Risk Factors for Youth Suicide

Home , Disembowelment, Double Suicide, Suicide by hanging, Suicide epidemic, Suicide in India, The Suicide (Seinfeld)

:t "-° DOCUXENT =ME

ED 334 502 CG 023 517

AUTHOR Davidson, Lucy, Ed.; Linnoila, Markku, Ed. TITLE Report of the Secretary's Task Force on Youth Suicide. VolUMO 2: Risk Factors for Youth Suicide. INSTITUTION Alcohol, Drug Abuse, and Mental Health Administration (DHMS/PMS), Rockville, MD. REPORT NO DHHCADM)-89-1622 PUB DATE Jan 89 NOTE 311p.; For other volumes in this series, see CG 023 516-519. PUB TYPE Collected Works - General (020) -- RepJrts - General (140)

MRS PRICE MF01/PC13 Plus Postage. DESCRIPTORS Adolescents; *Risk; *Suicide; Young Adults; *Youth Problems

ABSTRACT Commissioned papers by a work group on risk factors for youth suicide, which examined environmental, behavioral, socio-cultural, biological, and psychological factors associated with an increased likelihood of aicide among young people are presented in this document. The following papers are presented: (1) "Sociodemographic, Epidemiologic, and Individual Attributes" (Paul C. Holinger and Daniel Offer); (2) "Preparatory and Prior Suicidal Behavioral Factors" (Norman L. Farberow); (3) "Social and Cultural Risk Factors for Youth Suicide" (Carol Huffine); (4) "Family Characteristics and Support Systems as Risk Factors for Youth Suicide" (Cynthia R. Pfeffer); (5) "Contagion as a Risk Factor for Youth Suicide" (Lucy Davidson and Madelyn Gould); (6) "Stress and Life Events" (Eugene S. Paykel); (7) "Sexual Identity Issues" (Joseph HarrY); (8) "'Major Psychiatric Disorders' as Risk Factors in Youth Suicide" (Maria Kovacs and Joachim Puig-Antich); (9) "Pernonality as a Predictor of Youthful Suicide" (Allen Frances and Susan J. Blumenthal); (10) "Substance Use and Abuse: A Risk Factor in Youth Suicide" (Marc A. Schuckit and Judith J. Schuckit); (11) "Methods as a Risk Factor in Youth Suicide" (J. William Worden); (12) "Neurotransmitter Monoamine Metabolites in the Cerebrospinal F3lid ae Risk Factors for Suicidal Behavior" (Marie Asberg); (13) "Post Mortom Studies of Suicide" (Michael Stanley); (14) "The Neuroendocrine System and Suicide" (Herbert Y. Meltzer and Martin T. Lowy); (15) "Genetics and Suicidal Behavior" (Alec Roy); esid (16) "Summary and Overview of Risk Factors in Suicide" (FrederIck K. Goodwin and Gerald L. Brown).

*********************************************************************u* * Reproductions supplied by EDRS are the best that can be made * * from the original document. * *********************************************************************** Report of the Secretary'sTask Forceon grt CLA-11 1

LAPPF

BEST COPY AVAILABLE

U.S. DEPARTMENT OP EDUCATION Orrice or Educations; Rearamn and trnprouement EDUCATIONAL RESOURCES INFORMATION CENTER tERICI \Xrnis document nas been reproduced as ieceiuzio from the Person or organization origiruoting rt 0 Min Or Menges have been merle to improve reDroduchon dummy

Points ot roe* or opinions stated in thiscIrt meat do not necessarily rePtfent Official OEM Position or policy

U.S. DEPARTMENT OF HEALTH & HUMANSERVICES Public Health Service 2 Alcohol, Drug Abuse, and Mental Health Administration Report of the Secretary's Task Forceon - ...,...11 A S.....i...... 460

Edited by: Lucy Davidson, M.D. Markku Linnoila, M.D.

January 1989

MMIIMMEIMMIL. U.S. DEPARTMENT OF HEALTH & HUMAN SERVICES Public Health Service Alcohol, Drug Abuse, and Mental Health Administration The members of the Secretary's Task Force on Youth Suicide wish to acknowledge the extraordinary effort of the Executive Secretary, Ms. Eugenia P. Broumas.

Suggested citation:

Alcohol, Drug Abuse, and Mental Health Administration. Report of the Secretary's Task Force on Youth Suicide. Volume 2: Risk Factors for Youth Suicide. DIMS Pub. No. (ADM)89-1622. Washington, D.C.: Supt. of Doer..., U.S. Govt. Print. Off., 1989.

2-ii CONTEMS: Volume 2

Members of the Secretary's Task Force on Youth Suicide Members of the Work Gnxip on Risk Factors for Youth Suicide 3 Oveiview of the Work Group txt Risk Factors for Youth Suicide Summary of the National Conference on Risk Factors for Youth Suicide (David J.Kupfer, M.D.) 9 Commissioned Papers: I. Paul C. Holinger, Daniel Offer: Soclodemogmphic, Epidemiologic, and individual Attributes 19 2. Norman L Farberow: Preparatory and Prior Suicidal Behavior Factors 34 3. Carol Malik*: Social and Cultural Risk Factors for Youth Suicide 56 4. 4nthia R. Pfeffer: Family Characteristics and Support Systems as Risk Factors for Youth Suicide 71 5, Lucy Davidson, Madelyn Gould: Contagion as a Risk Factor for Youth Suicide 88

6.Eugone S. Paykel: Stress and Ufa Events 110 7. Joseph Hany: Sexual identity issues 131 & Maris Kovacs, Joachim Pulg-Antich: *Major Psychiatric Disorders* as Risk Factors in Youth Suicide 103 9. Allen Frances, Susan J. Blumenthal: Personality as a Predictor of Youthful Suicide 160 10. Marc A Schuckl4 Judith J. Schuckit: Substance Use and Abuse: A Risk Factor in Youth Suickle 172 11. J. Willlam Worden: Methods as a Risk Factor in Youth Suicide 184 12. Marie Asberg: Neurotransmitter Monoamine Metabolites in the Cerebrospinal Fluid as Risk Factors for Suicidal Behavior 193 13. Michael Stanley: Post Mortem Studies of Suicide 213 14. Herbert V. Meltzer, Martin T. Lowr The Neuroendocrine System and Suicide 235 15. Alec Roy: Genetics and Suicidal Behavior 247 16. Frederick K Goodwin, Gerald L. Brom: Summary and Overview of Risk Factors in Suicide 263 MEMBERS CW THE SECRETARY'S TASKFORCE ON YOUM SUICIDE

Shervert Frazier, M.D. Chairman Formerly, Director National Institute of Mental Health

Carolyn Doppelt Gray Dodie T. Livinvon Acting Deputy Assistant Secretary Commissioner, Administration for Chileren, Office of Hunuin Development Services Youth, and Families Office of Human Development Services M. Gene Handelsman (served until June 1986) Robert G. Niven, Formerly, Deputy Assistant Secretary (served until January 1986) Office of Human Development Services Formerly, Director, National Institute on Alcohol Abuse and Alcoholism Robert B. Helms, Ph.D. Acting Assistant Secretary for Planning Everett R. Rhoades, M.D. and Evaluation Director, Indian Health Service Office of the Secretary Health Resources aud Services Administration Jerome H. Jaffe, M.D. (served until June 1986) Mark L Rosenberg, M.D., M.P.P. Director, Addiction Research Center Assistant Director for Science National Institute on Drug Abuse Division of Injury Epidemiology and Control Center for Environmental Health and Stephanic Lee-Miller Injury Control Assistant Secretary for Public Affairs Centers for Disease Control Office of the Secretary Charles R. Schuster, Ph.D. Markku Linnoila, M.D. Director Clinical Director, Division of Intramural National Institute on Drug Abuse Clinical and Biological Research N.-tional Institute on Alcohol Abuse Robert L Trachtenberg and Alcoholism Deputy Administrator Alcohol, Drug Abuse, and Mental Health Administration

2-1 STAFF

Eugenia P. Broumas, Coordinator, Task Force Activities Special Assistant to the Deputy Director National Institute of Mental Health

Heather A Pack, Policy Coordinator Executive Secretariat Office of the Secretary

ALTERNATES

Dorynne Czechowicz, M.D. Assistant Director for Medical and Professional Affairs Office of Science National Institute on Drug Abuse

Jack Dure 11, MD. (served until July 1986) Formerly, Associate Director for Science National Institute on Drug Abuse

Paget Wilson Hinch Associate Commissioaer, Family and Youth Services Bureau Administration for Children, Youth, and Families Office of Human Development Services

Chuck Kline Deputy Assistant Secretary for Public Affairs- News Office of the Secretary

Arnold R. Tompkins Deputy Assistant Secretary for Social Services Policy Office of the Assistant Secretary for Planning and Eva !ration Office of the Secretary

2-2 WORK GRCUP ON RISK FACTORS FOR YOUTH SUICIDE

Lucy Davidson, M.D., Cochairperson Medical Epidemiologist, Division of Injury Epidemiology and Control Center for Environmental Health and Injury Control Centers for Disease Control

Markku Linnoila, M.D., Cochairperson Clinical Director, Division of Intramural Clinical and BiologicalResearch National Institute on Alcohol Abuse and Alcoholism

Susan J. Blumenthal, M.D. Chief, Behavioral Medicine Program Health and Behavior Research Branch Division of Basic Sciences National Institute of Mental Health

Dan Lcttieri, Ph.D. Psychologist, Division of Extramural Research National Institute on Alcohol Abuse and Alcoholism

Alec Roy, M.B. Visiting Associate National Institute on Alcohol Abuse and Alcoholism

Review PanelNational Conference on Risk Factorsfor Youth Suicide

David J. Kupfer, M.D., Chainntui Professor and Chairman Department of Psychiatry School of Medicine University of Pittsburgh Western Psychiat:ic Institute and C"..lic Pittsburgh, Pennsylvania

Gerald L Klerman, M.D. Professor of Psychiatry Cornell University Medical Center New York Hospital Payne Whitney Psychiatric Clinic New York, New York

2-3 George Murphy, M.D. Director, Psychiatric Outpatient Services Department of Psychiatry Washington University School of Medicine St. Louis, Missouri

Herbert Pardes, M.D. Chairman, Department of Psychiatry College of Physicians and Surgeons of Columbia University New York, New York

David B Pi Hemet, Ed.D. Associate Professor of Psychology Department of Psychology We Ile* College We Italy,,Massachusetts

Judith L Rapoport, M.D. Chief, Child Psychiatry Branch Division of Intramural Research National Institute of Mental Health Bethesda, Maryland

Lee Robins, Ph.D. Professor of Sociology in Psychiatry Department of Psychiatry Washington University School of Medicine St. Louis, Missouri

Ethvin Shneidman, Ph.D. Professor of Thanatolov University of Califoraia, Los Angel% Neuropsychiatric Institute and Hospital Los Angelis, California

2-4 OVERVIEW OF ME WORK GROUPON RISK FACTORS FOR YOUTH SUICIDE

INTRODUCTION The first goal of the Task Force on Youth three sources: the commissioned papers, the Suicide as stated by the Secretary was *to take review panel's work, and the reflections of the lead in coordinating activities about the conference attendees. suicide among various Federal agencies, Although research reviewed by the authors Congress, State and local governments, varied in quality as well as methodology, suf- private agencies, and professional organiza- ficient data were available to establish many tion.* The three work groups of the task characteristics as risk factors for youth forcerisk factors, interventions and preven- suicide. Those biochemical, psychological, tion, and strategies for the futurehave and social factors most clearly linked to youth worked toward establishing a model for the suicide were the following: kind of coordination and sequential progress envisioned by the Secretary. The research Substance abuse, both chronic and acute, co elusions and recommendationsreached in the contod of the suicidal act. Sub- by the Risk Factors Work Group build on this stance abuse was also tied to the exacer- foundation. ba tion of concurrent psychiatric disorders, themselves indicators of in- Another major charge to the task force was creased risk. to *assess and consolidate currentinforma- tion.* The work group generated a com- Specific psychiatric diagnostic groups prehensive list of potential risk factors, affective disorders, schizophrenia, and grouped them into specific risk factor borderline personality disorders. domains, and identified experts in each area Parental loss and family disruption. to review the scientific literature and write Familial characteristics including genetic the com- summary papers. In their papers, traits such as predisposition to affective missioned authors were asked to catalog, illness and the effects of role modeling. analyze, and synthesize the literature on factors linked to youth suicide. These papers Low concentrations of the serotonin me- clarified the environmental, behavioral, tabolite, 5-hydroxyindoleacetic acid (5- socio-cultural, biological, and psychological HIAA), and the dopamine metabolite factors which have been associated with an homovanillic acid (IIVA) in the increased hicelihood of suicide amoag young cerebrospinal fluid. people. The papers were presented at the Other risk factors include homosexuality, National Conference on Risk Factors for being a friend of family member of a Youth Suicide in Bethesda, Maryland, May 8 suicide victim, rapid socio-culturA and 9, 19E16. They were critiqued by a review c...ange, a history of previoussuicidal be- panel and opened for discussion and com- havior, impulsiveness and aggressive- ment by those attending the conference. ness, media emphasis onsuicide, and The following comments were distilled from

7 1) 2-5 Report of the Secretarys Task Force on Youth Suicide

ready access to lethal methods, such as specifying the theoretical models upon guns. which a study is based, (b) stating opera- The diversity of risk factors points to the tional definitions for all variables, (c) need for targeting intervention and preven- designing a research protocol that will allow individual exposures to be assessed, tion strategies. Our ability to address specific (d) fostering interdisciplinary collabora- populations at high risk for youth suicide will help focus rescue,: and evaluation com- tion, (e) using comparison groups, and (f) using nonclinical populations. ponents of planned interventions as well. Much of our biowledge of risk factors for While clear trends were evident, the avail- youth suicide has been extrapolated from able research made quantifiable estimates of studies of adults or studies of suicide at- relative risk a goal as yet unreached. Many tempters.Methodologic deficiencies of studies, while meticulously descriptive, lack- many studies ftirther limit our ability to ed comparison groups. Other lines of re- quantify that risk.Stronger studies of search had not been conducted for youth and youth suicide woukl provide the risk factor results were ectrapolated from adult popula- data upon which suicide intervention and tions. prevention efforts could be more reliably Recommendations for future research ap- based. proaches were derived from the authors' as- 2. Encourage collaborative behavioral and sessments of the studies that had been done biochemical research into risk Eldora as- in each risk factor domain. The types of re- sociated with youth suicide. search envisioned by the work group would parallel other efforts at suicide prevention Recently, behavioral and biological inves- and promote a more precise identification of tigators have advanced ou: knowledge of those young people likely to benefit from a risk factors associated with youth suicide. particular hitervention and of the cir- We expect that their combined efforts cumstances under which directed interven- would be synergistic. Their collo:Aration tions are most imperative. could facilitate the translation of laboratory data into clinically useful infor- CONCLUDING NOTE mation. 3. Conduct long term, prospective studies of The Work Group on Risk Factors for Youth completed suicides among youth. Such Suicide has examined attributes and ex- studies should bc multi-center efforts with posura; that are associated with an increased compatible data collection instruments. likelihood of youth suicide. The careful identification and exploration of these risk Statistically, youth suicides are rare events. factors forms the foundation for effective in- Collecting data at multiple sites could tervention and prevention planning. Given provide a larger sample size with power to the nature of the work group's task and the detect more subtle risk factors. Clarifying limitations of the data available for review, the developmental course of various risk most of our recommendations focus on the factors through prospective studies would research process itself. However, where data facilitate the timing of treatment options. were coinpelling, we have made the follow- 4. Establish surveillance systems for ing recommendations in specific risk faetor suicide attempts and suicide clusters. domains: The incidence, prevalence and charac- 1. Extend and verify knowlWge of youth teristics of suicide attempts among youth suicide by conducting well designed re- are unknown. Surveillance would provide search. Elements of sound design which population-based data to undersie.nd address current deficiencies include: (a) suicide attempts as a public health

2-6 1 1 Overview of the Work Group on Risk Factors

problem and to ffiuminate the relationship 8. Collaborate with the broadcast media in between suicide attempts and suicides. content analyses to identify the harmild and beneficial features of fictional Sificide clusters may represent an especial- suicide programs and news coverage of ly preventable type vZ youth suicide. On- suicide. going surveillance of clusters would allow for earlier detection and would address Some, but not all, fictional and nonfiction- the following issuac the proportion of al suicide programs have bivn associated suicides occurring in clusters, their relative with an increase in suicide attempts and frequency among certain age groups, suicides. Identifying which components of geographic differences, and changes in the the programming were associated with an pattern of suicide clusten over time. increase would allow for more responsible broadcasting. 5. Assist death certifiers in implementing uniform operational criteria for the 9. Increase gatekeepers' recognition of and determination of suicide and assist ability to refits potentially suicidal youth States in expediting their modality data by providing (a) infbrmation on acute recording and reporting. and chronic risk hectors for suicide, (b) heormation on behavioral manifesta- Nationally, death certificates comprise the tions of depression, schizophrenia, and primary data source for the epidemiologic conduct disoniers, (c) information on in- analysis of suicide. However, suicide is a dications and sources for reltrring youth mode of death particularly subject to at risk, and (d) training in communica- misclassification. Without explicit criteria tion skills fbr approaching and engaging for the determination of suicide, coroners youth at risk. and medical examiners do not reliably code suicide; medical examiners have es- Interested others, such as teachers, youth timated that half of all suicides may be in- activity leaders, clergy, and peers have accurately classified. Delays and errors in more opportunity for contactwith potcm- the vital registry system further com- tially suicidal youth than clinicians. They promise this data base. constitute a prime source for detection and referral of troubled youth. Deve/op valid and reliable mental health assessment instruments for youth. 10. Use risk hictor information to target intervention and prevention services. The personality variables and diagnostic groups associuted withadult suicide also Limits on the distribution of resources for appear to be associated withyouth suicide. suicide prevention compel us to direct our However, the lack of appropriate assess- efforts to those persons in greatest need ment instruments for youth is an obstacle and those most likely to benefit. Risk fac- to early identification and early treatment. tor information fosters prudentallocation of resources among those programs in- 7. Conduct biochemical research to identify tended to prevent youth suicide. neuroendocrine markers for suicidal behavior which can be measured repeatedly 11. Avoid sensationalized or roiLanticized oven time, inexpensively, and atlow risk reporting in media coverage of suicides as and burden to tbe subject. well as attention to violence and according celebrity status to the decedents. Biochemical markers may be able to identify persons at high lifetime risk for suicide Nonfictional media coverage of suicide during periods of low immediat: risk. has been associated with an increase in the Non-crisis-oriented interventions could number of suicides.Susceptible in- be implemented for these potential dividuals may be affected by positively suicides. regarded qualities of the suicide reported

2-7 '1 Report of the Secretary's Task Force on Youth Suicide

and by perceived rewards for the behavior. 12. Incorporate suicide prevention stiartegies into treatment aad outreach programs directed toward modifiable risk beton lbr youth suicide, such as alcohol and drug abuse. These youth problems are intrinsically worth treating. Adding suicide awareness to extant programs for troubled youth can increase our capacity to reach potentially suicidal youth without initiating costly new progams. 13. Provide increased psychiatric aid other youth service; commensurate with pre- dictable increases in the youth population. Population increases among youth, which have been associated with increasing youth services, rather than catching up after a deficit, may offset adverse effects of the population increase. 14. Improve health professionals' treatment of depression and ether psychiatric disorders associated wits youth suicide. Depression,conductdisorders, schizophrenia, and substance abuse are strongly associated with youth suicAe. The risk of suicide is increased when these disorders occur singly or concurrently. Many health professionals are rot trained in the effective treatment and management of these common disorders.

1 3

2-8 SUMMARY OF THE NATIONALCONFERENCE ON RISK FACTORS FOR YOUTHSUICIDE

David 1 Kupfer, M.D.

INTRODUCTION In the past fiftezn years, suicidal behavior dividuals, the Secretary's Task Force on among young people hasbecome an increas- Youth Suicide of the Department of Health ingly important public health problem re- and Human Services convened three nation- q uiring thedevelopment and al conferences to develop a strategar to deal implementation of detection and informa- with this pressing public health problem. tion strategies at the national level. In 1985, The first of these confer mces on RiskFac- suicide was the second leading causeof tors for Youth Suicide was heldin May 1986, death in young people and the rate of suicide in Bethesda, Matyland. At thisconference, among our nation's youthhad tripled over a number of distinguishednational and inter- the past thirty years. For example,between national speakers reviewed a variety of risk 1970 and 1980, 49,4% of the nation's youth fsctors for youth suicide. After the day and fifteen to twenty-four years of age committed a half of presentations, a groupof panelists suicide. Within this one decade, thesvicide responded to individual papers and to a num- rate for this age groupincreased 40 percent ber of general themes. These panelists were (from &8 deaths per 100,000 populationin David J. Kupfer, M.D. (chairman); Gerald L 1970 to 12.3 per 100,000 in 1980); while the Klerman, M.D.; George E Murphy, M.D.; rate for the remainder ofthe population Herbert Pardes, M.D.; David B.Pillemer, remained stable.Young adults twenty to Ed.D.; Judith L. Rapoport, M.D.; Lee twenty-four years of age had appromately Robins, Ph.D.; and Edwin Slineidman, Ph.D. twice the number and rate of suicides as They paid specific attention to the major teenagers fifteen to nineteen years old.This themes of the conference as well as sugges- increase in youth suicide is due primarily to tions for further strategies which might begin an increasing rate ofsuicide among young to connect some of the riskfactor areas with men. Rates for malesincreased by 50 per- detection and intervention approaches. cent (from 13.5 to 20.2 per100,000) Wm- pared to a 2 percent increase in females CONFERENCE SUMMARY 1970 (from 4.2 to 4.3 per 100,000) between all the and 1980, so that by 1980 the ratio ofsuicides The full text of the manuscripts by committed by males to those committed by speakers is included in this volume and I will females in this age group was almost five to only highlight particular points that the authors made which were either notdis- one. cussed completely or could benefitfrom fur- Based on these startling statistics and the ther emphasis. growing Rwareness of youth suicide by public health specialists and other concernedin- In their paper on sociodemographic,

2-9 4 Report of the Secretary's Task Force on Youth Suicide epidemiologic, and individual attrib, fes, suicide rate among their members. She dis- Holinger and Offer discussed the need for in- cussed social structures which might be sup- creased accuracy in epidemiologic data. portive and protective of individuals or might Both mortality data, local and national, and produce stress. Secondly, the culture as well also the population denominators from as social structure, influences the psychologi- which mortality rata are derived should be cal development of its members.Fmally, more reliable. They also argued that further through such mechanism as folklore and at- long term, cross-cultural comparisons of titudes about suicide and death, the culture suicide, homicide, and accident rates are may play a role in influencing suicide risk l'ac- necessary with a focus on period, cohort, and tors and the incidence of suicide itself. The age effects. manuscript focused nn social integration Gr In an extensive chapter on preparatory and the lack of it among various cultural group:, prior suicidal behavior, Farberow concluded including Hispanic Americans, American In- dians, Blacks and Asians. Dr. Huffine also that prior suicidal behavior of any kind is un- questionably as strong a risk indicator for empharszed that cultural attitudes and adolescents as it is for adults. On the other socialization may contribute to increased suicide rates in populatior: by romanticizing hand, indirect self-destructive behavior appears to play too xomplex a role in the per- micide, presenting death as a poAtive state, or influencing the psychological develop- sonality of the individual to serve as a reliable risk indicator. Suicide attempters and corn- ment of individuals in such a way that they are pietas among adolescents need to be ap- susceptible to suicide. propriately viewed as separate but In her chapter on family characteristics and overlapping populations.While pri,,r support systems, Pfeffer concluded that fan"- suicide attempts, threats, and suicidal idea- ly factor; associated with a high risk of youth tion may be exceilent clues for further suicide are related to experiences charac- suicidal behavior, it is also true that they are terized by the presence of intense levels of late clues in the progression towards suicide. stress. Such stress appears to be chronic and Certain behaviors have been identified that seems to occur at an early phase of the appear consistently in the histories of adoles- individual's life. One implication of this posi- cents who ether attempted or committed tion is that family risk factor research on suicide:school performance variables, in- youth suicidal behavior needs to adhere to a eludingcademic difficulties, disciplinary developmental perspective. Such an orien- problems, and truancy; and antisocial be- tation could facilitate the examination of havior, especially assauhiveness. Fmally, so- continuities and discontinuities in family cial isolation and impulsive behavior are also variable .. that may be precursors to youth often noted. Such factors might be of greater suicide. In identifying a number of tressors, value when they appear with specific Pfeffer pointed to the loss of social supports, evidence of active suicide potential. Other variability and parental functioning which factors that may play a role in suicide are run- also includes a violation of personal boun- ning away, suggestibility and imitation. daries. She concluded that perspective lon- Suicide notes may tell something about the gitudinal research design :. of h5gh-risk person's style and pattern of thinking. populations were indicated.Such studies Dichotomous thinking (cognitive thought may evaluate th.e long term outcome of processes) may serve as a useful clue in children who previously had suicidal tenden- reflecting increasing rigidity and loss of cies.These studies may also evaluate ability to seek alternatives. children who are the offspring of parents prone to abuse or affective disorders, or The manuscript by Huffine on social and cul- parents who have separated, divorced, or tural risk factors for youth suicide pointed to died.Finally, factors such as famPy disor- ways in which societies might influence the ganization, parental psychopathology and

2-10 1 5 Summary of the National Conference on Risk Factors family violence were vectots that could en- Since the interpretations of these associa- hance suicidal behavior. Efficient screeaing tions are not fully clear-cut, the implications techniques for identifying high-risk families for prevention are also not easy to sum- as well as screening devices for identifying marize. Most of the recent events may serve high-risk children and adolescents in such as shoals for high-risk periods when crisis in- situations are implied needs for youth suicide terventions might be attempted. prevention. Another approach is to plan in- In reviewing sexual identity ix les, Harry ar- terventions that focus on ameliorating fami- gued that suicide attempts among ly and disorganization parental homosexuals of both sexes are two to six psychopathology so that stress may decrease times more likely than in heterosexuals. This and stability may be enhanced. conclusion, however, is somew-,unclear In their chapter on contagion as a risk factor, due to a lack of control groups, especially Davidson and Gould argued convincingly for noncfinical control groups.Harry argues further investigation of suicide clusters. that it would be desirable to obtain both ex- After reviewing various epidemic suicide perimental and control groups for nonclini- reports in the literature and investing medial cal populations through population-based influence on suicide, they concluded that surveys. time-space clusters of suicide do occur and are not a new phenomena of the 1980's. In assessing major psychiatric disorders, Kovacs and Puig-Antich concluded that the Nonfictional media coverage of suicides ir "condition" of psychiatric patienthood in sociated with an increase in the observed adolescents and young adults is associated number of suicides over those expected, r-d with an alarmingly high mortality risk from susceptible individuals may be affected ny suicide. These psychiatric disorders include direct or indirect exposures to suicide. Ob- psychosis and manic-depressive illness. They viously, a number of qtmtions were raised, such as, what proportion of suicides may also suggested a need to focus on prepubertal children; since their suicidal intent may be occur in clusters and in what ways youth may profound, while their physical and cognitive be differentially exposed and susceptible to suicide contagion.Since youth suicide limitations may render them less lethal Al- though suicidal ideation occurs frequently clusters are of particular concern and may be among prepubertal children their rate of potentially more preventable, the sorts of prevention and intervention efforts for suicide remains quite low. This is piobably due to prepubertal child's lack of cognitive averting cluster suicides are extremely im- maturity and skills necessary to complete portant. Therefore, recommendations con- suicide. Therefore, this age group provides cerning m -dia coverage, further research an opportunity to study suicidality untrun- investigations, and the establishment of a cated by suicide completion. Suicidality of surveillance system for potential suicide very early onset may enable attempters to be clusters seemed valuable. selected for future studies who are at most In his review of stress and life events, Paykel risk and most closely approximate completed concluded that there still is a dearth of studies suicides.These children, therefore, are a of recent life events in the role of youth very interesting model for intervention. suicidal behavior. Relatively few studies ofe Finally, Kovacs and Puig-Antich concluded early parental loss or studies examining com- that more efficacious treatment and care of pleted suicide have been published. Studies psychiatrically Hi youths may be the most of older suicides and suicide attempters ex- feasible way to alter their risk of suicide. amining early loss due to breakup of paren- In their assessment of the role of personality tal marriage consistently show suicide and disorders and characteristics, Frances and attempt rates higher than in normal control Blumee.al strongly suggested that even groups or psychiatrically disordered controls. though conduct disorders and borderline

2-11 16 Report of the Secretwy's Task Force on Youth Suicide personality disorders are highly associated definitions. They also lead to the notion that with adolescent suicide, assessment of per- various prevention studies for conduct disor- sonality factors ha_ been impeded by lack of ders should be attempted more vigorously. standardized measures for these characteris- In essence, it appears that the same disorders tics in young people. In addition, assessment are predicting suicidal behavior in adults and of underlying personality at the time of a in children. suicide attempt is confounded by the distress In reviewing methods, Worden argued that experienced by the individual around the although the numbers may be few, suicide time of the event. However, from per- statistics should be diligently collected for sonality studies that have been done, a con- children under the age of 10. Methods of tinuum of the traits and disorders assmiated suicide should also be studied in context. with suicidal behavior in adok seer .e ap- Therefore, those who are collecting data on pears to be associated with such "i%.navior in suicide methods should use one of the exist- adulthood.Therefore, stability in per- ing scaks that account for the context of the sonality characteristics, such as impulsive- event. If that were to take place, we would ness and aggressiveness, appears to be have a better grasp of the relative lethality of present over the life cycle. It is proposed that various age, sex, an t. ethnic groups as well as certss!ri diagnostic categories from the DSM- important distinctions between methods III Infancy, Childhood, and Adolescence section correspond to, and may in some chosen by youth versus adults. ir dividuals eventually develop into particular In his examination of substance use and personality disorders in adulthcxxl. For ex- abuse, Schuckit suggested that controlled ample, schizoid disorder of childhood and substances and/or alcohol are frequently adolescence may become schizoid per- used as the means of attempting self-harm, sonality disorder; avoidant disorder of especially araong younger women. Alcohol childhood and adolescence may become is often taken as a prelude to the suicidal act. avoidant personality disorder, conduct disor- Adolescents, alcoholics, and drug abusers der may become antisocial personality disor- have an elevated risk for suicide attempts aad der; oppositional disorder may become completions. Children of alcoholics and of passive aggrosive personality disorder; and patients with deprosive or schizophrenic dis- identity disorder may become borderline per- orders mdy themselves be a elevated risk for sonality disorder. The presumption is that suicide attempts and completions. Almost all the childhood or adolescent condition is substances of abuse are likely to exacerbate diagnosed if the individual is under age the preadsting emotional or psychiatric dis- eighteen, and the adult personality diagnosis turbances. Efforts aimed at minimizing the is used after age eighteen whenever the per- risk for suicide should include educating sonality psychopathology has persisted at an young people and their families about the intensity sufficient to meet disorder criteria. need to refrain from all recreational In addition, these personality variables may psychotropic substances during times of also have biological correlates, i.e., serotonin mood swings or anger. Finally, the children deficiency related to increased impulsiveness of alcoholics may themselves have inherited and aggressiveness, and may interact with en- problems of impulsiveness or hyperactivity or vironmental factors. a propensity to misuse substances with subsequent mood swings, anger, and frustration The authors noted that the coexistence of in their own lives. depression and conduct disorder or borderline personality disorder may represent an In reviewing various biological factors, As- extremely risky combination of factors. The berg argued that a low output serotonin sys- implications of these findings suggest careful tem or perhaps even more likely a low clinical assessment and further development stability system" might render an individual of "kiddy' personality measures and better more vulnerable to self-destructive or impul-

2-12 1 7 Summary of the National Conference on Risk Factors sive action in time of erisis.Asberg also illness.Roy also pointed out that Kety's pointed out that althouba relatively little is recent data pointed to a genetic factor which known about the biocharkel linkages be- represents the inability to control impulsive tween the serotonergic system and suicidal behavior triggered by depression, stress, or behavior, the evidence that personality fea- other stimuli. Therefore, he concluded that tures (impulsiveness and aggressiveness) important goals were first the identification may represent intervening variables is highly of genetic factor, nod the examination of suggestive. At the present time, it might be genetic transmission of psychiatric disorders recommended that cerebrospinal fluid per se, and then the examination of an addi- ((SF) measures could be used as an aid to tive genetic factor relating to impulsive be- suicide risk prediction in highly specialized havior. clinical settings, but not on a larger scale. In summarizing the biological factors, Good- The next presentation reviewed receptor win discussed a cluster of biological findings studio; and the ways in which postmortem re- emphasizing again the difference between search might contribute to our under- persons with suicidal ideation versus those standing of the biological aspects of suicidal who are suicide attempters and completers. behavior. Stanley argued that postmortem He argued that the presence of - osychiatric research would necessitate interviewing next illness was the single most predictive factor of kin to obtain f T. needed personality for a serious suicide attempt. In pointing to descriptives and diagnostic information. He specific psychiatric illnesses, he felt that the also suggested that in an effort to maintain a key psychiatric diseases were affective disor- link between postmortem findings and the ders with an emphasis on hopelessness, al- clinical application of such findings, inves- coholism with an emphasis on loss of impulse tigators should obtain samples of postmor- control, and schizophrenia with an emphasis tem (SF wherever possible. on psychosis. In a third presentation on biological factors At the end of these presentations, the discus- with an emphasis on neuroendocrine aspects, sants each presented brief overview of their Meltzer suggested that the thyroid stimulat- thoughts as well as specific points on various ing hormone (TSH) response may be blunted chapters. in persons who commit violent suicide. Klerrnan stated that the conference con- Methodologically, he examined differences tained a number of omissions such as discus- between violent versus nonviolent attempts sion of some high-risk populations including and also persons who attempted suicide ver- the American Indians. He also pointed out sus those individuals with suicidal ideation. that relatively little of the existing He found a significant relationship between epidemiological methodology for assessing serum cortisol and hopelessness. His review risk has been applied to youth suicide. Tech- therefore, suggested that there may be niques such as attributable relative risk and relationships between cortisol and suicidal logistic regression, which are the standbys of behavior. epidemiology, have not been applied by and Finally, in reviewing genetic factors, Roy sug- large to most of these studies. Murphy rein- gested that the Copenhagen adoption forced the notion of expanding Shard's studies strongly suggest a genetic factor for recent study (Louisville, Kentucky) to other suicide independent of or added to genetic places. He also stressed Frances' argument transmission of psychiatric disorders. Inter- about comorbidity, such as, the combination estingly, support for this possibility comes of personality disorder and substance abuse. from the recent Amish studies which show Pardes discussed the need to find ways of that suicide was much more likely to occur fostering interactions among various dis- when an individual had genetic vul- ciplines working on this topic. Pillemer ar- nerabilities to both suicide and to affective gued that one way to try to firm up

18 2-13 Report of the Secretary's Task Force on Youth Suicide conclusions from correlational studies invol- ing several prospective studies with an inter- ves improving the database through more vention component might be the most careful matching of controls with suicide economical way to go in the long run. Which groups. Another strategy would involve ap- risk factors do we know enough about to plying survey methods to nonclinical popula- design such trials and which risk factors are tions. He also stated that intoxication with modifiable? These are crucial questions alcohol or drugs often precedes suicidal be- since part of the strategy of public health in- havior and that the suicide rate among sub- tervention is to identify risk factors which are stance abusers is much higher than in the modifiable. general population. It, therefore, might be In conclusion, consid,trable attention should possible to partially disentangle this risk fac- be paid to the differences between vul- tor from associated factors by conducting nerability factors and protective factors, the planned intervention studies. need for longitudinal prospective studies, Rapoport reemphasized the role of coinorbidity, and interdisciplinary research. psychiatric disorders, especially impulsive In understanding the application of risk fac- conduct disorders, in relPtion to suicidal be- tors we must plan for the education of havior. Robins suggested that we should try primary care practitioners and pay attention to study the whole spectrum of suicidal idea- to teachers in dealing with early detection tion through attempts to successful suicide and recognition of children in trouble. Other rather than keeping them as separate goals are planned interventions where the enterprises. Many factors are highly corre- disentanglement of risk factors is very ap- lated with suicide attempts and suicide com- propriate and the development of a child pletions but we do not yet know to what brain bank. Along these lines would be set- extent these are separate populations and to ting up criteria as well as the actual comple- what extent they are overlapping population of a national registry for suicide tions. The only way we can find out is by completer& With respect to some of the studying multiple variables simultaneously. methodology gaps, the notion of developing The intersection of two areas traditionally child personality measures would be impor- separred in psychiatric diagnosisthe inter- tant. nalizing disorders of depression and the ex- Davidson pointed out that we need to under- ternalizing disorders of conduct and drug stand much more about the circumstances of abusepredicts an explosive situation. This directed interventions, we need to study non- is an opportunity, therefore, to integrate clinical populations, and we need to replicate genetic and stress research. Another notion studies in children similar to what we have is to examine children who have had antiso- carried out with adults. An important point cial fathers and depressive mothers who to be emphasized in this report is the issue of would then be very likely to have a double weighting the various risk assessment factors genetic dose which might be the critical thing in the research reported. Shneidman and developing a bridge between risk factor stressed other points in the conference identification and intervention planning. (In a later presentation, Davidson pointed to five preceedings which have already been criteria for developing this bridge:the described. In particular, Schneidman argued prevalence of the risk factor in the popula- for "individual case autopsies.' tion, the strength of the risk factor (relative Kupfer suggested that several workshops be risk), bow the population at risk might be beid to examine the gaps in methodology. reached (identification), the acceptability of For example, there may be several areas the proposed intervention, and the effective- where we already have multiple assessment ness of the intemention (the percent of those instruments and we need to decide which is treated who will benefit). Shaffer has pre- the most appropriate. The notion of launch- viously pointed out that the implications for

2-14 1 9 Summaty of Pie National Conference on Risk Factors prevention rest very much on which modelpeople who suffer from an affective disorder for suicidal behavior is accepted, whether end their lives by suicide while the other 85 there is au overlap model or a continuous percent do not? Using this overlapping modeL mode' we may learn that the subgroup of affective disorder patients who commit suicide In assessing the various risk factor domains have a greater overlap of other risk domains that were covered in this conference, it be- such as increased hopelessness, impulsive- comes clear that the 14 or 15 risk factor ness, decreased social supports, a recent domains can be reduced in number. For ex- humiliating life experience, and/or an in- ample, these relating to CSFdeterminations, creased family history of affective disci, or neurochetnical receptors, and neuroen- suicidal behavior. docrine studies can be grouped together as biological risk factors. Also substance use From this example, it is clear that psychiatric and abuse can be included under the major diagnoses are key risk factors. Current re- psychiatric disease segment. Some of the search shous that affective disorders, con- first several risk factor domains can also go duct disorder, and substance abuse are the under a larger categoty concerning psyrhoso- psychiatric diagnoses most highly associated cial factom and social supports. In short, al- with suicide in young people. In the adult though these risk factor domains can be literature over 90 percent of persons who end teased out separately, one might argue that a their lives by suicide have an associated smaller set of risk factor domains might be psychiatric illness.The few studies on more useful for ascribing weights and at- adolescent suicide suggest high percentages tempting to deal with prediction and the es- as well. tablishment of various detection and Secondly, personality traits relating to intervention strategies. suicide, such as aggression, impulsiveness, Although one can provide different models, and hopelessness are intrinsically important it has been previously suggested that five in characterizing suicide since they may rep- domains organized as a matrix or multi-axial resent personality styles that cut across diag- set of domains may provide a simple model nostic groupings. In addition, this domain for looking at most of these risk factors. includes certain personality disorder), such Whether a model of risk should be a series of as borderline personality disorder and antiso- interlocking Venn diagrams or some other cial personality disorder, which me more additive model, it does appear that a major highly correlated with suicidal behavior and clinical research strategy will be the need to represent risk factors. 'the comorbidity (or develop weights for each of its major com- co-occurrence) of antisocial and depressive ponents.For example, in applying this symptoms appeats to be a particularly lethal model, the breakup of a relationship might combination in adults and young people. be a final huirlia'ing experience that triggers The third risk factor domain is concerned a depressive episode in a young person with with psychosocial factors, social supports, life a family history of affective disorder. Such ao events, and chronic medical illness. For ex- individual may also have poor social sup- ample, early loss, increased negative life ports, which interact with the other identified events, the presence of a chronic medical ill- risk factors to increase the individual's vul- ness, and decreased social supports increase nerability to suicide. the risk for suicide. The question is, at what level and in what de- In addition to these three risk factor domains, gree do each of these factors contribute to two others stand out. One is the identifica- suicide potential? Or is the degtee of over- tion of both genetic and family factors that lap of all factors the most significant predispose an individual to suicide. Previous criterion? Or we may wish to pose such ques- investigators have suggested that the tions as:What makes 15 percent of the genetics of suicide may be independent of the

2-15 2 0 Report of the Secretary's Task Force on Youth Suicide genetics in a family history relating to sped& component will involve the development of psychiatric disorders, such as affective disor- therapeutic strategies based on the known der or alcoholism. The final factor may be risk factors for suicide in this age group. The the neurochemical and biochemical variables research component will work toward the currently under active investigation in an at- goal of more accurate identification of tempt to identify either a biologic abnor- teenagers at risk of suicide. mality or a vulnerable state tor suicide.I would advocate that intervention strategies incorporate these factors into the research design (if only to track them) as an essential feature. This material may represent a somewhat personal viinv i.1" the proceedings of the risk meeting ard ma) need to be more 'objectively* reviev ted to achieve a consensus of the scope of tie problem, the risk factor characteristics, what kinds of detection and interventions can be carried out now, and what we can do to improve our detection and intervention strategics. On a broader note, it would appear that politically the time is ripe to advocate partnerships between various agencies and funding sources, e.g., Federal-State-local; foundation-Federal; university-community. Aside from the obvious problems of partnership, there is a tendency in the suicide area to promise too much too quickly. Perhaps if one does not only advocate research or education or intervention, one might have a better balanced shot at achieving the sus- tained longitudinal push one will need to redevelop this area of public health concern. Our own recent experience in the Common- wealth of Pennsylvania last winter was proposing a model center to capture all three components. We will establish at Westorn Psychiatric Institute and Clinic (WPIC), in conjunction with our ongoing program for adolescents and young adults, a Center for Teenagers at Risk to serve the wsstern Pen- nsylvania region. This center will have three major components: (1) outreach, education and prevention; (2) demonstration intervention and treatment programs; and (3) research on adolescent suicide. The goal of the mobile outreach component will be to teach targeted school and agency personnel to identify the signs and symptoms of depression and potential suicide. The intervention

21 2-16 COMMISSIONED PAPERS SOCIODEMOGRAPHIC, EPIDEMIOLOGIC, AND IN DM DUAL A1TRIBUTES

Paul C Ho linger, MD., M.P.H.,Associate livfessorof Psychiatry, Rush-Preskterian-St. Luke's Medical Center, Chicago, Illinois

Daniel Offer, M.D., Prvfessor and Chainnan, Department of Psychiady, Michael Reese Hospital and Medical Center, Chicago, Illinois

INTRODUCTION From an individual, clinical point of view, , it on an epidemiologic level The focus will be is difficult to overestimate the distressing, on completed suicides, and we will utilize a disastrous impact of the suicide of a young developmental model emphasizing early (10- person. The impact on parents, siblings, 14 years old), middle (15-19 years old), and friends, and the community seems almost in- late (20-24 years old) adolescence, with an expressible. However, the extent to which emphasis on the 15 to 19 and 20 to 24 year one views slf-destructiveness among the olds. The paper is divided into six sections; young as an issue in the public health and sections on literature, methodolov, data, epidemiologic realms seems dependent on discussion, and future research will follow the context and perspective. On the one this brief introduction. hand, suicide is the second leading cause of death among 15 to 24 year olds in the United LrIERATURE States (following only accidents) (National Center for Health Statistics. 1985) and, This section on literature will focus on the primarily because of the number of suicides, epidemiology and potential prediction of homicides, and accidents among young adolescent suicide. people, violent deaths* are the kading cause Of the many tasks of science, perhaps one of of number of years of life lost in this country the most important is that of prediction, (Holinger, 1980). On the other hand, young especially if such prediction can lead to effec- people have the lowest suicide rates of any tive intervention. Two types of studies over age group and are at least risk of dying by the past five years have begun to suggest that suicide (Holinger, Holinger, and Sandlow, prediction of certain violent deaths may be 1985). possible for some age groups. One type of The purpose of this paper is twofold. Fisst, study involved the use of a population model we will present and evaluate the (Holinger and Offer, 1983; Holinger and epidemiologic data related to suicide among Offer, 1984; Holinger and Offer, 1986; adolescents.Second, we will discuss the Holinger, Offer, Ostrov, et al., unpublished potential for the prediction of youth suicide data), and the second type of study utilizes whort analysis (Solomon and Henan, 1980; °Violent deaths refer to suicide, homicide, sad accidental deaths (Weiss, 1976). Hellon and Solomon, 1980., Murphy and

P3 2-19 Report of the Secretary's Task Force on Youth Suicide

Wetzel, 1980; Merman, Lavori, Rice, et .1., sive five-year interval it has a higher rate than unpublished data). the preceding cohort had at that age. Kler- man, et at. (unpublished data), noted a In our 1981 examination of the increase in similar cohort effect in their study of suicide rates among 15 to 19 year olds during depressed patients. the past two decades (Ho linger and Oft-, 1981), we reported that simultaneous with an There are both similarities and differences increase in suicide rates was a steady increase between the population-model and the in the population of 15 to 19 year olds from cohort-effect studies. The similarities lie in just over 11 million in 1956 to nearly 21 mil- the emphasis on recent increases in suicide lion in 1975. A subsequent study then related rates among the younger age goups. The the changes in the adolescent population and differences are Lie predictive aspects. The changes in the proportion of adolescents in cohort studies suggest that the suicide rates the total U.S. population to the adolescent for the age groups under study would con- suicide rates during the twentieth century in tinue to increase as they are followed over the United States (Holinger and Offer, time. Implicitly, the cohort studies also seem 1982). Significant positive correlations were to suggest that the suicide rates for younger found between adolescent suicide rates, age groups will continue to increase as each changes in the adolescent population, and new five-year adolescent age group comes changes in the proportion of adolescents in into being. The predictions by the popula- the population of the United States, i.e., as tion model are different. The population the numbers and proportion of adolescents model suggests that suicide rates for younger increased or decreased, the adolescent age groups will begin leveling off and suicide rates increased and decreased, decreasing, inasmuch as the population of respectively. It should be recalled that while younger people has started to decrease. In one might assume that the number of deaths addition, the population model suggests that from a particular cause will increase with in- as the current group of youngsters gets older, creases in the population, the mortality rates suicide rates will increase less than the cohort do not necessarily increase with an increase studies would predict. It is well known that in population because the denominator is male suicide rates increase with age in the constant (i.e., deaths/100,000 population). United States white female rates increase with age until about 65 and then decrease Cohort analyse; have also provided data to slightly (Kramer, et al., 1972; Holinger and demonstrate the increase in suicide rates Klemen, 1982). Therefore, one would ex- among the young (Solomon and Helton, pect an increase in suicide rates age consis- 1980; Helton and Solomon, 198% Murphy tent with this long-established pattern. and Wetzel, 1980; Klerman, et al., un- However, the current group of adolescents published data). Solomon and Helton and young adults make up an unusually large (198E), studying Alberta. Canada, during the proportion of the U.S. population. The years 1951 to 1977, identified five-year age population model suggests that the larger the cohorts, and followed the suicide rates as the proportion of adults in the population, the cohorts aged. Suicide rates increased direct- lower will be their suicide rates. Thus, the ly with age, regardless of gender. Once a population model would suggest that the cohort entered the 15 to 19 year old age suicide cates for the adult-mutations would range with a high rate of suicide, the rate for decrease over the next several decades com- that cohort remained consistently high as it pared with adult rates in the past, consistent aged. Min phy and Wetzel (1980) found the with the movement of the *baby boom" same phenomenon, in reduced magnitude, in population increase through those adult age larger birth cohorts in the United States. Not groups. This is not to say that the ratesfor only doe; each successive birth cohort start the older groups of the future might be ex- with a higher suicide rate, but at each succes- pected to have smaller suicide rues than the

24 2-20 P.Holinger Sociodemographict Epidemiologic...Alfributes older groups of the past. creases in suicide rates among the young, but these parallels were not consistent early in Other literature is also relevant to the imue the century (Vital Statistics of the United of violent deaths, population shifts, and States, 1979; Shapiro and Wynne, 1982). potential prediction. Positive relationships between population increases and upsurges in the rates of vRrious forms of violent death METHODOLOGIC ISSUES are described by Wechsler (1961), Gordon Methodologie Problems. Although we and Gordon (1960), and Klebba (1975). present elsewhere detailed discussions of the These findings were not supported by the methodologic problemsinusing work of Levy and Herzog (1974, 1978), Her- epidemiologic data to analyze violent deaths zog, et al. (1977), and Seiden (1984) in their (Holinger and Offer, 1984; Holinger, in reports of negative or insignificant correla- press, 1987), we will note the more important tions between both population density and issues here. crowding and suicide rates. Tho major types of methodologic problems The work of Easterlin (1980) and Brenner occur when using national mortality data to (1971, 1979), with extensive research of study violent death patterns: (1) under- and population and economic variables, respec- overreporting; and (2) data misclassification. tively, and other related studies (Seiden and Underreporting may result in reported Freitas, 1980; Peck and Litman, 1973; Kleb- suicide data being at least two or three times ba, 1975; Hendin, 1982) began to suggest the las than the real figures (Hendin, 1982; potential for prediction of suicide and other Seiden, 1969; Toolan, 1962, 1975; Kramer, et violent deaths. Turner, et al (1981), showed aL, 1972). The underreporting may be inten- that the birth rate increased with good tional or unintentional. In intentional un- economic conditions, and decreased with derreporting, the doctors, family, and friends poor conditions. Previous studies specifical- may contribute to covering up a suicide for ly examined the potential of a population various reasons: guilt, social stigma, poten- model to predict the patterns of violent tial loss of insurance or pension benefits, deaths (Holinger and Offer, 1984).This feats of malpractice, and so on. Unintention- model was also related to economic changes, al underreporting refers to deaths labeled with a suggested interaction of economic and "accidents," e.g., single car :ashes or some population variables (e.g., good economic poisonings, which were actually suicides but conditions leading to an increased birth rate were unverifiable as such because of the ab- with subsequent population changes) that sence of a note or other evidence. Studies of helped explain violent leath rates from an violent deaths among youth involve addition- epidemiologic perspective. al methodologic problems. There may be Other Factors. Summaries of other risk fac- greater social stigma and guilt surrounding tors for suicide among children and adoles- suicide in childhood and adolescence be- cents have been presented elsewhere cause of the intense involvement of the (Holinger and Offer, 1981; Seiden, 1969), parents at that age and the parents feeling but brief mention should be made here of that they have failed and will be labeled "bad three other variables: geographies, divorce parents." In addition, it may be much easier rate, and teenage pregnancy. With respect to cover up suicide in the younger age groups. to geographies, the western States have the Poisonings and other methods of suicide are highest suicide rates among adolescents, and more easily perceived as accidents in those the eastern States tend to have lower rates age g. :Hips than in older age groups. (Seiden, 1984; Vital Statistics of the United Two types of data classification problems States, 1979). The birth rates for teenagers exist. One involves classification at the na- and the divorce rate for all ages have in- tional level and the changes in this classifica- creased recently, paralleling the recent in-

2-21 Report of the Secretary's Task Force on Youth Suicide

tion over time. The cbangis in Federal clas- dude all U.& suicides among the age groups sification over time have been outlined in indicated. With the exception of Figure 1, various government reports (Dunn and data prior to 1932 are not utilized in the Shackley, 1944; Faust and Dolman, 1963a, present report as they are sample data and in- 1963b, 1965; Klebba and Dolman, 1975; Na- clude only death registration States and areas tional Center for Health Statistics, 1980; utilized by the Federal government during Vital Statistics- SpecialReports, 1941, any specific year. It v/86 only after 1933 that 1956). There has been little change over the all States were incorporated into the nation- century in Federal claasification for suicide. al mortality statistics (with Alaska added in The second type of data classification 1959 and Hawaii in 1960). problem concerns classification at the local Other Forms of Violent Death. Suicide, level, e.g., the legal issue involving the re- homicide (homicide mortality rates refer to quirement of some localities for a suicide those killed, not the killers), and accidents note as evidence of suicide; this practice both have been studied in aggregate (Weiss, 1976; decreases numbers and biases results be- Holinger and Klenten, 1982), and have been cause only the literate can be listed as having related in that all may represent some expres- committed suicide. sion of self-inflicted mortality (Wolfgang, Sources of Data. Sources of population and 1959; Menninger, 1938; Freud, 1901; Far- mortality data are noted in the respective berow, 1979). Homicide and accidents may tables and figures of this report. The data be self-inflicted in that some victims may used from 1933 to the present are for the provoke his or her own death by "being in the complete population, not samples: they in- wrong place at the wrong time" (Tauang,

Age Patterns of Violent Deaths by Type of Modality in the United States from 1900 to 1980.

e Suicide / go Homicide MVA NMVA

0- 0-1 14 5-14 15-24 25-34 35-44 45-54 55-64 65-74 7544 Age Groups (In Yeats) Sources of data: Same as for Table 1. Itor each type °Evident death, the data have been avenged over the ran 19)0-1980. MVA Motor-vehiele accident NMVA non-motomehiele accident Aft Figura 1.

2-22 PG P.Hollrtger: Soclodemagrephicr EpidemlologIc...Attributes

Boor, and Fleming, 1985; Wolfgang, 1959, however, the population of the adults (35 to 1968; Doege, 1978). Although suicide is the 64 years) in the United States increased most overt form of self-inflicted violence, steadily. These economic shifts could then homicide and accidents can be more subtle be seen to contribute t.3 the inverse correla- manifestations of self-destructive tendencies tions, with the population variable having a and risk-taking (Holinger and Klemen, coincidental, rather thtn etiologic, relation- 1982). However, this paper focuses primed- ship with thc violent death rates. The time lyon that most overt form of self-destructive- trends of violent deaths in the United Statzs ness: suicide. (e.g., the tendency of violent death rates to incrust in times of economic depression Methodologic Issues in Studying Adoles- such as the early 1930's ani decrease during cent Suicide sad Population Shafts. The war as in the early 1940's with World War II) general methodologic considerations were have been presented in detail elsewhere discussed above. The sources of population, (Holinger and Klemen, 1982). suicide, and homicide data are noted in the respective tables and figures of this section. With respect to population data, both the DATA figures and the correlations utilize the Epiderslokigic Data. Figure 1 presents age proportion of the population of a. given age patteri... of violent deaths by type of:mortality in the entire U.S. population (e.g., .he in the United States, averaged over the years proportion of 15 to 24 year olds in the entire 1900- 1980.Non-motor-vehicle accidents U.S. population). Homicide data as well as tend to have the highest rates, followed by suicide data will be noted, and the focus will motor-vehicle accidents, suicide, and be on 15 to 24 year olds. The coaelations homicide, respectively.orparticular impor- were derived as described rreviously tance to this paper are the age effects* seen (Hanger and Offer, 1984). in suicide. When male and female rates are In addition to the possibility discussed below combined, suicide rates can be seen to in- that there is a meaningful relationship be- crease steadily as age increases. Separating tween violent death rates and population male and female rates indicates that male shifts, one must also consider the possibility rates increase steadily as age increases, that either artifact or other variables are whereas female rates tend to increase to responsible for correlation. The possibility peaks during the 35 to 64 age range, with a that the correlations are artifact because of subsequent decrease.Perhaps the most change in Federal classifying of suicide and salient finding in Figure 1 is that children and homicide is until* as docribed above, the adolescents, despite the recent attention on comparability ratios for suicide and homicide the increases in their rates, have the lowest have been rather consistent over the suicide rates of any age group in the United decades. However, the possibility that Sta.(National Center for Health Statistics, another variable is involved, specifically 1984). That is, children and adolescents are period effects due to economic trends, needs at lower ri; ,f dying by suicide than any to be addressed. In the early 19r's (the other age fpap in this countiy. This finding starting point of these data, when the entire allows one to add= t1n question of adoles- US. population was included in the mortality cent suicide from a different and perhaps figutes), the mortality rates were at their more fruitful perspective: What factors Fab., probably because of the economic protect adolescents from suicide? And what depression. The violent death rates intrapsychic and external factors break down decreased for several yeats following, reach- to create a suicidal outcome in an adoles- ing low points during the early 1940's (World cent? War II). 'Age effects involve changes in specific rates of mortality or Mama over the life scam of the individual (Mulford, 1983). During the time of this decrease in rates,

f'7 2-23 Report of the Searetatys Task Force on Youth Suicide

Although the national mortality age group- during the 1940's and 1950's, increases from ings somewhat awkwardly separate various the mid-1950's through the 1970's, and the psychological deveitipmental stages, the age very recent tendency toward a leveling off of groups roughly correspond to la' 7, childhood this increase. and early adolescence (10 to 14 years), mid- Figure 3 presents suicide data for 20 to 24 dle adolescence (15 to 19 years), and late year olds. Rates for 20 to 24 year olds are adolescem and young adulthood (20 to 24 bigher than the younger age groups (current- years). Suicides are not recorded in national ty about 1$ per 100,000 population), men mortality figures for the 0-4 year age group, have rates higher than women, whites have and recorded suicid% for 5 to 9 year olds are higher rates than no whites, and white men very rare, usually less than 10 per year. tend to have the highest rates. Trends over Among the age groups under study here, 26 time show the familiar period effects:in- to 24 year olds have the highest elites, fol- creased rates during the 1930's, decreased lowed by 15 to 19 year olds and 10 to 14 year rates during the 1940's to about the mid- olds, respectively. 1950's, increases through the late 1970 s For 10 to 14 year olds, of interest are the low when suicide rates for adolescents were rates (les: than 2 per 100,000 population), hi.s.her than ever recorded in this country the recent increases in rates, and the fact that and a recent leveling off and decrease in boys tend to have higher rates than girls. rates. White boys have the highat rates. Data on Youth Suicide and Population Figure 2 shows the suicide rates for 15 to 19 Shifts. The data indicate that significant year olds, and here the trends are clearer. positive correlations exist between adoles- The rates are higher (currently about 8 per cent and young adult suicide rates and the 100,000 population), young men tend to have proportion of that age group in the United higher rates than young women, and whites States (Figures 4, 6, and Table 1). That is, in- higher than nonwhites. White men are at creases (and decreases) in the proportion of highest risk. In addition, the time trends, or 15 to 24 year olds are accompanied by in- period effects*, are more apparent, with in- shoed ufeets involve changes in rates of mortality or illness creased rates during the 1930's, decreases donne pestkuler itietorical period (Ifolford, 1983).

Correlation Coefficients Between Suicide and Hormcide Rates and Population Ratios for 18 to 24 and 34 to 44 Year Olds, United States, 1933-1983.

luicideBate tiomicideta Proportion of 15 to 24 Year Olds In Total U.S. Population + .34** + .41* Proportion of 35 to 44 Year Olds in Total U.S. Poptiation -.52* p<.001

Sources of suicide and homicide data: Vital Statistics - Special Reports Vol. 43 (for 19334953); Grove RD, bezel AM: Vital Statistics Rates in the United Stet= 19404960, U.S. Government Printing Office, 1969 (for 19544960); Vital Statistics in the United States, Mortality 19614979 (for 19614979); and National :stater for Health Statistics, unpublished data (for 19904910). SOUIVOs of population data: Grove Rd, Hetzd AM: Vital Statistics Rata in the United States: 1940-1960, U.S. Government Printing Office, 1969 (for 1933-1960); Vital Statistics in the Uni4 States, Mortality 1961-1979 (for 1961-1979); and National Center for Health Statistics, unpublished data (for 19604983). Table 1. 28 2-24 Manger: Sociodomographict Epidemiologic...Attributes

SUICIDE RATES 111.-1e ytIM OWS. SUMS 15 191011411113

14.

1$.

***tt 10111 $1-1 0 WHITE "V" Newman 0041 A liCaarnell MAW

.- 10141035 11140 1.,...°Aa 1550 1545 fan 475 1950 11103 YEAR Sources of data: Number of deaths: Unpublished data, National Center for Heatth Statistics (for 19334983). imams,Pc.e. `-tion, for rote derivation; Unpublished4Inw data, National Center for Health Statistics (for 19334983). Figur* 2.

1433 1436 1440 1448 1140 IWO lees 1570 t9T5 *NO 1443 YEAR

Sources of dat& Number of deaths: Unpublished data, National Center for Health Statistics (for 1933.1963). Population, for rate derivation: Unpublished data, National Center for Health StitiStial (for 1933-1983). 41111111MINIONIONIINIMPIMIIIIMMIONINIM Figure 3.

2-25 Report of the Secretarys Task Force on ituth Suicide

Suicide Rates and Population Change% United States, 1933-1982: 1544 Years Olds r"

0-0 won 0--.0%somame Via mi. is

elm *its *CM gess ,914 g nos Sources of data: Same as in Tabk 1. Rerrinted by permission: Am i Psychiatry 144:215.219, 1987.

Figure 4.

Figure 5.

2-26 30 PliolingerSociodemographicr Epidemlologic...Attributes

NODE RATES (ACRIAL ANDmac= meVOLUTION °NAMES. *IA YEAR MDR (111W STATES 1933-2000

Is335935 1940 5945 59150 MS ISSO UM 5970 5975 IMO YEAR Sault= o( data: Same a in Table l.

Figura O. creases and decreases, respectively) in their of 15 to 24 year olds are accompanied by insuicide rates. Opposite trends are seen for creases (and decreases) in their suicide and adult and older age groups, i.e., as the homicide rates. Figure 5 shows the relation- proportion of adults increase, their suicide ship between suicide and homicide rates and rates decrease. Homicide data follow the population changes among 35 to 44 year olds, patterns of suicide data for both younger and an age group whose trends are similar to the older age groups. These findings and result- adult groups in general (35 to 64 year olds) ing hypotheses regarding prediction should (Holinger and Offer, unpublished data). For be viewed with caution, because of the the 35 to 44 age group, rata can be seen to methodologic problems inherent in utilizing have time trends somewhat opposite to the national mortality data, as well as the number population changes:decreases in the of years required to adequately test such proportion of 35 to 44 year olds are accom- epidemiologic propositions over lime. panied by increases in their suicide and homicide rates.For 35 to 44 year olds, Figures 4 and 5 focus on suicide and homicide suicide and homicide rates were high during rates, and these figures present examples the early 1930's to the mid-1970's, and from 2 age groups to depict the differences leveled off recently. In contrast, the propor- between the younger and adult age groups. tion of 35 to 44 year olds in the population Figure 4 shows the changes over time in mor- reached high levels during the 1940's, tality rates for suicide and homicide rates for decreased throughout the 1950's and 1960's 15 to 24 year olds, and the proportion of 15 into the mid-1970's, and recently increased. to 24 year olds from 1933 to 1982 in the United States. The mortality rates and Prediction of Adolescent Suicide Rates to proportion of 15 to 24 year olds in the the Year 2000. Inasmuch as the population population can be seen to be rather parallel: shifts among adolescents over the next 15 increases (and decreases) in the proportion years can be approximated, based on the cur-

2-27 31 Report of the Secretary's Task Force on Youth Suicide

rent number of children and adolacents in olds, and it corresponds to the peak in the the United States (Current Population proportion of 15 to 24 year olds in the Reports, 1984), we can use the population population. model to tsy to ptedict the suicide rates for Understanding the recent leveling off and 15 to 24 year olds from the present time to decrease in suicide rates among adolescents the year 2000. Figure 6 presents the actual is aided by Figures 2 and 3, which provide a suicide rates for 15 to 24 year olds (1933 to race and sex breakdown of rates for 15 to 19 1982), their predicted rates (1933 to MOO), and 20 to 24 year olds. For all race and sex and the proportion of 15 to 24 year olds can groups among 20 to 24 year olds, the recent be seen to be rather parallel, as noted pre- leveling off and decrease in rates can be seen viously: somewhat high levels in the 1930's, in Figure 3. Similarly, in Figure 2, 15 to 19 decreases during the 1940's and .6950s, and year old nonwhite men and nonwhite women increases throughout the 1960's and into the also show recent decreases in rates. late 1970's. The actual suicide rates show a However, for 15 to 19 year old white men and peak in the later 1970's with a recent leveling white women, the rates do not appear to have off and slight decrease. The recent rates cor- leveled off recently. To examine the recent respond to the recent peak (nearly 1976) in trends among white 15 to 19 year olds, we ex- the proportion of 15 to 24 year olds in the amined in detail the rate of rise in suicide for United States. As noted earlier, this these two groups. We found that the rata; of relationship between suicide rates for 15 to increase show a statistically significant 24 year olds and their population shifts is decrease over the past several years for white statistically significant. men and white women in the 15 to 19 year The proportien of adolescents is expected to age group (Holinger, Offer, and Zola, un- decrease throughout the 1980's to the mid- published data). 1990's with a subsequent increase. These data for the proportion of adolescents in the DISCUSSION United States (Current Population Reports, 1984). The predicted suicide rates for 15 to Self-destructiveness among the young 24 year olds show a decrease throughout the presents us with something of a paradox. On 1980's and into the 1990's, corresponding to the one hand, young people are at lowest risk the decrease in the proportion of adoles- of suicide, and their suicide rates are lower cents. than for any other age group. Each year, only about one adolescent per 10,000 commits While at least decades will be needed to suicide, whereas the rates for older ages may evaluate the population model, recent data be ten times that figure. In addition, even lend some support to the hypothesis. The ac- among those adolescents identified as need- tual suicide rates for 15 to 24 year olds, 1977 ing treatment (Offer, Ostrov, and Howard, to 1983, were reported as follows: 1985), i.e., a high risk group, there is only 1977:13.6 about one suicide per 1,000 disturbed adoles- 1978:12.4 cents. This forces us to ask a number of ques- 1979: 12.4 dons: Why are adolescents at such low risk 1980: 12.3 of suicide? What intrapsychic and sociologic 1981:12.3 factors protect adolescents from suicide? 1982:12.1 ...sous= ot dux mud madam of the United States, 1977. 1919 (for 1977 to 1979 data) National Center for Health 1983:11.9 Statistics: Advance report, final mortality statistics, 1981. 1984:12.5 Monthly Vital Statistics Report, Vol. &Is No. 3 Sapp. (for 1960 and 1961 data); National Center for Health Statistice 1985:12.2 (10% sample)* Advance report, final mortality statistics, NM. Monthly Vital Statistics Report, Vol. 33, No. 9 Supp. (for 1963 data); The 1977 rate of 13.6 is the highest suicide and National Center for Health Statistics, unpublished rate ret.orded in this country for 15 to 24 year data, for 1963 to 1965.

2-28 32 P.M:flinger: Sociodemographice EpidemiologIc...Attributes

Why does suicide among youth seem to at- We projected the proportion of teenagen tract so much more msearch and media at- over the next 20 years and then made predic- tention than does suicide among age groups tions of the trends in suicide rates based on with much higher rates? Given the relative- those population changes. We began con- ly low rates of adolescent suicide, can the structing the prediction model up to the year issue of youth suicide be conceived of as a 2000 as described earlier. With the propor- major public health problem? tion of adolescents peaking in the late 1970's and then beginning to decrease, we sug- On the other hand, suicide is the second leading cause of death among adolescents. gested that the suicide rates among youth would peak then as well and then begin to Suicide, homicide, and accidents, all of which level off and decrease. This prediction con- may reflect self-destructive tendencies, are, trasted with predictions based on cohort in the aggregate, the leading cause of death studies which implied continued increases. for persons aged 1 to 39 yeats in the United The importance of the attempt at prediction States. Such deaths account for more years of life lost in this country than any other of suicide rates using the population model lies in the possibility of intervention and cause. In addition, suicide in particular ap- prevention on a large-scale, epidemiologic pears to create a ripple effect; it seems to level. have a li& -long traumatic impact on the survivors. It is our task to examine risk factors in youth suicide from primarily one perspective, the Throughout the 1970's, during studies of the epidemiologie There are other, obviously descriptive epidemiology of adolescent important risk factors, such as familial, suicide, we began focusing on the population biologic, and affective disorders. The shifts among various age groups over time. Specifically, we began examining the epidemiologic perspective may, however, provide a somewhat different view of the relationship between population changes concept of risk factors in general. While among the young and their violent death rates. We found that increases and white male adolescents are still at greatest risk of suicide among the young, the youthful decreases, respectively, in the proportion of population as a whole may be at less risk of adolescents in the U.S. population were accompanied by increases and decreases, suicide over the next decade if the hypothesis regarding the population model continues to respectively, in their suicide and homicide be supported. However, with the next in- rates, and we began publishing these findings crease in the prcportion of adolescents in the (Bolinger and Offer, 1982, 1984). The op- total United States population (probably posite trend was found for older people, i.e., beginning in the mid-1990's), the youthful as their proportion in the general population population as a whole once again may be at increased, their suicide and homicide rates decreased. The works of Brenner (1971, greater risk, as was the case during hv 1970%. Various preventive interventitm are implied 1979) on the economy and Easterlin (1980) by the population model when the suicide on population variables were particularly useful in rounding out the sociologic- risk increases for the youthful population. For =ample, as detailed below, high schools, epidemiologic perspective by which to un- folleges, mental health services, governmen- derstand these findings. As the data tal agencies, and businesses will all have a emerged on population changes and violent role in addressing this increased risk. deaths, the predictive aspects of the population model assumed particular importance. Adolescent Sukide and Population Shifts. Inasmuch as we knew the numbers of One seems obliged to attempt prelimimuy children and preadolescents in the popula- explanations of the findings at this point. At tion, it was possible to approximate the num- least three levels of interpretation seem bers of teenagers over the nod two decades. necessary and need to be subjected to further

2-29 Report of the Seaeterys Task Force on Youth Suicide hypothesis-testing: an epidemiologic- cause of a relative decrease in psychiatric ser- sociological level, a psychodynamic-clinical vices, and counseling that is available for level, and a nosological level. diagnosis and treatment. On the epidemiologic-sociological level, Potential Prediction:Psychiatric and suicide and homicide rates may increase with Publk Health Implications. One of the increases in the proportion of 15 to 24 year main features of published reports utilizing a olds for a variety of reasons; for example, in- population model has been the potential for creased competition for jobs, college posi- prediction of suicide, specifically among the tions, academic and athletic honors results in young (Hendin, 1982; nolinger and Offer, an increased number of adolescents who fail 1982, 1985): it was suggested that as the ab- to get such places (Holinger and Offer, 1982, solute numbers and proportion of adoles- 1984; Holinger and Offer, 1986).Such cents and young adults began to decrease in reasoning is consistent with Barker's (1964, the late 1970's and 1980's, the suicide rates for 1968) extensive data on large and small those ages (which bad been increasing over schools. In addition, the younger members the previous 20 years with the increase in the of the 15 to 24 year olds may be the least youthful population) would begin to level off powerful and attractive force in society with and decrease as well. This hypoihesis hits respect to political pressure, jobs, and so on. some support from the recent data noted ear- On the other hand, the adults in the 35 to 64 lier, showing a peak in adolescent suicide year old groups are much more powerful rates in 1977 with a subsequent decrease in politically and, with the exception of the the proportion of adolescents in the popula- older adults, attractive with respect to tion. Therefore, several researchers (Maris, employment (experience, schooling com- 1985; Holinger and Offer, 1982, 1984; Hen- pleted, etc.). Thus, the population increases din, 1982) have explicitly or implicitly in the adult age group may lead not so much predicted that with the leveling off and to increased competition and failure but decreasing of the population of 15 to 24 year rather to more economic benefits (greater olds would come a corresponding leveling off and more successful pressure on government and decrease in the s'incide rates of that age and union leaders to enlarge the job market, group during the late 1970's and 1980's. obtain more health services, etc.). There- There are important psychiatric and public fore, suicide and homicide rates would health implications in this model. Based on decrease with the increased population ratio current population projections, the decrease in the adult age groups. in the numbers and proportion of 15 to 24 Briefly, explanations for the two other levels year olds will be ending in the mid-1990's, follow somewhat similar reasoning. For ex- with another increase in 15 to 24 year olds ample, on the psychodynamic-clinical level, beginning at that time (Current Population depressed adolescents with marginal ego Reports, 1984). Thus, the population model capabilities and an inadequately internation- would suggest that the government, schools, alized sense of self-esteem may be at in- employers, health services, etc., should be creased risk of suicide during times when the ready to respond to that increase in terms of increased number of adolescents lead to increased psychiatric services, counselors, heightened competition for much-needed jobs, high school and college expansion, and external sources of self-esteem (e.g., so on. A preventive response would thus be academic honors, places on athletic teams, created, rather than "after-the-face reactive etc.). On the nosological level, when the model. proportion of young people is high, adola- The psychiatric implications of the popula- cents with thought disorders or major 4ffec- tion model are particularly important when live disorders may be at greater risk of suicide one considers epidemiologic data on the not only for the above reasons, but also be- number of adolescents who need help but do

2-30 34 P.HolInger: Sociodemograph14 EpidemfologIc...Affilbutes not get it Offer, Ostrov, and Howard (1985) death: (1) Further leng-term, cross-cultural found that 20 percent of adolescents in their comparisons of suidde, homicide, and acci- sample needed psychiatric help but that only dent rates with a focus on period, cohort, and 4 percent received help. Thus, 16 percent of age effects, and (2) cross-cultural studies that adolescents in their sample were in need of examined violent deaths and population treatment but did not get it With an in- changes to evaluate the predictive and creased number of adolescents in the popula- preventive aspects of the population modeL tion will come even higher numbers of Fourth, and finally, we have attempted in this adolescents who do not receive treatment, paper and others (Holinger and Offer, 1982, for whatever reason, and it would be espe- 1984) to examine violent death mortality cially critical that this issue of treatment rates primarily from the perspective of the availability and utilization be addressed single variable of population changes. We befixe periods of increase in the adolescent have focused on this variable because of its population. potential for prediction, with changes in population for the various age groups being IMPUCATIONS FOR FUTURE known years in advance. Yet, it is apparent RESEARCH that understanding something as complex as Four areas of future research appear par- violent deaths (whether from an intrapsychic ticularly important with respect to violent or epidemiologic perspective) requires a con- deaths, their potential for prediction, and cept of a general systems approach. From an population changes. First, it is critical that epidemiologic perspective, the work of the epidemiologic data be increasingly ac- Easterlin (1980) on population changes and Brenner (1971, 1979) and others (Wasser- curate, both in terms of mortality data (local and national) as well as the population bases man, 1984; MacMahon, Johnson, and Pugh, from which mortality rates are derived. 1963; Colledge, 1982; Courmier and ICIer- man, unpublished manuscript) on economic Epidemiology that utilizes national mortality variables have been particularly important. data is a relatively young field in the United States; only since 1933 have complete The relationship between economy and mor- population data, not just &milks, been avail- tality rates is well documented, with poor able for all States in the United States. economic conditions (as indicated by high Therefore, many more decades of data and unemployment rates) being related to higher study will be needed to test various modality rates (Brenner, 1971, 1979; Was- hypotheses involved in these epidemiologk. serman, 1984; MacMahon, Johnson, and trends. Pugh, 1963; Oolledge, 1982; Counnier and Merman, unpublished manuscript). In addi- Second, prospective studies are needed to tion, the birth rate m the United States test specifically tile hypotheses on the poten- (which, with immigation and incieased life tial for predicting violent deaths, using a expectancy, will be responsible for most of model of population shifts. To this end, the relevant population changes) tends to be mathematical models should be developed to inversely related to economir changes in the predict violent death rates based on projec- United States: bad economic craditions cor- tions of the future populatbn. These predic- respond to low birth rates and vice-vena tions of rates then could be measured against (Turner, et aL, 1981). Thus, an interacting the actual &dings over *he next several system emerges, within which violent death years.Such models are currently being mortality may be understood from ars developed. epidemiologic perspective; this system in- Third, at least two types of cross-cultural cludes such variables as economic conditions, studies of violent mortality would help en- birth rates, and population shifts. However, hance understanding of this leading cause of despite the well documented relationship between the economy and mortality rates, one

2-31 :4 5 Report of the Secretary's Task Force on Youth Suicide cannot predict future violent death rates tors associated with health and social adaptation in tho city from this relationship because of the difficul- of Rotterdam. tkban Ecology 2205-234, 1977. ta 14sNord Tit The estknation of ags, period end ty in predicting future economic conditions. cohort effects for vital rates. Biometrics 39:1311-1324. Thus, another important area of future re- 1083. M. Malinger PO Violent Deaths in the Unitad Stews search emerges: further work is needed to 1000-10110 M Epidentiologio ftdy of Suitt* Homicide determine if the population model discussed and Accidents. Nov York: OulliordPnw, 1087 (in press). in this paper will make possible predictions 21. Holinger PC, Hanger_,t) Sandlow J:Violent Oaths among children In the Wad States, 1000-1980 not only for violent death rates but also for spiderniologio study of suicide, homicide, and sooltion- tat deaths amang 544 year olds. Pediatrician *1149, economic conditions for specific age groups 10834985. (as per Easterlin's work (1980)), inasmuch as 22. Malinger PC, Klemm EH: Violent Deaths in tho the population shifts for certain age groups Unitod Stites, 10001975. Soo Sol Med 1&1923-1936. 1062. are known years ahead. 23. HaWrger PC, Moe D Porepectives of suicide in adolescent*, In Research In Community and Mental Hank Vokane 2 Ilimmone R. ad). Greenwich. CTIAI Press, 1081. pp. 139-157. REFERENCES 24. Malinger PC, Offst ikPrediction of adolescent suicide: A population model. Am J Psychiatry 139:302- 1. Basler RO:EcoloQical Psychology. Stanford, 307, 1982. California, Stanford Un Pfau 1961 25. Malinger PC, Offer D: Toward the pradiction of 2. Barker RO, Gump PV: jig School, Small School: violent deaths among the young. In Suicide Mom the High School Sin end Student Worbr. Stanfurd, Cantor- Young (Sudak. et of, ode) Now To*. Wright PSG, Ma nia, Stanford University Press, 1984. 28. Hotlngsr PC, Offer Cr:Ths spidomiology of 3. Brenner kik limo Series Analysis of Reimion- suicide. ho,fl4cld,, and =Mentor among adolescents. ships between Selected Economic and Social indicators. 1900-1980. In Advances in Adolsocent Menial Hoak Vol. Leir.114elic.f, Wyk*: National Technical information Ser. 1, Part El (Feldman R et). Gnomic:h. CT, JAI Press, 1988, pp. 119445. 4. Brennsr ktH: Mortality and Ms national economy. 27. Klobba Ai: Homicide trends in tiro United States, Lanost tartseava 1900-1974. Public Health Reports 90:105-204. 1976. & Coliodgo M: Economic es* and health. Soc. Sci. M. IleCteAJ, Dolman A& Comparability of mortality Med. 151919-1927, 1082. statistics for the month and eighth revisions of the knee- 8. Cormier P4.1 Korman OL:Unomployment and national Classification of Disesses. United States. Vital Ind male-fibmale labor force participation as detsrminants of Health Statistics, Wee Z No. a& Washington DC, US changing suicide rates of main and females in Ousbsc. Government Printing Moe. 1971 tInpubliihid manuscript 29. Kiernan Gt.; Laved Kt Roo .1 et al.: Birth-Cohort 7. Doege T: Aninjury is no accident NEJM 291509- fronds in raise of miler depressive disorder among rola- 510. 1978. lives of patient* with affective disorder. Arch Eton Psychiatry 42.8814103, 1081 8. etmn HI, Shockley W: Compwison of came of death assignments by ths 1929 and 1938 revtelons of tho 30. Kromer kf, Pollack ES, Redick RW, Locke RI: Men- knomational Ust Deaths In the UMW Stun, 1940. Vie! tal Disordera/Suicide. Cambridge, MA, Harvard University Statistice-Speolal Reports, WI. 19. No. 14, 1944. Press, 197Z 9. Easterlin Rk Birth and Fortune. New York, Casio 31. Levy L. Harcog k Effects of population density Books, 1961 and crowding an health and social adaptation In the Netherlands. J Health Soo Behavior 15:221440, 1974. 10. Farberow N: The Many Faces of Suicids. New York, Mo0raw4451, 1979. 32. Low L. Herzog A: Effects of powOlg on heaith and social adaptation in the" of Chicago. UUrt Ecol- 11. Faust MM, Dolman AB: ComparabIlity of mortality ogy 3327.354, 1979- statistic* for the fifth and sixth revillana thud Stoics. 1060. Wel Statisdos-Special Reports, Vol. 51, No. 2, 1983a. 33. Matrkfahon O. Jchnson S. Pugh TF: Relatkro of suicide rides to semis) conditions. Pubilc Health Reports 12. Faust MM, Dolman AB:Comparability ratios 78:25 5293. 1083. based on mortality statistics for the fifth and sixth revisions:United =e, _1950. Vital Statistics-Spedal 34. Maris Ft The adolescent suicide problem. Suicide Reports, Vol. 51, No. 3, 1983h. life-Threatening Belt 15:91-109: 1985. 13. Faust MM. Dolman A& Comparability of mortality 31 Monninger KA:Man Against Himself. New York statistics for the sixth and month revisions:Unita! Harcourt Brea* andCo.1938. States, 1968. Wei Statistics-Special Reports. WI. 51, No. 31 Murphy (3E, Wetzel RD: Suicide risk by birth cohort 4, 1981 In the United Mateo, 1949.1974. kch Gen Psychiatry 14. Flood & Tito Pepliopathrology of Everyday Life. 37:516823. 1960. London, The Hogerth Prow 1960 (1901). 37. Notions/ Conte( for Heatth Staffs** Muni Sum- 15. Gordon RE. Gordon MC: Social psychiatry of a mary for the United States, 1979. Monthly Vital Statistics mobile suburb. Int 41 Social Psychiatry 8:89-1011, 1980. Root* 2121 1980. 18. Mellon CP. Solomon MI: Suicide and sip in Albsr- 31 Notional Center for Minh Statioffor:Advance 11, Coned& 1951-1977. Arch Gen Psychiatry 37:505-510, Report final mortality statist 11. %ea Monthly Mal Statis- 1080. tics lispont Vol. 33, No.9. Sapp. DOHS Pub. No. (PHS) 1121 Puboo Health Nano% Hyattsville. PAX Deo. a% 17. Hendin H: Suicide In America. New York. WW Nor- 1084. ton & Ca.. a arer 0, Owed E, and Howard Kt: Epidemiology of 18. Herzog A, Levy I, %Wank Soma toological too- Mental health and monist Illness among adolescenta hi

36 2-32 BEST COPY AVAILABLE P.Holinger: Sociodemographicr Epidemic:logic...Attributes

Significant Advances in Child Psychiatry (Cill %hod). Now York: Bask Books kw, 1906, in pins. 40. Peck M, Litman RE: Wont bends In youthful suicide. Tribune Medics 14:13-17, 1973. 41. Seiden RH: Suicide mono youth: A review at the literature, tgoo 1987. Bull Suieldology (Supp!), 1950. 42. Belden Mt Death h the West - Amolonal analysis of the youthful suicide Mt. WestJMed 1 1954. 43. Seldom RH, Pelts* RP: Shifting patterns of dead- Verolaience. Suicide Life - Threatening Bah 10195-209,

44. Shapiro J, Wynn* Ek Adolescent alienation: &vitiating the hypoths.es. Social !tidiest= Research 11X423-4311. 1962. 45. Solomon MI, Mon CP: Suicide and age In Alber- ta, Canada, 1951-1977. Arch Gen Psychlatm 37:511-513, 19IXX 46, Too= JR Suicide and suicidal attempts In children and adolescents. Mt 4 Psychiatry 11e719-724, 1952. 47. Toolan JM: Suicide in children and adolescents. Mi J Psychiatry 29339-344, 1975. 43. Taring MT, Boor M. Renting A 4= aspecte of traffic accidents. MI 4 Psydthifty 1955. 49. Turner CW, Fenn MR, Cole AM: A social analysis of violent behavior. In Violent Be- griol °CMlearning approaches to Prediction, Managernent and Trostmentpot Re, ed) New York, enotner/Mexel,1981. 50. U.S. Bunau of the Census, Current Population Repot* Series P-20, No. 952, Projections of the Popula- tion of Ihe United States_by Pee, Sex, and Race: 1110to 2080. U.S. Government Met. Office. Washington, DC, 1984. 51. Vital Statistics of the United SOM, 1979: Mor- tality. Washington DC, US Government Printing Office, 1954. 52. Vital Statistics-Special Reports. Vol. 14, No. 2. Washington DC, US Government Polley Office, 194L 53Vital Statistics-Special Reporeg Ds= rates by raoe and sex. United States, 1900-1953. Vol. 43 Washington DC, US Government Printing Office, 1958. 53. Wasserman 1M: The Influence of economic busi- ness cycles on the United States suicide rates. Suicide Ufe-Threaning Belt 14:143-150, WM. 56. WechaW it Community growth, depressive disorders, and suicide. Mt J Sociology 67946, 1951. 58. Weiss NS: Recent trends in violent deaths among yowv adults In the United States. Mt J Epmiideology 103:8422,41 1978. 57. Wolfronnitchl:Suicide by means of victim- =rd .J Exp Ntydwietol 20:335- 5S. Wolfgang ME:Patterns in Criminat Homicide. Philadelphia, PA, University of Pennsylvania Press, 1988.

