 0.51) 0.56 (0.50, 0.63) 1.33 (1.02, 1.65) being diagnosed with hypertension. Compar- Overweight (BMI ≥25) 0.62 (0.53, 0.70) 0.79 (0.74, 0.84) 1.28 (1.08, 1.47) ing 2 groups with unequal access to medical Waist–hip ratio >0.8 0.82 (0.75, 0.89) 0.72 (0.66, 0.78) 0.88 (0.77, 0.98) care and thus unequal likelihood of diagnosis Hypertension (systolic ≥140 0.15 (0.09, 0.21) 0.22 (0.16, 0.27) 1.47 (0.78, 2.16) could lead to misleading conclusions. We or diastolic ≥90) chose to use a classification in both NHANES High cholesterol (≥240 mg/dL) 0.09 (0.04, 0.14) 0.11 (0.06, 0.15) 1.18 (0.31, 2.06) and WES-HS that was based only on mea- Low HDL (≥35 mg/dL) 0.09 (0.03, 0.16) 0.11 (0.07, 0.15) 1.18 (0.27, 2.10) sured blood pressure. Thus, our classification C-reactive protein (≥1 mg/dL) 0.20 (0.13, 0.27) 0.18 (0.13, 0.23) 0.90 (0.49, 1.30) actually misclassifies as nonhypertensive Glycosylated hemoglobin ≥6% 0.15 (0.10, 0.20) 0.22 (0.17, 0.28) 1.50 (0.82, 2.17) those participants in both samples whose Doctor’s diagnosis of blood pressure was controlled with medica- Hypertension 0.27 (0.20, 0.34) 0.19 (0.14, 0.25) 0.73 (0.45, 1.01) tions. In the same fashion, some of those par- Diabetes 0.05 (0.03, 0.08) 0.08 (0.04, 0.11) 1.41 (0.45, 2.37) ticipants we classified as not having elevated Breathing problems 0.29 (0.20, 0.38) 0.25 (0.19, 0.31) 0.89 (0.54, 1. 23) levels of glycosylated hemoglobin may actu- Arthritis 0.13 (0.09, 0.18) 0.37 (0.31, 0.44) 2.76 (1.66, 3.87) ally have well-treated diabetes. To some ex- Self-reported health poor/fair 0.30 (0.22, 0.38) 0.39 (0.33, 0.46) 1.31 (0.90, 1.72) tent, this approach may exaggerate the differences between the 2 groups. However, the Note.CI=confidence interval; BMI=body mass index; HDL=high-density lipoprotein cholesterol. higher prevalence of obesity in the WES sample is consistent with elevated rates of hyper- 1.02, 1.65), and to report a physician diagno- sylated hemoglobin levels, low HDLC levels, tension and glycosylated hemoglobin. sis of arthritis (37% vs 13%; PR=2.76; 95% low peak expiratory flow, low levels of physi- We compared welfare recipients in CI=1.66, 3.87). Women in the WES-HS cal functioning, and higher levels of C-reactive NHANES before enactment of welfare reform sample also had a higher prevalence of ele- protein compared with the national sample. In and welfare recipients in WES-HS to provide vated glycosylated hemoglobin levels (22% vs addition, rates of current smoking were higher some insight into possible changes in the 15%; PR=1.50; 95% CI=0.82, 2.17), hyper- and rates of smoking cessation were lower health status of welfare recipients before and tension (22% vs 15%; PR=1.47; 95% CI= than those in a national sample of women the after welfare reform. Although our data are 0.78, 2.16), and poor self-rated health (39% same age. Against this backdrop of poor consistent with a worsening of health status vs 30%; PR=1.31; 95% CI=0.90, 1.72). health, fewer physician diagnoses were re- after welfare reform, 2 limitations to our However, women in the WES-HS sample had ported than would be expected from the mea- study should be considered before such a a lower prevalence of waist–hip ratio ≥ 0.8 sured values. conclusion is firmly drawn. First, the small (0.72 vs 0.82; PR=0.88; 95% CI=0.77, Although a single measure of glycosylated number of NHANES participants who were 0.98) and were less likely to report a physi- hemoglobin or C-reactive protein, or mea- welfare recipients forced us to combine data cian diagnosis of hypertension (19% vs 27%; sures of blood pressure on a single occasion, collected over 6 years (1988–1994), thereby PR=0.73; 95% CI=0.45, 1.01). is not sufficient to diagnose diabetes, hyper- masking any ongoing secular trends. tension, or other diseases, in prospective epi- Second, we compared a national sample DISCUSSION demiological studies, single measures have (NHANES) with a sample from a single urban been shown to be predictive of high risk of county in a Midwestern state (WES-HS). In this sample of current and former wel- adverse health outcomes. Nevertheless, single Other factors aside from welfare reform may fare recipients, we found significantly higher measures cannot definitively characterize the differ between these groups. For example, at rates of hypertension, obesity, elevated glyco- clinical status of this sample of women under the time the sample was drawn, WES-HS par-

1256 | Research and Practice | Peer Reviewed | Kaplan et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

ticipants had been receiving cash assistance sive. The absence of well-designed studies of References for an average of 7.4 years; however, we do the physical health status of poor women and 1. Kaplan GA, Haan MN, Syme SL, et al. Socioeco- nomic status and health. In: Amler RW, Dull HB, eds. not have similar information on length of wel- their families as they weather one of the most Closing the Gap: The Burden of Unnecessary Illness. New fare receipt for the NHANES sample. It is sweeping social policy changes this country York, NY: Oxford University Press; 1987:125–129. also possible that the declining welfare has seen in decades represents an important 2. Kaplan GA, Lynch JW. Whither studies on the so- caseload before 1997 resulted in a pool missed opportunity. There is growing realiza- cioeconomic foundations of population health [editorial]? Am J Public Health. 19 97;87:1409–1411. of recipients who, in comparison with the tion that social and economic policy may 3. US Dept of Health and Human Services. The Ad- 1988–1994 NHANES respondents who re- have important influences on both the health ministration for Children and Families (ACF). 2002. ported being recipients of welfare, were es- of populations and the health disparities Available at: www.acf.dhhs.gov/news/stats/case-fam. sentially self-selected for poor health. Thus, within populations.27 Therefore, we need to htm. Accessed February 20, 2003. the comparison between WES-HS and study how important policy changes such as 4. National Council of State Legislatures. Welfare Re- NHANES samples is not without problems. welfare reform affect a population’s health form project. 2002. Available at: http://www.ncsl.org/ statefed/welfare/caseloadwatch.htm. Accessed Febru- We do find, however, that despite some eco- and make empirical data on health part of ary 20, 2003. nomic improvement among the total WES evaluationg the impact of such changes. 5. Danziger S. Introduction: what are the early les- sample since the introduction of TANF, a sons? In: Danziger S, ed. Economic Conditions and Wel- 36% increase occurred between 1997 and fare Reform. Kalamazoo, Mich: W.E. Upjohn Institute for Employment Research; 1999:1–10. 1999 in the percentage of WES respondents About the Authors 6. Blank RM. What Causes Public Assistance Case- who self-reported poor or fair health (from George A. Kaplan, Nalini Ranjit, Diem Tran, and John W. Lynch are with the Center for Social Epidemiology loads to Grow? Cambridge, Mass: National Bureau of 23.9% to 32.5%). and Population Health, Department of Epidemiology, Economic Research Inc; 1997. NBER Working Paper The results of this study suggest that the University of Michigan School of Public Health, Ann 6343. health of poor women who have previously Arbor. Kristine Siefert, Sandra Danziger, and Richard 7. Bennett NG, Lu H, Song Y. Welfare Reform and Tolman are with the School of Social Work, Trivellore E. Changes in the Economic Well-Being of Children. Cam- received cash assistance under welfare re- Raghunathan is with the Department of Biostatistics and bridge, Mass: National Bureau of Economic Research form is worse than that of a national sample Survey Research Center, and Elizabeth A. Young is with Inc; 2002. NBER Working Paper 9399. the Department of Psychiatry and Mental Health Re- of women of the same age and race. Al- 8. Danziger S. Approaching the limit: early national search Institute, University of Michigan, Ann Arbor. lessons from welfare reform. In: Weber BA, Duncan DJ, though some welfare recipients may have Susan Hudson is with the Genesee County Health Depart- Whitener LA, eds. Rural Dimensions of Welfare Reform. ment, Flint, Michigan. made social and economic gains after welfare Kalamazoo, Mich: Upjohn Institute for Employment Re- Request for reprints should be sent to George A. Kap- reform, many still lead lives characterized by search; 2002:25–49. lan, PhD, Center for Social Epidemiology and Population economic deprivation and demanding work, Health, 1214 S University Ave, Ann Arbor, MI 48104 9. Presser HB, Cox AG. The work schedule of low- family, and neighborhood conditions, with (e-mail: [email protected]). educated American women and welfare reform. few resources to alleviate any of these prob- This article was accepted October 24, 2004. Monthly Labor Rev. 19 97;120:25–34. lems. It is possible that welfare recipients who 10. Zedlewski SR. Work activities and obstacles to Contributors work among TANF recipients. In: The New Federalism: have economically and socially benefited National Survey of America’s Families. Washington, DC: G.A. Kaplan, N. Ranjit, and S. Danziger wrote the first Urban Institute; 1999. Available at: http://www.urban. from welfare reform’s large social policy draft, which was reviewed by all other authors. G.A. org/url.cfm?ID=309091. Accessed February 20, changes have also experienced improvements Kaplan, E.A. Young, D. Tran, and S. Hudson con- 2003. in their health. However, it is equally possible tributed to the design and conduct of the field protocol. N. Ranjit carried out the analyses. K. Siefert, S. Danziger, 11. Ong PM, Blumenberg E. Job Access, Commute, and that, in a context characterized by existing and R. Tolman helped in linking this study to the Women’s Travel Burden Among Welfare Recipients. Los Angeles, poor health and vulnerability, some recipients Employment Study (WES) and in interpreting WES Calif: Ralph and Goldy Lewis Center for Regional Pol- face worsening health and deepening health data. All authors contributed to the overall design of icy Studies, University of California–Los Angeles; the study, formulation of hypotheses, and interpre- 19 97. Working Paper 20. burdens that already limit participation in tation of results. 12. Danziger SK, Seefeldt KS. Barriers to employment work. A recent report indicated that welfare and the “hard to serve”: implications for services, sanc- recipients’ health insurance coverage has Acknowledgments tions and time limits. Focus. 2002;22(1):76–81. also decreased over time, adding even more Data collection and analysis were supported by an 13. O’Campo P, Rojas-Smith L. Welfare reform and reason for concern.26 award from the National Institute for Child Health and women’s health: a review of the literature and implica- Development (HD P50 HD38986) and by funds from tions for state policy. J Public Health Policy. 1998;19(4): Specific WES data (not shown here) indi- the Michigan Initiative on Inequalities in Health to 420–446. cate that the percentage of women not cov- G.A. Kaplan. Major funding of the Women’s Employ- 14 . Ensminger ME. Welfare and psychological dis- ment Study study was provided by the Charles Stewart ered by government or private health insur- tress: a longitudinal study of African American urban Mott Foundation and the Joyce Foundation. E.A. Young mothers. J Health Soc Behav. 19 95;36:346–359. ance at time of interview increased threefold, is funded by an award from the National Institute of from 6.8% in 1997 to 21% in 2001 (data Mental Health (MH01931). 15. Loprest P, Acs G. Profile of Disability Among Fami- not shown). Although the results from the We thank the participants in the Women’s Employ- lies on AFDC. Washington, DC: Kaiser Family Founda- ment Study. tion; 1996. current study suggest that the health of 16. Shern DL, Wilson NZ, Coen AS. Client outcomes, women of low socioeconomic status under Human Participant Protection II: longitudinal client data from the Colorado treatment outcome study. Milbank Q. 1994;72:123–148. welfare reform is poor and may have wors- This study was approved by the University of Michi- ened, these results cannot be seen as conclu- gan’s institutional review board. 17.Polit DF, London AS, Martinez JM. The Health of

