Stein, MA, Julie Hariton, MSW, and Sadye Bar- References Household Survey on Drug Abuse. Rockville, Md: tolomei, MSW; from the Rose F. Kennedy Children’s Substance Abuse and Mental Health Services Ad- Evaluation and Rehabilitation Center, University Af- 1. Juliana P, Goodman C. Children of substance ministration, Office of Applied Studies; 1998. filiated Program, Albert Einstein College of Medi- abusing parents. In: Lowinson JH, Ruiz P, Mill- 5. Fulroth R, Phillips B, Durand DJ. Perinatal out- cine, Herbert Cohen, MD, director, and Daniel Fried- man RB, Langrod JG, eds. Substance Abuse: A come of infants exposed to cocaine and/or heroin man, MPA, associate director for administration; from Comprehensive Textbook. 3rd ed. Baltimore, Md: in utero. Am J Dis Child. 1989;143:905–910. the Division of Substance Abuse, Department of Psy- Williams & Wilkins; 1997:665–671. 6. DeCristofaro JD, LaGamma EF. Prenatal expo- chiatry and Behavioral Sciences, Albert Einstein Col- 2. Black MM, Nair P, Kight C, Wachtel R, Roby P, sure to opiates. Ment Retard Dev Disabil Res Rev. lege of Medicine, Ira J. Marion, MA, executive di- Schuler M. Parenting and early development 1995;1:177–182. rector, Pat Sanders Romano, MS, deputy executive among children of drug-abusing women: effects 7. Chiriboga CA, Vibbert M, Malouf R, et al. Neu- director, Egidio Caparelli, director of business oper- of home intervention. Pediatrics. 1994;94:440–448. rological correlates of fetal cocaine exposure: ations, Patti Juliana, MSW, director of clinical serv- 3. Feig L. Drug Exposed Infants and Children: Ser- transient hypertonia of infancy and early child- ices, and Kathryn M. Williams, MSW, program man- vice Needs and Policy Questions. Washington, hood. Pediatrics. 1995;96:1070–1077. ager; and Emmanuel Shapira, MD, PhD (deceased), DC: Office of the Assistant Secretary for Plan- 8. Kliegman RM, Madura D, Kiwi R, Eisenberg I, and Karen Gore, chair in human genetics and direc- ning and Evaluation, US Dept of Health and Yamashita T. Relation of maternal cocaine use to tor, Human Genetics Program, Hayward Genetics Cen- Human Services; 1990:1–41. the risks of prematurity and low birth weight. J ter, Tulane University School of Medicine. 4. Preliminary Results From the 1997 National Pediatrics. 1994;124:751–756.