2-33

BEST COPY AVAILABLE PREPARATORY AND PRIOR SUICIDAL BEHAVIOR FACTORS

Nonnan L Fatherow, Ph.D., Cofounder, The Institute for Studies of Destructive Behaviors, and Suicide Prevention Center, LarAngeles, California

INTRODUCTION Suicide does not just occur. Experience has then on research-substantiated factors. shown that it is more often the end result of Finally, it must be remembered that, while a process that has developed over a period of considering each factor separately, the fac- time and within which have been many fluc- tors almost always will appear in a context tuations in the course of reaching the that will contain other clues, some of which decision to act against oneself. Fortunately, may affect considerably the significance or experience has also shown that during that the factor discussed. peziod, the individual engages in a number of behaviors that have become available as signs PRIOR SELF-DESTRUCTIVE and portents. There are, of course, many BEHAVIOR kinds of signs such as epidemiological, demographic, family characteristics, per- Prior self-destructive behavior, whether in sonality disorders, cultural factors, and the form of suicide attempts, threats, idea- others. Although inevitably there will be tion, or gestures, has been identified by many some overlap with other areas, this paper will investigators as a powerful indicator of com- focus on behavioral factors only, that is, acts pleted suicide risk (1,2,3,4,5). However, the and actions engaged in by the person that in- research was conducted primarily on adults dicate the potential for a self-destructive act in psychiatric settings, general hospitals, and to occur in the future. in the community. This paper will focus on children, adolescents, and youth. For the Because of the large amount of pertinent re- most part, studies reviewed are limited to search in the area of behavioral factors, this those whose subjects are aged 20 or below, review, necessarily, is arbitrary and selective. but occasional research that includes older For example, no effort has been made to ex- ages are included. The format will look at amine all the so-called suicide potential studies of completed suicides, then at- scales containing predictive items.First, tempted suicides within each risk variabie. there are no scales specifically constructed to The research on children or young adoles- evaluate suicide potential In adolescents; cents are presented first.In addition, the second, the behavioral items on the scales are studies are further divided on the basis of the almost all considered individually in this presence or absence of control or com- paper anyhow; and third, many of the scales parison groups. depend on clinically reported items rather

38 2-34 N.Farberow: Preparatory & Prior Suicidal Behavior...

PRIOR SUICIDAL BEHAVIOR peers.Shafii, Carrigan, Whittinghall and Derrick (6) at the University of Louisville Completed Suicides Control conducted psychological autopsies on 20 Groups adolescents, ages 12-19, who committed suicide in Jefferson County, Kentucky. The risk factors of previous self-injuries, Their highly relevant control group was made suicide attempts, threats, ideation, and ges- up of 17 matched-pair living peer friends, the tures have been grouped under the general same age and sex as the adolescent suicide, heading of prior suicidal behavior. Most of from whom the same extensive set of data the time, the prior suicidal behavior refers to were gathered. The researchers found that suicidd attempts. However, many re- the suicides were significantly more often searchers have not differentiated between likely to have a history of suicide threats suicide attempts and gestures, or between (p <.02) and suicidal ideation with expres- suicidal ideation and threats, and often, not sions of the wish to die (p <.02) than did the between suicide attempts and threats. When controls. such distinctions appear it is uncertain how much overlap has occurred, for suicide at- Rich, Young, and Fowler (7) also conducted tempts are almost always preceded by threats psychological autopsies. The subjects were and ideation. The best approach was all 283 completed suicides in San Diego deemed the conservative one of assuming County over a period of 20 months. The in- that all the forms of suicidal activity, verbal vestigators compared the 113 suicides under and behavioral, are equivalent for the pur- the age 30 with the 150 suicides age 30 and poses of risk evaluation and prediction. As over and found that prior suicide threats and seen later in this paper, this view is substan- attempts were high in all ages of completed tiated by other investigators.Dorpat and suicides. Prior 'suicide talk,* 71 percent, and Ripley's (5) review of studies of committed suicide attempts, 42 percent, were noted in suicides from the United States and England the under 30 age group, but were not sig- led them to the conclusion that between 20 nificantly different from the frequencies of percent and 65 percent of individuals who 63 percent suicide talk and 35 percent at- commit suicide have made prior suicide attempts found in the over 30 age group. The tempts. In an additional 15 studies (5) that comprehensive psychological autopsy study determined the incidence of completed of Maris (8) compared completed suicides suicides in groups of attempted suicides by (from Cook County, Illinois) with attempted followup, the authors found that the percent- suicides and natural deaths (from Baltimore, ages ranged from 0.03 percent in one short Maryland). The study covered all ages, in- followup study to 22.05 percent in the longest cluding 36 young adults, adolescents and folowup study. The researchers estimate children ages 10 to 29, among his study the incidence of committed suicide among population of 414 subjects. In analyzing his the suicide attempts to be between 10 per- data on all subjects with highly sophisticated cent and 20 percent, but add their belief that statistical methods, Maris found that the the actual number of attempted suicides who single factor with the greatest discriminatory go on to commit suicide is greater than the power (beta .64) was the number of prior percentage given. suicide attempts; however, that factor separated the two suicidal groups from the Relatively few investigators over the past 10 natural deaths, but not from each other. The to 15 years have used control groups in their interesting factor alai distinguished the com- studies of completed adolescent suicides, a pleted suicides was that they bad one rela- fact that does not surprise too much in terms tively serious prior attempt, while the of the problems involved in obtaining ap- attempters had several low-lethality prior propriate groups for comparison. One ap- suicide attempts. When Maris (9) later com- proach has been through the use of matched pared his group of 36 young completed

2-35 Report of the Secretary's Task Forre on Youth Suicide suicides with the remainder of his completed information from a wide variety of sources suicide group, using the same statistical pro- government, school, medical, sodal service, cedures, he found that the factor most sig- and family. The histories of his 30 cases dis- nificant in differentiating the young suicides closed that 46 percent, almost half, had pre- from the older suicides continued to be the viously attempted or threatened suicide, or greater number of prior suicide attempts in revealed their suicidal thoughts before their their histories. death, with 27 percent having done so in the 24-hour period before their death. Seidin's (10) study compared a group of 25 students at the University of California at Three studies, in widely separated parts of Berkeley who committed suicide over a 10- the world, using coroner's office reports and year period, with the entire UCB student hospital records, where available, of com- body population during the same period. In- pleted suicides, found that prior suicide at- formation from newspaper clippinp, police tempts were siptificant. Cosand, Bourque, files, university records and reports of friends and Kraus (12), gathered data on 315 suicides and acquaintances indicated there were between ages 10 and 24 that occurred in *numerous warnings' in almost every ease of Sacramento County, California, between the suicide, and 22 percent bad made prior years 1925 and 1979. They compared the suicide attempts. They bad also given subtle older youth with the adolescents aged 15 to warnings, such as making wry jokes about 19 and found that the older age group had killing themselves, or crossing out the word made significantly more suicide attempts emergency in the question "Whom shall we than the younger age suicides and that the notify in case of emergency?" on the medical females in both age groups had significantly history form, and substituting the word, more prior suicide attempts than the males. death. In another long term study, Marek, Widacki, and Zwarysiewicz (13) compared 76 cases of PRIOR SUICIDAL BEHAVIOR committed suicides among *juveniles" occurring between the years 11 to 1960, with 76 Completed Suicide - No Control similar cases occurring over 15 years, 1960 to Group 1974, in Cracow, Poland.Data were obtained from hospital records and coroners' Control groups for completed suicide are offices. In t! : more recent 76 cases, they often neither available nor feasible. Non- found prior suicidal attempts in 17 jxrcent of controlled studies are nevertheless valuable the cases and suicide threats and ideation in in providing information and generating an additional 13 percent.In a large-scale direction and hypotheses for further study. study of Ontario, Canada, Garfinkel, Cham- Some of the following suicide studies of berlin, and Golombek (14) examined all the children and adolescents are marked by ex- data available in the coroner's office records tensive exploration into background, per- for the 1,554 suicides aged 10 to 24 that oc- sona 1 ty, behavioral, physical, and curred between January 1971 and August communication factors. As in the case of 1978. They found that a history of threats, controlled studies, prior suicidal behavior is note-leaving, general manifestations of referred to often as a significant risk-in- depression, and other suicidal behavior "ac- dicator in the histories of the completed counted for the majority of prior symptoms" suicides the investigators studied. in the cases for which information on prior Shaffer's (11) study is well-known and often conditions was available. referred to because of its study population In summary, studies of the histories of com- completenessall the children and younger pleted suicides, whether control or com- adolescent suicides under age 14 occurring in parison groups are used or not, indicate that England and Wales over a 7-year period, prior suicidal behavior in the form of at- 1962 to 1968and its exhaustive search for

2-36 40 N.Farberow: Preparatoty & Prior Suicidal Behavior... tempts, threats, or ideation, or combination no such gesture was reportedby any of the of these, is one of the strongest indicators of controls. McKenry, 'fishier, and Kelley (19) high risk in adolescents as well as in adults. noted that prior suicide attempts were Percentages of frequency of this factor, sup- marked in their study population of 46 ported by good methodology and statistical adolescent attempters admitted to a univer- evaluation, vary from n percent to 71 per- sity hospital primarily for overdose& The cent in study populations from university, nonsuicidal comparison group from the same government, and city hospital settings;in hospital had no prior suicide attempts in their psychiatric and general populations; and history. In a study from England, Hawton, from coroner's office records. Osborn, O'Grady, and Cole (20) followas up 50 adolescents who were treated for suicide overdoses at the Oxford General Hospital PRIOR SUICIDAL BEHAVIOR Psychiatric Service in England. The inves- Suicide Attempts, Threats, Ideation tigators fouttd that 40 percent of the at- temptenversus none of the controls from a Control Groups sample. of the general populationhad either The presence or absence of prior suicidal be- taken an overdose or injured themselves havior has been explored in controlled prior to the attempt that brought them into studies even more frequently among suicide the study. attempters than among completed suicides. Some studies have been retrospective, others Clarkin, Friedman, Hurt, Corn, and Aronoff prospective. (21) studied two different age groups of adolescents with suicidal behavior in a New Two studies serve as landmarks in this area. York Hospital; the first group had a mean Jacobs' (15) and Teicher's (16) carefully age of 16 and the second a mean ageof 25.5. designed research compared 50 adolescent The researchers noted that 58 percent of the suicide attempters treated at Los Angeles adolescents and 75 percent of the young County General Hospital with a matched adults had made at least one prior attempt. control group of 31 adolescents at a suburban Apparently, prior suicide attempts appear in high school. They reported 44 percent of the significant numbers in the younger, as well as attempters had one or more previoussuicide the older, adolescents and youth. attempts, while there were no attempts among the controls. Theother landmark early study was a totat population study in PRIOR SUICIDAL BEHAVIOR Sweden carried out by Otto (17) on all the Attempts, Threats, Ideation No children and adolescents under age 21 who made a suicide attempt and came to the at- Control Groups tention of health authorities. A control Selected studies of young suicide attempters group consisted ofadolescents from the seen in psychiatrichospitals and in private general population matched for age, sex, and practice note the presence of prior suicide at- geographical region. Otto found that 16 per- tempts in the history. Shafii and Shard (22) cent of his study population ofthe 1,727 reviewed 340 cases of children and younger young suicide attemptershad made previous adolescents mostly aged 13 to 15, for suicidal suicide attempts. Prior attempts were more or severe self-destructivebehaviors seen on common among boys (21%)than girls (14%). an emergency basis atthe Child Psychiatric Services at the University of Louisville In three studies of overdosers, priorsuicide Hospital; Diekstra (23) reviewed 158 admis- attempts were frequent in the experimental sions of adolescent suicide attempters to a groups and absent in thecontrols. McIntire hospital in Holland. Both studies noted that and Angle (18) found that 26 percent of 50 the risk nf suicide is significantly increased poison center patients ages 6 to 18, bad made when a previous suicide attempt has been similar suicide gestures in the past, whereas

2-37 4 1 Report &the Secretaty's Task Force on Youth Suicide made. Diekstra found that previous suicide frequencies of suicidal behaviors that sub- attempts and suicide threats or ideation sequently appeared. =kid first and second in significance among the indicators. As in the previous studies of attempted suicides, the presence of a prior suicidal be- Schneer, Pedstein, Pnd Brozovsky's (24) in- havior is a good predictor of the index teresting study reported that the admissions suLida! behavior, which in turn, becomes a during one year for a hospital in Brooklyn at valuable due to piedict suicidal behavior in two different times, 11 years apart, indicated 'the followup period. Barter, Swaback, and that the number of suicidal adolescents had Todd (27) and Stanley and Barter (28) a/most doubled, from 13.6 percent t26.7 provide two reports on a followup study of a percent, although the census of admissions group of adolescents, under the age of 21, had remained practically the same. Con- hospitalized for suicide attempts in the comitantly, the percentage of admissions Colorado Psychiatric Hospiva over a 3-year with repeated episodes of suicidal behavior period. A followup interview on 45 patients had increased five-fold, from 4 percent to 20 indicated continued suicidal behavior by 42 percent. Gabrielson et al. (25) studied a spe- percent of the youngsters.Stanley and cial risk group, 14 pregnant femaies, arAl 13 Barter (28) extended the study to include to 17, who had made suicide atteuipts and psychiatrically ill adolescents hospitalized at were admitted to the Yale-New Haven the same time but with with no history of Hospital, and found that 21 percent had his- suicidal behavilr. The followup data indi- tories of prior suicide attempts. Crumley cated that suicide attempts occurred in 50 (26) reviewed the histories from his own percent of the experimental subjects and in private practice of 40 adolescents, aged 12 to only 16 percent of the controls, a highly sig- 19, in treatment following a suicidal attempt. nificant difference. He found a history of numerous prior attempts was common in 17, or 40 percent. Otto's (17) study, referred to earlier, was actually a 10-year followup of 1,727 child and In summary, retrospective studies of histories adolescent suicide attempters in Sweden. of adolescent suicide attempters, all except They were compared with nonsuicidal one from hospital populations where they children equivalent demographically. A were treated, indicated substantial levels of huge undertaking, it was possible only in a prior suicidal behavior, ranging from 16 per- country where data are routinely kept on its cer t to 38 percent. At least 11 the studies citizens from the time they are born until they Vrae controlled. die.Otto found that mortality was significantly higher in the suicidal group than in SUBSEQUENT SUICIDAL the controls, with 54 percent deceased in the BEHAVIOR experimental group and 1.8 percent deceased among the controls. Of the at- Followup Studios Control tempters who died, 81) percent committed Groups suicide. Repeated attempts were more common among boys than girls, 21 percent versus The preceding studies were retrovective; 14 percent. they started with groups of completed suicides or attempted suicide and then Four additional controlled studies reported sought information in the young people's his- on the occurrence of suicidal events follow- tories that indicated identifying or predictive ing treatment for suicide attempts. Cohen- behavior, especially prior suicidal behavior. Sandler and Berman's (29) well-designed The following firoup of studies use a prospec- investigation conducted with young suicidal tive approach, taking groups from their index children, ages 6 to 16, followed for 3 years 20 suicidal behavior and following them for suicidal children, 21 depressed but not various periods of time, noting the kinds and suicidal children, and 35 psychiatric controls,

2-38 4 2 N.Farberow: Preparsioty & Prior Suicidal Behavior...

=suicidal and nondepressed. Five percent SUBSEQUENT SUICIDAL of the children had engaged in multiple BEHAVIOR suicide attempts before the index hospitalization. Twenty percent of the Followup Studies Not Control suicidal children were suicidal again after dis- Groups charge. No further suicidal behavior was lvo researchers report followup studies but reported in the other two groups. do not use control groups. White (33) sur- Rauenhorst (31) followed a group of 50 veyed 50 consecutive patients aged 14 to 19 Qtucasian female attempters, ages 16 to 30, admitted to a general hospital in Birmin- and compared them with matched controls, gham, England, following suicidal overdoses. who had been treated for minor or acciden- Information on 40 of the cases in 1 year of tal trauma. Followup continued for 14 to 21 blowup indicated that 10 percent of the months after discharge. Among the 38 ex- patients had further self-poisoning episodes. perimentals and 44 controls located and in- Crumley's (26) group of 40 adolescents in his terviewed, 5 of the experimentals, 38 private practice who had entered treatment percent, and none of the controls hadmade following a suicide attempt yielded one a subsequent suicide attempt.One person patient, 3 percent, who committed suicide made several subsequent attempts. within 2 years following his treatment. As part of his larger study, Motto (32) fol- In summary, the followup studies of suicidal lowed for 10 years male adolescents who had behavior support further the usefulness of been admitted and treated in psychiatric in- prior suicidal behavior as an indicator of patient hospitals in San Francisco became of suicide risk. Further suicide attempts ranged a suicidal state or depressivericod. He from 10 percent to 50 percent, and com- focused on the 122 boys in this popuiation pleted suicides occurred in from 3 percent to who were ages 10 to 19 and found that i 10 percent subsequently. As one would ex- suicides, 9 percent, occurred within a mean pect, most of the study populations were of 38 months after discharge. Motto further hospital patients. The followup times varied reports that 43 percent of the 119 subjects considerably. ranging from several months to bad a history of one or more attempts before 10 years. the index hospitalization, and that 9 percent In general, from the studies of completed of those who reported no prior suicide at- suicides, attempts, threats, and ideation tempts went on to commit suicide, whereas suicidal behaviorwith and without control 10 percent of those with a history of prior at- groups and retrospectively in casehistories tempts did so. or prospectively throughfollowupsprior In a study in Oxford, England, Hawton, et al. suicidal behavior is a significant and valuable (20) did two followups of 50 adolescent over- clue for further self-destructive behavior. dosers aged 18 and under, at 1 month and 12 The risk factor seems valid both for children months after discharge. He compared his and adolescents and for boys and girls and group, divided into two sub-groups, one15 also seems to increase in significance with in- and below and the second 16 to 18, with a creasing age among adolescents. general population sample previously collected. In the year following the overdose, 14 percent of the patients made furthersuicide attempts, all but one of them in the 16 to 18 year age group.

4 3 2-39 Report of the Secretary's Task Force on Youth Stickle

RELATIONSHIP BETWEEN for both suicide attempts ...ad for suicidal COMPLETED SUICIDE, thoughts and threats, and with adola.cents SUICIDE ATTEMPTS AND referred for suicidal thoughts and/or threats but with no suicide attempts. The com- SUICIDE THREATS parison group was the rest of their large Most efforts to explore the relationship be- sample of disturbed adolescents not referred tween the various kinds of suiciJe have ap- for suicidal behavior. The groups were com- proached the problem from two pared on family, school, friends, leisure tiric directionssimilarities and diffmences be- interests, emotions, etc. using psychological tween demographic characteristics, suicitte tests, interviews and therapists ratings. methods and motivations, andior the extent Marks and Haller concluded that there was to which all the suicide behaviors occur in the little evidence to support the assertion that same persons. Stengel gad Cook (38) were those teenagers that threatened suicide were the fust to demonstrate the demographic and markedly different from those who at- personality differences between completed tempted it The results also indicated the and attempted suicides in their London same distinctions in personality and study. Those differences, that is, that at- demographic characteristics between com- tempters are younger, use less lethal mits and threats as had been found between methods, have more women than men, more clmmits and attempts. For suicide attempts, often occur where others can rescue, have the 3:1 ratio of females to males was con- been substantiated regularly. The role of in- firmed; for suicidal thoughts, the ratio was tention to dk has been found to be critical, more 2:1 female to male. There were ap- with the attempters more often motivated to parently enough sex differences to em- influence others than to die. The concept of phasize the need to study the sexes attempted suicide has been incorporated in separately. Kreitman's term, parasuicide, in which he Goldberg's (41) study aimed at identifying emphasized the nonfatal aspect of the act characteristics that may be the same for when an Irdividual deliberately causes self- people with suicide ideation, those who at- injury or ingests a substance in cccess of any tempted suicide and those who completiAi prescribed or generally recognized suicide. A variety of schedules were ad- therapeutic dosage* (63, page 3). Kreitman ministered to 489 persons and the presence and his colleagues (39) find their studies have or absence of thoughts of suicide during the confirmed the traditional view that previous month were related to the various parasuicides and suicides are epidemiologi- factors as the outcome variables. Goldberg cally distinct. Sex and age patterns are dif- concluded from the similarities reported by ferent, rates are very different and the subjects, that suicide ideation, attempts, motivations may be different At the same and completions, if not on a continuum, are time they note the close relationship be- at least overlapping phenomena. Persons tween the two, with one percent of with these different kinds of suicidal manifes- parasuicides going on to commit suicide tations, she felt might be essentially within one year and 41 percent of the completed suicides having a history of equivalent. parasuicide. Although Kreitman's poptga- In summary, the overlap to which all the tions refer to the general population, their suicide behaviors, commits, attempts, studies included ages down through 15. threats, and education occur in the same population has been noted at length in the Marks' and Haller's (40) large scale study on studies surveyed in the preceding sections of the relationship between attempts and this paper. While it is ponible, as Kreitman threats compared adolescents who had made (39) has done, to focus on the differences be- suicide attempts with adolescents referred tween the populations, it is the similarities

2-40 Nfarberow: Preparatory & Prior SuicidalBehavior... and overlap that seem more impressive. Itis No studies, except those of Farberow and his also unlikely that an 'overlap" group exists as colleagues, have specifically explored the a separate voup, as has beensuggested. In parameters of these behaviors from thepoint general, it seems justifiable to conclude that of view of determining the personality the presence of any of the overt forms of characteristics, attitudes, prevalence and suicidal behavior, such as attempts, threats demographic features of persons who consis- and/or ideation in the prior history of any in- tently engage in this kind of behavior. Far- dividual can be considered valuable in- berow (36) selected various groups of dicators of high risk for further persons engaging in indirectself-destructive self-destructive acts. The usefulness of the behavior (ISDB) and conducted controlled presence of high levelsof indirect self- studies of their characteristics. One specifi- datructive behaviors as predictive clues is cally involved young people, juvenile delin- much more uncertain. quents. Studies were conducted of groupsof patients who were diabetic, had thromboan- gilds obliterans or Buergees Disease, were PREPARATORY BEHAVIORS elderly and chronically Ill, hyperobese, or on While the risk factors reviewed above, prior renal hemodialysis.Control groups were and subsequent attempts, threats and idea- made up of cooperative patients in each dis- tion, rate as excellent indicators of suicide ease group. In addition, afile survey deter- risk, they have the disadvantage of octarring mined the presence and kind of ISDB in five late in the process of identification, andmark treatment groups: two drug rehabilitation the fact that a suicidal state has already ex- programs, a youth delinquency programand isted. The obvious aim is to identify factors two hospital populations, one ofcompleted that occur earlier, before any own suicidal and one of attempted suicide. Briefly sum- behavior has appeared, so that intervention marized, the major Jesuit; obtained from to prevent possible injury ordeath can take these studies characterized the patients with place. The following section reviews studies levels of ISDB as:high impulsivity, low that note activities aptly termed "preparatory frustration tolerance, present orientation, behaviors" that may be considered as risk fac- minimal tolerance for frustration or delay, tors for potential suicide. and little future orientation or concern. These characteristics werc frequently ac- INDIRECT companied by strong drives toward risk taking, pleasure seeking, and a need for ex- SELF-DESTRUCTIVE citemtnt.In the case file study, the BEHAVIOR methadone maintenance group had the Indirect self-destructive behavior (ISDB) highest ISDB scores, especially on noncom- has long been recognized as an important pliance with medical regimen, not making aspect of self-destructive behavior(34) but it recommended changes in life activities, and has not been systematically investigateduntil disregarding treatment requirements. AI- relatively recently. Menninger (35) cohol abuse was one of the most frequent developed the concept psychodynamically as forms of ISDB, appearing in from one-half to 'focal or partial suicide' while Farberow (36) one-third of the drug abusers. In the youth has conducted research on a number ofthe 'Aversion group, where the ages were 18 and behaviors felt to be indirectly self-destruc- under, 5 percent were noted to be problem tive, or suicidal equivalents. Whereas a num- drinkers and 20 percent to be drug abusers. ber of behaviors have beenidentified as Overt suicidal behavior in any of the groups indirect self-destructive behavior (36),only was absent or minimal, except amongthe the research by Farberow will bediscussed in elderly chronically ill. There was a high in- this paper. verse correlation betweensuicide potential and life satisfaction, leading to the con-

2-41 Report of the Secreteryt Task Force on Youth Suicide

elusion that ISDB among the elderly, chroni- there was a continuous deterioration of their cally ill was generated by feelings of isolation performance as they p-Igressed through and lass, variables also known to arouse school. The students were filled with doubts direct suicidal behavior. It may be that the of adequacy and were despondent over their elderly, chronically ill person uses ISDB asa general academic aptitude.1ga (42) wayof avoiding the stigma and taboos charac- describes the pervasive concern about gain- terizing overt suicide. The extent to which ing entrance into school and performing in the ISDB frustrates the hospital staff may school well enough to stay in it as primary also bring some feeling of power into anen- contributing factor in suicides of Japanese vironment in which the patients have, for the youth. Education is virtually the only means most part, lost control of their lives. for achieving security in a society that is high- In general, indirect self-destructive behavior ly status conscious. Preparation for the ex- as an indicator of suicide risk is behavior amination begins in early childhood for many Japanese. At least 80 percent of the children about which there is still far too little understood to function as a reliable clue. The dif- attend a neighborhood cram school for further preparation after school hours. ficulty seems to lie in its complexity in that different degrees of certain kinds of ISDB Shafii, et al. (6) report conflicting data :tout may serve different purposes in the same in- school performance from their studie.; of dividual, protecting the individual in some in- committed suicides versus attempted stances (as substitute behavior) or putting suicides. They found that poor academic him at greater risk (by becoming part of the performance or being a school drop-out did problem and making it even larger). Sub- not differentiate their suicide subjects from stance abuse seems to be at least a low level their control subjects. However, acting-out indicator of risk but will be useful only if ac- behavior resulting in disciplinary problems companied by other, more reliable in- and suspension from school did significantly dicators. differentiate the suicides from the controls. In contrast, Shafii and Shafii (22) found that SCHOOL PROBLEMS failure in academic performance was a marked characteristic of young suicide at- School plays an important role in the life of tempters, 15 and younger, when compared the young person, eventually occupying at with nonsuicidal psychiatric patients. least a third of the individual's day. As the second major social system in which children Pfeffer, et al. (43,44) found differences in and adolescents are involved, it adds its own suicidal behavior between inpatients and pressures and stresses to the traditional ones outpatients of latency age. As a result of a of family and home. A number of inves- higher incidence of multiple deficits in ego tigators of suicide have found school be- functioning in both the suicidal and non- haviors such as failures, discipline, and suicidal inpatients groups, at least three- truancy significant as risk factors for poten- fourths of each group tested below grade tial suicidal behavior. level. However, the suicidal children were much more worried about doing poorly in school than the nonsuicidal controls, 48 per- Academic Performance cent versus 19 percent. In contrast, among A number of investigators noted school the outpatients, both suicidals and controls failures and falling behind the appropriate worried about doing poorly in school, but grade among both completed and attempted showed no significant difference in the per- adolescent suicides. Seiden (10) found that centage functioning at thu appropriate although undergraduate suicides did much school grade level. better than their fellow classmates in terms of grade point average, 3.18 versus 2.50, The percentage performing poorly in school was noted as high (78%) in the study by

2-42 46 N.Farberow: Preparatosy & Prior Suicidal Behavior...

Barter, Swaback, and Todd (27); 58 percent tionally poor school records."fishier, Mc- by Hawton, et al. (20); and 35 to 38 percent Kenny and Christman-Morgan (46) felt that by Garfmkel and Golombek (37). Garfinkel deteriorating school marks and worty about and Golombek (37) found that more than failing school were typical of their adolescent half were experiencing failure in school or suicide attempters seen at a children's hospi- were drop-outs, but closer inspection indi- tal in Columbus, Ohio. cated that the severity of the attempt made a difference. Those attempts rated most Strong contradictory evidence is reported by Cohen-Sandler, Berman, and King (30) in severe were correlated most highly with suc- their well controlled study that divided sub- cess at school while minor or moderate jects into four developmental stages from ratings of severity of suicide attempts was birth to 15 years. Their control groups were more significantly associated with failure at depressed and nonsuicidal psychiatric school. They felt this corroborated results in children. They found, in contrast to the high a previous study of completed suicides in frequency of school adjustment problems which the suicides bad been associated with reported by other investigators, that school industrious and productive performances. refusal, poor concentration and poor school Stanley and Barter (28) found that distur- work failed to discriminate among the bance in school adjustment was an early and children.Indeed, school problems were frequent indicator of emotional disturbance reported generally for only 20 percent or in adolescence. However, they were not able fewer children in the entire sample. to demonstrate a significant difference in adequate school adjustment between their Disciplinary Problems experimental and control groups.In their Shaffer (11) found that it was primarily in this followup, they did find a difference with school behavioral area that children and patients who repeated suicide attempts after adolescents who committed suicide were ex- discharge having significantly poorer school pressing their emotional problems. The records than the control gimps. most frequent precipitant was a disciplinary Otto (17) noted school problems also in his crisis, occurring in 31 percent of his cases. extensive followup of suicide attempters. Five of his subjects had learned that a letter However, the levels of school problems he describing their anti-social behavior in school found for his experimental group, 7 percent, and their truancy was about to be sent to their was considerably lower than those reported parents. For had been in a Gght with another in the preceding studies. The nature of the child and two had been dropped from a school problems were primarily unsatisfac- school sporting team. Three bad presented tory school results, desire to quit school, school problems because of a dispute with problems relating to teachers and school one or the other parent. friends, and severe fear of examinations. Rohn, et al (45) found 35 percent of his Otto said that it is unusual for the school to suicide attempters were having behavior or be the direct cause of a suicide attempt. discipline problems. White (33) found ad- When an attempt does occur it is more like- verse school reports more likely to occur in ly to be in the higher grades where the the younger adolescents than in the older. demands may be too much for some children. He reported 67 percent were labeled as idle Marek, et al. (13) found school failures to be trouble-makers or a bad influence. When the primary motive for suicide deaths in their the subjects were in secondaty schools there study of juvenile suicides in Poland. were only 14 percent with similar adverse Among uncontrolled studies, Rohn, et al reports.Barter, Swaback, and Todd (27) (45) found that 75 percent of the teenagers noted disciplinary actions in 78 percent of hospitalized for a suicide attempt at the their suicide attempters. University of Maryland Hospital had excep-

4 7 2-43 Repod of the Secretiuys Task Force on Youth Suicide

Truancy or Drop-outs generally predictive of emotional distur- Shaffer (11) noted tniancy in 7 boys and 2 bance (which would include suicide) than primarily of suicide alone. However, school girls, and that 12 of the 21 boys had not been problems may be highly age-related, and so in school the day before their death. Three of these had been chronic school refusers less differentiating in the younger ages than in the older ages. While still an important while 5 had been absent for less than a week. ltvo had been sway from home and school clue, school problems rank more as moderately predictive, but dues that provide during the 24 hours preceding their deaths. Truancy was also noted as a problem by considerable additional supporting weight when other factors are present. Barter, Swaback, and Todd (27), Otto (17), and Garfinkel and Golombek (37). ANT1-SOCIAL BEHAVIOR Isolation and Withdrawal Anti-secial behaviors appear in many forms. Otto (17) noted a desire to quit school in 12 We have included in this category such be- percent of his subjects. Nilson (47) found a haviors as aggression, rage, hostility, number of school problems in her study of violence, delinquency, homicidal impulses, runaways.However, she felt that school fire-setting, stealing, and disobedience. problems and the process of running away References are primarily to behavior indiaz are interactive so it was useless to decide although at least one investigator, Mc- which was primary or to determine the Anarney (49) relates the tendency to nation- relationship of school problems to suicidal al attitudes. She sees Denmark, Sweden, and behavior in her runaway group. Japan as sulepreuing aggression and notes the higher rates of suicides occurring in their Wenz (48) conducted an interesting countries, and contrasn them with the Not- sociologi:-...1 study related to this problem by wegians who enjoy more freedom in express- looking for sociological correlates of aliena- ing anger and have a markedly lower rate of tion among 200 adolescents, aged 12 through suicide. 18, who were telephone callers to a crisis intervention center or suicide attempters Some of the most intensive work in evaluat- fruited in emergency medical facilities. He ing the role of aggression in suicidal behavior found significant negative correlations of has come from two sources, Pfeffer and her school performance with alienation in 20 colleagues (43,44,50,51) in New York, and ;ercent of the above-average students, 25 Cohen-Sandler, Berman, and King (30) in percent of the average students, and 56 per- Washington, D.C. Earlier studies from Pfef- cent of the below-average students.Like fer, et al. (43,44) on inpatient and outpatient Nilson, he assumes that school difficulties children of latency age compared four themselvas are not a primary factor in the groups: children who were suicidal recently, suicide attempt but rather problems that suicidal in the past, nonsuicidal recently, and have been present for a long time and are the nonsuicidal in the past. They found aggres- result of alienation. sion high in all groups, 90 percent for the suicidal groups and 80 percent for the non- In summary, almost all the researchers con- suicidal groups, a nonsignificant difference. firm the use of school problems and poor All thi children displayed severe aggression school performance as significant risk fac- that included fights, temper tantrums, a ten- tors. However, the evidence produced in the dency to hurt and teaie others, a destructive- well-controlled studies of Pfeffer, et al. ness toward objects, defiance, and (43,44) and Cohen-Sandler, Berman, and restlessness. Fire-setting and stealing were King (30) indicates that caution needs to be prevalent among both suicidal and non- used in evaluating the presence or absence of suicidal cSildren. When their desires were this factor, for school problems may be more not fulfilled, they displayed intense rage.

2-44 S ALFarberow: Prepatatoor 6 Prior Suicidal Behavior...

Often both suicidal and homicidal impulses which case they produce highly intense ag- were present, and to defend against the rage, gression. the children denied, projected and displaced In Washington, Cohen-Sandler, Berman, hostile feelings onto others. A simffar re- and King (30) noted especially the ap- search desirs applied to a group of out- pearance of aggression in their four progres- patient children who were latency-aged sive developmental life stases from infancy yielded exactly the same results (44). The to 15 years of age. They concluded that, as a sole specific indicator of potential suicidal result of experiencing a disproportionate behavior was an increase in hypermotor be- number of losses of all kinds, suicidal havior both within the inpatients and the out- children, in contrast to the experiences of patients. depressed and psychiatric controls, Pfeffer and her colleagues (50,51) then con- developed a greater loss of self-esteem and ducted two studies of children aged 6 to 12, an increase in aggrieved rage. Areview of with one group predominantly from a low so- their life events showed that suicidal children cial status and the second from a middle so- remained more intensely involved with fami- cial status. The subjects were divided into ly members and peers during childhood than four groups: those with suicidal tendencies the children in the control groups who were alone, those who showed assaultive tenden- more often separated from theirfamilies. cy alone, those with bothsuicidal and assaul- The frustrated investment of the suicidal tive tendencies, and those with neither children in these relationships was expressed suicidal nor assaultive tendencies. The as rage. Nearly two-thirds ofthe suicidal results shoved that the degree of expression children made homicidal threats, gestures of aggressive tendencies was a significant and, even attempts. predictor of group membership for each of The presence of aggressive, violent feelings the four groups of children. Aggression was in the young suicidal child was reported by greatest in the group classified as both assaul- Paulson, et al. (52) who found violence, in- tive and suicidal; lying, stealing, and truancy ternalized hate and anger, marked preoc- were present among both children who were cupation with fire-setting, and homicidal suicidal only and assaultive only. One impor- behavior toward family members and peers tant contribution to this study was the in 20 percent of suicidal children hospitalized delineation of two types of children: the first, for suicidal behavior. an amaultive and suicidal group whohad distinct ego deficits and exhibited rage episodes Aggressive, violent and anti-social behaviors and serious assaultive tendencies; the have been noted in varying proportions by a second, a group with relatively stable ego number of investigators of older adolescents. functioning who decompensated and were Among those with control subjects, Jacobs more liltely to become overtlydepressed (15) and Teicher (16) hypothesized that under extreme environmental stress. The rebelliousness of suicidal adolescents is one authors pointed out that from a theoretical phase they go through until they reach the point of view, the results did not deal direct- fmal suicidal stage. McIntire and Angle (18) ly with the question whether suicidal be- found a mean score for hostility significantly havior is an expression of inhibited higher for their aperimental group of self- aggression or aggression turned inward. poisoners than for control subjects. Tishler Rather, different groups of variables were and McKenry (46) found suicide attempters producing different patterns of behavior ex- between ages 12 and 18 had a significantly pressed as assaultiveness and suicidal be- higher score on the hostility scale of the Brief havior. These different sets can occur in Symptom Index when compared with a group isolation, producing groups of children who of nonattempters, and Otto (17) reported are assaultive alone orsuicidal alone. that one group of youngsters in his extensive However, they can also occur together, in followup of suicide attempters in Sweden

2-45 Report of the Secretary's Task Force on Youth Suicide could be characterized as showing increased overt aggression had the highest percentage irritability, aggressiveness, instability, and of suicidal thoughts, 16.5 percent Rosen- peevishness. Otto hypothesized that it was berg and Latimer (55) found evidence, in the depression that was hidden behind the anti- records of 77 suicidal adolescents at a State social activities such as vagrang, theft, and hospital, of patterns of sexual delinquency, truancy.Smith (53), comparing highly running away, truancy, and destructive be- suicidal teens with highly suicidal adults, haviors for the girls, and considerable acting- found that the high-risk teens were much out, running away, destructiveness, stealing, more overtly anvy, explosive, and less well anu dying authority for the boys. Alessi, et behaved than the adults. at. (56) found that 63 percent of juvenile Two of the three psychological autopsy delinquents considered seriously suicidal studies using controls, Shafii, et aL (6) and were committed for one or more violent Rich, Young, and Fowler (7) found that felonies and 31 percent had a history of as- anger, irritability, and outbursts charac- saultive in-program behavior. terized the subjects who committed suicide. Summarizing, anti-social behavior, espezial- Rich, Young, and Fowler classified 9 percent ly assaultiveness, when expressed along with of their under-30 cases as anti-social per- suicidal behavior, seems to be an earellent sonality compared with only 1 percent of the risk factor for suicidal potential. All of the group aged 30 and over. The under-30 aged researchers quoted noted its significance in group also experienced legal trouble more contrast to control groups.It seems likely significantly, p.001.Shafii, et aL(6) that anti-social behavior is more important as reported anti-social behavior including in- a clue in the younger ages, so long as it is ac- volvement with legal authorities, shoplifting, companied by suicidal expressions. It does fire-setting, fighting, school disciplinary not appear to be as significant in adult problems, drug selling, and prostitution sig- suicides, as indicated by Rich, Young, and nificantly more often in the suicidal group Fowler (7) and Maris (9). than in the control group, pic.003. Maris (9) reported that younger subjects who com- POOR IMPULSE CONTROL pleted suicide we.:e significantly more often AND ACTING-OUT motivated by revenge and likely to commit suicide based in anger and irritability than Impulsive and acting-out behavior are close- were older suicides. ly related to the and-social behaviors just discussed and are frequently included among In studies of committed suicide with no con- the list of various behaviors when anti-social trol groups, Shaffer (11) reported anti-social behaviors are identified by researchers. For symptoms in 73 percent of 30 children he example, Maris (9) found that younger studied. Crumley (26) found anti-social and children who completed suicide had more in- violent behavior a prominent feature in two tense deep feelings, more aggressive feelings, studies of suicide attempters from his private and higher levels of impulsivity and dissatis- practice, with inUnse rage appearing when faction with their life accomplishments than the teenager was disappointed when the per- older completed suicides. Shafii and Shafii son on whom he was leaning was unavailable. Hendin (54) related the increase in the (72) found the risk of suicide vots significantly increased when there waii impulsivity and suicide rate of young blacks in the United lack of regard for danger in the histories of States to their struggle to deal with conscious suicidal children am; adolescents. rage and murderous impulses. Impulse and acting-out behavior were both Among studies of attempted suicides with no noted by Crumley (26) in his review of his controls, Goldberg's (41) extensive private practice patients and by McIntire and epidemiological investigation of suicidal ideation found that persons who denied Angle (18) in their study of self-poisoners in

2-46 50 N.Farberow: Preparatory & Prior Suicidal Behavior... children and adolescents. The researchers in Academy Award-winning film, The Deer both studies considered the impulsivity and Hunter.Concomitantly, there has been the activity as part of the anti-social, hostile, widespread popularity of various fantasy and aggressive behavior that characterized games involving an array of violent activities, their populations. including instructions to the youngsters that they should try to become in real life the Among suicidal children of latency age, Pfef- characters that they are fulfilling in the garne. fer, et al. (43,44) found an inability to tolerate Now in its seventh year, the National Coali- frustration, delay actions, *Aerate deprivation on Television Violence (NCIV) has tion, or plan for the future. However, this been conducting lobbying and educational poor impulse control was present for both campaigns both among the public and the suicidal and nonsuicidal children. The re- government in an effort to control the searchers found an increase in hypermotor violence shown on television and the con- behavior, possibly providing an increased tinuing development of violent fantasy potential for dangerous acting-out be- games. They point out that prime time haviors. Finally, Pfeffer (57) noted in an ear- violence has increased over the recent years. lier study that an unrealistic, repetitive For the fall-winter season shows on the major acting-outc life-endangering and otn- television networks, ABC, NBC, and CBS in nipotent fantasies, such as being a superhero, 1980 to 1981 the average number of acs of was another important clue of high risk. The violence per hour was 5.6. By the winter of confusion and loss of reality from acting out 1985, the average had increased to 13.9 acts such fantasies were thought to arise from of violence per viewing hour. feelings of intense vulnerability and helpless- ma. To illustrate the process of imitation and the way in which violence on TV can stimulate In summary, poor impulse control and acting- similar acts of violence in young viewers, out behavior are often seen as part of the pat- NCTV gathered information on verified im- tern of behaviors included as anti-social. itations of the Russian roulette suicide that Imputivity appears to characterize suicidal was featured in The Hunter.Deer. The Coali- persons in the studies reviewed, although tion was able to find 41 shootings with 37 Pfeffer, et al. (43,44) found it did not dif- deaths. Most of the time the information ferentiate suicidal children from the non- about the event was attributed to newspaper suicidal children. In general, impulsivity and reports; in other instances it was attributed acting-out serve as good cr..ies especially to notification by a friend or the family. In when found in the context of anti-social be- each instance, the act was reported to have havior. It may be that, because of the lack of occurred within a short time of having seen preparation and planning, impute ity will the movie, or with specific identification of produce more suicidal events, but that they the action as an imitation of the mcwie scene are le s likely to be lethal. by friends or family. Of the 41 shootings, approximately half, or 20 of the victims were SUGGESTIBILITY below the age of 19. Involvement In Fantasy Games and Ncrv has also been concerned with toys requiring violence in fantasy, such as war games Imitation and fantasy games. A survey of 50 different In recent years increasing attention has been fantasy role-playing games found that all had focused tm the growing degree of violence a basic violent theme, such asmedieval sor- appearing in television programs, many of cery and combat, wild west battles, outer them screening at times when they can be space combat, post-nuclearholocaust watched by adolescents and children. Some- violence, urban thievery, espionage, assas- times the violence has involved suicidal be- sinations, samurai brutality, war violence, havior, as in the Russian roulette scene of the

2-47 5 1 Report of the Secretary's Task Force on Youth Suicide and others. Frequently the violence is in- games, such as war games and Dungeons and tense and gruesome. Dragons, become more and more engrossing One of the most popular and widely played and all-encompassing. games is the adolescent fantasy role-playing The games themselves become the reality. game Dungeons and Dragons (D&D). The Pfeffer (60) calls attention to this in her com- game has been claimed to be a major factor milts on ego functioning often seen in in causing at least 50 suicides and murders, suicina. children "As the pace, pitch, and according to the evidence accumulated by content of the child's play intensify, suicidal the NCTV and by a group calling itself children may lose the ability to discriminate Bothered About D&D (BADD), founded by between themselves and play objects. . the mother of one of its victims. The game Another aspect of ego boundary diffusion of requires attacks, assassinations, spying, theft, suicidal children is manifest when play ceases and poisonings. The players arm their and becomes a form of personal acting out characters with any of 62 different types of (p. 201)." weapons. An extreme number of monsters from horror and demonology are involved in- SOCIAL ISOLATION eluding 22 types of satanic demons and devils. Players can be cursed with 20 different types Social isolation appears in many forms in- of insanity. D&D is played in groups under cluding withdrawal, alienation, asocial be- the direction of a Dungeon Master, primari- havior, poor or inadequate peer ly by males, age 12 to 20, and there are an es- relationships, and similar behaviors.It is timated 3 to 4 million players in the United usually considered fairly often to be the States. The game can last for months and hallmark of a depressive syndrome. even years with the goal being not to get However, social isolation may appear in killed and to accumulate as much power as suicidal persons even when depression is not possible. Power is earned by the murder of opponents. Players can be good or evil and In their comprehensive description of the are encouraged to identify with their own development of the suicidal state, Jacobs character in the game. In the gimp of re- (15) and Teicher (16) identify one strge as a lated deaths, 22 or 44 percent of the suicides process of progressive social isolation from or murders involved voungsters aged 19 and meaningful social relationships. The fffst under. stage of exacerbated problems is followed by All the evidence about the game is primarily a period of dissolution and disappearance of anecdotal. The publishers of NCIV quote, meaningful social relations. The feelings of without sources, various authorities on isolation that follow then become both the whose work they rely for their evidence of the problem and the barrier that prevents resolu- harm from violent, fantasy role-playing tion of it.Withdrawal into self appears as games, violent toys and play, and the desen- gloominess, silence, and social alienation and sitization effect of continued exposure to the physical withdrawal frequently takes the role-playing violence. form of running away from home. Some- times the adolescent will try to reestablish a Although specific research is lacking on the relationship with a significant other in a relationship between magical thinking/fan- romance. About 36 percent of the suicide at- tasy and suicide, a hypothesis might be drawn tempters were found to be engaged in a from the results of the study by Gould (58). serious romance that was failing or terminat- Magical thinking implies a loss of ability to ing. iecognize the boundary between fantasy and reality. Thus, for young people or children Both Seiden (10) and Hendin (61) found in who have either lost or not developed the their studies of suicidal college students that ability to maintain the boundary, fantasy they were often asocial, withdrawn, shy,

2-48 52 N.Farberow: Preparatory & Prior Suicidal Behavior... friendless, and alienated from all but the found the variable with the strongest nega- most minimal social interactions. Often, they tive association was social contact with peers, almost totally absorbed themselves in their indicating the less the relational involvement school work, which may then have served as with peers, the greater the feelings of social a protective device. Hendin hypothesized isolation. Wenz concludes that alienation the psychodynamics as one in which the stu- sets in motion a train of events that leads to dents saw their relationship with their attempted suicide. As a result of alienation parents as dependent on their emotional, if there is an atrophy of interpersonal relations. not physical, death. The depression and The greater the communication blockage, isolation that developed was actually a form the greater the adolescent's sense of isolation of protection that sometimes shielded the in- and the more the person withdraus from fur- dividual and even ma& suicide unnecessary. ther contact with othets. The process is cir- cular and leads to ever-greater degrees of The adolescent suicide attempters treated by isolation and withdrawal. His reasoning is Shafii and Shafii (22) and Rohn, et al. (45) at strikingly similar to that of Jacobs (15). their university hospitals and the self- poisoners seen by McIntire and Angle (18) at Summarizing, controlled studies indicate their poison control center were charac- that social isolation consistently identifies terized as loners, preoccupied with self, in- the suicidal person when comparisons with volved in failed romantic relationships, nonsuicidal persons are available. It thus ap- withdrawn, and isolated. Rohn, et at (45) pears to be a good clue, but not in and ofit- found the social aspects were characteristic self; social isolation appeals in depressed of the boys in his study. However, Marks and persons who are not suicidal and in non- Haller (40), found that their suicidal girls depressed nonsuicidal persons.However, were more likely to have few or no friends when social isolation appears with suicidal during their dildhood, to be socially isolated, expression, it emerges as a significant high and to be ynable to talk about their personal risk predictor. problems with anyone. In two followup controlled studies, Stanley RUNNING AWAY and Barter (28) in Colorado found that Runaways are inherently a suicide risk group young people who went on to make further inasmuch as the behavior is associated with attempts after their hospital discharge were emotional disturbance, family conflicts and much less likely to have adequate peer strife, school and learning problems, social relationships, less likely to be living with their alienation, and other negative conditions, all parents, and to have more social agency con- of which are associated with suicidal risk. tact than those who did not make further Nilson's (47) study of 18 runaways compared suicide attempts.Otto (17) found social them with other court referrals who were not isolation much more prominent in his suicide runaways. Half the runaway group, 14, made attempters than in his nonsuicidal controls. suicide attempts or gestures and 12 showed He hypothesized that behind much of the suicidal ideation only. In the nonrunaway anti-social behavior may be found loneliness control group, only 2 of the 18 made an at- and isolation, listlessness, contact difficulties, tempt or gesture, while 11 showed signs or feelings of guilt, emptiness, and apathy. ideations. The difference between the two Wenz's (48) interesting sociological ap- groups is significant, p<.02. proach hypothesized alienation as a highly The prmence of a history of having been a significant factor among adolescent suicide runaway is noted by Rosenberg and Latimer attempters.Applying Dean's Measure of (55) and White (33) in their research on at- Alienation to adolescents who called a tempted suicides among adolescents, and suicide prevention center and adolescent at- Shaffer (11) also notes the presence of tempters treated at emergency rooms, he

2-49 3 Repott of the Secretary's Task Force on Youth Suicide runaways in the histories of younger adoles- The picture of the suicide note-writer ob- cents wha completed suicides. tained was impressively similar to that descnIed in clinical face-to-face contact, but Four well controlled studies sharply diverged on the usefulness of a history of running with the added special features of the suicide logic and style of thinking. away. Jacobs (15) considers it the third stage in his theory of the development of the Tripodes used a comprehensive approach, suicidal state, following the second stage cif covering all aspects of style of thinking. withdrawaL Marks and Haller (40) find the Neuringer and his colleagues (64) focused on presence of running away a significant dif- dichotomous thinking or the tendency to ferentiator between suicidal boys and emo- polarize thought in an extreme manner. tionally disturbed nonsuicidal boys, but not a Using the semantic differential with groups differentiating factor between the girls. The of suicide attempters, psychosomatic strongest evidence against running away as a patients and normal patients, Neuringer (65) distinguishing feature comes from hestudies found in two separate studies that suicide at- by Pfeffer, et aL (43,44) and Cohen-Sandler, tempters were overwhelmingly more Berman, and King (30).Their carefully dichotomous in their thinking on the activity designed studies find that running away is and potency factor scales than were the other more a symptom of emotional disturbance two groups of subjects. While polezation of than necessarily of suicidal risk. values was apparently a common characteristic of individuals in psychological stress, the In summary, the usefulness of running away as a significant risk factor, while seemingly dichotomimtion of the activity and potency positive, has been thrown into doubt by the factors seemed to be an exclusive hallmark of results. of Pfeffer's and Cohen-Sandler, et suicidal thinking. The results were even aL's studies.It is true, howev-tr, that the more extreme when only highly serious population of their studies are aung, rang- suicide attempters were compared with the ing in age from 5 to 14 at the most. It ms., be psychosomatic and normal patients. Neuringer and Lettieri (66) extended the that the usefulness of this factor changes with earlier studies by taking daily measures of older adolescents- At this point, the value of a history of running away as a suicide risk in- dichotomous thinking from high-risk, medium-risk, and zero-risk individuals over a dicator is useful, but not necessarily predictive unless accompanied by other three-week period following a suicidal crisis. self-destructive clues. The results were the same but the investigators were even more alarmed that the extreme dichotomous thinking did not seem to WINKING PATTERNS AND diminish over time. They raised the concern STYLES that the process of dichotomous thinking as- Shneidman (62) has pioneered in the study sociated with suicidal action might always be of suicidal thinking. His approach has been present, ready continuously to distort the an analysis of the logical substructures of syn- world in such a way that suicide is an ever- tax in the writings of suicidal people, primari- present possibility. ly through the comparison of genuine suicide Levenson and Neuringer (67,68) conducted notes with simulated suicide notes. Shneid- a number of studies on rigidity and constric- man stimulated a number of other re- tion of thinking in the suicidal individual and searchers. Using the matched pairs of real found that person to be more socially rigid and simulated suicide notes, Tiipodes (63) and inflexible, to show an inability to shift in applied 24 aspects of reasoning pauerns, in- problemsolving strategies (as did Lineham, cluding the cognitive maneuvers, and the et aL 69), to be consistently high in field de- idio-, contra-, and psycho-logic categories of pendency, and to have difficulty solving arith- analysis to the thinking of suicidal people. metic problems.

2-50 N.Farberm Preparatory & Prior Svicidal Behavior...

Summarizing, Neuringer (64) stated that notes in Los Angeles County from people suicidal individuals seemed to have difficulty ranging in ages from 13 through 90, with the in utilizing and relying on internal, imagina- median age in the 50s. Suicide notes come tive resources; to polarize their value sys- primarily from completed suicides; they are tems; and to tend to be rigid and constricted rarely obtained from attempted suicides. In in their thinldng. The process was the result a statistical comparison of attempted and of emotional stress affecting the cognitive committed suicides, Shneidman and Far- structures, and decreasing cognitive capacity. berow (73) found suicide notes in 36 percent of the committed suicides and only 1 percent Kaplan and Pokorny (70) tested the of the attempted suicides. hypothesis that the level of self-derogation in the thinking of individuals might predict sub- Shneidman (71) has held three different sequent suieslal behavior. In a longitudinal positions on the relationship of suicide notes study of junior high school students in Hous- to suicidal phenomena since he began their ton, they administered a questionnaire three study. At first, the notes would offer special times at annual intervals, and obtained opportunitia for observing the thinking and measures of self-derogation over periods of feeling that went into the act; second, that it time. By correlating them with the number is understandable that the views into the of self-destructive behaviors appearing be- thinking and feeling were so limked con- twt2n the measures, the investigators considering the stress under which the notes firmed their hypothesis that high level of were written; and third, that the notes' value self-derogation woe positively associated was increased measurably when seen in the with suicidal thoughts, threats, and suicide at- context of the history of the suicidal person. tempts. In such instances the note then 'illuminates many aspects of the life history.* In summary, style of thinking, while apparently useful and intriguing, is limited in its Stmeidman (74) reviewed all the studies that predictive value for suicide because few have been conducted on suicide notes up to professionals have the specialized 1976 and reported that research has been knowledge of logic and language to apply the carried out on the logic of suicide, changes in suggested analysis. Dichotomous and self- suicidaldynamics overage,the derogatory thinking seem like positive risk socioeconomic and psychological variables factors, especially after suicidal behavior ap- of suicide, suicidal life space, the emotional pears.Such modes of thinking serve as content, the effects of motivational level on evidence of increasing constrictim and language, comparisons with ordinary letters rigidity, boding ill for an individual's ability to to friends and relatives, language characteris- see alternative possible courses of action. tics, relations to persons and computer count of key tag words. SUICIDE NOTES Shneidman summarized by saying that, as a Shneidman (71) is also the person who dis- whole, the studies have indicated tha; it was covered and initiated the use of suicide notes possible to distinguish between genuine and and personal documents as a source of valu- simulated suicide notes, and that the genuine able insights, not only into the thinking and suicide notes were primarily characterized by reasoning styles of suicidal persons, but also dichotomous logic, and greater degree of hostility and self-blame, more use of specific as a means of determining his personality and emotional state immediately preceding the instructions to the sPrvivors, less evidence of thinking about how tine is thinking, and much suicide. evidence of the variety of meanings at- Researchers in this area have indicated that tributed to the word love. The content of the suicide notes do not appear until about age notes frequently reflect unrequited love, in- 15. Shneidman and Farberow (72) collected tellectual sell-assertion, shame and guilt re- r Jr 2-51 Report of the Secretary's Task Force on Youth Suicide lated to disgrace, the wish to escape from the CONCLUSIONS AND pain of insanity, the wish to spare loved ones SUMMARY from further anguish, and a sense of inner pride and autonomy connected to one's own A number of summary conclusions can be fate in the manner of one's own death. Often made about behavioral clues as risk factors. the writing is directed to his survivors-to-be 1.Prior suicidal behavior of any kind is un- as though he were going to be alive to super- questionably as strong a risk factor in- vise his wishes. dicator for adoleatents as it is for adults. It Peck (75) conducted a content analysis of is consistently substantiated in both suicide notes of suicide victims below the age retrospective and prospective studies, in of 35 (153% below age 25) in a large Mid- both well-controlled and no-control west city. He found that elements of fatalism studies, and identifies from 72 percent to were most prevalent in victims under the age 71 percent of completed suicides and from of 20, that they decreased for each of the sub- 16 percent to 38 percent a attempted sequent older age groups, and that they were suicides retrospm-tively, and from 3 per- more frequent among single persons. The cent to 10 percent of completed suicides author concludes that the young fit a fatalis- and 10 percent to 50 percent of attempted tic model of suicide, which is docribed as a suicides prospectively. condition of excessive constraint or regula- 2. Indirect self-destructive behaviois seem to tion that may trigger a reaction when the in- play too complex a role in the personality dividual moves from excessive regulation to of the individual to serve as reliable risk in- a state of alienation and a break in social ties. dicators. For some individuals, such be- This almost seems a definition of adoles- havior may play a protective role against cence. suidial behavior, whereas for other per- Edland and Duncan (76) categorized the sons, the same behavior can become a sig- notes written by about 23 percent of the com- nificant self-destructive behavior on its mitted suicides hi Monroe County, New own. Much more needs to be learned of York, into a system that focused on the the role of indirect self-destructive be- psychodynamics and attitudes toward death. haviors in the personality of the individual The authors felt that indications of such before they can achieve a consistent value thoughts and concerns in individuals, while as a risk factor. they were alive, should serve as significant 3. Suicide attempters and committers among predictive clues for high-risk potential adolescents still seem most appropriately suicide. viewed as separate but overlapping In summary, suicide notes have yielded much populations. information about suicidal persons, informa- 4. While prior suicide attempts, threats, and tion that has been useful iii identifying risk ideations may be excellent clues for fur- factors, such as dichotomous thinking, con- ther suicidal behavior, it is also true that striction, ambivalence,, and age-related they are late clues in the progression dynamics. Further exploration will undoub- toward suicide, indicating that a suicidal tedly yield more. As predictive clues, state has already occurred.Primary however, the usefulness of suicide notes prevention objectives encourage the iden- seems limited, for they are rarely dimovered tification of preparatoly suicidal risk be- until after the fact of a suicide and so serve haviors at ealer states that may prevent more to indicate what did happen rather then the development of an overt suicidal state. what might. 5.School problems have been identified consistently in histories of children and adolescents who have either committed or

5 i; 2-52 N.Farberow: Preparatoty & Prior Suicidal Behavior...

attempted suicide.These problems in- 2. lihnsidman ES, Farberow NL Clues to suicide. clude deteriorating academic accomplish- New York McGraw PO, 1957. 3. Shnekimen ES, Farberow Pa.:Statistical COM- ments, disciplinary problems, and truancy. =between attempied and con'tpiated suicides. In: NL, Simeldmen ES, eds. The Cry for HO. New However, they also identify general emo- York: Mcaraw HIN, 1951. tional disturbance and need other suicidal 4. Farberow NI., Shneldmen ES, Neuringer Case clues to be most useful as risk factors. history and hospiteliution factors In suloide of whist*,hospital patients. J hierv Ment Die net11u1.4arZ 6. Anti-social behavior, especially assaultive- 5.DorpmtTt, Noisy HS: The relationeNo behveen at- ness and thizatening people, and includ- suicide and committed suicide. Camp Psychlat ing rage and hostility, are highly useful 179W,E74-79. clues of suicide potential, especially 5. She'll PA, Cartigen 5. WhittingNII JR Derrick it Psychological autopsy of completed suicide in children among children and younger adolescents. and adolsecents. Am J Psychiatry 1985:142:1901-1954. 7. Rich CL, Young 0, Fowler RC Sen Diego skid* 7.Social isolation and impulsivity are also study: I. Yowv vs. old oases. Department of Univerek of Wfankr San Diego School of often noted. However, these behaviors 1954. (Unpublished observation). are found to characterize the general S. Marie it Pathways to suicide. Baltimore: Johns psychiatric population just as frequently as Hopkins Press, 1991. they do the suicidal group. These factors 9. Will Ft The adolescent suicide problem. Suicide & Lite Threatening Behavior 1991;15:9t-109. appear to be of greater value when they 10. &olden RH. Campus barratry. of student occur with other evidence of suicide suicide. J Abnorrn Psychol 1911(14 1: potential and help to substantiate the 11. Shaffer 0: Suicide in childhood and forty adoles- evaluation. cence. J Child Psycho! & Psychlat 104;15:375-291. 12. Cosand 5.1, Bourque lit Kraus": Suicide among adolescents in Sacramento County, 1950-1979. Adoles- 8.Suggestibility and imitation probably play cence 190217:917-930 a role in suicide. However, this has been 13. Men* Z, Vadoold J, Arrarysiewlez W: Suicides shown primarily in large-scale group and committed by mimes. RNIM$10 Sciences 1975.7103408. social reactions. The role suggesaility 14. GerlInkel BD, Chamberlin C, Golombek H: Com- Vnallotagar in Ontario youth. In: Proceedings of the 10th and imitation play, especially in violent Congress fat Suicide Prevention er. Crisis role-fantasy games and television Intervention. Ceram Cando: IASP, 1979125-131. 15. Jacobs J:?dehiscent suicide. New York: New violence, in stimulating suicidal behavior Yodr-Wley-interezienoe, 1971. in the individual is unclear. 10. Teich*. JD:Children and adolescents who attempt subide. Pullet OM North Am 19701711874911. 9. Suicide notes can be highly useful in 17. Otto LI: Suicidal eats by children and adolescents. delineating the suicidal person's style and Acta Psychlat Scandinav 197k232:7-123. pattdrn of thinking. Dichotomous think- 15. Mckitke MS, Angle Clt Psychological "blopay In selflaboMig of children and adolescents. Am J Die POd ing, especially, may save as a useful clue 197a 120. in reflecting increasing rigidity and loss of 19. MoKenry PC, Tlehler CL, Kelley C: The role of drugs in adolescent suicide attempts. Suicide & Life ability to seek alternatives as the person Threatening Behavior 199313:109-175. becomes more and more suicidal. 20. liewton K. O'Grady J, Osborn M, Cole D: Adoles- cents who take overdoses: Their characteristics, problems 10. Running away might also be considered and contacts with helping agencies. Br J Psychist another kind of indirect self-destructive 190214011(1-123. 21. Clarkin JF, Raman C. Hurt SW, Corn ft Aronoff behavior, possibly substitutive or protec- M: Affective and character pethology of suicidal adolescent and young adult Inpatients. J Clin Psychiat tive, possibly overtly self-harmful, probab- 10E14;45: tea. ly both.Because running away and 22. Mail PA, Shaft, SL Pathways of human devebp- suicidal behavior are both highly as- meat New Yob Thieme end Stratton, 193Z154-109. sociated with family conflict it is highly 23. Diskette RFW:Adolescent suicidal behavior: But/ding blocks for asocial teaming theory. Om Pfeffer Cr, likely they will be found together. Richman J, eds. Proceeding of the illth annual monSto of the Merkel, Association of Suicidology. New Yoh *My: AAS, 19103031. 24. Sdineer HI, Pedecein A, Brozovsky At Hospitalized REFERENCES suicidal adolescents: TWo generations. I/ Am Aced Child payhial 1975;142611-930.

I. Farberow NL Shneiciman ES: Ahem ,pted 25. Gebrielson 1W, Merman 1.1f, Currie Ja, Tyler NC, threatened and completed suicide. J Abnorm Sac Psycho! Joke) JP: Suicide attempts in a population pregnant as 11155;50X230. teenagers. J Am Pub Health 197M221119-2301. Report of the Secretary's Task Force on Youth Suicide

26. Crumley FE Adolescent suicide attempt. JAMA 51.Pfeffer CR Plutchik R. Wzruchi MS. Suicidal and 19792142404-2407. assaultive behavior in children: olassifeciation, measurement and interrelations. Am J Psychlat 198314t154-157. 27. Barter JT, Seebeck DW, Todd D.Adolescent suicide attempts: A followup study of hospitalized 52. Paulson MJ, Stone 0, Spat° it Suicide polential patients. Arch Gen Psychialty 1961119523-647. and behavior In children egg _4-12. Suicide & Life Threatening Behavior 19M:225-24a 28. ftift Si, Barter JT:Adoissoent suicidal behavior. kn J OtThweyd*d 197040:8748. 53. Smith it Family end individual oharacieriMice of suicidal adolescents anti adults. In: %tattooer C, Smith K. 29. Cohen-Sandler R, Berman AL: A followup study of eds. Proceedings of the tath annual meeting of the suicidal children. Proceedings of the 140 an- American ,IWIMon of Suicidology. Dallas, Texas: Al4S, =irreezadtingof the Medan Association of Suicidology, 19B3:58. 1981:4244. 54. Hendin it Youth suicide: A psychosocial person- 30. 03hert-Sandier R. Berman AL, King RA; Lilo stress PIPPAr ofosirded at Symposium on Adolescent and symptomatology: Determinants of suicidal behavior Suicide: Understanding end Respondky. Los Angeles, in children. J Am kad Child Psychist 1M21:178-188. 1985. (Unplialished observations). 31. Reuenhotet JfA: Followup of yotmg %women who 55. Rosenberg PK Latimer R: Suicide attempts by attempt suicide. Dis Nstv Sys 197t 3a792-797. children. Mental Hyg 196145036459 32. Motto Jk Suicide in male adoissoents. In: Sudak 55. Alessi N, MoMenue M, Brinkman A, Gremlin* k HS, Ford AB, Rushforth NB, eds. Suit:10ln the young. Boo- Suicidal behavior among variously/mile offenders. Am .1 tort: John WrighVPSO, 1984:227.244- Psychist 1984;141* 33. Whits IC Self-poisoning in adolescence. & J 57. Pfeffer Cf.t Oink* observations al ;day ofhowl- Psychtet 1974:1242445. taIlzId suicidal children. Suicide & Life Thrtning Be- 34. Duritheim E: Suicide. Glencoe, Illinois: Free Press, havior 1979;9136.244. 1951. 58. Gould RE:Suicide problems In children and 35. Idenringer K Man against himself. New York: Har- adolescents. kn J Psychother 196519:225246. court. Brook 1938. 59. Sebastian* ScieF, Rieder C. Bedt SE The or- 38. Farberow NL The many faces of suicide. New ganization of fantasied movement in suicidal ohildren and York: McGraw Hill, 1080. adolescents, in: Pfeffer CR, Richman,J eds. Proceedings of the 15th annual mesev of the Arnodcan Association of 37. Garfinkel BD, Goiombek H: Suicide behavior In Suicidology. New York MS. 1982:2546. adolescence. In: Goiombek H, Garfinkel BD, eds. The adolescent and mood disturbmioe. New York: internation- 89 Pfeffer, CFt The suicidal child. New York: Guilford al Universities Press, 1983:189417. Press. 1961 38. Stengel E. Cook N: Attempted suicide. London: 81. Hendin It Growing up dead: Student suicide. Am Oxford Univerefty Prose, 1958. J Psychother 1975;29327-338. 39. Kreitman N: Paresuicide. London: John Miley & 62. Shneldman ES:Psydmiftia: A personality ep- Son, 1977. ic patterns of thinking. kr Waput J, Lesser 0, eds. Issues in thematic epperoeption methods. 40. Maim PA, Heiler OL Now I lay me down fort= Springfield, unclE Charles C. Thomas, 1961. A study of adolescent suicide attempts. J Clin 1977:33:309.400. 63. Tdpodes P: In suicide notes. In: ShneldmanESRiescr,_ 00=04:Contemporary 41. Goldberg E: Depression and suicide ideation in Developments. New Mk Grune & Stratton, 1976. the young adult. Am J Psychlat 19111;138:38-40. 64. NeudrOir Current develop_ments in the study 42. lomat Suicide of Japanese Ifouth. Suicide & Life of suicidal Minting. In: Shneidman ES, ad. Suicidology: Threatening Behavior 19et:11:17-30. toOrtfatraly Developments. Now York: Grunt, & Strat- 43. Pfeffer (SR, Con* HA, Ruing* R, Jorrett I: Suicidal behinior In latency aged children: An empirical study. J 65. Neurinw C: Dichotomous evaluations in suicidal Am Child Psychlat 1979;18:679492. Individuals. J eftmM Psycho' 1981;25445-449. 44. Pfeffer CR, Conte HR Pfutchik R, Jerrett I: Suicidal 65. *winger C, Wiled DJ: nftkin, attitude and behavior in latency_ wed chNren: An outatatient popula- affect in suicidal individuals. Life T.atenlng Behavior tion. J Am Child Psychlat 198Ct 1970341 1971;1:108-124. 45. Rolm RD, Sades f44. Kenney TJ Reynolds BJ, 67. Levenson M, Neuringer C: Problem solving be- Heald FP: Adolescents who attempt suicide. J Pedlar havior in suicidal adolescents. J Consult Ctin Psythol 1977838438. 1971:37433-438. 48. Tishier C, McKenry P, Christman-Me an K.: 68. Levenson PA, Neuringer C: Suicide and field de- Adamant suicide ~10,: Some significant pendency. Omega 1974:5:181-188. Suicide & Life Threatening Behavior 1981;11:0892. Linehan MM. Franey A, Graham BJ, Chiles JA, 47. Nilson P:Psychological profiles of runawq Sadin P, Nielson SL interpersonal response styles of children and adolesoents. In: Wells CFI Stuart IR. eds. paresulcides. in: Proosedings of the 14th annuli meeting Self-tiestructive behavior in children and adolescents. of the knerican keociation of Suicidology. Abuquerque, New York: Van Nostrand, Reinhold. 1981:939 NM: AAS. 1981:28-30. 48. Wertz FW:Sociological correlates of alienation 70. Kaplan fa Pokorny AD: The solf-cierogation and among adolescent suicide attempts. Adolescence suicide - 1: Self-derogatior. as an antecedent of suicidal 1979;14:1940. responses. Social Science & Medioine 1978:M13-11a 49. McAnarney ER:Adolesosnt and young adult 71. Shneidman ES. Voices of death. New York: Harper stickle In the Lktited States-M*1061W oils:101sta! unrest? and Row, 1080. Adolescence 197%14:785-774. 72. Shneiciman ES, Farberow Soolopsychological 50. Pfeffer OR, Solomon 0 Mulch& R, favuohl MS. of suicide. In: David HP, B engelmann JC, IhWner k Suicidal behavior in Iatericy.age psychiatric In- eds.erep.ctives In personality research. New York: =de: A and case J AmAced Springer. 96047049a Psychial190t21:56441119.

5 g 2-54 N.Farberow: Preparatoty & Prior Suicidal Behavior...

73. Shne law ES, Reberost NL Stetaked ram- =between etternprad and committed euloklee. NE, Shneldmen ES, We. The ay for help. New Vet I"- lam MN, 10011047. 74. Atneldmon Et Wade notes reconsidered. h: Shneldmen ES, ed. Suladology: Contemporary Develop. =ob. New York Grano & &Mon. 1976.. 75. Peak IX: Taw* a thew of Waldo: A oese for modem fetellent. Omega 180041;11144. 70. Mend F Duncan CE: Suicide notes ki Moon* ftOr A twoottihroosiatelook (19304974 Jot Form- al° Moms 1073;1

2-55 SOCIAL AND CUL1URAL RISK FACTORS FOR YOUTH SUICIDE

Carol L. Huffute, Ph.D., Director of Research, California School of Professional Psychology, Bakeley, California

INTRODUCTION Our attention to suicide among young people Japanese dropped between e1960s and has been drawn by the extraordinary number 1970s, it has been rising again (2,6). In other of adolescents and young adults who killed Near Eastern and Asian countries, suicide themselves during the decade of the 1970s. rates generally have been low but increasing Although there is evidence that the rate of slowly and steadily over time (7). Exceptions suicide among youth is declining, it is still very are Hong Kong, where there is considerable high, and, more importantly, we are left feel- variation year by year (8), and Taiwan, where ing powerless to explain the unanticipated in- the suicide rate has dropped, from its mid- crease or to prevent such a social tragedy 1960s high, to the low point in about 30 years from recurring. In the following pages, 1 will (9). With the exception of countries such as briefly review the pattern of youth suicide in Japan, Thailand, and Sri Lanka, which have the United States over the past 25 yeas and, U-shaped curves, suicide rates in Asian where possible, compare United States rates countries tend to be high among adolescents and patterns with those of other countries. and young adults, and to decrease with age. Following that, I will summarize Emile Age-related patterns of suicide tend to be the Durkheim's explanation of suicide, which I same for both sexes in those Asian countries will then use to organize the literature on so- from which data are available. Differences in cial and cultural risk factors andio assess thz rates between the sexes tend to be much usefulness of explanatory models. smaller than in the United States. There is, From 1960, the rate of suicide among 15 to 19 in fact, almost parity in a number of year olds and 20 to 24 year olds increased countries, and in Taiwan, Singapore, and steadily until the late 1970s, when some Thailand, rates of suicide among female decrease could be seen in the older group. adolescents exceed those of male adoles- This pattern is most apparent among men, cents. but is also evident among women (1,2). The We have come to recognize the United United States tends to be about mid-way States and Canada suicide rate increases as, among the countries from whom data are at least in part, cohort phenomena. That is, available with regard to suicide rates (3), and the suicide rates for successive birth cohorts it appears that only Canada (4) and Australia are higher, at each age, than the rates for (5) have experienced a pattern of increase preceding cohorts (4,5,10,11). This means that similar to that of the United Stata. the jump in rates among adolescents and Although the suicide rate among young young adults in the 1970s should be 'echoed" in future suicide rate increases among the

69 2-56 4.