July 2005, Vol 95, No. 7 | American Journal of Public Health Kaplan et al. | Peer Reviewed | Research and Practice | 1257  RESEARCH AND PRACTICE 

Poor Urban Women: Findings From the Project on Devo- lution and Urban Change. New York, NY: Manpower Demonstration Research Corporation; 2001. Community-Oriented 18. Danziger S, Corcoran M, Danziger S, et al. Barri- ers to the employment of welfare recipients. In: Cherry R, Primary Care: Rodgers W, eds. Prosperity for All? The Economic Boom Health Care for the 21st Century and African Americans. New York, NY: Russell Sage Foundation; 2000:239–272. Edited by Robert Rhyne, MD, Richard Bogue, PhD, 19. O’Brien E, Waeber B, Parati G, Staessen J, Gary Kukulka, PhD, and Hugh Fulmer, MD Myers MG. Blood pressure measuring devices: recom- This book will give insight into: mendations of the European Society of Hypertension. BMJ. 2001;322:531–536. • How medicine, health systems, community leaders, and social services can be supportive as America’s public 20. Clinical Guidelines on the Identification, Evaluation, health practice continues to be restructured and redefined and Treatment of Overweight and Obesity in Adults. The • New models of community-oriented primary care Evidence Report. Washington, DC: National Heart, • Methods and interventions on population-derived health Lung, and Blood Institute, US Public Health Service; 2nd Edition 1998. NIH publication 98–4083. needs ISBN 0-87553-236-5 • Health promotion and disease prevention as part of the 21.Ware JJ, Sherbourne CD. The MOS 36-item short- 1998 ❚ 228 pages overall reorganization of health services form health survey (SF-36): conceptual framework and Softcover item selection. Med Care. 19 92;30:473–483. • Understanding how community-oriented primary care $27.00 APHA Members can complement managed care and community benefit 22. Hays RD, Sherbourne CD, Mazel RM. The RAND $39.00 Nonmembers programs 36-Item Health Survey 1.0. Health Econ. 19 93;2: plus shipping and handling 217–227. This book teaches skills and techniques for implementing a community-oriented primary care process and topics not 23. Knudson RJ, Lebowitz MD, Holberg CJ, et al. normally taught in health professional education. Changes in the normal-maximal expiratory flow- volume curve with growth and aging. Am Rev Respir ORDER TODAY! Disord. 19 83;127:725–734. American Public Health Association 24. National Center for Health Statistics. 1999–2000 Publication Sales Addendum to the NHANES III Analytic Guidelines. Web: www.apha.org August 30, 2002. Available at: http://www.cdc.gov/ E-mail: [email protected] nchs/data/nhanes/guidelines1.pdf. Accessed February Tel: 888-320-APHA 20, 2003 FAX: 888-361-APHA COPC03J5 25. Smedley BD, Stith AY, Nelson AR, eds. Unequal Treatment: Confronting Racial and Ethic Disparities in Health Care. Washington, DC: National Academies Press; 2003. 26. Mann C, Hudman J, Salganicoff A, et al. Five years later: poor women’s health care coverage after welfare reform. J Am Med Womens Assoc. 2002;57(1): 16–22. 27. Kaplan GA. Economic policy is healthy policy: findings from the study of income, socioeconomic status, and health. In: Auerbach JA, Krimgold BK, eds. In- come, Socioeconomic Status and Health: Exploring the Re- lationships. Washington, DC: National Policy Association; 2001:137–149.

1258 | Research and Practice | Peer Reviewed | Kaplan et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

To House or Not to House: The Effects of Providing Housing to Homeless Substance Abusers in Treatment

| Jesse B. Milby, PhD, Joseph E. Schumacher, PhD, Dennis Wallace, PhD, Michelle J. Freedman, PhD, and Rudy E. Vuchinich, PhD

Homelessness affects 3.5% to 6.0% of Amer- Objectives. Housing typically is not provided to homeless persons during drug icans at some point in their lives, and as abuse treatment. We examined how treatment outcomes were affected under 3 many as 11.4% of urban women of childbear- different housing provision conditions. 1–4 ing age are homeless. The point prevalence Methods. We studied 196 cocaine-dependent participants who received day of homelessness has been estimated at 1% of treatment and no housing (NH), housing contingent on drug abstinence (ACH), the general US population and 6.3% of peo- or housing not contingent on abstinence (NACH). Drug use was monitored with ple in poverty.5,6 Lack of child care and a his- urine testing. tory of physical and/or sexual abuse greatly Results. The ACH group had a higher prevalence of drug abstinence than the increase risk for homelessness.7 Homeless- NACH group (after control for treatment attendance), which in turn had a higher ness, in turn, increases risk for diseases in- prevalence than the NH group. All 3 groups showed significant improvement in maintaining employment and housing. cluding HIV, other sexually transmitted dis- Conclusions. The results of this and previous trials indicate that providing ab- eases, tuberculosis, asthma, and bronchitis.8 stinence-contingent housing to homeless substance abusers in treatment is an These diseases contribute to morbidity and efficacious, effective, and practical intervention. Programs to provide such hous- 9–11 premature death. ing should be considered in policy initiatives. (Am J Public Health. 2005;95: Providing effective drug abuse intervention 1259–1265. doi:10.2105/AJPH.2004.039743) for homeless substance abusers is a daunting challenge because of this population’s high morbidity, poverty, and social instability.12–19 in the field believe non–abstinence-contingent Act23 criteria for homelessness (they lacked a In particular, homelessness necessitates a con- housing to be an easier, cheaper, and more fixed nighttime residence, including shelters tinuous search for food and shelter, which se- effective way to provide treatment for home- or other temporary accommodations, or were riously interferes with treatment participation. less drug abusers, but until now no studies at imminent risk of becoming homeless), met For example, O’Brien et al.20 found that day have examined whether this approach is ef- criteria for cocaine dependence as outlined in treatment reduced drug abuse among non- fective and cost-efficient. the Diagnostic and Statistical Manual of Mental homeless participants but that no homeless In the trial, we examined how providing Disorders, Revised Third Edition,24 had used participants were retained in treatment. housing with and without an abstinence con- cocaine within the past 2 weeks, and scored Previous trials have shown that provision tingency affected drug use, homelessness, above 70 (2 standard deviations [SDs] above of housing for homeless substance abusers and employment among cocaine-dependent the mean) on 1 or more Symptom Check- while they are undergoing behavioral day homeless persons. We considered 2 hypothe- list–90—Revised scales at intake.25,26 Poten- treatment improved addiction outcomes and ses: (1) that the housed groups would have tial participants were excluded if cognitive im- resulted in less homelessness.21,22 In those better treatment retention and outcomes pairment precluded informed consent or if studies, housing was available only when par- than those without housing, and (2) that they required immediate inpatient medical ticipants were drug-free; that is, housing was those who received housing contingent on treatment. “abstinence-contingent.” abstinence would have better outcomes than The final study set consisted of 196 partici- Such a contingency differs greatly from the those who receive housing not contingent on pants randomly divided into 3 groups: those typical “housing first” approach to fighting abstinence. receiving abstinence-contingent housing (ACH; homelessness, in which housing is provided n=63), those receiving non–abstinence- with no abstinence requirements, on the METHODS contingent housing (NACH; n=67), and premise that housing provision alone will lead those receiving no housing (NH; n=66). All to improvement. In theory, non–abstinence- Participants groups received the same treatment during contingent housing for homeless substance Participants were homeless persons from 12 months. Participant demographics are abusers in treatment could be less expensive the Birmingham, Alabama, area with coexist- summarized in Table 1. and logistically simpler, because treatment ing cocaine dependence and nonpsychotic staff would not be required to monitor drug mental disorders who were clients at Birming- Intervention use and implement ejection from housing be- ham Health Care between September 1994 The intervention was divided into phase 1 cause of a failed urine test. Some authorities and November 2001. All met McKinney (months 1 through 2; day treatment), phase 2

July 2005, Vol 95, No. 7 | American Journal of Public Health Milby et al. | Peer Reviewed | Research and Practice | 1259  RESEARCH AND PRACTICE 

TABLE 1—Demographic Characteristics and Treatment Group Comparisons months, with individual counseling provided as needed. Treatment Group NH (n=66) NACH (n=66) ACH (n=63) P a Outcome Measures