Trends in Crime and the Introduction of a Needle Exchange Program

ABSTRACT Melissa A. Marx, MPH, Byron Crape, MSPH, Ronald S. Brookmeyer, PhD, Benjamin Junge, MSc, Carl Latkin, PhD, David Vlahov, PhD, and Steffanie A. Strathdee, PhD Objectives. This study sought to de- termine whether introduction of a nee- Needle exchange programs have been im- 14 months of operation, 3438 active injectors dle exchange program would be associ- plemented to help reduce transmission of HIV enrolled in the program, of whom 86% were ated with increased crime rates. and other blood-borne pathogens among in- African American; participants’ average age Methods. Trends in arrests were jection drug users1–4 and to increase the fre- was 42 years. compared in program and nonprogram quency of drug abuse treatment referrals5 areas before and after introduction of a among addicted individuals. Studies have Data Collection needle exchange program in Baltimore. shown that needle exchange programs do not Trends were modeled and compared via increase rates of drug use6 or increase num- Arrest records for the period February Poisson regression. bers of discarded needles or syringes7; because 1994 through October 1995 were obtained Results. No significant differences drug use has been associated with crime,8,9 from the Baltimore City Police Department. in arrest trends emerged. Over the study however, there are concerns that crime rates This enabled comparison of data 6 months period, increases in category-specific ar- may increase in areas surrounding needle ex- before and 6 months after introduction of the rests in program and nonprogram areas, change programs after their introduction.10,11 needle exchange program. The immediate im- respectively, were as follows: drug pos- We examined trends in arrests in Baltimore pact of the program was assessed, and sea- session, 17.7% and 13.4%; economically City before and after the opening of a needle sonal variations in arrests were examined in motivated offenses, 0.0% and 20.7%; re- exchange program. a subsequent 8-month period. Dates and lo- sistance to police authority, 0.0% and 5.3%; and violent offenses, 7.2% and 8.0%. Methods Conclusions. The lack of associa- The authors are with the Johns Hopkins School of tion of overall and type-specific arrest Study Population Hygiene and Public Health, Baltimore, Md. Melissa data with program implementation ar- A. Marx, Byron Crape, Benjamin Junge, David Vla- gues against the role of needle exchange In 1997, Baltimore City had 657250 res- hov, and Steffanie A. Strathdee are with the Depart- ment of Epidemiology; Ronald S. Brookmeyer is programs in increasing crime rates. (Am idents; the average age of these residents was 12 with the Department of Biostatistics; and Carl Latkin J Public Health. 2000;90:1933–1936) 35 years, and 60% were African American. is with the Department of Health Policy and Man- Approximately 50000 Baltimore residents reg- agement. David Vlahov is also with the Center for ularly used illicit drugs at that time, a substan- Urban Epidemiological Studies, New York Academy tial proportion of whom injected.13 of Medicine, New York City. In August 1994, the Baltimore City Health Requests for reprints should be sent to Stef- fanie A. Strathdee, PhD, Johns Hopkins School of Department opened a needle exchange pro- Public Health, Department of Epidemiology, 615 N gram housed at 2 locations. Program partici- Wolfe St, Room E-6010, Baltimore, MD 21205 pants were exempt from syringe possession (e-mail: [email protected]). laws within city limits. During the first This brief was accepted April 3, 2000.