C.Huffine: Social and Cultural Risk Fac!xs.. middle aged and then among the elderly. as vatying in the degree to which they in- Recognizing the phenomenon as a cohort ef- crease or decrease the probability of group fect is important because of the predictability members killing themselves.I will touch it provides, but also because it refocus= our upon three general ways in which societies search for explanation. Rather than look might influence the suicide rate among their only at what was, or is, going on at the time members; these might be considered sets of of the observation (i.e., in the era of the risk factors. 197(15), we seek a more pervasive factor that can affect the risk for suicide of successive 1. Social structures might support and waves of babies. And if, as seems to be the protect individuals, or they might produce case here, there is an extrPordinary jump, stress. what might we say about the life experiences 2. The culture, as well as social structure, of of this particular birth cohort that can help us a social group influences the psychological explain their amplification of the general development of its members. trend? 3. The culture produces, through such The suicide rates that have captured our at- mechanisms as folklore, attitudes about tention are those of the lutby boomets," per- suicide and death. sons born during the 195(15. This group of individuals stands out, not only because of its Most of the work in suicide has focused on absolute number, but also because it was the lust set of influences, and the review that preccded and followed by excepti7aally small follows reflects this emphasis. birth cohorts, the depression and the "baby bust" cohorts. In Europe, fertility increased BACKGROUND for a few years following World War II, but only Canada, Australia, and New Zealand ex- In looking at the literature produced in the perienced fertility trends similar to that of the past fifteen years or so, I have been im- United States (12). pressed with the continuing importance of the work Emile Durkheim (13) did so many My search for risk factors was guided initial- years ago, and I have been reimpressed with ly by the findings descnlied above: that in the the genius of that work. Much of what is said United States, successive birth cohorts have by contemporary writers about today's youth had higher suicide rates than antecedent and their suicide rates is a restatement, ones, that the increase in rate of suicide at generally in more specific terms, of some part ages 15 to 19 and 20 to 24 among persons born or another of the model developed by in the 195015 was especially sharp, and that the Durkheim. Durkheim described four basic only countries with similar patterns of youth *types" of suicide (anomie, egoistic, fatalistic, suicide over time are also among the few and altruistic), which represent extremes of countries demonstrothg a post World-War two processes he consideied essential to a II fertility pattern similar to that of the healthy social order integration and regula- United States.Subsequently, the search tion of its members. widened in response to (1) the discovery of other phenomena calling for discussion, and Social integration refers to the degree to (2) the wish to use a theoretical frame for the which members identify with the group. It is analysis that would ghe it coherence. a structural condition that binds persons together as members of a collective bodywith Sociocultural theorists do not presume to ex- which they identify and that provides part of plain individual instances of suicide. Rather, their identity. Too little or too much integra- they focus on rates and patterns of rates as tion, according to Durkheim, constitutes a phenomena in and of themselves. We tend suicide risk. Too little integration results in to see the social group as the context within individuals feeling isolated, set apart from, or which an individual acts, and to see contexts

f; I 2-57 Report of the Secretary's Task Force on Youth Suicide different from, their fellows. This state is tural risk factors for youth suicide using referred to as *egoistic* and is characterized Durkheim's model as a general frame for the by excessive individeation. In some form or discussion. another, this part of Durkheim's model has been the most widely used by contemporary REVIEW analysts of suicide rates.Its relative popularity may be attributable to the relative Egoism/Social Integration ease with which the concept can be opera- Without even looking at data, one might ex- tionalized and linked to psychological con- pect relatively high rates of suicide among cepts. adolescents and young adults in contem- Asocial group may be too integrated, as well porary societies because they are in stages of as not sufficiently so, in which case its mem- life characterized by identity change and con- bers are at risk of altruistic suicide. Here the solidation. Konopka (17) argues that adoles- problem is one of insufficient individuation; cence is a period 9f high risk because the the person may be said to have no identity individual undertakes a series of *firsts,* in- other than the social group or that which is cluding questioning childhood precepts that conferred by his/her social status. In such in- were provided by parents and which guided stances, individuals may be called upon to his/her life for so many years, reassessing old sacrifice themselves in the name of the group values, and evaluating potential new one& or to preserve it.Although some writers The adolescent needs to move beyond a very have dismissed altruistic suicide as irrelevant small interpersonal circle to find warm sup- for modern analyses (14,15,16), it is important, portive relationships, and to cope with the I suggest, for reasons I shall develop later. yearning for the warmth and support of family ties. Maris makes the point that adoles- The other dimension of importance to cence is a 'time marked by marginality, Durkheim's theory is that of regulation, the confusion, and ambiguity* (18, page 100). In social condition that *restrains the passions* suggesting that the greatest problem young of the individual.Current theorists have people have today is their uselessness, he tended to interpret social regulation in te? ms reminds us of Sabbath's (19) defmition of of norms. The idea is that individual aspira- suicidal adolescents as expendable children. tions, and the methods by which those aspira- In short, adolescence and young adulthood tions may be achieved, are defined by social are, in Western civilization, characterized by norms. A state of anomie is said to exist when loose and uncertain social integration and individuals do not have a set of clearly identity. defined goals, or when there is a disparity between goals and the possibility of their being Social integration and radon; of youth achieved because the social structure fails to suicide. Some authors have Eltauted the provide access to the means for goal achieve- steady increase in suicide rates over time to ment. Anomie conditions are generally seen concomitant changes in social intepation. as resulting from rapid social change, but it is Sudak, Ford, and Rushford (20), for instance, theoretically possible for them to be chronic suggest that increasing divorce rates and for some segments of a society. At the other diminishing importance of religious and end of the regulation continuum is fatalism, moral values have contributed to increased a condition created by excessive regulation. personal alienation (egoism, in our terms,) Here individuals' futures are foreclosed, and and, thereby, to increased suicide. Hawton lives are oppressively dictated.As with (2) also discusses the loosening of family ties altruistic suicide, fatalistic suicide as a over time as a causal agent in suicide among theoretical *type* has not been much used, the young, while Holinger and Offer (21) note and has been viewed as not particularly use- the 'Increase in broken families, and the im- ful. I turn now to a review of social and cul- plications of this increase for social integra-

2-58 C.Hufline: Social and Cultural Risk Frictors.. . tion of the young. They suggest further that feels of racism, among them, suicide. Seiden the large impersonal high schools of today takes a somewhat different approach but make it difficult for adolescents to find a with the same concepts. He suggrIsts that sense of self-worth and to establish changing racial attitudes in this country have friendships. led to increased individual freedom and, ironically, to decreased social integration for A final bit of evidence that social integration black people. Specifically, he argues that "as is a useful explenatoty concept for adoles- racial discrimination decreases, the stability cent suicide rates is the regularity with which of shared social relationships, the sense of minority group suicide rates are especially community based upon discrimination by a high in adolescence. It is only in adolescence common enemy, is likewise decreased" (24 and young adulthood that suicide rates of blacks approximate those of whites, and at Paile *- times they have exceeded them Data are available from two studies under- (20,2Z23,24).Similar patterns are found taken to test hypotheses about social integra- among American Indians (25,26,27) and tion and suicide among blacks. To test their Hispanic-American residents of the south- hypothesis that black adolescents living in a western United States (28).This relative black subculture would have lower suicide parity with white Anglo men in suicide rates rates than black adolescents living in "non- of young men of ethnic minority status sug- traditional" areas, Shaffer and Fsher (31) gests that the effects of uncertain social in- compared rates in two Southern and two tegration on all adolescents and young adults Northern regions. The specific regions were are compounded for minority groups. Inves- selected because blacks accounted for more tigatots of Indian suicide have been the most than 95 percent of their nonwhite popula- likely to discuss this possibility, describing the tion, and the investigators could then safely young American Indian as caught between assume that *suicide among nonwhites" two cultures. The profoundness of the iden- meant 'suicide among blacks." The results of tity problems tbe Indian might experience is the comparison confirmed the hypothesis. poignantly reflected in Tonkin's observation That is, blacks were underrepresented in the that, in British Columbia, native Americans suicides in the Southern areas, but they ac- may lose their Indian status when they live counted for almost the same proportion of off the reserve" (27, page 175). Resnick and suicide as of the population in the Northern Dizmang similarly describe the young areas. This finding helps put into perspective American Indian as caught between two cul- the extraordinarily high rates of suicide tures: 'he is unprepared for the one and feels among young blacks that Morris et al. (23) the other, toward which be is ambivalent, has found in Philadelphia. failed him" (29, page 886). In the other test of the hypothesized relation Social Integration and Rates of Suicide between social integration and black suicide, Among Youth. Social isolation is sometimes Davis (22) compared suicide rates and racial used, explicitly or implicitly, in discussing dif- composition of the 18 States with the largest ferences among subsets of young people in black populations. He does not interpret his terms of their potential for suicide. For in- finding of a negative correlation between the stance, social integration has been used to ex- percentage of blacks in the population and plain racial differences in suicide rates (20). the nonwhite suicide rate, but one might well Bush (30) posits the existence of a "blair make an argument that blacks living in states perspective' that can protect individuals where there are a large number of blacks are from the pain caused by racism. He suggests more socially integrated than blacks living that, to the extent that individuals differ where than are few other blacks. However, from, or move away from, the intragroup it should also be noted that among the ten perspective, they are vulnerable to the ef- States where the proportion of blacks in the

3 2-59 Report oi the Secretary's Task Force on Youth Suicide population decreased between 1970 and 1975, tially, the extent to which one's status con- the nonwhite suicide rate was as likely to go figuration is a configuration common within down as to go up over the same period of the population. Few teenage poops in this time. Among the eight States in which the society are married, so those who are have proportion of blacks increased or remained low status integration and, therefore, low so- stable, the suicide rate went up in seven and cial integration, and are at high risk for down in only one. sukide. We should note that Durkheim also would have expected high suicide rates It may be, as Shaffer and Fsher acknow- among married teens, especially men, be- ledge, that their and Davis' findings reflect cause early marriage forecloses options. In the disintegrative effects of migration, or the this respect, it would lead to fatalistic suicide. vulnerability to suicide of persons who migrate, rather than the protective effects of There is evidence that, among students, social integration. Direct evidence of migra- those who commit suicide are less well in- tion as a suicide risk is sparse and not very tegrated into their social groups than are the compelling.In Israel, suicidal behavior is others. Petzel and Riddle (36) observed that high among young immigrants, especially school nonattendance and multiple school those from Ma or Africa (32). It should be changs may be associated with suicidal be- noted, however, that Amir's data include at- havior. There is no consistent evidence that tempted, as well as completed, suicide and college students are more blrely than their that the rate of suicide for Israelis under the nonstudent age peers to commit suitide, but age of 19 years is extremely low. We should there is some evidence suggesting that stu- also remember Plat Sainsbury and Bar- dents of elite universities are at relatively raclough (33) found suicide rates of first- high risk (2,37,38). While one might suggest generation immigrants to be more similar to that these institutions are characterized by the rates of their countries of origin than to their emphasis on individualism and lack of those of the host countries. integrative mechanisms, Peck and Schrut (37) point out that the populations at elite One final point about racial differences in universities are likely to be more dispropor- suicide rates and social integration: In Dur- tionately male than are populations of other ban, South Africa, the rate of suicide among postsecondary institutions, and they are like- colored persons, those of mixed black and ly to be older. white heritage, is considerably higher than among the other major racial groups, In the descriptive material provided by Peck Africans, whites, and Indians. The differen- and Schrut from their study of suicide among ces are, however, greater among the middle college students in Los Angeles county, one aged than among the youth or the elderly can discern a pattern of behavior that sug- (34). gests failure to conform to the norms or expectations of the college student of the 1960s. Holinger and Offer (21) point out that, For instance, compared with other students, among teens, those who are or have been those who committed suicide were more like- married have much higher rates of suicide ly to spend a lot of time in solitary activities, than those who have not been married. Al- and to attend religious services and express though marriage is a primary form of social belief in an afterlife. They were less likely to integration and, therefore, protection use drugs and they were less experienced against suicide for adults, Holinger and sexually.Although we typically tnink of Offer's data suggest it is a suicide risk factor religiosity as inversely related to suicide (see for teenagers. This apparently anomalous below) and drug abuse as associated with it, finding would have been predicted by Gibbs we shout(' consider that failure to use drugs and Maren's extension of Durkheim's theory at all and attendance at religious services (35). They operationalize social integration were not normative behaviors for college stu- as *status integration," which means, essen-

2-60 C4 C.Huffine: Social and Cultural Risk Factors. dents in the 1960s, at least not in California. the sample.Smith and Hackathorn (43) These behaviors might, then, be seen as failed to find a relationship between religious evidence of failure to fit into the social group. integration and suicide across 69 primitive societies, although they found strong correla- Peck and Schrut, as well as Sanborn, San- tions between suicide and family integration born, and Cimbolic (39) who studied adoles- and economic integration.It may be, cent suicide in New Hampshire, found that, however, that their measure of religious in- compared with other students, those who tegra tion was not as valid as were the committed suicide were doing poorly measures of family and economic integra- academically. Shaffer (40) did not report on tion. the academic standing of the adolescent suicides occurring in England and Wales in There is no good evidence of a direct associa- the 1960s, but he did describe them as tend- tion between level of religiosity and risk of ing to be above average in intelligence as well suicide among individuals, because valid data as height. Nonstatistical analysis of the data on the religious views of the deceased are he had amassed on 31 cases of suicide led scarce. However, there is evidence from a Shaffer to suggest that they clustered into number of studies that religious involvement two personality types, the solitary isolate and is associated with low rates of sec-mance of the impetuous individual prone to aggressive suicide (42,44,45). The studies cited all com- outbursts. pared religiosity and attitudes toward suicide As noted earlier, religion was seen by in samples of young persons, a category much more liberal" toward suicide than older per- Durkheim as a key mechanism for social in- sons (46). tegration, and his findings of strong associations between suicide and religion have been We need to be very careful in interpreting replicated many times. These are findings of findings about *attitudes* toward suicide. higher rates of suicide among Protestants Respondents, especially the young, clearly than among Jews who, in turn, have higher differentiate between 'right* as a moral con- rates than Catholics. The interpretive prin- cept and right as a perquisite. Eighty-eight ciple of interest here is orthodoxy. In more percent of Boldt's adolescent respondents recent work the connection between religion said suicide is wrong (46). Yet, 35 percent of and suicide is less consistently located, and is them said it is a basic human right, and 47 per- lea clear than in early studies. For instance,, cent said people should not be restrained when level of religious commitment or invol- from killing themselves, or should be vement is controlled, investigators tend no restrained only under certain circumstances. longer to find differences in rate of suicide Minear and Brush warn that the liberal at- across religious denominations. Stack, for titudes of their sample of students toward the instance, found no correlation between the right to kill oneself "does not imply that a percentage of Catholics in the population majority of college students could personally and the suicide rate when he controlled for accept the idea of their own suicide. Most, in the divorce rate. The divorce rate was seen fact, could not imagine committing suicide* as a proxy for Catholic orthodoxy and Stack (42, page 321). interpreted his findings as follows: "We hold that the convergence of Catholic and Prates- Anemic Suicide tant normative structures is responsible for As noted earlier, anomie ccnditions are the lack of a relationship between generany associated, in the mind of the social Catholicism and suicide" (41, page 69). thsorist, with rapid social change. The Minear and Brush (42) attributed the lack of change needs to be of the type that leaves the difference in attitude toward suicide between social group, or some subset of it, without a Jews and agnostics in their sample of college set of *rules" by which members can regulate students to the absence of religious "ews in their lives.Without a set of expectations

f". 2-61 Report of the Secretaty's Task Force on Youth Suicide about what they should aspire to and how to stance, Resnick and Dionang (29) point out achieve the valued goals, the affected in- that the dominant culture in this country has dividuals are at increased risk for suicide. vacillated widely in its judgment of what In- The clearest example of this process I en- dians should be, and Indians themsehra have couMered is that praented in the very nice- been unable to make their own adjustments ly written article by Rubinstein (47) about an to the cultural changes they have ex- epidemic of suicide among Micronesian perienced.Reznick ard Dizmang point adolacents. Rubinstein descaes the trans- especially to th problems oioace-valued be- formation of Micronesian communities from haviors being no longer possible because of subsistence fishing and agriculture groupings the restrictions and limitations of raervation to reliance on a cash economy and govern- life, and to Mu loss of role models to young ment employment. Among the changa in men. The arguments they make in this work life this transformation has wrought is the ex- are, in large measure, illustrated in tinction of the traditional men's clubhouses Dizmang's earlier description of changes in and community-level nien's organizations. thelives of the Shoshone-BannockIndians Ilmse clubhouses and organizations had so- (49) and the apparent suicidogenic effects of cialized young males into their adult roles, thae changes on young men. In 1974, Diz- providing them a sense of social identity and mang, Watson, May, and Bopp (25) self-esteem, and compensating for the struc- presented the results of a case-controlled tural tensions and distance they encountered analysis of suicide on the Fort Hall reserva- in family relations. It was among young men, tion. Their goal was identification of per- of course, that the rate of suicide increased sonal factors associated with suicide. The dramatically and, as Rubinstein was able to childhood experiences of the suicide group demonstrate, among the young male resi- were much more chaotic than were those of dents of the district centers, which were the the controls, and they were much more like- foci of social change. ly to have attended boarding school and to have done so at young aga. One could argue Goldaey and Katsikitis (5) suggest that the that the experience of attending the off- surge in adolescent suicide among mid-cen- resenration boarding school, unless it is a step tury Australian birth cohorts is related to the in the process of assimilation into the social and erAlnomic changes in Australia dominant culture, is likely to accentuate the over the past 50 yews. Rin and Chen (9) suggest that the first of two surges in the cultural confusion experienced by the young American Indian. Taiwanese rate of suicide is related to the migration to Taiwan of 2 million mainland Resnick and Dizmang (29) point out that Chinese within a ten-year period. They see suicide rates vary tremendously across Indian the second surge as related to a period of groups, a point taken up in greater detail by modernization, industrialization, and chang- Shore (50), who presents data revealing ing value orientations. Socioeconomic chan- variation across tribes from 8 to 120 per ga raulting from political independence are 100,000. The figures he presents suggat that designated as likely causal factors in the suicide rates generally are low among south- marked jump after 1969 in the suicide rate in western United States tribes (Navaho and Sri Lanka (48). None of these investigators Papago) relative to northwestern tribes. Ex- provides an analysis of how socioeconomic or trapolating from homicide rates, Shore sug- political changes affected the lives of the gests that suicide rates are high among population stratum in which the suizide rata traditionally nomadic tribes and lower among have increased. agricultural village tribes. Although he does not offer interpretation or analysis, it is evi- The widely-recopized high rate of suicide dent that the controls of the Federal govern- among American Indians is often charac- ment, including movement of Indians to terized as a manifestation of anomie. For in- reservations, was more disruptive of the cul-

2-62 f; C.Huttine: Social and Cuftural Risk Factors.. . tural patterns of nomadic, than of the village- pm on the world in which men live. Thus, it based, groups. is feasible that the effects of the women's Some writers have suggested that social movement, insofar as they produce change in change during the 1960s resulted in anomie the suicide rate, are more evident among men than women. among black Americans and may account for increased suicide rates among blacks. Hen- There is, as noted at the beginning of the din, for instance, suggests that changes in paper, a condition other than rapid social what are considered acceptable forms of change that has become identified as anomie.. response to racial oppression can create That is the social condition in which struc- anomie situations for older blacks. "They be- tural mechanisms fail to provide oppor- come suicidal only when an adaptation that tunities for individuals to achieve culturally has been distinctly Negro in our culture fails valued goals. A number of authors assert them" (51 page 419). Seiden (24), although that this condition has been produced by the focusing on theories of social integration, high rate of fertility in the 1950s. Borrowing points out that, during times of marked social from Easterlin (52), they suggest that this change such as we experienced in the 1960s, large cohort of young people has been con- there is confusion about customs and moral fronted with increased competition for codes. One may infer that this resultant rewards and reduced opportunity to acquire anomie put young blacks at higher-than- them (20,52). usual risk for suicide. The concept of anomie has, then, been used Mother recent social change that might be less often, and with less attention to the expected to have disproportionately in- theory of which it is a part, than has egoism fluenced the birth cohort known as the "baby (isolation) in analyses of changes in suicide boomers* is the women's movement. With rate among American subpopulations.In the emphasis during the 1960s and 1970s on fact, one sees the concept of anomie acquiring and using access to educational and employed primarily in anthropological professional opportunitie-. changing sexual studies of traditional societies. It may be that mores, and altering thc nature of intimate as contemporary societies are not as vulnerable well as formal relations between the sexes, to rapid social change as are those like post- expectations about the American woman's war Micronesia, and/or that the effects of role, goals, and behavior were seriously chat- change are buffered by countervailing forces. lenged. The female members of the 1950s Except for the case of the nomadic American birth cohort grew up without the certainty Indian, the only United States group to which about their roles that had characterized their we can point as having demonstrated anomie mothers' lives. One might expect, then, that suicide in recent history is businessmen who they would demonstrate increased rates of responded to the crash of the stock market in suicide as adoles.As and young women. Al- the late 1920s by killing themselves. though rates of suicide among young white women have increased, the change has not Altruistic Suicide been nearly as great as for young men. The rate among 20 to 24 year old nonwhite As noted earlier, there has been a tendency among modern theorists to dismiss the con- women has increased tremendously, so that cept of altruistic suicide from serious con- it is now almost equal to that of their white sideration because of its lack of relevance to age and sex peers. This is an interesting and modern society. I consider this tendency un- potentially important phenomenon, but it wise for a variety of reasons. First, of course, probably cannot be taken as an effect of sex- we saw altruistic suicide re-emerge in Jones- role changes in the past 20 years. We should town with tragic results of astounding mag- not forget, however, that changes in women's nitude. Second, it appears to be very much a roles and expectations have profound chan- part of what drives suicide rates up in some

fi 2-63 Repoit of the Secretasy's Task Force on Youth Suicide societies. Third, where some form of altruis- towhich there is very limited access and, thus, tic suicide is part of the cultural tradition, it potentially an anomie situation, it also seems is likely to influence a society's suicide rate as that the young Japanese man's suicide is an "enabling factor,' a factor that makes motivated less by disappointment in his suicide acceptable, if not prescribed or ex- failure to achieve his goals than by his failure pected. to fulfill his responsibility to his family. Fuse (55) describes the tendency of many It may be that part of the impetus to dismiss Japanese to become overinvolved with their altruism as an important factor in suicide, is social role as 'role narcissism.' He means by the word itself. We tend to emphasize the this that the social role becomes eathected by connotation of benevolence in thinking of the individual as the ultimate meaning for this term. Yet, "altruism" also refers, perhaps life. Threat of loss of that social role creates even more strongly, to selflessness. The such shame and chagrin that the individual word was formed by Comte from an Italian may well choose to end his life rather than term meaning 'of, or to, others." It was trans- continue it in an altered or degraded social lated into English as "devotion to the welfare role.Iga (56) describes Japanese culture of others" and as an antonym to *egoism." In similarly, as one in which the basic social 1876, Marlborough described the law and structure is a small group and in which the duty of life in altruism as living for others supremacy of the group goal is stressed. Em- (54). My point is that the aspect of altruism phasis on education and the entrance ex- that constitutes a suicidal risk is the abnega- amination, which is the student's single tion of the self, the giving up of one's own opportunity to enter a university, have be- identity, or the dominance of the identity by come Japanese tradition.Iga vividly a single social role or social tie. If wv focus describes the pressures brought to bear by his on this aspect of altruism, we can understand parents, especially his mother, on the the acquiescence of hundreds of persons to Japanese adolescent to achieve a university the call from their leader that they murder education. The family may pay enormous their children and commit suicide. costs, financial and emotional, to prepare In societies such as the modern United him for the examination. The student who States, in which there is such strong emphasis fails does not just fail to realize a personal on autonomy and the development of in- goal; he fails his family and he fails, on behalf dividual identities, we seldom see cases of of his family, to achieve a group goal. My insuicide we might identify as altruistic. terpretation is supported by Hirsch (57), who However, in other societies, group loyalties describes the Japanese culture as consistent- ind family ties are such that the individual ly emphasizing an intense, life-long, parent- might be said not to exist and, in those child relationship. He identifies cultural societies, some suicides appear very much to values stressing self-discipline and subser- be sacrifices of the self for the group. For in- vience ot the self to a whole system of hierar- stance, Amir (32) says that suicide rata were chies leading to shame and guilt when role very high among the Israeli pioneers, and he expectations are not met.Hirsch cites attributes this phenomenon to a very intense DeVos in describing the need for children to social solidarity and the attribution of failure achieve and bring honor to the household in to individual shortcomings. order to fulfill their duty to their parents, and to repay them for their sacrifices. Although One can see traces of altruism in the very high there is no evidence that suicide is expected rates of suicide among the youth ofJapan. It of a young Japanese man who fails to gain ac- is widely accepted that the tortuous process of gaining access to higher education in cess to a university, the guilt he feels for failing his family, combined with the failure to Japan precipitates a good many suicides achieve the goal that has, for many years, among those who fail. Whereas it appears been his single focus, and cultural enabling that this is an instance of a highly valued goal

2-64 I j C.Huffina: Social and Cultural Risk Factors.. .

factors I will refer to later, constitutes a bership is maintained and against which serious risk factor for suicide. there is likely to be precious little protection. Hoskin, Friedman, and Carote (58) investigated an area of New Britain which had a Fatalistic Suicide high incidence of suicide. In describing the Fatalistic suicide is relatively infrequent and social structure of the communal groups, there is little evidence of high rates where they speak of the extreme importance to the one would expect them, e.g. among prisoners individual of personal ties and affiliations, servirig life sentences, victims of incurable and they discuss some historical rationale for debilitating disease. The one report of what the critical importance of the individual's af- I might call fatalistic suicide describes situa- fdiative tie to a small group of others. In their tions in which other risk factors are also investigation of suicide among the popula- praent. Gehlot and Nathawat (61) describe tion, they learned of the primacy of the loss suicide in India where the overall rate is quite of personal ties among the deceased. While low but where, relative to other segments of the authors seems to lean toward anomie as the Intlian population, the rate among young an explanation for the high incidence of people is quite high. The authors label a suicide, the social fragmentation they speak large proportion of young Indian suicides as of has, in fact, a very long history and could *performances suicides, but it is not clear just be seen as the reason why the individual what that term means and what its theoreti- Kandrian is so dependent upon relations cal backgrould and implications are. with a very small group of others for his/her However, in their case studies, I find continued well-being. In effect, I am sug- evidence of fatalistic suicide. They describe gesting that the role relation is one in which the deaths of two young women, well-edu- the individual is as role-cathected an are the cated professionals, who had been forced, by Japanese described by Fuse and Hirsch. The the absence of their husbands, to live with, individual's identity is extraordinar: y de- and under the control of, strict and tradienn- pendent on the role relationship. al in-laws. These women were unable to pur- sue their professional activities, they were In sum, I would argue that we not dismiss the unable to escape the situations they were in, concept of the altruistic suicide but, in fact, and they were unable, by virtue of their be wary of its potential for young people. education and/or personal characteristics, to Bbs (59), among others, describes adoles- fulfill traditional roles as subservient and cence as an individuation process and, there- dutiful daughters-in-law. In this sense, their by, an oppurtunity for the young person to suicides take on a fatalistic quality. Headley confront and resolve conflicts associated (7) similarly suggests that self-inflicted death with dependencyThere are a variety of among Asian women and young people tends ways, such as use of drup, that the adolescent to be fatalistic suicide. She focuses on the might avoid managing those development very limited opportunities these categories of tasks. An alternative available in the past 20 persons have. "The options for women to yr -vso is the cult. The cult provides al- refuse marriages they do not want, for sup- was to tte threatening and difficult ex- port should they be divorced, and for status periences of young adulthood, and it involves in society other than as wife and mother are relinquishing one's own sense of self which, limited (7, page 355).Thus, although for the troubled or fearful young person, can fatalistic suicide has long been discounted as be experienced as a relief (60). In so doing, a virtually empty cell in Durkheim's however, the young individual is trading off paradigm, excessive regulation may, in fact, the risk of egoistic suicide, which is likely to be an important cultural risk factor in some exist for a short period and for which there parts of the world. are social buffers, fo .. the risk of altruistic suicide which is present as long as cult mem-

2-65 Report of the Secretary's Task Force on Youth Suicide

Cultural Attitudes and Socialization as empathic with Mishima, who caine to be described as a martyr, an exemplar of what it Cultures do not contribute to suicide rates of means to be Japanese, and as dying splendid- populations only through the kinds of struc- tu:al variablea descriled in the preceding ly (63)- pages. They may also facilitate suicide in The goal of Fuse's paper was to explore the such ways as romanticizing suicide, present- *deep relationship between suicide and cul- ing death as a positive state, or influencing ture in Japan,* and he demonstrates quite the psychological development of individuals eloquently how *ritualized suicide such as in such a way that they are susceptible to seppuku was a culturally approred and rein- suicide. forced means of safe-guarding self-esteem and nonce' (55, page 63). Given that, he In the traditions of India and Japan, suicide has been an acceptable, even prescribed, pleads for restraint in what he sees 83 an increasing tendency to define suicide in terms practice. Suttee, the form of suicide among of psychopathology. Hirsch sums up his dis- Indian widova with which we arc familiar, cussion of the cultural determinants of was practiced by women who simply had no suicide by stating that Japan historically sane- other choices, or by those for whom the rioned ritualistic suicide and *condoned sacrificial pyre was a way to achieve social ad- other forms of suicide as acceptable means of miration. Another form of self-immolation, dealing with life's problems* (57, page 339). Joint., was practiced by Indian women whose Prominent 'among the cultural factors cited husbands had been killed in battle. Here, the by Tatai as underlying Japanese suicides are suicide was to avoid being raped by the con- Nan historical tradition of suicide as an querors (61). Thus, in India, suicide has tradi- honorable solution to harsh and difficult per- tionally been a way for women to resohre sonal solutions* and *a romanticizing of dilemmas created by their social role and the suicide as an escape from the stresses of life dependence on their husbands that role im- under a mantle of acceptability* (4 page 19). posed on them. Acmptabk forms of suicide in India were those associated with religious In two studies of suicide in nonindustrialized motives and committed by ascetics, and those societies, there is evidence that suicide is, in committed by the very old, the very feeble, or some form, ritualized or accepted. The the incurably ill (62). primary method of killing oneser among the Japanese culture also prescribed suicide Kandrian is hanging, and there appears to be a ritualistic procedure known to members of under some circumstances, but it was men the society. Kandrian children represent it in who were to kill themselves, and their doing their drawings (58).Rubinstein describes so was valued as demonstrating the highest the development of what he terms a *suicide of virtue& Suicide is, of course, proscribed in subculture* among Micronesian youth. It is Japan today, but the influence of the past is *a set (*coherent meanings which organize, clear in the conflicting and confusing reac- provide significance for, and contribute to tions to the suicide in 1970 of Mishima Yukio. the frequency of adolescent suicides in Proclaiming the values of the samurai, be in- Micronesia* (47, page 664). vaded the headquarters of the Japanese Self- Defense Force and, in front of the Cultural views of death that might facilitate commanding general of that force, Mishima or enable suicide are illustrated by writers committed seppuku, the time-honored about Japan more than others. Iga for in- ritualistic suicide by disembowelment. The stance, talks about how there is little separa- initial reaction to this event was to declare tion between this world and the aftenvorld in Mishima insane and fanatical. However, the Japanese culture, and bow the Japanese day after Mishimats suicide, the Japanese romanticize impermanence (e.g., treasuring people were asked to respect his motives, and most highly the most fragile of flowers, the various spokespersons described themselves cherry blossom) (56).Tatai (6) describes

2-66 70 C.Huffkle: Social and Cultural Risk Factors.. .

Shintoists as viewing death as an opportunity The risk factors found to be most closely and to become a kami, or spirit, and Buddhism as consistently associated with youth suicide in presenting the idea of reincarnation and sur- this country are those that, in some way, vival after death. operationalize Durkheim's concept of egoism. These risk factors reflect weak so- The only study I found that tested a cial integration and the concomitant isola- hypothesized relationship between per- tion that individuals experience. Weak social sonality and suicide across cultures is Smith integration may be endemic to adolescence and Hackathorn's (43) comparison of 69 and young adulthood in countries such as the "primitive and peasant" societies in terms of United States. The ubiquitous isolation felt suicide rates and a number of potential cor- by adolescents and young adults might be relates, including expression of affect, and amplified in minority groups, whose adoles- the importance of pride and shame. The cent members experience the combination of potendal predictor variables were selected age-related problems of egoism, and on the basis of review of existing literature problems created by the rejudice of the and availability of measures in the data files majority group, preventing integration of they used. The personality characteristics minorities into the social mainstream. are two of the three variables accounting for 69 percent of the variance in a 7-point suicide Whether or not adolescence is a period in rating scale. The data revealed that suicide which social ties are vulnerable and tenuous, rates are high where societies expect very many social critics feel that ours is an increas- restrained, or very open, expression of emo- ingly egoistic society, and they cite as tion, and where they place a great deal of im- evidence such social changes as increasing portance on individual pride and shame. divorce rates and decreasing importance in Certainly we see the emphasis on these people's lives of religion. Risk factors reflec- characteristics (emotional restraint and tive of weak social integration have been shame) in the description of socialization of used to 'explain" differential suicide rates Japanese children. it would be very informa- among subgroups of youth, es well as the in- tive to assess child-rearing practices in other crease over time of youth suicide in general. industrialized societies, and to test the The integrative function of identification hypothesis that cultures encouraging these with the minority group and its "perspective" characteristics in their populations have high has been identified as protecting young suicide rates when compared with societies blacks from suicide. Students who kill them- where moderation in expression of affect is selves tend to have histories of behavior that endorsed and individual pride and shame are is not normative; that is, they tend to look like not emphasized. social isolates. Religion was seen by Durkheim as a major SUMMARY form of social integration, and social theorists have tended to continue to see so- During the decade of the l970s, the United States witneffsed a startling increase in cial integration as a critical function of religion. It appears that the degree to which suicide rates among adolescents and young individuals are committed to orthodox adults. The increase has been described as religious beliefs continues to predict levels of part of a cohort effect, an increase in suicide However, rates in each successive birth cohort. It has suicide fairly consistently. religious affiliation is no longer a dependable also been attauted to the coming of age-at- index of orthodoxy and is, therefore, not use- risk of the baby boom generation, and the only countries showing similar patterns of ful as a predictor of suicide. change in youth suicide are those with similar Risk factors that can be clustered because post World War II fertility patterns. they reflect Durkheim's second major "type" of suicide, anomie, are las useful in analyz-

7 1. 2-67 Report of the Secretary's Task Force on Youth Suicide 11.1.11.=111=11110 ing suicide in the United States than are among some Asiatic subroups has been those discussed under the label *egoistic.* facilitated by Durkheim's predictions about However, anomie continues to be a produc- the effects of excessive social regulation. We tive concept when suicide rate changes in might be well advised to re-explore this part traditional societies are under consideration. of the theory. The societies in which anomie is readily avail- In sum, cultures contribute to high suicide able as an explanatory principle are those in rates by failure to provide supportive stsuc- which major and rapid social change can and tures such as those that integrate individuals does occur. 'The relative unavailability of the into the social group, by creating stress for concept for analysis of suicide in the United members of the social gsoup (e.g., rapid so- States may be a function of the resistance of western social structures to rapid major cial change leaving individuals without a set of norms to follow), by influencing the per- change. That is, these societies may be rela- sonalities of individuals through socialization tively invulnerable to anomie. practices (e.g., promoting repression of af- The problem in using the concept of anomie fect), and by promoting positive or accepting in reference to suicide in our society may attitudes toward suicide (e.g., ennobling cer- stem from the concept's resistance to tain forms of suicide or making the suicides operationalization rather than, or as well as, of certain persons heroic). society's resistance to rapid change. For instance, although writers may speculate on the CONCLUSIONS suicidogeniceffectsof changes (such as those in the status of blacb and women), we have The major conclusion I drew from this seen no specification of measures that would analysis is that we need more theory-driven reflect anomie, which then may be used in research. It is important that more and bet- comparative analyses of group suicide rates ter work be done in the following: or rates across time. Specification of theories, models, and Altruism as a suicide risk factor has tended to concepts related to suicide be dismissed by social scientists. I have ar- Operationalization of relevant concepts gued that this tendency is unwise, and that we into measures need to think once more about the risks posed when persons abdicate their egos for a Derivation of hypotheses from theories group identity and subordinate individual and models of suicidal behavior goals and desires to group needs and dictates. Testing hypotheses usin valid and reli- A second reason for not dismissing the able data and appropriate analytic tech- altruistic suicide as relevant in the world of niques today is that it may be a powerful enabling We need, for instance, to think about factor. There is evidence that where altruis- Easterlin's assertions that the fertility rates of tic suicide is part of a society's cultural tradi- one period determine the opportunity struc- tion, suicide continues, even in the face of ture 20 years later, and that the opportunity official proscriptions, to be seen as an aecept- structure influences the probability of suicide able, even honorable, way to deal with among young adults. What hypotheses can shame. The attractiveness of altruistic be derived from his model and how might suicide is Rely to be especially strong for rer- they be tested empirically? What oppor- sons at an age when they are confronted with tunities does his model provide us for being the painful hard work of individuation. more effectively proactive in preventing, or Evidence for the presence of risk factors that at least tempering, increases in rates of might predict fatalistic suicide is sparse and suicide among specified groups in the tenuous. However, interpretation of suicide population?

2-68 72 C.Huffine: Social and Cultural Risk Factors. . .

The second conclusion from this work is that & Gold/ley RD, Katelkitie M: Cohort analysis of suicide rates in Australia. Archives of General Pep:hist,' there is continued need for clarification and 40:71-74. Kea assessment of theoretical issues and ques- O. Tatal K: Japan. In LA Irtra_ (ed)Sde in Asia and the Near East Berkeley, i tnVit, University of tions. What, for instance, is thd relation be- California Pres& pp 1248, UM tween integration and anomie? I speculated 7. Headley LA: (Ed) Suicide in Asia and the Neer East earlier that contemporary societies might be Berkeley, California, Unimety of California Press, KS& S. Headily LA:Hong Kong. Ai LA Heade (ad) relatively invulnerable to anomie because of Suicide In Ada and the Near East Wm*, their resistance to rapid social change. We Whinny of California Press, pp BMW, have seen that contemporary societies tend 9. n H, Chen T: Taiwan. In LA Headley (edjniveSuicifesdev in Asia and the Near Ent Betkeley, California. to be characterized as relativelyegoistic. of California Press, pp 59418, 1983. This suggests that the more integrated a 10. Murphy GE, Witrel RD: Suicide risk by WM cohort In the United Slates. 1949-1974. Archive* of General society is, the more vulnerable it is to anomie. Psychiatry 37:59-823. Might this conclusion be derived from U. Dosand BJ, Bourque LB, Kraus JF: Suicide amino Durkheim's work or from that of the social adoleacents in Sacramento County, California Adolescence 17:917410, 1982. theorists who have followed in his footsteps? 12.Bouvier LF:America's beby boom I am not the first to suggest that social in- The fateful bulge. Population Bulletin38:1-35,Cratim tegration and social regulation are related 13. Durkheim E Suicide: A Study in SocIology JA Spaulding and G Simpson (Trans) New York, FRIO Primi processa. I do not believe,however, that 195. they are the same, and I feel we can profit 14. Giddons A: A tyrirettf suicide. European Jour- from trying to describe, not only how they in- nal of Sociology 7278- 15. Johnson BD. Durkhelm's one cause of suicide. fluence one another, but also how they are AMfiCal Sociological Review 30:87548& 1985.

independent. 18.Mastin WT: Theories of Variation In the Suicide Rate, in Gibbs JP (ad): Suicide. Kra York, Harper and We know that social systems are not static but Row, pp 7448. are dynamic. Might use of someof the basic 17. Konook. G:Adolescent suicide. Exceptional Children 491 . MB& principles of systems theory help us under- IS. MU* it The adolescent suicidshablem. Suicide stand variations in suicide rate across and Life- Threatening Behavior 18109, societies and time? For instance, might the 19. Sabbath JC: The suicidal adolescent-the expendable child. Journal of the Mellow Academy of Child recent emergence and proliferationof cults Psythiatry 8:272435. 1989. and fimdamentalist groups in the United 20. Sudak HS, Ford AB, Rushforth NB: Adolescent suicide: fro overview. American Journal of Psychotherapy States be understood as reaction to a social 3B:350483, 1984. system that has become excessivelyegoistic? a Wenger PC. Offer D: Perspectives on suicide kt What I am proposing here is that societies adolescence. Research in Community and Mittel Health 2139457, W. might be thought of as oscillating between 22. Davis it Black suicide in the seventies: CUrrent the extremes of the integration and the trends. Suicide and Effe-Threstening Behavior 9:131440, regulation continua. This way of looking at 1979. 23. Morris JB, Kovacs he, Beek AT, Wolff, k Notes societies and social change might constitute toward en ivitniMogy of suben suicide. Comprehensive a more productiveframework for assessing Psychlaby 1&537 W7-847 4. 24. Wen Rit Wty are suicides among blacks in- and interpreting change in rates of suicide creasing? KSMHA Health Reports 87:34. 1972. over time than any of themodels currently in 25. Diming LH, Watson J. May PA. BoPO .1: Adoles- use. cent suicide at an Indian Reservation. American Journela? Odhopsychiatry 4443-49. 1974. REFERENCES 28. Ogden M. SPealar Mi. Hill CA Jr:Suicides and homicides among Indian& Public Health Reports UM- Centersfor Disease Control. Violent deaths among ea 19m CZIn15-24 yews of age - United Slates, NW& PAW- 27. Tonkin RS: Suicide methods in Bdtish Columbian asndMortelity 14iieW Report 3t453487. KIM adolescents. Journal of Adolescent Health Care 502-25. 2. Newton K: Suicide and ?dimpled Suicide Among IN& Children and Adoiesoents. Bawdy Me, California, Sage 213. Smith JC, Mom JA, Aei alit CW: Comparison of Riblications, 1988. suicides among_ Anglo.' arkell losin five South- 3. Holinger PC:Adolescent suicide: An western slates. S 'te anu Life Threatening Behavior study of recent trends. Marken Journal 15:14-20, 1985. rItycnrallatrytire:784-788,19711. 29. Resnick HLP, Olzmwg LH:Observations on 4. Solomon MI, Helton CP. Suicide and so inAlber- suicidal behavior among American indium American ta. Cared& 1981 to 1977. Archivesof Remit Psydilatry Journal of Psychiatry 127882487, 1971. 37:511-513,

7 2-69 Report of the Secrettuys Task Force on Youth Suicide

30. Bush tikSuicide and blacks: A cormal 55. Fuse T: Suicide and culture in Japan: A study of framework. Suicide and Ufo-Threatening Behavior seppuku as an institutionalized form of sukide. Social 222, Iva Psychiatry l&fr-03.198G. X. Shaffer D, Fisher P: The epidemiology of suicide Iga M: Suicide of _Apneas Youth. Suicide and in children and young &dehiscent*. Journal of the Ufe-Thredenkm Behavior 1:1740, 1919. Americen Peademy of Chlid Psychiatry 2054$05, g. 57. Hirsoh ik Cultural detemilnants of suicide: The M*M &Wide wrim minom in isms!. The Israel perspective of the Japanese. Mental Hygiene. 3:337.030. Annals of Psychiatry and Mad Disciplines 0:2194109. 1973. 61 Hoskin JO, Friedman MIL Comte JE: A high incidence of suloide inmaeltprelitemte primitive sockity. 33. Sainsbury P, arradm0 B: Differences between Psychiatry 32:200-200. suicide rates. Nature 2204252, MM. 50. Blos P: The Molested Passage. New York, inter- 34. Mow F: Race and suicide in South Africa London, national Universities Pose, 1979. Routiedge & Kogan Paul, 1978. 00. %Which DK, Ungedeider J7: Destructive 35. Gibbs JP, Martin W7:Status integration and of the Cult Experience, In Peck MI., FarberowNI.,=A Weide. Eugene, Oregon, University of Oregon Press, RE (eds): Youth Suicide. New York. Springer. pp am, 1984. 38. Pettel SV, Riddle M: Molesoent suicide and cog- 01. Gehlot PS, Nathewst S& Suicide and family con- nitive aspects. Adolescent Psychiatry 9343-399, Ho. stellation in India. American Journal of Psychoihmapy 37. Peck AR.,_Schnrt k Suicidal behavior wrong col- 3T:273-278. 1961 lege students. K" ANA Health Reports 8&I40458. 0171. 03 Root India. In LA Headley (ec)Suicide MAW& and 38. MR K:Backgrounds of higher suicide rates the Near East Berke*, California, tWawigry of California among 'name unlverel students: A retrospective study Press, pp 210437, Mita deo=hvemv flare Suickle and Ulo-Threalening 03. Kristelter JL Mishima% suicide: A psycho-cultural 1518* analysis. Psychologia 18:5049, 1973. 30. Sanborn DE ilk Sanborn CJ, Cimbotio P: Two years of suicide: A study of adolescent suicide In igew Hempshire. Child Psychiatry and Human Development 1973. 4a Shaffer D Suicide in children and *arty adolescence. Journal of Child Psychology and Psychistry15:275- 291, 1974. 41. Stack & ROgion and suicide: A reanalysis. Social Psychiatry 15:55-70, MOM 42. Wear JD, Brush Lft The correlations of attitudes toward suicide with death anxiety, rel4Wry, and personal closeness lo suicide. Omega 11:317-M4, MO. 43. Smith CH, Hackathom L:Some social and psychological factors related to suicide Ito ilirimitive societies: 64 cross-cultural comparative study. Si.Mde and Ufa-Threatening Behavior 12195-241, 1982. 44. Best JB, Kirk WG: =and self-destruction. Psychological Record 3:t 45. Welter JW: 11gloefty, fear of death and suicide acceptabMly.Suicide and 1.110-Threatening Behavior 9:183-172, 1979. 46. Boldt Pk Normative evaluations of suicide and death: A crosesenerational study. Omega 13145457. 1982-16. 47. Rubinstin DK Epidemic suicide among Micronesian adolescents. Social Science and Medicine 17557485, Ian 48. Dissaiwake SAW, De Silva P:Sri Lanka. In LA Headley OW) Suicide in Asia and the Near East Bwieft, California, UMmrsity of California Press, pp 117-200, 40. Dizmang LH: Observations on suicidal behavior among the Shoslione-Bannodr Indians, Presented at 1st National Conk woe on Sufoldology. Chicago, MIL 50. Shore Jig:American Indian suicide - fact and fancy. Psychiatry 38:8501, 1975. 19. Hendon H:Black Weide. kchives of Gem* Psychiatry 21:407-422. 52. Eastedin Rk Binh and Fortune. New Ye*, Basic Books, Hendin H:Suicide among the a: rlity Psychodpentioe and rwgrapity, In Peck ML. NL, Utrnan RE (eds): Youth Nrm York, Springer, pp IDA 1985. 54. Old Ertglish DicPresstionary. Compact Edition Oxford, Oxford Whereby . WI.

74 2-70 FAMILY CHARACTERISTICS AND SUPPORT SYSTEMS AS RISK FACTORS FOR YOUTH SUICIDAL BEHAVIOR

Cynthia It Pfeffer, MD., Associate Profasor of Clinical Psychiany, Cornell Univers4 Medical College, and Chief Child Psychiany Inpatient Unit, New York Hospital-Westchester Division, White Plains, New York

The search for etiological, correlational, and studies, a large number have not focused early warning signs of child and adolescent primarily on family factors but describe them suicidal behavior has intensified, especially as aspects of the more extensive investigation because of the heightened awareness that of multiple factors associated with youth the incidence of suicide and nonfatal suicidal suicidal behavior.Since these studies behaviors have been increasing in the last evaluated factors after suicidal tendencies three decades. Current evidence suggests were expressed, they do not provide informa- that suicidal behavior is a complex multi- tion about family factors that are precursors determined symptom An implication of this for youth suicidal behavior. This issue can be concept of youth suicidal behavior is that in- studied in prospective longitudinal investiga- vestigations aimed toward elucidating the tions which, currently, arc absent in this field most important determinants of s of invatigation. havior-including diagnosis, persoifality Other important limitations exist in previous traits, family and environmental factors, and biological variables-require an integrated studies of family risk factors for youth suicidal behavior. Whereas most studies provide a approach that gives credence to the role of definition of suicidal tendency, the methods interactive effects of a number of variables. to evaluate the suicidal tendencies are often not described and certainly not systematic or Research Design Limitations uniform. The studies vary in the type of A variety of approaches have been used to suicidal tendency being invartigated. Some evaluate family factors associated with studies assess subjects with only suicidal idea- suicidal behavior. Nevertheless, knowledge tion, others evaluate suicide attempters, and gained about family rick factors for youth still others investigate subjects with a range suicidal behavior has been limited by the of suicidal tendencies that include suicidal designs of previous investigations. There is ideation, threats, and attempts. minimal information about family charac- Biases in selection of subjects may exist. teristics of youngsters who commit suicide Most studies use patients who were admitted and a paucity of prospective studies of child to medical or psychiatric emergency, in- and adolescent suicide victims. Almost all patient or outpatient facilities.There are the existing studies describing family features limited data on subjects in other settinp such are of individuals who exhibit nonfatal as schools and correctional facilities or for suicidal behavior.Furthermore, of these

75 2-71 Report of the Secretarys Task Force on Youth Suicide

those who drop out of treatment. There are prospective studies of high risk child popula- almost no studies of nonpatient populations. tions. For example, children who suffered The methcds ofcollecting data predominant- parental loss, family breakup, and environ- ly have used chart reviews or clinical inter- mental instability are appropriate subjects views. These chart reviews are limited by for investigations of high risk populations to informant, interviewer, and recording examine the precursor factors of suicidal behavior. variability. The clinical interviev- techniquo Anothcr issue to be evaluated is varied. Most studio did not use systematic whether there is a proximal association be- interview approaches. In some studies, the tween stressful live events and suicidal be- procedures for interviewing were not stated. havior among children and adolescents. The Only a few studies used standardiztA inter- next sections will discuss research on this view instruments or self-report ratings. issue. Finally, the statistics are reported in inconsis- Studies of Multiple Factors tent ways. For example, the greatest limitation is that most research papers do not Associatod with Youth Suicidal include complete statistical data needed to Behavior compare the findings from different inves- Table 1 outlines family factors in the cross- tigations. As a result, it is diffirult to mean- sectional studies of children and adolescents ingfully ware the effect sis of results. that erith,ated multiple psychosocial variables rtv.7.1ciated with suicidal behavior. verthelos, although the studies of family (Tables appear at the end of this chapter.) factors were limited by a number of Dermitions for suicidal behavior were clear- methodological features, the information ac- ly presented in most of these studies, but quired in these studies have a number of con- varied.All youngsters were studied after sistent trends that lend support to the they expressed suicidal tendencies. In most importance of family risk factors for youth studies, data were collected during interviews suicidal behavior.These findings will be of the youngsters and their parents. Most of described in the next sections. these studies were of youngsters admitted to psychiatric or medical emergency facilities. Studies of Adult Suicidal In this way, generalization of the findings to individuals the general population, medically ill, or un- Before there was an extensive interest in treated youngsters was limited. The sample studying youth suicidal behavior, the role of sizes ranged widely from 37 to 1010 early childhood experiences on suicidal be- youngsters. The comparison groups con- havior was investigated in retrospective sisted predominantly of youngsters evaluated studies of adults. These studies indicate that in the same setting as the index sample, so family instability due to early parental loss there was control over demographic and and deprivation is related to adult suicide and other variables that were not the specific nonfatal behavior (1-9). In fact, studies of focus of study. adult suicidal individuals suggest that the There was only one study of suicide victims critical time period for sensitization to the ef- (27). This pilot study utilized the psychologi- fects of parental loss is during the preadoles- cal autopsy method of interviewing relatives cent and early adolescent years (10-12). and friends of 20 children and adolescents These retrospective studies are important who committed suicide. There were 17 non- not only in suggesting ways in which early life suicidal controls who were matched by age, events may be associated with long term out- sex, race and social status to the suicide vic- come of suicidal symptom expression but also tims.Significantly higher prevalences of in pointing out the need for cross-sectional family breakup, violence, emotional problems, and suicidal tendencies were .e..;

C.R.PtelferFamily Characteristics & SupportSystems...

found in families of the suicide victimsthan patients who attempted suicide with38 the controls.Although the number of psychiatrie inpatient controls matched byage suicide victims was small in this study,the and sex. The authors determined that,al- types of family turmoil reported have also though there were no differences inin- been found in many investigations of nonfa- cidenfx of parental loss in the twogroups of tal suicidal behavior of children and adoles- adolescents, the suicide attempters hada cents (see Table 1). greater incidence of parental loss before they Among the studies of nonfatal suicidal be- were 12 years old than the nonsuicidal in- havior, three types of family factors have patients. Furthermore, parental discordex- been consistently associated with suicidal be- pressed as threats of parental separation havior. Fust, therewere family stresses in- and/or divorce were significantlymore com- volving changes in the composition of the mon for the suicidal adolescents. families because of losses, deaths, andparen- A further indication of the relationbetween tal separation/dMxce (13,17,21,23,2426). lack of social supports and suicideattempts Second, family violence especially involving was found when Stanley and Barter followed physical and sexual abusewas highly repre- up on the outcome of these hospitalized sented in these studies (21,22,24,26,28). adolescents. The previously suicidal adoles- Third, symptoms of family depression and/or cents were significantly morc likely toexpress suicidal behavior were found to be associated future suicidal behavior (Xh= 7.87, df = 1, with suicidal behavior of the children and p<.01). For example, approximately 50per- adolescents (13-15,17,22,23,26). The find- cent of the previously suicidal adolescents ings of these studies, therefore,suggest that continued to exhibit suicidal behavior after more indepth assessment of these three types hospital discharge in contrast to four adoles- of Lazily factor correlates of youth suicidal cents in the comparison group who showed behavior is warranted. postdischarge suicidal behavior. Stanley and Barter compared adolescents Life Stress and Social Supports who made a suicide attempt after discharge (N=15) with adolescents who Most of the previouslydescribedstudies in- were pre- dicate an association betweena variety of viously suicidal but made no suicideattempts after discharge (N=21) and with initially family problems and suicidal behaviorin nonsuicidal adolescents who did children and adolescents but they donot in- not make a dicate whether theoccurrence of these suicide attempt after discharge (N=24).The problems during specific periods of the adolescents who made a postdischarge youngster's life is an important element suicide attempt had less adequatepeer rela- as- tions than th other previously suicidal sociated with suicidal behavior. Ofnote is the study by Rosenthal and Rosenthal (24), adolescents (Xh = 6.61, df1, p.05) and the nonsuicidal controls (X4 which indicated thatsevere stresses of abuse = 5.02, df = 1, pic.05). In addition. the repeat suicide and parental rejection werecommon in the attempters were less likely to be living with histories of preschool children whoex- pressed suicidal ideas and/or attempts. their parents after discharge than,the other previously suicidal adolescents (X.4= 4.007, Only a few studies evaluate the relation be- df = 1, rt.05) and the nonsuicidal adoles- tween developmental periods, social support cents (XL = 4.7, df = 1, p<.05). These find- stresses, and child and adolescent suicidal be- ings support the notion that early and havior. Table 2 indicatessome of the recent continued lack of social supportsare impor- studies that address this issue. tant factors related to suicidal behavior in adolescents. Stanley and Barter (29), ina chart review, compared 38 adolescent psychiatric in- Cohen-Sandler, Berman and King (31) provided a longitudinal perspectiveon the

2-73 77 Report of the Secretary's Task Force art Youth Suicide

types and degree of stress in social supports The main findings revealed that 50 percent associated with youth suicidal behavior. In of the subjects with a history of early paren- this retrospective chart review study, 20 tal loss were preoccupied with suicide and 18 suicidal psychiatric inpatients, 5 to 14years percent made one or more suicide attempts. old, were compared with 21 depressed non- There was more suicidal ideation among suicidal child inpatients and 35 psychiatric in- those subjects who experienced both an early patient controls. The two comparison loss and greater family instability after the groups, who were hospitalized in the same in- losses. When there was greater family inpatient unit, provide good controls for the ef- stability before, during, or after the parental fects of depression and other factors loss, suicidal ideation was more frequent than emaciated with 12,Nchiatric inpatients. Only in those subjects who experienced a restora- 38 percent of the depressed inpatients tion of family stability after the loss. engaged in suicidal behavior but 65 percent of the suicidal children were depressed. Among the rare studies of adolescent suicide, Thus, depressed children were not necessari- an innovative approach was taken by Salk ly suicidal, and not all of the suicidal children and colleagues (32) in their study of prena- were diagnosed as depressed. tal, birth, and neonatal distress factors and their relation to adolescent suicide. The The results revealed that the suicidal birth records of 52 adolescents who com- children had higher lifetime stress scores mitted suicide before age 20 were compared than the other children and that by thc age of with the records of 52 adolescents whose 4-1/2 years, the suicidal children began to ex- births preceded and 52 adolescents whose perience more family stresses. Furthermore, births followed those of the index subjects. in the year prior to hospitalization, the The results indicated that the suicide victims suicidal children had significantly higher had more prenatal and postnatal problems degrees of stress than the other inpatients. than the two control groups. The three most This study suggests that measures of life common problems found among the suicide stress can distinguish suicidal from non- victims were respiratory distress for more suicidal children and that particular types of than one hour after birth, no antenatal care stresses were important. Loss of adult sup- before 20 weeks, and chronic disease of the port was an important factor and was a con- mother during the pregnancy. The types of sequence of hospitalization of a parent, chronic oiseases for the mothers of the death of a grandparent, and birth of a sibling. suicide group included chronic persistent Furthermore, problematic parenting could anemia, rheumatic fever at age 12 residual be inferred from the fmding that the suicidal heart murmur and arthritis, chronic anxiety children were more hlely to have parents with treatment, kidney infection, polio, who abused drugs and/or alcohol. chronic hypertension, asthma, high blood Adam and associates (30) used a high risk pressure, repeated gonococcal infection, population to determine the relation be- muscular dystrophy with incapacitation, mul- tween early losses and suicidal ideation. tiple surgery, extreme obesity, infectious They intaviewed 41 students, referred to a hepatitis, and fibroid uterus. While it was university mental health service, who noted that socioeconomic factors or family reported the death of at least one parent conflicts did not distinguish the groups, no before age 16 years, and compared these stu- specific information was provided about dents to age-, sex-, and religioo-matched stu- specific psychosocial variables before or at dents at the same clinic; the control group the time of birth or during the lifetime of the included 35 students with a history of paren- adolescents. This limitation makes it diffiafst tal separation and/c r divorce before 16 years to evaluate the results with respect to of age and 61 students from intact homes. whether the reported features of infant pre- and postnatal distress were indicators of more fundamental features of psychosocial

2-74 78 C.R.Pfeirer: Family Characteristics & Support4stems... stresses associated with adolescent suicide. tempted suicide, two made suicide gestures, Nevertheless, this study reinforces others and eleven exhibited suicidal ideation. that indicate the important relation between These two studies support the need for fur- early childhood stress and risk for adolescent ther research into the association between suicide. family violence and youth suicidal behavior.

Child and Adolescent Abuse Parental Psychiatric Symptoms Although violent deaths attributed to acci- and Disorders dents and homicide are the leading causes of A number of studies evaluating multifactoral death in adolescents (33), the relation be- elements associated with childhood and between family violence and youth suicidal adolescent suicidal behavior have suggested havior has been relatively neglected as a topic that the parents of suicidal youngsters exhibit for investigation. 1\vo studies, described in a variety of symptomsthat include depres- Table 3, provide information about this issue. sion, violence, alcohol and drug abuse, and Deykin, Albert, and MeNamarra (35) sur- suicidal tendencies (14-17,22-24,26-28). veyed the Massachusetts Department of So- These fmdings are in keeping with a number cial Services records for evidence of contact of studies of adult suicidal individuals that in- with this agency for abuse and/or neglect for dicate a consistent relation between family 159 adolescents admitted to an emergency suicidal behavior, family affective disorders, service after a suicide attempt. For each of and an individuars suicidal behavior (36-38). these adok-scents, two age- and sex-mc:ched Surprisingly, even in view of the important comparison subjects treated for other medi- implications of these fmdings, there has been cal conditions in the same emergency service a relatively minimalnumber of investigations were studied. The suicide attempters were of the family pedigees for psychiatric approximately 5 times more likely to have problems among suicidal children and contact with the MassachusettsDepartment adolescents. of Social Service. The strength of this study Table 4 highlights two recent investigations is that the data from the social service depart- of types of parental psychopathologies found ment were unbiased because the contacts among suicidal youngsters.These two were recorded before thesubject's emergen- studies (39,40) suggest that parental cy room visit and bypersonnel who had no symptoms of depression, alcoholabuse, awareness of the futurepsychological status suicidal impulses, and chronic psychiatric ill- of the subjects. The study provides strong ness are importdutiezturet associated with evidence for an association between child the histories ,..)f adolescents who attempt abuse and/or neglect and adolescent suicide suicide. These studies suggest that further attempts.However, it could not discern investigation of the psychiatric problems of whether there were differential effects of parents of suicidal youngsters iswarranted. abuse or neglect with regard to adolescent suicidal behavior. Mother approach of study is to use a high risk population involving the offspring of An earlier study (34) used a high risk popula- parents with psychiatricproblems. Most of tion to evaluate self-destructive behavior in the current research employing this 60 physically abused children, 30 neglected methodology involves children of parents children, and 30 normal children. In this with affective disorders. These studies vary study, Green noted that there was a higher with respect to the types of comparison incidence of self-destructive behavior in groups, methods for assessmentof the abused children than in the two comparison youngsters, ages of the youngsters who were groups.Five of the abused children at- evaluated, time of parental assessment in relation to the parents' degree ofillness,

2-Y5 79 Report of the Seaetety's Task Force on Youth Stilcide criteria for diagnosing the youngster, and de- of cases, child care problems were identified gree of blindness of raters.Nevertheless, before the suicide attempt. However, there consistent trends in the results of these were no differences among the mothers for studies have emerged. other clinical chracteristics such as history of Children with an effectively disordered psychiatric treatment, previous suicide at- parent are likely to exhibit behavior tempts, alcoholism, and drug addiction. The problems and psychopathology (41-49) and main finding; suggest that mothers who attempt suicide may be more at risk for serious have a distinct risk of developing an affective disorder (47,50-52). Child-rearing patterns child care problems involving child abuse of effectively disordered parents include dis- than mothers in the general population or for turbances in mother-child attachment (53), mothers who are at risk for depression. An lack of parental encouragement of the child's implication of this study is that there may be ability to regulate affects (48,53,54) and un- a relation between parental violence and stable caretaking characterized by parental parental suicidal behavior.Furthermore, hostility, abuse, rejection, and separations since the children were all 5 years of age or (41,48,49,54,55). younger, no information was provided about the children's potential for suicidal behavior, L. spite of these findinp, the relation be- a behavior that other studies have associated tween youth suicidal behavior and parental with child abuse. affective disorders and/or suicidal behavior has been a relatively unexplored issue. Table There is only one study of the effects of 5 highlights the results of the existing studies. parental suicide on children (57), and it has Only two studies of effectively disordered a number of limitations. The main parents (44,47) mention that there is a higher methodological shortcoming is that data about the children were obtained from the association of suicidal tendencies in children of depressed parents than nondepressed surviving parents rather than by direct assess- control parents.For example, rates of ment of the children. The results indicate a higher frequency of psychiatric referrals for suicidal ideation of 6.5 percent to 9 percent and lower rates of 0.9 percent to 3 percent the children of the suicide victims than for for suicide attempts were noted for children comparison children whose parents were of depressed parents. These studies did not recruited fivm general medical practice and evaluate whether there were cenain paren- matched to the suicidal parents for age, sex, tal characteristics, such as the presence of and marital status. No chibilen of the suicide suicidal tendencies, that distinguished the victims, however, exhibited suicide attempts. depressed parents with suicidal younpters Family life was more unstable before the from those depressed parents without parental suicide than among the comparison suicidal children. families. This feature suggested that the effects on the children may be related not to The relation between parental suicidal be- the parental suicide as a sudden isolated dis- havior and quality of child care has been min- aster but rather to a chronically stressful en- imally studied. However, some preliminary vironment in which the suicide was a major data exists on the relation between parental event.Since this study had serious suicide attempt and child abuse (see Table 5). methodological problems, the results must Hawton, Roberts, and Goodwin (50 looked be considered to be quite tentetive. in the child abuse bureau records for reports on mothers who attempted suicide, mothers DISCUSSION at risk for depression, and control mothers who were not depressed or suicidal. There This review of renarch on family variables was significantly more child abuse reported and social supports as risk factors for suicidal for the suicide attempter mothers than either behavior in children and adolescents indi- of the comparison mothers. In the majority cates that there is much variation in the re-

2-76 0 C.R.Pfeffer: Family Characteristics & Support Systems... search methods, populations studied, types underlying the relation of these family of suicidal behavior investigated and manner characteristics and social supports to youth in which statistical results are reported. In suicidal behavior. For example. early loss of fact, most studies are of nonfatal suicidal be- social supports may be an important stress havior smd, therefore, most of the findings that affects the development of personality mu4t be considered to be about putative risk characteristics and/or bioloecal systems and factors. mowever, a number of consistent thereby enhances chronic vulnerability to trends have emerged when the issues are ap- suicidal behavior. On the other hand, an praised from the multiple perspectives of in- acute loss of social supports may alter exist- vestigations of suicidal younpters, offspring ing psychological and/or biological function- of parents with psychopathology, and ing.This may create a temporary and an retrospective accounts of adults. acute crisis in ego functioning that affects the quality of a youngster's affect regulation, im- The studies reviewed in this report support pulse control, judgment, cognition, and fan- the notion that the nature of family charac- tasies. teristics and social supports are important factors associated with youth suicidal be- Another mechanism may involve genetic vul- havior. Family factors associated with high nerability. In this case, stress involving fea- risk are related to experiences characterized tures of parent-child discord may be of by the presence of intense levels of stress. secondary importance.For example, the Such stress, suggested by the investigations mechanism involving the relation between of children, adolacents, and adults, appears parental violent abuse and youth suicidal be- to be chronic and/or occurs at an early phase havior may be determined by some factor, in the life of the individual. An important im- such as a biological correlate, and the stress- plication of these findings is that future re- ful parent-child experiences may be of secon- search on family risk factors for youth suicidal dary value.In connection with genetic behavior should adhere to a developmental vulnerability, there must be an interaction perspective. Such an orientation can with levels of experiential stress to promote facilitate the examination of continuities and risk for suicidal behavior. discontinuities in family variables that may be Evaluation of these merhanisms requires dif- precursors to youth suicidal behavior. ferent research strategkes than those used in Studies suggest that a variety of stresses exist. the investigations reviewed in this report. One type is related to loss of social supports Prospective longitudinal research designs of through death, parental separation and/or high risk populations are indicatoi. divorce, mobility involving school changes, studies may evaluate the long-term oilt and problems with peer relationships. A of children who previously had suicidal ten- second type of stress is associated with dencies, as well as evaluate children who are variability in parental functioning that is liked the offspring of parents prone to abuse, af- to parental psychopathology. Thisaffects fective disorders, separation/divorce, or the quality of the parent-child interactions death. The time required and cost invoktied and/or may necessitate the temporary ab- in these irvestigations can be manageable, sence of the parent who requiresinterven- and the yield of potentially valuable information that removes him or her from the home. tion may be great. It also involves violation of personal boun- From a more theoretical perspective, youth daries best characterized when a youngster is suicidal behavior is a multidetermined witness to, or the victim of, sexual or violent symptom with many associated factors.The abuse. studies described in this review focus mainly Important shortcomings of the studies on risk factors and do notaddress issues in- descnIed in this report are that they do not volving protective variables against suicidal address hypotheses about the mechanisms behavior.Factors associated with inval-

2-77 R I Report of the Secretary's Task Force on Youth Suicide nerability of an individual, such as ways in 2. Dorpat TI., Jackson JK, and RFiley HS: &Oen bones and attempted and oompleted suicide. Archives of which adaptive skills can be enhanced, must General Psychiatry 1985; 12:213218. be documented. Such an approach may be 3. Greer, S: Parental loss and attempted suicide: A consistent with a theoretical framework for NM* report. British Journal of Psychiatry 1985; 112:465- 470. youth suicidal behavior that suggests oppos- 4. Levi LO, Pales CH, Stein M, and Sharp VH: Separa- ing vectorial components related to the ex- tion and attempted sulckle. Archives of Genera/ Psychiatry pression of the suicidal symptom.For 1955; 15159484. 5. Crook T, and Rasidn k Association of childhood example, qualities of a supportive individual and parental loss with attempted suicide and depression. such as an abffity to provide empathy, consis- Journal of Consulting and Clinical Psychology 1975; 43:277. tent availability, capaiity to set limits and 8. Luscomb RL, Oum GA, and Patsloints T: headlin- offer structure, andbility to gratify in- ing factors In the relationship between Ole stress and Wald* attemptia4T1445wr turnal of Nenrous and Mental dividual needs to enhance self-esteem may Disease 1990; 1 . be important social support vectors that 7. Goldney RD: Parental loss and reported childhood stress in young women who attempt suicid. Acta prevent suicidal behavior. The factors Psychiatric* Scandanavia 1981; 84:34e. described in this review such as family disor- 8. Roy k Earty parents) death and adult depression. ganization, parental psychopathology, and Psychologial Medicine 1983; 131-5. 9. Weems JA. and Wdrow LEarttf parental dis- family violence are vectors that enhance cipline and adult seitdestructive aci.s. The Jeurnal of h suicidal behavior. This model of opposing vous and Mental Disease 1985; 173:74-77. vectors, illustrated below, may elucidate ap- 10.Hill, OW The association of childhood bereavement with suicidal attempt in deOla Illness. British proaches to assessment and interventions for Journal of Psychiatry 1989: suicidal risk. 11. Birtchnen J: The relationship between attempted cticide, depression and parental death. British Journal of Finally, important implications for preven- Per:Misty 1970; 11&307-313. tion of suicidal behavior among children and 12. Adam KS, Bouckoms A, and Steiner D: Parental loss and family stability In attempted suicide. Archives of adolescents can be derived from these GOMM Psychiatry 1992; 391081-1085. studies. There is a need for efficient screen- 13. Teich*/ JD, and Jacobs J: Adolescents who attempt suicide: Preliminary findings. American Journal of ing techniques to identify high risk familia. Psychiatry 1988; 1=1248-1257. Also of value would be screening devices for 14.Pfeffer CR, Conte HR, Plutchik R. and Jerrett 1: Suicidal behavior In latency-age children: An empirical identifying high risk children and adolescents stud . Journal of the American Academy of Child in such situations. Another approach is to Psychiatry 1979; 18:P79462. plan interventions that focus on ameliorating 15.Pfeffer CR, Conte HR, Plutcht:: R and Jerrett I: Suicidal behavior In latencrage children: An outpatient family disorganization and parental population. Journal of the knerican Academy of Child psychopathologies so that stress may Psychiatry 1980; 19:703-710. 18. Car1son OA, and Cantwsll DP: Silk:kit behavior decrease and stability may be enhanced. and depressicn In children end adolesoents. Journal of the American Academy of Mild Psychiatry. 1982; 21:381- 388. 17. Garfinkel BD, Frogs* A. and Hood J:Suicide at- REFERENCES tempts in children and adolescents. American Joucnal of Psychiatry, 1982; 139:1257-1281. 1. Greer & The relationship between parer/NI loss and attempted suicide: A control study. British Journal of le.Miller Min Chiles JA, and Barnes VE: Suicide m- Psychiatry 19134; 110.698-705. tempters within a delinquent population. Journal of Con- sulting and Oinical Psychology. 1992; 4:491-498. Risk Factors for Increased and Protective Factors for Decreaud Youth Suicidal Rkk

increased Youth Dsc reseed 'Usk Facials For:41- Suicidal -0.Protective Factors For-ct.I Suicidal Risk Risk Riak I. Um of Social Support 1. Presence of Social Support a. Death a. Empathy b. Parente Separatioa/Divorce b. aystant Availability c. c. Linsit iif,ceigt)mbakr d.EnvirctmenettinlialStructure 2. Variability in Parental Penctiunins 2. Individual Adaptive Skills a. Affgthe Disorders a. Appraisal ci Stress b. Suicidal Tendencies b. 5**1 Abernathy Solutions c. Alcohol Abuse c. High Frustration Tolerance d. Stir-esteem 3. Violence 0. Good Impulse COMM a. Sexual Abuse b. Physical Abuse ,ERMIONIMIN.MM11111111.11=11.111PMMI.S01.1.1.

2-78 82 C.R.Pfehier Family Characteristics & Support Systems...

20. Pfeffer CR, Solomon 13, Plutohik R. Mizruchl MS, 41. %Wiseman MM, Paykel ES, and Merman CIL The and Weiner k Suicidal behavior in latency-age psychiatric woman as a mother. Social Psychiatry, 1972; liftman* A replication and wow validation. Journal of cP=d tha American Aoadsmycf Child Payohistry, 1902 21:584- 42. Garmszy N: Children at risk; the search for an- 562 tecedents of schizophrenia U. On oing research 21. Kooky ft Childhood suicidal behavior. Journal of issues, and intervention. Sdzophmnia But- Child Psychology and Psychishy, 1963 24:457-468. 01,1119A;9;55-125. 22. Pfeifer CR, Mutat* R. and Misruchl MS: Suicidal 43. Garmazy N, and Streitman $ : Children at risk. The and assauttive behavior In children:ClassificsAlon, saarch for the antecedents of schizophrania I. measursment, _and interrelations. American Journal of models and research methods.SchisaphrenCIMe=1, Psychiatry, UM; 140114-157. 1974; 8:14-90. 23- KOMI CR, Zuckerman Si, Rutohlir I% _end Miscued 44. Wainer Z, Weiner A, McCrary D, and Leonard Mk MS: Suicidal bahavior in normal school chmrm: A oom- Psychotherapy in children of inpatierits with depression; parboil with child psychiatric inpatients. Journa: of that A oontroliod study. The Journal of Nervous and Mental Dis- Amwican Academy of Child Psychiatry, 1984; 23416423. ease, 1977; 164:400-413. 24. Rossrahai PA, and Rosenthal & Suicidal behavior 45. Decina P, Kestenbaum CJ, Farber S, Kron L., Gar- In preschool children. Amnon Journal of Psychiatry, vin M. Sackeim HA, and nave RR:Ctinical end 1964; 141.620-52E nsychologIcsl assessment of children of bipolar Journal of Psychiatry, 1903. 14e548. 25. Taylor EA. and Stansfelo bk Children who poison themselves. I. A clinical comparison with psychiatric controls. British Axonal of Psychiatry, 1984; 145127435. 46. Weissman MM, Lockman JF, Merikas KR, Gammon OD, and Prusoff Bk Daprestion aM- anxiety 26, Myers KM, Burke P, and McCaulay E: Suicidal be- disorders In parents and children: Results from the Yale havior hospitalised loreadolescant children on a Family Study. Archives of General Psychiatry, 1984: unit Journal ofthe Amarican Acadarny of Child 41:845-852. , teak:4:474-480. 17. Weissman MM, Prusoff BA, Gammon GD, Merikan- 27. Shaft' OA, Carrigan 8, Whittinghall JR, and Darrick gas KR, Lockman JF, and Kidd KK: Psychopathology in k Psychological autopsy of completed suicide in children the children (ages 6-18) of depressed ancl normal parents. and adolescents. Arned: in Jammu: of Psychiatry. 1985: Journal of nw American Academy of Child Psychiatry, 142:1061-1064. 1984; 23:78.81. 28. Pfeffer CR, Newcom J, Kaplan G, Misruchl M ,S and Zan-Waxier C, McKnaw DH, Cummings EM, Plutchth Ft Suicidal behavior in adoliscant payohlatric in- Davenport YB, and Radko-Yanow M: Problem Whaviors patients. 1986; presented at the Annual Meetna of the and past interactions of young children with a manic- American elsychistric Association, Washington, D.C. daprass2zZejarent American Journal of Psychiatry. 1984; 21/.Stanley EJ and Bener JT: Ackiescent suicidal be- 141: hai4or. Amarican Journal of Orthopsychlatry, 1970; 40:87- 49. Williams H, and Carmichael k Depression in mothare in a mute-eh:do usban Industrial municipality In 30. Adam KS, Lohrenz JO, Harper D, and Strainer 0: Whom*. Aetiological factors ad affects on infants and Early parental loss and suicidal idaation in urn Putty at* preschool children. Journal al Child Psychology and dents. Canadian Journal of Psychiatry, 1982; v.mai. Paydliatry. 1985; 26:277-08. 31. Oohen-Sandier R, Berman AL. and King Rk Life 50. Cytryn L, MoKnew DH, Bard* J411, Lamour M, and stress and syrnptomatology: Determinants of suicidal be- Hamovitt J: Offspring of patients with affective disordars bawler in children. Journal c d the Myriam Academy of D. Journal of the American Acadamy of Child Psychiatry, Child PsYchildry, 1952; 2178-186, 1982; 21:389-391. 32. Salk Li Lipson L, Stung WO, Reilly BM, anu Lova! S. Lockman JF. Weissman MM, Prusoff BA, Caruso RH: Ralationship of =tarnsl and prenatal conditions to KA, Winking** KR, Paula DL, end Kidd KK: Subtypes of evantual adamant suicide. The Lancet, 1985; March 15. depression:Family study perspectIve. Archives of General Psychiatry, 1984; 41; 33. Monthly Vital Statistics Report-National Center for Human Services, 19134; 33. 52. Wordstar WR, Warman GL, Kellar MO, Likvori PW, and Padorefsky DL: But are they ? Validity at DSM 34. Ocean AH, Self-dastructive behavior in Wailed children. American Journal of Psychiatry, 1978; 135:579- iii=ande=nAt of ah"IdenPsychiatry, 1 142687-691.familY study. 53. Gaensbauer TJ, Harmon RJ, Cytryn L, and Mc- 35. Deytin EY, Alban JJ, and Mc:Namur& 11: A pilot Knew DM: Social and affectiva development in Infants study of tha effect of exposure to child abuse or neglact with a manio-dapressive parent. American Journal of an adolescent suicidal behavior. Amaricar. Journal of psychiatry, igos; 141:223-229. Psychiatry, 1985; 142:1299-1303. 54. Davanport 103, Zahn-Waxiar C, Adiand ML and 36. 1=efr3E, and Wain! RD:Family history of Mayfield k Early child-rearing practices in famines with a suicidal mono suicide attemptars. Tha Journal manic-cfspressive parent American Journal of Psychiatry, of Nervous and Mental Masse, 1962 170:86-90. 1984; 141:230-234. 37. Roy k Family history of suicide. Archivas of 55. Ohodsian M, Ulla* E, and Wolkind 8: A ion- General Psychiatry, 1963; 40171474. gitudinal study of maternal ftrusion and chlid behavior 38. Ray k Famihr history of suicide in manic-depres- problams. Journal of Child Pridtology and Psychiatry. siva patianta. Journal of Affactive Disorders, 1985; 8:187- 19N; 25:91-109. 189. 56. Hawton K. Roberts J, and Goodwin 0: The risk of 39. Pallier CL and McKenry PC: Parental negative child abuse among malhars whonlitezst suicida. British salt and adolescent suicide attempts. Journal of the Journal of , 1905; 148: Amarican Academy of Child Psychiatry, 1983. 21:404-406. 57. Shaphird DM, end Barraclough BM: The after- 40. Friedman RC, Corn R, Hurt SW, Rho! B. Schaff* math of parental suicidal tor children. British Journal of J, and Swirsiry S: Famigi.chirla of nine's in the sariously Per:1110y, 1976: 129:267-278. suicidal adolescent: A I nal of Onhopsychiatry, 1964;53=.. Am.rican Jour*

BEST COPY AVAILABLE 3 2-79 Studies of Multiple Factors Associated with Youth Suicidal Behavior

Definition Time of Suicidal Population Sample Comparison of Data Citation Behavior Studied Size Group Study Source Test Rnuits Statistics

03) 'Nohow. J. D. and None Oren 20 adolescent 55 noneuioldal 1554 Inleninte el 44% at slides alleneris had a IMO, or No oompadson Amato, .1. Adoissoents Inpatients with *claimant In- adolescents Mend Woo ~treed ar camminsd beam *ha allempt &Odd* suicide el- patients matched and peon& suicide and In 251,111m woe a suicide samba polinansoy findkvo. too age, me, seir. eillMit by a pm* PoNyeedio soda dosulbsd. elealatible. I got SES looketen ~Om t2Z124114257.

04) Feethe, Thoughts ar acts 42 suickial 58 children nosuiddel 107* treardero at No Merano to parents' operations at Owls, H.R., POMO; which may Ned to children. 12 yore ol* children tf.11 chadon and suicidelreaqard nanilulcidd (7n1) R, and JOWL I. dadh or oda* In psycNlfIcIn mon& th Outddol behei/or hi Iury. A amino ot patients. Where at suicidal children ay -11 t..2.11 p.c .03 talencrega cNithen: suicidal What% &premed. an eincialcal study. IL no defined thoi aseeciodon Moen scaly at child t -2.el. P .011 ilEttethEASMS eluded non suicide, Woke od paninial Oars& alga 107itilice70- sufddakofficklei rdon and adoldeltandenalee. idpee Owed% mild 411%oulddd cOldien and in nen- 1.02.0t.p.cM and *edam d- Wad& *edam monied about poor eem" school potenswinir&

(UR Pieter, C.R, Suiddat bohemia' id suicidal m- 30 child 20 noniulcidel 1077- freer** et No Oloranca In waned &voodoo, included thoughts ono& poychlelde out- eutpetlents. pawn and poodeoldo hospillislonoicahoedisc FL, Jarred, L &MOM indoor sets ankh patients. ego & chlkiren. abuse. mold operaeori% beherior Ni Islancy-ses may hod lo death 12 yam In otaisede, and child Male In MOM Madmen oditiopori er seit-infuo. A ma childnin. p < .05 PoMiNiori- ealectoim at hooped °NW. SWOON ~an hod Mons perallei faiiLeittLessel% suicidal behaider **Add Melon then nansulcidal Heft 110703.710. was defined. childran.

(to Wien. GA and Suicide enamor 22 youngOo 102 daildnao 80 nonoulMiot I. 1577- *mime at 50% old* elm* had Oaks *eh Not given. Canto& D.P. Suit** van an oat leading sampted and adobe- tempter& 19711 children end depression end elochellsin compared la behavior and doom- esury. suicide. cane In went* 50%ln nonoloriapor row no dit clan In chedien and psychloic in- Orono below glow. scloissoente.I. Amos, patient and out- Mat_CMILEIretit MOO am leiU 21211146S

(17) S.D., Dellemle sett-In- 505 children 1010 children 505 nen-sulcsidel 1970- Hospital Mort Suicide elhimplin had 11101e1analy From, A. and Hood. Noted Inluo eah and adorn- and adobe. Youncloiolo 1577 oda,. too at mental Illness 131.1%, N. 442) V2.. 122.3, .1. Suicide stoma in censdous lecielen °onto in enver- cents. matched tor ag, then ricoallsmpito (119.0k NOP 452). dr -1, peln children and isione- la dia. gamy mice odic and limo of SulcIde ~Om Put mote lionly cote. &Aga tar outdid* A- evaluation. lacy *medical Info pi." 445) x2.3.153. !Ma& IM %mos (mon than nondesmptec, (44.1914. N 401). pe .05 136e1287-litet. se 154 ),ere - Eluickte allecipOno had moo tandy hi* onge 0-21 tory at WOW MOW* P-11114, N 443) X2 - 23.95, then nondilompleos (1.1%, N.442). ce-2,p<.01 Suicide. ~Om had mos palm* urismplorient (KM N then x2.asa nonallompleo p.7%, N MC cr.1.is< .01 Suldde atemiplent hed moo pond, F 4 shame In .o 444 then nonimplies X2.1203. P.437). dt-3.p<.001

Table I.