Gender, n (%) .98 Participants were assessed by interviewers Male 50 (76) 50 (76) 47 (75) . . . not aware of the participants’ treatment group Female 16 (24) 16 (24) 16 (25) . . . at baseline and 2, 6, and 12 months after Age, yb (SD) 38.2 (7.4) 40.9 (7.2) 38.4 (6.7) .055 baseline. Participants were given backpacks Race/ethnicity, n (%) .25 containing personal hygiene items after base- African American 59 (89) 63 (95) 55 (87) . . . line assessment and were paid $25 for each European American 7 (11) 3 (5) 8 (13) . . . completed follow-up assessment. A nonpar- Veteran, n (%)c 11 (17) 12 (19) 11 (19) .97 ticipant “tracker” was tasked with locating Education, y (SD)d 11.7 (1.6) 12.2 (1.8) 12.4 (1.8) .22 participants and transporting them to the Longest full-time job, mo (SD)e 51.6 (46.3) 53.5 (40.9) 65.9 (69.6) .76 follow-up assessments. The tracker was paid $25 for each completed follow-up assessment. Note.NH=no housing; NACH=non-abstinence-contingent housing; ACH=abstinence-contingent housing. Drug testing was conducted with the aP values based on χ2 test (df=2) for categorical measures and on Kruskal–Wallis test for continuous measures. P tests were 2-tailed. bOne NACH participant had missing data for age. OnTrak TesTstik (Roche Diagnostic Systems, c Ten participants (3 NH, 3 NACH, 4 ACH) had missing data for veteran status. Somerville, NJ). To encourage compliance, d Four participants (2 NH, 2 NACH) had missing data for years of education. clients were paid $3 per specimen during eEighteen participants (5 NH, 9 NACH, 4 ACH) had missing data for longest employment period. months 5 through 10 and $5 per specimen during months 11 and 12 for 4 randomly selected specimens out of the 8 scheduled uri- (months 3 through 6; work therapy and after- reestablished with 2 consecutive drug-negative nalysis tests. If a urinalysis result was incon- care group meetings), and phase 3 (months 7 tests, after which ACH participants were im- clusive, the original specimen was tested 1 or through 12; aftercare). Our manual-based mediately moved back to program housing. 2 more times, and the 2 most consistent re- cognitive–behavioral (CB) day treatment22 NACH participants remained in housing as sults determined the data point. Urinalysis was provided to all participants. All partici- long as they gave scheduled urine specimens, test results were used to implement the hous- pants were treated during the same types of regardless of results. Exceptions occurred being contingencies and to construct 2 drug group meetings, in the same offices, and by cause of severe misbehavior at housing sites use–dependent variables: (1) the proportion the same counselors. Day treatment was Mon- (e.g., damaging apartments, on-premise drug of participants, by group, whose urine tests all day through Friday, from 7:45 AM to 2:00 PM, use, having overnight guests). Such behavior, were drug-negative each week, and (2) the and included lunch, transportation to and which was observed in only 22 participants, longest string of consecutive weeks of absti- from housing, individual counseling, and was grounds for immediate housing restric- nence attained by each participant. Data from psychoeducational and therapy groups. tion for both ACH and NACH participants specimens obtained at follow-up are not re- During the study period, urine testing was for 1 week, during which the participant was ported here. conducted for cocaine, marijuana, and alcohol placed in a shelter. Homelessness and employment were mea- use. Specimens were obtained from randomly During phase 2, ACH and NACH partici- sured with the Retrospective Interview for selected participants twice weekly during pants were charged a monthly rent of Housing, Employment, and Treatment His- phases 1 and 2 and once weekly during $161.00; participants earned rent money tory completed by each participant.27 Days phase 3. During phase 1, ACH participants through work therapy or other employment. employed was the number of days the partici- received rent-free housing after 2 consecutive Neither ACH nor NACH participants were re- pant was fully employed (i.e., worked at least drug-negative urine tests. NACH participants moved from housing for not paying rent. All 7 hours per day) the last 60 days. Homeless- received rent-free housing in similar buildings participants were offered work therapy at a ness was measured by number of days in a different neighborhood after 2 consecu- minimum wage of $5.25 per hour. Continua- housed during the past 60 days. tive urine tests, regardless of test results. In tion of work therapy was not contingent on 19 97 (midstudy) the ACH and NACH groups abstinence, but the housing contingency re- Treatment Attendance and Study switched housing units with one another to mained in place for ACH participants. Retention balance any housing or neighborhood influ- During phase 3, ACH and NACH partici- Treatment attendance was recorded in a ences. NH participants received no program- pants could remain in program housing if counselor-verified daily service log in 15- provided housing. housing slots were available, but no absti- minute increments. A complete treatment day For ACH participants, a drug-positive urine nence contingency was imposed for any of was defined as participation in 10 or more in- test resulted in immediate removal from hous- the participants. All groups were offered af- crements per day during phase 1 and in 4 or ing to a shelter. Abstinence was considered tercare group meetings once weekly for 6 more increments per day during phase 2.

1260 | Research and Practice | Peer Reviewed | Milby et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

Study retention was measured by number of follow-up assessments completed.

Analysis We employed an intention-to-treat analysis for the drug abstinence, housing, and employment outcomes, meaning that, in our analysis, we included all people for whom baseline data was available, regardless of each person’s participation in treatment. Some of the scheduled urine tests were missed by participants. To determine whether different coding schemes for the missed tests would produce different substantive conclusions, we conducted 3 types of analyses: (1) assuming missing data were drug-positive, (2) treating missing data as missing, and (3) carrying forward the previous observation. Although absti- Note.ACH=abstinence-contingent housing; NACH=non-abstinence-contingent housing; NH=no housing. nence prevalence—the proportion of participants who were abstinent—differed across FIGURE 1—Weekly prevalence of abstinence among participants during treatment and these approaches, the substantive conclusions aftercare. were consistent. Thus, we report results obtained by the standard approach of treating missing data as drug-positive, with the excep- days employed from baseline to 6 months (4.98), 6.60 (6.32), and 8.32 (6.79), respec- tion that data missing for administrative rea- and from baseline to 12 months were ana- tively. Analysis of variance showed an overall sons (test scheduled but not conducted owing lyzed by treatment group and overall with group effect (P=.0020), with strong evidence to computer problems, daily test schedule repaired t tests. of a difference between the NH and ACH ceived too late, tests scheduled on clinic holi- groups (P=.0004), but no evidence of a dif- days) were treated as randomly missing. Too RESULTS ference between the NH and NACH groups few urine specimens were collected for analy- (P=.058) or between the NACH and ACH sis beyond month 6. Abstinence Outcomes groups (P=.091). For our estimate of weekly abstinence Figure 1 shows the abstinence prevalence On the basis of previous experience,29,30 prevalence, we used generalized estimating for weeks 1 through 24; it also shows a clear we defined participants who completed 20 equation (GEE) extensions of generalized lin- separation between the ACH/NACH groups or more treatment days during a single ear models to assess differences between and the NH group, with the former showing phase as high attenders for that phase. Par- housing groups and to determine whether consistently higher abstinence prevalence ticipants who had fewer than 20 treatment group differences in abstinence were modi- than the latter. Among the housed groups, days during a phase were defined as low at- fied by treatment attendance.28 We used an ACH participants had higher abstinence prev- tenders for that phase. During phase 1, initial model to estimate abstinence as a func- alence than NACH participants during 22 of mean (SD) days attended for the NH, tion of housing group and study phase. We the 24 weeks, but this difference was small NACH, and ACH groups were 18.3 (12.2), used Wald tests based on Liang and Zeger’s and not statistically significant. Table 2 shows 26.7 (9.3), and 24.9 (11.5), respectively, procedure28 to test for group differences the statistical results of the GEE model: dur- with strong evidence of differences between across each phase. We used a second model ing both phase 1 and phase 2, the ACH and the NH group and the NACH group (P < to estimate abstinence as a function of hous- NACH groups had significantly higher absti- .0001) and the NH group and the ACH ing group, study phase, and treatment atten- nence levels than the NH group, whereas the group (P =.0010) but no evidence of differ- dance. To estimate consecutive weeks of ab- ACH and NACH groups did not differ from ence between the NACH group and the stinence, we used an analysis-of-variance one another. ACH group (P = .35). After summation model to test for differences between housing Because a string of consecutive weeks of across both phases, mean (SD) days attended groups, with and without control for treat- abstinence could include the treatment phase for the NH, NACH, and ACH groups were ment attendance. transition, phase was not included as a factor 32.0 (28.4), 56.6 (29.9), and 50.1 (29.1), We used Wilcoxon rank sum tests to as- in the analyses. Mean (SD) consecutive weeks respectively, with strong evidence of a differ- sess group differences in days housed and of abstinence (out of a possible 24 weeks) for ence between the NH group and the NACH days employed. Changes in days housed and the NH, NACH, and ACH groups were 4.48 group (P < .0001) and the NH group and the

July 2005, Vol 95, No. 7 | American Journal of Public Health Milby et al. | Peer Reviewed | Research and Practice | 1261  RESEARCH AND PRACTICE 