December 2000, Vol. 90, No. 12 American Journal of Public Health 1933 cations of arrests and up to 5 criminal charges allowed to change in both areas. The hypothe- percentage change in overall arrests was higher were abstracted. ses tested were that changes in intercepts and in program (11.4%) than in nonprogram (7.6%) On the basis of input from law enforce- slopes would not significantly differ in pro- areas. However, there were no significant differ- ment, crime, and drug abuse experts, as well as gram and nonprogram areas before and after ences in arrest trends by category after program hypothesized associations of charges with nee- initiation of the needle exchange program and introduction relative to before program intro- dle exchange programs, arrest charges were that changes in arrest trends in program areas duction in program vs nonprogram areas (P>.05). categorized as follows: (1) drug possession, would be similar to changes in nonprogram (2) economically motivated offenses, (3) re- areas. We tested hypotheses using a likelihood sistance to police authority, or (4) violent of- ratio test with 2 degrees of freedom, account- Discussion fenses. Drug possession offenses included pos- ing for overdispersion.15 session of drug paraphernalia and distribution/ We found that increases in drug-related possession of heroin or cocaine. Economically arrests were not more pronounced in needle motivated offenses consisted of property theft Results exchange program areas than in other areas of (e.g., nonvehicular breaking and entering, bur- Baltimore after establishment of the program. glaries, vehicle break-in/theft) and prostitution, Overall, there were 53848 drug-related Although there were some differences in considered means of financing drug use. Re- arrests in Baltimore City during the study pe- category-specific arrest trends in areas of close sistance to police authority was defined as as- riod. Before introduction of the needle ex- proximity to the program relative to outlying saulting a police officer, resisting arrest, or vi- change program, there were 2500 drug-related areas, these differences were not statistically olating parole/probation; these offenses were arrests per month. After introduction of the significant. seen as indicators of increased frustration pos- program, there was a slight increase in the num- If the needle exchange program had di- sibly resulting from law enforcement practices. ber of drug-related arrests to 2775 per month. rectly influenced rates of drug use, a dispro- Violent offenses included homicide, assault, Wide fluctuations seen in monthly aver- portionate increase in drug possession arrests rape, and armed robbery, which were consid- ages of drug possession arrests citywide were would have been expected in program areas rel- ered potentially linked to drug trafficking. evidenced by high extradispersion values (co- ative to nonprogram areas.Although increases We defined the area of maximum pro- caine: 5.3; heroin: 9.8) in the Poisson model. in heroin and cocaine arrests after the program gram impact with data from an ongoing eval- Overall, the mean number of monthly arrests had been established were slightly more pro- uation of the program. We determined that 76% for drug possession rose slightly in program nounced in program than in nonprogram areas, of participants reported walking to the program areas, from 150 (range: 100–190) to 175 trends were not significantly different. Vari- site and that travel time for these individuals (range: 110–270). Average numbers increased ability in heroin and cocaine arrests reflected in averaged 15 minutes or less (median: 10 min- gradually in nonprogram areas, from 1020 the high model extradispersion values might be utes).14 At an estimated speed of 2.0 mi per (range: 825–1240) to 1160 (range: 925–1370) explained in part by “police sweeps,” which are hour (3.2 km per hour), 84% of participants per month. common and variable in Baltimore, especially were estimated to live within a 0.5-mi radius of Frequency of arrests for economically mo- in drug trafficking areas.Anecdotal reports in- the program site. Therefore, areas within a 0.5- tivated offenses remained constant in needle dicate that police sweeps were occurring early mi radius of the 2 program sites were com- exchange program areas, averaging 30 per after program introduction, and we hypothesize bined and designated as “program areas,” while month before and after introduction of the pro- that these sweeps may have contributed to the in- areas within the city limits but outside of these gram (ranges: 25–40 and 15–40, respectively). creased number of drug possession arrests ob- radii were deemed “nonprogram areas.” Arrests for economically motivated offenses served in program areas at that time. increased in nonprogram areas from 240 If the program had indirectly resulted in Data Analysis (range: 180–260) to 300 (range: 230–70) per increased drug use rates, we would expect to month over the same period. see drug users committing, and being arrested To examine the impact of the introduc- Similarly, numbers of individuals resisting for, a relatively higher number of economically tion of the needle exchange program on arrest arrest remained consistently low in program motivated crimes in program areas than in non- trends in Baltimore City, we assessed the num- areas, averaging 30 per month before and after program areas. Our data did not support this hy- ber of category-specific arrests before and after program introduction (ranges: 25–40 and 25– pothesis. In fact, a decrease was observed in program introduction. Mean numbers of 45, respectively). However, in nonprogram numbers of arrests for break-ins and burglaries monthly category-specific and overall arrests areas, the average number of individuals re- in program areas after the opening of the needle for program and nonprogram areas were cal- sisting arrest increased slightly from 300 per exchange program, whereas a slight increase culated (1) over the 6-month period before pro- month (range: 270–350) to 325 per month was observed in nonprogram areas. gram introduction and (2) over the 14-month (range: 285–370) during the same period. If the needle exchange program had in- period after program introduction. Percentage Average numbers of arrests for violent of- creased drug users’perceptions of lawlessness changes in mean numbers of arrests were then fenses dropped in program areas from 90 in areas of close proximity to the program, an calculated. (range: 70–100) to 80 (range: 70–100) per increase in instances of resisting arrest might To formally assess trends in monthly ar- month after introduction of the program. In- have occurred. However, numbers of arrests rests by proximity to the program site, we used creases in arrests for violence were seen over for assault on a police officer decreased in pro- Poisson regression models that considered the same period in nonprogram areas; the num- gram areas while increasing slightly in non- overall and category-specific arrests. A re- ber of such arrests increased from 820 (range: program areas. The opposite was true for num- gression line was fitted to log E(Yt), the log of 670–920) to 890 (range: 710–1100) per month. bers of arrests for parole or probation violation, the expected number of monthly arrests at Table 1 summarizes percentage changes in which increased slightly in program areas and month t, which allowed for different slopes and overall arrests and category-specific arrests in decreased in nonprogram areas. None of these intercepts in program and nonprogram areas program and nonprogram areas in the period after differences were statistically significant. before initiation of the needle exchange pro- introduction of the needle exchange program rel- If introduction of the needle exchange gram. At initiation, intercepts and slopes were ative to the preprogram period. The unadjusted program had resulted in a perception of anar-