BEST COPY AVAILABLE

Report of the Secretary's Task Force on Youth Suicide JALfiepit ti.1511-144 v Vtgbi s § dIuln i 11111111'111 1 II 1 liii I 6111111111 111:11/1111I mg 1III 11 II I 1111III] e ihillillutiJifIfihlilt

2-82 C.A.Pleffer Family Characteristics & Support Systems... elvin'(41515 4 a 1 I. 9:i I

eetell 111111114 g Wit do oh;Via; Ifigidit 2-83 Report of the Secretary's Task Force on Youth Suicide c Et op) in eilc lea lit et< ft 11111; iii 2-84 C.R.Pfeffer: Family Characteristics & Support Systems... two.,1:7;

pit- 1 glit211 As. Enda1111151 2-85 Report of the S3eretary's Task Folcie on Youth Suickle li I f iti ii f i la hi IA 10100h011105Iivolitiotilli WI Op IIIIMglith 1 1 d 111111 fangMit i /111111II 1 HOfl 1; hI a h 111111101 111"9 IiHill Ih lb/ninth 1.1 1 ifilillig 2-86 I IA 1 iciiiIIIIII Suicidal Behavior of Children with Depressed and/or Suicidal Parente

Definition Time of Suicidal Population Sample Comparison of Data Citation Behavior Studied Siza Group Study Source Test Results Statistics

(44) Weiner, 1 Weiner, No *Nalco given. 75 Wee 227 ea normel 1975 inerMerme at 7% ol children at dimmed pawls met A.. McCrory, D. and children at 29 children at at went% Newman aline tor depronson. Leonard, NA parents Mani- parw t. ers were Suiddal Ideas more common In Psychopelhology In tided tor not haeptalleed probends pep than control childon pc .02 Nekton at Inpatients depieselon. or depressed. (i114. Nu Merman In suicide 'newts NM depression: a con- imam proton* (3% or coerce children doled etudy. 'Lang: (194- or aistsliferal gams ierr; 154:406- 413.

(47) Weissman. NAL. No definition Wren. 107 children 00 194 eilliken 87 children ot 40 Cuestiorindre Children of depressed parents hed dote Pune, BA- Gononon, depressed nannal parents ackninlotered emplane (WA Von children at normal pc .01 G.D., Meniangss. Ka. Pawls OW* (ageo of children to parent parents (19.1) and 00111111diegnonse et .01 locranen. J.P. md of children OAS 11.18 years). morn child. (34,2 end &I feepeolledh. Cdiviion POdtk K.K. Yam). MN disposed ot children at Paychoostheld9Y in deposed pawls swe: major dom. the ciddren nes 0- don O&M, slinlion deal disorder t8) ot depredired end Cia.31111 one separsion annialf (10.3%). normal parents. ,Lka, betisdar was repotted In 61iPLIlit ftrObie. childnon at deposelems but nal a/ nor- 1954, 217844. male. Mat atildisn at deprsookse moiled aikido/ Wass. Mee children at demeans Ilweelensd NOW% OAS of children at depreselmi ellerMed suicide.

(50) Holston, K., Ns definition Oen. 1 14 mochas 223 45 wend 1961- Load died Abuse doeumenssd In 20% at edemples 22- 8.40 Roberts, J. end Good- din d Wee mothers vAo gm 1902 abuse *dee mothers and 0% of men% dt.t. pc.01 nen, G. The risk ct child one child av 5 birth In earn* records and Abuts ruler before enemata An N- ahum among mothers ar under, In Mae end times words at *mot modem than She maws st Ask X2 3.09 seto attempt suicide. emergency sir- as prebend local for damask*. rft, p4 .05 Oft J,or Psychig. wins tor suicide mothers, and tet childrerVe 1985; 14e405-459. allenept. maim, eith a hospitat chad me 5 ar under *honor* N tor clepreesion.

(57) Shephent MIA Documented 30 children 185 150 children 1970 2 intenkrer No dilld at suicide Acem dimpled and amedough. D.M. suicide. whose parent sham parents WM sunreing suicide. one child made threats. Nigher X2 .- 5.32 The allommeh of won- comnOted em makhed parents, one frequency at peychiskIc *mob tor d I- 1. pc .02 tel suicide tor children suicide new* nelfr the suicidal erase** a kw children ot suldde *time than controls. 9dLAZIY019. 1976; child roe 2-17 parent ke sp. weals Mu Moist sspaisNon and disewmony 121121174715. yews old. Nit, meld Niko deed% second wester before roulade (50111 than In X2 11.911 ond drawl from InterVew nee coital% rim 1, p4.01 general vodka 5-7 yews Oar relibistek death.

Table S.

BEST COPY MAKE CONTAGION AS A RISK FACTOR FOR YOUTH SUICIDE

Lucy Davidson, M.D., &LS., Medical Epidemiologist, Division of Injwy Epidemiology and Contm4 Center for Envimamental Health and Injwy Contra, Centers for Disease Control, an4 Clinical Assistant Professor in Porhiatry, Emory Univenity School of Medicine, Atlanta, Georgia

Madelyn S. Gould, Ph.D., M.P.H., Assistant Pmfessor of Clinical Social Sciences, in Psychiatry and Public Health (Epidemiology), Columbia Univers* Colkge of Physicians and Surgeons, and Research Scientist, New York &at, Psychiatric Institute, New York, New WI*

INTRODUCTION "The sole approach to the youth suicide Until recently, contagion as a risk factor for problem lies in recognizing beforehand the puth suicide had not stimulated much public susceptible individuals and in their proper awareness or research interest.However, management." Harry Bakwin reached this widely publicized clusters of youth suicides in conclusion in his seminal article entitled, places such as Plano, Texas; Westchester "Suicide in Children and Adolescents" in County, New York: and Omaha, Nebraska, 1957 (1). In 1987, we are still struggling with have focused attention on the possible role these tasks and, during the intervening three of contagion in suicides that occur close decades, our task has expanded enormously. together in time and space. Suicide has become the third leading cause Suicide contagion is a hybrid term that ap- of death for persons in this country aged 15 pends to suicide the medical meaning of con- to 24 (2) and reducing youth suicide is a tagion as the transmission of a disease priority objective for the Department of through direct or indirect contact. Although Health and Human Services (3). We are still suicide is not a disease, per se, applying the working to clarify risk factors for youth infectious disease model to suicide contagion suicide in order to identify *susceptible in- can clarify for whom and through what sorts dividuals' and develop effective suicide of contact the likelihood of suicide is in- prevention programs. In this paper we focus creased. Components of the infectious dis- on the processes by which one suicide be- ease model that illustrate analogous factors comes a compelling model for successive in suicide contagion are host susceptibility, suicides or facilitates other suicides. We modes of transmission, degree of virulence, review two bodies of literature: (1) reports and dose dependency. of suicide epidemics or clusters for evidence of contagion and possible mecharlsms and Host susceptibility measures an individual's (2) research on the effects of suicide stories intrinsic ability to ward off or resr.. an illness. in the mass media. The term "suicide con- For example, an immunized child would not tagion" has been used to describe this pat- be very susceptible to measles, even if ex- tern. posed to an outbreak of measles at school.

1CU 2-88 L.Davidson:Contagion as a Risk Factor for Youth Suicide..

For adolescents, host susceptibility to suicide inapparent infection to severe clinical illness Ismultidetermined. Genetics play a partwe or death (4). PCISCMS susceptible to suicide know that depressiona common antece- contagion may have an "inapparent infec- dent to suicide and that some forms of tion" and not be registered among suicides or depretnion have a strong genetic component. suicide attempters. The consequences of Good baseline emotional health might be suicide contagion may be inapparent, either like an effective immune system in being because the illness was arrested before the highly capable of warding off challenges. appearance of effects (suicides or suicide at- Fmally, the cognitive and affective ability to tempts) or because the apparent effect was identify and speak about feelings may make unobsencd, ignored, misclassified or un- an individual less susceptible to suicide. reported (an actual suicide recorded on the death certificate as an accident). Inapparent Infirtious agents have difkrent modes of infections are not available for study and transmission. We can classify routes az direct their absence may result in an under- (person-to-person) or indirect. Person-to- statement of the significance of contagion as person spread may be implicated in sub- a risk factor for youth suicide. sequent teen deaths following the suicide of another member of the same social network. The suicide of someone famcus, such as Evidence of Suicide Epidemics c Marilyn Monroe, may be an indirect ex- Clusters posure to suicide for millions of people. If contagion does play a role in suicidal be- Thus, various suicide contagion pathways havior, one might expect clusters of suicides may ocist: direct contact or friendship with a to occur. That is, isII excessive number of victim, word-of-mouth knowledge, and in- suicides would occur in close temporal and direct propagation through the media. geographic proximity. Evidence for suicide contagion has been reported in accounts of Infective agents differ in their degree of epidemic suicides from ancient times virulence. Beta-hemolytic streptococci are through the 20th century. In England in 665, more virulent than other types of streptococ- distraught persons crowded to the seaside ci. Similarly, for youth suicide, the virulence cliffs and threw themselves over. They of the agent may be greater when the first preferred a speedy death to the lingering tor- death in a potential cluster is that of a highly ture from the plague which was rampant at esteemed role model, such as the class presi- that time. In 1190 in York over 500 Jews dent, rather than a loner who was always per- committed suicide to avoid religious persecu- ceived as odd ar disturbed. tion. In 1928, an epidemic of 150 suicidal The likelihood of an infection dose de- drownings in the Danube occurred during 2 pendent. Not all persons who consume salad months. The epidemic was finally conrolled contaminated with staphlococci will get food by establishing a 'suicidal flotilla,* a boat poisoning, but those who had two helpings squadron to patrol the river (1). Youth have are much more likely to become ill than the not been immune to epidemic suicide. Bet- ones who only tasted the salad. The risk to ween May 1908 and October 1910, 70 an individual youth for suicide may increase children io one schuol district in Moscow as the number of suicides increases in his or lulled themselves (5). her peer group or in the community. The More detailed accounts of epidemic or seventh youth suicide in a widely publicized cluster suicides can be examined for charac- series of seven is likely to have been wore ex- teristics that suggest that suicide contagion posed to suicide than his rredecessors and, in may be part of the etiology. Table 1 sum- effect, to have received a higher dose. marizes the reports that we have compiled Infectious processes, though, can result in a from the literature, newspaper accounts, and wide variety of clinical effects ranging from our personal knowledge. There is no sys-

2-89 Report of the Secretary's Task Force on Youth Suicide

tematic surveillance or reporting system for are under the influence and power of ex- suicide clustets. These accounts have selec- ample more than precept.* He attributed the t ion biases that affect their repre- outbreak of suicides to *the force of imitation sentativeness. They are merely descriptive being so great and acting prejudicially on studies. No comparison groups or statistical weak-minded persons or on those analyses are included. Furthermore, most of predisposed to mental disorders* (13). these studies describe suicidas among adults. Another 19th century writer who eonsidered Youth may be differentially exposed and sus- the role of the potential suicide's intrinsic ceptible to the characteristics of suicide con- vulnerability and the impact of outside events tagion presented in these reports. said that 0...it is difficult to determine how much was due to the psychopathic tendencts A number of studies highlight the choice of of the actors...and how much to the external identical methods among suicides in a cluster circumstances which probably served only as (6-12). Most striking is Walton's report of an unintentional death by antifreeze poisoning, the spark applied to the inflammable material* (11).Contemporary researchers which was headlined in the evening paper. have also considered the impact of poor Five suicides from antifreeze poisoning en- sued. One decedent was found on the bed baseline emotional functioning on susceptibility suicide in a cluster (6,9,14-16). Emo- next to the newspaper featuring the article; tional well-being, then, could lessen and another told her husband before here death that she drank the antifreeze because she emotional disability increase host susceptibility.However, the proportion of non- read about it in the paper (6).Seiden cluster suicides with psychiatric problems described five cases of suicide by jumping that occurred within a month on a college (17) may not differ from proportions reported in case series of duster suicides. campus. Etch successive case was given extensive news coverage (7).In these two The mechanisms most often associated with series, as in a report of suicides in Great epidemic suicides among those who are sus- Britain by burning (9), decedents were not ceptible are imitation and identification (15, acquaintances, and the route of contagion 18-22).Imitation represents the action was presumed to be indirect via the mass derived from an identification with another media or word-of-mouth. person's needs, the identification being conscious or unconscious (73). Ward and Fox Subsequent suicides in clusters in which the studied a suicide epidemic on an Indian victims use flamboyant methods automatical- reserve, during which 8 adolescents and ly access a degree of celebrity. This notoriety young adults died from among the 37 families may convey the illusion of immortality in a in that community. The researchers ex- way that counteracts the potential suicide's amined the way that one suicidal youth can more reestic appreciation of the finality of serve as a role model to be imitated: *The death. Nalin has described the *aura* that a stimulus of one suicide could suggest to particular method of suicide may assume. In Guyana, suicide by malathion insecticide others a similar mode of escaping an in- tolerable life situation* (14). The role model poisoning became associated, through press reports, with unrequited love. Suicides, in also may be an unintentional death, as in the this case, were making a publicly recog- series of antifreeze suicides reported by Wal- ton (6). Sacks and Eth explored the idea of nizable statement through their choice of pathological identification in a cluster of method. The potential suicide attempter also may have imagined that attendant suicides among hospitalized patients. They newspaper publicity would manipulate held that these pathologjcal identifications were *fostered by the individual's past history others (12). that may contain many points of common ex- In reviewing epidemic suicide, Forbes perience* (19). Winslow observed that *all human actions

2-90 .12 L.Davidson:Contaglon as a Risk Factor for Youth Sulckia.

Although imitation and identification may be results.Without a clear and replicable powerful mechanisms, emphasizing these definition for suicide clusters, we cannot un- components exclusively, as in the term 'copy- dertake surveillance.Surveillance could cat suicides,* trivializes the many other fac- determine what proportion of youth suicides tors contributing to suicide. The simplistic appear to occur in dusters and how repre- notion that one person merely copied sentative these case-series findings may be. another's suicide does not explain why the Suicide clusters are more commonly suicide was copied by that particular in- reported now, but without comprehensive dividual and not be hundreds of others who surveillance, we cannot be certain that these were similarly exposed. Suicides remain mul- episodes occur any more frequently than tidetermined events even when they occur in would be expected by chance variation. clusters. Explaining cluster suicides by imita- Without reference to a comparison group, tion alone does not take each decedent's sus- descriptions of the demographic and ceptibility and stresses into acount. psychological characteristic; of suicides that occur within the context of a cluster are Some researchers have considered the social speculative. What may appear to be a ubiq- environment and the ways in which disrup- uitous characteristic of cluster suicides may tions in that environment or negative social not differentiate them from noncluster expectations within the milieu may foster suicides or nonsuicidas and, therefore, may cluster suicides. Rubenstein looks at rapid be of limited value in preventing this par- sociocultural change as an environmental setting that fostered clusters of youth ticular type of death. suicides (20). Cluster suiddis in hospital settings have been attributed to social disrup- Studies in Progress tion creating anomie or a sense of Researchers are currently attempting to ad- hopelessness (24-26). Hankoff felt that the dress the problems identified in the previous prospect of secondary gain (i.e., reassign- section. Field studies are using a psychologi- ment or relocation) was the major environ- cal autepsy protocol with several cluster out- mental influence for a cluster of suicide breaks and including comparison groups and attempts among Marines. The environmen- detailed analyses of the relationships be- tal response to the first attempt provided sig- tween the suicides in an attempt to identify nificant secondary gain; this became the possible mechanisms of contagion. The harbinger of other attempts, which were Centers for Disease Contra-A (CDC) and the finally averted by minimizing the secondary New York Psychiatric Institute are conduct- gains (8). ing two such studies using the psychological autopsy--a procedure that involves These accounts of epidemic suicides indicate reconstruction of the life style and cir- that temporal and geographic clustering of cumstances of the victim, together with suicide does occur. Cluster suicides appear details of behaviors and events that led to the to be multidetermined, as are noncluster death of that individual' (27). Data collec- suicides, but imitation and identification are tion and analysis are still in progress for these factors hypothesized to increase the studies.Additional suicide clusters have likelihood of cluster suicides. Among those been reported to the CDC. Sonic: recent susceptible, the route of exposure to the known clusters are highlighted in Table 2. model may be direct or indirect. The nature of existing research, however, limits con- These recent clusters indicate that it is not clusions that might be drawn, and also sug- necessary for the decedents to have bad gests areas for further clarification. direct contact with each other.In the Westchester County outbreak, indirect The absence of a standard operational defini- knowledge of the suicides appears to have tion for the time and space parameters of a been obtained through the news media. suicide cluster limits our ability to compare

2-91 0 3 Report of the Secretary's Task Force on Youth Suicide

Other clusters had a mixture of members proportion of suicides occur in clusters; (3) from one social network and individuals who whether clustering of suicides is were unknown to each other direcely. predominantly a phenomenon of youth; and Among those who knew another decedent (4) whether the proportion of duster out- the degree of acquairtance variedfrom breaks is increasing. The analyses will also closest friends to those in the same school or provide guidelines on the time and space church who knew of each other but had little parameters that should define a suicide direct personal contact. cluster. Methods may be similar for most deaths A limitation of statistical time-space cluster within a cluster, indicating a possible under- analyses is that they cannot indicate whether lying imitative mechanism. The clearest im- clusters are due to the influence of a model itation of method is seen in a cluster of suicide, or whether the model merely hap- suicides by jumping from an expressway over- pened to be the first individual who com- pass in Seattle. Jumping from overpr mitted suicide in response to conditions that bad previously been extremely rare in.at then led others to die. Field studies are bet- community.Identical methods, however, ter suited to identifying the mmhanisms of may not always reflect direct imitation of the clusters. A goal of the time-sphce cluster another decedent in the cluster. Although analytic study is to identify a representative all of the Wind River suicides were by bang- sample of clusters for future complementary ing, cultural factors may have predominated field investigations. in that choice of method. Hanging has been the method favored by most native American Media Influence ruicides in that community. Most of the research on imitative suicide has Time-space cluster analysis is another typ :. of focused on the impact of suicide reporting in ongoing study in which epidemiological tech- the mass media. This research strategy ex- niques are used to detect and statistically as- amines the possibility of contagion being sess temporal and geopaphic clustering of transmitted indirectly through the media, in suicides (Gould MS, Shaffer a A study of contrast to direct, person-to-person time-space clustering of suicide. RFP #200- propagation. Phelps conducted a prototypic 85-0834 (P), Centers for Disease Control, ecological study of media influences on 10-85-4/87). Several epidemiologic techni- suicide in 1911 and concluded: ques had been developed to examine the oc- The practically universal increase in the currence and significance of time-space mortality of suicides of late years, clusters of diseases (28-30). These methods however, can be demonstrated by official are being adapted to establish clustering. figures of at least comparative accuracy, They can demonstrate an excess frequency of and as this increase historically parallels suicide in certain times and places or show a that in the number and percentage of significant relationship between the time and sensational, crime-inciting books and space distances between pain of suicides. newspapers, at least a semblance of posi- These: techniques are being applied to US. tive evidence of the relations of Ike two mortality data on suicides occurring during is thereby affordedthough not for a mo- the two 5-year periods 1978 to 1982 and 1955 ment, of course, can the open-minded to 1959, and also to data from a consecutive student of the painfully complex problem series of adolescent suicides in the Greater of the increase in suicide forget the fact New York Meuppolitan area in 1984. The that the suggestion of printers' ink is but principal aims of the study are to determine one of the many factors involved (32). whether outbreaks of suicide are real; that is, (1) whether clusters are occurring more fre- More current studies of the impact of nonfic- quently than by chance alone; (2) what tional suicide reporting will be reviewed

2-92 10 4 L.Davidson:Contagion as a Risk Factor for Youth Suicide..

(Table 3A) as well as the impact of fictional tion of suicides.Bonen and Phillips did suicide stories (Table 3B). report, however, an excess in suicides that could not be predicted by any day-of-the- Nontictionsi Suicide Stories week, month, year or holiday effect and confirmed the excess .svith two analytic strategies Phillips and his colleagues have provided increasing evidence suggesting that imitative (38). behavior follows media coverage of nonfic- Stack found no relationship between the tional suicides (33-37). They reported that monthly national suicide rate and the amount prominent newspaper coverage of a suicide of television coverage per month on suicide has thc effect of increasing suicidal behavior stories (46).This lack of a relationship, within the readership area of the newspaper. however, may have been an artifact of the The magnitude of the increase is related to methodology used in the study.Monthly the *attractiveness' of the individual whose rates of suicide may not be sensitive enough death is being reported and the amount of to detect imitative influences, since a con- publicity given to the story. This finding has tagion effect of the media has been reported been replicated with data from the United not to extend beyond 10 days (35,38). States (38) and from The Netherlands (39). Methodological artifacts may also account In addition, Wasserman found that a sig- for Littmann's not finding a relationship be- nificant rise in the national suicide rate oc- tween suicide-related newspaper reports and curred only after celebrity suicides were the occurrence of subway suicides in Toron- covered on the front page of the New York to. Littman reported that there was no sig- Tunes (40). nificant excess of newspaper reports before subway suicides in both epidemic and non- Sex- and age-specific imitative effects have epidemic years (47). There were overlaps, been noted by Barraclough, Shepherd. and however, in the 'before time periods for one Jennings (41), who found an association be- suicide with the *after" time periods for tween reports of suicide inquests in a local another suicide, making independent ex- paper anti the subsequent suicide of men amination of the "before period impossible. under 45 years of age. Further support for a The examination was limited to subway sex- and agt. 4pecific effect was reported by Motto (42), who found a reduction in suicides. An examination of all suicide,s might have yielded a different result. suicides among women younger than 35 years of age during a newspaper strike in Detroit. The core independent variable in Stack's This specific reduction was replicated in study was the number of seconds of television another city (43). coverage of suicide stories included in the 6 o'clock news. There is evidence that it is not Although these investigations support the only the amount of coverage, but also the role of imitative behavior in suicides follow- type of story that has an impact on sub- ing nonfictional suicide stories, results of a sequent suicides (37,40,45).Articles, number of studio have demonstrated no ef- reports, features, and editorials are likely to fect (44-47). Baron and Reiss reported that have differential effects. In another study, the findings of Bonen and Phillips indicating Stack commented on the tennr of news a significant imitative effect of nonfictional reporting that may offset imitative effects television news stories, were due to statisti- (45). He found no increase in U.S. suicide cal artifact and the timing of media events rates after widespread coverage of the Jones- (44). Upon reanalyzing Phillips' data, they town mass suicides. He attributed the lack of reported that the media events had their ef- imitation to the labeling of decedents as cul- fects only during periods when suicides were ti.the presentation of many deaths there already high; the variables measuring the as involuntary, and the horror conveyed by purported effects of the media events were postmortem photos. actually capturing regularitia in the distnbu-

1 5 2-93 Report of the Secretary's Task Force on Youth Suicide

Differential susceptibility to the imitative ef- 11-episode weekly series that was presented fects of the media may also reflect selective by the BBC in Edinburgh to dramatize the coverage in even routine reports. Shepherd suicide prevention work of the Samaritans. and Barraclough analyzed reports of suicides The series resulted in a significant increase in appearing in the Portsmouth News between new client referrals to the Samaritans in the 1970 and 1972. They found that longer 4 weeks following the programs. If the series reports were written on violent suicides than had produced a preventive effect, the rise in on less violent suicides and that violent referngs should have been associated with a suicides were more likely to stimulate multi- fall in completed suicides. To the contrary, ple reports (p.001). Suicides of the very suicides did not decrease during the 10-week young or very old were more often reported period following the series. Moreover, this than suicides of other-aged persons. They period did not show the decline that was regarded this distortion of the news as a evidenced in corresponding weeks in com- publishing commitment to entertainment parison years. These resu!ts suggest a and to the belief that violence is intrinsically deleterious effect of fictional suicide stories nemsworthy (48). in the media. Gould and Shaffer (55) examined the varia- Fictional Suicide Stories tion in youth suicide and attempted suicide Very little research has been carried out on before and after four fictional television films the impact of fictional representations of that were broadcast in the fall and winter of suicide. In the context of an epidemiological 1984/1985. They reported that the observed study of childhood suicide, Shaffer impli- number of attempted suicides after the cated this mechanism as a precipitant in one broadcasts was significantly greater than ex- of the 30 consecutive suicidal deaths he peeled and that there vin significant excesr studied (49). A teenage victim was found of completed suicides after three broadcasts. dead with copy of Graham Greene's novel Their findings are consistent with an imita- Brighton Rock, in which the young, central tion mechanism. character commits suicide. There have been anecdotal reports of suicide rates increasing Critique of Methods in response to the publication and popularity The major limitation of the studies in which of other novels and poems. Publication of investigators examined the impact of media Goethe's The SCRIM of Young Wenher in coverage of suicides is that all have employed 1774 launched a fad among young men of aggregate data (see Tables 3A and 3B). A wearing blue tailcoats and yellow waistcoats major constraint of such a design 6 that it Mce Werther and, in many cases, imitating his cannot demonstrate whether the suicide vic- suicide (50,51). tims were actually exposed to the media Results of recent st Alies that focus on the ef- events. There is always the danger of an fects of media coverage of fictional suicide ecological fallacy, therefore, which involves stories are controversial. Kessler and Stripp making spurious individual-level inferences (52) failed to replicate Phillips' (37) finding from aggregate relationships. that fictional television suicide stories on As early as 1911, Hemenway outlined a study daytime television serials--or *soap operas" to determine the effects of newspapers on triggered imitative deaths. They attributed suicides. He proposed using coroners to col- the discrepancy to Phillips' misspecification lect data in such a way as to avoid making con- of the dates of 8 of the 13 television suicide clusions from aggregate relationships. stories, invalidating Phillips' one attempt to Coroners would obtain the following infor- examine the impact of fictional suicide mation: stories. 1)Dates of prominent publication of Holding (53,54) examined the impact of an

2-94 L.Davidson:Contagion es a Risk Factor for Youth Suicide..

details of suicides, with the method OtIt. selected by the unfortunates. 2) Dates The strength of the association is indicated of subsequent suicides with special by the reports that suicide stories bad larger references to the grouping of cases ac- effects on suicide rates than did day of the cording to methods. 3) Direct evidence, week, month, or holidays, which are variables by asking at inquests for information as known to affect the suicide rate (38). Fur- to the possible relationship of the suicide thermore, reports of suicides by celebrities being investigated as to previous cases resulted in a large increase in suicides (40). either read about or known of (56). The final criterion suggested for judging a As Table 3 indicates, Hemenway's proposals causal association is whether it is coherent or have not been acted upon. consistent with existing knowledge. Media Despite the limitation imposed by the use of coverar of suicides is associated with an in- aggregate data, the investigators in the crease in subsequent suicides. This increase studies taken as a whole, have employed in suicides relates to the amount of pub!icity rigorous statistics, comparison periods, and and is restricted to the area in which the control variables.Their findings meet a stories are publicized. This is consistent with number of criteria that assist judgments the consensus of laboratory findings that about the causal significance of associations. mass media violence can elicit aggression Five criteria for judging causal relationships (58). The association is also consistent with are time sequence of variables, consistency of a number of mechanisms of contagion, such associations on replication, strength of as- as imitation and familiarity with the idea of sociation, specificity of association, and suicide. These mechanisms will be discussed coherent explanation (57). in a subsequent section. Results of several studies (33,34,36,39,40) es- In summary, growing evidence forcefully tablished the time sequence of the variables, supporta the contention that imitative for instance, that the increase in mortality oc- suicides follow media coverage of nonfiction- curred only after the media events. The al suicides. The effect extends to both suicide stories, it was shown, did not occur newspaper and televbion coverage. Some of during a "suicide wave,* but before it. the inconsistencies that exist among studies could have arisen as a result of significant Consistent fmdings in support of an imitation methodological differences among them. hypothesis were reported by the most inves- Although there is some evidence that fiction- tigators, despite their differences in method, al suicide stories have an impact, Hale infor- location, and types of variables. A number of investigators examined an excess of deaths mation is available and the results of following the appearance of suicide stories available studies are contradictory. (33,34,36,38,40,41,55). Others examined the decrease in deaths during the cessation of Mechanisms of SuicideContagion newspaper stories (42,43). Different types of Mechanisms underlying the phenomenon of control periods were employed, varying from contagion have not been studied in the con- contro1 periods immediately before the text of cluster suicides.In social-learning suicide stoty (36), to control periods in dif- theory, however, behavioral scientists have ferent years (34,42), and indirect control ccnstructed a foundation on which many periods used in time-series analyses (40). aspects of suicidc contagion may build. Ac- Both quasi-experimental designs (33) and cording to this theory, most human behavior regression analytic strategies (36) were is learned through observation and modeling employed. Dapite the consistency in the (59). People learn from example. Imitative findings, however, the possibility of con- learning is influenced by a number of factors, founding variables cannot be entirely ruled including the characteristics of the model and

17 2-95 Report of the Sectetatys Task Force on Youth Suicide the consequences or rewards associated with investigations of the news media and their the observed behavior (59). Models who relatbriship to suicide and (2) work in the be- possess engaging qualities or who have high havioral sciences indica:a four conclusions: status are more likely to be imitated. Be- Time-space clusters of suicide occur, haviors depicted as resulting in gains, includ- have been reported among various age ing notoriety, are more effective in groups, and are not a new phenomenon. prompting imitation. Nonfictional media coverage of suicida Consistent with these principles, Phillips and is associated with an increase in the ob- his colleagues have reported that the mag- served number of suicides over the num- nitude of the increase in suicide behavior ber expected. The increase may not be after prominent newspaper coverage is re- uniform for all age-sex groups. lated to two factors: (1) the "attractiveness* of the individual whose death is being Susceptible individuals may be affected reported, and (2) the amount of publicity by direct or indirect exposures to suicide. given to the story.Lilrewise, Wasserman Imitative learning is fostered if the model found that the national suicide rate rose sib- is held in high regard and if rewards are nificantly after suicides of celebrities were expected for the behavior. mported on the front page of the New York Times but not after less prominent suicides Remaining questions are legion. Only sub- (40)- stantial research will resolve them. Some of the questions for future research follow: People cannot learn much by observation unless they attend to the modeled behavior What proportion of suicides occur in (59). A number of factors, some involving clusters? Are :lusters more common in the observas' characteristics, regulate the certain age groups, geographic locations, amount of attention to wibessed or reported or times?Is the proportion of cluster hehavior. Research thus far has only rough- outbreab increasing? ly sketched the 'sag characteristics that may In comparison with other age groups, in yield a greater susceptibility to imitating what ways are young people exposed to suicide. Sacks and Eth proposed as one such and susceptible to suicide contagion? characteristic a histoiy of similar past experiences that lead to *pathok ;Ical iden- Which characteristics of model suicides are most likely to cause an increase in tification" with the victim (19). suicides? In addition to imitative effects, the occur- Which combination of host susceptiLility renae of suicides in the community or in the and contagion factors are most lethal? media may produce a familiarity with, and acceptance of, the idea of suicide. Rubinstein What sorts of prevention and interven- postulated this mechanism in his study of a tion efforts could most effectively avert suicide epidemic among Micronesizn adoles- cluster suicides? cents. Familiarity with suicide may eliminate Does media coverage mate new suicides the *taboo* of suicide. It may also lower the or accelerate suicides that would have oc- threshold point at which the behavior is curred anyway? manlested and may introduce suicide as an acceptable alternative response or option to Youth suicide clusters are a particularly life stresses (20). grievous loss of life and are potentiafly more preventable than single suicides.Interim CONCLUSIONS recommendations are needed even though our knowledge base has sobering gaps. Fwe A review of (1) reports of suicide epidemics, basic recommendations follow: of ongoing studies of suicide clusiers, and of

10 2-96 L.Davldson:Contagion as a Risk Factor for Youth Suicide..

News media representatives should be Med Sci Law 1979; 18:2314. encouraged to avoid romanticizing 7. Ssiden Rit Suicidal behavior on a ocl- romo:atioilsuif in: Fathom?, (ed.,, of fourth suicide, emphasizing violent aspects, and conference for suicide prevantied, 1957; 300- making celebrities of petsons who die by 5. 9. Hankoff La An epidemic of attempted suicide. suicide. Comprehensive Psychiatry 1901; =944 9. Ashton VR, Donnan 84 Suicide by burning as an News media representatives should be epidemicishenomenon: An analysis of 02 deaths and M- invited to collaborate in studies to iden- ountain England and Waleski 1P86. Psychol Med 1981: tify the destructive and constructive com- 11:7354. 10. Crawford JP, Willis JH:Doubts suicide in ponents of fictional accounts of suicide. psychiatric hospital patients. Br J Psychiatry 1956; Before a decision is made to broadcast 11Z12314. 11. ROVIIISky k Epidemic suicides. Boston Medicel such a story, these components should be and Surgical .bumal taaa; 136.2384. identified and assessed, and a warning 12. Nalin OR:Epidemic of suicide by malathion that the program might adversely affect poisonin3 g in Guyana. Tropical & Geographical Medicine some persons should precede any such 1973; 2544. 13. Winslow F:Suicide considered as a mental broadcast. kWarnic. Bulletin of the Medico-Legal Congress. New York, 1895, 334.51. In research stedies, variables should be 14. Ward JA, Fox J: A suicide epidemic on an incllan operationally defined, comparison reserve. Can Psychin Assoc J 1977: 22:4234 15. Nismi 1:The time-space distances of suicides groups used, and individual rather than committed in the lock-up in Finland in 19834987. Israel aggregate- only exposures assessed. Annals of Psychiatry and Related Disciplines 1070: 16.39- 45. Surveillance for potential suicide dusters 18. Robhins 0, Conroy RC: A cluster of adolescent suicide attempts:le suicide o .agious? J Adolf= should be established so that potential Health Care 1953 ; 3:253-6. clusters could be averted and existing 17. Flobins E: The final months. New York: Oxford clusters kept from spreading.Surveil- University Prete, 1981. 18. Rosenbaum M:Crime and punishment-the lance would alert researchers to suicide suicide pact. kch Gen Psychiatry 1903: 40:97942. clusters which might be more thorough- 19. Sack* U, Eth & Pathological identification as a ly investigated. cause of suicide on an1nortient unit Hasp & Cernmunity Psychiatry 1981; 3k Intervention efforts should be directed 20. Rubinstein DH:Epidemic suicide among toward those who are most exposed Micronesian adolescents. Soc Sol Med 1993; 17:85745. 21. Bunch J, Sarraciough & The influence of parental (either directly or indirectly) to the death anniversaries upon suicide dates. Fir J Psychiatry *model" suicide and toward those who 1971; 116.821-28. 22. Crawford JP, Willis JH:Double suicide in art most susceptible, for example, those psychiatric hospital patients. Br J Psychiatry 1980; whose emotional health is poor and lit 12314. those who strongly identify with the per- 23. Hinsie LE, Campbell Ri: ftchinlo dictionary 4th Whim. New York: Oxford tkOnlift Pries, 1970. son who has taken his or her life. 24. Kahn* MJ:Suicide among patients in mental hospitals. Per2h1aby tea 31:3243. 25. Anonynous: A suicide epidemic in a REFERENCES hospitel. Diseases of the Nervous %stem1977;2317T 28. Kobler AL. Stolland E The end of hope: A social- 1. Bainvin H: Suicide in children and adolescents. J cfinical study of suicide. London: The Free Press of Glen- Pedistr 1957; 50.74949. coe, 1984. 2. Centers for Disease Control:Violent deaths 27. Farberow NL Neuringer C: The social scientist as 15-24 years of age-Urritett States. 1970- coroner's deputy. J of Forensic Sol 1971; 10:1540. illn19771.11=98X3t453-7. 20. Eden/ v, Myers MH, Mantel N: A statistical 3. Public Health *Woe: Proing health/prevent- problem in span and time: Po lelikeinla cases came in ing disease: Objectives for the nation. ~skvft, D.C.: clusters? Biometrics 1904; Sept.:02830. U.S. Government Printing Offioe, 1980. 29. Knox G: The detection of speoe4me interactions. 4. Haden TP, Custom NH:(eds). Massachusetts Applied Statistics 1984; *25-29. General Hospital handbook neonatal hospital 2nd edition. Mean, Mass.: PSGPublishinrrriftr: 30. Mantel N: The detection of disease clustering and 1987. a generaxzed regression approach. Cancer Pee 1957; 27:20S-20. 5. Papaw NM: The present epidemic of school suicides in Russia. Newel Nostra (Kazan), 1911: 19:312- 31. Wallenstein 3: A test for detection of clustering 55, 592448. Craf tkr4. Am J Epidemiology 1900: 111:367-73. O. Walton EW: An epidemic of antifreeze poisoning. 32. Phelps ES: Neurotic books and newspapers as

1 9 2-97 Report of the Secretaty's Task Force on Youth Suicide

factors In the mortality of suicide end crime. WNW the 1011; 12.25643. Myriam Academy of Medicine 1011; 12.264-305. Ft Sower, kt Causal thinidng in the NAM edema: 33. PhiNips Ot The influenoe of suggestion on suicide: Concepts Ind *Wei** in epidemiology. New York: Ox- Substantive and theoretical kopticatIon of the ~her et- ford University Rees, 1973. fact. /On Sociological Review 1974; 30440454. 55. Comstock B: Television and human behavior: The 34. MOOS DP: Suicide, motor vehicle fatale*, and key studies. Sense Monk*, C4: Rend, 1975. the mass media: Evidence tomud a theory of suggestion. Bandure k Social learning theory. NM Jersey: S0010100Y 1970; 54:1150-74. Prentice-Het 1977. 35. Phillips DP: /*plane =kiwi*, murder, and the mass medlis Towards a theory of Imitation and sugges- aMotto Jk Suicide end suggestibility-the role of tion. Social Forces 1980; 58100144. the press Mi J Psychiatry 1967; 1242524. 38. Bonen KA, PhiNipe DR Suicidal motor vehicle fatalities in Detroit A repNostion. P411 J Sociology 1951; 87:40442. Sr Phillips DP:The knpact of fictional television stories on U.S. adult fatalities: New evidence on the effect of the mass media on violence. Am J Sociology 1054 07:134040. 8011011 KA, PhiWps DP: Malik. suicides: A nation- el study of the Or* of televlsion news stories. Am Sociological Review 1052;aloe. Genteboom HBO, de Hem D:Gepublioserde asiknoordert en verhogilig von stem* door zelfmoord en ongelukken in Nedsdand 1972-1000. Mena en MeeiechsP- pij 57:5540. 40. Wasserman IM:Imitation and suicide: A mexamination of the Weither effect Ms Sociological Review 1054; 042748 41. Barraciough B, Shepherd Ds Jennings C:Do newspaper reports of coroners' frquests incite people to commit suicide? Br J Perthistry 1077; 131526%32. 42. Motto Jk Newspaper influence on suicide. kr Gen Psychiatry 1070; 23:14:14. 43. Blumenthal 8, &rgnor L:Suicide ants rite3terapaps, A replicated sh*. J Psychiatry 1973;

44. Baron JWReise PC:RtiT3124.o Mips and WWI. Am Sociological Review INS; 45. Stack & The effect of Jonestown on the suicide rate. J Soo Psychollikit 1191454. 46. Stack S.:The effect of suctgeetion on suicide: A reassessment Paper mad at the Annual Meetings of the Anwdcan Sociological Association, San Antonio, Texas, 1054. 47. Uttmann St Suicide epidemics and newspaper reporting. Sukide and Life-Threatening Behavior 19135; 15:4340. 48. Shepherd D, Barraoiough BM Suicide reporting: information or entertainment?Brit J Psychiat 1070, 1=28347. 40. Sheen?: Suicide in ohadhood andM Moles- awe. J Child Psycho! Psychiatry 1574; 15: 1. 50. Spends( 8: Foreword to JS Goethe: The sorrows of young ~Mr. New Weir. New American Library, 1062. 51. Goethe J8:Reflections on Worth's. New York: New American Library, 1052. 52. Kessler RC, Stipp ItThe Impact of fictional television suicide stories on U.S. fatalities: A replication. Am J Sociology 1004; 00:15147. 53. FloWlek. The B.B.C. Iferriendear missend he effects. Brit J Psychiey 1974; 124:470-2. 54. Holding Tk Suicide and The Befrienders. British Medical Journal 10753:751-753. 55. Gould MS. Shaffer 0: The impact of suicide in television movies: Mono, of imitation. N Eng! J Med 1056; 315.0004. 4. To what extent aro suicide end other crimes the person due to suggestion from the press? of the Medan Academy of Medicine

2-98 1 o L.Davidson:Contagion as a Risk Factor for Youth Suicide..

1 1111 hi s al h. 111111 1111111111111111 tA}111111 ill 1 hi711 011 0 Ili lilt mad 1'11 2-99 Report of the Secretary's Task Force on Youth Suicide IP41114,,

Ii4111111 ii' I hg.1111111111 4 A 2-100 CHARACTERISTICS OF RECENT SUICIDE CLUSTERS

Social Number Relation to of Ago First Date Last Date of Mother in LOOS11011 Sulos oeIJ)e stb Death Methods Cluster

Rano, Texas s TM 1448 2f23/83 WM 4 Gunshot Some IF 4 0101Cift rnonoxicie

Werdohestar and Putnam 6" 5,4 13-19 214134 3113184 3 N rantahoi None Counties, Nav Yora I Caonrb monoxide

Clear Lake, Texas 8 ISA4 14.10 819184 I0/11/84 3 Gunshot Soma IF 2 Hanging I Carbon monoxide

Seattle, Wahine Ion 3 3/41 217*-42 T/W85 7114165 3 Jumping from Unknown expreeeney ammo

we Fiver. *Yarning St 984 14,25 WWI* 10/1/86 9Moving Some

Omaha, Nebraska 3 3M 1848 UM 2/7/88 1 OunItICA Some IF 2 Cuenione

An unirdentioned hanging Woo occurred 1' 4 other suicides of tribal member, ape 17-34 mowed between I/2/85 and 10/18/85. NI ware by hanging. IP 8 Table 2.

1t! 0 1I Report of the Secretary's Task Force on Youth Suicide ii

2-102 WV?

gill "1 1

il M

i2 IiiiiiiilliV IIIIIIPItil 0z tillitgiiiii Ilifi 1114i ii. 4101111 ? S1 il z 2i ilII 111111111 111 rtillifl ilkP. [FA ;1 Jpim 11

1101 qa 111111 I 6I IV nil 1 C(1) Eillii1Pro1111V lipilli94111114 i 8 1 If y 5ii lib 1 li enn: 'PAW dritire 1Tliltirib0 i,04 Pt/IP11 I ll t3

Asomommw

-eppIns qinoA 101 ioped obv e seuoleewoo:uospmea7 Report of the Secretary's Task Force on Youth Suicide 96.4C i 111111111 th I IPiv I 11Hot I 11rlLjfltj 1111111 jIjJiiIith till pi lilt I

2-104 L.Davidson:Contagion as a Risk Factor for Youth Suicide..

2-105 STUDIES EXAMINING MEDIA INFLUENCES ON SUBSEQUENT IMITATIVE SUICIDES* A. Nonfiction! Suicide Stories

Reread and Papule- Gemparison Supped of Otation don Studied Methodology Ciroup/Perlod Findings Shitistice krittation

Waseennan, ISM Monthly IL& suited, raise Blinded Philips' (WM Ilst For queekudednowdsl Theme,* dee In Weide, diet Proboby at *note Ad Yee tot lee through laTT. hod page sulealse et the Now York wialiolo the eaparknen- 31I at Ito dB dew II* Immo ti 40 dose b .00027I. Thee (s-4e cessd. Also kW tat podod weed* nom duel° the oelebty nods/ leat). Inin* soda coded ohother the sulphite were ro- MorIllehodier Ms suicide suicides. There sera a mean deo edattoto,, Ihe dope we& *and caldwese. door, Sy =idol 133.Seuketse Odd* dent rsonesteng tho ChresPelderttnenid design se eel prided vise the earns suicides oriMedinmani celebrity story wee V asmtSNIWciyW. mondpdat and eub- celebIllse, a died SS tor raglan- -*** a< Mi. Used maw) bathe 114* CO MICIUMI IWO. Udine al palled oeletelliso. Wa ase Wily*. Included duralion at un- COMO! WWII drre ci 303.1 atter tho meek% ot en In- ondornent de enothat Was andyste. lerndlonal colotody. =vallebet. ctrollingon for W00% end loam roxdogi- cal study.

"A taw dudes on towed sulddor ore kick/. .01(304en PM" Mt; Pldopo, 1974. A sOidy thst ~lined de hoped ot iniaderaulddo dodos an subeequent Osten. Manse 1eS0) vow nal Included. TM endings from the MO study do supped a theory et Indetion end suggrosion.

Table 3 continued.

1 21;

BEST COPY AVAILABLE STUDIES EXAMINING MEDIA INFLUENCES ON SUBSEQUENT IMITATIVE SUICIDES B. Fictional Suicide Stades

Period and Popula- Comparison Suppoitof Citation tion Studied Methodology CiroupiPeriod Findings Statistics ltation

Gould snd Shatter, Adotement suicide* and The wariatIon in Weide and al- 2-tseek periods beam The obetined number at at- The MIMI moll ow of ce VS 190. (53) attempted sdes aged tempted suicide beton, and after 4 and eter each movie tornSod suicides Woking the Soolpfs offor the broanceits 19 yews and younger mide4cosivrision movies broad- wore compered. kt ad- teletrielon males wet signiecent- es-esswoe sig. during the period WHIMS cast In the tett ot 10114 wide**, of Mon. obeened num- ly greeter than erepecte4 . and a Mort* gnaw' then the through WM in thei 1905 wee wantined. Ecological ber* ot suicides and signincen4 mese of completed mew before the brad- Greeter New York etudy. attempts during the suicides vms found siteu 3 broad- awl, (14, id- (t htetropolitan Area. WV period* elm arra. %Wel bias, meats- de.% p.03). Ms observed compered with ear- lion et medical omorninmo. or proportionallempled petted number, hoopltol pommel two mak* suicides to occur during derived from the to account for the Increase In the 4 24see4 periods/0w averege of ea vreelm tempted and completed the mcr.4sis, 40% (I1.119) Itatordlyorof than expected n.711) 4-JO7,beiidon the binornlel diebett)on). The mean num&P ot completed out:10e athsr 3 broadoeste (4.33, ad le) wee eiplacently greater then !Mb mem beim the 3 broadoests(I, id I) csme.P.mi.The obemed number a/ completed sulcides(14 iiiter the 3 hidedomts vies greeter then the e xpected numbs/ (7.44) (p...02 based on the bino- mial diebibution).

Holding. 1974 Commuted sudes and The Imriation in deaths before, The 4 weeks before the Awnsweekly suicide 'Stamp. None Yes i975 MAO) undetermined deaths in during. and after lhe Betrienriers,* series Woodcut date admissions Incemied by 13% Edinburgh during the an I 1-emeode woody series on each yam were used as during the liefrienders swim same 304see4 period in BBC in 1972 ries examined. As the bemlinee. Num- and by 22% during the Waring 191394973. described in Holding (1971). the ber* of *wets arid 4 weeks. Serneritin nos client Wee &mottled the suicide suicide attempts during Worm% incromied Ny 112% prevention vaxk of the Sternentena. and es the wise duelnedieWas a..*140% In Caresponding mole in can- dates wens compered the mid 4 woke. wean yams (1035-1971. end vrith baseline numbem 1973) were examined. Ecological for wet% year. study.

Table 3 continued.

" c) BEST COPY AVAILABLE Report of the Secretary's Task Force on Youth Suicide

r.1. 2-108 L.Davidson:Contagion as a Risk Factor for Youth Suicide..

z i (") IIilidit ce. romf

Jillil Pin i0 I *Ill g t4 4101111 X I i Mg* II 1ainthi hiIh qbh01440044 Hmilhollthilihmfillohh z ipolobt ra tidoolf z0 a hhohithii 1 M A a fit til Fs hIl flisihghil

II its

2-109 STRESS AND UFE EVENTS

LS. Payke4 MA, MD., F.RCP., F.RC Psych, Professor of Psyvhiatry, University of Cambridge, Addenbrooke's Hospita4 Cambridge, England

RECENT UFE EVENTS AND SUICIDAL BEHAVIOR Methodology are of such magnitude as to require a sys- This paper reviews studio of suicidal be- tematic and probing interview, rather than a havior in youth in relation to recent and early self-report checklist. Review of reliability stressful life events. By a "'recent life event* studies (1) shows that self.report methods we mean a change in the external social en- such as the questionnaire used by Holmes vironment that can be dated approximately. and Rabe (2) tend to give low reliabilities; in- A life event represents a change, in contrast terview methods, usually employing semi- to a chronic difficulty or problem, such os a structured format with considerable probing, bad marriage or chronic poverty. The give moderately high reliabilities. Interview change is external and r3t just one of percep- methods atso produce better patient-inform- tion: increased worry over work is not a life ant concordances of the order of 0.8, and event unless it reflects some actual change in relatively little retrospective fall-off of event circumstances. One *internal event"' is physi- recall as time periods extend back. cal illness, which is externally verifiable and Psychiatric disorders may produce new carries major implications for change of life events, such as loss of job, which are conse- pattern. quences rather than causes of illness. To Adequate study of recent life events has en- eliminate these from study, two approaches tailed solving a number of methodological have been adopted. One is to confine atten- problems (1), the most prominent of which is tion to time periods preceding symptomatic retrospective reporting of events. The ordi- onset. The second is to concentrate on *in- nary inaccuracies of recall may be mar:dried dependent* events (3)--those which, in suicidal patients by the effort to give mean- evaluated in terms of their specific cir- ing, in terms of life experience, to such a cumstances, appear highly unlikely to have major occurrence as suicide, and by pes- been brought about by the patient. simism, quilt, and other misperceptions due Alternative methods of quantifying the stress to psychiatric disorder. in events have included consensus scaling The technique for data collection is probab- and a summation to total-stress scores (2), in- ly cruciaL The complexities involved in elicit- dividual judgments of contextual threat (3), ng information, accurately dating and categorization of event: into groups occurrences to the relevant period, and depending on their qualities (1). The dif- deciding whether the threshold and defini- ferent methods, in practice, produce rather tion for a specific life event have been met, similar findings.

2-110 1:13 E.S.Payket Stress and Life Events

The methodology ef chronic stress is less well children and adolescents. The studies that worked out.Brown and Harris (3) have do exist of the young often only look in pass- studies i'difficulties"lontstanding rather ing at stress, and with deficient methodology. than recent stressesand have successfully An issue that arises in young adults is the ex- used methodology parallel to that of life tent to which increased numbers of life events. There are, however, fewer studies of events may reflect a more generally unstable reliability and validity in this arca, or in the lifestyle prone to self-induced life change, closely related area of social support as a and itself rooted in personality. A number of protective factor (4). It can be particularly papers have hinted at this, and detailed ac- difficult here to be sure that the stress is ex- quaintance with some yuung suicide at- ternal: to separate perception and reality. It tempters does suggest a generally chaotic and can also be difficult to make sure that the impulsive life. The same issues can arise in stress is truly independent of the person. relation to life events and other disorders. Personal resources influence the creetion of Oneway to tackle them is by followup studies social networks and close relationships and using recovered patients as their own con- the finding of solutions to long term trols (10) to rule out the possibility that as problems, so that social isolating and chronic many events might occur at any other period, problems may reflect personal qualities as irrespective of onset of disorder. Brown's well as external circumstances. methodology of independence of events (3) should control for this element but it is hard Additional Problems in Relation to to make the judgment of independence from Youth Suicidal Behavior personality.Interpersonal arguments and Some additional problems arise in studies of separations, which are common in young suicidal behavior in youth.First, the suicide attempters, often reflect contribu- methodology of life events in children has not tions from both sides of the relationship. been as well worked out. Adult life event lists are not appropriate. Some adaptations of Studies of Suicide Attempts and scaling to children's life event lists have been Recent Life Events described (5,6,7) although they have The literature on recent life events mainly in- received only limited application in studies of volves suicide attempts, in samples un- suicidal behavior (8,9). selected by ab.t, but usually with large Second, the retrospective detailed interview- representation of young adults. Studies are ing method that is usually employed and summarized in Table 1.(Tables appear at validated in life stress studies can only be ap- the end of this chapter.) Four studies have plied to suicide attempts.For completed made comparisons with general population suicide, the principal witness is no longer controls. In one study of adolescents aged 14 available. Other sources of information may to 18, Jacobs (11) compared suicide at- be usedinterview with relatives or access to tempters and normal controls for events over various kinds of recordsbut these are likely lifetime. The time periods nearer the at- only to be reliable in relation to the most tempt showed an excess of events, particular- major events, namely, bereavement. ly in the weeks or months before the attempt, when there were more break-ups of relation- Studies are therefore predominately of at- ship, illnesses or injuries, and pregnancies. tempted rather than completed suicide. The differences between these two groups are Among studies of adults, Paykel et al. (12) in- well known and it cannot be assumed that terviewed suicide attempters for life events findings valid for one are valid for the other. in the six months before the attempt. Com- Also, most studies are of adults over the age parisons were made with matched general of 25, do not analyze by age group, and when population controls and with matched they extend to younger ages, still tend to omit depressives who were interviewed for the six

2-1 1 1 Report of theSecteterysTask Force on Youth Suicide months prior to onset. Suicide attempters trauma, broken homes, and peer acceptance reported four times as many events as in the change- general population and one and a half times Among studies of adult suicide attempters as many events as did depressives in the not restricted in age, Paykel et al. (12) found period prior to onset. There was a marked that suicide attempters had experienced peak of evarts in the month before the at- more events than depressive controls, par- tempt, and often in the week before. The ex- ticularly in the month before the attempt. cess over general population controls This excess was confined to threatening involved most types of life events. eventeategories: Undesirable events, events Cochrane and Robertson (13) used a less scoring as mote stressful in a scaling study, or satisfactory method, a self-report checklist, events outside the control of the patient. and stadied only male subjects. This study Slater & Depue (IS) compared depressives did undertake separate analysis of subjects who made moderately serious suicide at- under 25. Total stress scores for the year tempts with other depressives. In the year before the attempt and the number of life preceding the attempt, particularly between events VICre much highec in depressives than onset of depression and the attempt, there matched general population controls, and were higher rates of independent events and the excess was equally apparent in the two of exit events involving departure of some- samples: under 25 and over 40 years of age. one from the immediate social field of the It particularly involved unpleasant events subject. Luscomb et aL (16) used a self- and disrupted interpersonal relationships. report inventory to study male sukide at- temptersadmittedtoVeterans Isherwood et al. (14) also used a modified Administration hospitals and patients with Holmes-Rabe methodology.Suicide at- no history of suicide attempts. Using a num- tempters showed much higher stress scores ber of events and scores for perceived stress, than general population controls or a second frequency-of-events rated high in stress, erdt control group of drivers involved in events, desirable events, and undesirable automobile crashes. events, the researchers round some differen- Several studies have made comparisons with ces, with a particularly high rate of exit patient control grotr. Only one study ex- events. However, differences ..ere confined amined life stress and suicidal behavior in to subjects over 35 and, most markedly, those children, and life event methodology was over 50; the differences were not present in limited. Cohen-Sandler et aL (8) compared those Icy to 34 year olds. 20 children admitted to an inpatient O'Brien & Farmer (17) compared life events psychiatric unit because of suicide attempts in the six weeks before interview of suicide and threats with depressed, nonsuicidal attempters who had taken overdoes of children and with nondepreised children ad- medication, compared with young people mitted to the same unit. Life events over the visiting general practitioners for various whole of the life span were ascertained from complaints. Most life events were much the case history charts, a method that might more frequent in the suicide attempters. be vulnerable to unwliability in the original Patients were followed up at three months recording. Mean stress scores increased over and a year. At three months, there was no developmental periods, particularly in the decrease in life event rates, but at tware srtidal sample. In the hvelve months prior months there was a decrease. This was in the ro admission, the suicidal sample had ex- only study in whkh subjects served as their perienced higher stress scores than either own controls, confirming that not all the life control group. The suicidal group had ex- event elevation before the attempt was due perienced more death of a grandparent, separation, divorce, remarriage and to unsAanging life style. hospitalization of a parent, psychologjcal Three studies have been limited to separa- E.S.Paykel: Stress and Life Sents lions, both recent and early. Levi et aL (18) in Table 2. None of the studies have specifi- examined actual threatened disruptions of cally addressed youth suicide, End most interpersonA relationships in the preceding samples have been over 24 years. Bereave- year among suicide attempters, patients with ment, of a parent or spouse, is an event which suicidal thoughts, and nonsuicidal patients. can usually be ascertained accurately. Bunch Suicide attempters exp4rienced more (27) interviewed informants concerning separations than the nonsuicidal group, bereavemen: in the previous five years in whereas those the patients with suicidal suicides and general population controls. thoughts were intermediate. In a replication There was a significant excess among suicides study of working class subjects, Stein et al. (10 in the last two years. The difference par- found more recent separations among ticularly hvolved deaths of mothers and suicide attempters than psychiatric controls. spouses. Men appeared more vulnerahte to Greer et al. (10) found that disrupted inter- loss of a mother (especially if unmarried). personal relationships in the last six months MacMahon and Pugh (28) used death certifi- were more common in suicide attempters cates to compare timing of deaths from than in psychiatric or medical controls. suicides and other causes in widows and widowers. Suicides showed a clustering in One study (21) using a multiple regression the few years following death of spouse, and ana4sis found that life stress on the Ho lmua- particularly in the first year. Rabe scale related significantly to suicide intent, but a study in adolescents (22) failed to In a third study in general population, Hag- find this. nell and Rorsman (29) compared recent events among suicides Loin the prospective In some other relevant, uncontrolled studies, Lundby cohort study, matched nonviolent Power et aL (23) found that severe events, as- deaths, and general population controls. certained over a six month period, peaked in Seven of 20 suicides experienced stressful life the month before a suicidal attempt, but non- events in the two weeks before death, com- severe events did not.Suicidal intent, as- pared with none of the people who had sessed subjectively and objectively, natural deaths. Viewed over the year before correlated with total life event stress, but death, the suicides showed more changes of lethality of attempt did not. Katschnig (24) living conditions, work problems, and objea found a peak of threatening events in the losses than the normal controls, and more ob- three weeks before the attempt. ject losses than the people with natural In controlled study, but of a less recent event, deaths, for whom work was not relevant be- Birtchnell (25) found that more psychiatric cause of the nature of their terminal illness. patients with a recent suicide attempt had ex- Some of the events in the suicides app:ar to perienced death of a parent in the preceding been consequences of psychiatric ill- one to five years, than in nonsuicidal Less, iather than independent events. psychiatric controls. Other studies have used psychiatric patient In a controlled comparison, Paykel et aL (26) controls.Humphrey (30) studied male studies suicidal feelings in the general suicides, homicidal offenders, and patients population. Subjects reporting suicidal feel- hospitalized with neurotic disorders but with ings in the last year experienced more life no suicidal or homicidal histories. The study events, particularly undesirable events. examined losses over a lifetime rather than purely recent ones. Exeuding early losses, Studies of Completed Suicide nnd the suicides had significantly more evidence Recent Life Events of student, occupational, marital, and parental loss than did the neurotic patients; A small number of studies have examined homicides tended to be intermediate. Infor- completed suicide, depending usually on in- mAtion on neurotic patients was obtained terview of relatives. Studies are summarized

1 2-113 Report of the Secretary's Task Force on Youth Suicide from hospital charts, which might not be catastrophes. Separation, divorce, and death comparable with the psychological autopsies are no uncommon in the general population; on the suicides. together, they form the end of all marriages. Case control studies ignore base rates. In cir- Pokorny & Kaplan (31) interviewed relatives cumstances where the causative event is of psychiatric inpatients at a Veterans' Ad- common and the disorder uncommon it is ob- ministration hispital, patients who sub- vious that most event occurrences are not fol- sequently committed suicide. Suicides were lowed by disorder.Suicide attempts and more likely to have had adverse life event be- suicide are rare occurrences in the general tween discharge and suicide than patients population; in children, they are even rarer. who, over a comparable time period, did not Their predictability from any kind of single commit suicide particularly when scores causative factor is recognized to be low (36). during the hospitalization had been high on a measure of defenselessness, mainly reflect- In studies of life events, conventional es- ing depressive content. timates of magnitude of effect depend on the time period used, since there is a consistent Borg and Stahl (32) also compared tendency for differences between subjects psychiatric patients who committed suicide and controls to diminish as time periods are (in varying time periods up to two years fol- extended retrospectively further back. The lowing presentation) with matched effects of life events are most marked soon psychiatric controls.There were no sig- after the event and decay with time. This fact nificant differences for the individual life renders difficult comparisons with long-ac- events analyzed from case notes, and overall, ting factors such as demographic risk factors the controls bad experienced more events, or early loss. Further problems are the ten- although the suicide victims h KI reported dency of different studies to use different more deaths. time periods, different ways of assessing Fernando and Storm (33) undertook a event stress, and analyses that are sometimes similar comparison. They found a sig- categorical and sometimes quantitative. nificantly greater frequency of losses in the Fmdings also vary with the type of event last year; these included divorce, separation, analyzed, precluding a single summary index. illness or death of a first degree relative or One useful epidemiological measure of mag- friend, and loss of job. nitude (37) is the relative risk of disorder in Murphy et al. (34) in an uncontrolled study those exposed to a causative factor and those also found that alcoholics who committed not exposed. An approximation, the relative suicide tended to have recent loss of close in- odds, can be used in case control studio. Ap- terpersonal relationships. Humphrey et aL plying this measure to studies of psychiatric (35), examining the sequence of events in disorder (37) suggests values of around 6.0 case histories of former psychiatric patiehts for the risk of depression in the six months who committed suicide, identified a charac- following the more stressful classes of life teristic sequence starting with drinking event, with considerably lower values of 2.0- problems, followed by difficulties with fami- 3.0 for schizophrenia. Values fall off con- ly, sex, friends, and work. This sequence was siderably with extension back of the time regarded as reflecting the lifestyle of thr.: periods. suicide, suggesting alcohol problems leadirig Table 3 gives relative odds for suicidal acts, to gradual social deterioration. from studies using general population controls which report data in a suitable form for Magnitude of Effect computation. For suicide, values range from It has often been pointed out that the recent 4.6 to 6.5, but for periods of one to two years: life events implicated in psychiatric disorder, for suicide attempts, from 6.0 over 6 months although stressful, usually fall short of major to 10.0 over 1 month. In general, these ..ug-

2-114 1 :17 E.S.Payket Stress and life Events gest effects that are higher than for depres- psychiatric patients was in children and one sive onset. in young adults; another found differences only among older and not younger aduhs. Overall, these effects are moderate in mag- Various stressful events are involved. One nitude, suggesting an important effect but flaw in existing studies is that few have con- very far from an overwhelming one. A com- sidered separately events that are inde- parison is provided by acute infectious dis- pendent, in the sense of not previously being eases where risks of disorder nre high early caused by the psychiatric disorders which after exposure then fall off rapidly. Using may precede suicidal behavior. The one this comparison, risks are dramatically lower study that did so found that effects of events than for disorders such as chickenpox after and bereavement, one event involved, is al- exposure in those who have not acquired im- most always independent. Lifestyle effects munity, but comparable to tuberculosis, are not ruled out, but one study did find a fall- where there are many modLying factors (38). off of events on followup, supporting a If effects were summated over a lifetime in clustering at onset.It cannot be excluded suicide, they would be higher for persisting that lifestyle contdbutes to the events even if associations, such as demographic variables they do cluster at onset: even if this were the and, personality, than for recent life events, case, the events nevertheless may be although in short periods the life event effect pathagenic in their own right.Effects are can be dominant (39). moderate in magnitude, higher than for There has been considerable study of inter- other psychiatric disorders, although actions with potential modifying factors such predominately short term. as social support in depression (3,38), but Possibilities for prevention are limited by the relatively little similar study in suicidal be- rarity of suicidal behavior. Many life events. havior. Slater and Depue (15) found poorer at least in adults, are inevitable consequen- social support in suicide attempters than con- ces of the life cycle and interpersonal trols, but much of this was due to prior exit relationships, and cannot easily be events. A rich literature relates social isola- prevented.However, as will be seen in tion to suicidal behavior, although it is not so desceptive studies of children and adoles- clear the extent to which some of this associa- cents reviewed in a later section, some events tion might reflect personality and previous may be consequences of living in very dis- psychiatric illness, and the extent to which it turbed family settings and might be prevent- acts specifically as a modifying factor to the able.Otherwise, preventive efforts must consequences of recent life events. focus on modifying the consequence of the event. Major events can be used to signal a Conclusions Regarding Recent period of high risk whui crisis intervention Stressful Life Events may be indicated, although in adults, so far. There are few studies of recent life events this approach has not proved useful in and youth suicidal behavior that use rigorous modifying behavior in adult repeat suicide at- methodology. Including all studies irrespec- tempters (40). tive of subject's age, findings are that life events strongly and consistently precede EARLY LOSS AND SUICIDAL suicide, attempted suicide, and suicidal feel- BEHAVIOR. ings. For suicide attempts, four studies, one of adolescents, also show stressful life events Methodology more common than in the general popula- Studies of early environment and suicidal be- tion; such life events are more common in at- havior have mainly concerned loss of a parent tempters than in depressives, in mixed in childhood, by death or other causes. Care- psychiatric patients, and in medical patient ful controls and matching are required (41). controls. One of the comparisons with other

2-115 19 Report of the Secretary's Task Force on Youth Suicide

Rates of childhood bereavement tend to beseparation, and a possible association with higher in older subjects, born in earlier severity of depression (45,46). Most studies decader,, as death rates in young adults, who of suicidal behavior deal with suicide at- are patents of young children, have eieclined tempts rather than completed suicide. progressively through this centuty. Divorce In one of the few studies of completed rates, on the other hand, have risen. Death suicide, Paffenberger et al. (47) used college rates also tend to be higher in lower social records and death records of former students classes and in certain areas. Higher rates of at Harvard and the University of Pennsyl- childhood bereavement will also be found in vania to examine antecedents of 381 suicides conditions associated with greater parental and 652 matched controls over .117 to 51 year age. Reliable information is difficult to ob- followup. An earlier publication (48) had tain. Finlay-Jones et al. (42) found that only used a smaller sample included in this later parental death and marital breakup were analysis.Maternal loss did not predict reported reliably in an eight-month test- suicide, but paternal loss by death did so, with retest study: these made up only one-third of paternal loss by separation showing a trend reported childhood separa:;ons. at 10 percent significance.The effect of Childhood bereavement effects are general- paternal death was a relatively small one, ly assumed to be mediated environmentally, with a relative risk of 1.6, only reaching sig- but could indicate common genetic influen- nificance by the large sample size, but its ces on parent and child, such as suicide in ef- specificity and causative importance were fectively disordered parents.For loss in reinforced by an absence of any similar ef- other ways, such as marital separation and fects on accident deaths, another group divorce, the influence of personality patterns where familial lifestyle might have been of becomes more plausible.Parental chaotic importance. Roy (49) competed case notes lifestyle may contaute much to early mari- of 30 chronic schizophi enics who committed tal breakdown and even early parent death, suicide with 30 chronic schizophrenics who and could well represent genetically in- did not. The rates fss loss of parent before heritable traits manifested in the next age 17 by death or separation were closely generation both in lifestyle and in suicidal be- comparable. Studying patients with recur- havior.Tsuang (43) found that, among rent affective disorder, the same author (50) families of schizophrenic and manic depres- found that more of those who committed sives, risk of suicide in relatives was higher suicide had early parental loss than those who when the patient had committed suicide. did not.Studying 90 psychiatric patient suicides with mixed diagnoses, presumably Retrospective studies of qualitative aspects including some or all of the above (51), he of early upbringing are even more difficult, in found only a trend at the 10 percent level for view of the likely retrospective distortions parental loss by death or separation before and the limited possibilities for validation. In age 17. an elegant study, Wolkind and Coleman (44) showed that recollections of the quality of relationships between parents in childhood Suicide Attempts varied with the mood state at the time of the Table 4 summarizes a number of studies of interview, whereas reports of separation suicide attempts and early loss in adults. from parents in childhood did not. Only two of these separately analyzed younger subjects under 30.Four studies Completed Suicide made comparisons with general population samples. lvo of these used medical patient The plentiful literature on early loss and controls.For the purposes of recent life depression suggests a weak association be- event studies, these are best regarded as a tween parental death and later aepression, a patient control group since there is evidence stronger association with parental loss by

1 2-116 E.S.Payket Stress and Life Events that life events may precede some medical was reported in 77 percent of the former and disorders and hospitalization. For early tow 20 percent of the latter,remarkably high there is no similar evidence ard it is difference. Bruhn (58) reported more kas in reasonable to regard medical patients as a suicidal patients, than in nonsuicidal out- normal control group. patients or inpatients. Greer et al. (20) compared suicide at- Greer (59) studied patients with neurotic and tempters, nonsuicidal psychiatric patients sociopathic disorder. These with a history of and medical patients and found significantly a suicide attempt bad more evidence of higher rates of early separation among at- parental loss for at least 12 months before the tempters, particularly for loss of both parents age of 18; they had experienced such loss par- and loss under the age of 4. Analysis con- ticularly before the age of 55 and more com- firms that the differences were significant monly were deprived of both parents. The separately for death and for separation or analyses were significant both in patients divorce. Crook and Raskin (52) compared under 30 and 30 & over.Reanalyzing depressed inpatients who had histories of Greer's data to examine the nature of loss, suicide attempts, depressivei without such loss by death just failed to reach significant histories, and general population controls. (233% attempters, 24.3% controls, p = They reported a significant excess of loss in whereas loss by divorce or separation was sig- the suicidal group by divorce, desertion or nificant (173% vs 8.3% p <.05). Gay and separation, but not by death.Detailed Tonge (60) found significantly more separa- figures were not given. Goldney (53) com- tions before the age of 15 in new consecutive pared female suicide attempters aged 18 to referrals to a psychiatric department with a 30 with a small sample of women attending a history of a suicide attempt than in those community health center. There was no dif- without it. fermce for parental death but more loss by divorce and separation. Adam et aL (54) Hill (61) examined case notes of depressed patients, comparing those who had made a studied suicide attempters and general prac- suicide attempt and those who had not, for tice controls. Again, there was a slight but not significant difference for deaths, but early parental deaths. Overall rates were not much different in the two groups but there there was a significant excess of loss due to was some excess of deaths in certain sub- divorce and separation.Overall, all four studies confirm that parental divorce or groups: patients of both sexes who had lost their fathers when they were aged 10 to 14 separation is a risk factor for suicide attempts in adults but, with one acception, leave and women who had lost their fathers when they were aged 15 to 19. doubtful the impact of parental death. Birtchnell (25) studies psychiatric patients Other studies have made comparisons with with a history of attempted suicide. Fooling psychiatric controls. Many have not distin- data about all parental deaths and il- guished the type of loss.Farberow (55) reported no difference overall in the in- legitimacy, he found more such events had This was cidence of separation in small samples of occurred in suicide attempters. mainly due to parental deaths than occurred suicide attempt= and nonsuicidal patients when the patients were age 10 to 19, although but more loss before the age of 6. Moss and this difference failed to reach significance. Hamilton (56) reported separation in 60 per- In a later, similar study (62), he found no dif- cent of seriously suicidal patients as opposed ferences for loss at any sge in childhood or to 15 percent in other patients, both those who were potentially suicidal and non- adolescence. suicidal.Walter (57) studies early loss in Levi et al. (18) studies the effect of parental patients with suicidal threats or attempts, and separation of six months, comparing in nonsuicidal depressed patients. Early loss psychiatric patients who had made a recent

2-117 Report of the Secrefaty's Task Force on Youth Suicide suicide attempt, who had had suicidal feel- separations, and usually including temporary ings, or who were nonsuicidal. The first separation, all seven have shown an excess in group welt significantly more Rely to have suicide attempters.There is a clear and esperienced a separation than the third, with strong association with loss by other means the second group intermediate. The dif- and a weaker, but probable, association with ferences were particularly distinct when the las by death. separation occurred before the age of 7. Stein et aL (19) replicated this study in work- Suicidal Feelings ing class attemptens, analyzed separately by Some additional studies have been made of sex and race.There were significant dif- suicidal feelings. Adam et aL (64) found in ferences for loss up to the age of 7 for all four students attending a student mental health groups, and for loss up to the age of 17 for all service that suicidal ideation was associated except black males, for whom the differences with loss of a parent by death and divorce or were suggestive and the sample small. separation. Ross et al. (65) studied a mixed Both these studies also examined separations sample of students, medical patients, and in the last year and have been included in state employees, and found that those with Table 2. Both sought to examine whether suicidal feelings more often reported their there was a particular aggregation of patients parents had separated. Goldberg (66) in an with the combination of early and recent loss, epidemiologica' community study of 18 to 24 suggesting that early loss acted to sensitize year olds found a significant association with the patient to recent loss. Brown and Harris loss of mother before the subjects reached (3) have presented 'evidence that this is the age 16. case for depression in women, although not Retrospective study of qualitative childhood all the evidence is consistent Interactions in environment is very liable to selective fal- these two studies were tusted by partitioning sification. Ross et al. (65) found that those the total chi squam in neither study were with suicidal feelings reported parents who they significant, although inspection of fre- favored millings, were unstimulating, guilt- quencies does suggest some interaction in engendering, rejecting, and unaffectionate, the predicated direction. Greer et al. (20) abusing, and punitive. Goldney (53) found used a different mode of analysis and found that young women attempting suicide were both suicide attempters and nonsuicidal more likely than normal controls to report psychiatric patients tended to show this pat- parental quarrelling, frequent disageements tern but medical controls did not, which with their parents, fmancial problems at would be consistent with an effect in various home, poor childhood physical health, and psychiatric disorders. Some confirmation of various negative parental characteristics. an effect was found in an uncontrolled study These findings may say something important (63) in which suicide attempters were inter- regarding subjects' perceptions rather than viewed about early separation and recent in- their real environments. terpersonal loss. 'There was an association between early and recent loss for females, but Studies In Children and not males. Adolescents Fooling these comparisons with psychiatric Few studies of children and adolescents com- controls, four studies have found more early paring early loss in suicide attempters and parental death in adult suicide attempters control groups have used adequate samples than in psychiatric controls, while two have and methods. In the study already shown for failed to find this association. Three studies recent life events in Table 1, Cohen-Sandler have found a difference for parental separa- et aL (8) compared children aged 5 to 14, ad- tion or divorce and none have failed to. mitted after making suicidal acts of threats Among studies not distinguishing =MSC of with admitted children who were depressed

141 2-118 ESPayket: Stress and Life Events but not suicidal, or not depressed. Data were time of loss and sex of parent or patient do obtained from case histories. When corn- not report consistent findinp. pared with the two other groups, the suicidal More controlled studies of children and children showed, during infancy and preadolescents are needed.Stronger effects school years, a higher incidence of separation might have been expected if events decay from parents; during early childhood, more over time; early childhood events are not far parental divorce; and, during late childhood, in the past for the child, as they are for the more separation and divorce of parents. adult. Parental loss is also clearly an event of However, Jacobs (11) found only small dif- major salience in general terms for the child. ferences between adolescent suicide at- Effects are moderate in magnitude. In adult tempters and general population controls in the incidence of separation or divorce of comparisons with the general population where relative odds are computable from the parents during childhood, with considerable data given, the figures are:completed difkrences for more recent break up. suicide and parental divorce o: eeparation, Stanley and Barter (67), comparing sukidal 1.6 (47); attempted suicide and parental and nonsuicidal hospitalized adolescents, death 23 (20); attempted suicide and paren- found that parental low through separations tal divorce or separation, 3.9 (211); 3.8 (54). and divorce occurred in both groups, al- Although lower than relative odds obtained though more common before age 12 in for recent life events computed over six suicide attempters. months, they are impressive for effects acting over prolonged times. Conclusions Regarding Early Loss Overall, there is an association between early There are few adequately controlled studies loss and suicide attempts, mere marked than of youth suicidal behavior, with no systematic for psychiatric disorder in general and more study of completed suicide in the group marked for early loss by separation or divorce under 25. The very few studies in adult than by death. The association with death groups do suggest that early loss charac- could be due to parental suicide and genetic terizes suicides, when they ate compared transmission of suicidal behavior, since with psychiatric controls. parental suiade is not usually excluded from Regarding attempted suicide, one study of parental death, but this is unlikely to be of children does show a considerable increase sufficient magnitude-to account for the dif- of separation from and between parents, al- ferences. A more plausible explanation though methodology was not ideal, and other might be in an associated risk-taking lifestyle, studies provide less support. Studiesofadult but there is a prima facie case for regarding the death itself as pathogenic. suicide attempters show higher rates of early loss by separation and divorce than the Loss for reasons other than death, is likely if general population in all studies, but in- permanent, to be due to breakup of parents' creased parental death in only one of four. marriage.Here, conclusions can be less Comparisons between adult suicide at- clear. The effect is larger and more consistent than for early parental death. This might tempters and psychiatric controls consistently show more loss in the former when the reflect fan dial lifestyle towards acting-out, study examines separation, divorce or loss whether genetically or environmentally based, rather than causation. However, not distinguished by nature, although in some studies, the differences are weak and selec- marital strife that precedes breakup and the tive to certain groups. Studies of parental subsequent uncertainties of having two death also tend to show an excess, but to a separated parents may be more persistently weaker degree and with some negative harmful than the major loss itself, even by death. studies. Studies that specify the child's age at

1.12 2-119 Report of the Secretary's Task Faroe on Youth Suicide

OTHER STUDIES OF SUICIDAL (72) reported on 597 suicide attempts in subjects aged 12 to 20 in New York Qty. There BEHAVIOR IN YOUTH was a high incidence of family disorganiza- The studies reported so far have beention. In 21 percent of the cases, the parents predominately of adults, reflecting the were not living together. The author felt that paucity of good controlled studies in young most attempts were sudden, precipitous adults, adolescents, and children.Studies reactions to stresful situations. that lack control groups are unsatisfactory in Barter, et al. (73) studied case notes of 45 life event collection or other methodological suicide attempters under 21. Twenty-three aspects can supplement these. A selection of had lost one of both natural parents, and of studies is reviewed rather than a comprehen- 21 cases 'Acre natural parents were living sive survey. together, 11 had marital problems. School One recent study is reassuring. Rich et aL performance was almost uniformly poor. Fif- (68) (in press) compared 133 completed teen of the attempts grew out of an argument suicides aged 15 to 29, with 150 aged 30 and with parents and 14 were associated with a over. There was no difference in the number break-up of a relationship. Only in 9 cases of stresses at the time of death, although the did the case notes not indicate a precipitant older group had more illnesses and the cause. younger group had experienced more un- Rohn et aL (74) studied 65 adolescents who emplornent and legal problems, with more attempted suicide. Fifty-nine percent came separations regarded as suicide precipitants from one-parent families, with prolonged ab- when present. The presence and salience of sence of a parent in 11 percent more; 25 per- stresses is likely to be related to the life cycle. cent were not living with either parent at the Even in this study only 31 of the suicides were time of the attempt. Seventy-five percent unda 25.. had very poor I chool records. Tischler et al. There have been many uncontrolled descrip- (75) studying 108 adolescent suicide at- tive studies, mostly in adolescents rather than tempters, found that most frequently cited younger children.Shaffer (69) made a precipitants were family problems (52%), detailed study of 31 completed suicides aged problems with the opposite sex (30%) and under 15. Common precipitants were dis- school problems (30%). Almost 50 percent ciplinary crises, Bghts with peers and disputes reported that at least one of their parents had with parents or with friends of the opposite been divorced. Only 49 percent were living sex, bereavements. Only in 10 percent were at home with both parents at the time of the there no precipitants. Many suicides took event place during a period of absence from schooL These studio were American. Similar rind- Most children were living with one or both inp have been obtained in Britain. White parents and early loss was not common. (76) studied 50 adolescents admitted to a Atnir (70) studied Israeli suicides aged 10 to general hospital following overdose. Seven- 18. Common motives recorded were quar- ty percent were experiencing difficulties with rels with parents, with family members and important persons in their environment, 8 others.In only 3 percent were motives percent academic worries, 6 percent concern recorded as unknown. Seventy-six percent regarding physical illness. In 10 percent no came from intact families. cause was known to the patient Fifty per- Studies of suicide attempters are plentiful cent had undergone separation from parents Toolan (71) reviewed 102 admissions to Bel- before the age of 15. levue Hospital under age 18 but mostly over Hawton and colleagues (77,78) studies 50 age 12. There was a high incidence ofbroken adolocent overdosers. Family backgrounds homes; only 32 were living with both parents, were disturbed. 36 percent living with a single with paternal absunce common. Jambliner