TABLE 2—Proportion (SE) of Participants Who Were Abstinent During Phases 1 and 2 of As previous studies of homelessness have 21,22 Treatment: Results of Generalized Estimating Equation Model observed, our participants missed scheduled urine tests for administrative and other Prevalence Within Attendance Group (SE) reasons. Data missing for administrative rea- a b c Abstinence Prevalence (SE) Low Attenders High Attenders sons were comparable across groups, ranging Phase 1 Phase 2 Phase 1 Phase 2 Phase 1 Phase 2 from 17% to 21% in phase 1 and from 8% to 11% in phase 2. Data missing for other rea- NH 0.41 (0.044) 0.18 (0.036) 0.25 (0.038) 0.08 (0.016) 0.56 (0.053) 0.25 (0.038) sons differed across groups and increased NACH 0.60 (0.038) 0.40 (0.043) 0.42 (0.066) 0.21 (0.038) 0.64 (0.035) 0.51 (0.048) throughout the study, with phase 1 missing ACH 0.69 (0.045) 0.46 (0.046) 0.45 (0.063) 0.29 (0.048) 0.82 (0.029) 0.60 (0.047) rates of 43%, 21%, and 14% and phase 2 Note.NH=no housing; NACH=non-abstinence-contingent housing; ACH=abstinence-contingent housing. missing rates of 67%, 43%, and 35% for the a Phase 1 Wald tests showed an overall treatment effect (P<.0001), with pairwise differences between the NH and ACH NH, ACH, and NACH groups, respectively. Al- groups (P<.0001) and the NH and NACH groups (P=.0015) but no evidence of a difference between the NACH group and the ACH group (P=.17).The results for phase 2 also showed an overall treatment effect (P=.0001), with pairwise differences though these rates are less than ideal for study between the NH and ACH groups (P<.0001) and the NH and NACH groups (P=.0003) but no evidence of a difference purposes, sensitivity analyses indicated that between the NACH group and the ACH group (P=.29). our results are robust across all approaches to bAmong low attenders, phase 1 Wald tests showed an overall treatment effect (P=.010), with pairwise differences between the NH and ACH groups (P=.0080) and the NH and NACH groups (P=.032) but no evidence of a difference between the handling the missing data (data not shown). NACH group and the ACH group (P=.69).The results for phase 2 also showed an overall treatment effect (P=.0009), with pairwise differences between the NH and ACH groups (P=.0007) and the NH and NACH groups (P=.010) but no evidence of Homelessness Outcomes a difference between the NACH group and the ACH group (P=.21). cAmong high attenders, phase 1 Wald tests showed an overall treatment effect (P=.0005), with pairwise differences between Results for days housed are shown in the NH and ACH groups (P=.0004) and the ACH and NACH groups (P=.0011) but no evidence of a difference between the Figure 2 and provide strong evidence of NACH group and the NH group (P=.26).The results for phase 2 also showed an overall treatment effect (P=.010), with within-group housing changes from base- pairwise differences between the NH and ACH groups (P=.0024) and the NH and NACH groups (P=.0011) but no evidence of a difference between the NACH group and the ACH group (P=.19). line to 12 months for all groups (P < .0001) and for each group (ACH, P <.0014; NACH, P<.0006; NH, P<.021). No significant within-group differences from baseline to 6 ACH group (P = .0005), but no evidence of series of linear models with an ordinal mea- months were found for any group, and the a difference between the NACH group and sure of attendance that included treatment groups did not differ significantly from each the ACH group (P =.21). days in both phases. Participants were classi- other at any time point. We used a second GEE model to examine fied as low attenders if their attendance was the abstinence prevalence as a function of low during both phase 1 and phase 2, as in- Employment Outcomes housing group, phase, and attendance. As termediate attenders if their attendance was Results for days employed are shown in shown in Table 2, attendance modified the low during one phase and high during the Figure 3 and provide strong evidence of housing group effect. In both phase 1 and other, and as high attenders if their atten- within-group employment changes from base- phase 2, low attenders in the ACH and dance was high during both phases. A model line to 6 months and from baseline to 12 NACH groups had greater abstinence than including housing group, attendance, and an months for all groups combined (both P< low attenders in the NH group, with no evi- interaction term showed no evidence of ef- .0001). No evidence of significant differences dence of a difference between NACH and fect modification (P =.34 for the interaction). between the groups was found at any time ACH low attenders. Among high attenders, A model without the interaction term point. Separate analyses within each group the phase 1 data showed a clear trend for showed that attendance acted as an interven- from baseline to 6 months showed significant higher abstinence prevalence in the ACH ing variable for the housing group effect. differences for 2 of the 3 groups (ACH, P = group, followed by the NACH group, with the This analysis provided strong evidence of an .28; NACH, P <.0001; NH, P =.052). Sepa- lowest abstinence again in the NH group. The overall housing group effect after adjustment rate group analyses from baseline to 12 phase 2 results for high attenders were com- for attendance (P =.0080) and an effect of months showed significant employment parable to the results for low attenders, but attendance on abstinence (P < .0001). The changes for all groups (ACH, P =.0014; the point estimates suggested greater differ- mean adjusted consecutive weeks of absti- NACH, P =.0006; NH, P =.021). ence between ACH and NACH high atten- nence for the NH, NACH, and ACH groups ders than was observed between ACH and were 5.28, 4.68, and 7.32, respectively, with Retention During Follow-Up Assessments NACH low attenders. evidence of a difference between the ACH The percentage of follow-up assessments Results for consecutive weeks of absti- group and the NH group (P =.024) and be- completed did not differ between the 2 nence differed somewhat from results for ab- tween the ACH group and the NACH groups housed groups, but more participants were re- stinence prevalence. To assess the effect of (P = .0031), but no evidence of a difference tained in the housed groups than in the NH attendance on the relationship between hous- between the NACH group and the NH group group: 66.7% for ACH, 62.7% for NACH, ing group and sustained abstinence, we fit a (P =.51). and 42.4% for NH (χ2 =8.19; df=2; P=.017).

1262 | Research and Practice | Peer Reviewed | Milby et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

60 portant role of housing in treatment of home- ACH less substance abusers. The positive effects of NACH NH housing were maintained after control for the 50 contribution of attendance to abstinence. The results related to our second hypothesis, that abstinence-contingent housing would 40 improve outcomes more than non–abstinence- contingent housing, were more ambiguous. 30 The ACH group had a consistently higher abstinence prevalence than the NACH group during the 24-week study period; this differ- 20 ence was not statistically significant but may Days Employed in Past 60 Days in Past Employed Days be clinically important. However, by the more rigorous measure of consecutive weeks of ab- 10 stinence, the ACH group was significantly more successful in maintaining abstinence 0 than the NH group, whereas the NACH 02612group was not more successful than the NH Months group. Moreover, for both abstinence vari- Note.ACH=abstinence-contingent housing; NACH=non-abstinence-contingent housing; NH=no housing. ables, the ACH group’s abstinence levels were FIGURE 2—Mean days housed at baseline and at 3 follow-up assessments. clearly superior to those of the NACH group after control for attendance, especially among high attenders early in treatment. Employ- 60 ment and housing outcomes were similar to ACH NACH those seen in our previous study, which also NH showed gains from baseline to 12 months 50 with no differences between groups.32 The consistent trend of higher weekly ab-

40 stinence prevalence in the ACH group compared with the NACH group and the significantly higher weekly abstinence prevalence 30 among high attenders in the ACH group compared with the NH group (but not in the NACH group vs the NH group) all are consis- 20 tent with our second hypothesis and support Days Employed in Past 60 Days in Past Employed Days the clinical and administrative value of pro- 10 viding abstinence-contingent housing during drug treatment. In a previous study we conducted,31 ACH and NACH participants 0 shared an apartment complex, a situation that 02612 resulted in clinical and management difficul- Months ties. In that study, several homeless clients = = = Note.ACH abstinence-contingent housing; NACH non-abstinence-contingent housing; NH no housing. assigned to ACH refused the free housing FIGURE 3—Mean days employed at baseline and at 3 follow-up assessments. (a response previously unobserved in more than a decade of work with homeless substance abusers), stating that they were not yet Gender-Specific Effects DISCUSSION “strong enough” to resist using drugs while In exploratory analyses, we examined the fellow NACH clients used them regularly. Ad- relationship of gender to the outcomes of Greater abstinence and attendance in the ditionally, Birmingham Health Care incurred housing, employment, and abstinence. We housed groups compared with the nonhoused property losses owing to actions of NACH found no evidence that gender acted as an ef- group, for both abstinence variables (preva- participants, and some NACH clients experi- fect modifier or a confounder in the relation- lence and continuous weeks of abstinence), enced incidents of being robbed and other ship of treatment with any of the 3 outcomes. supports our first hypothesis regarding the im- crimes.

July 2005, Vol 95, No. 7 | American Journal of Public Health Milby et al. | Peer Reviewed | Research and Practice | 1263  RESEARCH AND PRACTICE 

In our previous studies,21,22 abstinence observed greater abstinence at 15 months’ all data analyses and assisted in data interpretation and among the housed group was clearly superior, follow-up in the abstinence-contingent writing. M.J. Freedman supervised day-to-day implementation of the 3 experimental groups, data collection, but the provision of housing and the absti- voucher group than in the non–contingent and data review and participated in manuscript prepa- nence contingency were confounded for the abstinence voucher group. Maude-Griffin and ration. R.E. Vuchinich coordinated urine database housed group. Those studies and the current colleagues35 also observed treatment differ- preparation and, with D. Wallace, participated in data review, data interpretation, and manuscript preparation. study suggest that provision of abstinence- ences for cocaine-dependent outpatients at contingent housing may both support higher long-term follow-up after day treatment. Acknowledgments abstinence and be easier to manage. Thus, These treatment successes using CM suggest Data collection and manuscript preparation were sup- available evidence supports serious considera- that long-term abstinence may be obtainable ported by the National Institutes of Health (grants tion by public agencies serving homeless sub- in cocaine dependence treatment with behav- DA08475 and DA11789, respectively). The following individuals contributed substantial stance abusers of a policy to require absti- iorally based procedures. However, no study assistance: Jonathan Dunning, MA, Sonja Frison, PhD, nence-contingent housing during drug with a cocaine-dependent homeless popula- Stefan Kertesz, MD, Cecelia McNamara, PhD, Max treatment. tion has yet shown persistent abstinence gains Michael, MD, Mary A. Plant, PhD, Stewart Usdan, PhD, and Catherine Ward, PhD. Unlike previous studies,21,22 participants in beyond 6 months. the ACH and NACH groups in the current The most important limitation of this study Human Participant Protection study could work and earn a stipend when concerns the missing data. We were unable to All procedures and research reported here were re- they were not abstinent, and ACH partici- obtain sufficient assessments of abstinence viewed, approved, and monitored by the University of Alabama at Birmingham institutional review board for pants could use their earnings to purchase beyond 6 months, and many scheduled urine Human Research. other housing while barred from program- tests during months 1 through 6 were missed provided housing because of nonabstinence. by participants. High rates of missing data are References This ability to access and use funds for sec- a significant concern. However, 3 conven- 1. Link BG, Susser E, Stueve A, Phelan J, Moore RE, ondary when a person was between absti- tional analytic approaches for dealing with Struening E. Lifetime and five-year prevalence of homelessness in the United States. Am J Public Health. nence and access to treatment program activi- missing data yielded consistent results in 1994;84:1907–1912. ties may have weakened abstinence support group comparisons. Furthermore, we reported 2. Reardon ML, Burns AB, Preist R, Sachs-Ericsson N, by weakening the effect of the abstinence the results with the most conservative of the Lang AR. Alcohol use and other psychiatric disorders contingency on housing allowance, especially procedures (coding missed tests as drug- in the formerly homeless and never homeless: prevalence, age of onset, comorbidity, temporal sequencing, during months 3 through 6, relative to the positive), an approach that underestimates and service utilization. Subst Use Misuse. 2003;38: earlier studies. actual abstinence for all groups. Finally, miss- 601–644. This trial’s demonstration of the positive ef- ing data and less-than-ideal analytic proce- 3. Link B, Phelan J, Bresnahan M, Stueve A, Moore R, fect of housing, with or without an abstinence dures for missing data were not unique to this Susser E. Lifetime and five-year prevalence of homelessness in the United States: new evidence on an old requirement, could be interpreted as favoring study but are chronic problems in the sub- debate. Am J Orthopsychiatry. 1995;65:347–354. arguments for minimal or no barriers to hous- stance abuse treatment field. Better methods 4. Webb DA, Culhane J, Metraux S, Robbins JM, ing entry during the rehabilitation of home- for addressing these concerns would have im- Culhane D. Prevalence of episodic homelessness less persons. However, observed benefits ac- proved this study, just as they would improve among adult childbearing women in Philadelphia, PA. Am J Public Health. 2003;93:1895–1896. crued among persons who consented to all such studies. 5. Burt MR, Aron LY, Douglas T, Valente J, Lee E, participate in an addiction treatment trial, so Provision of housing during day treatment Iwen B. Homelessness: Programs and the People They it is unknown whether housing would confer of homeless substance abusers was associated Serve. Findings of the National Survey of Homeless Assis- similar benefits on addicted persons uninter- with greater drug abstinence. This study has tance Providers and Clients. Washington, DC: Urban In- stitute; 1999. Available at: http://www.huduser.org/ ested in treatment. clear implications for treatment and housing publications/homeless/homeless_tech.html. Accessed The current study is part of a series in- of homeless substance abusers. March 23, 2005. tended to examine the impact of applying con- 6. Burt MR, Aron LY. Helping America’s Homeless. Washington, DC: Urban Institute; 2001. tingency management (CM) interventions to 7. Herman D, Susser ES, Struening EL, Link BL. Ad- substance use disorders. In a comparison of About the Authors Jesse B. Milby, Joseph E. Schumacher, Michelle J. Freedman, verse childhood experiences: are they risk factors for CM and CB treatment in polydrug abusers, and Rudy E. Vuchinich are with the University of Alabama adult homelessness? Am J Public Health. 1997;87: Rawson and colleagues33 found that whereas at Birmingham. Dennis Wallace is with Rho Federal Sys- 249–255. CM interventions produced better short-term tems Division Inc, Chapel Hill, NC. 8. Smereck G, Hockman EM. Prevalence of HIV in- Requests for reprints should be sent to Jesse B. Milby, PhD, fection and HIV risk behaviors associated with living outcomes, CB treatment produced better long- Department of Psychology–CH415, University of Al- place: on-the-street homeless drug users as a special term outcomes. However, when Higgins and abama at Birmingham, 1530 3rd Ave S, Birmingham, AL target population for public health intervention. Am colleagues34 conducted a trial that compared 35294-1170 (e-mail: [email protected]). J Drug Alcohol Abuse. 1998;24:299–319. This article was accepted August 2, 2004. voucher reinforcements that were contingent 9. Martens W. A review of physical and mental health in homeless persons. Public Health Rev. 2001; or not contingent on abstinence, in a popula- Contributors 29:13–33. tion with less severe addiction and rehabilita- J.B. Milby and J.E. Schumacher originated the study 10.Plumb JD. Homelessness: care, prevention, and tion problems than the current study, they and supervised implementation. D. Wallace completed public policy. Ann Intern Med. 1997;126:973–975.