1934 American Journal of Public Health December 2000, Vol. 90, No. 12 TABLE 1—Changes in Numbers of Arrests Before and After Introduction of the Needle Exchange Program (NEP): NEP and Non-NEP Areas, Baltimore, Md, 1994–1995

NEP Non-NEP Mean No. of Mean No. of Mean No. of Mean No. of Arrests, Arrests, Arrests, Arrests, NEP vs Time 1 Time 2 Change, % Time 1 Time 2 Change, % Non-NEPa, P

Overall 278.3 299.4 11.4 2221.8 2475.4 7.6 .40 Drug possession 147.2 173.3 17.7 1018.8 1155.6 13.4 .32 Cocaine 101.5 117.8 16.0 743.3 818.0 10.0 .34 Heroin 59.8 80.2 34.1 342.3 433.5 26.6 .30 Paraphernalia 17.5 17.4 –0.4 150.2 135.2 –10.0 .39 Economically motivated 32.5 32.4 0.0 240.8 290.6 20.7 .29 Break-ins and burglaries 27.0 24.1 –10.6 209.7 225.4 7.5 .25 Theft from vehicles 1.8 3.0 63.6 10.8 21.6 99.8 .26 Prostitution 3.8 5.5 43.5 20.7 46.1 122.9 .43 Resistance 32.8 33.2 0.0 305.2 321.4 5.3 .38 Assaulting officer 11.3 9.6 –15.5 81.8 86.1 5.3 .30 Resisting arrest 16.3 17.6 7.0 128.7 147.5 14.6 .37 Probation/parole violation 10.7 11.6 8.5 138.8 134.8 –2.9 .36 Violence 89.0 82.6 7.2 817.2 882.3 8.0 .34 Rape 3.8 4.6 21.1 38.8 45.6 17.5 .40 Murder 5.0 6.0 20.0 48.2 66.3 37.6 .38 Assault 79.0 70.6 –10.7 724.3 767.4 5.9 .35 Robbery 16.7 19.1 15.8 150.5 174.3 14.4 .41

Note. Time 1=6-month period before NEP implementation; Time 2=14-month period after NEP implementation. Arrest categories and types are not mutually exclusive and thus will not sum to overall drug-related arrests. aBased on likelihood ratio test derived from Poisson regression model.