2-120 1 3 E.S.Paykel: Stress end Life Events parent and 12 percent with neither parent, stases, which were most commonly preoc- higher than national figures. The common cupations about school failure, disturbed problems identified at the time of the over- friendships, fears of pareigal punishment, doses were arguments with parents, and school and family crises. Nor were there problems with nhool or work, including un- significant differences in family situations, employment and problems with boy or girl with high incidences of parental separation, friends. Eleven percent had problems with parental absence and abuse home atmos- physical health. pheres in both groups.Comparing delinquent adolescents with and without a Seiden (79) reviewed the literature on the previous history of suicide attempts (Miller special case of university student stress and et al.) (83), found no difference in the in- suicidal behavior. Most studies find students cidence of absent parents. However, Gar- to have higher suicide rates than their non- finkel et al. (84), comparing case record academic peers. The degee to which this is information on 505 suicidal children and related to academic stress or other self-selec- adolescents presenting at an emergency tion factors is not clear. room and 505 nonsuicidal cases, found the at- Otto (8)), in a study of suicidal attempts in tempters to show significantly more un- children and adolescents related to school employed fathers, employed mothers, and problems, concluded that they were of rela- parents absent form the home. tively low importance. However, recent An English study (85) compared children ad- suicides in school children in Japan reported mitted to a general hospital after overdoses in the popular press suggest that findings may with matched psychiatric outpatient refer- be different in a culture in which school rals. The suicide attempters showed more children are under considerable academic family disturbances, with more evidence of pressure. poor relationship between child and father or Comparisons with psychiatric controls tend mother, lack of warmth, and discord, but no to suggest that these fa:tors are only weakly greater evidence of other stresses such as specific to suicidal behavior, and characterize poor living conditions, migration, stress a variety of disturbed adolescents. Mattson within the family, stras at school, or other et al. (81) studied 75 suicidal children and ortra familial stress. adolescents referred as emergencies to a Overall, these studies uniformly confirm high child psychiatry clinic. Only about half came rates of early loss, broken homes and from intact families, 27 having experienced precipitant stress in adolocent suicide at- parental divorce or separation, 6 loss of one tempters, but suggest lower rates of broken parent, 8 loss of both. However, this was not horn= for completed suicides. Stresses par- different from other children referred to the ticularly center around family problems, clinic.The common triggering situations break or threat of a break in relationships, were conflicts with parents and loss of and school problems. Physical illness is un- heterosexual !Lye objects, with school common as a stressor. The stresses are those problems, sexual conflicts, and pregnancy in which might be expected to loom important- a smaller proportion.In the nonsuicidal ly at this stage If the life cycle. children these were los common, except for conflict with parents, whereas physical injury Controlled comparizions with other and exacerbations in chronic physical illness psychiatrically disturbed adolescents show were MOM common. similar rates of recent stress and fairly similar rates of family breakdown and early loss, in- Pfeffer et aL (82) studio 42 children with dicating that these phenomena are not high- suicidal ideas, threats or attempts and 16who ly specific to the suicidal, but characterize the were nonsuicidaL There were no differences psychiatrically disturbed. The rates are between the two groups in the type of recent probably above those in the general popula-

144 2-121 Repott of the Seaetety's Task Force on Youth Suicide tion, although better studies are needed to psychiatric disturbances; in particular, confirm that. they often rapidly follow interpersonal dis- One issue does emerge clearly:the ruptions. precipitant life stresses in adolescents are to 5. Studies of early loss in older suicides and some extent bound up with the disturbed suicide attempters consistently show rates family backgrounds: both are facets of dis- higher than in normal control groups or turbed families. A second issue often other psychiatric disorders because of described is the impO:kiveness of the suicide break-up of parental marriage. Relative attempts, raising the question of how much risks are moderate in magnitude; less than personality traits of impulsiveness and acting for recent events but apparently long-en- out contaute to the ultimate consequence. during. For parental death, the effects are less marked; rates are probably raised but GENERAL CONCLUSIONS only a little. 6. Uncontrolled studies of adolescent suicide A number of general conclusions and direc- attempters and the few comparisons with dons for further studies emerge form this general population controls suggest par- overv"--v. ticularly high rates both of recent life stras 1. There is a dearth of studies that assess the and of earlier family disruptions. impact of employing careful methodology Precipitant stresses pardcularly involve and controlled comparisons of recent life family problems, break-ups of relation- events in the role of youth suicidal be- ships, and school problems. Physical ill- havior. ness as a streasor is uncommon. Recent stresses often appear to a out of dis- 2. There are more studies of early parental turbed family settings. loss, but even here, there are not many controlled comparisons or attempts to 7. Comparisons of adolescent suicide at- date the family break-up to examine the tempters with other psychiatrically dis- extent to which the associations depend turbed adolescents show only small on recent or earlier family break-up. differences, suggesting that the early losses and disturbed families are relatively 3. There are few studies involving completed nonspecific, and the recent stresses only suicides, rather than suicide attempts. partly specific. The former is more difficult to study, but the many differences in demographic and 8. Interpretations of these associations are epidemiological characteristics of the two not fully clear cut. In careful studies of populations mandate separate studies. recent life events, it is clear that much of the excess in events is independent of 4. Studies of recent life events in older causation by the patients and represents a populations consistently indicate in- genuine clustering of external stress creased rates before suicide attempts and before the suicidal act, rather than simply suggest the same for major events, before an enduring aspect of lifestyle that would completed suicides. In terms of relative be found at any other time. Nevertheless, risks, effects are moderate in magnitude, generally disturbed lifestyles and, for but tend to be short term, with rates of adolescents, family settings not in their events in suicide attempters (and other control, do put subjects more at risk of psychiatric patients) converging with major events. those in control groups as time periods are extended retrospectively beyond a year. 9. For early loss, the small differences in Suicide attempts probably bear the parental death and considerably larger dif- strongest and most immediate relation- ferences in loss by marital break-up are ship in time to recent life events of any susceptible to two alternative explana-

2-122 E.S.Payket Stress end Life Events

tions, both of which suggest that some- 11. Jacobs J (H) Miasmal suicide. %ley-Inter- thing other than the loss itself is science New York 1871. 12. Pavlof ES, Prusoff BA, Myers Jit Suicide attempts pathogenic. One interpretation is that the and recent Me events: A controlled comparison. koh. findings to some degree reflect a tenden- Gen. NyoNst. 1975; 3t3ZP43?. 13. Cochra_ ne Ft, Robertson k Stress in the lives of cy to an acting-out personality and life pwasuicides. Social Psychist. 1975; 10.181-171. style, genetically or environmentally trans- 14. ishsmned J, edam KS, Homblow Aft Life event mitted down the generations. The second strir factors, suicide attempt and auto-so oidem prodMty. J. Psychosom. Res. tat 2%371483. is that the prow Ng marital strife and ar- 15. Slats( J. Dews RA: The contribution of environ- guments are mote dtsruptive to the child, mental events and scold to 'Mous suicide attempts in disorder. J. Abnormal on a long term basis, than is the loss of a Psycho,. 1 1; parent itself. 15 Lusoomb RL, Gawp AC, Petslokes AT: Mediating factors in the Wetland* bidween Nas stress and suicide la Neither recent nor euly environment can attempting. J. Nervous end Mental Diseases 198; 108:844- be regarded as sufficient causes by them- WO. 17. O'Brien SEM, Fanner RCM The role of life events selves. They, particularly recent events, in the &fuof episodes of self in Farmer A. dearly interact with other personality vari- Hirsch S The suicide syn. -London. Cream Heim pp. 124-130. ables related to acting-out and impulsive 18. Levi False CH, Stein M. Sharp Vit Separation behavior. and attempted suicide. Arch. Gen. Psythlat 1900; 15:155- 184. I I.Implications for prevention are not easy. 19. Stein kg, Lew MT, Glasberg RA: Separations in black and Wie suicide attempters. Arch. Gen. NOV*. Partly, this is a general problem in relation 1975; 31:815-821. to the occurrence of rare consequences, so 20. Greer S, Gunn JC, Koller Kkt fratiological factors that most occurrences of a cause will lot in attempted suicide. B.M.J. 1008 t 1352-1357. be followed by disorder. In addition, most 21. Papa LL Responses to life *vents as ples of suicidal behavior. NUlliftg Rte. 1979; 2t362 of the recent or early events are not 22. Friedrich W, Reams R, Ambit J. Depression nd preventable. At best, the recent events suinidel Idestion41:3-4V adolescents. J. You9s & might serve as sip& for high-risk periods 198St 11: .;.:31. Pow KG, Cooke Mt, Brooks DN:Life stress, when crisis interventions might be at- medical lethality, and suicidal RdsM. Br. J. Payahlat. 1085; tempted. 147:855459. 24. KaNchnig It: Measuringlifostmes: a of two methods, Farmer R, mei (eds)Thccit= Syndrome, London, Groom Helm 108% pp 118422. REFERENCES 25. Sinchnell J: The relationship between aisempted suicide, depression and went death. Br. J. Psychist. 1. Paykel ES: Methodological aspects of Me events 1970; 118.307413. resswoh. J. Psyrhosom. Res. 198% 27:341,152. 20. POW ES, Myers X. thidinthal JJ. Tanner .1: 2. Holmes TH, Reba RH: The social readjustment Suicidal feeling in the general palmation: A prevalence rating male. Pcychosom. Res. 1957; 11:213415. study. Br. J. Psyddat. 1974; 38771478. & Brown GW, Hwris T: Social Odgins of Depression. 27. Bunch JRecent bereavement in relation to 1978; Landon:Tavielock Publications. suicide. J. Psychosam. Res. 1972 1e381405 4. Paykal ES: Ws events, social support and dinkel 28. McMahon BR Pugh TR Suicide In the widowed. mohistdo disorder, in Sermon Samson BR (ads): So- M. Epidemiology 1985 81:2341. litary march end einillostion. Martino 29. Mignon 0. nommen EtSuicide in the Lundby NIIhilfrrHague.NetherWia, 1905, pp 321-347. study: a controlled prospective kwestigation of stressful 5. Coddington Oft The significance of life events as life events. Neuropsyrthobiology 1985 8319232. Wore In diseases ci children. J. Psychosom. 30. Hurriphnw Jk Sookil loos: A comparison of 18205-213. suicide victims, homicide offenders and non-violent 8. Monaghan J, Robinson J, Doge J: The ohildren's Dim Nor. System 1977; 3%157-100. life event inventory. J. Psychosam. Res. 1979; 00358. 31. Pokorny AD, Kaplan HEI:Suicide following 7. Yeaworth R, York J. Hussey M. Ingle M. Goddwen Walsoitionneric hortalication. The interaction effects oi T: The development of an adolescent Me change event and adverse Nis events. J. Nem Mont. sale. Pdolesoenoe 1988, 15e9147. Dia. 1975; 182119425. 8. Cohen-Sandier R, Berman AL, King Rk Life Ness 32. Borg SE, Stahl PA:Prediction of Suicide. A end eyrnptornstalogy: atunVwft of suicidal behavior os ive study of suicides and oontrois among In children.Ant kad. Child MOM. 111188; 21:178488. patients. Acta Psychlat. Sand. 1988; 05:221- Fer WE: Gifted adolescents, stress, and Ws change& Mol.sosno. 1981; 18:974985 33. Fernando s. Moan V: ildickle_kinf,n9 patients ofdistal general hospital !yogic,.Ar14198trr: 10. Paykel E% Recent life events and clinical dews* 14:881472. Mon, in Gunderson EKE, Rohe RH (eds) Life Stress and illness. Spdngfield, lino* Charles C. 'Mamas 1874, pp 34. Murphy GE, kmstrong .Aft. Hermele SL, Fischer 134-18& JR. Ckindenin MY: Suicide arid Alcoholism. Marini-

BEST COPY AVAILABLE 2-123 14 Report of the Secretwys Task Force on Youth Suicide

sonal loss colifirmed as a predictor. Arch. Gen. Psychist. 58. Bruhn JO:Broken homes among attempted 1979; 36:6549. Nickles and psychiatric outpatierfte: A comparative study. J Men Splint 10&772-779. 35. FlumphreyJA, Puccio D. Newandsr GD, Casey TM: M of the of selected events kt the lives se. Graer & Parental loss and stimt suicide: A of a : A preliminary noon. J. Nem. huller report. Br J Psychiat Mt 1 it Mont. Die. 154:137440. 60. Gay PM, _Tang, WL: The ists affects of loss of 38. Rosen A: Detection of suicidal patients: an ex- parents In childhood. Br J Psychiat 1037; 113:753-759. ample of some Nmitations In the preclollon of infrequent 81.Hill OVA The association of childhood bereave- events. J. Consul. Psych& 1934; 1&397-401 mentwith suicidal attempt in deptessive Innis% Sr J 37. Paykel E& Contribution of Me owns to causation Psychist 1009; 115:301-304. of psychlabI Vines% Psycho!. Med. 1978 &245-253. 82. Bitiohned J: Some familial and clinical characteds- 38. Pseksl Elk Recent Vie events in the doe of female suicidal psychiatric patients. Br J Perchiat of dtpateM disorder% In Depue RA (ad). Thepsyofioblol- 1981; 131081-390. coy of depressive disorders: implicetions for eve effects 83. Lester 0, Beck AT: Early loss Its a possible `sen- of stress. New York Academic Prue 1979, pp 245-262. sitizer to leer loss in attempted suicides. Psycho! Rep 39. Paykel ES, Holtman JA: Utto events and 197& 194121422. don: a psychiatric view. Trends inNeuroscience:era; 64. Pdarn KS, Latvians JG, Harper et Suicidal ideation 478-481. and permits' loss. A preliminary research report 1973; 40. Hirsch SR, Walsh C, Draper R Parasuicide: A 18:96400. reView of treatment Iniervendons. J. Affect Dia. 1004 05 Roes MW, Dept JR. Campbell RL Parental tow- *290311. ing pommy end suicidal thoughts. Mta Peyohlat Sand 41. Grenvlite-Grosemen KL The early environment In WA 67:429433. affective disorder, in Coocsn & WO A (ads). Recent 06. Goldbery EL Depression end suicide Ideation in Dsvslopmsnts in Affscve Disorders, Royal Medico- IN. young adult. Am J Psychist 1981; 138.3540. Association 1068. 07. Um* ES, Barter JT:Adolescent suiddal be- 42. Finlay-Jones R, Scott R. Ouncen,lones P, Byrne 0, havior. kn J Ordwpsycl*Wy 197* 40:87-06. Henderson & The reliability of reports of sixty separations. Poet N2.0). Rohl& 1081; 152741. 68. NM CL. Young 0, Fowler RD: San Diego suidde study:I. Young vs aid Cases Arch. Gen. Psychlat (in 43. Twang MT:Risk of suicide in the relatives of press). manias, depressives, and timings. J. Clin. 191M 44:308400. 70. knit N: Suicide among minors in Israel. Israel Mn. Psychlet. and Related Dlscrplln.a 1973; 1121008. 44. Wbildnd IX Coleman E: Adult psychiatric disorder and chketood **edemas. The vandity of retrospective 71.+Toolan JM: Suicide and suicidal stlempto in data. Br. J. Papilla. 1982% 143:188491. chikiren and adolescent% Mt. J. Psychiat. 1052 111%719- 724. 45. Pai4tel E& Life events and ewly wwkwyrwmt, in Paykel ES (ed) Handbook of Affective Disorders. -Mn- 72. Jacobziner H: Attempted suloides hi adciesoenos. burgh. QuuolM Livingstone 1062, pp 146481. JAMA 1965; 191:101.405. 46. Uoyd UN Events and Depressive Disorder 73. + Barter JT. Ss/aback DO. Todd * Adolescent Reviewed. I. Events=disposing factors. Arch. Gen. suicide attempts. A follow-up study of hospitalized Psychlat 1980; 37* patients. kW Orm. Psychlat. 1908; 18.523427. 47. Paffenbarger RS, 14Ing OH, MNAL: Chronic dis- 74. Roan RD, Series RM, Kenny TA Reynolds BJ. ease In former college students. DC. amiraWw_Wa In Heald + FP:Adolescents who sttempt suicide. J. mthlitrAmpnjd.iscetiketatheuic:dava!v5iloogiootVal death in Pediatrios 1977; 00:8304 3t. 75. Mohler CL, McKenty PC, Motruols.C: Adoleeoent 48. Paffenbarger RS, Mtn* DP: Chronic diseases In suicide attempts: Some Suicide and femur college students: NI precursors of suicide In early Ufe ThrosteningBehavicreitfar Al:t86412. end midese Rh. Mi. J. Pub. lisaith 1968. &10204035. 78. Mite FiC:Self Poisoning in Adolescents. Br. J. 49. Ray A:Suicide in chronic schizophrenia. Br. J. Psychist. 1974; 1242445 Psychist. 1002a; 141:171-177. 77. Newton K. CI, Grady J. Osborn M. Cole tk Adoles- 50. Roy A.Suicide in recurrent affective disorder cents who take overdoses:Their oherioNtistk a, patients. Can J Psyohist 1984; 28.319-322. s and contacts with helping agencies. Br. J. 198t 10118-123. 51. Rw A:Risk factors for suicide incasychistrict patients. Amh Gen Psychlat 1982, slew 78. Motion K. Osborn N, O'Grady J. Cole D:Clas- sification of adolescents who take overdoes*. Br. J. 52. Cook T, Raskin A: Association of childhood lumen- Psychist. 196t 10124431. tat tee wfth attempted suicide and depression. J. Consul Oin Psyohal 1101; 43:277. 79. Seiden RH: Suicide among youth: A review of ihe literature 1900-1987. Supplement to the Bulletin of 53. Goldney RD: Parental loss and reported childhood Suicidology. U.S. Gov't. PMftng Moe, Washingtan, D.C., stressin you% Irma who attempt suicide.Acte Psychiat 1000. Scan 1981; 04:d4-50. 80. Otto U: Suicidal attempts mode by children and 54. Adam KS, BOUCkorne A. Streiner D: Parental loss adolescents because *school problems. Acta Peedlehica and family Waft in attempNd suicide. Arch Gen Som. 1985; 54:348.358. Psychlee 198a 3&10111-1085. 81. Monaca A. Havaldne At Suicidal behavior as 55. Fwberow NL:Personality patterns of suicidal chNd psychiatric emergency. Noh. Gen. Psychist. 191* mental hospita patient% Gent Psychal Monogr 1060; 42:3. 2*1001-108. 58. Mow LM. Hamilton DM: The psyclioltierapy of the 82. Pfeffer CR, Conte HR. inutchOr R. Jewett 1 Suicidal suicidal patient. Mt J Psychist 1955; 112:814819. behavior in latencyge dtlidren, kn. Aced Child Psychiat. 57. Walton JH: Suicidal behsvior in depreeeive bless: 1979; 1&879492. A 0.1* of aetiological factors in suicide. J Mont Sol 1958; 83. Miller ML, Chiles JA, Barnes VE:Suicide at- 104:M-891. tempts's within a delinquent popuistion. Consul. Clin.

1.17 2-124

BEST COPY AVAILABLE E.S.Paykel: Stress and 1.1fe EVents

Psycho,. Mk 81481408. 84. Owinkol D Prom A. Hood ,k Sulddo In Ogden and adolognmia. Am. J.Payoldrfa 1M125742811. 88 Taylor EA, Manaffold 8k Chadian who poison Ihsonaolvot I. A dialed oompeloon with trata. fk. J. Payable. 1984; 145127-18/

113 2125 CONTROUED COMPARISON OF LIFE EVENTSAND SUICIDE ATTEMPTS ti Cartels Sample Mae Method Findings 1. Genesi Population Canby* a) Studies at Chldren

Adolescent ampler, N a 31 Interview Excess of events In weeks/awes before Jacobs attempt, pardoulady brisk of relationship, (1971) Control adolescent* N 31 PhYsical Dm% MANY, PROnlinalf b) Studies of Adults

Store of Ws events over six months. POW st al P41sinplers N 53 Interview for record Controls N 53 life events Widely last month. (19T5) wet typos st event Higher stress some and number of life Oachrene & Robertson Atlempiers N 100 self lima Controls N 100 oheddist scores events over I year. Especially unpleasant (1075) events, disrupted interpersond Equally for under 25, over 40. Higher stress than both control group*. Isheiwood st al Attemptere a 150 folf report egn) Controls a 200 Total stress scores Time not stalsd. Automobile accident drivers 100 2. Other Patients a) Studies of Children

Higher mean ideas scorn separation. divorce Cohen Sandier st el Suloicird children N20 From cue remade remarriage, hospitalization of went, (1102) Depressed N 21 Other pitychlablo N 35 broken homes, othw Mmes. b) Studies of Adults

Attempters experienced more events in Pay,* et al Attempters a 53 kderview for recent Depressives a 53 life events threatening classes: undesirable; higher (1975) rated stress, uncontrolled. Attemptsm expedenoed mom independent eater &NW. Primary depresehe Interview Adam:lbws N a 14 events exits In )1wir before attempt, particu- (1951) larly batmen onset of depression and attempt. Depteselve N 14

Table 1.

BEST COPY AVAILABLE Controlled Comparison of Life Events and Suicide Attempts (continued)

Comrole Sam Size Method

Luscomb et al Male attempters N 47 Self report More events, perceived 011$6, high stress events (1990) Psychiatric patients N 51 Checklist exits, desirable, undesirable events. kw previous year, no attempter-contral differences under age

O'Brien & Fanner Overdoesre N sa 197 interview Higher rates In previous 6 weeks for most events. (1960) General practice intenders N mg 152 Fall off on one year followerp. Levi et al Suicide attempters N l 40 interview Attempters experienced more separations in last (1966) Suicidal footings N so 40 Separations only year. Suicidal feelings intermediate. Non-suicidal N 40 Stein et al Suicide attempters N NB 165 Interview Memptere experienced more separations in last (1974) Other psychiatric N 185 Separations only year. Significant except for black women where sugges6ve. Greer et a! Suicide attemptere N 148 Intenview only disrupted Attempters experienced mom dbruption In last 8 (1968) Other psychiatric N 148 interpersonal relationships months than either groups. PsychisMo controls Medical controis N Ea 148 intermediate.

Table 'I concluded.

1. )"'

BEST COPY AVAILABLE STUDIES OF RECENT UFE EVENTS AND COMPLE1EDSUICIDE

Study Semple Comparison Group Method Meg* General Population Controls

More logs of parents spouse in previous 2 years. Bunch Suicides N75 General Populatkxt N75 Intenrisw of informants. Bereavement oft/ Maim eapocially urtmeiried mom vulnerablo to (1972) loss of mother. Clusiedng of suNect death by suicide In 4 pars Suicides N 320) Nonsuicidal deaths N320 ',oath Certificates Machtshon & Pugh following WOW* death, especially previous year. (1005) of spousw More changes of Vying oonditions, work problems Hagnell & Rosman Suicidas N 28 Neural deaths N25 Bowies uncioar General population controls N 50 and 6ufect loss than general population In last (1900) year: more object loss than natural death.

Psychiatric Patient Controls

More losses, not all moant, than In Humphrey Psychiatric patient Homicidal °Bonder* N mg 02 Psychological autopsy, Of* Moodie and neurotics. Hortgoidos IMermediate. (1977) suicides on %nowt* Neurotic patients N 76 N as 08 (Mikis only) hawks, charts Ma" adverse events afW dischargo, Pokorny & Kaplan Male psychiatric Patients who did not commit 1IteMew nth relative aikido N20 Motive perticidarly wilefe higher defenseless, (10n) paVeM suiddes piobably refloating depression, during N 20 hospitalization. Psychlwio patient Patients who did not commit Came records No significant difffamnos although tendency Borg & Stahl to more deaths. (1902) suicides N 34 suicido N 34 More losses. Fernando & Storm Patient suicides Non4uloide patients Case records (1984) N 22 N 22

Table 2.

BEST COPY AVAILABLE --ft RELATIVE RISKS IN COMPARISONS OF UFEEVENTS-- IN SUICIDAL AND GENERAL POPULATION SAMPLES

Event Rates Relative Author Time Period Event Patient Controls Odds Suicide Bunch (1972) 2 years Bereavement 7/95 7/150 6.6 Magna il & Rorsman (1980) 1 year Work problems 14/28 10/56 4.6 Object loss 11/28 4/56 8.4 Suicide Attempts Paykel et al (1975) 6 months Undesirable events 32/53 21/53 6.0 1 month Any event 37/53 10/53 10.0

Table 3. 10 3' 0 Studies of Early Loss In Adult Sulelda Attempters

Study Sample Compadson Sample Time of Loss Mena Genial Population Controls

Greer et al Suicide attempters Noneulddal psychiatric patients Death of parent under 15 Greater incidence than both comparison groups. (1006) N so 158 N 158 Medical patients N so 158 Separation, divorce Greater incidence than both comparison groups. under :5 Particularly under 4, and loss of both parents.

Crook & Raskin Depressives with Non-suicidal depressives N 115 Death of pwent under 12. No difference. Higher Incidence In suicidal. (1975) history of 'utak% General population controls N 285 Loss for at least one year attempt N 115 due to separation, divorce, desertion. No significant difference. Goldney Female suicide Women attending a community Death under 18 More loss In suicidal. (1981) attempters aged 1840 health Mire (N no 25) Divorce, separation N 110 No significant difference. Adam Suicide attempters General practios controls (N 102) Deaths under 25 N NI 98 Divorce, separation Significantly higher In suicidal. Psyclatrio Patients Controls No significant difference overall. Farberow Suicide attempters Non-sulcidal patients (N 32) Permanent Ws under 19 More loss before 8. (1950) N 32 Stilaide threat (N 32) OM k seriously suicidal vs 1516 in each Moss & Hamilton Seriously suicidal Potendely suicidal Inpatients N 50 Permanent loss in early life comparison group. (1055) inpatients (N 50) Non-suicidal patients N 50 Significantly more. Walton Psychiatric, patients Non-sulcidal patients N 60 Temporary loss under 15 (1958) with suicide attempt or threat N so 430 More loss in suicidal. Bruhn Suicide *Hampton Pep:blind* outpatients N 91 Temporary loss under 15 (1962) N 91 Non-suicidal psychiatric Inpatients N 50 More loss.. Both over 30 and under 30. Greer Neurotics or sociopathic Other neurotic or sociopathic patientsAll loss for 1 year under 15. More loss (p < .06) (1966) patients with history of N 385 Death attempt N13.1 More loss in suicide attempters. Gay & Tong, Psychiatric patients Non-usicidal psychiatric patients Separation for at least (1967) with history of suicide N oo 382 six months prior to 15 attempts Ng. 111

Table 4. I 5 3 BEST COPY AVAILABLE SEXUAL IDENTITY ISSUES

/mph Harry, Ph.D., Associate Pmfessor, Depamnent of Sociology, Northern Illinois University, De Kalb, Illinois

Study Selection Criteria In this literature teview, it was necessary to ages among the female groups were 12 create several criteria for selection of studies (N=57) and 5 (N=43).All four gxoups and for emphasis of certain types of studies. generally studied young people in their twenties or thirties.In a much larger study in I. Since much of the literature dealing with which the four comparison groups were sexuality and suicide does not specifically matched on age, race, and education, Bell focus on adolescent or youthful popula- and Weinberg found the following percent- tions, it was decided to include all studies ages of subjects who bad ever attempted dealing with the relationship between suicide:homosexual men, 18 percent suicide and a particular aspect of sexuality. (N=686); heterosexual men, 3 percent Whereas this decision increased the num- (N=337); lesbians, 23 percent (N=293); ber of studies to be reviewed, the final heterosexual women, 14 percent (N=140) result was largely a study of youth suicide. (2). Similar sIgnificant (ANOVA) differen- This arose because many studies of ces in their t, ;stories of suicide attempts have sexuality are also stud.= of the young. been found in comparing lesbian and nonles- 2. Those studies with control groups have bian female prisoners, although means or been given greatest emphasis. However, percentages were not presented (3). single-group studies are also included. Turning to studies of clinical populations, 3. Primary emphasis was given to studies of one study of 500 psychiatric outpatients attempted and completed suicide with less found that 50 percent of 12 homosexuals attention to suicidal gestures, suicidal compared with 13 percent of 488 non- ideation, or self-mutilation. homosexual patients had attempted suicide (4). A 6 to 12 yeir followup study of these 4. Little attention has been given to clinical obsetvations, interpretations, and studies patients reported that two of the homosexuals had committed suicide (5). reporting one or two cases. This suicide rate was 17 times greater than the age- and sex-specific death rate for that Homosexuality State. Another study of 60 homosexual men The literature clearly and consistently shows with a mean age of 20 found that 32 percent that homosexuals of both sexes attempt had attempted suicide at least once (6). As suicide much more often than do might be expected, the percentages who had heterosexuals. Saghir and Robins found that attempted suicide are higher among the clini- 7 percent of their 89 homosatual males and cal than among the nonclinical populations. none of the 35 heterosexual controls had at- The literature also speaks clearly to the na- tempted suicide (1). The respeciive percent- ture of suicide attempts by honiosexuals, al-

1 5 9 2-131 Report of the Secretary's Task Force on Youth Suicide though that differs by gender. Five of the six of the problems of these troubled youths. homosexual male attempters reported by Saghir and Robins had made their attempts Confusion Over Sexual Identity before the age of 20 during conflict with family members or within themselves over their The topic of confinion over sexual klentity incipient homosexuality (I).Five of the has been here interpreted to mean seven lesbian attempters had made their at- transsexuals. Althouga it could also refer to tempts during their twenties daring a depres- the case of young homosmuals attempting to sion following the break-up of a relationship. accommodate themselves to their sexual Bell and Weinberg similarly report that orientation, that has been dealt with in the homosexual men in their study generally at- previous section.All except one of the tempted suicide at an earlier age than the les- studies dealing with transsexuals and suicide bians (2). The homosexual men were more are single-group studies lacking a control likely to report that their attempts were re- group.In the one exception, male l a ted to trying to deal with their transsexuals still living as males had made sig- homosexuality than was the case for the les- nificantly more suicide attempts than male tsanssexuals living as females, homosexual bians, whose attempts were more related to psychiatric patients, homosexual non- the break-ups of relationships. patients, and heterosexual nonpatients (12). The above data indicate that the period of Also, the two transsexual groups were sig- coming to an acceptance of one's nificantly higher in suicidal thoughts than homosexuality--"coming out"--is a period were the other three groups. The sample that may be accompanied by a heightened sizes in this study ranged from 19 to 25. risk of suicide attempts. The average age of Turning to the single-group studies, coming out among homosexual men has been Walinder reported that 20 percent of 30 found to be 18 or 19., (7,8,9) this seems to Swedish male transsexuals and 8 percent of coincide with the time when homosexual 13 female transsexuals had made attempts men are at risk of attempts. Two studies of (13).These percentages should be con- large nonclinical populations of homosexuals have reported that just under half of the sidered minimum estimates because the author only reported attempts documented respondetis agreed with the item: "Before I in hospital records, thereby excluding those came out, the idea that I might be not leaving such records. Another study of homosexual troubled me a lot" (8,10). In a 72 English transsexuals (55 men and 17 recent study more intensively analyzing the women) reported that 53 percent had made Bell and Weinberg data, Harry found that, attempts. Four (5.7 percent) had completed among the homosexual men, being troubled suicide (14). Person and Ovesey reported over one's homosexuality during adoles- that six of ten male transsexual patients were cence was related to subsequent suicide at- preoccupied with suicidal thoughts and two tempts (11). Among the men of both sexual orientations, childhood cross-gender be- had made attempts (15). haviors were related to attempts. Among all Although these studies dealing with four study groups, being a loner during transsexuals either lack a control group or do adolescence and general adolescent unhap- not report means or percentages of those piness were related to attempts.It thus who had made attempts, the percentages seems that a number of the pre-adult charac- available indicate that transsexuals may be at teristics and experiences of homosecuals may higher risk for suicide attempts than other create special difficulties for them and that, groups at risk, e.g., homosexuals, and much for some, suicide may seem preferable to higher than the general population. As a other solutions. Studies suggest that trying matter of caution when discussing the at- to grow up homosexual in a culture organized tempts of traossexuals, it is important to dis- for heterosexuality Rely contributes to many tinguish between past attempts and threats,

2-132 J.Herry: Somali Identity Issues attempts, and self-mutilation to induce a promiscuity and suicide attempts in a popula- therapist to undertake desired medical action of 105 adolescents who were pregnant tions. Such threats and acts seem to be fair- (20).The 14 (13 percent) who had at- ly common among clinical transsexuals. tempted suicide during the two years follow- However, as far as possible, such attempts ing pregnancy were more tritely to have had have not been included in the percentages a venereal disease. While the presence of cited above. venereal disease indirectly suggests promiscuity, the 14 attempters were also significant- Acquired Immune Deficiency ly more Mt* to be single and have a number of other problems. Hence, isolating these Syndrome (AIDS), AIDS-Related girls' sexual activity as a contdbutory factor Complex (ARC) in their attempts should be viewed with ex- Since the phenomenon of AIDS is quite new treme caution. and since the associated medical prospects are frequently changing, little can be said Victims of Physical and Sexual about the relationship of AIDS to suicide at- Assault tempts or completions. Approximately 70 to 75 percent of persons with AIDS or ARC are The strongest relationship between suicide male homosexuals.However, there have attempts and physical abuse was reported in been cases reported of both attempts and an Australian study comparing 20 suicidal completions (16,17,18).Also, one and 50 nonsuicidal children under 14 from a homosexual man who had been unsuccessful child psychiatric hospital (21). Sixty percent with suicide attempts devised an interesting- of the suicidal children versus 4 percent of and successful-means: having sex with per- the nonsuicidal had been subject ed to paren- sons known to have AIDS (19). tal physical abuse; 65 percent versus 4 percent had also witnessed physical fights With our current medical knowledge of between their parents. However since one is AIDS as a classical terminal illness, one might more likely to be interested in predicting expect suicide and attimpts. If attempts and from abuse to attempts, rather than the completions are not directly due to the medi- reverse, these percentages should be made in cal aspects of AIDS, they may also be the in- the other direction. In that case, the percent- direct result of societal reaction to the age among the abused who were suicidal was individual with AIDS. For example, one 86 versus 14 among the nonabused. Viewed homosexual man with AIDS attempted in this way, these data appear to show a mas- suicide after his employer fired him for sive and significant association between having the illness (18). Similarly, AIDS can abuse and suicidality. However, this associa- result in evictions, loss of a lover, and loss of tion is very likely an artifact of the clinical a sex life. ATTIS can also effect an involun- populations studied.Suicidal children are tary dischs tre of an individual's probably much more likely than nonsuicidal homosexuality to others. Hence, these in- ones to come to clinical attention with direct effects of the illness potentially could problems. Also, abused children are probab- enhance the possibility of suicide. No infor- ly more likely to come to clinical attention mation is currently available on any link be- than are nonabused children with problems. tween ARC and suicide. Children who are both abused and suicidal are thus more likely to come to clinical atten- Sexual inhibition or Promiscuity tion than either the simply abused, the simp- Computer searches of the STEC data base, ly suicidal, or children who are neither. Psychological Abstracts, and Medline Hence, abused sad suicidal children will be facilities produced nothing on this topic. far over-represented in clinical populations and thus suggest an apparent association One study suggested a relationship between even when none may exist in thegeneral

1I 2-133 Report of the Secretary's Task Force on Youth Suicide population. These data suggest that clinical been raped once indicates a possible populations may be very inappropriate for relationship between suicide attempts and studying the association between two being multiply raped (25). All subjects were phenomena when both of those phenomena from a hospital rape center. The multiply are major correlates of, or criteria for, admis- raped women were significantly more likely sion to a clinical population. to have attempted suicide (52 percent vs 16 percent), to be younger, poorer, lona% im- Another study compared 60 abused children, 30 neglected ones, and 30 nom:Unica! migrants to the city, and to have had more children (22). 'Their respective rates of self- psychiatric treatment. Although there may destructive behaviors were 41 percent, 17 be an association between multiple rape and percent, and 7 percent. The percentage attempts, attribution of causality here is among the abused who had made attempts clouded because of the apparently disor- was 8.5 (figures for the other two groups ganized histories of the multiply raped vic- were not given).Again, since both self- tims. Both their suicides and rapes may be destructive behaviors and abuse are both symptoms of disorganized personal histories. A suggestive study of 13 young (18 to 23 likely to be brought to either clinical or offi- years) military servicemen who had been cial attention, the reported associations may well overstate the strengths of the true as- raped by other military personnel found that sociations. two (15 percent) had made subsequent suicide attempts out of feelings of deflatiois The studies of sexual abuse and suicidal beof their manhood (26). A study of prison haviors are also from clinical populations. sexual violence in New York State men's Herman and Hirschman compared 40 prisons found that 38 percent of 107 inmates women who had had incestual relationships who had received verbal threats to physical with -heir fathers, with 20 women who had assault for sex had made suicidal gestures had seductive, but not incestual, relation- (27). This percentage was more than twice ships with their fathers (73). The women in that among 45 sexual aggressors recorded in both groups were private psychiatric inmate fdes and 17 times that among tnnhar- patients. The researchers found that 23 per- rassed prisoners. These data provide some cent of the former and 5 percent of the latter presumptive evidence that, at l::ast among had attempted suicide; the difference was young males, being subjected to same-sex statistically significant. A single-group study sexual harrassment or abuse in situations of sexually abused girls from 201 families where there is little escape may precipitate served by a protective service agency found suicide attempts. However, it is pouible that that eight girls (4 percent) from these those prone to suicide may also be subjected families had also attempted suicide (24). All to more threats and assaults. attempts occurred when the girls were 14 to 16 years old. Since the families of these girls Pregnancy-Related Suicklality were generally chaotic and conflictful, it is difficult to attribute direct causality to the It is vety difficult to conclude anything about abusive behaviors of their fathers. Also, the the association between suicide and preg- percentages of attempts in this study and the nancy because of the diverse types of studies, one previously mentioned (22) are not par- the varying definitions of the term *pregnan- ticularly high when compared with groups cy-related,* the lack of adequate coLtrols for discussed earlier. Although there may be a age, and the small sample sizes. relationship between sexual abuse and 1. Types of Studie& The studies divide into suicide attempts, these studies cannot show those that show the percentage of preg- one. nancies among attempters, those that A study comparing 25 women who had been show the percentage of attempters among raped several times with 92 women who had the pregnant, those with adequate control

2-134 162 Many Sexual Identity Issues

groups, and those of completed suicides. tions describing this study leave it unclear as to whether the number in the sample, which 2. Definitions of "Pregnam-Related.* The was three-quarters female, was 20, 48, or 68. definitions include pregnant during a suicide or attempt, postpartum attempts Pregnanties Among Attempters and Con- or suicides, belief that the attempter was trots. It seems that in only two studies did the pregnant, and overdue menstuation. 'marchers bother to obtain a nonclinkal These varying definitions make any con- control group matelmd for age with the at- elusions from this literature almost impos- tempters. Jacobs compaird 50 adolescent sible. attempters, of whom 38 were female, with 22 Since suicide com- female age-matched adolescent nonat- 3. Controls for Age. tempters (22). Matches were made on age, plete's are generally middle-aged to old race, sex, and maternal education.The data and both attempters and pregnant women showed that 21 percent of the attempters are generally young, controls forageand were either pregnant or believedthemselves fairly precise onesare mandatory. to b.: at the time of the attempt compared The above-listed problems occur in the scien- with none of the controls. Since the figure of tific literature about pregnancy and suicide in 21 percent is very close to Teicher's figure of many, if not all, possiblecombinations. 22 percent in the previous study and since the Below, we proceed through the most com- two authors worked together, it is unclear if mon groupings. they are actually describing different sets of Attempts Among the Pregnant. One study respondents. The 22 or 21 percent figures of these studies are much higher than the figure of 105 pregnant girls under 18 at an urban of 6 percent of attempters who are pregnant hospital found that 14 (13.3 percent) had attempted suicide at least once during a 2-year reported by Whitlock and Edwards. However, since the former studies include followup (20). Clearly, these atterrnts are postpartum and beyond. The attempters both real and suspected pregnancies while the Whitlock and Edwards study includa were significantly more likely to besingle, only real pregnancies, no conclusions can be Catholic, to have had venereal disease, and to have come from higher socioeconomic drawn. arMs. This study lacks a control groupof Birtchell and Floyd compared 107 female at- adolescent nonpregnant girls. tempters with 110 female nonattempters Pregnandms Among Attempters. Compar- with a control for age (32). Of the attempters, 12.1 percent comparedwith 2.7 ing 30 pregnant attempt= with 453 non- percent among the nonattempters, were pregnant attempters from Australia, either pregnant or overdue in their Whitlock and Edwards concluded that since menstrual cycles. Ten (77 percent) of the 13 6 percent of suicidal women are pregnant pregnant attempt= were unmarried. Be- compared with 7 percent of women who are cause the dermition of "pregnancy-relater pregnant in the population at any given time, employed in this study includes being ncy is not a predictor or inhibitorof In their study, the menstrually overdue, comparison with the atteiapts (28). immediately preceding three studies is, once pregnant women were less likely to be married than the nonpregnant (53 percent vs 41 again, impossible. percent). Teicher found that whventy-two Studies of Completed Suidde. A study of percent of all suicide-attempting girls com- completed suicides in Minnesota for the pared to zero percent of control girls were years 1950-1965 foundthat 1 percent either pregnant or believed themselves to be (14/1019) of female suicides were pregnant pregnant" (29,30). It is difficult to determine (33). The study reported that suicides per the sample size on which this 22 percent 100,000 population were 16 for men and 4 for figure is based because the various publica- women during this period.For pregnant

2-135 Report of the Secretary's Task Force on Youth Suicide women, suicides were close to 1 per 100,000 Conclusions live births (14/92,982). None of these 14 pregnant women were unmarried. The Homosexuality. Homosexuals of both sexes are two to six times more likely to attempt author concluded that pregnancies during suicide than are heterosexuals.Data on suicide were extremely low and that pregnan- completed suicide do not exist. Kinsey et al. cy served as a protector against suicide. Ten estimated that 10 percent of all males and 113 of the 14 pregnancy-related suicides in this to 112 that of females are predominantly study occurred postpartum. While this fur- homosexual (35,36).However, because ther reinforces the author's conclusion, it Kinsey's data included large numbers of also suggests that the postpartum period may prisoners, Gebbard later revised Kinsey's es- be the time of greatest, although still ex- timates to 4 percent of white males with at tremely low,&ILIt shoulkt be noted from least some college education and 1 to 2 per- this study that using live births as a proxy for cent among all adult females (37). The data the number of pregnant women is a more ac- show that these two populations are marked- curate denominator than using the total ly at risk of attempted suicide during late universe of women in the reproductive age adolescence and early adulthood. range (15 to 45). Confusion Over Sexual Identity. Whereas A later study disputing the conclusion that studies with adequate control groups are pregnancy is an inhibitor of suicide presented lacking, the percentages from the single- data on 8 pregnancy-related suicides among group studies strongly suggest that a total of 47 suicides in New Mexico (34). transsexuals may be an extremely high risk lvelve and a half percent (4132) of female group for attempted suicide. Adequate data suicides of ages 15 to 34 were pregnant. This do not exist for completed suicide. age range was provided by the authors to coincide better with the principal years of AIDS. Only case reports exist. However, childbearing. While the 12.5 percent figure there are strong reasons to suspect that is considerably higher than that from the suicidal behavior may be very common Minnesota study, the numerator (four) is so among AIDS sufferers. low that the percentage is extremely unstable Sexual Inhibition or Promiscuity. There It should be from a sampling viewpoint. has been no research on this topic. If re- noted that all of the eight women who com- search is undertaken, it is important to attend pleted suicides were married and that their to the sexual behaviors of both men and means of suicide were principally guns and women to prevent continued bias on labeling hanging. of promiscuous behavior in reference to Despite the dreary inconclusiveness of the women only. various studies on pregnancy-related Vktims of Physical or Sexual Abuse. Al- suicides, one finding does emerge. Most though the data on this topic come mainly prerant attempters are single, whereas from single-group studies, there does not ap- most pregnant completers are married pear to be a markedly high rate of attempts (20,29,3Z33,34). Also, the means of suicide among abuse victims. The two groups of for the single attempters are drugs primarily abuse victims in which there might be an whereas the means for the completers are elevated rate are women who have been mainly guns and hanging. These data suggest raped on several occasions and men who that the two groups are different popula- have been raped, or sexually assaulted by tions, although with some overlap. Hence, other men. being in a pregnant population at risk of attempt may only marginally affect the risk of Pregnancy-Related Suicide. The literature completing suicide. on this topic is too conflicting in its findings to conclude anything. However, the two sub-

2-136 164 JBarir Sexual Identity Issun groups who maybe at risk are single pregnant events, the difficulty arises of obtaining suf- women and married postpartum women. ficient numbers of such persons in a general The former group may be at risk of attempts, survey to analyze. This difficulty can be over- while the latter group may be at risk of com- come by focusing the suivey on limited seg- pleting. ments of the general population such as high school students or pregnant women. Also, Recommendations tor Future the numbers for analysis can be increased by Research asking about suicidal behaviors not only of the respondent but also of a limited set of Control Groups. We need control groups to close relatives. This recommendation of sur- determine whether a given population has a vey research does not extend to possible re- high or low tisk for suicide. search on AIDS or ARC and suicide since, Nonclinical Control Groups. While com- given the infrequency of both of these paring suicidal with nonsuicidal patients is phenomena, survey raearch on this topic convenient, it can only snow that one special- without massive and expensive samples be- ly defined population may differ from comes impossible. However, given the con- another specially defined population. siderable potential of AIDS as both a direct and indirect cause of suicidal behavior, one In the case discussed earlier where one ex- would hope that research on the relationship amine the auociation between attempts and be soon undertaken. some other clinical characteristic, both of which serve as criteria for becoming a clini- As a concrete example of my proposed form cal patient, it is !tidy that one may find total- of research, I suggest a survey of the general ly false associations. When introduced to the population which includes a question on clinical literature, such associations can mis- sexual orientation. To date this has not been lead others into pursuing, or funding, lines of done. Until the 1960s, and especially the research that may be ultimately unprofitable, 1970s, homosexuality was a topic studied lar- especially if pursued in clinical settings. gely by clinician& Beginning with the worb of Evelyn Hooker, there began a continuing Nonclinkal Experimental Groups. It would stream of studies based on nonclinical be most desirable to obtain both experimen- samples of homosexualtwthe results of which tal and control groups from nonclinical differed much from those reported by populations through conventional methods clinicians (40).However, all studies of of survey research. For example, it has been homosexuals to date still involve non- possible to profitably study family violence probability samples, hence, estimates of rates through survey methods, including even of suicide can only be approximate (41). The telephone surveys (38,39). Such research time may be ripe for a survey of the general was an immense advance over earlier re- population that include a question, possibly search of family violence based largely on placed among the demographic items, on clinical samples. Two major hurdles to con- sexual orientation. This would permit more ducting survey research on attempted or accurate, if still imperfect, estimates of the completed suicide come to mind. Fust, there true risk of suicidal behaviors among such is the intrinsic sensitivity of asking persons population& about the suicidal behavion of themselves or of other family members. However, since it has proven possille to ask people in suivey REFERENCES situations about the crimes they have com- 1. Seghlr M Robins E:Male end female mitted, deviant sexual activities, and family homosexual* Baltimore: Mama and MM., 1117% violence, it should also prove possible to ask 118-, 276-277 2.Bell A. %%IWO PA: Nomosexuallties. New Yode about suicide. Second, since attempted and Oman and Schuster, 1918:450. completed suicide are infrequent-to-rare

1 f; 2-137 Repott of the Secretaty's Task Force on Youth Suicide

3. Clermont C. ErvIn F RoMns A, Piulohik R Satinet- 30. Teicher J, Jambe JAdolescents WI° attempt C:Epidemiological studies of female prisoners: suicide. Mr. J. Psychiatry 1980; 1221248-1257. Homosexual brii.ra J. Nom Mont Ole. 1977: 18425- 29. 31. Jacobs Ik Adolescent suicide. New York May, 1971:87-435. 4. Vitrodruff R. Clayton P. Guts & Sukids and psychiatric diagnoses. Dis. New. Slit 32. Sklohnea J. Royd & Further menstrual chraoleds- 1972"3171I- goof suloideattempters.4 Psychosom. Res. trilk 1&81- 95. 5. Martin R Mellow R. Gure S, Clayton P: Modality In a fo#4 5006arlatdo outpatients. kdt Gen. 33 Boma k Criminal aborting, deaths, Psychiatry 1; VWdeaths, and suicides InprognanciM dirtste , 98:358-387. 8. Roesler T, Deleher R:Youthful male homosexuality. JAMA 1972 2191018-1023. 34. Goodwin J. Herds D: Suicide in pregnancy: The Hedda Gabler syndrome. Suicide and We-Threatening 7. Dank & Coming out in the gay world. Psychiatry Behar. WM; *105-115. 1971: 34:180-197. 35. lOnsey A. PomiKw1V. Martin 0 Sexual behaviors 8. Awry J, DeVeN W: The social organization of gay in tho human male. MirWelphis: W.B. Saunders, males. Nwo& Praeger 19*04. 1948:851. 9. Troiden R:Becomingay. (Dissertation). 38. Kinasy A, Pomeroy W, Martin C, Gebhanl P. Stonybrook, NY: State of Ww York. 1977. p. Sexual behavior in the human female. Philadelphia: Ka 183. Saunders, 1910473474. 10. Hwy J:Gay children grown up. New York: 37. Gebhard P. kicidenee of overt homosexuality in Praeger, 1942154. the Willed States and %intim Europe. In: LMnitood, 11. Harry J:Parasuicide gender, and gender et Notional Institute of Mental Health Task Force on deviance. J. Het and Social Behar. nea 24:350581. Honvesxuality, Final rAtand Background Paper& Washington, DC:GPO. 12. Langsvin Ft Plinth D. Steiner a The clinical profile of male transsexuals iving as females vs those 38. Straus U. Odin R, Sleinmelz & Behind dosed living as males. Arch. Sex May. 1977; 0:143.154. doors. Garden Oty, NY: Doubleday, 1080. 13. Walinder J: Transsexualism. Goteborg, Sweden: 39. Schulman U. A noway of *ousel rebuitgo= Scandinavien UnNersity Books, 1987. women in Kentucky. New York Lau Herds 1979. 14. Huxley si, Brandon 8:Partnership in transsexualism, part tPaired gird non-paked groups. 40. Hooker E:The adjustment of the male own Arch. Sex Balmy. 1981; 14:C133-141. homosexual. J. Prof. Tedmiques 1957. 21:17-31. 41. Harry J: Sampling gay men. J. Sex Res. 1900; 15. Person, E. °veiny L: The tranesexusi syndrome in 2221-34. msdes:I primary transsexualism. kn. J. Psychotherapy 1974; 2&4-21. It Anonymous: AIDS likely motive for double suicide. Edmonton Journal. Oct 25, 1985 (Reuters). 17. Anonymous:Suicidal AIDS victim saved. Son Francisco Chronicle, Mayil, 11185. 18. Anonymous: Gay man with AIDS attempts suicide after being fired. Advocate, Jan. 21, 1988, Issue #435. 19. Frances R, Wkstrom T, Alms V:Contracting AIDS as a moons of committing suicide. Mr. J. Psychiatry 1985: 142858. 20. Gabdelson L Kerman I., Curds J. Tyler N, Joke! J: Suicide rounWn a mrirdon pregnant as teenagers. Am. J. Public Hula 1970; 00.2,019-2301. 21. Kooky it Childhood suicidel behavior in battered children. Am. J. Pephistry 19133 24:457-488. 22. Green A:Self-destructive behavior in battered children. Am. J. Psychiatry 197& 135:579-552. 23. Herman J. Hirschman L: Families at risk for father- daughter incest. Mi. .L Psychlaby 1981: mama 24. Goodwin J: Suicide atlempts in sexual abuse via- time and thek mothers. Child Abuse and Neglect 1981; 5:217-221. 25. Ellis E, Meson B, Calhoun K: M examination of differences between muWpl and eingle-Inoldent viotkns of sexual assault. J. Abn. Psycliol. net 91221.224. carairr P. Eddleman Same-trex rape of non-in- men. Mt J. Psychiatry 1904; 141575479. 27. Lockwood D; Prison sexual violence. New Yodc El- sevier, 1930:80-74% 28. Midair F, Edwards E: Preen/ix., and attempted suicide. Comprehensive Psychiatry 1988: a1-12. 29. Michel. J: A solution to the chronic problem of living: Adoleecent attempted suicide. In: &hooker J, ed. Current Issues in adolescent psychiatry. New York: &un- nor- Mazol,197&12/1-147.

2-138 Risk Factors Potentially Associated with Adoleecent Suicide of Attempted Suicide

Outcome Experimental Control eigralieencee Mk holm variable Group Group fbiultt TWO

14211113011XISSIfty Sok 964nboos N Homosolueloo. *Adds con:nork SV *alp my mon. 11% X2 Now Yoe% 19711, Simon A Sohuslor, p. elbobb 293 140 olcol=on, 48D dile* Smoin 31% 2391; ISWilkt bmilo. 14%

Cam& a. EMN p., Polkh, A, oulddo 27 sollkopadsd leobbno mob, 1N00/A Pkaoltb,1114Mholl. C. Epkismlobal- allsmpb lostabo. V WO. wornonVlirairploon ;lea= nal ob Mid= of boob Ohm= WNW WOW* homoommal beludon. J. Nob Wad h Own OM 19Th 11142101

Harp J. Ponrulokb. ghldors & f4o44$ aikido Moms op BuI$ mo WW1 A dowamol Lopinow dadortoo. JUNO I Soo Bohm t&Ift disobb Wobbly oboe Weinberg Mob am mud 4,15V999 24.1180.91. bar moblod b iambs

Magin. C941311pr, Ft. Gun. 11. It goi mob 490 after ipirp aomplob Xt. Ramon ahren, P., Malilly In a Oollosmip of solekbo outpollonlo 0109.4110 mob ouldbo KO m* oulpobleo. Ash Oen We cam PayoNey111* MSS& MOM

Flosolor. T. Ibbhoo. R. Y4400.9 hob suicide ID young (* nun. 312% mode none homoonswIty. .1KM/11117Z 21e19111. allompla fpw ftIS allomplo KG&

3oglik, M. Robin% E.. Mole end suicide so pay mom 35 stalptd mom OW mon. 7111 Xa %mob homcoonally. BIIMMOIN Wil. abonple 37 Mebane 43 dmIGht toms drol* rolmon. bore ft MS% 11,73; 111279.277. 01110 MSc *WO Com^ OIL

Womb", R. Chflon. P.. Gosh & wields 12 goy mob 499 nonim guy mina OM mono Sublito Oomph mid popohloho dlot alornp10 011001101111 obpollonlo ohm 1316 MONA Me i Mow io 111714 MO- MK Table 1. Report of the Secretary's Task Force on Youth Suicide

2-140 Risk Factors Potentially Associated with Adolescent Suicide or Attempted Suicide

Outcome Experimental Control Significance Mk Factor Citation Variable Group Group limas Teets*

Pregnancy- Bank A. *Mind abodfon drew complaled 14 pregnant O&M We births suicide roe of none Mated gleglirnale Pregnsney in mummy. J suicide suicides in Minn.. In Minn., 1 Pregnant Cbci aarl Oro 1937; 9114150-357. 1950-1555 MUMS lemon is about MORONS

Elincimell..1., Floyd, & Further suicide 10? *nab si- 110 fanmie non- 1311*8% of none inensbui chameNdsike oi suicide al- allempla tempiers allemplent contra, sem tensplers. J Psychosom Rse 11175; msichad tor ow WPM cff 19:014& menstrually omit*

Gabdelson, t.. Mennen. L. Cm* L, subsequent 105 under 1$ none 13:311 made none Ver. 14.. Jebel./ Suicide ellempte In suicide pnegnwe patients allenmts in 2 populalion pregnant se Monagers. Wen** yews eller ArnJ Alba Welt 1975; attoseam. owe birth

Cloodoln..1.. Harris. 1:). &Aside In pet completed $ pmgnant or 47 female 17% Ilentele none nervy. Waldo and Llie-rweelenky suicide Plattpmturn Weds, in aikidos are Beher..1971% 51015413: during suicides in New Mexico Paereill ar magerancy New Mexico mulpse -... ce prutparturn

Jeurtbsol. Mokenent suicide. Nes Pagralla Or 311 adolescent 22 fames age- 2114 is OM ot none Yale ay, 1971; 5743: Waisted tamale anampisre makhed corm* lentaie conkob sell lo be Nue pregnant or belisted see to be

Teichst..1. A solution to the chronic PregnantOf teens* adoisecera 22 ternste 2211 wars 014 none problem ci hang: @dolmens al- Wood snetspienc N adolescents at controls sere templed suidde. In: Schools', .I. Co- ea to be 20os 418 or OS Pregnant rent issue* In saloiseced psychiatry. Neer Vat Eleurmsr-Maapt, 1915: 129- 147.

Whitlock F.. Weenie. E Pregnancy bekig 30 Pregnant 453 nonpreg- MS ot Olsrnpt- none and anompums suicide. Cornprehen- illegnera allemplere In nerd atternplass ere me pregnant oko Poyohlesy IN& 51-12. Australis in Aimbills vs ?II ot sii women In Augusta aged 15-43

Table 1 continued. 171 172 Report of the Secretary's Task Force on Youth Suicide II9 tifli Nit 1111 II

2-142 1041 "MAJOR PSYCHIATRIC DISORDERS" AS RISK FACTORS IN YOUTH SUICIDE

Maria Kovacs, Ph.D., Department of Psychiatsy, Western Psychiatric Institute and Clinic, University of Piusbwgh School of Medlcifle,Pittsbuigh, Pennsylvania loaquim Puig-Antich, M.D., Department of Psychiatiy, Western Psychiatric Institute and Clinic, Univerdty of INttsbwgit School of Medicine, Pittslnugh, Pennsylvania

The aim of this review is to assess whethe. among them. But, the resultant figures are major psychiatric disorders constitute risk best interpreted against base rates derived factors in youth suicide. As the data on this from demographically comparable popula- matter are limited, we have chosen to present tions.Finally, the retrospective, case-con- succinct reviews of three different sources of trolled study of select samples of youth information that can illuminate this relation- suicides could shed light on the contributing ship, namely: ID the types of psychiatric diag- role of mental illness.Strictly speaking, noses found among youths who have however, both the latter approach= identify completed suicide, (2) psychiatric diagnoses correlates or associated features of suicide. among youth who attempted suicide, and (3) the relationship between familial aggrega- Although there has been increasing concern tion of suicidal behaviors and familial ag- about youth suicide in the United States and gregation of major psychiatric disorders. The other countries (1-.. ), the evidence about its limitations of each data source and of each association with specific psychiatric diag- group of studies will be discussed in each sec- noses is neither scientifically unassailable nor tion. unequivocaL The scarcity of conclusive data is partly a reflection of the enormous technical, practical, and ethical problems that are Psychiatric disorders and youth inherent in the study of suicide, including the suicide extremely low base rate of its occurrence. The strongest evidence about the suicide risk Additionally, data from earlier publications paced by psychiatric illness would have to cannot be readily interpreted because, in the come from prospective studies of clearly last 15 years, there have been substantial defined psychiatric cohorts. Ideally, such a changes in the use of psychiatric diagnoses in study would entail at least two different diag- juvenile cohorts. Nonetheless, the available nostic samples; a suitable, psychiatrically un- findings are clinically alarming and suggest affected group of controls; and a period of that major psychiatric disorders play an im- followup sufficient for the outcome of inter- portant role in youth suicide. est (suicide) to occur.Alternatively, a retrospective study of a consecutive, un- Suicide among psychiatrically Ill selected sample of young people who had youth: Prospective studies died by suicide could provide data about the incidence of various psychiatric disorders The mere fact of "patienthoodff is apparently

175 2-143 Report of the Secretwys Task Force on Youth Suicide associated with a higher than expected rate there were no suicides among the 'antisocial of suicide across the age span. Morrison (6), personality' cases, although well had con- whose database consisted of 12,104 patients comitant alcohol and drug abuse and most (7 (age range not given) treated over an 8-year out of 9) were hospitalized because of intetval (19724980) by a San Diego-based *depressive symptoms and suicide attempts? private practice group, computed the cohorts We used the same computational procedure suicide rate by age and sex (controlling for on data from King and Pittman (8) data who years of followup). As shown in Table 1, the determined the 6-year followup status of 65 patients had a consistently higher suicide rate consecutively admitted patients (aged 12 to than similarly aged males and females in the 19). This group was an earlier cohort from general county populaqon. This higher rate the same university hospital that provided was most marked for patients aged 19 or cases in the study of Weiner et al., with index younger, who committed suicide at roughly admissions from mid-1959 to mid-1960. Vol- 40 times the expected rate. lowup status was ascertained via structured There is evidence that suicide among youths interviews with the patients themselves, is associated with certain major psychiatric adult informants, andior hospital records. disorders more than with others. Weiner et The diagnoses (listed in Table 3) were al (7) reported on the 8 to 10 yew followup derived according to symptomatic criteria status of 77 psychiatric inpatier ts whose specified by the authors. On followup, all index hospitalization occurred bet xv.,n 1965 available data were used to determine the and 1968, at a mean age of 16. Or tollowup, diagnoses. However, the findings were a comprehensive interview verified initial presently only for 51 casts who fell into two diagnoses, according to the Feighner criteria. general categories: affective disorders (91 = The data were gathered from the patients, 26) and 'other* disorders (n = 25). The one significantothers, or records of case of suicide was a patient whose diagnosis rehospitalizations. The authors provided the was 'organic brain syndrome? This study is number of cases in each diagnostic group notable because it illustrates changes in diag- who committed suicide 0, .:r the followup in- nostic practices and/or in the actual manifes- terval. Using these figures, we computed the tations of psychopathology. For example, number of suicides in a followup year, and the authors specifically statt4 that none of then, the portion of each diagnostic group the patient met criteria for "addiction and al- who committed suicide in a year. Then, we coholism? In contrast, in a current inpatient multiplied the resultant figure by 100,000 to adolescent sample, one is far more likely to estimate the met suicide rate per 100,000 diagnostic population per year. Please note Suicide Rates in a Sample of Psychiatrically Hospftalted Adolescents that these rates are only 'rough estimates' in at &Year Follow-up part, because inpatients may not represent (Adapted from King 8 Pittman, 1970) the entire population of cal= with a given AgetbdW diagnosis. We realize the shortcomings of Olimmosisio hb. Rats this method, but the rates are easier to com- Affective disordw 28 pare this way. (depressionfinanla) Schizophrenia/ 5 .1=1. As Table 2 indicates, schizophrenia, bipolar schisophreniforrn affective illness, and primary unipolar syndrome depression in adolescence were associated Organic syndrome* 12 .188 1,38W (OBS,ownnalthe dliorder, 100,000 with astonishiney high rates of suicide (the Sydenham's chorea) table does not include the "alcoholism only' Sociopathic personality 4 MEM. diagnosis, because there were only two such Other 4 cases, nor six others who had a variety of different diagnoses). It is of interest here that Table 3.

2-144 176 ALKovacs: °Major Psych loft Disorders" as Risk Factors...

Suicide Rates in a Sample of Private Practice Patients and in the ConvapondIng General San Diego County Population (¼dapted from Morrison, 1984) Rate/100,000/Year Patients County Pot& 0-19 Female 63.05 1.66 Male 173.01 3.99 20-24 Female 80.81 10.71 Male 205.66 23.17

Table11.

Suicide Rates In a Sample of Psychlatrically Hospitalized Adolescents Adapted from Weiner ci al., 1979)

Diagnosis at Follow-up Follow-up Suicide/fur (n;age z' or.eat) Length(1) No.* Rate/100,000

Schlzorhrenla (13; 14.5 yrs) a5 yrs Female (7) .00 Male (6) 235 3.921 Bipolar Affective (12; 15 yrs) 9.0 yrs Female (3) .00 Male (9) .333 3.700 Unipolar, Primary Depression (16; 15 yrs) 10.n yrs Female (10) .01 1,008 Male (6) .00 Antisocial Personality (9; ?) Female (4) .00 Male (6) .00 Uo Mental illness (8: ?) Female (6) .00 Male (2) .00 Undiagnosed Illness (8; ?) (probably adjustment reaction) ? Female (?) .00 Male (?) .00

I? dal not given; *mundord b moist dip

Table 2.

2145 177 Report of the Secretary's Task Force on Youth Suicide find the latter conditions than "organic years).Again, using our rough computa- syndromes." tions, and taking the maximal followup interval (14 years), we obtain a frequency of .143 Based on an average of 3 years of followup suicides/year in this group of 28, which then (up to June 1970) of 227 drug-using adoles- yields a suicide rate of 510/100.000lyear. cent girls who had been admitted to a remand home in London during 1966-1968, Noble et al. (9) assessed the outcome and correlates of Studies of youth suicides In the this condition. Using those data, Miles (10) general population computed the cohorts suicide rate as The retrospective method of identifying 3,500/100,000/year. However, a careful read- psychiatric risk factors for suicide suffers ing of the original publication reveals this from numerous constraints, including ascer- rate to be suspect because the suicides were tainment and interviewer 'Ass, incomplete cited for all admissions commencing in databases, and probable distortions of his- 1964which was 2 years before the study toric information because of the informants' sample's admission date--and included *need* to explain the suicidal act. Nonethe- deaths up to one year after the cut-off date less, the findings do deserve consideration, for the study samples followup (p .501). although for youth suicides, estimated rates Moreover, the number of all admitted cases of mental illness are extremely variable, rang- was not provided. ing from 92 percent (14) to 12 percent (15). All We will closc this section by noting three thinas considered, psychiatric disturbance other studies which underscore that the seemed more likely to be detected if study in- likelihood of suicide being detected varies vestigators used more extensive or systematic with the length of followup and the nature of information sources. sample, including the type of diagnosis. For According to Rich and associates (14), example, Annesley (11) who described the 2 'psychiatric illness is a necessary (but insuffi- to 5 year followup status of 362 previously cient) condition for suicide." Using the San hospitalized adolescents (aged 7 to 18), Diego Coroner's list for the period of reported no suicides among them. This was November 1981 to June 1983, they located 133 particularly notable because the admission consecutive cases of suicide under 30 years of diaFrioses were wide-ranging, covering age. These cases were compared with 150 various psychotic disorders, sociopathy (i.e., consecutive cases aged 30 ard older, sampled i'behavior disorders", neuroses, and obses- from November 1981 to September 1982. sional states. Likewise, Strober and Carlson Multiple information sources were used (12), who conducted a 3 to 4 year followup of (92% of the sample) to determine the most 60, hospitalized, depressed adolescents valid responses to an extennive, structured in- (diagnosed according to RDC), aged 13 to 16 terview; tten, consensual, DSM-Ill diag- years, found no suicide among them. noses were derived. Subsequently examined On the other hand, in a highly select sample, toxicology reports might be appended to the followed for a period ranging from 1 to 14 diagnoses. For our purposes, the interpreta- years, 2 out of 28 patients were found to have tion of the findings were constrained because died by suicide (13). The patients, who bad a the data were grouped by cases aged "under diagnosis of manic-depressive illness, were 30" versus "over 30" and the exact age range specifically examined because their first at- was not reported. Although the second tack occurred before age 19 at the mean age paper on this cohort (16), provided an age dis- of 15 (range: 13-18). The patients entered the tribution by 5 and revealed that 44 percent (p, study as a result of having been hospitalized = 59) of the 133 suicides were in the 25 to 29 at a particular institution; the followup took age range, diagnosis data by age were not in- place after the index the admission (mean cluded. This is problematic because, for ex- age at followup: 41 years, range: 23 to 69 ample, a particular diagnosis such as cocaine

2-146 1 7 ALKovacs: Waior Psychiatic Disorders* as Risk Factors... abuse may be more prevalent among persons Mother methodologically complete study 25 to 30 than under 25.Without more reported a very high rate of "psychiatric specific age by diagnosis data, the exact symptoms' and a notable incidence of prevalence of a diagnosis among early psychiatric contacts among children who adolescent suicides cannot be determined. killed themselves; unfortunately, no diagnoses were provided (17).In this total Nonetheless, as Table 4 indicates, 92 percent population survey of suicides in England and of those "under 30' had psychiatric diagnoses Wales during 1%2-1968, there were 30 une- (comparable to 91% of the older cases), and quivocal youth suicides (aged 12 to 14). Using substance abuse was very prevalent (much coroner, school, psychiatric, and social ser- more so than among the older suicides). Ac- vice records, Shaffer (17) found that 30 per- cording to a subsequent report (16), al- cent of the victims (11 = 9) bad been in together 53 percent of the younger cases had contact with a psychiatrist; 20 percent more a diagnosis of substance abuse. Affective dis- (11 = 6), although not referred to a orders were relatively frequent (a = 47), but psychiatrist, were recognized as having *con- less than in the contrast group (ft= 78), and duct or emotional problems"; and all but 13 52 percent of the 'under 30" cases bad a past percent had *psychiatric symptoms" of the history of psychiatric care.It should be "antisocial," 'emotional, affective," or pointed out, however, that the high overall *mixed" (emotional plus conduct) kind. rate of disturbance in this study's cohort included "minor* DSM-111 conditions, such as The foregoing findings were echoed by a "adjustment disorders" or dysthymia (which recent study that combined the :.3tal popula- has been viewed as a form of "minor" depres- tion survey and case control methods (18). sion). Moreover, the use of multiple diag- The Jefferson County (KY) coroner's office noses per case accounts foi the large, was used to locate all cases of youth suicides absolute frequencies in Table 4. (aged 5_ 19 years) from January 1980 througb June 1983; 24 cases (aged 12 to 19) were found, Psychiatric Characteristics of 133 Younger and their families were then contacted. The Suicides (aged <30)in San Diego County families of 20 (83%) agreed to cooperate. (Adapted from Rah et al., 1988). Extensive questionnaires, checklists, and in- Poet-Mortern N cues with Diagnoses* ventories were used with families, friends, Diagnoms-05M411 and relatives to determine the victim's Substance Use Disorder psychologic profile- Similar data wert col- Drug Use sa lected on matched-pair controls (friends of Noohol Use 72 the deceased). The 20 suicides differed from Affective Disorders 47 the 17 controls on the following variables of Atypical Depression 32 interest to us: frequent use of nonprescribed Ma/or Daxession drugs or alcohol (70% vs. 29%, p <.02); Sch4opttninkt Disorder 8 having exhibited 'antisocial behavior," in- Other Psychoses 18 cluding shoplifting, physical fights, or dis- Antisocial Personality 12 ciplinary probbms in school (70% vs. 24%, p My Diagnosis (92%) 122 <.003); and having had previous psychiatric treatment (45% vs. 24%, p <.04). However, Psychiatric Care (%) despite the high rate (65%) among the Past 52 suicide cases of inhibited/withdrawr,bverly At time of suicide 23 sensitive 'personality," they did not differ from the case-controls (25%) in this regard. *Table does not Include all the diagnoses that were Although Bourque and associates (19) focused on long-term trends in suicides Tabli 4 among females only, the data are useful to us

1 7 ;) 2-147 Report of the Secretary's Task Fame on Youth Suicide because they are analyzed by age.In this (n = 3) in the cohort, the manner in which methodologically careful study, the authors these diagnoses were ascertained and the examined the coroner's records in Sacramen- exact meaning of the terms are ambiguous. to County (CA) aver a 55-year period; both The final total population survey we sum- case ascertainment and data retrieval ac- marize was conducted in Upper Austria, for curacy were verified. In the final tabulation, the years 1977 to 1979 (21). The families of all data on Asian and Black women were ex- 47 'adolescents" (age criteria not given) who cluded because of the small sizes of these committed suicide in the 3-year period were subsamples. The authors appeared confi- contacted; 29 families were finally inter- dent in their findings particularly because, viewed within two to four years of the suicide. since the 1950s, the county coroner's staff has Based on a multi-item questionnaire, the been trained to systematically gather data on deceased youths (aged 9 to 20) bad the fol- several psychosocial variables. The variables lowing attributes: l4\percent (a = 4) bad a of interest to this paper included: 'despon- history of psychiatric treatment (including dency,* *emotional problems,"alcoholism,* two psychoses*); *2/3 had shown behavior dis- or *drinking problem,* and prior psychiatric turbances;* 7 percent (a = 2) were *begin- treatments. The data for the most recent 10- ning stage" drinkers; and 10 percent (a = 3) year period (19704979) are the most per- were 'unmistakably drug addicts.* tinent. During that time, 40 females, aged 15 to 24, were identified as suicides, of whom 85 Unfortunately, an extensive exposition of percent had shown evidence of desponden- another cohort (aged 17 or younger) is cy; 45 percent had emotional problems or uninterpretable for our purposes (22). For *unbalance;* 5 percent had recent arrests; 25 reasons best known only to the author, the percent had prior psychiatric treatment; psychosocial data on all official cases of none had evidenced pnablems with alcohol. suicidal acts byyouths in Israel (for 1963-1966) were combined for the fatal-outcome and In contrast, Sathyavathi (15) found a low rate nonfatal-outcome cases (n = 21 and n = 343, of *mental illness* among children who killed respectively). We must also take note of a themselves in Bangalore, India, during the widely cited, mostly narrative-report by Jan- years of 19674973. In this total population Tausch (23) on suicide among school survey of suicides reported to the police (a = children in the New Jersey public school sys- 1834), the author found 45 cases aged 14 years tem during 1960-1963.Although it was old or younger and 351 cases aged 15 to 19; only probably the first US. survey of its kind, its the findings on the younger aged group were description of 41 child suicides (awd 7 to 19), summarized. The records were scrutinized did not include information on psychiatric or for the presumed *causes* or *motives* of the mental illness.Additionally, the study did fatal act.In 10 cases, the cause was *un- use a preselected sample (rather than the known.*Categorization of the remaining general population), although the specific records showed that in 12 percent, *mental ill- method by which cases were ascertained was ness was reported as the cause of suicide.* not well descnbed. Unfortunately, the author had no details as to the nature of the mental disturbances. Table 5 summarizes some of the salient data from the studies we just reviewed. Fmdings Another suggestive set of findings was on psychiatric variables (diagnoses, condi- reported by Marek et al. (20), based on their tions, symptoms) are extremely difficult to suivey of all documented suicides in Cracow, synthesize because of the range of linguistic Poland, for the period 19604974. There were labels used, the lack of definitions, differen- 76 cam of youth suicides (aged 8 to 18). Al- ces in data retrieval methods, and significant though note is made of "mental disease such changes in "diagnostic habits* across the as schizophrenia, epilepsy, and charac- years. teropathy* (a = 3), as well as drug addiction

2-148 1 0 ALKovacs: Psychlatic Disorders° as RiskFactors...

Suicide Attempts and Psychiatric made previous attempts or threats (29). At- Diagnoses tempts also predict future attempts; thus, about one quarter of adults who attempt Given the scant diapostic information on suicide bad at least one previous attempt completed youth suicides, we decided to examine also the evidence available on the (34 relationship between psychiatric diagnoses Tbe data for youth are similar. As in adults, and attempted suicide in the same age group. sex ratios are different for completed and for This strategy has been criticized because attempted suicides.For every completed many attempters never complete suicide suicide in children and adolescents, there are (24), and because systematic differences in at least 3 to 4 attempts in boys and 25 to 30 demographics and diagnosis have been found in girls (31). Among adolescent attempters, between attempters and completers in both the male/female ratio is approximately 1:4, youth and adults (24 to 27). Nemtheless, it but in preadolescence it is 11 (31-33). In ap- is also true that there is overlap between the proximately 9/10 cases, drug ingestion was the two groups. For example, from a prospective method of choice (31,32,34). In two large viewpoint, suicide attempts among adults studies of youth attempters 16 percent and 37 also represent a risk factor for completed percent of their subjects had made at least suicide (28). It is known that up to 75 per- one previous attempt, and 6 percent and 17 cent of adults who committed suicide had percent of the samples, respectively, had

Prevalence of Psychiatric Variables Reported In Total or Partial-Population Surveys of Youth Suicides

ty.tud "4y Reference List Number) 14 Variable 18 17 19 18 15 20 21 Location U.S. U.K. U.S. U.S. Indla Poland Austria Years: 19 81-83 62-68 70-79 80-83 67-73 60-74 77-79 N described 133 30 ao 20 45 78 29 ACP ranfie ti< 12-14 15-24 12-19 "< 14" 8-18 9-20 Psych. data source: records-coroner X reconds-other herviews Psych Rx (96) Past 52 26 46 14 Current 23 30 Mental/...notional problems/iliness (96) 92 + 60 + 46 + 12 4 Behavior/conduct disturbances (96) 70 67 Psych symptoms (16) 87 85 65 Drinking problem (96) 53 + + 0 IDrug problem (96) 4 10

Tablo 5.

1 s 2-149 ,

Report of the Secretary'sTask Force on Youth Suicide

made two or more attempts (31,32). More focusing on very youngattempters cannot be important for this review, in Otto'sstudy (31), found in the literature. child and adolescentattempters who ultimately lolled themselves inthe 10 to 15year followup periodwere more likely to commit Psychiatric characteristics ofyouth suicide in the firsttwoyears after the attempt, who attempt suicide but the risk continued allthrough the fol- Bergstrang and Ow (39) collectedcharts on lowup, and repeatedattempts constituted a In7 patients under theage of 21 years who major predictor of suicide inthe boys in this had presented to all hospitals inSweden be- sample. The suiciderate on followup in male tween 1955-1959 for attempted suicide. attempters was 11.3 percent,a 7-fold increase Evidence of parental mentalillness (28%), over the conttol group. In the femaleat- parental alcoholism (15%), andfather's ab- tempters, the suicide ratewas 3.9 percent, a sence (44%) were characteristics of the 2-fold increaseover the control group. Fol- group. Otto (40) reexamined a probablyrep- lowup showed that about 19percent of the resentative subsample of 484cases that male repeated attempt=at intake, versus yielded enough informationto make a only 8 percent of those withoutprior at- psychiatric diagnosis.Neurotic-depressive tempts, subsequently completedsuicide. reaction was diagnosed in 30percent of The predictivepower of early attempts for cases, neurosis in 23 percent, schizophrenia future suicidal acts has also beenreported by in 12 percent (19% in boysand 9% in girls), Stanlgy and Barter (35) andby Cohen- manic depression in 5percent, and etindler et al. (36) in youthand by Tefft et al. psychopathy in 13 percent (18% ofboys and ,J7) in adults, among others. 0% of girls). Only primary diagnoseswere considered.In an attempt to delineate Them is some evidence that suickialattempts a in childhood and early presuicidal syndrome in children andadoles- adolescence may be cents, Otto (41) could only ascertain that particularly associated with latersuicide. Tr Otto's study (31), besides the during the 3 months before theattempt, the history of symptoms of some mental disordersname- repeated attempts for boys, otherfactors associated with later completed ly, anxious and depressivene...rosis, suicide among tchizophrenia, and manic these attempterswere male sex, length of depressionhad prior functional impairment (which worsened considerably in about 50percent of was, in patients with these diagnoses. turn, associated with attempts underage 13 Personality years, and with major psychiatric disorders characteristics showed much lesschange during the preattempt period.in a 10 to 15 like manic depression andschizophrenia, in both boys and girls), and the year followup of the original sample (31), the use t3f active investigators selected (violent) methods.Girls who used active a control group methods in their attempts accounted matched for age, sex, and geographicalbirth for only site. Attrition rate of the 7 percent of all girlattempters, but appeared attempter sample unique in that they was only 10.4 percent. The attempterson fol- were also over- lowup had committed represented among those whose first more legal offenses. suicidal Mental conditions in military health attempt occurred before theage of 13; they clas- were more likely to have left sifications were 10-fold higher (53%)in the a note and to attempters. have attempted suicide by hanging Unfortunately, no systematic than by analyel of t'lle predictive overdose. This divergence between power of psychiatric intent diagnosis for suicide outcome and lethality has also beenfound among was provide& suicidal prepubertal childrenwith major In a cross-sectional study basedon chart depression (38). It is apparent thatpre- or reviews from a pediatricemergency room peripubertal attemptsmay carry the highest (ER), Garfinkel et el (32) identified50 risk for future suicide completion.Unfor- children and adolescents who hadmade 605 tunately, good prospective diagnosticstudies suicide attempts, and comparedthem with

2-150 M.Kovacs: °Major Psychlatic Disorders' as Risk Factors...