1264 | Research and Practice | Peer Reviewed | Milby et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

11.Hwang SW. Is homelessness hazardous to your ment outcome as a function of treatment attendance health? Obstacles to the demonstration of a causal rela- with homeless persons abusing cocaine. J Addict Dis. tionship. Can J Public Health. 2002;93:407–410. 19 95;14:73–85. 12. Corrigan EM, Anderson SC. Homeless alcoholic 30. Schumacher JE, Usdan S, Milby JB, Wallace D, women on skid row. Am J Drug Alcohol Abuse. 1984; McNamara C. Abstinent-contingent housing and treat- 10:535–549. ment retention among crack-cocaine-dependent home- 13. Galanter M, Egelko S, De Leon G, Rohrs C, less persons. J Subst Abuse Treat. 2000;19:81–88. Franco H. Crack/cocaine abusers in the general hospi- 31. Milby JB, Schumacher JE, McNamara CL, Usdan SL, tal: assessment and initiation of care. Am J Psychiatry. Wallace D, Frison S, et al. Abstinent vs non-abstinent 19 92;149:810–815. contingent housing for cocaine dependent homeless: 14 .Kessler RC, McGonagle KA, Zhao S, et al. Life- effects of housing together or apart. Paper presented at: time and 12-month prevalence of DSM-III-R psychiat- College on Problems of Drug Dependence 62nd An- ric disorders in the United States. Results from the Na- nual Scientific Meeting; June 17–22, 2000; San Juan, tional Comorbidity Survey. Arch Gen Psychiatry. 1994; Puerto Rico. 51:8–19. 32. Milby JB, Schumacher JE, Wallace D, et al. Day 15.Koegel P, Burnam MA, Farr RK. The prevalence treatment with contingency management for cocaine of specific psychiatric disorders among homeless indi- abuse in homeless persons: 12-month follow-up. J Con- viduals in the inner city of Los Angeles. Arch Gen Psy- sult Clin Psychol. 2003;71:619–621. chiatry. 1988;45:1085–1092. 33. Rawson RA, Huber A, McCann M, Shoptaw S, 16.Rahav M, Link BG. When social problems con- Farabee D, Ling W. A comparison of contingency man- verge: homeless, mentally ill, chemical misusing men agement and cognitive-behavioral approaches during in New York City. Int J Addict. 1995;30:1019–1042. methadone maintenance treatment for cocaine dependence. Arch Gen Psychiatry. 2002;59:817–824. 17.Rahav M, Rivera JJ, Nuttbrock L, et al. Character- istics and treatment of homeless, mentally ill, chemical- 34. Higgins ST, Wong CJ, Badger GJ, Ogden DE, abusing men. J Psychoactive Drugs. 1995;27:93–103. Dantona RL. Contingent reinforcement increases co- 18.Regier DA, Farmer ME, Rae DS, et al. Comorbid- caine abstinence during outpatient treatment and 1 year ity of mental disorders with alcohol and other drug of follow-up. J Consult Clin Psychol. 2000;68:64–72. abuse. Results from the Epidemiologic Catchment Area 35. Maude-Griffin PM, Hohenstein JM, Humfleet GL, (ECA) Study. JAMA. 1990;264:2511–2518. Reilly PM, Tusel DJ, Hall SM. Superior efficacy of 19.Rosenheck R, Leda C, Gallup P, et al. Initial as- cognitive-behavioral therapy for urban crack cocaine sessment data from a 43-site program for homeless abusers: main and matching effects. J Consult Clin Psy- chronic mentally ill veterans. Hosp Community Psychia- chol. 1998;66:832–837. try. 1989;40:937–942. 20. O’Brien CP, Alterman A, Walter D, Childress AR, McLellan AT. Evaluation of treatment for cocaine dependence. NIDA Res Monogr. 1989;95:78–84. 21. Milby JB, Schumacher JE, Raczynski JM, et al. Suf- ficient conditions for effective treatment of substance abusing homeless persons. Drug Alcohol Depend. 1996; 43:39–47. 22. Milby JB, Schumacher JE, McNamara C, et al. Ini- tiating abstinence in cocaine abusing dually diagnosed homeless persons. Drug Alcohol Depend. 2000;60: 55–67. 23. Stewart B. McKinney Homelessness Assistance Act (1987), Public Law 100-177. 24. Diagnostic and Statistical Manual of Mental Disor- ders, Third Edition, Revised. Washington, DC: American Psychiatric Association; 1987. 25. Derogatis LR. SC-R-90: Administration, Scoring and Procedures Manual. Baltimore, Md: Clinical Psycho- metric Research; 1977. 26. Derogatis LR, Cleary PA. Confirmation of the di- mensional structure of the SCL-90-R: a study in construct validation. J Clin Psychol. 1977;33:981–989. 27.Drake RE, McHugo GJ, Biesanz JC. The test-retest reliability of standardized instruments among homeless persons with substance use disorders. J Stud Alcohol. 19 95;56:161–167. 28. Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika. 1986;73: 13–22. 29. Schumacher JE, Milby JB, Caldwell E, et al. Treat-

July 2005, Vol 95, No. 7 | American Journal of Public Health Milby et al. | Peer Reviewed | Research and Practice | 1265  RESEARCH AND PRACTICE 

Written Parental Consent in School-Based HIV/AIDS Prevention Research

| Catherine Mathews, PhD, Sally J. Guttmacher, PhD, Alan J. Flisher, PhD, FCPsych, Yolisa Mtshizana, BA, Andiswa Hani, BTech, and Merrick Zwarenstein, MBBCh

More than 60% of the 39.4 million people Objectives. We examined the process of obtaining “active,” written parental living with HIV/AIDS live in sub-Saharan 1 consent for a school-based HIV/AIDS prevention project in a South African high Africa. In South Africa, 1500 new HIV infec- school by investigating (1) parental consent form return rates, (2) parents’ recall tions occur daily, 60% of these in persons and knowledge of the research, and (3) the extent to which this consent proce- 2 aged 15 to 25 years. Adolescence, when sex- dure represented parents’ wishes about their child’s involvement in the research. ual activity generally commences but behav- Methods. This cross-sectional descriptive study comprised interviews with parents ior patterns are still malleable, is a critical time of children in grades eight and nine in a poor, periurban settlement in Cape Town. for HIV prevention.3,4 Schools are particularly Results. Within 2 weeks, 94% of 258 parents responded to a letter requesting important for adolescent sexual behavior in- written consent and of those, 93% consented, but subsequent interviews showed terventions, especially where health services that 65% remembered seeing the consent form. At the end of the interview, 99% are inaccessible.4 Yetinsub-Saharan Africa, consented to their child’s participation. Conclusions. These findings challenge many of the assumptions underlying ac- few school-based AIDS prevention programs tive written parental consent. However, they should not be used to deny adoles- exist, and even fewer have been evaluated 4 cents at high risk of HIV infection the opportunity to participate in prevention trials. using randomized, controlled designs. Rather, researchers together with the communities in which the research is under- Because adolescents may not legally con- taken need to decide on appropriate informed consent strategies. (Am J Public sent to participate in research in South Africa, Health. 2005;95:1266–1269. doi: 10.2105/AJPH.2004.037788) parental consent is needed. South African research ethics committees accept “passive” parental consent procedures. Parents/guardians tive consent procedures in school-based re- the trial was to assess the efficacy of a story- are informed (usually by letter) about the na- search on youths’ risk behaviors produce less telling and drama-based intervention for stu- ture of the research, their permission for their representative samples and underrepresent dents in grades eight and nine. The interven- child’s participation is sought, they are alerted minority and high-risk youths in the sample, tion took place in 1 school while a second to their right to refuse this, and they are given in comparison with passive procedures.9,10 school served as a control school. The paren- directions for whom to contact if they should The National Institutes of Health funds tal consent investigation took place at the in- wish to do so. In the absence of such notifica- South African school-based HIV prevention tervention school. tion, it is assumed that consent has been research but requires active consent, so we The intervention school was the only high given. have examined the process of obtaining ac- school in a poor, periurban settlement of ap- In the United States, by contrast, “active” tive, written parental consent in the South proximately 1200 plots and 10000 people in written parental consent for adolescents’ par- African context. This research was under- the Southern Cape Peninsula, South Africa. ticipation in research is required by the US taken in 2003 in a high school participating The settlement area is well demarcated and Department of Health and Human Services, in a school-based HIV prevention controlled geographically isolated from other high-density the US National Institutes of Health, and most trial. We investigated (1) parental consent areas. Apart from its relative geographic isola- ethics review boards.5 The parents/guardians form return rates, (2) parents’ knowledge and tion, the area is typical of poor periurban set- must be informed and must positively notify recall of the research, and (3) the extent to tlements in South Africa, characterized by the school or researcher in writing that they which active, written parental consent repre- high unemployment rates, shack dwellings, permit their child to participate in the re- sented parents’ wishes about their child’s in- and a high risk of HIV infection. The preva- search, or else it is assumed that consent has volvement in the research. lence of HIV infection among the 330 preg- been denied. nant women attending antenatal clinics in the In the United States and Canada, many METHODS settlement during 2003 was 21% (oral com- parents fail to return consent forms, and so munication, Keren Middelkoop, 2004). even though research shows that the over- Study Population and Setting The parental consent investigation was whelming majority of “nonresponding” par- This research was undertaken in a school conducted with the parents/guardians of all ents approved of their child’s participation, it participating in a school-based HIV preven- the 258 children in grades eight and nine at is presumed that consent was denied.6–8 Ac- tion controlled trial during 2003. The aim of the high school (median age, 15 years).