chy, increased violence might be expected. policing practices in program areas; however, R.S. Brookmeyer directed and supervised data analy- However, violent assault arrests decreased in no record of official changes in policing prac- sis and contributed to the writing of the methods sec- program areas while increasing slightly in non- tices specific to program areas was found. tion of the manuscript. B. Junge conceived and planned program areas. Violence trends in program vs Our data are consistent with those gath- the study and performed the preliminary data analy- ses. C. Latkin directed study progress and reviewed nonprogram areas were, again, not statistically ered in a study conducted in Boston, Mass, in the final manuscript. D. Vlahov oversaw study progress different. which no differences in arrests were observed and contributed to major sections of the manuscript. In conducting this analysis, we assumed in needle exchange program areas and non- S.A. Strathdee assisted in interpretation of the sta- that coding of arrests was uniform across dif- program areas.19 Our data also corroborate re- tistical analysis and contributed to the writing, edit- ferent areas of the city at different times. How- ports from a study of New Haven, Conn, crime ing, and revision of the final manuscript. ever, even if this assumption were invalid, there trends20 and results from a multisite study of is no reason to believe that differences in cod- Manhattan, New Haven, San Francisco, Bos- ing would vary by region. In addition, we es- ton, and Portland, Oregon, crime trends.21 The Acknowledgments timated crime trends using arrest data. While lack of increases in arrests after the establish- We gratefully acknowledge financial support from the this approach may be subject to bias16 and may ment of the needle exchange program in Bal- National Institute on Drug Abuse through grant 09237. We appreciate the assistance provided by Keith limit the conclusions that can be drawn, police timore is consistent as well with survey data Harries of the Department of Geography, University department arrest data are considered superior showing that frequency of injection did not in- of Maryland, Baltimore County; Peter Beilenson, com- to self-reported crime and self-reported arrest crease among program participants during the missioner of health for Baltimore City; Michele data in that both of the latter measures may be same time period.14,22 Brown, Lamont Coger, and the Baltimore City Health subject to response bias.17,18 In conclusion, based on results of analy- Department needle exchange program staff; and Elise Arrest data may also be superior to crime ses of Baltimore City arrests, needle exchange Riley, Steve Huettner, Mahboobeh Safaeian, John Vertefeuille, Heena Brahmbhatt, Jennifer Mulle, and data because drug-related crime is often “vic- programs do not appear to be associated with Hanne Harbison of the Johns Hopkins School of Pub- timless” and therefore underreported. The va- increases in crime rates. This suggests that such lic Health Needle Exchange Program Evaluation. lidity of using arrests as a surrogate for crime concerns should not be a basis for formulat- could be ascertained by calculating the degree ing policy regarding these programs. of correlation between arrests (as reported by References police) and drug-related crime. However, this 1. Vlahov D, Junge B. The role of needle exchange method would also be subject to bias because Contributors programs in HIV prevention. Public Health Rep. it relies on counts of drug-related crime. M.A. Marx and B. Crape participated equally in the 1998;113(suppl 1):75–80. Trends in crime, as measured by arrests, writing of the article. M.A. Marx assisted with study 2. Des Jarlais DC, Marmor M, Paone D, et al. HIV are also likely to be affected by secular factors design, directed study progress and planning, served incidence among injecting drug users in New (e.g., demographics, community policing prac- as a liaison with the city health department, and wrote York City syringe-exchange programmes. Lancet. the manuscript and revisions. B. Crape assisted in 1996;348:987–991. tices). These factors were not taken into ac- study planning and design, served as a liaison with 3. Hagan H, Des Jarlais DC, Friedman SR, Pur- count here, which is also a limitation. In addi- the city police department, compiled and analyzed the chase D, Alter MJ. Reduced risk of hepatitis B tion, some officers may have altered their data, and contributed to the writing of the manuscript. and hepatitis C among injection drug users in the

December 2000, Vol. 90, No. 12 American Journal of Public Health 1935 Tacoma syringe exchange program. Am J Pub- 10. Maginnis RL. Needle exchanges are bad medi- nan D. Evidence of self-report bias in assessing lic Health. 1995;85:1531–1537. cine. Paper presented at: Annual Meeting of the adherence to guidelines. Int J Qual Health Care. 4. Strathdee SA, van Ameijden EJ, Mesquita F, Congressional Youth Leadership Council; June 1999;11:187–192. Wodak A, Rana S, Vlahov D. Can HIV epidemics 1998; Washington, DC. 18. Embree BG, Whitehead PC. Validity and relia- among injection drug users be prevented? AIDS. 11. Fay C. Needle ‘exchanges’ often aren’t. Wash- bility of self-reported drinking behavior: dealing 1998;12(suppl A):S71–S79. ington Post. April 26, 1999:A18. with the problem of response bias. J Stud Alcohol. 5. Heimer R. Can syringe exchange serve as a con- 12. 1990 census: 1997 update. Available at: http:// 1993;54:334–344. duit to substance abuse treatment? J Subst Abuse www.census.gov. Accessed February 20, 1999. 19. Case P. First Year of the Pilot Needle Exchange Treat. 1998;15:183–191. 13. National Admissions to Substance Abuse Treat- Program in Massachusetts. Boston, Mass: Mass- 6. Normand J, Vlahov D, Moses LE. Preventing HIV ment Services, the Treatment Episode Data Set achusetts Dept of Public Health; 1995. Transmission: The Role of Sterile Needles and (TEDS). Washington, DC: US Dept of Health Bleach. Washington, DC: National Academy and Human Services; 1996. 20. O’Keefe EK, Khoshnood K. Preliminary Re- Press; 1995. 14. Vlahov D, Junge B, Brookmeyer R, et al. Re- port on the New Haven Needle Exchange Pro- 7. Doherty MC, Garfein RS, Vlahov D, et al. Dis- ductions in high-risk drug use behaviors among gram. New Haven, Conn: City of New Haven; carded needles do not increase soon after the participants in the Baltimore needle exchange 1991. opening of a needle exchange program. Am J Epi- program. J Acquir Immune Defic Syndr Hum 21. Lurie PR, Bowser B. The Public Health Impact of demiol. 1997;145:730–737. Retrovirol. 1997;16:400–406. Needle Exchange Programs in the United States 8. Drug Use Forecasting 1993 Annual Report on 15. Aitkin M, Anderson D, Francis B, Hinde J. Sta- and Abroad. San Francisco, Calif: University of Adult Arrestees: Drugs and Crime in America’s tistical Modeling in GLM. Oxford, England: Ox- California, San Francisco; 1993. Cities. Washington, DC: US Dept of Justice; ford Science Publications; 1989. 22. Brooner R, Kidorf M, King V, Beilenson P, Svikis 1993. 16. Inciardi J. Heroin use and street crime. Crime D, Vlahov D. Drug abuse treatment success 9. Leukefeld CG, Gallego MA, Farabee D. Drugs, Delinquency. 1989;25:335–346. among needle exchange participants. Public crime, and HIV.Subst Use Misuse. 1997;32:749–756. 17. Adams AS, Soumerai SB, Lomas J, Ross-Deg- Health Rep. 1998;113(suppl 1):129–139.