505 nonsuicidal patients who had come to the lethality is very unpredictable. In fact, the same ER at the same time. Prior substance class of agent ingested was the most impor- abuse, past history of psychotherapy, and tant variable to predict the medical serious- current (not specified) psychiatric diagnoses ness of the attempt in this impulsive group. (most frequent symptoms were dysphoric af- Strober (42) compared 250 suicidal inpatient fect (55%) and aggressiveness and hostility youths aged 10 to 18 years with an equal num- (41%)) were signiftcantly more common in the attempters. The families of the at- ber of nonsuicidal inpatient controls matched for age, socioeconomic status, and tempters had significantly higher rates of his- sex. From a diagnostic point of view, emo- tory of mental illness (especially drug and tional, neurotic and conduct disorders alcohol abuse), suicide, paternal unemploy- (depression was "hidden* within thate two ment, group or foster placement, and ab- categories, according to the author) were sence of father.Only 36 percent of the attempts were judged to have carried a more prevalent in the suicidal group, al- moderate or high danger (criteria not though statistical significance levels are not reported Alcoholism was only present in the provided). These were associated with tang- suicidal group (5%). But Carlson and ly history of suicide and current psychiatric Cantwell (43), in a sample of 102 clinically symptoms, although this analysis included 100 patients twice and it is therehre flawed. referred children and adolescents, found that suicide attempts occurred with and without After a followup for up to 9 years, 8 of the the syndrome of depression. Attempters did 505 attempters had died by hanging, over- not differ from nonattempters in any par- dose, or motor vehicle accident, while the 5 ticular diagnostic category. deaths in the control group were due to medical disease. Several studies have included only adolescents. In such studies, *neurotic* depression, Brent et al. (34) also used chart reviews to conduct or personality disorders, and al- study the characteristic.. of 131 consecutive coholism, are the most common psychiatric suicide attempts by 126 children and adoles- diagnoses (44,45). As new assessment cents presenting in a pediatric emergency methods become more generally used, room between 1978-1983. The median age several reports have emphasized the wes 14 years. A bimodal distaution was relationship between suicidal attempts and noted, with approximately one-third of at- major depression in adolescents (34,46-48). tempts noted to be medically serious, which All these papers have also emphasized the in- usually involved a high degree of planning crease in alcohol and drug abuse as an as- and use of a psychotropic agent (rather than sociated diagnosis to major depression, and an over-the-counter agent) (34). The follow- also as a precipitant to the attempt. A similar ing variables were associated with a medical- point has been made by McKemy et AL (49) ly serious attempt: male sex, family history of and by O'Brien (50). One study found a 5- affective disorder, high suicidal intent, and a fold excess of patients with epilepsy among diagnosis of affective disorder, either in isola- attempterrs of whom 69 percent were under tion or associated with substance abuse. The 30 years old. Phenobarbital was the most fre- following distribution of diagnoses were quent overdose (51). Brent (52) replicated found:affective disorder (39%), conduct this finding recently in adolescents, indicat- disorder (26%), substance abuse (23%), and ing further that 8/9 epileptic attemptem were adjustment disorder (26%). Many of the receiving phenobarbital as their main an- Multi- patients had multiple diagnoses. ticonvulsant.Studies of adolescent delin- variate analysis of these data suggests the es- quents show similar agreement. Depressive a istence of two groups of attempters: symptomatology and suicidal behavior were dysphoric, effectively disturbed, hopeless found to be associated among 48 delinquent group whose lethality is very much a function girls aged 13 to 11 years. All 23 girls with his- of intent, and an impulsive group whose

1 3 2-151 Report of the secretwys Task Fome ort Youth Suicide tory of suicidal attempts were rated as that for adults, whereas hopelessness has moderately or severely deputed. In turn, been shown to be the key link between depression and suicidality way mainly as- depression and suicidal behavior (53) and sociated with antisocial and neurotic per- also predicts later suicide in samples of sonality patterns. Some of the girls showed depressive patients (64). no suicidality in spite of marked depressive Therefore, it appears that affective disor- syinptomatolog (53).Alessi et al. (54) ders, alcoholism, and other drug abuse play reported similar findings with a sample of in- the same central role in regard to suicidal at- carcerated adolescent delinquents of both tempts among youth, as they do in completed sexes. Those with major affective disotders suicide in youth and in the rest of the age or borderline personalitydisorders spectrum. Schizophrenia appears more represented the highest suicidally and com- lated to suicide than to attempts, whereas mitted the most serious attempts. But there personality diagnoses may relate more to at- also is substantial evidence that various other tempts. The primacy of unipolar and bipolar characteristics (i.e., aternalizing symptoms, affective disorders and alcoholism, as well as impulsivity and borderline personality) may schizophrenia and schizoaffective disorder, also be related to suicidality in the absence of as the diagnosis most oftenassociated with an affective syndrome (34,55-57,17). adult suicide has been known since the 1960's Studies of only preadolescents with suicidal (65.70). But there appears to be no doubt behavior had to address suicidal threats and that there is also a prominent component of ideation, especially those with a definite plan, irritability, aggression, impulsivity, border- because actual attempts in this age group are line features, 1-nd conduct disorder among rare.Evidence suggests that suicidality seriously suicidal young people, which is before 13 years may constitute a very high risk similar to findings in adults (71-73). It should for later suicide (31). Myers et al. (58) com- be noted that many of these symptom are by pared chart reviews of 61 suicidal and 287 no means inconsistent with thediagnosis of nonsuicidal preadolescent inpatient an affectke disorder (74) andespecially children. Depressive disorders were present bipolarity (75). Further refinement of diag- in ?0 percent of suicidal children and only 3 nostic instruments and improvements in percent of the controls. This was the only clinical skills to diagnose mixed bipolar Wness diagnosis for which the rate significantly dif- in adolescents, which probably is the com- fered between the two groups. Other vari- monest presentation of bipolarity inadoles- ables that differentiated the groups were cents, are likely to bring about further family history of suicidal behavior and abuse advances in the clinical characterization of of the mother (usually by the father). In a this aggressive-impulsive subgroup. Never- series of uncontrolled studies, Pfeffer et al. theless, it is also likely that a subgroup of im- (59,60,61) have repeatedly found that pulsive suicidal adolescents will remain prepubertal children with suicidal ideation, outside the major classical diagnostic threats or acts, frequently present symptoms categories. Their nosolog and biolog will of depression, hopelessness and law self-es- require further study. teem, psychomotor activity, conduct disorder, and that the parents present suicidal behaviors and/or depression. Ryan et al. (38) found that amover prepubertal mrior depres- Familial patterns of suicidality and sive children, 46 per ;cot present with persist- psythiatrio dieorders ent and severe suicidal ideation involving at it is clear that suicidality cuts across several least a conerete suicidal plan. Kadin et al. psychiatric diagnoses, the most frequent (62) found that hopelessness related more being bipolarity, Aonbipolar MDD, al- closely than depression to suicidal intent in coholismandsubstanceabuse, prepubertal inpatients, a finding similar to

2-152 iii.itivacs: °Major Psychiatic Disorders° as Risk Factors... schizophrenia, and perhaps some personality with a definite family history of suicide in a disorders includin& borderline and antiso- fust or second degree relative, and 5602 in- cial. In addition, several of the stuthes pre- patients without such family history found viously cited have noted the frequency of that &most half of the index group had at- suicidal behaviois in the relatives of suicide tempted suicide, and that more than half had attempters and victfins.Mother way to an affective disorder. A family histcuy of study the relationships between specific suicide increased the risk for suicidal at- psychiatric diagnoses and suicidality is to tempts across the following diagnostic review their patterns of association in categories:schizophrenia, unipolar and families, to determine if the familial transmis- bipolar affective illness, dysthymia, and per- sion of suicidality overlaps with, or is inde- sonality disorders (77). Similar findings were pendent from, the psychiatric disorders just reported by Twang (78) comparing suicide mentioned. Ibis strategy is possible beense rates among adult schizophrenics, manics, most of the diagnoses involved, as well as depressives, and surgical controls. Suicide suicidality, have been shown to aggregate in rates were high for all psychiatric patient certain families. Although most of the data groups, and the relatives of patients who had come from adult studies of affective disor- committed suicide had higher risk of suicide ders, the findings are relevant to this review than the remaining relatives. because there is increasing evidence of diag- Egeland et al. (79) reported that, among the nostic continuity between youth and adults in Amish, 78 percent of all suicides in the last most of the diagnoses involved, as well as in centuly were accounted for by four extended suicide, and because young adult, probands pedigrees with high density for major and relatives, were also included in many of unipolar and bipolar affective disorders, the studies. which together account for only16 percent of Most of the evidence indicates that although the Amish population. Nevertheless, other some diagnoses (i.e., major affective disor- Amish pedigrees with heavy loading for af- ders) and suicide may be associated within fective illness present no suicidal acts among families, familial transmission for suicidality their members. Similarly, Linkowski et al. and for psychiatric diagnoses are separate (80) in a study of 713 major depressive in- and probably independent. Overall, the data patients found that a family history of violent suggest that, given a positive familylistory of suicidal behavior was associated with the suicide, a superimposed major psychiatric same in the proband, and the effect was more disorder constitutes a serious risk factor for marked in bipolar women. In a sample of 50 suicide. bipolar patients, those with a family history of suicidality also had suicidal attempts, and In a study of adult primary major depressive associated alcoholism in the parents or in the probands, only age of onset of affective illness and the secondary diagnoses of al- probands increased the risk (81).From coholism or an anxiety disorder in the another viewpoint, the findings of Murphy proband were independently related to and Wetzel (82) point to the possibility that in the presence of a similar family history ot higher familial aggregation for major depres- suicidal behavior, the presence of affective sion in relatives. On the other band, disorder or alcoholism, or both, in the suicidality or the presence of any of the fol- proband will substantially increase the lowing clinical characteristics during major person's risk of completing suicide compared depressive episodes in the proband showed to cases with other or no psychiatric diag- no independent relationship to familial ag- noses. gregation for affective illness:any of the Research Diagnostic Criteria depressive subtypes, recurrent depression, or hospitaliza- CONCLUSIONS tion (76). A chart review of 243 inpatients The available figures suggest that the condi-

2-153 1 Report of the Secretwy's Task Force on Youth Suicide tion of psychiatric *patienthood* in the nostic value of these conditions among adolescent and young adult years is as- adults. This body of information also sug- sociated with an alarmingly high mortality gests that "patienthood* is associated with risk via suicide.This conclusion is ines- elevated suicide rates; that the risk is probab- capable in light of the general population ly higher for previous inpatients than out- base rates for death-by-suicide for these age patients; and that, although the mortality groups (83), and in comparison with the figures even within diagnostic groups vary, suicide mortalities associated with most affective inners, schizophrenia, alcoholism, psychiatric disorders among adt.hs.Using and disorders involving drug addiction yew global and "rough* estimates, psychiatri- probably early the highest relative risk in this cally disturbed youths may be running a risk regard. of suicide about 200-fold (or more) the rate For example, in this recent 5-year followup of their general population counterparts. report of 4,800 consecutive, first-admission And psychiatric illness in the young also VA psychiatric inpatients, Porkorny (84) seems to pose a suicide risk that may be up to found a suicide rate of 279/100,000 five times higher than the rates reported for patients/year.In a similarly recent study, adult patients (see below). Morrison (85) estimated the suicide mor- Although changes in diagnostic practices in tality among outpatients, using a large child psychiatty and in the use, definition, and private practice sample (n rs 12,500); the meaning of psychiatric labels make it exceed- overall rate was found to be 120/100,000/year. ingly difficult to compare the risk value of The rates for particular psychiatric condi- various conditions, tentative conclusions can tions are difficult to interpret because of be made. The *classic" mood and thinking their variability. The variability is especially disorders, namely, affective illness and evident with data that were presented in per- schizophrenia, are prognosticators of suicide centages or frequencies without adjusting for mortality. Because much of the supporting the length of the followup. Miles (10), who, evidence deAves from studies of youths with in his comprehensive review of this body of a history of inpatient hospitalization, the im- literature estimated the overall risk for plication is that the major factor could be the severalconditions,arrivedat presence of psychosis. Among the affective 230/100,000/year suicides for deprasives; disorders, this would further imply that 270/100,000/year for alcoholics; and up to a 3- bipolarity in this age groups is likely to be a fold higher suicide rate for adult opiate ad- major factor (12). The addictive disorders dicts. However, the variable data ti lies with may also carry a high risk value, although the which Miles (10) had to contend can be readi- study of these conditions has been con- ly illustrated in reference to schizophrenia. strained by the fact that their incidence and Recently reported suicide figures for this prevalence may be far more subject to socie- diagnosis yield rates ranging from a low of tal factors than seems to be the case for af- 2031100,000/year (86) to a high of fective disorders and schizophrenia. Fmally, 456/100,000/year (87). data are scant on the prognostic value of *neurosis" or personality disorders for suicide Given the available evidence, therefore, it is in the preadult years. In no small measute, difficult to say whether affective illness in the relative absence of such evidence adults is associated with a lower, higher, or probably reflects long-standing debates equivalent suicide risk than schizophrenia, about the diagnostic validity of these condi- for instance. However, the two riOst recent tions in juveniles. studies of unselected adult patient samples give every indication that different condi- Our conclusions about the risk value of tions make differential contributions to the psychiatric illness for suicide in the younger cohort's overall suicide rates. In Pokorny's years is underscored by the data on the prog- (87) sample, for example, diagnosis-specific

2-154 10 M.Kovacs: 'Major Psychiatic Disorders° as Risk Factors... suicide rates varied from a high of subjects, such "chance* survival may repre- 6951100,000/year for affective disorder to a sent lower intent. Although cluster analyses low of 71/100,000/year for organic brain of data on adult suicidal attempters reveals a syndrome, with other conditions occupying group who used more violent methods, with the middle ranges (e.g., 187/100,000 for al- more intent to succeed (71), nevertheless, robotism).Likewise, in Morrison's adult adolescents and adults with the highest intent outpatient series (85), the suicide rates are stB1 much more likely to kill themselves, ranged from 42/100,00Wyear for "unipolar' af- thereby removing themselves from prospec- fective disorder, through 318,1100,000/year for tive scrutiny (64,88,89). However, in bipolar illness, to the peak of 411/100,00Wyear prepubertal children, there is a discrepancy for schizophrenia.Suicide rates for *per- between intent and lethality which makes the sonality disorders" as well as for the situation quite different. While suicidal idea- 'neurosis" have been also estimated. The tion with a plan occurs among prepubertal overall trends suggest that these diagnoses children with MD13 almost as frequently as are also associated with elevated risk rates, in their post-pubertal counterparts (46% vs. although by no means as high as those cited 49%) (38), the suicide rate among prepuber- above (e.g.,10,85). tal children remains quite low (3,17), in spite of its secular increase among adolescents and In light of the available information on the young adults. Prepubertal children's lack of various disorders in adults, it is more than cognitive maturity and skills necessary to likely that the major psychiatric ewes also complete suicide probably accounts for the contribute differentially to the overall tate of low rates (3,17). Therefore, this age group suicide among children and adolescents. provides the opportunity to study suicidality Furthermore, the diagnostic fmdings in at- naturally untruncated by suicide completion. tempted suicide in youth point in the saute The findings reviewed earlier, showing that diagnostic directions as those for complaxi attempts before age 13 seem to cany a bleak suicide, except for a larger overall proportion prognosis for later suicide, also suggest that of cases with *personality disorders.' Given prepubertal suicidality may be much more the diagnostic problems associated with representative of completed suicides and these conditions, not the least of which is that deserve special study.Suicidality of very mixed bipolar illness can be egsily missed, early onset may enable the selection for fu- even when the patient can b. inteiviewed, it ture studies of those attempters who are most is not unthinkable that this could also occur at risk, as the closest approximation to com- with post-suicide diagnoses. Thus, identify- pleted suicides. ing a group of child and adolescent suicide attempters who resemble suicide completers This approach would be consistent with the may be very useful, not only for therapeutic findinp by one of us regarding differences in and preventive purposes, but also because GH responses and REM latency, and in their this population may most closely ap- patterns of association, among suicidal and proximate suicide completer&Moreover, among nonsuicidal prepubertal children with ft.= these attempters, we could learn the major depression, both during the episode most about the affective, cognitive, and and in the recovery drug-free state. Such biological process that characterize the findings may relate to the evidence for the as- young patients who actually do commit sociation of a functional CNS serotonin suicide. deficit in adult suicides (90-103), which may be familial (102). Furthermore, the evidence Some have proposed focusing on patients for familial aggregation of suicidality is also who have engaged in suicidal behavior of consistent with higher risk with miler age of high lethality and intent, on the assumption onset, especially if genetic factors are at least that only chance circumstances allowed their partially responsible for such familial trans- survival; however, in adolescent and adult mission (104406).

17 2-155 Report of the Secnitteuys Task Force on Youth Suicide

It is difficult to integrate the pertinent data youth suicides with known histories of on youths who have killed themselves with psychiatric cuefrom one-third to about the information on the pmgnostic value of one-half of the samplesare far higher than psychiatric alma in juveniles. In part, this is the mental health referral rata in the general a consequence of the methodologic youth population. But the figures on the problems posed by the use of coroner's suicides are comparable in this regard both records and "psychologic autopsies.* In this to the portion of psychiatrically identified ex post facto database, probably the most cases who have been referred for treatment 'hard core* indicator of psychiatric illness is and the portion of maladjusted youths who a positive history of mental health care. aztually received mental hailth care. However, even those figures can be only ap- Our conclusions must be viewed in light of proximations of the prevalence of mental M- the *methodologie problems that confound ness in the cohorts because treatment is a review of this body of literature. First, using mediated by a variety of factors including conventional referencing methods, it is not recognition that a disorder exists, the possible to access every study of mentally ill availability of mental health treatment youths that may have found suicide among resources, awareness of such resources, and the sample (e.g., 7), because of the way in social influences that determine referral pat- which the studies were apparently indexed. terns. With the above caveat in mind, we Second, in some of the available prospective used known rata of treatment referrals for studies of prychiatrically ill juveniles, the ab- pediatric cohorts to interpret the pertinent sence of data on suicide could mean either data on completed suicide among youth. that this outcome was not observed or that In their review of epidemiologic studies, the investigators did not look for it (e.g., Gould and associates (89) noted that "almost 11,108,109). Third, some of the published data all psychotic children and adolescents are are inconsistent, which is most evident in known to some treatment facilitiy." But only multiple publications regarding the same approximately 1 percent to 49 percent of cohort. For example, the followup time in- "maladjusted" children receive mental health tervals and the sample sizes riay be dis- care. If we eliminate figures that included crepant (6,74) or case frequencies with treatment by general medical personne certain diagnoses may not exactly correspond pediatricians), 30 percent is about the upper (14,16). Should it be desirable to conduct a limit of trrated cases among *maladjusted" meta-analysis on the reported findings, such youths. Turning to information about rates methodologic issues need to be taken into ac- of referral, Costello (107) found that, from nunt. about 2 percent to 5 percent of younpters In summary, there is significant evidence that seen by primary health care providers major psychiatric disorders constitute an im- (pediatricians, family practitioners) are portant risk factor for completed and at- ieferred to mental health specialists. She tempted suicides in children and adolet also estimated that 50 percent is tte modal The efficacious treatment and care of referral rate for youths who had been defini- psychiatrically ill youths may be the most tively identified by their health care providers feasible way to alter their risk of suicide. as having a psychiatric problem. The reported rates of psychiatric treatment or contact among youths who have com- BIBUOGRAPHY mitted suicide in the United States and 1. Rushforth NS, Ford AIL Sudek HS, st al: Increas- ing suicide rates In adolescents and young adults In an Britain (see Table 5) therefere suggest that urban community (12118-1282). Teets of hypotheses from thae you.hs were unlike normal pclliatric national data. Ai i3ulcide In the Young. Edited by Hudak populations, but similar to maladjusted or HS. Littleton, MA, John %Wight, 1004. pp. 411411. 2. Mods It The adolescent Weide problem. Suicide disturbed juveniles. That is, the portions of Ufe Threat May 1591-10110,10115.

2-156 1 S S M.Kovacs: °Major Psychiatic Disorders" as Risk Factors...

3. Shaffer 0, Fisher P: The epidemiology of suicide 27. Stallgel E: Suicide and attempted suicide, Mac- in children and ouera adolesoents..1 Am Poszi Child Gibbon 8 Kee. Bristol UK. 1965. Psychiatry 20:54, 1981. 28. Wan AA Nielsen 0, Blite-Brahe U, Hansen W. !Col- 4. McClure GMG: Recent Vends hi cuicide amongst mos t.: Attempted suicide in Denmark 8. Assessment of the young. Br J Psychiatry 144:134138, 1084. repeated auk:1(W behaviour.Acta Psychlatr Spend 5. Rubinstein DH:Epidemic suicide among 72:389404, 108& Micronesian adolescents. Sac So Med 17657465, 1983. 29. Schneldman EL Faberow NL Clues to suicide. & Morrison J: Suicid_ein peWhWic patients: Age Pub Health Rep 71:105, 1958. distribution. Suicide Life nuM Behav 14:52.58, 1984. 30. Schmidt Df, O'Neal P. Robins E: Eviduation of suicide attempts as a guide to therapy. J Am Med Assoc 7. Weiner A. Weiner Z. Fishman ftPsychiatric adolescent lopeffsnts. Eight4o4en.year fotiovoup. Arch 155:549, 1054- Gen Psychiatry 381196-700, 1979. 31. Otto U: Suicidal n end 16014110,11% a follow-up study. Acta Scuuffnavice. Sup- 8. King LI, Pittman GO: A elx-year follow-up study of plementurr: 233, 7-123, 65 adolescent patients. Natural history of affective disorders in adoloacence. keit GM Psychiatry 22:230.235. 32. Garfinkel BD, Posse A. Hood sk Suidde attempts 1970. in children and adolescent& Pin J Psychiatry, 139.141. 1972. 9. Noble P. Hart T. Nation ft Correlates sr d outoome of liNcit dnrg use by adolescent girls.Br J Psychiatry 33. Nichol it, Guichon D: Atlempted suicide armov 120:407404, 1972. children and adolescents in 1966. BC Med J14:139-141, 10. Miles CP: Conditions rwedleposing to suicide: A 1972. review. J New Mint Dia 184:231.248, 1977. 34. Brent I): Correlates of medical lethality of suicidal 11. Ann/Way PT: Psychiatric Wen in adolescence: attempts in children and adolescents. J Amer Acad Child Presentation and prognosis. J Men: So 107:269-278, Psychiat kt press. 1961. 35. ftirdw EJ, Batter JT: A Adolescent suicidal be- 12. Strober M, Carlson G: Spier illness in adoles- havior. Arn J Orttyddd 40:8745, 1970. cents with miler depression.Clinical, genetic, and 35. Cohen-Sandisr R, Berman AL, King RA: A follow- psychophennecologio predictors in a three- to four of hospitalized suicidal children. J Aced Child prospective folkve-up investigation. Arch Gen up:ysctud21:306408 1982. 39:549-555, 1982. 37. TON BM, Pederson MA, Bablglan Ha Patterns of 13. Olsen T:Follow-up study of maniodepressive death among suicide attempters, *psychiatric patients first attack occurred before the age of 19. and a general population. Arch GenPsycitiatn11:81r4.- Acts Psycida- Spend (Suppl 182) 37:4541, 1981. 1161, 1977. 14. Rich V., Young 0, Fowler RC: San Diego suicide 38. Ryan ND, Pulg-Mtich J, Rabinovich H, Robinson study: I. ng vs old oases. Arch Gen Psychlat 43:577- 0, Arnbroalni PJ, Nelson B, !Yong.: 5: Tffo clinical *lore 582, 1986. of or depression in children and adolescents. (Sub- ndftd). 15. Sathysvathl K:Suicide among children in Ban- galore Indian J Pedlar a 149457, 1 39. Bergatrand 00, Otto U:Suicidal attempts in adolescence and childhood. Acta Paediatric.% 51:1e.28, 18. Fowler RC, Pkh cL. Young D: Sin 049* sololdo N. Substance abuse in young oases. Arch Gen 1962. Psddat 43.962465, 1998. 40. Otto U:Suicidal attempts In adolescence and childhood. States of mental Illness end petsenslity vari- 17. Shaffer D: Suicide In childhood and early adoles- able& Ma Paedopsychiatrica, 31:397411, 1984. cence. .1 Mild Psychol Psychisby 15:275-291, 1974. et .Otto Chem* in the behaviour of children and 18. Shaft! M. Carrigan S,Whitinghill JR, et al: adolescents preceding suicidal attempts.Acta Psychological autopsy of completed Wade In children Psychistrice Scandinavia*, 40:388405, 1084. and adolescents. A,m J Psychiatry 142:1081-1084, 1985. 42. arbor Et:Social environment and suicidal 18 Bourque 12, Kraus JF, Cound BJ: Attributes of children and adolescents:A comparative study, in suicide in females. Suicide Life Threat Behav 13:123438. Dominion et Suicide, pp. 008415 Pergamon Rm. 1983. 1981. 20. Monk Z. Wdaold J, Zwarysiewicz W:Suicides 43. Cashion GA, Cantwell D. Suicidal behavior and committed by minors. Forensic Sal 7103-108, 1978. depression In children and adolescents. J Arner Acad. 21. Lebthuber F. Schony W, Fisher F, et a Study on 21:361-368, 1982. suicides committed by adolescents in Upper Austria 44. White NC: Self-poisoning in adolescents.firlt J covering a period of three years. Depression et Suicide Psychist 134-24-35, 1974. 652-655, 1981. 45. Hawton K. Osborn M, Grady J. Cole 0:Brit .1 22. Amir kt Suicide among minors in Israel. ler Ann Psychiat 40:124431, tam Psychiatry 11219-269, 1973. 46. Crumlay FE: Adolescent suicide attempts. J Mi 23. Jan.Tausch J: Suicide of ohildren 196043. New Med Assoc 241:2404.2407, 1979. Jersey public school students. Unpublished manuscript. undated. Department of Education, Trunton, NJ. 47. Crumley FE:Adolescent suicide attempts and mew:hale. Texas Med 7110245, 1962. 24. Clayton PJ:Epidemiologlo and risk factor* in suicide.In Psychiatry Update, American Psychiatric 45. Robbins RR Need, NE. Depressive symptoms Press, Washington, D.C., 1983, pp.405-428. and suicidal behavior kit adolescents. Amer J Psychlat 2& Shaffer D Dsprss&on , mania and suicidal acts in 142.585492. 10eff. child end adomt psychiatry: Modem Approaches, 49. MoKenty PC, MAW CL, Miley C: Tao nolo of Ed. by M Rutter and I. Hersov. London, Maokeell, 1965. drugs in adolescent suicide *tempt*. Suicide and Life- Threatening Behavior 11116-175, 1063. 25. Robins E. Bohm,* EH, O'Neal 12: Some interrelations ofsocial factors and clinical diagnosis in attempted 50. O'Brien JP: increase in suicide ottemc4o_b7 clni9 suicide: A study of 109 patients. Amer J neychist. 114- ingestion: The Boston experlenoe,19644974. P0011Gen 22i-231.1957. Psychlat 34:1185-1108 1977.

BEST COPY AVAILABLY 1S9 2-157 "

Report of the Secretary's Task Force on Youth Suicide

54. Newton K, Fago J, Mersecir P:Association be- Psychist 130460-455, 1973. tween epilApsy and attempted suicide.J Neurol 74. Pulg-Antich J: Major depression and conduct dis- Neurosurg PwroNt 43168.170, 1980: order korpuberty. J Amer Azad Child Psych's! 21:302- 52. Brant lk Ctrerrepresentation of epilepilos in a con- W, 1 secutive series of suicide atlenestses at a chlicken's how* 75. Akiskal HS, Downs J. Jordan P, Watson S, 1978.1983. J Aced Child Pephist 25242-2464 D, Pruitt DB: Affective disorders in referred 1088. %IsraYndyounger siblings of manic-depressives. keit 53. Gibbs JT:Depression and suicidal behavior Gen Pwrdi.* 42:996-1003, tan. anxicie delfictuent females. J Youth Adobe 10:150-167, 78. Weissman MM, Menkangas KR, Wokramaratne P. 1981. Kidd KK Understanding the clinical heterogeneity of 84. Neal NE McManus PA, Brickmen A, major depression using family data. Arch Gen L Suicidal behavior smarts seriousjuvenile= 43:438434. 1988 Am J Psychiat 141:286-217, 1984. 77. Roz. Familaiston of suicide.Arch Gen 55 OilesJkMuhsrMLcoxGB :Depression in an Pep:dist .071-974, 1 adolescent deNnquiont gyration kch r3en Psychlat 70. Twang MT: Rsk of suicide in the relatives of 37:1179-1184 igen ehirophrenice, manias, de,. aseNes, and controls. J Can Crumisy FE Adolescent suicide attempts and bor. Psychist 44:308-400, 1083, &dine personality disorder: clinical feature,. Southern 79. Egeland JA, Sussex JN: Suicide and family load- Med J 74-684444 1081. ing for affective disorders. J Am Med Assoc 254:015415. 57. Connell Pk Drug addiction:Adolescent dm; 105. taking. Poo Rey Soo Med 58489412, 1065. Linkowski P, diMaertelaer V. Mendlewica J: 68. Myers KM, Burke P, M-.*AW E Suicidal Who* 4" Suicidal behaviour in myor depreasive Muss. Acta by ImOd preadolescent ~tot on a Psychistr Scand 72.233 1985. unit J Amer Aced BIM Psychiel 24:474-480,ratiatda 81. Johnson OF, Hunt 0: Suicidal behavior In big: 59. Pfeffer CR, Conte FA Plutchik R, Arran 1 Slicidal manic-depressive patients and their famWes. behavior in latency-age children: An empkicel study. J prehensive Psychlat :0159-1134, 1979. kner Aced Child Payibliat 1E67902. 1979. 82. Murphy 0.. Wetzel Rit Family histoty of suicidal 60. Pfeffer CR, Conte HR, Plutchlic R. Jarrett I: Suicidal behiMor among suicide attempters.4 New Mont Dis behavior in latent:rev children: An outpatient popula- 170:8640, 1982. tion. J Amer Aced CNN Psychiat 19703-710, 1080. 83. U.S. Bureau of the Census: Statistical Abstract of 81.Pfeffer IA Yuckermen S. Plutchlk R Miouchl MS: the United States: 1979 (100th Edition). Washington, DC, Suicidal behavior in normal whoa! children: A corn- U.S. Government Printing Mk*, 1979. glid°with child psychiatric inpatients. J Amer Aced 84. Pokorny AD: Prediction of suicide in psychiatric 11Psychlat2t418-423. 1084. patierffs.Report of a prospective study.keit Gen 82. Kazdin AE, French NH. tinia AS, Esveldt-Dawson K. Psychiatry 40:249-257. 1983. Shed* RB:pi=snesegitelbreedesion, and euick41 in- 85. Morrison J:Suicide in a paythiatic practice tent inpatient children. J populations. J Ciln Psychiatry 1982. ConsualliraPsycho! 504410, 1963. 88. Black Winokur 0, Warrack 0:Suicide in 83. Beak AT, Kovacs M, WAillonan k s c h i z o i d yenta: The lowa record linkage study.11 COn and suicidal behavior: An overview. JAmii°641$1;alocas Psychiatry 44k14-17, 1985. 234:1148-114 1075. a Pokorny AD: A follow-up study of 818 suicidal 64. Beck AT, Steer RA, Kovacs M, Garrison B: atients. Am J Psychiat 1221 IOW 118, 1964 Maness and eventual suicide: A 10-year prospective %Idoiffatients Itkftpd with suicidal ideation. Am J O. Pierce DW A predictive validation of a suicide in- 14,Z5WM 1985. snt scale: A tholes: f011ovr-up. Brit J Psychlat 139:391- 398. 1981. 611, Dorpat TL, Rpley HS: A study of suicide in the Seattle wee. Comp, Pftelft 1900: 89. Gould MS, Wunsch4410g R. Dohrenwend BP: For- about Ma prevalence, treatment, 68 Robin E. P.twphv E. Wilkinson RH, Gardner S, of and significance of psychiatric. disorders in KayesJ Some ciInlci considerations In childrenthe United Slats*, In Minis! Rpm In the UMW suicide based on a study of 134 successful . Amer 3tates: EpidernW.Wgic Estimates. Edited by Dotuenwend J Pub Health 40:M=, 1959. OP. Gould MS, Link B. et al. New York: Praeger. 1980. pp. 87. Guze 08, Robins E Suicide and primary affective 944. disorders. Br J Payolft 117:437434 19'70 00. Aaron H: Symptom patterns in unipolar and 68. Bans:dough% Burch J, Nelson% Sainsbury 12: A bipolar depression correlating with monoamine metabo- hundred oases of 'Waldo: Clinkai aspects. Brit J Psychiat lites in tly cerebrospinal fluid: P. Sulkies. Psychlat Ras 128:358-373. 1974. 3125436- 1980. O. Twang MT:Suicide in schizophrenics, manic 91. Asberg M. Traskman t., Thom P: &HIM in the depressives, and surgical controls. Arch Gen Psychlat cerebrospinal fluid.A biochemical suicide predictor? 35:153-155. itrr& Arch Gen Psychist 33:1193-1197, 1978. 70: Twang MT, Dempsey GM, Fleming Jib Can ECT 92. Banki C, Molnar G. Fellete I:Correlation at in- ppraezturemature death and outside In lochizoaffective' dividual symptoms and other clinical variables with J Mact Olsord 1:187-171, 11979. cerebrospinal fluid amine msfoirgilles and tryptophan in depression. Arch Psychlatr New 229:346-353, 1981. 71. Paykel ES, Rassaby: Classification of suicide st- by duster analysis. BrttPsychist 13a4642. 03. Bank! C. Arato PA, Popp 2, Kura PA: Biochemical %mem1 markers Ir. suicidal patients.Investigations with cerebrospinal fluid amine metabolite* and neuroen- 72. Crook T. Rodin A, Davis 0: Factors associated destine tests. J Affect Dis it341.350, 1984. wipIntsatiemfited.psych:M*1AM tref.26.1182inad 94. (Zeeland I., Milberg A, Asberg PA, Traskman daixtend MarL., Thom Bertilseon Wring Platelet -3. Weissman M, Fox K. Merman G:Hostility and adctivityand monoamine metabolites in depression associated with suicide attempts. Am J cerebrospinal fluid in depressed and suicidal patients and

2-158 (I)

BEST COPY AVAILABLE ALKovacs: *Major Psychiatic Disorders° as Risk 7ac1ors...

In healthy controls. Poyohlatry Res 121-29, 1981. 95. Tradcman 6 Tybiva Asbert_ Bertlissontango 0. .fthia1g 0: Collect rhe 06F of depresesd and suioldsi Went). Arch Gen Psychist 37761-767, 1090. 96. van Prase NM: Cepression, suicide and the metabolism of aerator& In the brain. J Nitwit Dls 42715-290. 1962. 97. Brown G, Goodwin F, Ballenger .1, Gayer P. Major L.: Aggression In human correlates with fluid amino metsbollise. Plyoklat Res1:131.0rirmin.ai 913. Brown 0, Ebert M, Jimorson 0, Klein W, Bunn W, Goodwin F: ression, suicide, and :Rstationships to metabolites. Am J Psychlat 130741-746, 1999. a uniwila 111, Roy A Guthno S. Indices of sorotonin metabolism in vMmt offsnders, arsonists, and alcoholics. Presented al The Now York Academy of Sciences. Con- ferenceon Psychobiology°, Suicideillshavior. New York, NY, September19, 1995. 100. Traskman-Bendz L., Asberkt, Sohalling 0: Serotonerglo function and suicidal Miavior In disorders and neuroses.Promoted atTheMNsarVoik AwAwny of Sciences, Conference on ftdto of Suicidal BOuo4m, New York, NY, Srst 19, 1961 101. Liaem L, Tuck JR, Alborg PA, Soslia-Tomba earthman L.: iimmdde, sulolds and CSF 5-HIPA Acta Poyohlatr Soand 71230-236, 1285- 102. Sedan 0, Figatonsatring, Nybedl H. II FA. Wods-Holgodt healthy 1m- tween concentrations of monoamino metabolites in cerebrospinal fluid and tangly history of psychiatric morbidity. Brit J Psychist 133206474, 1990. 103. van Prase HM:Sionlflcance of biochomical parameters in go dnosls,lrestment anafrovention of dopresalves. Bid Payable( 12101-131, . 10e. Jusi-Nleisen N, Inclobech T:A twin study of suicide. Acta Genet Med Osmond 19207410. 1970. 106. TIMM MT: Genetic, teflon In suicide. Dis Nerv Sys 38:490-501, 1977. 106. Wonder PH, Kety SS, Rosenthal D. Schulaingor F, Ortmenn J, Lunde I: Psychlablo disorders In the biological and adopgve familiss of adopted Individuals with 81- fective disorders. Arch Gen Psyohlat 43:923429, 1986. 107. Castolio EJ:Primary owe pediatrics and WM psychopathology: A review of diagnostic, treatment and MAIM! practices. Pediatrics. In poss. 105. glvowitz Foreotson J, Goldstein 0. et al: A tot- low-up study of hospitalind adolescents. Comp Psychiatry15:35-42, 1974. 109. Warren W: A study of adolescent psychiatric in- and the outoomo six or more yoors lator. P. The ritilotitsupstudy. J Child Psychol Psychiatry 6:141-160, 1965.

BEST COPY AVAILABLE 191 2-159 PERSONALFTY AS A PREDICTOR OF YOUTHFUL SUICIDE

Allen P. mces, M.D., hofessor of Psychiatry, Conseil University Medical Center, New York Hospita4 New York New York

Susan Blumenthal, M.D., Chief, Behavionil Medicine Program, Health and Behavior Research Brunch, Division of Basic Sciences, National Institute of Mental Health, Rockville, Maryland

INTRODUCTION This paper will summarize the limited avail- tions. We will conclude with suggestions for able literature on the personality risk factors future research, current clinical practice, and associated with youth suicide and will outline prevention. The most interesting question the methodological difficulties inherent in that emerges from this review is the degree this line of investigation. Personality disor- to which the personality factors that predict der research has recently flourished greatly youth suicide are equivalent to factors that because the Diagnostic and Statistical also pertain to adult suicide. This question Manual III (DSM III) provided a separate has important theoretical, clinical, and axis for personality diagnosis and specified prevention implications. explicit criteria defining each of the personality disorders.This has led to the Personality Predictors of Adult development of reliable semistructured in- Suicide Behavior terview instruments to assess personality dis- The two DSM III personality disorders most orders in adults. Preliminary findings also suggest that personality disorders may in- clearly associated with adult suicides, both completed and attempted, are the borderline fluence, in important ways, the presentation, (BPD) and the antisocial (APD) (1). Suicide course, biological test results, and treatment rates for several-year followup studies of response of various Axis I conditions. Thus BPD patients are reported at 4 percent (2) far, however, there has been very little sys- and 7 percent (3); on a 15-year followup, the tematic nmearch on personality assessment rate was 7.5 percent (4). Several studies sug- in children and adolescents, and there are gest that the comorbidity of BPD with affec- many inherent conceptual and practical tive andfor substance abuse disorders results obstacles to any precise determination of the in particula- ly lethal combinations (5, 7). Al- personality risk factors for youth 3uicide. though most self-destructive behavior in We will briefly review personality variables BPDpatients is pkobably nonlethal in intent, associated with suicide in adults, summrxize a substantial portion ofBPDpatients do research on the personality variables as- eventually die by suicide, usually in young sociated with youth suicide, and outline a adulthood. number of the pertinent methodological Reported rates of suicide attempts in APD problems and some of their possible solu- individuals vary considerably (11%-4696),

2-160 A.Frances: Personality as a Predictor of Youthful Suicide perhaps because of differences in underlying (24,25,56,57,58). Studies using the Eysenck base rates of APD and of suicide attempts in Personality hwentory fairly consistently find the samples studied and the fact that most high neuroticism, psychoticism, and introver- studies did not use DSM III criteria (8-10). sion scores (59-62). The major limitation of It is thus difficult to generalize the findings available personality dimension studies is across studies. It is estimated that 5 percent that each has tended to assess in isolation of APD individuals eventually die by suicide only one or a small number of dimensions so (8-11). APD may also predict for frequent that we don't know the eevee and direction and recurrent attempts (12-14). These may of convariation among them and the amount occur in response to anger and frustration in of total variance they explain. Dimensional interpersonal relationships and in order to personality trait measures may also be dif- manipulate otheis (15,16). The comorbidity ficult to obtain in ordinary clinical situations. of APD with affective and/or substance Recently, a particularly fascinating connec- abuse disorders may, as with BPD, result in tion has emerged between the personality more frequent and more lethal attempts dimension of aggressive impulsivity in (6,17,18). suicidal and violent individuals and the The psychology literature has employed a biological finding of low central nervous sys- different strategy to determine the per- tem serotonin turnover (63-67). The as- sonality predictors of suicide. Rather than sociation holds up in patients with impulsive assessing the presence of a categorical per- personality disorder, even in the absence of sonality disorder in suicidal individuals or the an Axis I diagnosis of affective disorder rate of suicide in those with personality dis- (68,69). The serotonin dysfunction appears order, many psychology studies have to represent more a trait than a state condi- measured specific personality dimensions or tion (70)It has been postulated that a traits in suicide attempters and/or com- central problem in serotonergic metabolism pleters. Attempters and completers appear may contribute to the individual's impul- to be different in their personality and in sivity, aggressiveness, and suicidal potential, other characteristics. Attempters have the which then may be released in the presence more disturbed personality profiles and are of clinical depression. also more likely to be young, female, to lack an Axis I diagnosis, and to commit public, im- Personality Predictors of Youthful pulsive, suicidal acts using less serious means Suicide (19).Most of the personality dimensions This paper reviews the available research that have been studied apply only to suicide literature on personality traits and disorders attempters and may not generalize to corn- in adolescent suicide attempters and corn- pleters. pleters. Most of the literature on personality The following personality traits seem to be applies to suicide attempters and may not particularly characteristic of suicide at- generalize to completers. The four studies tempters: aggression or hostility (20-29), im- on completed suicides in youth are retrospec- p u lsivi ty )), social withdrawal or tive and do not utilize standardized per- interpersonal difficulties (31-37), low self-es- sonality measures. In studies of attempted teem (38-45), dependency (21,26,27,46), suicide, assessments frequently are brief and hopelenness (47-49), external locus of con- often are made in crisis settings. Patients in trol, rigid cognitive style, and poor problem these studies who are not admitted to the solving (54). The many studies that have hospital are difficult to follow and frequent- tested the ability of the MMPI to differen- ly do not want to discuss their suicide at- tiate suicidal patients have been inconsistent tempt. Assessment of personality occurs at (55), and the same is true for studies of the variable time intervals from the attempt and association of hysterical tria:is and suicide

I 93 2-161 Report of the Secretary's Task Force on Youth Suicide may be confounded by the presence of an sessed for conduct disorder in suicidal Axis I diagnosis and by stress. youngsters are supported by the additional studies to be reviewed soon that have found Voir few of the studies reviewed have used systematic and rigorous diagnoses of per- aggressive, impulsive, and irritable personality traits to be more generally associated sonality disorder since the focus of attention has usually been on Axis I disotders. A with suicide. variety of different control groups have been Porderline Personality Disorder (BPD): used in some studies but not in others. Be- Although the construct of borderline per- cause suicide is an event with relatively low sonesty disorder has not yet been carefully frequency, thm are few prospective studies investigated or validated in adolescent in the literature. Furthermore, there are ex- patients, there are several interesting tremely few studies of suicide attempters in a preliminary studies suggesting that it is often population with identified personality disor- present and can be reliably diagnosed in ders. This review focuses on studies using adolescent suicide attempters (78-81).It standardized personality measures and as- also appears that the comorbidity of DPD sessments with appropriate control groups. and other disorders in adolescents is par- The most frequently used method of assess- ticularly la:ely to predict for more frequent ment has been the questionnaire, but issues and more lethal suicide attempts. Friedman of reliability and validity are inf. Iquently ad- et al. (82) found that among 76 adolescent in- dressed, and the findings and measures have patients, those who met criteria for both not been replicated in other studies (11). BPD and for major affective disorder were Conduct DisorderStrictly speaking, the most suicidal.In this same sample, Clarkin et al. (78) found that adolescent childhood conduct disorders cannot be considered the met equivalent of personality suicidal patients were equivalent to their adult counterparts in the prevalence of per- disorder since the majority of cluldren who sonality disorders (defined by a duration qualify for this diagnosis do not go on to display a pattern of adult categotical disorder criterion of one year). Crumley found that BPD was the most common personality dis- (71). Norethelem, we discuss conduct disor- order in a group of hospitalized adolescent:. ders here because they tend to be relatively and that this diagnosis usually coodsted with stable and are, by far, the sturdiest predictors major depression (67%) andior substance of adult antisocial personality. The fact that conduct disorder is a major risk factor for abuse (77%). Alessi et -!. (81) found a high both yoLth suicide and for adult antisocial prevalence (35%) of BPD in a sample of in- carcerated juvenile offenders who would personality disorder (which itself predicts for presumably also meet criteria for conduct adult suicide) suggests that this diagnosis disorder. The BPD diagnosis strongly deserves special treatment and preventive at- predicted for greater frequency, seriousness, tention. and lethality of suicide attempts. The as- Conduct Isorder appears to be strongly association between BPD and suicide held up sociated with both su:cide (72-74) and with in both the Friedman (82) and the Alessi (81) suicide attempts (75-77). Conduct disorder studies even when the suicide item was itself is much more common among male suicide eliminated as a criterion for making the BPD victims, and the precipitating event for the diagnosis, thus removing the risk that the as- episode is often a disciplinary crisis. There is sociation was merely a tautological artifact a frequent comorbidity of conduct disorder resulting from the fact that the DSM III with affective, substance abuse, and border- definition of BPD includes one criterion line personality disorders, and the frequency devoted to suicidal behavior. Pfeffer (76) and lethality of attempts increases with the and colleagues found that BPD was the most degree of comorbidity. The few studies frequent diatoosis among 48 preadolescent reviewed here that have systematically as-

2-162 Lot A.Frances: Personality as a Predictor of Youthful Suicide inpatients who were brIth assaultive and dimensions so it is impossible to determine suicidaL the degree to which the various dimensions can vary. Predictors are reported tobe sig- The place of the BPD diagnosis remains con- nificantly different in large groups of suicidal troversial in adult psychiatty and this is even individuals compared with nonsuicidal in- more the case for adolescentpatients who dividuals, but the absolute and comparative have received less systematic porsonality dis- predictive powers of the variables for the in- order assessment.Nonetbeiess, it seems dividual patient are not calculated. Studies likely that a diagnostic construct tapping of personality traits are usually not coor- charactetological instability usefully predicts dinated with studies of personality disordets for suicidal behavioe and thir can be reliably so it is impossible to determine thedegree to assessed fairly early in life. which these are correlated. Finally, there is It is of great theoretical interest, and also of the problem of comparison grovps. Many practical importance, that antisocial and bor- studies of personality traits of suicidal in- derline personality disorders that are most dividuals compared them to normal controls. associated with adult suicidal behavior are Unless subjects are equivalent in their also, in their adolescent form, (that is, con- psychiatric diagnosis, it is impossible to deter- duct disorder substituting for antisocial) the mine the degree to which a given finding in most common personality disorders predict- the suicidal group is specific to suicide or ing adolescent suicidal behavior. Moreover, whether it represents a trait more generally the personality trait measures associated characteristic of psychiatric patients. with suicide are similar in adults and adolescents. METHODOLOGICAL ISSUES Personality Traits: In contrast to the rela- The Ability to Assess Personality in Youth: tive paucity of studies having systematic Before we can confidently determine psychiatric diagnosis, a number of studies whether particular petsonality disorders or have focused on personality traits in suicidal personality traits are useful predictors of youngsters. The personality traits most com- suicide, we must address the more fundamen- monly found in suicidal adolescents are tal question about the degree to which per- equivalent to those found in adults and in- sonality assessment is meaningful in the clude aggressiveness (83-87), irritability (89- younger age groups. How doespersonality 92), low frustration tolerance (83,84), social diagnosis in children approximate that in isolation (83,87,92-98), hopelessness and adults? helplessness (91,104), poor self-concept 91,99-101), sexual conflic4 (93,101,102), The DSM III defmition of personality disor- probkin solving (100,103), resentful- der requires an age of onset that occurs by ! ess(88), and external locus control (105). It adolescence or earlier and & continuous is of interest that these personality traits course throughout most ofadult life. The (especially aggressiveness, irritability, low relationship of personality disorder diagnosis frustration tolerance, and resentfulness) are in children and adolescents to that in adults fully consistent with the personality diagnosis addressed specifically, and in some detail, tic categories (i.e., conduct disorder, border- in the introduction to the DSM III per- line personality disorder) most often found sonality disorders section.It ix proposed in youthful suicides. (without any great empirical support) that certain diagnostic categories from the DSM As is the case in studies of personality dimen- III Infancy, Childhood, and Adolescer-e sec- sions in adults, there have been several major tion correspond to, and in effect eventually problems in studies of such dimensions in develop into, certain personality diserders adolescent& Gmerally, a given study focuses (e.g. Schizoid Disorder of Childhood or on only a small numbr rof possibly important Adolescence into Schizoid Personality Dis-

195 2-163 Report of the Secretary's Task Force on youth Suicide order; Avoidant Disorder of Childhood or complicated ways with personclity function- Adolescence into Moidant Personality Dis- ing; they may cause personality dysfunction order; Conduct tworder into Antisocial Per- or, conversely, personality dysfunction may sonality Disorder; Oppositional Disorder predispose to state conditions, or, in many in- into Passive Aggressive Personality Disor- stances each may influence the other. The der, and Identity Disorder into Borderline Axis VAxis II confound presents major Personality Disorder). The presumption is problems in adult personality disorder diag- that the childhood or adolescent condition nosis, but there are several reasons to sup- will be diagnosed if the individual is under pose that it is an even more difficult problem age 18 whenever the personality in childhood and adolescence.First, the psychopathology has persisted at an intensity various Axis I conditions have been teas sufficient to meet disorder criteria. Adult clearly and definitively described in younger personality disorders without a correspond- patients and, in this age group, may be more ing childhood or adolescent category (e.g., likely to present in atypical or individual ways hystrionic or paranoid) can be applied in (perbaps influenced by developmental fac- childhood or in adolescence "in those un- tors). Second, children have a less extensive usual instances in which the particular traits track record on which to deride whether the appear to be stable. When this is done there problems are more state- or mow trait-re- is obviously less certainty that the personality lated. This combination of atypical Axis I disorder will persist unchanged over time.* presentations and a limited Axis II data base makes it doubly difficult to determine with Indeed, there are a number of reasons to be any certaintynether a particular symptom concerned that personality assessment may or behavior (say irritability and/or poor con- be less stable over time and predictive of fu- duct) arises from an Axis I syndrome (e.g., ture behavior in younger individuals. Since depression) or instead represents the early the past is the best predictor of the future in onset of stable personality features that may most things, including behavior, it makes beczome manifest, for example, es conduct sense to assume that the more of the pot one disorder or, ultimately, as antisocial per- has available, the more accurate the ptedic- sonality. tion will be.One is on statistically safer ground predicting that an individual with 30 The confounding of situational factors and previous criminal offenses by age 30 will soon personality disorders is also a problem in commit more crimes than that a lust offense adult petsonality diagnosis but may be even at age 16 will be repeated over and over more difficult in childhood because children again. This general threat to the stability of tend to be more influenced by their social en- personality aasessment in early life is en- vitonment and its role expect ,tions. For ex- hanced even further by this= more specia- ample, when a youngster presents with a confoundt (personality/state; per- conduct disturbance, it is difficult to deter- sonality/role; and personality/developmen- mine whether he is responding to a disturbed tal) which are also inherent problems in adult family environment or to peer pressure or personality diagnoses but become especially whether this is the beginning of what will be- problematic in assessing youths. We will dis- come an antisocial personality disorder. in cuss each of these in turn. the first two instances, the appropriate diagnosis would be adjustment disorder with dis- It is well documented that current state fac- turbance of conduct, with the expectation tors in adults (particularly the presence of a. - that the conduct problems will be self-limited companying mond disorder) greatly if the precipitating stressors are removed. A influence personality ratings by causing diagnosis of conduct disorder or personality retrospectively distorted reporting of pre- disorder implies that the behavior is more vious behaviors. State conditions, especially specific to the individual and likely to be those that are chronic, may also interact in

2-164 1 G Anances: Personality as a Predictor of YouthfulSuicide stable and manifest across different social cient stability to suggest the value of early situations and role expectations. detection, treatment, and prevention. Developmental changes constitute the third The best, although partial, solution to the specific confound complicating personality methodological problem raised in this sec- issaament in children and adolescents. It is tion is to develop semistructured personality often difficult to predict prospectively disorder interviews adapted for children and whether a given behavior represents a stage- adolocents. This is analogous to the fairly specific manifestation that the child is likely recent development of specialized AxisI in- to "grow out or or whether it is the beginning terviews adapted especially for children of a stable and lifelong pattern of personality (KIDDIE SADS).It will be necessary to functioning. For example, as a group, adoles- define more specifically the behavioral cents in our society are probably morenarcis- criteria for childhood personality disorder as sistic, troubled by identity problems, and this is not handled with sufficient clarity in prone to conduct disturbances thanthey will DSM IIL Personality assessment in children be as adults. The diagnoses of narcissistic, will always be more difficult and less predic- borderline, or antisocial personality disorder tive than in adults, but such assessment will therefore will be applied prematurely and improve as it becomes mote systematic and too liberally if these are based on a smallslice as empirical data accumulate. of developmentally influenced adolescent Possible Relationships Between Personality behavior. Rather, such diagnosis should be Variables and Suicide: Establishing a cor- based on a longer and wider strip of life ex- relational relationship between personality perience beyond the confmas of the develop- variables and suicide does not alone establish mental epoch in which such behaviors are the direction of causality. We will discuss the less pathologic and specific to the individual. several possible relationships that may be in- Despite all these methodological cautions volved: concerning personality diagnosis in a. Definitional overlap:suicidal behavior childhood, there is evidence that some per- may form an inherent partof the personality characteristics consistent across time sonality disorder dermition, just as suicidal can be detected fairlyearly in life. A number behavior is included within the definition of studies indicate that marked individual dif- for major depression. For example, DSM ferences in aggression become manifest early III includes reference to suicide within the in life (certainly by age 3) and remain stable criteria sets for both the borderline and to a degree that approximates the stabilityof hystrionic personality disorders. the LQ. There is also abundant evidence that Whenever suicidal behavior is included the presence of conduct disorders in within the definitional set for a disorder, it childhood is unii:ormly obtained in the his- is inevitable that there will be some con- tories of individuals who go on to exhibit nix-Hon between that disorder and suicide. adult antisocial personality. (Note, however, To establish that the relationship is real that the majority of childhood conduct disor- and not just definitional, one must der youngsters do not grow up to be antiso- demonstrate empirically that the criteria dal.) (1064 l ).These data suggest that set for the personality disorder predictsfor although many children with conduct disor- suicide even when the suicide item is ders 'grow out' of them, kaany others do not. omitted. A childhood diagnosis of conduct disorder predicts both for childhood suicide and also b. I' -rsonality disorder directly predisposes for adult antisocial personality. Thus, in the to suicidal behavior or to a form ofsuicide persocality areas that most reliably predict attempt (e.g., if one considers theimpul- youthful and adult suicide, the pertinent per- sivity and aggressivity of BPD as a direct sonality variables have demonstrated suffi- cause of suicidal behavior).

2-165 197 Report of the Secreteuys Task Force on Youth Suicide c. Personality disorders may predispose to Attempters vs. Completers:In adults, Axis I disorders (e.g., depression), which suicide attempters and suicide completers then independently increase the risk for seem to constitute two separate, but overlap- suicide. ping populations. A previous attempt carries d. Personality disorders may exert an in- an increased risk of eventual suicide (2% in fluence on the expression of the Axis I one year and 10% lifetime), but attempters condition so that the suicide risk is in- do not greatly resemble completers in creased (e.g., most depressed patients do demographic, diagnostic, or personality vari- not suicide, the presence of personality ables. The relationship of attempters to disorder may increase the vulnerability to completers needs to be defined for the child suicide in depressed patients. and adolescent populations. The degree to which data on attempters (who are much e. Axis I conditions (e.g., depression), espe- easier to study) can be extrapolated to corn- cially in chronic presentations, may pleters remains unclear. Moreover, the data predispose to behaviors that are indistin- gathered after a suicide attempt may not ac- guishable from personality disorders or curately reflect the individual's presuicidal may exacerbate personality characteristics functioning. Studies on personality charac- so that they present at the disorder level teristics in successfully completed suicides and/or the combination may interact to in- will have to depend upon informant methods crease risk for suicide. of data gathering that are now being 1. Personality traits (e.g., impulsivity or ag- developed for &adult personality assessment. gressivity) that cut across the categorical Informant methods also may be very useful Axis II personality disorders may for childhood attempters who are not very predispose to suicidal behavior. reliable reporters of their own personality characteristics. g. The covariation between suicide and personality disorders may be based on chance Comparison Group Selection: Many studies or on the covariation of each with some attempeng to isolate the characteristics of other underlying factor. suicidal patients compare them with normal controls. This is a serious methodological A number of different types of studies and limitation, given that factors in the suicidal analyses are necessar3r to establish the nature patients may be secondary to their psychiatric of causality. The first step is to document disturbance and not I-pa ticularly specific that the prevalence of person.lity disorders to, or predictive of, suicide. To relate a risk is higher in suicidal patients compared with factor specifically to suicide, it is necessary to nonsuicidal patients, controlling for Axis I use a comparison group that is matched on diagnosis. This would establish that there is diagnosis (as well as other possibly pertinent a greater than baseline or chance comor- variables) so that suicide is the only uncon- bidity. The degree of independer spezific trolled variable in the comparison. Risk fac- contribution to suicide of the pertinent per- tors may vary by sex and age groups. sonality disorders (BPD and APD) can be determined by comparing their rates of Interactions with Environmental Variables: suicide with those that occur in other kinds Personality variables usually have been con- of personality disorder and also comparing sidered in isolation from the environmental the rates that obtain for them with and variables with which they may interact in im- without co- abid Axis I disorders. Studies portant ways. Future studies will have to should alsi compare that predictive power of redress this simplification. It will be neces- the categorical DSM III disorder system with sary not only to tap the personality variables dimensional measures of pertinent per- and the environmental variables associated sonality traits (e.g., aggressivity). with suicide but also to determine the specific interactions between these variables that

2-166 I P S A.Frances: Personality as a Predictor of Youthful Suicide heighten risk (e.g., angty impulsivity in a bor- Suggestions for Current Clinical Practice: derline petsonality disorder interacting with Youngsters who present with conduct disor- the loss of love object or the conduct disorders and* the impulsivity associated with der individual who has been caught in a mis- botderline personality disorder deseive an demeanor). especially thorough diagnostic evaluation that specifically assesses for the possible Predktive Power: Suicide is a rare event, is presence of Axis I conditions (e.g.,affective associated with many correlates, and may disorder and substance abuse). Suicide risk result from heterogeneous causes. It seems and lethality appear to be highest for patients unhiely that any variable, or grouping of vari- who present with comorbid combinations of ables, will ever have a high predictive power Axis I and posonality disorders so that ag- for suicide, especially in cross-sectional gressive treatment of the depression and/or evaluation. By isolating personality variables substance abuse is crucial in this group. The associated with suicide, we can probably im- - sence of specific stress=(trouble with prove our ability toidentify a group of law, loss of love object) that are par- younptets at high suicide risk who deserve ficularly associated with suicide in per- extra treatment and preventive efforts. As- sonality-disordered youngsters should ater sessment is unlikely ever to be very success- the clinician to increased risk and provide a ful in predicting which specific youth is at target for immediate education andinterven- high risk to attempt suicide in the very near tion. Although effective treatments for con- future. duct and borderline disorders have yet to be documented, certain promising leads deserve DISCUSSION further clinical and research attention. Thanks largely to the increased reliability af- Suggestions for Prevention:Robins and forded by DSM III, petsonality disorder re- Earls (112) have suggested a promising research in adults has recently been flourishing search design to test the ability of a special and promises to provide incteasing clarity on pievention strategy to reduce the incidence the relationship of Axis I and Axis II disc. - of conduct disorder. They would select a ders and in the interaction of these with group oi children who have atleast a 50 per- suicidal behavior. However, the technical in- cent morbid risk of developingconduct dis- novations in personality assessment order (i.e., those who have at least one developed for adults have not yet been trans- antisocial parent). Since some of the origins lated into improved assessments of of conduct disorder may occur early (prena- childhood personality features.Although tal exposure to neurotoxins and distress, the diagnosis of personality in youngsters is postnatal trauma and illness, and parental inherently problematic in the many ways we deprivation) and since the earliest manifests- hay.; outlined, it is likely to improve gteatly tion may occur during the preschool years, as research attention turnsin this direction. the preventive intervention would be The results of this literature review suggest designed for early delivery. The sample that the personality features that predict would be selected from pregnant women youthful suicide are very closely related to with antisocial disorder. The intervention those that are associated with suicide in aim to offset known risks for conduct disor- adults (borderline personality, conduct dis- der by providing adequate prenatal and order/antisocial personality, impulsivity, ag- pediatric care, offering high risk infants a gression, social withdrawal).This would special curriculum to increase language and seem to confirm the continuityof personality social skills, and providing training and sup- factors throughout the life cycle iss possible port of the parents. The researchdesign contributors to suicide risk. would include randomization to a special developmental center or to a no-intervention control group. If the intervention were sue-

1 n 2-167 Report of the Secretary's Task Force on Youth Suicide cosful, Robins and Earls would expect sig- It seems crucial to develop programs of nificant differences to be evident by age 3 and prevention and treatment to counteract the that these would be substantial through age disturbing cohort effect for conduct and af- 10.Positive results detected at this point fective disorder Ent may be responsible for would presumably reduce the longer term the increasing ra Le of suicide in younger age risk of antisocial personality and also of groups. youthful adult suicide, although such determination would require additional lon- Specific conclusions derived trom our review include: gitudinal study. This strategy for studying the effects of primary prevention for conduct dis- The personality predictors of youthful orders would appear to be feasible, cost-ef- suicide arc equivalent to the personality fective, and likely to have an impact on predictors of adult suicide. youthful suicide rates. There is stability of adolescent per- Suggestions for Future Research: The most sonality diagnoses into adulthood. immediate need is to develop methods to n- Conduct disorder and borderline per- um childhood personality disorders and personality disorder are the most important sonality traits.Available personality personality disorder risk factors for definitions and assessment instruments adolescent suicide. designed for adults have been developed only recently and must now be adapted for Certain dimensional personality and cog- use in youngstem. The stability and predic- nitive traits (e.g., aggressivity, impul- t ive power of childhood personality sivity, hopelessness, social isolation) may measures and their relation to Axis I condi- be important predictors, perhaps cutting tions must be determined empirically. Once across categorical personality disorder assessment problems have been addressed, it diagnosis. will be important to evaluate treatment and The comorbidity of Axis I and personality prevention programs. diagnosis increases the frequency and lethality of attempts. CONCLI ISIONS There may be a contributoty role of a Personality disorder diagnosis has only family history of antisocial behavior, sub- recently achieved reliability, and empirical stance abuse, and/or affective disorders studies have only just begun to demonstrate to suicidal behavior. the predictive power of personality variables. Comorbidity plus family history may in- Thus far, almost all the available research has crease lethality (both through genetics been conducted among adults, and we must and environmental reinforcers). recognize that personality assessment in childhood and adolescence is difficult and There is a need for better personality subject to inherent limitations. Nonetheless, measures and assessments for children it is fascinating that the very same personality and adolescents. variables that are associated with adult There is a need to establish whether the suicide are also associated with youth suicide, biological (particularly serotonergic) suggesting that preventive effortn focused on correlates of personality and suicidal be- pemonality variables should be targeted to havior found in adults also apply to early identification, treatment and preven- adolescents. tion, in high risk populations. Interventions that are effective in reducing personality risk It is necessary to determinfr.; how per- factors are likely to reduce the suicide rates, sonality disorder/traits interact with not only in youthful populations but also in other risk factors (i.e., family history, these same populations as they grow older. biological abnormalities, lack of social

2-168 2;:t1 AFrences: Personality as aPredictor of Youthful Suicide

supports, other psychosocial risk factors,

presence or affective disorder) to in- 18, IMwd NO, Bonuowft Mk 1980. Factors assodated crease risk for suicidal behavior. with suicidal behavior hi polydrug sham. J. CNn. Psych. 41(11)279485. Perhaps most important of all is the pos- 19. °Mon Ikt 1995. Suicide. Psych. CM. N. Amer. sibility that programs of primary and 8(2):203-2 4. 20. Crook T, Raskin A & David IX 1978. Factors as- secondary prevention that succeed in sociated with atlemptrod suicide among hospitalised reducing the morbidity of personality dis- depressed patients. Psychol. Mod. 5:381-308. orders in the young may thereby reduce 21. Payial ES, Meng kt 1971. Suicide =01- lowing acute depression. J. Nem. Met Ole. 1 not only the youthful suicide rate, but 22. Henderson AS, !Wigan J,Devideon J. el al: 1977. also the rates of adult personality disor- A typology of parasuicide. Br. J. Psych. 13101441. ders and the rates of adult suicide. 23. Conte HR & Plutchlk It1974. Personality and background characteristics of suicklal mantel pants. J. Psych, Roe. 10131-108. REFERENCES 24. Vinod& KS: 1981 Personality charactsristics of at- 'woad middies Bt. J. Psych. 11t1143-1 180. Rance* .A,FvsrMOailchijPersonsitty and 25. Mutely V14:1989. Personality and the nature of Subic% In PsychobIology of Suicidal Behavior Od. J. suicidal atkimpts. Br. J. Psych. 115:791490. Mann, M. Stanley) New odc Academy of Seisms, New 25. Bkichnell J:1961. Sem* familial and clinical York; in pram characteristics of female suicidal psychiatric Wants. Br. 2. Mika! HS, Chen SE, Davis GC, et al: 1985 Bomar- J. Psych. 138481-390. One: An adjective in search of a noun. J. Pin. Psych. 4542- 27. Pallis DJ & Binchnsli J: 1977. Serious of suicide 411. sttsmpt In relation to pwsonallty. Br. Psych. 130253- jr., Jonas JM. Hudson, st al: 1983. 259. The of DSM-111 Eardertine personality disorder. 28, Weissman MM, Fos K & Kiernan OL 1973. Ho- Arch. Gen. 41k23-30. eft end depression *sweated with suicide attempts. Am. 4. Stone, Mk 1900. Long Term of Bordw- J. Psydt. 130(4):450-454. line Personality Disorder. Journal of Personality Disorders 29. PP.lwp k 1970. Traits, attitudes and symptoms in Pealls). a group of stamptsd suicides. Br. J. Psych. 118:475482. 5. Fps M, Frances A. Sullivan T, el al: (Unpublished). 30. Collor. PC:1975. Ploortalty chs.rectsdstice Bordsrline personality disorder and alfccihms disorder Int- found among itouthful female suicide OwieL J. Ab- pact of comorbidity on suicide normal Psycho]. 85(3):324-392. 0. Rounsaville BJ, %%Immo V libber H & Wilber 31. Toppi P & Rernikoff kt 1982. Perceived peer and 0 NU Iftwopmdlyof in Vested family ritationships, hopetseeness and bras of mu-0i as opiate addicts. Arch.am.1=1311:101-180. factors In adolescent suicide attempts. 7. Rkidman RC, Amnon MS, Caddo JF, et al: 1983. of suicidal behavior in depressed borderline In- 32. Nelson", Wan EC & Mickel% MT: 1977. In- =Am. J. Psych. 14*1023-1026 terpensonal altitudes of suicidal Individuals. Psychological Reports 49193-989. 8. Meddooks RD: 1970. A Ode yaw failowup of un- 33. Farberow N, Dswies A& 1987. An hem Offwentla- treated psychopaths. Br. J. Nab. llekilt 1-516 don Instals of MMPla of suicidesl nsuropeychlaide hospi- 9. ibbins IN: 1900. Deviant Children Grown Up. tal patients. Psychological Reports 2611074117. & Mins, Baltknore. 34. Vain A. Sinai R & Wks It 1972. Adolescents in 10. Woodruff RA, Jr., Clayton PJ & Ouzo SB: Suicide crises: saltation of qusstionnaire. Am. J. Psph.129:574- attempts and psychlablet diagnosis. Da. Ner. Syst. 33:817- 577'. 821. 35. Rushing: 1989. Devlanos, interpersonal relations 11. Mks k 1977. ConditionsgsdAosing to suicide: and suicide. Human Relations 22(1):01-78- Areview.Nwv. Mont Dle. 164: 1 38. Meyhryer AH, Hskmat H &MOM R 1977. Some 12. Owvey MJ & &soden F: 1000. &kids attempts In correlates of cordemplatsd suicide. Psychol. antisocial personality disorder. Comore. Paphietr. 21 Crp810-1.8aof21:1201-129. (2)140449. 7. Flood R & Surf 0 190. A ex- 13. Morgan HG, Barton J, Pea LS, st al:1978. ammation of psychiatric cue moor& of pitiants who sub- DelthersIe self-harm: A followup study of 279 pedants. Br. madly committsd suicide. Bt. J. Psych. 114:443452. J. Psych. 120:301-380. 38. Roes MW. Myer JR & Campbell RL 1983. Nan- pa:Luatisse P & Horton it1974. The rspetition of tal raring pantos and sudal thoughts. Acta. Psychlatr. : A oompedson of dime cohorts. Br. J. Psych. Scand. 07:429433. 100-174. 39. Want RD: 1975. Self conospt and suicidal Intent. 15. Robins E, Schmidt EH & motto P: 1057. Some in- Psychological Reports 30:279-282. terrelations of social factors and clinical diagnosis In at- 40. Farberow NL & McEvoy YL Sett Sulold among tempted suicide: A study of WO patients. Am. J. Psych. patkints with diagnoses of anxiety reaction or 114:=1-231. npacNon in vowel msdical and surgical J. Ab- 10. Batchelor, IRC: 1964. Psychopathic states and at- normal Psychol. 71:287-20. tempted skid*. Br. Med. 1:13424347. 41. Slosh L & Weisbauch .61:1972. Suicide: an 17. Robins, IN, Murphy GE, Wilkinson RH, et al: 1959. epidemiological study. Dis. New. Syst. 3923-29. Some GNAW considendleons In the prevention of @mid* 43. Kimono DK & Cawfud CS: 1986 Ssif-evalua- based on a *dye. 134 successful suicides. Am. J. Public tons of suicidal mental health patients. J. CM. Perthol. Health. 4Ik9. t279279.

2-169 201 Report of the Secretairs Task Force on Youth Suicide

43. Wilson LM, &taught JN, MieldmIns RW, Bony Kt.: 87. Benidom, V** M, Papp Z, et st: Cersbrospinal 1971. The wan Waldo attempter and witooncopt. J. fluid magnesium and caicium related to amine motabo- Clin. Psychol. 27:307-309. Nies, diagnosis and suicide attempt. BioL Psych. 950:103- 44. Nawkw 0 1973. 'etude tweed self in 'Waldo! 171. individuals. Me-dwededrp Behavior. 4:118-108. 68.Brovm GL, Goodwin RC, Banger JC, et 1979. 45. Nmeinger 0 1974. Soffenctelher-appcalsols by Aggression in humans correlates with CSF metabolites. suicidal, psychosomatic and nonnal hoopitsked patients. Psych. Rse. 1:131. J. Consult. Clin. Psychot 42:300 (10. Brown CIL, Ebel ME, Go. sr PF, et at 1982. Ag- gression, suicide and serotonin:Relationships to 0-11F 45. &Men, t./ & Wing CD: 1982. Suicidal tendon- amine metabolites. Am. J. Psych. 139631438. cies among college student" Psych. Otiartedy 38.55 O- 70. Pam It 1983. Life at tisk: Markers of suicidelity and depression. Psych. Dowel. 1:81. 47. Bedrosian RC & Beck AT: 1979. Cognitive aspects of suicidal behavior. Suicide and Life-ThreatoWng Be- 71. Robins LN: Study ohlicMoo_ d predictors of adult sp- haviot.10117-98. ike:Ida behavior. Psychological Medicine 8:1311422, 1978. 48. Beck AT:1983. Thinking and deprossional 72. Shaffer 0: Suicide In childhood and early adoles- =tit centent raid oognItive Melton. MO. Gen. cence. J. Child Psycho!. Psychiat 15:275-201, mit 73. Pfeffer CR: Selkiestruotha behavior in children 49. Hake ft, Bergman E, Omit AT & Book Ft 1973. and adolescents. PsycliIalr. OK of North Mot* VoL 5, Hopelessness, depression and attempted suicide. Am. J. No. 2, 19115. Psych. 130(4):458-49). 74. Shaffer D: Diagnostic Considerations in suicidal 50. Went FV: 19776102= =noes, sex, behavior in children and adolescent*. J. Me. Acad. ChM and suicide potential. 40127-928. Psych. 21:414415, 1982 51. Boor M: 1070 Relationship of koomatextemal 75. Carlson C. Cantwell DP: Suicidal behavior and control and United Steles suicide toss. 1915-73. J. Can. &wilain children and adolescents. J. Mi. Acad. of Psychol. 38(4):795797. siPmcnillatry21:381-388, 1982. 52. Patelokas AT, Slum GA & Luecomb RL: 1979. Cog- 78. Pfeffer CR, Plutchik R, flizouchl MS: Suicidal and nitre characteristics of suicide Simpler*. J. Consult & assaultive behavior In children: classification, measure- Clin. Psychol. 47(3):470484. ment and interelations. Mv. J. Psychiatry. 53. Nearing*: 0: 1904. Rigid thinking In suicidal in- 77. Cohen-Sandier R, Barnum AL, letv Rk Life Strom dividuals. J. CormulL Psycho!. tic8458. and symptomatology. Determinants of Stdid behavior 54. WW1* DB & Oum ak 1982. *Wick% ideation in in childmn. J. Nn. Acad. Child Psych. 21:178.188$ 1082. its=so popuisSon: A test of a model. J. Consult. & Ciin. 713. Outdo T, Fdedman R Kin S, Coen R & knonoff 800I):111041911. Affective end character pathology of suicidal Etches- cent and young adult inpatients. Psyohlatr. 45:19-21 1954. 85. Eashvood hiFt, Henderson RS & Montgomery 1M: 1978. Personality and parasuloldal methodological 79. Crumley FE Adolescent suicide ettempte and bor- probkans. Med. J. Aust. 1:170-173. derline personality disorder: clinical features. Southern Medical Journal 74:5484549, 1981. Goidney RD: 1081. Are young women who attempt suicide hystsvicart Br. J. Psych. IM:41-148. 80. Crum* FE: The adolescent suicide attempt A ST. Farberow 1950. Personality patterns of J. suicidal mental hospital patients. Gen. Psycho!. Psychotherapy 381511-185, 1 Monograms 42:3-79. SI. Alessi NE, McManus M, Brikman A & Grepetine L: Ocoton J, Post R. Lam* J: 1983. Identification of suicide .0ers _by moos of MMPI profiles. J. Glib. J.°1"18t1Psycti;sib:1A::::Zemcf:9 1;11;1°.91:919sYchiatilluveni:disardWb1"Am. Psychoi. 311(41):812,871. R. Friedman Rt, Clarldn JF, Com Ft OSM id end cf- SI Roy k 1978. Selfmutilation. Br. J. Mod. Psychol fective_pathology hesphoilzed adoisatents. Nerv. 51:20140a Mont. Dis. 175:511-521. 1902. 83. Ptna SV & Riddle ktAdolescent suicide: 80. Mini $: 1978. Pomona* correlates of suicidal litrussosal and cognitive a9ponle. tendency among Iranian and Turidan studeMs. J. of Ada.9:343398, Psycho/. 98:1151-163. 84. Cantor PC:Personality characteristics found St. Colson CE: 1978. Neurodoism, extraversion end among youthful female suicide atternotars. repressionesnsitication in suicidal opting, students. Brit. Abnormal J. Soc. & Oln. Psycho!. 11:8080. Psychol. 85:324-32S, 1978. 85. Goldberg EL: Dejernssion and suicide ideation In 82. Paulin DJ & Jenkins JS:1977. Extraversion. the young adult. kn. J. Peyohlote. 138.1, "981. neurotic:km and intent in attempted sudes. Psycimlogi- cal R torte 41:1022. atTishier CI. I McKenry PC: Inuopyschlo symptom dimensions of adolescent suicide altempters. 83. TraokmanPabeig BertNeson L & *strand L: 1981. Moms** mstaboilss in cerebrospinal fluidand 87. Hawton K, Cole D, M: Motiva- sudal behavior. Arch. Gen. Psych. 30081-08. tional impede of deliberate self ing In adolescents. Psychlut. 141:288-291, 1 154. van Prase It 1982. Depression, suicide and metabolism of serolonkv in the brain. J. At Dle. 4:278-290. W. Lester Suicide as an aggressive act A re0a- I& Sabot M, Tnekman L & Theron P: 1978. 5 HIM in don with a control for noun:410km. J. Geri. Paphol. 79:83- the fluid: A biochemiod suicide prediction? Arch. Gen. Psych. 01190119?. 39. Haider L: Suicidal attempts in children and adoles- 85. rixottak1081trylom pawns in unipolar and cents. Brit. J. of Psychist. 114:1113-1134, 1030 bipolar eith monamlne metabo- 90. Wan JM:Suicide and suicidal attempts in lites in the oefebrospinalw" Sw7dNgolds. Psych. Res. 3:225- children and adolescents. Am. J. of Psychiat. 118:716-724, MI 1962. AFrances: Personality as a Predictor of YouthlUlSuicide

91. Marks PA & Helier DL Now I lay me down for keeps: A etWfamadoMecen1 suicide attempts. J. CM. Psycho!. 31k 9Z Jacobs .1: Adolescent Suicide. New York: Wrierks- tersolenoe. 93. Peck ML & Schad k duloidal behavior among college students. HSMHA Health Reports 88149-158. Luidanowloa NMempled suicide In children. Acta Psych. Sosndinavios 44:415435. 91 Schrift k Some typical patterns in the behavior and background of adolescent We who attempt sude. Arm J. Psychlst. 125:107412. 98. Barter JT, %%back DO & Todd 0:Molest:lent suicide attempts: a follow-up study of hospitalizsa patients. Arch. Gen. Peyohistr. 193.7, 1088. 97. Yoko A, SlnsvR&MNmK Moleeoents or valuation of a quesbonnaks. Mt. J.Peyote*. 12e574- tin 98.Ifieltrel WD, Novlano V & Hatcher: Adolescent failure during mondary socialization: A study of army trainee casualties. J. Psychiatr. Res. 13125-13& 1077. 90. WO, M$, We CR, YAW RL & Soleicht Recurrent adolescent Wddal behavior. Pediatrics 0005- 0133, 1977. 100. Levenson M & Newinger C: intropunitiveness in suicidal oddments. J. of Projective Tech. end Pereonalitv Assessment 34409411. 101. Wiseman LA: Atiornined suicide in adolescents: A suicide prevention center in Rhode Island Is in urgent need. Rhode bland Medical Journal M449451,1989. 102. Sabbath JC: The suicidel adolescent - the= able child. J. of Mi. Acad. ChM PeyoNsay & 1909. 103. Merman JS & Tucker GJ: Wort 'she peraistelft suloidal patient. Comprehensive PsyrAistry 14:71-79, fen. 104. Memos FT & Weisz AE: The psrtonal future and suicidal ideation. J. of Nam and Menu Du. 153.244-250. 1971. 105. Goldney RD: Locus of controi in young women who hue attempted suicide. J. of WC. aniMenL Disease 70:4. 108. Farrington DP: The family backgrounds of sive ruts. L Herm, M. Berger & 0 Shaffer (eds. Ni- and ',discoid dki~1 In children. ggles=: Perqamon Press, 1978. 107._ Moore DR & Ankur JL: Juvenile delinquency. In T.H. 011endidc & t,t Hereon Handbook of Chiki Psychopathology. New York. Flown Prom 1983. 108. Olweus D: Aggression end peer acceptanos in adolescent bow Uwe ehort-lerm longitudinal studies of ratings. Child Dowftpment, NM 48:1301-1313. 103 Otwaus =ftr of '11st reaction ;teems In males: A review. Wolin, 1979, 118.1152- 1375. 110. Rolf kt Childhood scold Interactions and young adult bad oonduct.Journal of Abnormal and Social Psychology, 103, 03333437. 111. RA JD & Yen RD: Childhood aggression and scold adjustment as antecedents of delinquency. Journalof Abnormal Odd Psycho/ow, 1004, 1Z 111-120. 112. Earls F: Towards the Prevention of orders. kr The MIMI Review of Psychiatry .R. Hales & Frames) American Psychiatric Press, D.C. 1987.