1266 | Research and Practice | Peer Reviewed | Mathews et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

TABLE 1—Information Conveyed to A parental consent letter was composed to tion and make an appointment for an in-person Parents/Guardians: Southern Cape include clear information about the research interview. If there was no biological parent in Peninsula, South Africa, 2003 (Table 1). It was translated in Xhosa (the the child’s home, we invited the guardian who home language of the majority of the stu- assumed primary responsibility for the child’s • The purpose of the study is to prevent HIV infection dents), and both the English and Xhosa ver- care and schooling to participate. If the par- through a storytelling and drama-based sions were pretested to ensure that they were ents/guardians did not have a telephone, the education program conducted at the school understandable and conveyed the appropriate fieldworkers visited them to request consent for • The organizations involved in the research are the messages. A consent form (in English and their participation and make an appointment University of Cape Town, the South African Medical Xhosa) was attached to the letter. for the interview. The fieldworkers were Research Council, and the Pulpit Drama Group A member of the research team visited blinded as to whether the parents/guardians • The children will be asked to fill in questionnaires each class in grades eight and nine and de- had been classified as a consenter, refuser, or about their preparedness to protect themselves scribed the research to the students, in their nonresponder. Interviews were conducted ei- from HIV infection, and about their sexual behavior home language. She explained that parents/ ther at the school or at the home, depending • Students will not write their names on the guardians would have to indicate in writing on the preference of the parents/guardians. questionnaires whether or not they wished their child to par- During the interview, the fieldworkers • Students’ responses to the questionnaires will be ticipate in the research before students could showed the parents/guardians a copy of the confidential and their teachers will not see their be invited to participate. The researcher gave consent letter and form (which was easy to responses each student in the class a copy of the con- remember because, unlike other letters from • Participation is entirely voluntary, and no sent letter and the attached consent form to the school, it had been printed on brightly col- disadvantages will accrue to the parents or their take home to their parent or guardian. ored paper) and asked whether the parents/ child if they prefer that their child not participate The researcher revisited each class 3 days guardians had seen it, read it, signed it, and • There will be no HIV testing later and collected returned consent forms. returned it, and what they remembered about • If a child is younger than 21 years, he or she will The students who had not returned consent the research. Then the fieldworkers carefully only be allowed to participate if the parent or forms were given a second copy of the con- informed the parents/guardians about the re- guardian has consented to this participation by sent letter and form and a reminder of the search, making sure to correct any misunder- completing, signing, and returning to the school importance of returning the form. Remaining standings they had about the research. Once the consent form attached to the letter forms were collected in class 1 week and the parents/guardians clearly understood • The parent/guardian or child may withdraw consent 2weeks after the initial letter was sent home. the research, the fieldworkers asked them at any point during the study, by notifying the The research team then compiled a list of whether or not they wished their child to par- school or the research staff, and no penalties will students in each class, noting those whose ticipate. The fieldworkers had been explicitly accrue to them parents had given written consent (consen- trained to conduct the interview so that the • Parents/guardians are welcome to contact the ters), those whose parents had denied con- parents/guardians did not feel any sense of school principal or the researchers if they have sent in writing (refusers), and those whose coercion to consent, or any need to please the any questions or would like to discuss the project parents had not returned the consent form fieldworkers by consenting. (contact details were provided) (nonresponders). In the control school, a passive parental RESULTS consent procedure was used. Obtaining Informed Parental Consent Parental Consent Form Return Rates After obtaining permission from the West- Investigating Parents’ Responses All 258 students in grades eight and nine ern Cape Education Department and the to the Research were included in the study and given consent school principal, we addressed a meeting of Two weeks after the initial consent letter letters for their parents. Within 2 weeks, the the school governing body (consisting of par- was given to the students, the research team students returned 243 (94%) of the 258 pa- ent and teacher representatives). After obtain- sent letters to all parents/guardians in the in- rental consent forms, of which 225 (93%) in- ing consent from this group, we sent a notice tervention school informing them that a field- dicated parental consent and 18 (7%) paren- to all the parents of the students in grades worker would telephone or visit them to re- tal refusal (Figure 1). eight and nine, inviting them to learn about quest an interview about their opinions about the research at an evening meeting at the the research and their child’s participation in Demographic Details of Parents/ school. Approximately 60 of the 258 invited it. It explained that no disadvantages would Guardians Who Were Interviewed parents attended. Table 1 describes the infor- accrue to them or their child if they chose to Many students (73; 30%) did not live with a mation conveyed to them during the meeting. refuse to be interviewed by completing and biological parent, in which case their guardian Parents were also informed that they would be returning a form, attached to the letter. was interviewed. We interviewed 246 (95%) asked for an interview to discuss their involve- The fieldworkers then telephoned each par- of the 258 parents/guardians. Only 76 (31%) ment in decisions about school-based research. ent/guardian to request consent for participa- parents/guardians had completed grade 10 or

July 2005, Vol 95, No. 7 | American Journal of Public Health Mathews et al. | Peer Reviewed | Research and Practice | 1267  RESEARCH AND PRACTICE 

FIGURE 1—Consent obtained through returned forms compared with consent obtained during an interview with parents/guardians.

higher. Most parents/guardians (177; 72%) was important for their child to participate be- in our study knew that their school had been were employed, predominantly in domestic cause it was a program sanctioned by the allocated to the intervention arm (as opposed work. Few parents/guardians (91; 37%) had school. Only 2 of the 18 (11%) parents/ to the comparison arm) of the trial. We sus- ever spoken with their child about HIV. Many guardians classified as refusers based on the pect that some students signed “on behalf” of parents (151; 61%) had attended a meeting at consent forms preferred that their child not their parents, mostly in their enthusiasm to the school during the last academic year, but participate. We interviewed 12 of the 15 nonre- participate in the intervention. Ideally, paren- only 57 (23%) knew the name of the principal. sponding parents/guardians, and all 12 (100%) tal consent needs to be obtained before allo- Only 65 (26%) parents/guardians stated that said they wished for their child to participate. cating schools to the intervention or compari- they had attended the meeting called by the The fieldworker identified 87 parents/ son arm of a trial. Our finding of a high school to explain the nature of our research. guardians who reported that they had not seen prevalence of parental consent may not be the consent letter and form. However, the stu- generalizable to other studies (especially if Parents’/Guardians’ Knowledge and dents had returned a signed form for 81 they are descriptive as opposed to interven- Recall of the Research (93%) of these 87 parents/guardians (Figure tion studies) nor to other communities. An- Only 159 of the 246 parents (65%) stated 2). Most (71; 88%) of these forms indicated other potential limitation is that it might have that they had received the consent letter and consent, and only 10 (12%) indicated refusal. been difficult for parents/guardians to refuse form. Of those who received the letter, many consent during the interviews conducted by were unaware about what was in it (Table 2). DISCUSSION fieldworkers, resulting in a social desirability bias. However, our careful selection and train- Did the Consent Form Reflect Parental This study has several limitations. At the ing of the fieldworkers, and the frank appreci- Wishes? time of seeking parental consent, the students ation expressed by parents for the opportuni- When the fieldworker asked parents/ guardians whether they wished their child to TABLE 2—Recall of the Research Among Those Parents/Guardians Who Stated That They participate in the research, 243 (99%) of the Had Received the Consent Letter (n=159) 246 parents/guardians interviewed stated that they did (Figure 1). All 243 wished their child Did Parents/Guardians Know That Correct Knowledge, No. (%) to participate so that they would be better pre- Students would not be tested for HIV? 76 (48) pared to protect themselves from HIV. Of After consenting, they were allowed to withdraw their child from the research at any time? 98 (62) these, 18 (7%) wanted their child to share the Students would be asked to answer questionnaires? 106 (67) new knowledge about HIV prevention with Teachers would not see students’ completed questionnaires? 49 (46)a other family members; 11 (5%) wished their Students would not be required to write their names on the questionnaires? 30 (28)a child to participate because, as parents, they found it difficult to talk about HIV at home; aOnly the 106 respondents who knew that students would answer questionnaires were included in the denominator. and 16 (7%) specifically stated that they felt it

1268 | Research and Practice | Peer Reviewed | Mathews et al. American Journal of Public Health | July 2005, Vol 95, No. 7  RESEARCH AND PRACTICE 

Health and Primary Health Care, University of Cape Town, Cape Town, South Africa. Sally J. Guttmacher is with the Department of Health Studies, New York Univer- sity, New York, NY. Alan J. Flisher is with the Department of Psychiatry and Mental Health, University of Cape Town. Merrick Zwarenstein is with the Knowledge Translation Programme and Department of Health Policy Management and Evaluation, Faculty of Medicine, University of Toronto, Toronto, Canada. Requests for reprints should be sent to Catherine Math- ews, Health Systems Research Unit, Medical Research Council, PO Box 19070, Tygerberg 7505, South Africa (e-mail: [email protected]). This article was accepted March 27, 2004.