Giving Means Receiving: The Protective Effect of Social Capital on Binge Drinking on College Campuses

ABSTRACT Elissa R. Weitzman, ScD, MSc, and Ichiro Kawachi, MD, PhD

Objectives. We tested whether Binge drinking among adolescents and neighborhood collective efficacy (which sub- higher levels of social capital on college young adults in college is a prevalent problem sumes concepts of social capital).17,18 campuses protected against individual affecting upward of two fifths of the college In this study, we sought to examine risks of binge drinking. student population.1 Public and private agen- campus-level patterns of participation in Methods. We used a nationally rep- cies are now supporting efforts to reduce it voluntary activities (an indicator of social resentative survey of 17592 young peo- and related harms.2 Newer efforts include capital) in relation to binge drinking in ple enrolled at 140 4-year colleges. So- social-ecologic interventions to change indi- college. Campuses with high levels of so- cial capital was operationalized as vidual and environmental factors,3,4 reflect- cial capital may provide the patterns of individuals’ average time committed to ing theories that individual and community interconnectedness and mutual obligation volunteering in the past month aggre- characteristics shape youth alcohol abuse.5–7 required for collective regulation of de- gated to the campus level. One such community factor may be social viancy in a group. Although social capital Results. In multivariate analyses capital. may have little effect on (or even encour- controlling for individual volunteering, Social capital is a contextual character- age) light drinking, it may protect against sociodemographics, and several college istic describing patterns of civic engagement, binge and problem drinking. characteristics, individuals from cam- trust, and mutual obligation among persons.8 puses with higher-than-average levels of Recent attention to it has been spurred by social capital had a 26% lower individ- the work of Coleman in sociology,9,10 Put- ual risk for binge drinking (P<.001) than nam in political science,11–13 and Kawachi 14 Elissa R. Weitzman and Ichiro Kawachi are with the their peers at other schools. and colleagues in public health. The latter, Department of Health and Social Behavior, Center Conclusions. Social capital may using aggregate rates of participation in vol- for Society and Health, Harvard School of Public play an important role in preventing unteer associations and survey measures of Health, Boston, Mass. Ichiro Kawachi is also with binge drinking in the college setting. (Am social trust and reciprocity as measures of the Harvard Medical School, Boston. J Public Health. 2000;90:1936–1939) social capital, found that state-level social Requests for reprints should be sent to Elissa capital varied with all-cause mortality,14 vi- R. Weitzman, ScD, MSc, Department of Health and 15 16 Social Behavior, Harvard School of Public Health, olent crime, and self-rated health. Oth- 1633 Tremont St, Boston, MA 02120 (e-mail: ers have found that juvenile delinquency and [email protected]). violent crime varied with differences in This brief was accepted April 20, 2000.

1936 American Journal of Public Health December 2000, Vol. 90, No. 12