Z03 2-171 Wet: SUBSTANCE USE AND ABUSE A RISK FACTOR IN YOUTH SUICIDE

Marc A. Schucldt, M.D., Professor of Psychiatiy, University of California Medical School, and Director, illcohol Research Centcr, San Diego Veterans Administration Medical Center, San Diego, California

Judith I. Schucldt, San Diego, California

INTRODUCTION This paper reviews the relationship between often use the Research Diagnostic Criteria the use and abuse of substances and adoles- (111GC) or the Third Dia ;nostic and Statisti- cent suicidal behaviors. Before presenting cal Manual of the American Psychiatric As- the actual data, it is important to address sociation (DSM III) (4,5). Reflecting these some relevant problems of definition. more rigorous and reliable classifications, whenever possible emphasis is placed on For purposes of this overview, childhood and findings from current studies. adolescence extend to age 24, although the majority of studies have focused on people Discussions of behavior associated with sub- 19 years old or less. Suicidal behavior can instance abuse, however, must go a step further clude intense thougtts of wishing to be dead; in classifying subjects.In the course of attempts or "gestwes* can relate to any level misuse of drugs or alcohol almost all types of of dPliberate self harm; while completion of behavioral aberrations can develop, includ- suie.de requires documentation througha ing severe states of anxiety, depression, acute coroner% report or interviews with 'Psig- psychoses, and severe confusion (1,2,6). nificant others.' Using diagnosis to indicate prognosis and to help in selection of treatment, it is important Alcohol and substance use connote intake of to attemapt to establish a hierarchy of these drugs without a&sociated major life problems. One approach is to differentiate problems (1,2). Information about use is dis- between primary and secondary illness (3,7). tinct from data on alcoholism or drug abuse A primary psychiatric label is assigned when which relate to heavy intake of drugs or al- an individual fulfills criteria for that disorder cohol and documentation of serious and per- and has no major preexisting psychiatric sistent related life problems (1,5). problem. An example would be an adoles- Unfortunately, in many studies specific cent who met the requirements for drug criteria are not clearly stated. abuse and who bad no antisocial personality Also of central importance to this discussion disorder (ASPD), major depressive disor- is the definition of psychiatric disorders. ders, rchizophrenia, etc., prior to the onset of Many of the earlier studies set forth no severe drug related difficulties. A secondary rigorous criteria for the syndromes being label is assigned when an individual meets describe.i. The more recent investigations criteria for a discrder only after another

0.% 2-172 204 M.Schuclidt Substance Use and Abuse major psychiatric illness was present. An ex- THE EPIDEMIOLOGY OF ample of this phenomenon is the man or SUICIDE ATTEMPTS AND woman with an ASPD (e.g. the onset prior to COMPLETIONS the age of 15 of pervasive antisocial difficulties) who, at age 18, went on to develop The following discussion distinguishes be- severe alcohol-related life problems;this is a tween the prevalence of suicide attempts and case of primary ASPD and secondaryal- completions. Within ezuh section, informa- coholism and the prognosis is hlely to be that tion is rust given regarding rates in the of the personality disorder, not alcoholism general population, and this is followed by (6,8-10). figures among adolescents.Each section concludes with data on changes in the rate of With these caveats in mind, we turn to a dis- these phenomena over the years. cussion of some risk factors associated with adolescent suicidal behavior.Section II briefly reviews the prevalence of suicide at- Attempts tempts and completions among young About 10 percent (10,000/100,000/yr) or people in order to plact. into perspective the more of people in the generalpopulation data that follow.Section III reviews the reported suieklal feelings over the prior year, direct relationship between substance use or including 2.5 percent who had more intense abuse and suicidal behavior, emphasizing the thoughts (11-13). Counting all age groups, application to adolescents. Section IV looks actual suicide attempts are observed at a rate at indirect associations betwent substance between 100 and 800/100,000/yr, with use or abuse and suicidalbehavior as they re- women age 15 to 24 years standing out with late to other primary diagnoses including tbe high figures shown in Table 1 (12,14,16). ASPD, borderline person3lity, affective or !!..k men, the 15 to 24 year olds also depressive disorders and schizophrenia. predominate, but in each age group the ac- This includes brief comm:nts on the ties be- tual rates are about half those seen in women tween suicidal behavior and substance abuse (12,14,15,17-22). in adolescents and family environment or family historj of psychiatric disorders. Fmal- Since the peak age for suicide attempts is be- ly, Section V synthcsizes the information and tween 15 and 24 years old (14-19), it is not offers some clinical and research implica- surprising that a number of studies have focused specifically on the attempt rate for tions.

Suicidal Behavior by Sex and Age per 100,000/yr(12-15,17,19522130,31)

ATTEMPTS COMPLETIONS ftal Male Female Male Female <15 26 92 <1 <1 15-24 423 786 16 4 25-44 267 598 24 10 45-64 152 257 26 14 65 + 70 41 33 11

Table 1.

205 2-173 Report of the Secretary's Task Force on Youth Suicide children and adolescents. Looking first at "epidemics* of suicidal behavior among more anecdotal data on children who have young people (36), completed suicide is sought help, it has been estimated that 3per- especially rare for children under age 15 cent of a consecutive series of young people (15,30.34 coming to private practice or psychiatric outpatient settings had ever attempted suicide. The prevalence of suicides in the United The same is true for 10 percent to 30 percent States and Canada appears to have been of adolescents coming to emergencyrooms, stable during the 1950s (20,_2). However,as and the rate of attempts increases to between reported for attempts, suicidal death began 10 percent and 50 percent among young to increase in the 1960s. The overall rate in psychiatric inpatients (18,20-22). These 1961 was 5/100,000/yr for men and 1 for figures are probably inflated because of the women in Alberta, Canada; es .tost double troubled nature of the populations observed. the figures from 1951 (32).Self-inflicted death rose to 25 and 5 for the two sexes by The high prevalence of suicide attempts in 1971, on to approximately 32 and 5 by 1976 any age group, including adolescents, is not a (32). In general, the U.S. suicide rate rose new phenomenon (13,16,17,23-25). Most of from 5.2 to 133/100,000/yr from 1960 to 1980 the liter sture focuses on more anecdotal (31).Other investigators have also docu- reports; between 1972 and 1980 there was a mented an increase of at least two- to three- 5-fold increase in adolescentsseen for fold between 1960 and 1980 (18,30,31,37,38). suicide attempts in a Louisville, Kentucky psychiatric hospital (26), and an almost dou- SUBSTANCE USE/M1SUSE bling of adolescent suicidal behaviorwas AND SUICIDE RISK: A DIRECT seen between 1970 and 1975 in a New Haven, Connecticut emergency room (14). ASSOCIATION Alcohol and drug use and violent death (ac- Suicide Completions cidents, homicide, and suicide) apotential- Suicide completion is a much rarer ly related in a number of ways (39-41). Drugs phenomenon than suicide attempt. The of abuse, especially brain depressants (e.g. ratio between the two depends barbiturates, antianxiety drugs, and alcohol) upon the and brain stimulants (e.g. amphetamines, definitions used, but is at least 10 to 1 (17,27), cocaine, weight reclining products) can im- and could be as high as 100 or more to 1 in pair judgment, increase levels of hnpulsivity, some groups (28,29). There is, however, an and are capable of producing severe mood important connection between attempts and disturbances, including temporary, intense completions, because as many as 50 percent or more of completers have attempted and suicidal depressions (1,6,42). Any sub- suicide in the past (27). stance of abuse can also exacerbate a preexisting state of psychopathology, includ- As shown in Table 1, overall about 13 people ing increasing the level of hallucinations or per 100,000 of the general population died by delusions in psychoses, enhancing anxiety, suicide in 1980, with a male preponderance and increasing levels of depression (1443- of between 2 or 3 to 1, and a peak rate for 45). The following sections briefly review in- men age 65 or older (15,30-33). As infre- forma tion on the association between quent as completed suicide is in the general substances and suicide attempts or comple- population, it is even less common among tions in general (Section A), as well as in adolescents. However, self-inflicted death adolescents (SectionB). has long rated as the third leading cause of mortality in this otherwise healthy group, Alcohols Drugs, and Suicide especially for youth with histories of Attempts: A General Discussion isychiatric care (18,21,34,35). Despite isolated cases of apparent short-term One obvious association between substances

2474 M.Schuatt Substance Use and Abuse and suicide is the use of alcohol or drug; as drug abusers as well, with at least 15 percent vehicles for the suicie -I act.Historically, admitting to past suicide attempts. The ai- drug overdoses of prescription or over-the- timated lifetime suit :de completion rate counter drugs have been a favored among some types of drug abusers exceeds 10 mechanism in suicide attempts, especially percent (1,53,56-59), including a four-fold in- overdoses with brain depressants in women creased rate over the general population (14,46). The use of drugs as the mechanism among amphetamine abusers, as well as a of attempt by young people is equally strong. high of suicide among cocaine and heroin ad- In one consecutive series of 505 adolescents dicts. Drug abusers who express extreme and children seen in a pediatric emergent.? feelings of depression or hopelessness may room for a suicide attempt, 88 percent h. I be at exceptionally high risk (60). used drug overdose, as had 78 percen te dO Considering the lifetime rate of suicide com- percent of suicide attempting youth reported pletion among substance abusers, it is not in other samples (1,22,4749), although some surprising that studies of patients who have authors have reported lower rates (30,31). committed suicide have noted that a high In any age group, intoxication with alcohol or proportion have a history of substance drugs oft. n immediately precedes suicidal misuse. Using rigorous criteria, 20 percent to behavior (17,39,50-52). In some instances, 50 percent of suicide attempters or coin- this relates to suicide attempts among al- pleters were found to be drug abusers, while cohol or drug abusers, but in others it reflects between 15 percent and 50 percent of com- the use of substances as part of the attempt pleted suicidai in several studies were al- itself or in an effort to *screw up enough coholics (27,33,63-68). courage' to carry it out. In any event, one In summary, from the studies in the general study reported as many as 70 percent of male population and focusing on all ages there ap- and 40 percent of female suicide attempters pears to be a close relationship between al- had consumed heavy doses of ethanol before cohol and drug use or abuse and suicide. The the act, with average resulting blood alcohol association is supported by followup studies concentrations (BACs) of almost 150 mil- of alcoholics or drug abusers, evaluations of ligrams per deciliter (mgAIL) for the men and the characteristics of suicide victims, and fol- over 100 mg/I:IL for the women (12).Alcohol lowup evaluations of psychiatric patients. MIS found to have 'contributed to thedeath,* through perhaps impaired judgment or exag- gerated mood swings as well as through ef- Alcohol, Drugs, and Suicidal fects on vital systems, in more than half the Behavior Among Adolescents autopsied casts in one study in Washington, Analyses of data for adolescents also docu- D.C., and more than two-thirds of the ment a close association between substances suicides in another study in New York City and suicide, as outlined in Table 2 (35,69,70). (61,62). This conclusion is supported by studies of There is ample documentation of a high suicidal youth, evaluations of substance suicide rate among substance abusers them- abusers, and through observation of young selves. At entrance into an alcohol treat- psychiatric patients. ment program, at least 20 percent of Adolescents who fulfill criteria for drug alcoholics report histories of suicide attempts abuse or who have relatively heavy drug in- (53). Prospective folowup of identified al- take patterns have an increased rate of death coholics have revealed a three-fold or higher overall, including high rates of suicide. A 10- increased rate of suicide completion year 'blowup of two groups of teenagers (41,54,55), with Miles estimating a lifetime (one from the general population and the rate as high as 15 percent (56). second identified because of prior drug use) The association with violent death extends to revealed a two- to seven-fold increased death

2-175 .iNt207 b:4it4 Report of the Seaetary's Task Force on Youth Suicide

rate among boys w'h histories of drug stancesabused per individuaL misuse, and an almost two-fold to eight-fokl increased death rate among girls (71). Ap- The association between suicidal behavior proximately half of this increase in death rate and drugs is just as strong among adolescent was from suicide. psychiatric patients.Robbins and Alessi studied 33 teenage psychiatric inpatients who The relationship between drugs, alcohol, and had histories of prior suicide attempts, look- suicide in young people is corroborated when ing for the relationship between alcohol or populationsidentified because of suicidal be- drug use and suicidal behavioral (18). They havior are evaluated.Patel reported that used an analysis of correlation that evaluates among suicide attempters age 12 to 19 years the degree to which two factors change at the old, 41 percent of the boys and 19 percent of same time (thehigherthe correlation, the the girls had beendrinking immediately greater the similarity in change). A history of before theattempt (13), and Garfinkel found alcohol abuse correlated with the number of a ten-fold higher rate of recent alcohol or past suicidal `gestures' at a 0.42 level drug use in 505 adolescent attempters than (p<.001), accounting for 25 percent of the for controls (22). Shan/ noted "frequent use variance or range of the number of gestures of nonprescription drugs oralcohor among in this group. Similarly, the ;An-relation be- 70 percent of the 20 teenagers who com- tween alcohol abuse and the seriousnas of mitted suicide in the Louisville area between past attempts was 035 (p<.01), and alcohol 1980 and 1983 (26), while almost halfof the problems correlated with the level of medi- suicides aged 15 to 19 in Erie County, New cal seriousness at 0.36 (p<.01). Overall, the York, had alcohol in their blood (50). The association between a history of alcohol study of suicides under age 30 in San Diego abuse and the occurrence of a suicide at- found that more than 75 percent abused tempt was 028 (p<.05). A history of drug drugs or alcohol, including between a third abuse correlated with the number of suicide and a half for whom these diagnoses were the 'gestures' at 032 (p<.05), and with the medi- primary illnesses (31). Among those with cal seriousness at 0.26 (p<.05). Those drug problems in the San Diego sample, 79 authors conclude that "substance abuse in percent had abused marijuana, 45 percent depressedadolescents appears both to in- cocaine, 34 percent amphetamines, and crease the risk of multiple attempts and to about 25 percent each had abused opiates, add to the medical seriousness of the at- sedatives/ hypnotics, or ballut.inogens. tempt.' There was an average of three to five sub-

AIIIMENIMMINIMMIEW Suicidal Behavior and Substance Abuse in Adolescents are Linked (13,1142,26,30,31,50,71)

Among heaw substance users: Four foldincreasedsuicidal death rate Among adolescent suicide attempters: Ten fold increased substance use 30 percent drank before attempt Among adolescent suicide completers: 70 percent used drugs frequently 50 percent had alcohol in blood 75 percent ft criteria fa; drug or alcohol use disoiders Among young psychiatric patients: Suicidal behavior and substance abuse correlate

Table 2.

2-176 208 M.Schuckft: Substance Use and Abuse

Several factors complicate the interpretation problems. Thus, an important association of these data. Pint, adolescence and early between substance abuse and suicidal be- adulthood are ages of maximal alcohol and havior is mediated through personality disor- drug use (1,72-74). Second, as discussed by ders and psychiatric illnesses that can be seen Weissman (75) people with multiple in adolescents and that sometimes run in problems are more likely to seek care than families. those with one problem alone. Therefore, substance users who also have depressive Personality Disorders symptoms and multiple life crises are the ones most hiely to be identified and to be a Over the years, attempts have been made to use clear-cut criteria to outline psychiatric part of studies. The apparent close associa- disorders for which good followup data are tion between substance use and suicidal behavior in treated groups might not available (3-5). Unfortunately, with one or completely generalize to substance users in two exceptions, the progress for personality labels has been less impressive than for the the general population. other categories of the DSM IIL In summary, there is much evidence connecting alcohol and drug use with suicide at- Personality disorders in general are likely to tempts and completions in both adults and be associated with suicidal behavior. Anec- dotally, character and behavior (or per- adolescents.This includes an increased prevalence of alcohol and drug use prior to sonality) disorders are among the most frequent diagnoses for soldiers or sailors with suicide attempts, a marked increase in risk for suicide attempts and completions among suicide attempts (17,19,29,76). Among psychiatric patients with histories of self- drug and alcohol abusers, an overrepresenta- tion of drug and alcohol abusers among harm, between 35 percent and 80 percent are suicide attempters, and a correlation be- noted to have some type cf personality disorder (21,77-79), although few studies used ob- tween suicide attempts and a history of drug jective criteria to identify the personality or alcohol abuse among psychiatricpatient& problem involved. MORE ADIRECT EVIDENCE It is probable that the specific label of the an- CONNECTING DRUGS AND tisocial personality disorder (ASPD) is close- ALCOHOL MTH SUICIDAL ly tied to suicidal behavior. The diagnostic framework for this problem has changed over BEHAVIOR the years, evolving from the psychodynamic Section I described how careful psychiatric concept presented by Cleckley (80) to a more diagnostic labels can give important clinical precise constellation of symptoms outlined information on a patient's probable prog- by Robins and colleagues and adopted (with nosis and treatment needs (1,3). To meet some modifications) by the DSM III these goals, however, it is necessary to distin- (5,81,82). As presently used in most studies, guish between primary disorders and those the label of ASPD connotes an individual illnesses that develop only after another with antisocial problems in multiple life areas preexisting psychiatric problem was estab- beginning prior to the age of 15 and continu- lished (secondary illness). This is especially ing into adulthood; these are problems that important for alcohol and drug abuse, be- cannot be explained solely by alcohol or drug cause the prognosis and rehabilitation needs use histories. Subjectively, theseindividuals can be quite different for primary and secon- are likely to be impulsive, have difficulty con- dary misusers (1,6,9,45). forming to the expectations of others and learning from mistakes, and show impair- This section highlights a number of primary ment in establishing long-term relationships. psychiatric disorders in which both suicidal behavior and substance abuse are common Approximately 80 percent of ASPD patients Report of the Secretary's Task Force on Youth Suicide have a history of alcohol abuse and as- depressivl disorders or, in the earlier studies, sociated serious problems (6,83,84), and *depressive neurosis.* An affective disorder these men and women carry an elevated risk is seen in 50 percent to 65 percent of adult for drug related pathology as well (82-86). suic:de attempters (19,20,53,100,103), as ASPD patients are also more likely than the well as in between 55 percent and 95 percent general population to attempt and complete of suicide attempting adolescents and suicide (1,35,68,83). One in four have his- children (18,22,93,104). The association be- tory of suicide attempts, and during a 5- to 6- tween suicide completion and depressive ill- year followup, 5 percent had died by suicide ness is also high for all ages, with about half (5184 of suicide completers noted retrosp *vety to have had major depressive disorders Two other personality disorders appear to be associated with both substance abuse and (64,101,105). Synthesizing this information, suicidal behavior.First is the DSM III Miles projects that the life-time risk for completed suicide among major affective disor- syndrome of somatization disorder that grew der patients is at least 15 percent (56). out of a concept of hysteria and Briquet's disease (3,5,88,89). Patients with this lisorder Another psychiatric illness with an elevated are usually women who at an early age risk for suicide attempt and completion ap- develop somatic complaints in multiple body pears to be schizophrenia. This diagnosis is system, including numerous conversion especially important in young people be- symptoms (neurological symptoms other cause the onset of the process is usually in the than pain with no known medical basis). teens or twenties (3). While definitive con- Severe mood swings are common and suicide clusions are jeopardized by the marked varia- attempts occur at a much higher rate than in tion in diagnostic criteria utilized in different the general population (3,89,90). Women studies, it appears that 12 percent to 15 per- with this disorder may have a 15 percent cent of adolescent or adult suicide at- prevalence of concomitant use of substances tempters have schizophrenia (20,21), while 3 for "recreation* or use outside of normal percent to 10 percent of suicide completers prescribing practices (89). Therefore, con- fulfill criteria for this disorder (64,101,105). sidering the relatively early age of onset, this may be a second example where substance This section is included both because each of misuse and suicidal behavior are tied these problems is a risk factor in suicidal behavior, and because secondary substance together through a separate primary illness. abuse during the course of these disorders Finally, an early onset personality disorder can exacerbate symptoms (1,6). As many as characterized by severe mood swings and an two-thirds of manics and one-third of severe inability to handle life stress has been depressives escalate their drinking while ill, a described as the *borderline personality* problem that could intensify depressed feel- (91,99).Subjectively, these patients sham ings and impulsiveness and might increase many characteristics with individuals with suicidal behavior (45).Similarly, ASPD or somatization disorder, demonstrat- schizophrenics who increase intake of brain ing impulsiveness and frequent mood swings. depressants or stimulants are likely to ex- In addition to an increased rate of suicide at- perience a worsening of their poor judgment tempts, these men and women are also rore and psychotic thinking (107,108). Finally, it likely than the general population to misae is possible that heavy drinking or drug use alcohol or other drugs (93,94). might exacerbate mood swings or intensify almost any major medical disorder (1,109). Major Psychiatric Disorders The most frequently reported diagnoses as- Substance Misuse, Suicidal sociated with suicidal behavior are affective Behavior and the Family illnesses, usually unipolar or bipolar major The previous subsections have documented

2-178 !2 M.Schucklt: Substance Use and Abuse

that few adolescents who attempt or com- sons and daughters of alcoholics adopted plete suicide are free of major psychopathol- away close to birth and raised without ogy.This finding impacts on two other knowledge of the biological parent's problem observations: the high prevalence of family (112-114). Similarly, family, twin and adop- instability and the high rate of psychiatrical- tion studies indicate that major depressive ly M relatives among suicidal young people. disorders (especially bipolar, manic depressive clisease) are also genetically influenced During the 1970s it was common for authors (115,116), and the data supporting the prob- to emphasize the observation of broken or able importance of genetic factors in chaotic homes among suicidal youth schizophrenia are equally impressive (116). (19,101,105). Interpersonal difficulties were often named as an immediate precipitant for For at least one of these disorders (al- suicide attempts in adolescents (19), and coholism), the greater the number of alcoholic relatives and the more severe their many of these young people were reported problem, the greater the probability of an to have observed suicidal behavior within earlier onset and more intense course for their immediate farnilies.Indeed, there is evidence that self-destructive acts run in those children who develop the disorder (117,118). families (111). For example, one evaluation of 243 psychiatric inpatients concluded that Thus, there is another level of association be- those with a family history of suicide were tween substance misuse, psychiatric disor- more likely than their coevals to have them- ders, and suicidal behavior in children and selves attempted suicide (50 percent vs 22 adolescents. Alcoholism or other illness in percent, p<.0001) (110). parents are relatively common findings among suicidal children and teenagers (110), Few clinicians doubt the interaction between it is possible that substance abuse or ,suicidal behavior, chaotic homes, and a fami- psychiatric disorders in these parents might ly history of suicide attempts or completions. The causal nature of the relationship is, have contributed to the increased rate of however, less obvious. Not only are most broken Immo; and chaotic childhood lifestyles for young people; these children may children raised by their parents, but they also have a high risk for early onset of the disor- get their genetic material from them. Con- sidering the close relationship between der itself (117,118). Therefore, the suicidal suicidal behavior and psychiatric illness behavior in these children may sometimes reflect their own early onset of illness as- described in the prior sections, if the major sociated with disordered mood or judgment. psychiatric disorders associated with suicide are themselves genetically influenced, then In summary, substance abuse or psychiatric part of the familial nature of suicidal be- disorders in parents could contribute to the havior could relate to genetic factors increas- suicidal behavior in young people in at least ing the predisposition towards similar illness two ways. First, some of the self-destructive in parents and children. problems in these children could have been influenced by the models set by the rearing Three of the disorders carrying the highest parents, as well as the child's reaction to the lifetime risk for suicide attempts and comple- anger and frustration engendered by the be- tions do each appear to be genetically influenced. The importance of biological, havior of the ill parent. Second, some of the suicidal behavior observed in the children genetic factors in alcoholism is supported by may reflect the inheritance of a predisposi- the familial nature of this disorder, the 60 tion towards the generically influenced ill- percent to 80 percent rate of concordance for ness itself, with concomitant suicidal risk alcoholism in identical twins of alcoholics associated with the disorder and not just the compared to a risk of approximately 30 per- specific childhood environment. cent for the fraternal twins, as well as the four-fold increased risk for alcoholism in

411 2-179 Report of the Secretarys Task Force on Youth Suicide

SUMMARY AND ly than the general population to observe CONCLUSIONS suicidal behavior in their parents, more likely to suffer chaotic homes during childhood, This paper has presented data and also at- and themselves have a genetically increased tempted to stimulate thought and discussion. risk for both suicidal behavior and substance The emphasis has been on the potential con- misuse. tribution of alcohol and drugs to suktidal be- From this review it is appropriate to conclude havior in children and teenagers. that, through both direct and indirect There is an important relationship between mechanisms, intake and abuse of substances substance use or misuse and violent behavior is a potentially important risk factor in in adults. Controlled substances and/or al- suicidal behavior early in life.The relation- cohol are frequently used as the means of at- ships between these substances and self- tempting self-harm (especially among harm is rather complex, and careful data younger women), and alcohol is often taken collection will be required if we are to under- as a prelude to the suicidal act, thus con- stand more about this clinically relevant as- tributing to impaired judgment and impul- sociation. sivity_ Also, the lifetime risk for completed It is possible to speculate about some of the suicide appears to be 15 percent among al- clinical implications of the data reviewed in coholics and about 10 percent for drug this paper. First, through numerous direct abusers. and indirect mechanisms, adolescent alcohol It is not surprising, therefore, that the as- and drug abusers have an elevated risk for sociation between substance intake or abuse suicide attempts and completions. Second, and suicidal behavior is also observed in whatever the mechanism, children of al- adolescents. Whether studied in the general coholics and of patients with depressive or population or among groups in treatment, schizophrenic disorders may themselves be at substance abusing young people have a sig- elevated risk for suicide attempts and com- nificantly increased rate of self-inflicted pletions.The same may be true far children death; youth identified because of their of drug abusers, although less data is avail- suicidal behavior frequently use and abuse able to substantiate that conclusion.In nonprescription drugs and alcohol; and there working with these families the potential is a close relationship between substance dangers for self-harm in children should be misuse patterns and the number and severity recognized, A third clinical implication of of suicide attempts. the data comes from the recognition that al- Suicidal behavior early in life is also as- most all substances of abuse are likely to exacerbate preeidstinl emotional or psychiatric sociated with substance problems in more in- disturbances. Therefore, heavy intake of any direct ways. Diagnoses such as the ASPD of these controlled substances or alcohol may and "borderline personality' carry high risks be important risk factors increasing the for both self-harm and substance misuse. It suicidal propensity of young people in crises, is also probable that heavy drinking or drug as well as those with major psychiatric disor- use during major depressive episodes or in ders or personality disturbance&Efforts the midst of other psychiatric problems can aimed at minimizing the risk for suicide exacerbate problems and might contribute to should include educating young people and a suicide attempt or completion in these high their families about the need to refrain from risk individuals. A third indirect association intake of all substances of abuse during times bctween substance abuse and self-harm re- of mood swings or anger. Fmally, physicians lates to the probable importance of genetic must learn to be careful in prescribing factors in the development of alcoholism and psychotropic agents or drugs of potential early onset psychiatric disorders. Thus, for abuse to teenagers with emotional or example, children of alcoholics are more like-

2-180 212 M.Schuckft: Substance Use and Abuse psychiatric disturbances because there is 11. LuldanowIce etAttempted suicide In children. evidence that in many instances the suicidal Acta Psychiat. &and. 1968 4441545. 12. Paykel ES, Myers JK, Undenthal JJ, Tanner J: overdoses involve prescribed drugs (119). Suicidal ferv in the general popriation; A prevalence study. Brit. J. Psydde. 1974; 124:4804. The research implications of this review are 13. POW AR Roy M.Wison, Gat Self-polsoning and also apparent. Studies of suicidal behavior in alcohol. The Lancet 10714 21090402. 14. Wader L, Weissman MM, Keel SV:Suicide at- adolescents should rigorously document tempts 1970-75 Updating a United Stets etudy and com- drug and alcohol use patterns in the recent parisons with international trends. Brit L Psyl3sL 1979: past, evaluate individuals for major primary 1321804 15. Jarvis etc, Femme RO, Johnson FG, Whitehead and secondary psychiatric disorders including eax and =home in selfury. J. Mk Soo. those related to substance abuse, and docu- Bohm 1976; 17: 16. Men increase In suicide attempts by drug ment the prtmence of these problems in first- ingestion. Pah. Gen. Psychlat. 1977; 3411654. degree relatives.Investigators must also 17. WAsman elk The epidemiology of suicide at- take care to not assume that an association tempti. 1980-1971. Noh. Gen. Psyclwat 1974; 30:73748. between two factors (e.g., a suicide attempt 18. Robbim OR, Masi we Depressive syirums and suicidal behavior in adolescents. kn. J. Psychlst. 1985; in a child and alcoholism in a parent) proves 142:58842. that it is the disturbance in the home environ- 19. Wlitte HC SeligleonIng In adolescents. Brit J. ment that caused the attempt. While the en- Psychiat 1974; 124:2 20. Geiser SH, Masterson JF: Suicide in adolescents. vironmental stressors probably contribute An. J. Prichist 1957; 11&4004. greatly, other important avenues of influence 21. Tootan JM:Suicide and suicidal attempts in children and adolescents. Ant. J. Psychiat. 1982 118714,- must also be considered. For example, it is 24. possible that children of alcoholics may 22.Garfinkel BD, Freese A. Hood J: Suicide attem_ 1:10 themselves have inherited (or acquired In children and adolescents. Am. J. Psychlat 1102; through in utero damage) problems of impul- 139:125741. 23. Flygnested TitProspective study of social and sivity, hyperactivity, or a propensity to misuse gryyceZattrio aspects In self-poisoned patients. Acta substances with subsequent mood swings, Scan. 11:482; 88:13Oft 24. Walkman N, Schreiber ht Perasulaide in young anger, and frustration in their own lives. Edinburgh women, 1988-75. Psychol. Med. 1979; 9:409. 79. 25. Holding D, Duffy JC, Kreitman N: Paruulcide In Edin seven-year review, 1968-74. REFERENCES Da J. Psychlat 19Th *53443. 1. Schuckit Mk. Ekug and alcohol abuse: A clinical 26. Shatil M, CarrIgan S. VAtitlinghill JR. Derrick k. knideunt:24= and treatment. 2rul ed. New York: Psychological autopsy of_ kehrd suicide in cetildren Coveration, 1984. and adolescents. Am. J. Pft 1088; 14Z10614. 2. Schuokit MA:Alcohol and Alcoholism. In: 27. Ovenstone 64 : Spectrum of suicidal behaviors In Petersdorf, R.G. Adams, RD., Braunwald, E. eds., Edinburgh. Wit J. Prevent Soc. Med. 197* 2727-35. Herrisoprervites of Internal meritane, llth et New 2& BainvktFt Teenage suicides. U.S. Medicine 1973; York: Co, In press. Oct tt 3. Goodwin OW, Gun SitPsychiatric diaLthosis. atHeiberg A, Gerfein AD: Predicting suicide gestures New York: Oxford University Prom 1964. In a Need recruit population. Military Medicine 1976; 4. Spitzer 11., Endicott S, Mins E Research diag- 141:5274. nostic ~a. Pah Gen. Psychiat 1978; 36:77342. 30. Rah CL, Young 0, Fowler RC: San Diego suicide 5. Diagnostic and Statistical Manual of Mental Diem- study: I. Young vs old cases Arch. Gen. Per:Nat. (in dem, 3rd et Washington, D.C.; Antedcan Psychiatric As- press). sociation. 1980. 31.Statistical Abstrect of the United States: 1084 8. Schuokit Mk. Neohallem and other (104th Witten). Washington. D.C.. U.S. Bureau of the Cen- disorders. Hospital and CommunityPsychas4h114:1°3-. sus, 1983; 79-86. 34:1022-7. 32. Solomon MI, Helton CP: Suicide ar2d agsi In Alber- 7. Gun Sa The need for tough mindedness in ta. Canada, 1951 to 107T. keit. Gen. PsyNnattaw psychiatric thinIdng. So. Med. Journal 1970; 6a1182-71. 37:511-3. 8. Schmidt MA, Wnokur G: A short-term follow-up of 33. GE: Oinicel IdentifIcetion of suicidal risk. %omen alcoholics. Mumma of the Nervous System 197t Arch. Gan. 1974 273584. 33:872-78. 34. RdilIus PA: Deaths among child and adolescent 9. Schmidt Mk 11* clinical Implications ofp=z patients. k:ta PayrWat. Spend. 1984; 70X119- IVA= among alcoholics. Itrch. Gen. 35. PC: Violent deaths wt the young: 10. VaNlant CI: Natural history of male alcoholism: Recent in suicide, homicide, and dsmbi. Pm. J. alcoholism the cart bkori=thy? Presented at the - Psychiat.1979; 138:11447. American Annual Meeting, Toron- 36. Curry 8:Suicides el young Indians called to, Ontario, M1y5, 1982. sPidemlo. Um. Part 1, 1985 Oa 12:4.

213 BEST COPY AVAILABLE 2-181 I L

Report of the Secretary's Task Force on Youth Suicide

37. Meares R, Kraljuhin C, Bonfield J:Ade/49cent 84. Robins E The final months: A study of the lives of suicide. Aust. Fem. Physician 1983; 18144. 134 persons who committed suicide. New Eng. J. Med. 38. Weiss N& Recent trends In violent deaths &nom 1982 306:111 7. Imliv adults kt the thilled Stem. Mt. J. Epidemio. 1911k 66. Murphy GE, kmstrong JW, Hormel* SL, Fischer 103:40= JR Ciendenin WW: Suicide and alcoholism. Ach. Gen. 39. Lester 0: MON and suicide and homicide. J. Peyotdat. 19* 36:8541. Stud. No. 1960 41:12204. 68. Crawshaw R. Bruce JA, Biker PL, Greenbaum M, Lindemann JE, Schmidt DE:An epidemic of suicide 40. Wolfe M:Patterns In criminal homicide. among physicians on probation, JAM.A. 1904 243:1915- PhliodsipNs tWvsrW of Penney/oft Pees, 1958. 7. 41. Schmidt_ MA Gunderson EKE: Suicide in naval 67. Borg SE, Stahl M: Prediction of suicide. A prospec- service. kn. J. Psychlat. 1974; 131:132841 tive etticry of suicides and controls among4rychiatric 42. Weill CV:Changiing patterns of methaqualone patients. Mt' Psychiat. Scand. 1964 85121 abuse. A survey of 248 MOW& JAMA. loft 4.W4. 68. Morrison ,NtSuicide in a mdtiatric practice 43. Langevin R, Paltioh 0, Orchard B, Handy I., 'Meson population. J. Clip. I 1902 43: -34842. A The role of alcohol, drugs, suicide attempts end situational strains in homicide committed by offenders seen for 60. kk:Kenry PC, nettles' CL, Kelley C: The role of mdlistric assessment. Acentrolled Ardy. Acta Psychlat. drugs in adolescent suicide attempts. Suicide Lite Threat Scald. liet 88.22042. Bohm. la& 13:168-76. 44. Schmidt Mk The ICftly of psychotic symptoms 70. RD: Alcohol in association with suicide in alcoholic& J. Ohl Psychiat 10ft 43:534. and Mem suicide in young women. Med. J. Aust 1981; 221 7. 4& Schuoldt Mk Alcoholism and affective disorder: Genetic and clinical knplications. Am. J. Peychlat. (in 71. Benson G. Holmberg MB: Drug-misted modal/ press). in young people. Acta Psychlat. &end. 1984; 70:52544. 46. 1Nhitiock Fk Suicide In Brisbane, 1958-1973. The 72. Ryser PE: Students and drug abuse, 1974 and drug-death epidemic. Med. J. Past. 1975; 14:737-43. 1980. J. School Health 1983; 53:4354. 73. Schuckit Mk Overview:Epidemiology of N. 47. Donnell HM: Attempted suicide in schoolchildren. coholism. hi Schuoldt, !AA., ed. Aloahoi worm and Mod. J. Austr. 1974 11105410. problems. New Brunswick, N. Rattles University Press, 48. Haider I: Suicidal attempts in children and adoles- 1965:1-42, (Series in psychosocial epidemiology, VW. 5). cents. Brit J. Psychlat. 1968; 1 4:11334. 74. Smart RG, Goodstadt MS, P.diaf EM, Sheppard MA, 49. Ghodse AH: Deliberate seff_-poisoning: a sWn Chan GC: Trends in the prevalence of alcohol and other London casualty departments. Or. Med. J. 1077; &pa use among Ontario studenta 1977-1993. Canadian 50. Abel El" Zeidenberg P: liaa, alcohol and violent J. Public Rib. 1 78.15742. death: A postmortem study. J. Stud. Alcohol 1985; 75. Weissman MM: Alcoholism and depression: 46:22841. !=*mites/ Presented at the seventh annual N- Symposia?: "Diagnosis and Treatment Cur- 51. Haberman PW, Baden MM: Alcoholism and rent by the Department of violent death. Oust. J. Stud. Ala 1974; 35:22141. PsychiprtryiellgentalCarnbi="11H4pital.Boston Pwk Plaza, 52. Mayfield 0, Montgomery * Alcoholism, atoohol March 3. 1984. intoxication and suicide attempts. Amh. Gen. PapahiaL Schuckit MA, Gunderson EKE The clinical charac- 1972; 27:349-53. teristics of personality subtypes in Navel service. J. Clin. 53. Woodruff RA Clayton PJ, Gun SB: Suicide at- Psychiat 1979; 40:1754. tempts end psychiatric diagnosis. Dis. Nem. Sys. 1972; 77. Yowler PG. Gibbs JJ, Becker Hk On the com- 33:817-21. munication of suicide ideas. Arch. Gen. Psychlat 1960; 54. Thorminsson Ak among men alcoholics 3:61241. in losland, 1961-74. J.StudPA?no.971; 40:70448. 78. Leese SM:Suicide behavior In twenty adoles- 55, Berglund M: Suicide in alcoholism. Arch. Gen. cents, Brit J. Psychiat 1969; 115:47940. Psychlat 41188491, 1904. 79. Diming LH, Watson J, May PA Boat Jciirmohileat 56. Wes CP: Conditions predisposing lo suicide: A cent suicide at an Indian reservation. Am. J. review. J.Nov. Mont. Dis. 1977; 104:231-48. 1974; 44:434. 57. Moilhoff 0, Schmidt a Deets resulting from 80. Cieckley H: The mask of sanity. St. Louis, C.V. drugs of abuse. Forensic Sol. 1970; 73140. Mosby, 1955. 58. Kalant H, Kalant 0.1: Death in amphetamine Mrs: 16.Robins LN: Deviant children grown up. Baltimore: Causes and rates. Can. Med. Assoc. J. 1975; 8:299404. Mame and Wiliam 03. 1968. 59. Lundberg GD, Gardott JC, Reynolds PC, 82. Robins IN: Sturdy childhood predictors of adult RH, Shaw RF: Cocaine-related death. J.Forens7roaV. antisocial behavior:Replications from longitudinal 1977; 224024. studies. Psychological Medicine 1978; 8:611-22. 60. Wateon JM: Glue sniffing. Two case reports. Int. J. Schuckit Mk Alcoholism and soolopaihr Diag- Addict. 1979; 14:45144. nostic confusien. Quart J. Stud et 1973; 34:10744. 61. Riddkdr_ L Luke JI: Alcohol-associated deaths in $e. Virkkurteo M:Alcoholism and antisocial per- the District of Columbla-a postmortem study. J. Forensic sonality. Acta PsychiaL Soand. 19* 59:493001. Sok 19* 23:493-50a 86. Fowler RD Uskow 0, Tenn* VL; NOWC Illness 52. Novici! LF, lismrnlinger E: A study of 128 deaths and alcoholism. Presented at the National Council on Al- in 1,4sw York Oity ooweellonal facilities (1971-1978): Sri- coholism convention. Washington, D.C., May 1978. nth:tenons for prisoner health care. MM. Care 1978; SI Cadoret R, VadITtif RN, Troughton E: Clinical dif- 10:749-758. ferenoss between antisocial and primary alcoholics. 63. Beskow J: Suicide in mental disorder in Swedish Presented at the National Counc6 on Alcoholism annual men. Acta Psychist. Scand. 1979; 227:131-138. meeting. Washington, D.C., May 1982. 87. Maddocks PD: A five year followirp of untreated psychopaths. Brit. J. Psychic. 1970; 116;5114

2-182 214 BEST COPY AVAILABLE M.Schuckft Substance Use and Abuse

$8. Bibb RC. Ciao SB: 1=411.1 a psychiatric hospi- 112. Sahucklt Mk Genetics and the risk for alcoholism. tal. Pm. J. Psychist.1972; I JAMA, 1985; 254: 26144. 69. tee* CE, Helier J, ClorOrger CR, Croughen 113. Goodwin OW: Alcoholism and genetics. kcir. Whitman BY: Psychiatric diagnostic predispoeitions to al- Gen. Psychist. 1985; 421714. coholism. Compmhensive Psychiatry 1982; 2345141. 114. Sohuokft MA: Studies of populations el high risk 90. Gun SEI, Woodruff 124: Misled* and for alcoholism. Psychiatric Developments 1988; 23143. antisocial behavior. Arn. J. Proate 1971: 12195740. 118 Gmhon ES, Surma, WE, IlIcklut1 JF If..an Ear- 91. Gm:Wm JO, Kolb JE: Disoriminating features &Krogh M, DeBauche Bk ITIO Inhodtano. of Emmy, dis- of badman* patients. Am. J. Psychist 1978; 1351924 ordenc A review of data and of hypotheses. Behavior 92. Gunderson JG, Molt Git The Werke* between Genetics 1975; 8227481. borderline pemonality disorder and affective disorder. Am. 118. Schuckft MA:Trait and state) markers of a 4. Peyohist. 1985; 14227748. to .1n:4ctuas,a,Judd, Eit,itrovP"rne,P., eds. foundations of cantos, 93. Crumley FE: Adolescent suicide attempts and bor- psychiatry, Vol. 3. J.P. Upping:4o1l 1985: 1- derline pemonalit/4491sorder: Mica! Features. South. 19. Med. J. 1981; 74: 117. Schuckit MA: Relationship between the °ours, of 94. Aldskal SF Davis GC, Purantien VR. primary alcoholism In men and family history. 4. Stud.Ate. =lPA, .a4ciective in 1N4:45:14. c:fa:noun.J. . Psychlat1985; fk41 1* Frances RJ, Thrwm (I. Stocky 5: Studies / femalel 95. Frances A. Clarkin S, Gilmore IA Hist SW, Brown and nontemitial alcoholism. kch. Gen. Psychlat. 1980: ft Rsilabllityof criteria* borderline persordisorder: 37:5644. Aboorrourarison of DSM-111 and the diagnostic ftuview for patients. Mi. J. Psychist 1984; 141:10h04. itr.Prescott IF, Highley MS: 01148 prescribed for self poisoners. Br. Mod. J. 19ft 1:18334. 98. Pope He, Jonas JM, Hudson 41, Cohen OM, Gunderson JG: The validity of WWI borderline i;rer- milky disorder. Arch. Gen.PsOM 1953; 40:2340. 97. Schulte SC, Goldberg SC: Borderline personality disorder: New Findings on Pharmacotherapy. Psychopharmacology &IN6n 1964; 2055440. 98. Maks, HS, Iferevenlan BI, Dads GC, Pang 0, Lamm! Pt TM nosologic status of borderline pwpzerralf anderlysomnogmphic study. Am. J. 1965; 14214. 99. PAcklan.3 M, Lerner H, Barbour C: Assessment of bordedine stivolpsomelology in hospitalized adolescents. J. Ant Acad. Child Psychlat WM: 231155414. 100. 1.1min P, Gibbons A.: Psychiatric diagnosis in self- poisoning patients. Psycho!. Med. 1979; 6:501-7. 101. Barraciough B, Bunch J, Nelson 13: A hundred cases of suicide. Brit J. Reptile'. 1974:125:355-73 102. Porot M, Clouded A. Collett kt Suicidal behavior of adolescents. Psychiatric de l'Enfant 1968; 11:31749. 103. Reeves JC, Large FIG, Harriman It Paresuidde and depression: A comparison of clinical and questionnaire diagram Aust. NZ J. Psychlat 1965; wan 104. Friedman RC, Clarkin J. Corn R DSM4 and affective pathology in hospitalised adolescents. J. New. Mont. Os. 1982 170:511-21. 106 Dorpat Tt., Jackson JK, Roby HS: Broken homes and attempted and completed suicides. Arch. Gen. PsycNat 1955; 12213218. WEL Weinberg 5:Suicidal intent kr adolescence: A hypothesis about the role of physical %rest Journal of Pediatrics 1970; 77:57949 107. Schmidt MA, Wnokur 0: Alcoholic halludnosis and schizZel A negative study. Ildt. J. Psychlat 1971; 119: 108 .Segal DS, Schuckit Mk Animal models of stimulant induced psychosis. In:Cress* I, (ad). Stimulants:neutWwnical, behaviorst, and clinical perspectives. New York; Raven Press, IWO 13149 109. Saghir MT. Robins E. Walbran 0, Girr Homossxusllty Ill Psychiatric disorders and amuity In the male hommexual. Amer. 4. Psychiat 1970; 128:1079- aa. 110. Roy k Family history of suicide. Arch. Gen. Psychist 1983; 40:4714. 111. Murphy GE, Wetzel RD: Family history of suicidal behavior among suicide attempters. J. Nem. Ment. Cis. 1902 17011540.

BEST COPY AVAILABLE

4.1 2-183 1144" 215 METHODS AS A RISK FACTOR IN YOUTH SUICIDE

J. William Warden, Ph.D., Assistant Professor of Psychiany, Harvard Medical Schoo4 Boston, Massachusetts

In this paper we will consider methods of places or drowning, whereas relatively more suicide as a risk factor. Although other risk white males use firearms or die utilizing carb- factors already discussed in this conference on monoxide. White females are more like- may be more important than methods as ly to bang themselves than black females. precipitants and precursors of suicidal be- Overall, a higher proportion of females to havior, wne is more important than methods males poison themselva irrespective of race. when it wmes to the risk of death. Some (Fredrick, 1984). It is interesting to note that methods lead to almost certain death while suicide statistics in the United States are not other methods are more uncertain as to out- kept for children under 10 years of age. This come and they portend more reversibilityand implies that suicide is not seen as an option rescuability. We have seen a recent rise in for young children by those collecting such youth suicide, especially with regards to com- information. pleted suicide. Is this increase in completed Paulson at UCLA disputes this fact. Study- suicide due to an overall increase in suicidal ing children who were seen at the UCLA behavior of is it due to the fact that youths Neuropsychiatric Institute from 1970 to who engage in suicidal behavior are using 1974, Paulson found 34 children ages 4 to 12 more lethal means of self-destruction? A who demonstrated suicidal behaviormostly better understanding of methods may suicide threats and attempts. The mean age provide us with an answer to this question, as of his sample was &2 years. Suicidal behavior well as providing us further insights into the in this age group involved more ;Glf-abuse dynamics of youth suicide and possibly offer- and bodily mutilation than self-poisoning. ing some clues for prevention. Examples of self-abuse were cutting, stab- How do adolescents and young adults kill bing, burning, and jumping either from high themselves? Overall, those in the 10 to 24 places or in front of moving vehicles. age range most frequently kill themselves by Children exhibiting the most mutilating as- firearms and explosives, with guns being the saults on themselves frequently came from most Imminent means. The second most highly disorganized familii% and were also frequently used method of suicide is by hang- the children with the highest ideational ing, strangulation, or suffocation.Self- violence. It is interesting to note that Paul- poisoning by ingesting solid and liquid son did not fmd significant gender differen- substances is the third most frequent method ces in ideational violence among his young of self-destruction.(Holinger, 1978). Al- subjects. (Paulson et al., 1978). though there are few major differences in Pfeffer also studied suicidal behavior among choice of method between black and white younger children. In investigating children youths, there are some distinctives. Propor- ages 6 to 12 who were inpatients in the child tionately more black males than white males psychiatry unit of the Bronx Municipal hang themselves or die by jumping from high Hospital and who demonstrated some type of

2-184 t.. 216 J.W.Worden: Methods as a Risk Factor in Youth Suicide suicidal behavior, she found that jumpingwas sample and in another country. the most frequent choice of these young at- There is general agreement that incidents of tempters. Out of the 42 subjects in her study, self-poisoning become more frequent as jumping was chosen by 38 percent of the children increase in ar. There is less agree- children. Of those remaining, 25 percent chose self-poisoning, 19 percent burned ment as to what substances they ingest. Haw- ton compared adolescent self-poisoners to themselves, 13 percent cut themselves, and 6 adults and found that adolescents more fre- percent ran into oncoming traffic. (Pfeffer quently ingested non-opiate analgesics cad et al., 1979). In a subsequent study of 65 les frequently took psychotropic drugs. The children admitted to the child psychiatry unit use of akzihol as an adjunct factor in an over- of New York Hospital Westchester she also dose was found less frequently among adoles- found jumping from high places to be the cents than among adults. This was especially most frequent method and this was found in true in the younger groups of children. Most 2$ percent of the cases. (Pfeffer et al., 1982). young children tended to ingest drufound Although jumping is a frequent method for around the house rather than drugs specifi- these very young attempters, self-poisoning cally prescribed for them. (Hawton, 1982). should not be overlooked.McIntire and Hawton also found that the ingestion of Anglediscoveredaninteresting psychntropic drugs increased as the females phenomenon when investigating rept 'its in his b. udy became older, but the same was from 50 poison control centers in the United not true for the males studied.Fredrick States and the United Kingdom. In the 6 to (1984) reported that barbiturates and tran- 10 age range, self-poisoning was more fre- quilizers were the most frequently ingested quently found among males (63%) than substances for completed suicides among 15 females (37%). This is the oppsite of what to 24 year okls in 1979. etadney in looldng they found in their older group of children, only at female self-poisoners ages 18 to 30 11 to 18 where self-poisoning by females far found that anti.depressant medications were exceeded that by males. This same gender often ingested as overdoses. This viss espe- differences can be also found among adult cially true for attempts judged to be highly self-poisoners where the number of females serious both from the standpoint of medical exceeds males. McIntire and Angle offered lethality and the ratings from the Beck Intent no explanation for the difference they found, to Die Scale which was given to these 109 but they did conclude from their study that women in the study. (Goldney, 1981). One self-poisoning in a child over the age of 6 is might conjecture that younger children are rarely accidental.(McIntire & Angle, more likely to ingest substances found 1971b).It should also be noted, however, around the home. As children row older that completed suicide for these very young they may be in psychiatric treatment and chiL ma is rare. receiving medication which is available to theia if they decide to overdose. Hawton & Goldacre studied cases of self- poisoning in a population of young people The reason- eiriven for choice of method are ages 12 to 20 years who were admitted to always of h est. Some suirlde behavior is hospitals in the Oxford area of "angland. well thought out and the methods selected They also noted that self-poisoning was well in advance. Othek suicides are impulsive higher for females than for males at all of and the person may turn to the nearest per- these ages. (Hawton & Goldacre, 1982). It ceived lethal substance at hand. To the ex- would have been useful if the investigators tent that the choice is rational, there was an had included the younger children from ages interesting study conducted by Marts with 6 to 12 in their study to see if the reverse male approximately 700 college students. The to female phenomenon that McIntire found mean age of his respondents was 193 years. could have been replicated in such a large He asked the students to rank order nine

217 2-185 Repon of the Secretary's Task Fort:e on Youth Suicide

means of self-destruction according to their with stricter gun-control laws had lower rates acceptability and to suggest which methods of suicide by firearms but higher rates of they thought were used moat often by males, suicide by other means. Howmf, in an ear- by females, and why. Both males and females Her report, Lester and Murell (1980) found in his study rauked self-poisonink, as the most that the toul suicide rate was lower in Stew acceptable. Firearms were ranked higher by with aricter gun-control laws. Thus it is pos- men than by women. Women were more sible that when firearms are less available, hiely than men to rite "lack of pains as the persons turn to other methods which may reason for their choice. In citing additional leave more time for intervention. reasons, "Ervailability" and "knowledgeability" were associated with self-poisoning for Marks and Abernathy (1974), in trying to explain both gender am! individual differences females; with firearms for males. Similarly, in choice of method, have posited a sociocul- 'efficiency* was related to self poisoning for tural perspective of differential socialization females and to firearms for males. (Marks, which takes into account not only availability 1977). but several other factors which also influence Although self-poisoning is the frequent such a choice. Marks and Abernathy choice of the adolescent suicide attempter, it respond negatively to what they term the .anks thin in the eau= of suicidal deaths in sipt-ychological perspectivea perspective the 10 to 24 year old population. Most young which suggests that the difference in choice people who succeed in taking their own lives of method solely reflects a difference in in- do so with guns and explosives. For young tent to die. This perspe-...zive implies that the people ages 15 to 24 the rate of suicide by higher incidence of handgun use in male firearms increased 97.1 percent from 1966 to suickies indicates that males are more intent 1975 while the rate of suicide by other means on killing themselves. increased 72.4 percent for the same period. Marks and Abernathy argue against this The significantly large increase in death by psychological perspective on several firearms may reflect the 14.5 percent in- grounds. For instance, if it were correct, then crease in the availability of legal handguns we would expect to find different methods during that same time period (Seiden & emp:oyal by female attempters than female Freitas, 1980, Boyd, 1983). completers. However, the use of poisons is To what extent, then, is availability a deozr- high in both groups.Furthermore, one mining factor in the choice of a method of would expect those areas of the country in seif-destruction?There has been con- which the more deadly methods, e.g. siderable discussion as to the availability of a firearms, are preferred to also have the method, both as a determinant for choice and highest suicide rates. However, such is not as a possible way to reduce suicidal behavior, the ewe. such as gun control. Marks and Abernathy Marks and Abernathy have identified at least (1974) minimize physical availability as a three factors beyond availability which may determinant. The most available methods influence the choice of a suicide method. are not those most frequently employed. For The first of these is the socbcultural accept- example, rope with which to hang oneself is ability of the method. Certain methods have more readily available than firearms, yet greater or lesser acceptability within the in- firearms are more frequently used. Obvious- Lernalized social and cultural norms of the in- ly, however, availability is a necessary dividual, and this will be reflected in the precondition in the choice of a method. choice of a method. Examples of this may be There is evidence that when a certain method reflected in the distinctiyes mentioned ear- becomes unavailable some persons wishing lier between young black and white males to suicide will switch to another means. and between young black and white females. Lester and Murell (1982) found mat States The second factor is the person's knowledge

2-186 218 J.W.Worden: Methods as a Risk Factorin Youth Suicide of the methods. A person is more hiely to Min groups of women who prefer firearms to employ a method with which he or she is other methods are any more %berated* than knowledgeable and familiar. The third fac- those women who prefer poison or some tor is personal or %vial accessibility. For ex- other means of selfdestruction. Liberation ample, heroin may be physically available is probably a poor choice of terms. Changing from a street supplier but socially inacces- role definitions does not necessarily imply sible because of the person's social standing liberation. Location within the social struc- as a police officer. In this example heroin ture is a better term than liberation and fits would be a less likely method for an overdose. within the socioctdtural perspective. Chan- ges in traditional role expectations are per- Research by McIntosh & Santos (1982) also vasive in our society and its culture and this supports the idea that selection of a method needs to be taken into consideration when is dependent upon a constellation of explaining gender changes in patterns oi sociocultural factors. They found that no suicidal behavior. generalization could accurately describe the range of suicidal behavior displayed by There Are important implications to be various sex, racial, or ethnic groups.Al- drawl from the sociocultural perspective though they did not specifically study youth with regard to suicide methods and youth. suicide or break down their sample by age, The most important is the need to determine their point is most applicable to our discus- what sociocultural factors differentiate sions of youth suicide in this cenferencs. We youths and adolescents from the adult must be careful not to draw overgeneralized population. Are different methods more or conclusions about youth suicide methods less available to youths than to adults? Are without more closely examining the various certain methods more culturally acceptable socic.ratural influences on these young men to youths? How do these sociocultural fac- and wmen and how these influences might tors vary by age within the youth population? differ by age, sex, social position, cultural There is no specific research in this area background, geographical location and per- though it is clearly needed. We recommend sonal histoiy. Much more research is needed research which will clearly differentiate bc- tr: understand the selection of suicide tween adults and youth and carefully ex- methods by youth. amine the influence of sex, age, geography, cultural values, social status, intent to die, One important factor influencing gender familiarity and availability of methods, and preferences for suicide methods may be *an- how these factors relate to the choice of self- (Marks & Aber- ticipatory socialization? destructive methods in these populations. nathy 1974). Traditionally boys have been exposed to and encouraged to participate in Some suicidal methods obviously cagy more violent games and activities to a greater de- risk of death than others. Because of this, gree than their female counterpart& An ex- clinicians are apt to urnominal classifica- ample would be the game ci cops and tions when they discuss the relative lethality robbers. This may account, in part, for the of a suicidal act.Words like 'gesture,* greater incidence of firearm use by males. *moderate lethality,* and *serious suicide at- One could speculate that changing role tempt* are still used when talking about a definitions for women may account for the person's suicidal behavior. However there increase in the use of firearms in female are many suicide attempts of intermediate suicide&It is interesting to note that the lethality which cannot be dichotomized easi- majority of young women in the U.S. Army ly into *gestures* and *serious' attempts. This who suicide (73%) do so predominantly with is especially true in the area of self-poison- firearms (Datel & Jones, 1982). Taylor & ing. Wicks (1980), while supporting the sociocul- In 1972, Weisman and Worden developed a tural perspective, find r.a evidence that cer- scale known as the Risk-Rescue Rating Scale

2-187 Report of this Secretary's Task Force on Youth Suicide in an attempt to describe and quantify various social sub-cultures and geographic lethality in suicide attempts. (Weisman it locations. Being able to rate methods within Worden, 1972). This scale is based on the as- their context and to be able to scale the ob- sumption that the probability of death is sub- servations would clearly aid in our attempt to stantially influenced by what a person does to understand youth suicide across the age himself and the context in which he does it. cohorts. Any suicide attempt entails a calculated risk. Second, such a rating scale also helps one to But because any attempt must also take place assess relative lethality for a person who has in a psychosocial context or within a specific made mukiple attempts. It is well known that set of circumstances, survival may depend those who have made previous attempts are upon the resources for rescue as well as upon at higher risk for completed suicide. Rating the specific form of the attempt. Jumping off each attempt with the same scale enables one a high bridge into the river beneath has to see whether lethality is ascending, decend- similar risk to physical damage whether it is ing, remaining the same, or forming some dom.; at 3 p.m. or 3 aan. However, rescue fac- other pattern. In one study Worden found tors are different in the early morning hours that one could predict the lethal level of sub- when it is dark and there are fewer people sequent attempts in adults by rating previous around. The Risk-Rescue Rating Scale as- attempts in combination with various other sesses five factors of risk and five factors of demographic information.(Worden & rescue and enables the clinician or re- Sterling-Smith, 1973). searcher to quantify the lethality of implementation in any suicide attempt. A third use for a lethality rating scale is to better understand the complex issue of intent to There are other scales similar to the Risk- die. Measurg4 g intent is often very difficult. Rescue that look at the relative seriousness Patients do not always give an accurate ac- of methods. Recently, Smith, Conroy, and count of their intent to die from an attempt. Ehler at the Menninger Foundation created There are both CODSCi0133 and unconscious a nine-point interval scale for assessing the factors that lead to distortion. Patients may relative lethality of a suicide attempt. (Smith say one thing when taken to the emergency et al., 1984). This is a well developed scale room and have a totally different story a day and their updated toxicity chart is an im- later. Also, ambivalence is present in most prover -ent over the older one developed by people trying to hurt themselves. On the ex- Sterling-Smith end used by Weisman and treme ends of the spectrum intent to die is Worden. rather easy to discern. The young girl who in- The ability to quantify a suicidal act by look- gests 25 aspirin and immediately tells her ing at methods in context is valuable from mother is clearly in the low lethal range. The several standpoints. Fast, you can compare young man who takes a gun into the remote one group of suicide attempters with woods and shoots himself in the head is ob- another. We know for example that young viously in the high lethal range. These are females more frequently use self-poisoning obvious. But there br a whole range of be- than young men, but young men do poison haviors that fall into the middle ranges of themselves. Even though there is las self- lethality and using a rating scale that looks at poisoning among males, is the lethal level of methods in context would be very useful in their attempts higher than for females? future studies of youth suicide to better un- Using a scale hie the Risk-Rescue Rating derstand the issues of intent especially in could help answer this. We presently kno w those falling inlo the middle ranges of the types of methods used by youth in dif- lethality. ferent age categorks and the percentage of In any discussion of youth suicide methods use for various methods varies widely for dif- one should not overlook automobileacci- ferent age groups possibly for youth in dents as suicide equivalents. In the 1970s we

2188 220 J.W.Worden: Methods as a Risk Factor In Youth Suicide did a study of automobile fatalitia in the changing. This would have implications greater Boston area under a grant from the both for prevention and for prediction of Department of Transportation. One part of future trends. the study investigated the human factors as- We also believe that methods should be sociated with there accidents. A number of studied in context. If those who are col- these fatal accidents involved young drivers lecting data on youth suicide methods from 18 to 25 years of age. Our retrospective would use one of the existing scales that analysis, fashioned after the psychological accounts for the context of the event, we autopsy, revealed a number of probable would have a better grasp on the relative suicides especially in cases involving young lethality for various age, sex, and ethnic men who died in single car accidents. (Sterl- group as well as important distinctives ing-Smith, 1976). between methods chosen by youth as What are the implications and recommenda- compared to adults. This would greatly tions from our investigation into youth enhance our understanding and take us suicide methods? There rue several: beyond our current state of listing Although the numbers may be few, methods only in the grossest of forms. suicide statistics should be gathc ed on children under 10 years of age. REFERENCES In the U.S. Vital Statistics we were able 1. Card, J.J;Lethality of suicidal methods and to find methods by age and methods by suicide risk: Two distinct concepts. Omega 197n 5:37-46. 2. Bo1A, J.H: The Increasing rate of suicide Py sex but not methods by age by sex. We firearms. Wv England Journal of Medicine 1003; 3013:Sti- recommend that statistics be made avail- 874. able in this fashion. a Date!, WE., Jones, F.D: Suicide In United States Army personnel. Military Medicine1902; 147:843-847. The octant literature on youth suicide 4.Fmchick, C.J: Suicide In ymv minority grotropsc: sons, in Sudak, H.&, et al. SAM0 and the Young. with regards to method has several John Wight. 1004; pp. 31-44. shortcomings. There is a lack of control 5. Goldney, R.D:Attempted suicide in young studies. These are necessary so that we women: Correlates of lethality. British Journal of Psychiatry 1001; 13.497-505: might determine more accurately to what O. Newton, K:Allem_ pted suicide in children and extent youth suicide differs by age and adolescents. Journal of Child Psychiatry 1992 23:497- from the adult population. 503. 7, Hinton, K.. Catalan, J: ktempted Suicide. Ox- For example, are different methods ford: Oxford University Prins, VW. 8. Newton, K., Goldacre, kg: Wapitil admissions for available to youth than adults? If so, at adverse effects of medicinal agents (mainly self-poison- what ages and how does this discrepancy ing) among adolneoentil In ffle Oxford region. British Jour- nal; nit int141:100-170. arise? g. Newton, K., et el:Adolescents who take overdoses: Cheractedetios, problems, and contacts with help- We believe that the selection of a self- ing agencies. British Journal of Psychiatry 1982: destructive method depends upon a con- 140..1 le.12a 10. Holinger, P.C:Adolescent suicide:An stellation of sociocultural factors. rdemlological suglyareo.f recent trends. Aintriesn Journal However, just what and how these fac- Psychiatry 1978; 1 154-750. tors are has not been well documented. 11. Ktogsbrun, F:Too young to des:Youth and There is a need for well developed suicide. 0011t00: Houghton Win, 1978. 12. Lester, D., Murrell, M.E: The preventive effect of studies that will identify what social fac- strict gun control laws on suicide and homicide. Suicide tors are influencing the choice of method and Lib Threatening Behavior 1092 12131-140. 13. Marks, A., Mornay, T: Toward a sociocultural for the youth population and how these On moaned W-MMMon. Ufe-Threatening factors might differ by sex, by age, and remhttir1/74;4:3-17. from the adult population. It is also im- 14. Malks,A; Sex differences and their effect upon cut- tura! evaluation of methods of solf.dsstruodon. Omoga portant, once these factors have been 1977: 5:05-70. isolated to know how they might be 15.McIntire. M.S., Angle, C.FtSuicide u seen in poisoncontrol centers. Pediatrics 1971; 45:914-922.

221 2-189 Report of the Secretary's Task Force on Youth Suicide

18. McIntire, M.S., Angie, C.R I. the poisoning ac- ddentalt An ever presern question beyond the nay childhood years. Mimi Parables 1911; ifk414-417. 17. McIntosh, J.L. Santos, J.F: Changing patterns in methods of suicide by rape and sex. Suicide and We- Threatening Behavior 19tit 12.221433. 18. Penis, D.J., Bwraolough, ELM:Seriousness of suicide attempt and future risk of suicide: A comment on Card% paw. Omega 1977; 8: t41-149. 19. Paulson, M.J. Slone 0, Spode, M.k Suicidal potential and behavior In children ss 4 to 12. Suloide and life-Thrsalenkg Behavior 197It 20. Pfeffer, C., et al: Suicidal behavior in children:An outpatient population.Jouriantallincsig: Merlon Academy of Child Psychiatry 1980; 18703410. 21. Pfeffer, C., et al: Suicidal behavior latency-age children:An_ empirical study. Journal of the American Academy of Child Psychiatry 1979; 18:8791112. 22 New, C..et aSuicidal behavior in latency age psychiatric Inpatients: A repilootion and cross validation. Journal of the American Mademy of Child Psychiatry 1982; 2158410. 23. Raiford, N.B., et ak Increasing suicide rttes in adolescents end staring adults in an urban conyntuity. In Sudak, H.S. et re. Suicide and the Young. Boston: Mtn Wight. 1984; pp. 4548. 24. Selden, RH., noises, R.P:Shifting patterns of deadly violence. Suicide and Uio.Thr.atsnlng Behavior 1980; 10:105-209. 25. Shaffer, 0: Suicide in childhood and salty adolescence. Joumsl of Child Psychology and Psychiatry 1974; 15:275-29f. 28. Smith, D., Conroy, RW., Ehler, 8.1):Lethality of suicide attempt rrOm scale. Suicide and Mr-Threatening Behavior 109.4; 14:215N2. 27. Sterting-Snith, R5 An analysis of drivers most responsible for fatal accidents versus a control simple. Springfield, Vk National Technical information Service, or& 29. Taylor, M.C., Woks, JAV: The choice of weapons: A study of methods of suicide by sex, race, and region. Suicide and Life-Threatening Behaviortot10:142-148. 29. Weisman, AD., VONden, J.W: ahlit414010 riag in suicide assessment. kit:hives of General Psychiatry WM 28:553480. 30. Warden, J.W., Storting-Smith. RS: lethality patterns in multiple suicide attempts. Life-Threatening Be- havior 1973; 395404. 31. Warden__ .J.W: Lethality factors and the suicide attempt in Shneldman, E.S. Suicidolospi: Canterninkorwy Developments. New York Grune and Siration,

ACKNOWLEDGEMENT The author is grateful to Michael S. Worden for assistance in preparing this paper.

BEST COPY AVAILABLE

2-190 222 STUDY N AGES POPULATION FINDINGS

Ce idnay 1951 109 1030 Women admitted to largo city Them WM no strewne eanyhelon butwasn ego and intent to die. (Seck Seale) Fontein weal hospital fey setivaleceing. Then, was a strew. cawaSdian Moen NW* to die and miasma el the arampt. Was, of the maws athowts booked anlidapwasst mediations.

ao 13-111 Admaakee to Wood Genial °wowed to adult sersolion, te. addesarde more *swam* lack non-opide analgssies end Heatitala lees hsquady Molt psyshcfropie dap. laxtal was less *Med to madam smong adoMments then in adtes. sweaty toSisyesm Mat clays drugy wad yam found mad the house end not preacelbed for the pas%

hamon 082 12-20 Genital Hamad admissions. Satoeleconim was higher far amass Men for main. Of frequent use sae wysigssics and Oct Maas 1952 Csdadaife 1974.79 and-pyrelice, the use of wads Inewneed by op tat males end demand by me for female*. The use et pridrotrople flags Inersesed the women gat older but such vas net Wafer the

Hotinget 1978 uraneen 1024 Wa. vitsi stabelles eamMeefd Polon Other 1014 8% G"0% "PritiFIZZ. -314 1549 13% 7% 19% MS 3% 2024 13% 8% 19% 0314 0 The anal* number ef milsollenkig in Me 1014 age meg be due ito Maim much bah/Ma as eacoldanlet.' Flamm Ses, end Homing wen* woad moil fisquantly by maim Istipokoning we not tram* used by femeles.

Mointro I Angle t, 103 848 Repots from 50 pollen wan in 7911 of the pelsoninge were suicide misled. 1971(10 3514 PA U.S. ant U.K., 1958.1058. Setwafradm via highs, among nabs Men amass in Ms )ummy ages p.m) but the 85% F craw ewe mimed In the alder group (114.11. SelhocisenIng mom Was diparaw4d vAth ege Wes It lnereaseci alth op for ealeasiany. Them I. speaufrian that Me aides blade may hem 1031wd to ether mane then poison.

Wintry.Angle 1.100 018 Repoos from 50 poison owt.-.4 1971(li) maws In U.S. end UK 010 Uff"--Egfil 1.13 31% 09% 4-111 20% 72% IT-% WI MS As In adults, selfpoisoners ale mom foam* female. Set-palsoning in a child ever Spars Ow I rerely aocidanad. (bmblksatee, n lI,us1 1enquii.ss$ were used mast freffuenfry hy el age raga The nod mast used dug woodmen which woo used ks equally at all mow

Mane 1977 003 Mean age OW College Students from South in mane paleanoty of said* bolh same teased selfixibanIng the most 1.4 208 19.5 Sf111. and Non-South dakabfe. Roams awe untied highae leit men than women. Women visa ma* Moly than F 1112 men My painfsesness fix their choice al method. Men woman% May le yhoom method, because et tholt atoeselbeg at stitokimay.

a Table t aco 224 223 BEST COPY AVAILABLE STUDY N AGES POPULA170N RNDINGS

Documented skids seempte Irwatdd more sedebesee and badly multelars then Paulson WM 34 Mean we-62 yrs. awn at M -23 UCLANeurapeychlatdc selisdeankig. Deliabuse Intatred cull% GOMM& burning erd juneskte In frontal mad% F-11 Melbas 19704974 ushIctee or from MO deose. Allsough them were no est strand dremnase In Idediand deism% Moe see a delleant relelonehlp between tangy dlecsgerdmlors. Mogul Mall% end e mulletkv useulL mg, Redd 1976 42 042 BMX Muni:Mel Moped. Jumpine wee Me most frequent choke at ellemeters (31116. flobassed by Norsoisoning Won so* -9.0 CMS Rya. 1;1valve. burning ea% *Me Dal. and needle kilo Wag 0,14. NewI1 lumping ma the mod *mese Ideedsm or threat. only ane lumped In trod at bale. Rad et el. 13 6-12 MO 0, Me semelnIng f 2. Owes woo 1 ise-poloceirs 1 dabbing, 2 hengIngm end rt used In 259I at el ROW de 6-12 ChIlden Winded to** child Jullirgle from illh 01100114110 Ms Mae *desert method used end I eds se -46 psyching, tine at Pew Yost Clele. F-17 14osoltd, Weekheeler : March 001114une Mt Sneer tem 30 12-14 condeted *kids* In Engemd - Wet fremserd method used wee carbon mancedde. & Weise lamins. -More nude then %melee hanged thsenestme. - More hmeho then mom used eelldebaring.

1624 Mimed eldidicell The nee al suicide by Seem* Increased Mt% tram 111M41115 Mile the nide at decide by Sweden & IWO undo.% new mare Inueseed 72.4% tor Me some pedod. Die etadlowdy low Mateo Indepth by dreams may ceded Me 143% Mentes* In the dellebny atisist heffideune dufing thesone time pedal.

,. Table I concluded.

r BEST COPY AVAILABLE ..r. 2,4t) NEUROTRANSMITTER MONOAMINE MU/SOLUTES IN THE CEREBROSPINAL FWID AS RISK FACTORS FOR SUICIDAL BEHAVIOR

Marie Asberg, M.D., Professor, Depalment of Psychiatry? and Psycholov, KarolinskaHospital, Stockholm, Sweden

Introduction The idea that brain biochemistry may con- studies of brain tissue obtained at autopsy, tnbute to a person's decision to take his own and measurements in cerebrospinal fluid life is fairly recent. In a bibliography (I) of re- (CSF), blood platelets and plasma, and urine. search on suicide published between 1958 Monoamines and their precursors, cataboliz- and 1967, only five out of 1267 titles deal with ing enzymes, and degradation products have biochemical subjects. Suicide has been con- all been measured; and hormonal processes ceived of as an exclusively human behavior, thought to be controlled by monoamine which presupposes intentionality and a con- neurons, and the reaction of the various sys- cept of death, and whose biological back- tems to challenge, have been studied. This ground is remote and irrelevant. review will focus on measurements of CSF Recently, however, two lines of study have concentrations of monoamine metabolites, suggested that some instances of suicidal be- particularly the serotonin metabolite, 5- havior may indeed have biological correlates, hydroxyindoleacetic acid (5-HIAA) and the which obtain not only in conjunction with dopamine metabolite, homovanillic acid depression, but even perhaps when no (HVA). depressive disorder is apparent. Two clusters of biological factors have emerged Cerebrospinal Fluid that tend to correlate with suicidal behavior, Measurements as an indicator of namely variables associated with a Brain Events neurotransmitter, the monoamine serotonin CSF metabolite concentrations ttave been (5-hydroxytryptamine, 5-141), and variables associated with certain neuroendocrine widely studied to clarify the turnover of the monoamines in the brain. The advantage of functions.This paper will review the evidence for a relationship between the CSF studies is that spinal fluid is com- paratively easily obtained with little discom- serotonin and suicide attempts or completed fort to the patientusually by means of a suicide. lumbar puncture (LP), a routine procedure in neurological investigations. Biochemical Methods There are disadvantages as well. The con- The biochemical investigation techniques centration of the metabolites of serotonin used in the field almost entirely derive from and dopamine, 5-HIAA and HVA, depend, studies of depressive illness. They include inter alitt, on the subject's sex and age (2-4),

2-193 22 7 Report of the Secretary's Task Force on Youth Suicide and on body height (5-7). The dependence Stanley and coworkers (25) of a strong cor- on body height is presumably due to an active relation between 5-HIAA concentrations removal from the CSFof the acid metabolites measured post molten: in the frontal brain as they flow from the brain ventricles down cortex, and in the lumbar spinal fluid, sup- to the lumbar sac where the CSF issainpled. ports the contention that OF 5-HIAA in- The concentrations of 5411AA and HVA 1.!=d reflects brain events; interestingly, the decrease along the route of the CSF from the correlation between HVA in brain and in cerebral ventricles to the lumbar sac (8-9). A (SF was lower and was not statistically sig- concentration gradient is seen even within nificant. the small volume usually drawn at lumbar The concentration of the acid monoamine puncture (1042). metabolites in the CSF is a function not only Metabolite concentrations also vary of their production rate, but also of their seasonally (7,13), and with the time of the day removal by an active transport mechanism (14). Most important for clinical studies, the from the cerebrospinal space. That the concentrations of 5-HIAA and HVA are transport mechanism can be blocked by drastically altered by treatment with certain means of probenecid, has been taken ad- psychotropic drugs. Many antidepressant vantage of in attempts to obtain more valid treatments lower CSF5-111AA (4,15-19), and estimates of transmitter turnover (26). The neuroleptic drugs usually increase HVA (20- probenecid technique has been described in 23). Some factors of importance for the con- detail by van Praag et al. (28). The centrations of 5-IIIAA and HVA in lumbar probenecid technique, while removing one CSF are summarized in Table L(Tables source of error, introduca other problems, begin on page 207.) however.Wit'd lower concentrations of probenecid, the blockade of the transport Clearly, when groups of subjects are com- mechanism is incomplete, and may vary pared, confounding factors must be control- within individuals because of differences in led or taken into account, as their influence probenecid metabolism. Higher probenecid is sometimes substantial and may lead to er- concentrations often cause nausea and roneous conclusions. Thus, if consistently vomiting, and may also alter central more CSF is drawn from one group than from neurotransmitter turnover. another in a comparative study, the average concentration of 5-HIAA and HVA will One of the advantages with the probenevid naturally be lower in the group from whom technique is that concentrations of the less CSF was taken. On the other hand, a monoamine metabolites are increased, which true difference in concentration may be hid- places less heavy demands on the analytical den if, for instance, control subjects are taller methods. With the very sensitive methods on the average than experimental subjects. available today, concentrations in the The difference in average CSF 5-HIAA be- nanornole range can be measured with satis- tween depressed patients and healthy con- factory precision, and most investigators rely trols of equal stature, is numerically smaller on baseline measures of the metabolites than the difference between tall (>180 cm) rather than on probenecid-induced ac- and short (>160 cm) subjects, irrespective of cumulation. whether they are healthy or depressed (7). The concentration of a transmitter metabo- Measures of Suicidal Behavior lite in the CSF is at best an indirect measure In comparison with the advanced biochemi- of the turnover of the parent amine in the. cal methods used in the studies to be brain, and it has been argued that, for ex- reviewed, the approach to measuring suicidal ample, 5-HIAA concentrations reflect behavior has been much less sophisticate& events in the spinal cord, rather than in the The reason for this is probably that, rather brain (24). However, the recent finding by

2-194 228 MAsberg: NeurotransmitterMonoamine Metabolites... than being designed to deal with suicidal be- et al. (29) unexpectedly found patients with havior, most studies were focused on depres- low concentrations of the serotonin metabo- sive illness and relied on procedures lite to represent an increased incidence of developed for measuring severity of deprm- suicide attempts (defined as any deliberate, sion. Diagnostic inventories and depression self-inflicted injuty, regardless of the lethality rating scales often contain an item dealing risk involved, that the patient had thought to with suicidal ideation and tendencies, and entail a death risk). such ratings have been used in some studies. The findings of previous research by van In other studies, the occurrence of 'suicide Praag and !Cori* (30) and Asberg ali.1 attempts' has been related to the biological coworkers (4,31) (see also Gibbons and Davis variables. While some studies use more or (32)) had indicate( that the concentrations less explicit, operational definitions of the of the metabolite were bimodally distributed term suicide attempt, others do not even at- in depressed patients, suggesting the exist- tempt a definition. Although, in very few ence of a biochemical subgroup of depressive studies, ratings have been made of intent and illness characterized by disturbed serotonin lethality of a suicide attempt, heuristically the turnover. In the study by Asberg et al. (29), most useful classification seems to be accord- 40 percent of the patients with low CSF 5- ing to the method used in the attempt (active, HIAA concentration had attempted suicide violent, or passive, nonviolent) perhaps be- during their current illness, as compared with cause of its high reliability. 15 percent in patients with normal 5-HIAA. Moreover, the attempts were of a more The time span involved also varies from one determined nature with a preference for ac- study to another. Some investigators have tive, violent methods in the low 5-HIAA considered the incidence of any suicide at- patients, whereas those in the high 5-111AA tempts in the patient's history, while others groups were confined to drug overdoses. have focussed on attempts during the current 1vo deaths from suicide occurred during the illness episode. In the former approach, the study period, both in low 5-HIAA patients. biological measures are assumed to be stable over time, a controvetsial assumptionwhich The relationship between CSF 5-HIAA and will be discussed later. suicidal behavior was confirmed by Agren (33), who studied depressed patients and Only a very few investigators have examined measured suicidal behavior by means of the the possible predictive value of biologjcal suicide behavior scales in the Schedule for variables for ultimate suicideunder- Affective Disorder and Schizophrenia standably so, considering the low base rate of (SADS). These scales do not differentiate suicide and the time and cost of the investiga- suicidal ideation and suicidal acts. In the As- tkai. berg et al. (29) study, low CSF 5-HIAA was None of the available studim has dealt with not correlated to suicidal ideation, only to the question of youth suicide.Although suicidal acts. Argren's choice of method may there seems to be little reason to believe a thus have weakened the correlation, which priori that a correlation between a biological nonetheless was statistically significant. variable and suicidal behavior would be These early studies did not take into account limited to a certain age group, the issue the relationship between CSF 5-111AA and remains to be empirically examined. such interference factors as sex and body height. Men tend to lave lower CSF 5- CSF 5-hydroxylndoleacatic acid HIAA concentrations than women, and they (541M) are also more prone to use violentmethods In a study of possible clinical correlates of if they attempt suicide. The sex factor could, CSF 5-HIAA in depressul patients, Asberg however, be ruled out in a subsequent con- firmatory study by Traskman et al. (34), who

r 2-195 -229 Report of the Secretary's Task Force on Youth Suicide

adjusted for interference factors by analysis have not been included in Table 2, which of covariance (ANCOVA). summarized the relevant studies. More recently, the relationship between 5- A well-designed, nonconfirmatory study was HIAA and suicide has been confirmed in performed by Roy-Byrne and coworkers Dutch depressed patients studied by van (43), who studied Americoin patients, most of Praag (35), who found a highly signiftat in- them more or less treatment-resistant, creased incidence of suicide attempts in referred to a research center specializing in patients with low probenecid-induced ac- the study of depressive disorders. No sig- cumulation of 5-HIAA The association be- nificant relationship was found between CSF tween 5-HIAA and a violent mode of the 5-HIAA and suicide attempts (over the attempt was not confirmed, however. individual's lifetime), possibly owing to the In a British study of depressed patients, high proportion of bipolar (mank depres- Montgomery and Montgomery (36) also sive) patients in the group. Suicidal unipolar found more suicide attempts in patients with patients tended to have lower CSF 5-HIAA low CSF 5-HIAA concentrations (using the than had nonsuieidal unipolars, but the num- cot-off point between love and "normal* 5- ber of such patients was too small for statis- HIAA suggested by Asberg and coworkers tical analysis. The biological correlates of suicidal behavior may thus differ between (N))- bipolar and unipolar disorders, a conclusion Among depressed patients in India, also reached by Agren (44). Palaniappan and coworkers (37) found a significant correlation between CSF 5-HIAA Another difference between the study of concentrations and suicidal tendencies es- Roy-Byrne et aL (43) and those of Asberg et timated by SON= on the item Suicide in the al. (29) and Traskman et al. (34), is that the Hamilton Rating Scale. A rating scale index former considered suicidal behavior over the of suicidal tendencies was also used by Leek- patient's entire life span. In the Asberg et man et aL (38), who found an association with (29) study, the significant association with 5-HIAA which was confined to patients with CSF 5-HIAA was restricted to suicidal be- disturbed reality testing. havior during the index illness episode. The discrepancy suggests that CSF-HIAA values Banki and coworkers (39) found a relation- may not be stable over time in suicidal in- ship between low CSF 5-HIAA and suicide dividuals, a poa..ibnity that will be discussed attempts in Hungarian female patients, a in further detail below. relationship that was confined to those who had used active methods. Lopez-Ibor et al. Only about half of those who cocunit suicide (4) (1985) reported a relationship between are retrospectively diagnosed as having suf- suicide attempts and low CSF 5-HIAA in fered from a depressive syndrome, as sug- Spanish patients, irrespective of the method gested from the thorough psychological used in the attempt. A further confirmation autopsies performed by Beskow (45) and As- gard (in preparation). Several groups have in Swedish patients was provided by Dimon and coworkers (41), using the same methods studied the relationship between suicide at- as in the original Asberg et al. (29) study. tempts and MT 5-HIAA in other diagnostic categories. Traskman et aL (34) found CSF There are also, however, some nonconfir- 5-HIAA concentrations to be loco= in non- matt:a)? studies. Vestergaard and coworkers depressed suicide attempters (mainly (42) mention that among depressed patients patients with personality disordas and minor studied by them, suicide and suicide attempts affective disordeis). Brown et al. (46-47), were equally frequent in individuals with low studying two pimps of men with personality and high CSF 5-MAA. Since they do not disorders, found more subjects who had provide any further information, their data made a suicide attempt at some point in life among those with low CSF 5-HIAA.