Contributors M. Zwarenstein, C. Mathews, S. Guttmacher, and A. Flisher originated the idea and wrote the protocol. C. Mathews, Y. Mtshizana, and A. Hani supervised the data collection and analysis. All authors contributed to the final article.

FIGURE 2—Fate of consent forms sent to parents/guardians by means of their child. Acknowledgments Thanks to Linda-Gail Bekker, Fiona Mendelson, the participating school, and the Western Cape Education Department, who made this research possible. The ties afforded their children by the research, this standard of consent, if widely applied, South African Medical Research Council and the New York University Challenge Fund funded this research. suggests that this was not the case. would artificially lower the proportion of par- Three assumptions are implicit in the ap- ents who wish to have their child included in Human Participant Protection proach of obtaining active informed consent low-risk studies of educational and informa- This research was approved by the New York Univer- from parents of adolescents participating in tional interventions. This applies to both non- sity institutional review board and the University of school-based research. These assumptions are responding and refusing parents: for both Cape Town ethics research committee. that (1) having a signed form in which active groups the vast majority of parents gave con- informed consent is indicated provides suffi- sent once fully informed. Active consent re- References 1. Global summary of the AIDS epidemic, Decem- cient evidence for the actual existence of such stricts access to interventions, places children ber 2004. UNAIDS Web page. Available at: http:// consent, (2) the person providing informed at odds with their parents, and skews the re- www.unaids.org/wad2004/report_pdf.html. Accessed consent understands the study, and (3) a writ- sulting studies in ways that are likely to overes- January 10, 2005. ten refusal of informed consent reflects the timate the effect of interventions because of 2. Gilbert L, Walker L. Treading the path of least actual wishes of the parent/guardian. How- volunteer bias. We support a return to the use resistance: HIV/AIDS and social inequalities—a South African case study. Soc Sci Med. 2002;54:1093–1110. ever, in the present study none of these as- of passive consent procedures for low-risk 3. Eaton L, Flisher AJ, Aarø LE. Unsafe sexual be- sumptions were supported. studies of educational and informational inter- haviour in South African youth. Soc Sci Med. 2003;56: These research findings should not be used ventions (many of which are consistent with 14 9–165. to deny adolescents at high risk of HIV infec- the objectives of the new South African school 4. Kaaya SF, Mukoma W, Flisher AJ, Klepp K-I. tion the opportunity to participate in prevention curriculum), and we suggest that a carefully School-based sexual health interventions in sub-Saharan Africa: a review. Soc Dynamics. 2002;28(1):64–88. trials. Rather, the ethical conundrum that is worded, pretested parent/guardian consent 5. Protection of Human Subjects. 45 CFR §46 (2001). raised by this research (weighing the benefits of form be developed for each such study. Fur- 6. Ellickson PL, Hawes JA. An assessment of active developing effective widespread HIV preven- ther, we suggest the use of methods other than versus passive methods for obtaining parental consent. tion programs at schools against the harms re- asking students to solicit their parents’ consent. Eval Rev. 1989;13:45–55. sulting from imperfect parental informed con- These could involve sending letters home in 7. Esbensen FA, Deschenes EP, Vogel RE, Arboit K, sent procedures) needs to be acknowledged. the post, or giving letters to parents when they Harris L. Active parental consent in school-based research. Eval Rev. 1996;20:737–753. Researchers need to explore the values and register their child at the school at the begin- preferences of the communities in which the re- ning of the academic year, and providing infor- 8. Baker JR, Yardley JK, McCaul K. Characteristics of responding-, nonresponding- and refusing-parents in an ado- search is conducted and to work together with mation to parents at meetings held to discuss lescent lifestyle choice study. Eval Rev. 2001;25:605–618. these communities to decide on appropriate school fees (which are well attended). 9. Esbensen FA, Miller MH, Taylor TJ, He N, Freng A. strategies to gain the kind of informed consent Differential attrition rates and active parental consent. that is appropriate for studies of this nature. Eval Rev. 19 9 9;23:316–335. The current implicit gold standard of con- About the Authors 10. Anderman C, Cheadle A, Curry S, Diehr P, Shultz L, Catherine Mathews, Yolisa Y. Mtshizana, and Andiswa Wagner E. Selection bias related to parental consent in sent is active written consent by parents. We Hani are with the Health Systems Research Unit, South school-based survey research. Eval Rev. 19 95;19: believe that in South African school settings, African Medical Research Council and the School of Public 663–674.

July 2005, Vol 95, No. 7 | American Journal of Public Health Mathews et al. | Peer Reviewed | Research and Practice | 1269 MARKETPLACE Annual Meeting and Exposition Dates and Sites:

2005 • November 5–9 • New Orleans, LA 2006 • November 4–8 • Boston, MA 2007 • November 3–7 • Washington, DC 2008 • October 25–29 • San Diego, CA 2009 • November 7–11 • Philadelphia, PA

3rd International Congress on Developmental Origins of Health and Disease (DOHaD)

November 16 - 20, 2005 The Westin Harbour Castle Hotel Toronto, Canada

Abstract submission begins May 23, 2005

Visit http://www.dohadsoc.org for the latest information

July 2005, Vol 95, No. 7 | American Journal of Public Health Marketplace | 1271 at the 133rd Annual Meeting and Exposition

Public Health CareerMart is the place to find and post jobs in the field of public health. This full-fledged career guidance and management resource center offers APHA members many benefits. Please go to www.apha.org/career for discounts on job posting and access to the Evidence-Based APHA’s Career Coaching Services. Policy & Practice Due to the overwhelming request from employers and job seekers, Public Health CareerMart is expanding! Additional computers and interview booths have been added to the new Everything APHA in the Public Health American Public Expo. Prior to the Annual Meeting, job seekers and employers may submit Health Association Public Health CareerMart their resumes and job announcements online at www.apha.org/career. This 800 “I” Street, NW is the perfect opportunity for employers and job seekers to meet face-to-face! Washington, DC 20001 [email protected] For booth reservation form go to www.apha.org/career/

1272 | Marketplace American Journal of Public Health | July 2005, Vol 95, No. 7 July 2005, Vol 95, No. 7 | American Journal of Public Health Marketplace | 1273 1274 | Marketplace American Journal of Public Health | July 2005, Vol 95, No. 7 P OSTDOCTORAL F ELLOWSHIP IN C ARDIOVASCULAR D ISEASE E PIDEMIOLOGY

NIH-sponsored two- to three-year fel- JOB OPPORTUNITIES lowship is available emphasizing research methods in the epidemiology of cardiovascular disease (www.epi.umn. ILLINOIS edu/academic/fellowships/cvdepfel.sh tm). Fellows contribute to and gain Epidemiologist: The SIU School of and employment is contingent on the competency in designing, adminis- Medicine, Department of Surgery, result of a criminal background tering, and analyzing cardiovascular Division of Cardiothoracic Surgery is investigation. Interested individuals population studies and can seek an seeking an individual to pursue research should submit a letter of interest MPH degree. Candidates must have in the area of cardiothoracic diseases. accompanied by a Curriculum Vita , a a doctoral degree and permanent U.S. residency status. The candidate should have a Ph.D. in statement of research interests and the Epidemiology with a strong record of name of at least three references to: Inquiries: Dr. Aaron R. Folsom, high quality scientific productivity and Alice Costa, SIU School of Medicine, P. Division of Epidemiology & be willing to collaborate with research O. Box 19638, Springfield, IL 62794- Community Health, School of programs of other divisions in the 9638. Deadline to apply is June 30, Public Health, University of Department of Surgery and with basic 2005 or until the position is filled. SIU Minnesota, Suite 300, 1300 scientists. Applicants must be self- School of Medicine is an AA/EEO South 2nd St., Minneapolis, MN motivated to develop an externally employer. 55454-1015 (email: funded research program. Sufficient [email protected]). support to allow the right candidate to succeed will be provided. This position The University of Minnesota is an equal opportunity educator and employer and en- has been designated security sensitive, courages applications from people of color.

July 2005, Vol 95, No. 7 | American Journal of Public Health Job Opportunities | 1275 HAROLD R. JOHNSON CHAIR IN GERONTOLOGY Assistant Professor Position in The University of Michigan School of Social Work seeks to appoint a nationally/internationally recognized scholar the Graduate Program in to the Harold R. Johnson Chair in Gerontology. This en- dowed chair seeks to advance understanding of the needs Medical Education for the of older adults in health care systems. Latino Community Applicants should have substantial intellectual accomplishments, including scholarship and grants in areas re- University of California, Irvine lated to the quality of life, health outcomes, and the care of older adults and their families—especially those with The University of California,Irvine is seeking nominations and chronic illnesses and disabilities. applications for an Assistant Professor position in the Department of Environmental Health, Science, and Policy in For more details please visit: http://websvcs.itcs.umich. the School of Social Ecology, who will teach in the interdisci- edu/jobnet/job_posting.php?postingnumber=027005 plinary PROGRAM IN MEDICAL EDUCATION FOR THE LATINO Prospective candidates should submit a letter of applica- COMMUNITY (PRIME-LC). This program has been created to tion, curriculum vitae, three letters of reference, and three train physicians with the knowledge, skills and attitudes to samples of written work. Please submit applications for serve the medical needs of the underserved Latino commu- this position by October 1, 2005 for full consideration to: nity. In order to equip students with the medical, academic, and advocacy skills necessary to advance the health agendas Paula Allen-Meares, Dean and Norma Radin of that community, PRIME-LC coordinates two existing de- Collegiate Professor of Social Work/Professor of grees: the MD offered by the UCI College of Medicine,and the Education, University of Michigan, School of Social MS in Environmental Health Science and Policy offered by the Work, 1080 S. University, 4720 SSWB, Ann Arbor, MI Department of Environmental Health, Science, and Policy in 48109-1106 the School of Social Ecology. The coordination of those two The University of Michigan is committed to a policy of programs, including use of some summer sessions, will allow non-discrimination and equal opportunity for all persons regardless of students to complete both degrees in five years. race, sex, color, religion, creed, national origin or ancestry, age, marital status, sexual orientation, disability, or Vietnam-era veteran status in REQUIREMENTS employment, educational programs and activities, and admissions. In addition to a doctorate in a field relevant to the new training program, the successful candidate will be someone who can play a leading role in the growth and further development PhD Job Advertisement-Center of the Graduate Program in Medical Education for the Latino Community. The specific field of doctoral training and re- for Healthy Communities search and teaching interests is open. For appointment at an The Center for Healthy Communities, Department of Family and advanced Assistant Professor rank, evidence of success in se- Community Medicine, Medical College of Wisconsin, is seek- curing extramural funding to support research and graduate ing an Assistant professor to work on externally-funded commu- students is desired. Candidates should submit (1) a letter de- nity-based participatory research projects, federally supported scribing research and teaching interests; (2) curriculum vitae curriculum development projects, and to participate in the edu- including email address; (3) up to three representative publication of medical students and residents. cations; and (4) contact information for five references to: Qualifications: A doctorate in public health, health services research, sociology, urban studies, health psychology, adult educa- John M.Whiteley, Search Committee Chair tion, health education, social work or other relevant field is re- Attention: Margaret Wyvill (room 205) quired. Candidate must have experience working in underserved Department of Environmental Health, Science communities and understand how community organizations im- School of Social Ecology pact. A strong commitment to community-based participatory University of California, Irvine research is also needed. Experience in grant writing, both federal Irvine, CA 92697-7070 and foundation, is required. Current projects in the Center for Healthy Communities include The position will remain open until filled. Applications re- health advocacy, health literacy, healthy aging, cancer prevention ceived by September 15th, 2005, will have priority. The program, rural substance abuse and mental health research, on- University of California, Irvine is a top-ranked public univer- line educational module development, youth and domestic vio- sity dedicated to research, scholarship, and community serv- lence research, and elder abuse research. The successful appli- ice. Founded in 1965, UCI is among the fastest-growing cant will be expected to collaborate on some of these projects and University of California campuses,with more than 24,000 un- develop their own externally funded research. dergraduate and graduate students and about 1,400 faculty Salary and benefits commensurate with experience. members. It is located in dynamic Orange County, between Interested applicants should forward a current CV or resume and Los Angeles and San Diego, six miles from the Pacific Ocean. three letters of reference to Veronica Ruleford, Faculty Programs The University has an active career partner program, is an Coordinator for the Department of Family and Community equal opportunity employer committed to excellence Medicine, Medical College of Wisconsin, 8701 Watertown Plank through diversity, and has a National Science Foundation Road Milwaukee, WI 53226. Phone: (414) 456-4243 E-mail: ADVANCE Gender Equity Program. [email protected]. EOE

1276 | Job Opportunities American Journal of Public Health | July 2005, Vol 95, No. 7 Associate/Full Professor, Public Health (Tenure Track) Division of Epidemiology and Biostatistics

The Division of Epidemiology and Biostatistics at the University of Arizona, Mel and Enid Zuckerman College of Public Health seeks AJPH WEBSITE a well-established epidemiologist for a tenure-eligible faculty position to participate in the College's teaching programs and to contribute actively to the College's research and service mission. The online Journal joins the many The College offers epidemiology degrees at the MPH, MS, and online benefits offered to APHA PhD levels. Candidates with research experience in areas includ- members at www.apha.org, including: ing infectious diseases, genetics, and cardiovascular epidemiol- • full issues of The Nation’s Health ogy are encouraged to apply. Outstanding candidates with expertise in any other sub-specialties will be considered. The •APHA Membership Directory College is an active partner in several regional multi-disciplinary, and community-based research projects that address health issues • Annual Meeting information relative to border communities and indigenous populations. Interested applicants should visit http://www.uacareertrack.com and search Job #32258 for a complete listing of application instructions. Review of applications will commence September 1, 2005 and continue until the position is filled. Check out the Journal at As an equal employment opportunity and affirmative action employer, the University of Arizona recognizes the power of a diverse community and encourages www.ajph.org applications from individuals with varied experiences, perspectives and backgrounds including minorities, women, persons with disabilities and veterans.

Physician Epidemiologist The State of New Hampshire is seeking a Physician for the position of State Epidemiologist for the Division of Public Health Services (DPHS). The New Hampshire Department of Health and Human Services is seeking applicants for the position of State Epidemiologist and Chief, Bureau of Disease Control and Health Statistics for the State of New Hampshire. This position is directly responsible for all medical, programmatic, health data collection and assessment, and financial activities for the Bureau of Disease Control and Health Statistics, with a total of eighty staff members. The State Epidemiologist will report to the Director of the Division of Public Health Services and exercise direct responsibility for: the communicable disease control and surveillance sections, immunization, environmental health tracking, health statistics and data management section and food protection section. The individual will develop and maintain an innovative, integrated system of disease control and health data management that serves the needs of medical providers, policymakers and the citizens of the state. A degree in medicine from an accredited school of medicine or school of osteopathy is required along with a graduate level of training in epidemiology. A degree program such as an MPH or extended epidemiological experience as exemplified by an EIS fellowship is preferred. Five years experience in public health management with demonstrated ability for management and supervision is also required. The individual must hold or be eligible to hold a medical license within the State of New Hampshire. The position is located in Concord, NH, which is the state capital, a small city approximately one-hour drive from the ocean, lakes, mountains and Boston that offers an attractive lifestyle in a family-friendly community. The Physician Epidemiologist is a 37.5-hour per week minimum position with 15 vacation days per year, with days available at the start of employment. We offer an excellent benefit package, which includes fully paid health/dental care for the employee and dependents, 12 paid holidays, 15 sick days per year, state retirement plan, tuition reimbursement and life insurance. EOE For further information, please contact Jennifer Ritchings at 603-271-4612.

July 2005, Vol 95, No. 7 | American Journal of Public Health Job Opportunities | 1277 Dynamic new department leadership and an energetic multi-cultural faculty group and an active research and training center seek to recruit three MD/DO/PhD colleagues to join an expanding research group in the Department of Family and Community Medicine at the University of Arizona. Those who are interested in joining us in the tri-cultural southwestern city of Tucson may access the position descriptions at https://uacareertrack.com

We are looking for two experienced researchers with a track record in funded research,who will continue their own research and collaborate with and mentor our junior faculty to become successful independent researchers. Research interests in the Department include diabetes,obesity,tobacco/substance abuse, integrative medicine, health services research, and health disparities particularly as these problems are relevant to medically underserved populations. A new electronic medical record system will provide unique opportunities for health systems research. Rank is open. Salary is competitive based on experience. This is posting #32154.

We are also looking for an experienced researcher with a track record in funded research and administration, who will continue their own research and collaborate with other colleagues, to serve as co-director of the Native American Research and Training Center (NARTC). Research interests preferred include diabetes, tobacco/substance abuse, health services research, integrative medicine, and health disparities particularly as these problems are relevant to American Indian/Alaska Native populations. Rank is open. Salary is competitive based on experience. This is posting #32297

As an equal opportunity and affirmative action employer, the University of Arizona recognizes the power of a diverse community and encourages applications from individuals with varied experiences, perspectives and backgrounds.M/W/D/V.

1278 | Job Opportunities American Journal of Public Health | July 2005, Vol 95, No. 7 Medical Officer (Epidemiologist) Centers for Disease Control and Prevention Salary: $88,893 - $111,753 Location: DHHS, Centers for Disease Control and Prevention, Coordinating Center for Health Promotion, Center for Chronic Disease Prevention and Health Promotion, Division of Reproductive Health, Women's Health and Fertility Branch, 2900 Woodcock Blvd., Atlanta, GA 30341 Mailing Address: 4770 Buford Hwy, NE Mailstop K-34, Atlanta, GA 30341 Type: Full Time Medical Officer (Epidemiologist) The Centers for Disease Control and Prevention (CDC) is seeking a Medical Officer for the Women's Health and Fertility Branch to conduct a variety of epidemiologic studies and clinical trials that address birth outcomes and perinatal HIV transmission among HIV-infected women. Currently there is a large, complex clinical trial in Malawi assessing perinatal and postnatal HIV transmission among HIV-infected women that will enroll more than 2500 mother-infant pairs and involves a factorial design.This study which involves administration of antiretroviral medication to both mothers and infants requires close clinical monitoring and careful evaluation of side effects. A medical officer with current clinical HIV expertise and experience with clinical trials is urgently needed to play a leadership role in providing timely consultation to clinicians in the field, monitoring safety concerns including the preparation of reports for the Data and Safety Monitoring Board,and communicating with pharmaceutical companies that have donated drugs to the study. This study also includes 9 laboratory sub-studies that will require a wide variety of laboratory techniques and a medical officer with a basic science background in immunology. Full civil service benefits are provided, including insurance and retirement. The position is located in Atlanta, Georgia. Funding for relocation is not available. Applicants are required to be U.S. citizens. CDC is an equal opportunity employer and its facilities are designated as smoke-free. By July 31, 2005, please mail or e- mail your curriculum vitae, copy of transcript and medical license to Denise J. Jamieson, WHFB, CDC, 4770 Buford Highway, NE, MS-K34, Atlanta, GA 30341; (770) 488-6377; [email protected].

Required Education: Medical or Osteopathic degree with board certification in pediatric infectious diseases. Training or experience in immunology, HIV care and treatment, clinical trials, public health and epidemiology is desirable.

July 2005, Vol 95, No. 7 | American Journal of Public Health Job Opportunities | 1279 1280 | Job Opportunities American Journal of Public Health | July 2005, Vol 95, No. 7 Boston University School of Public Health

Chair of the Department of Social and Behavioral Sciences

The Boston University School of Public Health invites applications for the position of Chair of the Department of Social and Behavioral Sciences. The mission of the department stresses the study of social, cultural, environmental and behavioral factors that influence the health of individuals and populations, with particular emphasis on intervention strategies. The department places equal value on teaching, research, and practice-based service in order to accomplish this mission.

The department's 14-member faculty is multidisciplinary. Working together, it has developed a structured, tiered curriculum organized by topical clusters for the department's large, longstanding MPH program and for the recently initiated DrPH program. The departmental curriculum includes 28 courses. External research funding to faculty for grants and centers totals over $5,000,000 for 2006.

Candidates must have a doctoral degree with formal training or extensive experience in public health or a related field. Qualified individuals will bring an established research program and experience in teaching, practice, and management.

Interested applicants are invited to submit a curriculum vitae and a cover letter in confidence. The cover letter should briefly describe the ap- plicant's research, teaching, practice and management experience.

Materials should be sent to Roberta F. White, PhD.; Chair, Department of Social and Behavioral Sciences Search Committee; Department of Environmental Health-T2E; Boston University School of Public Health; 715 Albany Street, Boston, MA 02118. Email: [email protected].

Applications will be accepted through September 15, 2005.

Boston University is an equal opportunity employer. We encourage applications from women and minorities.

July 2005, Vol 95, No. 7 | American Journal of Public Health Job Opportunities | 1281