2-196 2 0 MAsberg: Neurotransmitter Monoamine Metabolites. van Praag (48), andNinon and coworkers Low concentrations of HVA in suicidal (49), found a similar association in depressed patients have been reported by schizophrenia. This finding is somewhat Traskman et aL (34), by Montgomery and against the odds, considering the report by Montgomery (36), by Palianappan (37), and Sedvall and Wode-Helgodt (5) that a sub- by Roy et aL (56). In the Roy et aL (56) study gaup of schizophrenic patients(those with of 27 depressed patients, the association be- a family history of the disorder)have abnor- tween low HVA and suicidal behavior was mally high concentrations of the metabolite. much stronger than that between 5-HJAA Both suicide studies are well-designed with and suicide, which did not reach statistical carefully selected, matched controls. significance. In his 1980 study, Agren (3.$) Patients with a depressive disorder superim- found no association between HVA and any posed on their schizophrenia were of the SADS suicide scales. In Agren's later deliberately excluded from van Praag's (48) (44) and larger patient group, he reports an study. Roy et al. (51), however, found no dif- association between low HVA and the ference in CSF 5-HIAA concentrations be- lethalityof suicide attempts made prior to the tween chronic schizophrenic subjects who current episode. had attemptec: suicide at some time during Banld et al. (39), on the other hand, found their life, and those who had not made such suicide attempts to be less dearly related to attempts. Lower CSF 5-HIAA concentra- HVA than to 5-HIAA. In particular, their tions were, however, reported in suicidal depressed patients who had taken drug over- patients with schizophrenia than in non- doses had significantly higher HVA than bad matched controls by Banki et aL (39), who nonsuicidal patients, whereas HVA was low also found similar relationships in al- in attempters who had used violent methods. coholism and adjustment disorder. The bulk of the evidence would thus seem to sup- The studies of HVA in CSF in relation to port the notion that potential forsuicidal be- suicidal behavior are summarized in Table 3. havior is reflected in low concentrations of Interestingly, none of those who have studied CSF 5-111AA-- even when no major affective nondepressed groups have reported any as- disorder is apparent. sociation between CZFHVA and suicidal behavior. Thus, Brown et aL (47) 'mind no auociation in their patients with personality CSF Concentrations of disorders. Leckman et aL (38) report no as- Homovanillic Acid (HVA) sociationintheirdiagnostically The average concentration of the dopamine heterogeneous group, and Traskman et\al. metabolite, HVA, in CSF is reduced in (34) found an association only in those of depression (52-53), and more consistently so their patients who fulfilled research criteria than is CSF 5-HIAA, to which HVA is never- for a diagnosis of depressive illness. Ninon et theless strongly correlated.Whether the al. (57) found no association in their correlation between the two metabolites is schizophrenic subjects. A possible inter- due to their sharing the same transport pretation of the findings would be that CSF mechanism, or to a functional connection be- concentrations of HVA are related to tween the parent amines is notknown. A suicidal tendencies, but only in conjunction functional connection would seem to be in- with a depressive illness. Studies of patients dicated because of the consistent finding with bipolar depressive illneu would seem to that, L addition to reducing 5-HIAA, drugs be particularly interesting in the context. that interfere with seratonin turnoversuch as the antidepressantsclomipramine (54), zimeldine (19), and citaloprain (55)-change HVA concentrations in CSF, while having no known direct effects on dopamine neurons.

, 2 3 2-197 Report of the Secretary's Task Force an Youth Suicide

Noradrenaline and were closer to normal in suicidal, than in non- 4-hydroxy-3-methoxy- suicidal depressed patients.Melatonin phenyiglycol (HMPG) production is dependent on prevalent light- ing conditions and thought to be regulated by In contrast to the evidence relating suicide to beta-adrenergic neurotransmission. serotonin,ther elationshipwith Serotonin is a precursor of melatonin, al- noradrenaline is less clear.In depressed though 'tittle is known of any correlation be- patients, Agren (33) reported a negative cor- tween the concentrations of the two relation between suicidal tendencies and the compounds in humans. CSF concentration of the noradrenaline metabolite, 4-hydroxy-3-methoxy-phenylgijco1 (11MFO). Brown and coworkers (46) found Post Mortem CSF Measurements a positive correlation in subjects with per- in Suicide Victims sonality disorders, which was not reproduced The concentrations of monoamines in their study (47) of borderline patients. (serotonin, dopamine, noradrenaline and In two studies of mixed diagnostic groups, adrenaline) after death by suicide were Ostroff and associates (58-59) measured the measured in suboccipital CSF by Kauert et al. (67), who somewhat unexpectedly found ratio between noradrenaline kind adrvadline (the NA:A ratio) in urine and found a increased serotonin concentrations in the suicide victims. Their finding has however, relationship between a low ratio and suicidal received support from preliminary findings behavior.Within a group of suicide at- by Arato et al. (68), who report significantly tempters, Prasad (60) found the NA:A ratio higher concentrations of 5-HIAA in lumbar to be signifwantly lower in those who used and suboccipital CSF obtained post mortem violent methods in the attempt. from suicide victims Other Substances in the CSF: The CSF autopsy studies are summarized in Cortisol and Magn9sium Table 4, which also contains summaries of some studio of monoamines and metabolites The relative robustness ofthe association be- in brain tissue from suicide victims. The post tween CSF 5-HIAA and suicide tendencies mortem CSF findings may prove crucial for has inspired investigators to examine the cor- our understanding of how alterations in relations with other biological markers than serotonin transmission predispose to suicidal the amine metabolites. Traskmui et\al. (61) behaviorthat is, if they can be confirmed thin; measured cortisol concentrations, but and are not due to any of the sources of error found no abnormality in suicide attempters. that mar autopsy studies of suicide victims Depressed patients, on the other band, had (such as delay between death and discovery significantly higher CSF cortisol than bad of the body, the influence of drugs, mode of healthy control subjects dying and agonal state). Banki and coworkers (62) found a relationship between suicide attempts and low CSF Prediction of Suicide from CSF concentrations of magnesium. There was a Measures strong positive correlation between CSF magnesium and CSF 5-HIAA. Interestingly, Those who commit suicide and those who magnesium concentrations in CSF are merely attempt it differ notoriously in many important respects, evai if there is an over- strongly correlated to CSF melatonin con- lap between the two populations (69).In centrations (63). Melatonin in plasma may in semal studies, subsequent mortality from turn be related to suicidal tendencies. Beck- suicide among suicide attempters has Friis and coworkers (64) reported that noc- amounted to about 2 percent within a year turnal serum melatonin concentrations, after the attempt (70). Although this is a known to be decreased in depression (65-66), 232 2-198 M.Aeberfp Neutotrammitter Monoamine Metabolite considerable increase in suicide frequencyserotonin have been reported to be disturbed over that of the general population,suicide in depressive illness. Among them are the is a rare event even in this group. concentrations of the precursor, tryptophan, in serum and its ration to other amino acids Estimating suicide risks° as to be able to take transported by the same mechanism (74), the appropriate precautions is one of the most binding of the antidepressant drug im- difficult tasks of the practicing psychiatrist, ipramine to specific sites in blood platelets and many attempts have been made to create (75-76), the uptake of serotonin by the rating scales and inventories for the purpose. platelets (77-78), and the concentrations of Most of these have not been very successful serotonin in plateles and plasma The uri- (71-72), which may be due, at least partly, to nary output of 5-HIAA, on theother band, is the low base rate of suicide and other statis- of little interest since it is strongly influenced tical problems (73). by diet, varies from day to day, and is uncor- Among a well-known risk group for suicide, related to CSF 5-HIAA (79). namely pada ts who have made a suicide at- So far, there are very few studies of other tempt, those with low CSF 5-HIAA were 10 serotonin-related markers in relation to times more likely to die from buicide than the suicidal behavior, and little is known of the remainder (34) (see also Table 5). Roy et al. interrelations between them. These (56) reported a relationship to exist between relationships need to be clarified, both with low concentrations of HVA in the CSF and a view to understandingtheir physiological subsequent suicide in depressed patients, significance and for practical diagnostic pur- regardless of whether previous attempts poses. have been made. These findings suavest that inclusion of biological variables in the clini- Interotingly, there are no clear-cut relation- cal assessment of suicide risk might increase ships between imipramine binding and its precision. serotanin uptake in depressed patients (80). Mese two possible serotonin markers may To judge from the studies putoished so far, thus reflect different aspects of serotonin there is fairly consistent evidence that low function. The relationships reported be- concentrations of CSF 5-HIAA are as- tween CSF 5-HIAA and platelet MAO ac- sociated with an increased rate of suicide at- tivity have not been consistent (81-82). tempts, and may be a risk factor for suicide in individuals with a psychiatric history. There Monoaminergic neurons are known to be in- is also evidence relating low HVA concentra- volved in the chain of events resulting in the tions in the CSF to suicide, although so far release of many hormones, including cortisol. only in depressed individuals. The details of this have not yet been worked out, but the data from the Meltzer et oL (83) Among the many questions raised by these study of 5-hydroxytryptophan-induced ftndings, a few will be discussed here: how do release of cortisol strongly suggest a func- the (SF risk factors correlate with othce tional connection between the serotonin sys- potential biological risk markers; what can be tem and HPA axis. inferred about the processes whereby a disturbed serotonin system may predispose to The available human data do not, however, suicide; and how this knowledge can be ap- show any negative correlations between plied in preventing suicide. markers of the serotonin system and the hypothalamus-pituitary-adrenal axis, such as might be expected if they reflect an identical Correlations Between Possible risk factor for suicide. Thus, CSF concentra- Biological Risk Factors tions of cortisol and of 5-HIAA have been Apart from the 5-111AA concentrations in shown to correlate positively, though weakly CSF, a series of biological markers related to (61), or not at all (62,84). Both Carroll et al.

ft .01.10. .. : 233 2-199 Repott of the Secretaty's Task Force on Youth Suicide

(85), and Banki and Arato (86), found a posi- for the platelet serotonin uptake (93-94), and t ive correlation between postdexa- for the platelet 3H-imipramine binding (95- methasone cortisol and 5-HIAA. 96). There is some evidence that a seasonal Interpretation of the results of Carroll et aL rhythm, very similar to that observed for (85) is, however, complicated by the fact that serotonin in the hypothalamus, may also exist spinal fluid was drawn after the administra- for CSF 5-HIAA (7), tion of dexamethasone, which raii.es CSF 5- }MA concentrations (87). The evidence from four published followup studies of depressed patients (52,97-99) is Among other potential markers of serotonin, summarized in Table 6. 5-111AA concentra- the ratio of 1-tryptophan to other neutra. tions in CSF appear to remain fairly stable amino acids was positively correlated to over limited periods in normal subjects, and postdexamethasone cortisol (88), whereas in depressed patients re-admitted for relapse vmax of serotonin upt .ke into platelets of depression (99). Recovered depressives, (which is reduce in depression) tended to be whose concentrations are normal during ill- negatively correlated to an abnormal DST ness, also remain stable over prolonged (89). Preliminary reports suggest that the 5 periods, whereas in depressives with low 5- hydroxytryptophan-induced cortisol release HIAA during Him= the concentration some- may be related to CSF 5-HIAA (90). times increases with recovery, though it Gold and coworkers (91) report an inverse remains in the low range in most cases. correlation between CSF 5-HIAA con- A possible interpretation of available data is centrations and the magnitude of the in- that there is a subgroup of depressed crease in throid stimulating hormone (TSH) patients, characterized by concentrations of reaction to administration of thyrotropin CSF 5-HIAA that are not only low but also releasing hormone (TRH). The negative less stable over time. If this type of unstable correlation between CSF 5-IIIAA and the serotonin system is associated with an in- TRH/TM-test also appears in a study by creased vulnerability to illness, and with a fur- Banki and coworkers (39), where it is com- ther decrease in release during illness, the patible with their finding of more normal emergence of bimodal distributions in dis- TRII/I'SH- responses lin suicidal than in non- eased populations is easily explained. suicidal patients. In line with the 'instability' hypothesis, are findings from two patients in whom tepeated Stability of CSF Concentrations of lumbar punctures were made, and who sub- 5-HIM over Time sequently committed suicide (Asberg and Related to the question of the usefulness of coworkers, in preparation). In both cases, biological markers as risk factors for suicide, there was a substantial reduction in CSF 5- is their stability over extended periods. HIAA from one puncture to the next. The above-mentioned finding by Arato et aL (68), Unfortunately, CSF studies of recovered of higher CSF 5-HIAA in CSF from suicide depressives are rare. Such patients are often victims than in controls may also be in line maintained on drugs for extended periods, with an instability hypothesis. and those who are not, even if available for lumbar puncture studies, may well be non- representative cf the depressed population. Low CSF 54IIAAa Vulnerability Marker? A further compliattion in followup studies is that most serotonin-related variables also Low concentrations of 5-HIAA in CSF occur seem to vary seasonally. Seasonal rhythms not only in depressed and suicidal people, but have been shown for the serotonin con- also in perfectly healthy subjects (53). This centration in the human hypothalamus (92), suggests that low CSF 5-HIAA is not a marker of the state of depression, but rather

234 2-200 MAsberfr Neurotransmitter Monoamine Metabolites... an indicator of vulnerability. Supportingthe in those cases where the victim is a spouse vulnerability hypothesis, van Praag and de (107). (In the United States, the suicide rate Haan (100) 'mind an increased incidence of among murderers is lower, around 4 percent depressive illness in relatives of patients with according to Wolfgang (108).) low CSF 5-HIAA, compared with those of Several investigators have tested the patients with normal 5-HIAA concentra- hypothesis that aggression dyscontrol is the tions. This finding is reminiscent of the ob- link between serotonin turnover and suicidal servation by Sedvall and coworkers (101) that behavior. Brown and associates (46) found a CSF 5-H1AA concentrations were lower in life pattern of aggressive behavior in subjects healthy subjects with a family history of with personality disorder and low LSF 5- depressive illness than in healthy subjects HIAA. without such antecedents. Preliminary data from twin studies by Sedvall and coworkers Further support for a relationship between (102) further support familial involvement in serotonin and violence came from three CSF concentrations of the monoamine me- studies of murderers.Linnoila and tabolites. cmorkers (109), found lower CSF 54HAA in violent offenders whose crimes were un- Serotoning Aggression and Suicide premediated.Lidberg et al. (I10), found lower CSF 5-HIAA in homidde offenders If serotonin transmission is permanently low who had killed a spouse or a lover than in or unstable, it is conceivable that this maybe those who had killed someone of less emo- manifested in other ways than in suicidal ten- tional significance (usually a drinking buddy). dencies. The often quite unpremediated, im- Lidberg and coworkers (111) also found very pulsive and violent character of mry of the low CSF 5-HIAA concentrrtions in three suicide attempts in low 5-HIAA patients cases, where suicide attemptei s had killed, or gave rise to the suggestion (29) thatthey attempted to kill, their children. might have difficulties in controlling aggressive impulses. The hypothesis was supported A relationship between s-aotonin And ag- by the association known to exist between aggressive behavior in alcoholics was also found gression in animals, and serotonin turnover by Branchey et al. (112), who studied the ratio (summarized by Valzelli (103)), as well as the of tiyptopban to other neutral ',amino acids in links between anger and suicide proposed by serum. They found significantly lower ratios, classic psychoanalytical theory (104404 compatible with a deficiency of brain serotonin, in those subjects who had been ar- 'Aggression' is a somewhat nebulous con- rested for assaultive behavior than in other cept. The word has many meanings, and alcoholics or in nonalcoholic controls. some aspects of aggression arehardly amenable to empirical study. Aggression, in the sense of verbal threats or violent acts Suicide and the Biology of aimed at causing injury to others or to Personality oneself, has, however, been studied in Interestingly, some personality features that suicidal individuals. lims, Weissman et al. seem to be prominent in patients who at- (106), found that excessive hostility was tempt suicide, are also associated with CSF characteristic of suicidal depressed patients, concentrations of 5-HIAA. These per- and Brown et al. (47), found more overt ag- sonality features often reflect impulsivity and gressive behavior in subjects who had made problems in the handling of anger. In normal suicide attempts. people, low CSF 5-HIAA appears to be associated with vitality, social dominance and One of the strongest predictors of suicide is In Great Britain, a 30 percent easily aroused anger, as shown by Zucker- murder. man et al. (82), andSchalling et at (in suicide rate is reported among murderers In psychiatric patients, low after the act. The risk of suicide is greatest preparation).

'""%r 235 2-201 Report of The Secretary's Task Force on Youth Suicide

CSF 5-HIAA has also been associated with A better understanding of the biological and high vitality, self-reported impulsivity and psychological links between serotonin turn- psychopathy- related features (47,113414), over and suicidal behavior might also open and with high hostility and anxiety in ratings up new approaches to the prevention of based on Rorschach protocols (115). Correla- suicide. Serotonin transmission can be con- tions with HVA, when reported, are general- trolled, with drugs or amino acid precursors, ly parallel those with 5-HIAA, but are and possibly by dietary changes, and it would weaker. seem important to test such treatment regimens in patients with a high suicide Implications tor Suicide Prevention potentiaL It is also possible that an increased understanding of the psychological processes Their association with a heightened risk of that are controlled by serotonin neurons suicide suggests that markers of serotonin could be used to develop more specific may be valuable in a clinical context. Low psychotherapeutic techniques than has concentrations of CSF 5-11IAA in suicide at- hitherto been possible. tempters, for instance, were connected with a 20 percent mortality from suicide within a year, which suggests that the combination CONCWSIONS may be one of the strongest suicide predic- In a number of studies, a relationship has tors hitherto identified. The number of false been shown to exist between low MP con- positives is, however, still very large. centrations of the serotonin metabolite 5- Although it seems Rely that (SF determina- HIAA and an increased *incidence of suicide tions might help in the assessment of suicide attempts in psychiatric patients. Although risk, there arc problems in applying the tech- most studies deal with depressed patients, nique in a clinical setting. Owing to the many there is fairly strong evidence that this factors that influence csF concentrations of relationship exists in other disorders as well, 5-IIIAA and HVA, the spinal tap procedure particularly in personality disorders and pos- must be standardized to an extent that is rare- sibly also in schizophrenia. Some reports ly practical in a busy clinic. Furthermore, the suggest that bipolar (manic-depressive) dis- patients must be hospitalized overnight, and order may be an exception. most difficult of all, they must have been off Low concentrations of the dopamine meta- antidepressant and neuroleptic drugs, and bolite HVA may also be associated with lithium, for several weeks prior to the punc- suicide attempts. Although this may to some ture. extent be accounted for by its correlation Usually, the spinal tap is easily tolerated by with 5-HIAA, there is probably more to it, the patient, and the post LP headache that since unlike 5-111AA, the association may be afflicts about a third of the subjects is not a confined to depressive disorder. major problem The procedure sometimes Both markers have been associated with an appears to pose greater problems for the increased frequency of ultimate suicide, but psychiatric staff, who may feel that it is too there is a need for further prospective *medical* and out of tune with the type of studies. therapeutic relationship they wish to establish with the patknt. Low CSF concentrations of 5-HIAA may reflect a low serotonin output, or ponselly a Thus, though there is an obvious need for low stability serotonin system, which may in new, more easily accessible markers of the turn be a vulnerability factor. In most in- state of the serotonin system, in centers with dividuals with low CSF 5-HIAA, this vul- access to the relevant analytkal procedures, nerability will never be manifested in a routine spinal taps may nevertheless be a real suicide attempt. A suicide attempt Ls unlfice- help in clinical management. ly to occur unless the individual finds himself

2-202 t:-236 M.Asberg: Neurotransmitter Monoamine Metabolites... in a situation which he conceives of as bolhes in human cerebrospinal fluid and their relations to age and sex Neurophermoology 1971; 10.111515472. hope for the desperate, or when he is without 4. Alborg kt, Beason L., Tuck D. Cronhotm B, future. Adverse events may have created ibis via F. Indalerakte metabolise in the cerebrospinal fluid of depressed patients before and during vestment with situation, or the individual's perception of nortdplOine.Clinical Phraseology and Therapeutics the situation maybe colored by depressive ill- 197% 142/7-2911. ness. Previous experience of adverse events 5. VAKIls-Helgodt B, WWI 0. Corretations between height of subject and concentrations of monoamlne me- (e.g., during childhood) is liable to render the tabolise in osrebiuspinal fluid from psychotic men and women. Commus in Psychopharmacology interpretation of current adversity more 197t,t117483. ominous. Whether this state of affairs leads O. Paben tet, Balloon L Serotonks in depressive id- to a suicide attempt, is to some extent deter- nue - Studio of CSF 5-HIM. In:Saletu B et al., eds. Neuro- peeehophermaoology. Oxford-New York:Per- mined by the quality of the person's social gamon Press, 1971k1 150111- support network, which may attenuate the T. Aaberg M, Balloon L Rydin E, St:hailing D, Thom P, Tradonen-Benclz L Monoarnine metabolites in effect of adverse events, or render the suffer- cembropinid fluid ht relation to depressive Mass. ings of depressive illness more tolerable. suicidal behavior and peracnalfty. In: Morn mt B. Burrows G. Lader 141, Urssrd. 0, El, MuMsy 0, eds. Recent ancsö1n Narro-psychophannecology. Oxford A low-output serotonin system (or perhaps and New York: Pergamon Pros, 1951:257471. even more likely, a low-stability one) might I Wawa HC, Pahcroft OW, Crawford TEL Con- an vulnerable centrations of PsyWanftdOasoeW sok' and homovenii- render individual more to self- lic add In the cerebrospinal fluid of the dog before and destructive or impulsive action in time of durino treatment with probenecid.Life Sciences 1 oak& 1571-1575. crisis.This characteristic of the serotonin 9. Moir ATB, Pahoroft OW, Crawford MB, Swinton system may have a genetic basis, or it may be D. Wawa HC. Cerebral metabolites in owebnal fluid as a Woolemical approach to tree brain.ftin acquit' ed. 19709*35?-365. Although little is known of the processes 10. Sync L., Kraemer H, Sack R, it al. Gradients of blogenio amine metabonto in oreLvospinal fluid.Dis- linking serotonin with suicidal behavior, eases of the Nervous System 1975,39:13-19. there is some evidence that personality fea- 11. Jeoupoevic N Lachovic 2, Stsfoeld 0, Bulat M. Nonhomogeneoue _distribution of erhydroxylndoletic tures such as impulsivity and difficulties in sold and homoventlic add in the lumbar cerebroepinal handling aggression may be important inter- fluid of man. Journal of the Neurological Sciences 1977;31:185471. vening variables. 19. Ballston L Alborg U Lantto 0, Scalia-Tombs G- P, Treekman-Bendt L, Tybring G.Gradients of CSF measures are currently used as an aid to monoamlne metabolites and =lad in cembroOnal ftuld suicide risk prediction in some highly special- of psycNatric patients and heafthy controls.Pairdlistry Research 191091:7743- ized clinical settings. They appear less likely 13. Loamy MP, Mahe RC, Davis KL Seasonal varia- to be useful on a larger scale, because of the tion: of human hraftir CSF neurotransmitter metabolite need for strict standardization of the proce- concentrations. Psychiatry Research 1994;12:79-87. 14. Mooted F, Raffaele R Falespede A. Pad R Cir- dure. An important research task would (*dim variation in 5-hydroxy4ndolooetio acid levels In seem to be to identify other markersof the human cerebrospinal fluid.European Neurology 1 991 20134-13311. serotonin system that can be measured 15. Bowers MB Jr. Cerebrospinal fluid 54iydroxy4n- repeatedly over time in large groups of sub- doles/A.9a add (5-itAA) and homovantillo sold (tWA) fol. =apjegneprobeneold in Woofer depressives treated with ject& .Psychopharmacclogia 1972:23:2633. 18. Pot RM, Goodw'on FL Effects of amitriptyline and The potential for treatment and prevention imipramine on amine metabolites in the cerebrospinal of suicide remains to be explored. fluid of depreelagnts. Archives of General Psychiaby 1974;30: 17. Muscatels 0, Goodwin FK. Potter WZ, Owe MM, Markey SP. knipremlne end desipramine In plasma and spinal fluid.Relationship to clinical fesp00110$ and REFERENCES serolonin metsbollsm. Archives of General Psychiatry 1978;38.021-1125. 1. Farberow NL Bibliography on suicide and suicide 1897-1957 1958-1907. Washington DO Ns- 18. Treakman L Asberg M, Balloon L, et al. Plasma Imis:111uteof Mental Health, 1909; DHEW publication levels of chlorlmipramineandittext=7 metabolite no. pro 1979. during _treatment of biochemical and cilnioel WedsofdiCtii:Olimpounds.Clinical Phar- 2. Bowers MB Jr, Gedsode FA. The relationship of macology and Therapeutics 1979:M0041a monoamtne metabolites In human cerebrospinal fluid to age. Nature Mit 919:1250-1257. 19. Bedlam L, Tuck JR, SWIM B. Biochemical effects of zimelidine In man. European Journal of Clinical 3. Gottfrbe CO. Gottfdes I, Johanson it Olsson R, Pharmatelogy 1980.18:453-487. Persson T, Roos B-E, Spat/cm R. Add monomine meta

23 7 2-203 Report of the Secretary's Task Force on Youth Suicide

20. Persson T, Roos B-E.Acid metabolites from 39. Banid CM, Prato PA, Popp Z, Kyles M. Biochemical monoamines in cerebrospinal fluid of chronic markers in suicidal patients.investigations with schizopluenics. British Joumei of Psychiatry 1909 11595- cerebrospinal fluid amine metsbolltm and neuroen- OIL domino tests. Journal of Affective Disorders 1084:&341- 21. Chase TN, Schnur JA, Gordan EX Cerebrospinal 350. fluid monoamine catabolites in drug-Induced 40. Lopez-tor M k, 8.1:14kilz J, Perez do los Moe eigg=ree.disorders. Mauro- peyohopharmacology JC. Biologiod oorrelations of suicide end in 1 , major depressions (with melancholia): ft- 41.0mireisloMaset ggroria.n.intilsolx:ttt aited doles:MID acid and NNW in cerebral fluid, dexamethasone suppression test and &pouffe sychotio patients treateT1 Za; phenothiazinn. Witrto Nrequan. Neuropsycholdology 19732t3094119. 67-74. 23. Fyro B, Wode-Helgodt B, Borg 8, Bedroll G. The 41. EdmanAmbers M, lavender 9, Saha% D. Skin effect of chlorpromazine on hamvuWft acid levels in conauctanos NriMon and ogre fluid 5- acid in sukAdsi kr:hives of osrebrospinal fluid of schizophrenic patients. =121deptistralgt(in Psychopharmacologia (Barl.) 1974;W:267-2N. press). 24. Bulat PA, ZhrkovloB.Orkpin of 5-hydro*n- WfalerVilard P. Sorensen T. Hoppe E, Rafaeleen dolma* acid in the spinal fluid. alum. 1971 ;17aM- 0J, Yates CM, Moslem N. Blovenic amine metsbolltes 740. in ostebrospinal fluid of patients with affective disorders. Acta psychkAca scandinavlos 1076;58.6896. 25. Stanley PA, Traskman-Bendz L Dorovini-Ze K. Correlations tvetween aminergic metabolites sinful- 43. Royasine P, Post RM, Rubinow DR, Unnolla PA, Savard R, Davis D CSF SHIM and personal and family taneously obtained from humen CSF and brain.Ufe history of Weide in alloWvNy II potion* A negative Sciences 1986;371279-1286. study. Psychiatry Research 1031293-224. 25. Fame BF., *atom R. 5.hydroxylndoimostio acid (and hontove acid) levels in the cerobrospinal fluid 44. Perm H. UN at risicMotors of suicidallty in after probenecid application In patients with manic- depression. Psychiatric Developments 195 at87-104. depressive psychosis.Pharmacologla Clinica 45. Boskow J: Suicide and mental disorder in Swedish 19M1:153-15$. men. Acta psychiatric. scandinavica 1919;SuppiZI7. 27. van Praag HM, Korf J, Schut D.Cerebral Mk Brown GL, Goodwin FK, Ballenger JC, Goyor PF, monosminee and depression. An kmetlgation with the Major LF:Aggression In humans correlates with probenecid technique. kchives of General Psychiatry cerebrospinal fuid amine metsbolites.Psychiatry 1973;28927431. Research 19791:131-139 29. Goodwin FK, Post RM, Dunmr DL., Gordan EK. 47. BrOWn OL, Ebert MH, Gayer PF, et al. "egression, fluid amine metabolite' in ~v. iliness: suicide, and suotanin: Relationships to CSF amine me- TheoCersil=oldtechnique.American Journal of tabolites. American Journal of Ptrachistry 1952139.741- Psychiatry 197313D73-79. 745. 29. M, Traskman L, Thoren P. 5-HIM in the 48. van Prase HM. CSF 5-HIAA and suicide in non- fluid: A biochemical suicide predictor? k- depressed sohlzophrevice. Lancet 1993*977476. chives General Psychiatry 1971%3&1193-1197. 49. Mnan PT, van Kammen DP, Sahel/an it Limon* 30- van Prase HM, Korr depressions PA, Bunney WE Jr, Goodwin FitCSF 5-hydroxyln- wfth and without disturbances in the deleacetio acid in suicidal schizophrenia patients. motabrIsm: A biochemical classification American Journal of Psychiatry 1954;t4t. macaagia (Berl) 1971;191411-152. 50. Sedvsol 0. Viotio-Fisigodt B. Absnent moncemine 31. Asbarg M, Thom P, Traskman L, Benham L, metabolite levels in C3F and family history of Ringberger V. llierotonln depresolore - A biochemical sub- schizophrenia.Their relationships in schizophrenic group within the affective disorders?Science patients. Archives of General Paphistry 198007:1113- 1976;191:4711-480. 1110. 33. Gibbons RD, Davis JM. A note on the distribution- 51. Roy A. Ninon P, Mazonson A, et al. CSF al form of the Alborg et ei. CSF monoamlne data. Acta moncramine metabolites in chronic schizophrenic patients psychiatric* scandinevice (in press). who attempt suicide.Psychological Medicine 33. lyeprerneln Symptom patterns in unlpolar and 1985:15:3351440. bipolar with monoamine metabo- 52. Post RM, Ballenger JC, Goodwin FK. lites in the comb!le=.tial:prrets U.Suicide. Psychiatry Cerebrospinal fluid studies of neurotransmitter function in Research lint* manic and depressive Mmes.In:Mind Jti, ed. 34. Traskman L, Pabere_PA,_ Bertlisson L, *strand L of cembrospkwal fluid. I. New Yak Plenum Monoamine metabolites In0SF and suicidal behavior. N- ruff: 1=717. OM* of General Psychiatry 1901;381131435. 53. Asborg PA, Bertileson L Mortemart B, Scalia- 33. van Prue HM. Depression, suicide and the meta- Tomb* G-P. Thoren P. Traskmen-Bedz L CSF bolism of serotonin In the twain. Journal of Mahe Die- monoamine metabolites in melancholia. Acta orders 1=4275-290. psychiatric. soandinevia. 1956;59:201-219. 54. Asirrg PA, RinOlow V-A Sjoqviet F, Thoren P, 35. Montgomery SA, Montgomery D. Pharmso:21; Traskman L. Tuck JR.Monoamlne metabolites in cal prevention of suicidal behavior. Journal of =tiviallhflieldand ferule(*) u=dinhibition during Disorders 19124:291-29111. Pharmacology 37. Palaniappan V, Ramachandran V Somasundaram andTherapeutloci1977°11nirr:20 O. Suicidal ideation and blogaenio amines in depression. 55. Bjerkervelatit L. Edman G, Rydd ftenkM Indian Journal of Psychiatryus3,2525e-291 SedvaliG, Wiese. FA. Clinical and ~mama effects of 311. Lockman JF, Chamey DS, Nelson CR, Henkver dteloprem, * selective 5-HT rouptake inhibitor - A dose- GR, BMWs MB W. CSF tryptophan, 5-HIAA and WM in response iMady Wooed pollen* Psydiopitanneocl- 132 patients characterized by diagnosis and clinical state. ogy 1985;87: Recent Advances in Neuropsychopharmacology 1961;31:299497.

BEST COPY AVAILABLE 2-204 238 MAsberg: Neurotransmitter Monoamlne Metabolites- oorrixosfdRolA, Agren H. Picker D, et at.Reduoed Mk BMW Mr& Langer fa, maroon R. Sechta D, fluid concentrations of homovanillia acid Zerifian E. T Wed imipramine binding sites are and hormonal* acroxvindoluoitioacd ratios decreased hoszt=itunftealed depressed patients. kt depressed patient* to auloidality end Silence 1 dezsmelhasons nonsuppression. Journal of 78. Paul SM, ROW M, Skolnidc F, Ballenger JO, Psychiatry (In press). Goodwin Fit Deposed paints have dimmed binding 57. Nino PT, van Kennon OP, Unnolia M. Letter to of Wigged knipramine_ to Oditke serolanin 'transporter: the Edict. Montan Journal of Psychiatry 198514214$. Archives rg General 1961 alk 1315-1317. 55, Ostraff Oiler E, Bonen It Ebersole E, Hart- 77. Titania, J, ?Mahon E. Deoressed uptake of 5- nese L Pillion J. Neumendocrine tisk factors of suicidal nYcirallkle in blood platelets from depressed behavior. Amedoen Journig of Psychiatry 19821391323- pidierft Nan 1975:262806-500. 1,215. 71 Caxton A, lahovotaittg$0nWood K.Platulet 5- 50. Ostroff RB, Geier E, Holmes* L, Mason J. The in depreealve laws narepinephrine- to-epinephrine ratio in patients with a his- =a1M1PliwnmicallAt197107:165.1111 tory of sukids att227empts. American Journal of Psychiatry 79. Wilson L, Tyteing 0, Braithwahe R Trackman- 1985;14Z224-. Bondi L, Pabsrg M. Urinary exciedon of 60. Pond AL Nsuroendocrtne differences between loisaaetio acid - no relationshitoth5litirlic:i violent and non-violent paracuicides Neuropsychobial- cerebrospinal fluid.Acta scandinavica ogy lama 157-isa. tan:WI-00498. 51. Treakman L, TOIfte. Albin P4, Bedlisson L. 80. Niemen R. &ley MS, Bombing F, Socha, D. Lent* 0, Scheibe D. Itoi in the CSF of depressed Zarifian E, tenger 3H-Imipramine and and suicidal patients.Archives of General serotortin upteicein platelets from depressed and PsYchkdrY1901%37:7111767. control volunteers.Payohopitemiscalogy 9S277:332- 62 Bead CM, Vojnik M, Papp Z, Balla KZ, Nato M. 335. Cerebrospinal fluid megnselum and calcium related to 81. Oreland t., %%erg A, Paberg PA, et al. Platelet MAO amine metabolites, diagnosis end suicide attempts. activity and monoemine metabolites in cerebrospinal fluid Biological Pap:Natty lettattemn. in and suicidal patients and in healthy controls. Psychiatry Research 1081;4:21-20. 63. Beckmann H. WeBabn L Gattaz WF. Molatonin kwmunoreactivity in oereol fluid of schizophrenic 82. Zuckerman kt, Bolinger JC, 'Amerman DC, patients and healthy controls.Psychiatry Research DI., PoetI.A corrigNome lest b hunwro or thebioMurigghr- 1064;11:107-110. cal models of sensation sssidng, knPuleM1Y, bnd en/4/1. in: Zuckerman kk ed. of Sensation - 64. Beck-FrUs J, Kletimen BF, Veda B, 0 al SWUM ing, kmpuldvity and PixTói .Hinsdale New Jersey: melaionin in relation b anical variables in patients with major depressive disorder and a hypothesis of a low Laurance Mean Atecutistas 1963:22944S. melatonin syndrome. Acta peyontainca scandinexice 8,3.Meltzer HY, Perlin. R, Tricou BJ, Lowy PA, 10887t:310331#. Radon= A. Effect of 6-hydroxylrypeophan on serum cortical levels in mew affeatke distudere.5.Relation to 65. Welloterg L, Beok-Rils L Veda B. Penman U. mid* Psychoels, and depossive:319461nplome. kchives Melatoninloortisol ratio in depression.Lancet of General Perinkttry 1084;41 1979:1:1351. 84. kninoll Mt. kitbag M Beidsson L Booth P. 08. Claustrat 0, Chazat 0, Brun J, Jordan 0, Suedes Marlensson 8, Traskmartamdz L CSF monoamlne me- G. A chronobiological study of meialonin and carting tabolites and the dexamethasone suppression test searetkm in depressed subjects: Plasma melatonin, a Manusalpt 1901 biochemical marker in major depression.Biological Psychiatry 1064;11k1215-1228. N. Carroll BJ, Oroden JF, Haskett R, et al, Neurotransmitter etudes of neuroendocelne pathology in et Kamm 0, Glig T, Beenmengsr W, %Nan W. deposition. In: Svensson TH, Cartoon A, ads Blogamic Postmortem bloom* mines in CSF of suicides and con- amines and affectIve disorder. Acts psychiatric* wan- trots Poster, 144h Confines of the Caesium Internation- dinavioa 1000.11(Suppl 280):183-191 ale Neuro-Psychophannacologicum, Row= 1984. 88. WM CM, Arato M. Amine metabolites and OIL Amp ta, Mut &Solon), P, Somoavi _antakiel neuroendoccine responses related to 9=232.and 19...4adVer K Postmaten neurochemical imie of suicide. Journal of Menlo Disorders 1 behavior. Abstract First European Research of SuicidalBehavior.81=kr19121,1 87. Bold CM, kets M, Papp hunt IL The influence TarkimMunichFRCL of dexamethesone on osobrospinal Agd monoemine *bogies and ooltisol bridges* patients Pharmsocp- (XL Olenw E, Cook NC. Attempted suicide. London: Chapman & He, 1951 eYddidde $8. Joseph MS. Brewenan TD, Noss VI, Stebbins GT. 70. -7Mw R. Evaluation of suicide prevention after Plasma Letryptophant neutral amino acid ratio and attempted W. Ma psychiatric* swanks. 1975; dercarnethasone suppression in depression. Psychiatry Sup,* MO. Resesich 1084:11:185-192. 71. Polcomy AD. Prediction of suicide in= 89. Meltzer HY, kora RC, Tricou BJ, Fang VS. patients of General Psychiatry tan . Serotonin uptake In blood platelets and the 72. Buik F. Kurz A. Moiler 114. Suicide risk eosin: do clexantsihnone suppression lest in depressed patients. they help to predict suiddal behavior? European kohives Psychiatry Research 196t8:41-47. 00 Ps .ydy and Netilalogical Sciences 1985235:153- 90. Meltzer HY, Koenig J1, Lowy M, Koyame T, 157. Robenson AG. Serolonergio oneasmikto challenges 73. Cohen 1L Statistical approaches to suicidal risk- in affective disorders Abstract, Mi %told Congress of factor analysis. Proceedings of the New York Academy of Biological Psychiaby, Philadelphia, Penn. USA. Sep 9.13, Soleness (in press). 109& 74. Nitod K, Coppen A. 9100hemical abnormalities In 91. GOO PW, GoOdwin FK, VA* T, Reber R. Pitultery depresehre Diem ft9lophsi and =response to ftwobv01-410aft hormone in In: Curzon_ CI, ed. The blochembft of distur- Onset Relatlon to spinal fluid amine meta- OgdwOw: John Valey k Sam Un.. 1900 :1343. bolites American Journal of Psychiatry 1977;134:1028- '031.

239 2-205 Report of the Secretaty's Task Force on Youth Suicide

92 Carlsson A, Bionnerhoint L, trAnblad B. Seasonal 111. Ucltran) L. Mien M. -Stoneman LB 5- and cirosdian monoemine vaditions b human brains ex- hydroxylndoissostic mid levels in suicides who amined wands& In: Svensson TH, Carlsson A, eds. have load trek chblren. Land 1084. @Jogai* mines old illative disorders. Oats psychiatric& 113. Branoliejkli Branchey Shaw S, Lieber CS. soindinavica 2110)150- , end aggression in alcoholics and VAnsiusibeAlfdahtat R. Seesonallty in blootrimi- ZrnillitiOnhlpto plasma amine acids.Prishisby oil determkidions: a come of Wino* and a clue to the Research 1984; 12:219426. temporal Molding* of affectivs illness.Psychiatry 113. amid CM, Arab M.Relationship betwoen Research 1971t111340. owebrospinal firrid amber metabolites, wounds:vino 94. Pam RC, Kneel L, Meltzer HY. Seasonal variation and personsOly amnions (Msolte4imany scab of eemeonin uplift In normal win* and depressed fin%el inrgtodsin Mist=lents. Acta psychiatric. scan- patients. Olologiod Psychiatry 103409:795404. &mica 1 96. Earley D, Deemedt 0, Moulins A, Msndiewice 114. &hailing LI, Asberg M. Edman a Lavender S. Im Cfroannual variations in the dimity of Vitiated knipramine pithily, nonconformity and sensation essidno as related Wane sass on blood plaids% in man. Neuropsychoblel- to biologics( merlon for winsrability. Clinical ogy 1993;ith 101-102. msoology 1984;7(Suppl. 1):748-747. 98. Winks( PMAVarehJJ, Stancer HC, Pentad E Vint it" Ryan E, SchMeiD, Poboni M. Rorichaoh rOw CK. fissions/ variation In qaMot 3141mipramine binding: In depressed and with low levels of IF =rvalues in oci and &paned population& hydroindoleacetioscriertrierebrospinalfluid. lieseerch 111640 1: 127-131. Psychiatry Research 11111:47229243. MIL Johanson B, Ron B.B.fehydroxybdoissostio AbtrmalV"of PeallIndo=st=Pglia=s14.in k- and homovaniltio sold levels in the cerebrospinal fluid of chives of General Pairehlatry 197226474479. healthy votuntenrs and patients with Parkinson's 98. ven Prue HM.Significance of bloohomical syndrome. Lib Sciences 198M1449-1454. parantaters In the dwells, troWnewt, and prevention of 117. Andersen 0, Johanson BB, Svonnerholm L truth* disorders. Biological Psychiatry 1977;1Z 101- Monne:nine metebolitles in suceessive simples of spinal fluid.A comparison between healthy weluntairs and 99. Traskman-Bendz L Asbrg M. Bertlisson L Thwen pa;:ents with multiple wierosis. kta neurologic* scan- P. CSF monosmine metaboOtes of depressed patients dinevica 1981:n247-254. cchunalcisibeee and aftw recovery. Ass psychittica scan- 118. Haat C, Astrom J, Sedvall G. Psychic disturban- 1964:09:333-342. ces In adult coeliac dissaws.ILReduced central 100. van Prate HM, de Haan S. Central seroknin monownine metabolites and signs of deposition. Scan- tabollern and the frequency of depression. Pewhistry dinavian Journal of Gastmentwology 198217:25-28. Resewoh 1979;1:219424- 119. Andersson H, Roos B-E.Increased ie.* of 5. 101. Sodvall G. ',Afro B, Guliberg B, NybackvItrifs hydroxylndoleacetio sold in owebronoinal fluid from Infan- FA. Wode-Heigodt O. Relationships in Malty tile hydrocephdus. Expedenlia 1905.22749441. between concentrations of monoamin metebolites in 1. Post RM, Kotin .1, Goodwin FK, Gordon E. owebrospind fluid end family hielory of psychiatric mor- afty and cerebrospinal fluid amine meta- bidity. British Journal of Psychitry i0M13M303474. DrnlatagodvsWm. American Journal of Psychiatry 197%1311674a 102. Sidvall 0, haus L Nyback H. et al.Genetic studies of CSF moneamine mslabolilss. Kt Media E. et 121. Mean C. Snore B, Berdisson LSills of lumbar al., eds. Frontiers in binational and phwmacologioelre- levels of monoemine metabolites. search in depression. New York: Raven PIUS 1094:79- E=1:influencesthe Editor.Nohives of General Psychiatry 85.(Advances in Biochemical Psychopsharmacology, 198239:1445. 122. TraskmanBenciz L Depression and suicidal be- 103. Vaizeill L Psychobiology of aggression end havior - a biochemical and OVwmooloOmi ;tidy (Disser- violence. Pow York, Raven Press 1961. tation). Stockholm: Kar. oUnska institute, woo. ei 104. Freud S. Trent und Melancho0e(1917). In: Freud 123. SloquIst 5. Johansson B. A comparison botween A. Mixing E, Honor W, it al., eds. Sigmund Freud fluoremetric and mass fvagm.ntograpltlo determinations Gianni* Werke. %fa 10. 87th Ed. Frankfurt am Main: of homovanlilicr add and 5.hydrnxvindols.oetio acid In Fischer, 197Sk428445. human cerebrospinal fluid.Journal of Neurochemistry 105. Abraham K. *ouch eines Entwicirkmgsges- 197t31:f41-625. chlohte der Wide aid Grund der Pephosnalyse 124. Musidet FAI, Jowl/to FM, Korf J st aL Correlations cher Storun on. Leipzig: intornationaler between a ftuorimetrio and mass irasmontographic S. ed. Niue k. method for the determination of-methoxy-4- en:Wohen roarmlyee; vol. 4. itraopzt=deacitio sold and two mass fragmen- ts the dalsrmination of 3 methorry-4- 109. Weissman M, Fox K, *man (IL Hostility and =henylethylsne glycol in cerebrospinal fluid. depression associmed with suicide attempts. American Neurochemistry 1979:U191-194. Journal of Psychiatry 1973030450455. 125. Mejor LF, PArphy _S, Gordon E. Effects 107. %fest 0.1. Murder followed by sulcido. London: of and web* on '0.4ted metabolites of Heinem nn 1985. no ephrine, dopamine) and serotonin in human 108. an ME. Patterns in criminal homicide. Lan- Dere respinal fluid.Journal of Neurochemistry don: tford LhMmity Press, 1986. 19795-=1. 108 Lind% M, Widninen PA, Sdreinin PA, NuuMa 129. Kap YO4, EtIert MH, Weigh 14, Lake R. Abnor- Rimon A. Goodwin rA. Low serebroWnal fluid 5- malities in CNS melamine meft'n In anorexia ner- hydrinylndoisecotio add concentration Muenthdro kn- vosa. Archives of General Psychiatry 1994;41:360455. phrhe from nod/flout*, violent behavior. Life Sciences 127. aweless D. Traskmarrilendi I, Gilroy 1911%3&2809-2014. J. The Influence of the and *Ong positions on the 110. Lfterg L, Tuck JR freberg U Scalle-Tombe =of 5-HiMand A krmbar cembroepinal Seminar L lione, suicide and OSF 5-HiAiL Acta psychiatry 1084; 111k 1585-1 889. psychiatric& scandinavIca 196511:230-236.

2-206 240

BEST COPY AVAILABLE MAsberg: Neurotransmitter Monoamine Metabolites...

120 Shaw ON, Camps F E, Ecoleston EG.S. 132. Codwan E. Robins E., seratonin hydrowybyptunle In the hindtmin of demob. sues. levels In brain: A comparison of &Addeo and Mesh Journal of Psychiatry 103711&145?-1411. alooholic suickfto with controls. Psychiatry 121 Downs MR, WWI WE Jr, Colbum RW, et el. 107%11 ANN. Notadronaline, 541ydroxyhyptemine, and fihydroxyln- 133, Ikekow J, Gotthies 00, Roos IA Waled 0, doleacetio acid In hIndbadnesuicidal patients. Lancet DsWmination of monownine and menamins metabolites 10502:80151013. In the human brain: post modem studies in s group of subsides and in a control group. PAL Psychlatrict Scan- 130. Pars CPS, Young OFM, Rice IC, Stacey 113.5- notadranalltre and dopamine in dinavioa imesavo. and misdate nucleus of con- 134. Owen F, Cross AJ, Crow TJ, et al. Bmin 54412 tra and of patients oommitling WM* by oodles receptors and subside. Land 100321251 poisoning. Lana* lasktise-las. 135. Korpi ER, Kleinman JE, Goodman Si, et al. 131. Uoyd KO, Fraley U. Dock JHN, Hornyidevrios 0. Serotonin and Ihydrontndolosestle acid concentredons Serotonln and Ihydrostykidolosostio acid in discrete In different brain regions of Welds victims: comparison saes of the Windom of suicide victims and control in chronic schizophrenic patients with suicide es cause of crtlostrot In:Costs E, Gesso OL, Sandlot PA, eds. death. Archives of General Psychiatry (In press). n: New vistas. New York Ravsn Press 10/4:387 307.(Advances in Biochemical Psychopharmatelogy: vol. 11).

Some factors of Importance for the concentrations of 5-HIM and HVA measured In lumbar spinal fluid samples.

Factor Possible control measure

Subject's age and sex (24,6) Matched controls or ANCOVA Subject's body height (5,6) !dem PhysL.al Illness, e.g. Parkinson's disease (116), Physically healthy subjects multiple sclerosis (111), adult coeliac disease (118) and hydrocephalus (119) Drug treatment (4, 15-23) Drug-free subjects Time of day (14) Samples always drawn at the same time Time of year (703) Patients and controls matched for season Dlet a Fasting subjects or controlled diet Physical movement (120) Bed rest prior to lumbar puncture Subs ct's position at lumbar puncture b Same position always used Amount of CSF drawn (122) Same amount always drawn Harding and storage of samples identifimtion procedures for all samples Analytical method (123-125) Best available method

a. It cannot be excluded that a diet particularly rich (or poor) in monoamine precursors might influence the metabolic concentrations. Thus underweight anorexia (lemon patients have sivtificantly lower 5411AA than weight-rammed and long-term weight-recovered anorectics. (126) V. :ikvert ci sL (10) report that the concentration padient of CSPS-111,AA was cosi in the decubitus positIon. Oateless et it (127), on the other hand, found similar gradients in sitting and tying patients.

Table 1.

241 2-207 Studies of CSF 5-filAA in Relation to Suicidal Behavior

CSF sampling and Confounding Author Subjects analysis factors Manure of su)oldatity_ Result

Men a el. (29) hoepltsilated &premed stendwatec% 024453 not contmiled diempled or completed suicide lots 541t0A in the 15 atempisa, plena sea India Mew episode Otelladle, Mow mina adore methods

Stow et it. 1979 (411) 22 men vah poioneilly sandordirec% not coarcited Wears Way at Wade alma 5+2,AIhe 1 1 suicide abounams decoder Summitry

Aaron 1980 (33) 32 depressed Adana elendirdiza% na ooncotied SADO sukidaty wales made, conoludon 5MM GC419 mtil suicidally sonar

Tresernan et it 1981 gig 30 wields Whomplom (11 ateldsallts0I contraloci by NOMA rasa asempled a comploted 5411A4 ewer In both osingodes it MOlor depraeltra 22 Mot CIIC492 tor sea son and tody Wade atemplas dun In hstaby mama psystdado discase Wahl

:1=101/11crh)45 mni.a heathy controls

Leaman et UM (36) tS2 psychlain pateals. otonitorMsoct. ohm not airecited nom aikido/ Wootton or num madam comtladon nes euidda idesdon N NW' &VW* probeancir% Summitry at mg ma% In the 78 psychotic pedants

DrvitiL (a) 12 pitIirh bordeane stonderaza% not coral:Sled Ole time Way of suicide attempt Iona 544AA In the 5 slampters porecnaly dleador GGAIS

Montacrmery snd patients lab endow not rowed not contralled hang et sukrktei act man atemptem among patents nal be Massomey 1902 (36) noun depreseion Cliff 841tAA

%en Pampa 1992 (36) 203 depressed pedonts stenderclized, disc not °unbolted wont suicide edema rINIcandy man suicide ettempters probates:It fluomenetry amone Plitenb rah Ion COP 11411AA

Pato/Wpm tot at. 193 40 hospealized &praised LP proceduns not tufty not crarectied outside Item in the Hamilton need** canehdort beMeon CSF it 649.AA end auk** wow (37) patents describec% Mau /catty Rana Seas

Aaron 1983 (44) 110 depressed patients standandlart catAis not combated BADS euicideilly scales la SHIM assocaled nth mow of 0~4 suicklel kleotion

Table 2.

243

24 2 BEST COPY AVAILABLE Studies of CSF 141IAA in Relation to Suicidal Behavior

Ct3F oarnpfing confounding Author Subjects and analysts %MOM Pilouuro of suicidatity Result

flareyrna et a. too 04 $21epolm 13 lin00111, dandedlest nal cantrollod lestime Way at RAdde Met* no modeion alitt 11411.4A In bipolar paean% In Orme flummery or WIC Mame Phases el 'bete

eon Ram ism (40) 10 nondopeamed schlep inendirdtrod. meshed tar ay went Wok* allanr* loner OW PIM ear mbeneld In ramie who allerzeled ofter probonookt and au seed* allernetes weld* In reeponaa le flummery Imporallm hiluolnelona 10 ranallekel ecalza. Owes, 10 cantres

Owe el al. 1964 (30 141 tsmala hasten* (36 slendenflizet adlused for sue and want saki& alawne novae oonestIon lath 044141 In ell domes* 44 3uoremstry WY Wahl k, diagnose; veva pellculady eel violent tonleenrene, 33 di- /474XNA allaneb azhate 24 with allotment dlsoodsc 40 praMoudy repailat)

Ninon at al 10114 (40) salcalet, 4 noneutaidai eandedized; mated for ape, sem Olsen, history or sued& elempt law 5411AA In wick* stemma lichkeihrliede patients HMG and e)alsal ehararistalles

Lapea4bor a al. 1904 (40) 21 deemed patent' tiandwitizet canbaltd seeds allsmak suicidal kleseon MOM saimpts and War sulakelity floommsay reed en Me Hamann Ore* we scam in pedants nel few 6-14A4 de MVP *non

Roy s( al. 1006 (51) 56 patients *eh ehronie standardized; Ousted le age and Wee* Nem of suicide *Met no Merano* h Seen beteren 27 schizophrene HPLC baly WO by AN- altentatere end 27 noneremplete COVA

Edman az al. tee (4t) 7 Made stentetwe ath standardizot% nuseed tor ea; ego recent puled* Mane lower 541tAA In ettemplers venous peychietna dies- CIC4,49 end body IWO dera 7 healthy control*

Ray a al. 1910 peg 27 damned meek 22 standardized; ectitated far esit and !Veen* Macey re seeds same lowet HVA In the 14 attempters healthy cantata PIPLO see by ANOOVA i Table 2 concluded.

BEST COPY AVAILABLE 245 244 Studies of CSF HVA in relation to suicidal behavior

CSF sampling anti ConfounTng fac- Author Subjects analysis b3111 Measure of sulckiality Result

Brown et al. if179 (sit) 22 men vAth personalty standardised; aucrometry not preened Mistime Way ci Weide Mempt no consistIon nooted decider

Aaron 1960 ($24 23 dammed pewits illandentbot WAN not matrolled 1108 suicidally males no alonbloant constetlons

Yi*tononuIsL 1981 (34) 30 suicide Pampa (5 elanderdaid GOAN controllsd by ANCOVA moat samoted ot compisied liaw then controb in doomed molar deprooke, 22 Ohm for ego sal and body suicide ettemphse, nondsomed strenoters psychiatric &orders Might amllar to controls exchaling achttophronla end alcoholism). 43 he* thy nontlein

to:low et at. 1N1 (30) 132 parchlade paint% standerdiad diet proboneckknot cottatrned Ilem sdal kiestice on nine consalion waled swami dignoess flootometry meg mote

ataran et al. 1452 (47) 12 Miens 01 bordedhe standardizet 0C-149 nitc Maim history of suldde Mon* no conelatlon personeay disorder

liontgornsty and 49 MOM PM en- not moiled not conbolled Way of WPM ad mom allinnMers among patients aMs ClIF HIP Monillemilf IOW (10) &venous dapression Paionappan et al. ION 40 hospitsibid deomed LP procedure not Ray nal =menisci Weida Mem in the Herndon neva* cormidan Maw MP et KVA and &Ode woo ($7) malenra demodbe* Nosonsary Roane Rode

Agren IMO pils 110 deposed pliant, standard:44 not controlled 8A08 suicidelity sodas Ian KVA minalied van high iiielsay at itiC418 NINO atempts lo meant epode

Banal at al. 1034 (39) 141 Waft Implied, (38 slandentireck NOON for age end recont mid*, ascot Inghet KVA fot dog ~dm csou Wen dammed. 48 Nato !biometry bodif Nicht bY $$4' ths depremion 'Prow cdocabie no Ostyak 31 alcoholo, 24 CCM dem out miocistion sath adiudmeM disorder. 45 poMouely repodial)

Won et el. 1905 (57) 8 suicidal,nonsuicidal dendankaick mached fix mac op Weft* halo/ of NO* nelempt no dinsonae to HVA belsonn Iht two echizaphrenic laden* HPLC and physical condition 9161062KIVOS matched ha op end eas

Roy el al. 1958 (51) 54 patients Ph Monk dandardimO eluded for age and lifetime Mabry al midi* Om* no diRtirenai to HVA bedoen 27 wadaticihrenla HPLC atiempas and 27 nonataimptste Mhs4ahe t4Y AN.

Roy et at. 1958 (55) 27 depoted Petienb. 22 standardbod; adimald for ma and Wakes Mao/ ot WON Pawl lower 54844 in the 10 attenolso. Mhot healthy control, HPLC age by MOON the dllferance vao not staddically elpfloant

Table 3.

246 BEST COPY AVAILABLE 247 WO,

Some Controlled Studies of Monoamines and Their Metabolites by Brain Tissue and Cerebrospinal Fluid from Suicide Victims

Author sublati Tissue Result (Concentrations In suickle victims compared to controls)

Sher et al. 'OW OW) 22 suicides. 17 controk Hindbmin &KT Imes

Bourne al it Nei (UM 23 suicide% 25 oonlage HIndteeln 5447 Omar, 5411AA reduced, NA almlar

Pars el al. IWO (1301 21 suicides, 15 cent** Brelnaiern, caudate, 541T reduced, NA and DA *Vier hypolharnue

tag.d al el. WI pro 7 suicides, 5 controls Set mph* nuclei 544T Meer In nude{ raphe dandle end °Weals Inledor 54MA WNW

01~ el len (132) sulcidae, 12 conacee Seteral bob mem 5-HT Winner

Beersaw et at Me (134 23 macaw 52 conbola Severst brain mew 544T Wear efiet adiusbnersi br deasnoe In postmodern deft; NA, CM, 545U and le/A iitI

Men si a Ism (134) 7 NAM" la controls Frardsi calm 544VIA *Mar

MO at et Ian (lM 30 adltrepiverece pc % dead Dans Bawd Wain WW1 544T beer In hspolhalernue In non-echarepreenb suicide Wen* 544IM and TRY *raw suicide), la weeNteplwarde euIddee, 29centrols

Moan et N. 1904 (57) OD suicides and cordrole CetsbreSpinai mid 5 HT. NA, NM higher, DA and A imam,

Armo et at IMO (MD Nal geed Cerearceplead tIud 5-HIAA Nigher

Abbreviations: 5-HIAA so 5 hydroxylndoleacetio acid, 5-HT Es 5 hydroxytrypternfne (serotonln), A adrenaline, DA dopamine, HVA hornovanillic acid, NA sv noradrsnatine, NM vs normetsnephans, TRY In tryptophan.

Table 4.

248 249

BEST COPY AVAILABLE Mortality from Suicide within One Year after Admission to Hospital hi Some High Risk Groups

Patient category Number Percentege suicides

Patients admitted to Intensive care unit after suicide attempt 45

Patients admitted to it psychiatric clinic stet a suicide attempt. 42 2% CSF 544IM above 90 nonomol/L

Ditto, CSF 5411M below 90 nanomol/L 34 21%

Data an suicide attempters admitted to the psychiatric department of the Wolin** Hospital derive from the studies by Alborg et al. (29), Traskman et al. (34), and Edman et al. (41). Data on patients admitted to the intensive care unit of the swne hospital after a drug overdose are given for comporison. Table 5.

Changes In CSF 5.HIAA after Recovey from Depression

Interval between Author Type of 5-HIM measure N examinations Results

Coppon at al. 1972 (97) Baseline 3-HIM 3-59 w Stable over time

von Prow 1977 (99) 5-HIM after probeneold so 6 mo 2 % of patients with low OSP SHIM during illness normalized; the rest remained stable

Post et al. 1990 (52) 544IM after probertecid 11 several months Stable over time

Traskmen-Bendz et al. Baseline &HIM 11 2-7 yr Increased concentrations at foilow-up In patients whose levels were 1094 (99) low during Mow stable In the remainder Table O. 250 BEST COPY AVAILABLE 251 POST MORTEM STUDIES OF SUICIDE

Michael Stanley, Ph.D., Depamnents of Psychiatry and Phannacology,Columbia University and New York State Psychiatric Institute, New York, New York

Suicide is a major cause of death in the and their metabolites, and the findings from United States for the adolescent and young studies that have employed the more recent- adult population. In the 13- to 24-year age ly developed technology of receptor binding. range, it ranks as the second leading causeof death. Furthermore, a dramatic increase in The advantages of human post mortem neurochemical studies must be balanced the adolescent suicide rate has occurred in the last 15 years. In an effort to counteract against those disadvantages inherent in these this growing rate, recent research has begun investigations.In post mortem research numerous confounding variables may con- to focus on the identification of youths "at tribute to inconsistencies within a study or risk" for suicide and prevention of suicide. across studies. In this review, we will assess These studies, which are few in number, have principally investigated personality, the impact of some of these variables (e.g., psychosocial and diagnostic factors. age, post mortem interval, and regonal brain However, suicide research on adults, focus- dissection) and describe their impact on ing on the same factors has had limited suc- several of the neurochemical measures cess and the suicide rates have notbeen described. significantly reduced. Predictors are typical- There are several lines of evidence that sug- ly overinclusive and identify many individuals gest an association between serotonin and as 'at risk' who never commit suicide.Fur- suicidal behavior (2). thermore, the association of these predictors with suicide is too weak to have much utility Because of the involvement of 5-HT in suicide and because 5-HT is a substrate for in the clinical setting. Thus, the traditional monoamineoxidase (MAO-A), Mann and approaches to identifying suicide risk have been inadequate. Therefore, while Stanley (6) thought it would be of interest to conduct a post mortem study of this enzyme psychosocial and diagnostic factors are important to examine in the adolescent and in a series of suicide victims. young adult population, the studyof suicide Two previous studies had examined post in this age group may benefit from an alter- modem MAO activity in suicide victims. native approach that takes into account One study reported no differences in MAO neurochemical factors. To date, there have activity compared with controls (7). A second been no studies of the neurochemistry study found reduced MAO activity in suicide in adolescents. patients where the suicide was associated with alcoholism (8). Both these studies, in The purpose of this paper will be to review the post modem biological findings in the contrast to the study of Mann and Stanley (6), included a significant proportion of field of suicide. This review will include criti- patients who had died by carbon monoxide or ques of studies that have examinedenzymatic findings, concentrations of biogenic amines drug overdose that may have altered

252 2-213 Report of the Secretatys Task Force on Youth Suicide neurochemistry and employed a singk sub- number of studies that have made a strate concentration, a method that is less in- retrospective diagnostic analysis of informative and less sensitive than enzyme dividuals who have committed suicide (941). lcinetic studies. In general, these studies found that in addition to the diagnosis of depression, in- Mann and Stanley (6) assayed MAO-A and dividuals classified as schizophrenic, 13 in the frontal cortex of 13 suicides and 13 alcoholic, and having povonality disorders controls using labeled 5-HT and were also represented. Ito, it is of both phenylethylamine (PEA) 83 substzates for theoretical and practical importance to note MAO-A and BI respectively. The suicide vic- that suicide victims typically represent a diag- tims we studied generally died by determined nostically heterogenous group of individuals. and violent means, with the exception of one With regattl to biochemical findings within overdose. There were no significant dif- this population, the diagnostic heterogeneity ferences between the suicide and control suggests that differences in neurochemistry group with respect to factors such as age, sex may be more related to suicidal behavior and post mortem intoval. rather than to depression per se. The results of this kinetic study show no sig- It should also be mentioned that none of the nificant difference between the groups for studies descri'bed below had youth suicide as either substrate (5-HT or PEA). There was their focus. a significant positive correlation between age and MAO-B Vmax for both groups. There A total of 11 studies have investigated the was no correlation between post mortem in- concentration of 5-HT, 5-HIAA, or both in terval and MAO enzyme kinetics. several brain regions of suicide victims (12- In addition to measurements of the The series of suicide victims included in this n). serotonergic system, 3 of the 11 studies also study are distinguishable from those of other report findings for the noradrenergic and studies of brain MAO in as much as those dopaminergic systems. who died by overdose woe largely excluded, thereby avoiding the potential probkm of With regard to findings reported for the drug effects contaminating the results. serotonergic system, 7 of 11 studies have Other studio have suggoted that MAO ac- reported significant decreases in the levels of tivity was reduced in alcoholic suicides but 5-HT, 5-HIAA, or both. In general, not in non-alcoholic suicides (8). The data decreases were noted in the area of the brain suggest that the reported lowered brain stem (Raphe Nuclei) and in other subcorti- MAO activity in alcoholic suicides, if con- cal nuclei (e.g., hypothalamus). Lloyd et al. firmed, may be related primarily to al- (15) measured 5-HT and 5-HIAA in Raphe coholism rather than to sukidal behavior. Nuclei of five suicides and five controls. Three of the five suicides had died by drug In many of the post mortem studies, which overdose. They found no significant dif- have measured the concentration of 5-}IT or ference in 5-HIAA levels between the two its principal metabolite, 5-Hydroxyin- groups. There was, however, a significant doleacetic acid (5-HIAA). it is important to reduction in 5-HT levels for the suicide point out that some diagnostic information group. Pare et aL (14) determined was available for the suicide victims. These norepinephrine, dopamine, 5-11T, and 5- data indicate that approximately 50 percent HIAA levels in suicide victims who Lid died of the suicide victims were diagnosed as en- by carbon monoxide poisonkig. They found dogenously depressed: the remaining cases no significant difference between the two carried a variety of diagnoses including groups for norepinephrine, dopamine, and 5- schizophrenia, personality disorders, al- HIAA. They did report a significant :educ- coholism, and reactive depression. These tion in brainstem levels of 5-HT, for the diagnostic gtoupings are consistent with a suicide group. Shaw et al. (12) found lower

2-214 253 M.StenterPost Mortem Studies of Suicide... brainstem levels of 5-HT in suicide victims the number of sites (or their density) can be compared with controls, a statistically sig- induced by either chtunic exposure to a nificant difference. However, it should be chemical agent (e.g., antidepressants) or noted that about half of the suicide group deprivation of the particular amine by its died by barbiturate overdose and the other removal (e.g., !calming). Recently, binding half died by carbon monad& poisoning. assays that appear to be associated with pre- (ItniPramine) and Post-(sPiroPelidnl) synap- More recently, Korpi et aL (17) reported sig- tic 5-1IT neurons have been developed nificant decreases in the hypothalamic con- (23,24). Imipramine binding sita have been centration of 5-HT of suicide victims characterized in platekts and various regions compared with nonsuicide consols. Similar of the brain. Sonw of the experimental findings mere reported by Gillis et al. (16). evidence linking imipramine binding with 5- They noted that 5-HT levels were significant- HT is that (1) radioautography studies of 3- ly lower in the hypothalamin of suicide vic- H imipramine binding sites show distribution tim compared with controls. similar to serotonergic terminals (25); (2) Three stndies have reported significant chemical and electrolytic lesions of the reductions in the levels of 5-HIAA in suicide Raphe nucleus cause a significant reduction victims. Bourne et al. (13) measured in serotonin level and in the number of im- norepinephrine, 5-HT, and 5-HIAA in the ipramine binding sites (26); (3) the use of an hind-brain and found significantly lower irreversible ligand results in reduced 3-H im- levels only for 5-HIAA. Beskow et aL (18) ipramine binding and serotonin uptake (27); measured dopamine, norepinephrine, 5-HT, (4) the potency of antidepressant drugs to in- and 5-HIAA in brainstem areas of suicide vic- hibit uptake is significantly corre- tims and controls. They noted significant lated with their potency to inhibit 3-H reductions in 5-HIAA levels for the suicide imipramine binding (28); (5) serotonin group. The results of these studies are sum- only neurotransmitter known to inhibit(Rll) marized in Table 1. (Tables and figures ap- imipramine binding (28,33); and (6) there is pear at the end of the chapter.) a similar pharmacglogic profile between brain and platelet ('H) imipramine binding In many of the foregoing studies, factors such sites (27). as death by overdose or carbon monoxide poisoning, extensive post mortem delay, and The clinical significance of imipramine bind- lack of age-matched control groups figure ing was provided by the studies of Langer and significantly in the interpretation of these coworkers who reported decreases in the &dings. These variables may also account in number of binding sites in the platelets of part for the lack of uniformity of findings depressives (29). The combined association among the post mortem studies. In addition of imipramine binding with 5-HTfunction, as to these potential sources of error, the levels well as the significant reduction in binding of monoamines and their metabolites are density in depressives, suggested the pos- known to be influenced by Won such as sibaityof alterations in imipramine binding in diet, acute drug use, alcohol, etc. While it is suicide victims. To test this hypothesis, Stan- possible to control for the acute influence of ley et al. (30) determined imipramine binding these factors in (SF studies, for obvious in the brains of sukide victims and controls. reasons this is not the case in post modem as- Because of the problems previous research sessments. In an effort to minimize the im- groups bad encountered conducting post pact of the aforementioned variable, we mortem studies, we took particular care in decided to examine a systemreceptor bind- selecting cases for this study. Thus, there higwhich has been shown to be generally were no significant differences between the nonresponsive to these acute influences. two groups with respect to age, sex and post mortem intervaL The suicide victims chosen Binding studies have shown that chann,es in for this study bad died in a determined man-

2-215 254 Report of the Secretary's Task Force on Youth Suicide ner (e.g., gunshot wound, hanging, jumped the cortex of suicide victims compared with from height) and, as is the general practice of controk In contrast to the findings cited these researchers, the control group was above, one study has reported an increase in chosen to match for sudden and violent imipramine binding in the brains of suicides deaths. compared with contrul (33).Possible explanations offered to aderess this discrepant The findings indicated &significant reduction finding include single point analysis insWad in the number of imi binding sites in of saturation isotherma acd inadequate frontal cortex (suicides Th gs 330 ± 39 matching of factors such as age, gender and finoleimg/proteirc controls = 5871. 75 post mortem interval.In summary, five finole/mg/protein) with no difference in published pst mortem studies have binding affinity (Kd) (Fig. I). The results of measured imipramine binding. Thus far, this experiment seem to be consistent with four of the five studies reported a decrease the accumulating evidence sugpsting the inin imipramine binding and one study found volvement of 5-1I1' in suicide. Specifically, an berme. And, as was the case in those reduced imipramine binding (associated with post mortem studies that measured levels of presynaptic terminals) may indicate reduced 5-HT and 5-HIAA, none of the aforemen- 5-1IT release and agree with reports of tioned studies had youth suicide as their reduced post mortem levels of 5-1IT and 5- focus (2). HIAA in suicides as well as Iowa levels of 5- HIAA in the CSF of suicide attempters. In addition to assessing of post mortem presynaptic function of the 5-/U system in Since the completion of this study, there have suicide, Stanley and Mann also been four other studies that have measured post-synaptic 5-HT binding sites using imipramine binding either in suicide victims spiroperidol (5-1IT2) (34). 5-HT2 binding in or in depressive persons who died from animals has been shown to change in natural causes. response to chronic antidepressant treat- Paul and coworkers (31) measured im- ment and lesioning of 5-1IT nuclei (26,34). ipramine binding in hypothalamic In this study, suicide victims were compared membranes from suicides and controls. Both with controls, and, as in previous studies, groups were matched for age, gender, and both groups were matched for age, sex, post post mortem interval. Imipramine binding mortem interval, and suddenness of death. was significantly lower in the brains of the Also, care was taken to select subjects who suicide victims compared with controls. This had died by nonpharmacologic means. group also measured desipramine binding in the same samples and noted no significant The study found significant increase in the difference between the suicide and control number of 5-HT2 binding sites in the frontal group. They interpteted this finding as argu- cortex of suicide victims with no change in ing against the possibility that the reductions binding affinity (Fig. 2). they had observed in imipramine binding Because many of the brains had also been could be attributed to a drug-induced effect. used in the previous report on imipramine Perry and colleagues (32) measured irn- binding by Stanley et al. (30), the researchers ipramine binding in the cortex and hip- were interested in assessing the degree to pocampus of depressed individuals dying which these measures of receptor function from nonsuicidal causes. 'They reported a correlated. -I found that the numberof significant reduction in imiprainbe binding binding site (ax) for 5-HT2 and im- in the depressive group relative to a non- ipramine was negatively correlated.This depressed control group that had been finding is of interest because it closely paral- matched for age, sex, and post mortem inter- lels the experimental observations noted in val. Cf0W et al. (20) also reported a sig- animal studies. Brunello et aL (26) lesioned nificant decrease in imipramine binding in

2-216 ALStat4ey:PostModem audios at Suicide... the Raphe nucleus of rats using the 5-/IT was nonpharmacological (3). selective neurotoxin 5,7-dihydroxytryp- Scatchard analysis of the binding data indi- tamine. Uva weeks following such lesions, 5- cated that there were no significant differen- HT levels were significantly reduced. The ces in the mean number of binding site same members found significant reduc- (Bmax) between the two groups (mid* vic- tions in imipramine binding (associated with tims, 493 fmole/mg protein, and control sub- presynaptic serotonergic terminals) with sig- jects, 492 fmolefing protein) (Kii) between n fica nt increases in 5-HT2 binding the means of the two groups (suicide victims, (postsynaptic). They suggested that the in- 14 pM, and control subjects, 1348 pM) (Figs. crease in 5-11T2 binding might reflect a com- 3 and 4). pensating increase in postsynaptic binding sites secondary to a loss of presynaptic input. Correlations between Bmax or Kd and either Extrapolating to human data in suicide vic- the suicide victims or control subjects were tims where Stanley et al. observed an in- not significantly related to factors such as age crease in postsynaptic binding sites es well as and interval between death and autopsy. a decrease in presynaptic binding sites, it may However, when both groups were combined, be that the functional consequences of this Bmax was significantly correlated with time receptor arrangement could result in an between death and autopsy (r = 35, p< .02). overall hypofunction of this system. Thus, Comparisons between Bmax value& of reduced levels of 5-11IAA in the CSF of suicide victims who died by violent means suicide attempters as well as reduced levels (gunshot wounds, hangings, or jumping from of 5-1IT and 5-HIAA in the brains of suicide height) and of controls who had died either victims would be a logical consequence of a by violent or nonviolent methods revealed no hypofunctioning serotonergic systems. significant differences. Variations in mus- Subsequent to the study done by Stanley et carinic cholinergic binding as a function of al. (34), there have been twa additional the time of day that individuals died have reports of 5-HT2 binding in suicides. Owens been reported (36). The Bmax values for the et al. (19) reported an increase in 5-HT2 combined samples (suicide victims and con- binding in nonmedicated suicide victims. trol subjects) were examined at eight, Crow and colleagues (20) found no change in sepatate 3-hour intervals by one-way analysis 5-HT2 binding between suicides and controls of variance; none of the intervals significant- (2). ly differed from each other. In addition to examining serotonergic bind- No other studies have estimated QNB binding sites in suicide victims, muscarinic binding in suicides and controls. Kaufman et al. ing in this group was also measured (35). The (37) determined QNB binding in three brain rationale for this assessment was based in regions (including frontal cortex) in suicide part on the seVerb-IICSOfChOlinCreesen- victims and found no differences between the sitivity with affecrivt disorders and the high groups for any of the regions studied. In con- incidence of individuals diagnosed as having trast to our findings and those of Kaufman's, an affective disorder who subsequently com- Meyerson and colleagues (33) reported a sig- mit suicide. nificant increase in ONB binding in the frontal cortex of a small group of suicides not In this study, muscarinic binding was es- adequately matched for factors such as age, timated using the reversible antagonist 3- sex and post mortem interval (2). quinuclindyl benzilate (QNB). Samples of frontal cortex from 22 suicide and 22 controls More recently Mann et al. (38) have matched for age, gender, post martem inter- measured beta adrenergic receptors in val and suddenels of death were used in this suicide victims in the hope that such studies study. As previously, care was taken to chose might indicate the functional status of central a majority of cases where the cause of death catecholamine neurons in suicidal behavior.

258 2-217 Report of the Secretwys Task Force on Youth Suicide

It has been suggested that down-regulation They also found that the ratio of 5-HIAA/5- of beta adrenergic receptors may be linked HT, an estimate of serotonin turnover, was with the therapeutic effect of antidepres- uninfluenced by age. Severson did note a sig- sants and that chango in these receptors may nificant positive correlation between im- also relate to the neurochemical substrate of ipramine binding and age (range 17-100 suicide and depression. years). Severson's and Stanley's findings of a positive correlation between these variables The researchers measured beta adrenergic is of interest because the findings are in the receptor binding in the frontal cortex of opposite direction of those reported by suicide victims and controls using Langer and coworkers (23) for imipramine dibydroalphrenolol (DHA). There was a 73 binding in the platelet. Langer (23) reported percent increase in beta adrenergic receptor that platelet imipramine binding decreases as binding in suicide victims compared with con- a function of age. These discrepant findings trols. are of interest because imipramine binding in In addition to this study, Zanko and Beigon the platelet and the brain were thought to be (39) t. an increased number of bind- identical. Thus, findings such as these raise ing sites (' ) with no change in IQ in a questions about the validity of peripheral small series of six suicide victims and matched measures as indices of central systems. A sig- controls. In contrast with the above studies, nificant age-related decrease in 5-Nr2 bind- Meyerson et al. (33) reported no alteration ing sites in frontal cortex (r = -0.42, N = 34, in DHA binding in suicide victims. Thus, two p <.01) was observed in the study of Mann of three studies measuring beta adrenergic and colleagues (38). A statistically sig- receptors report an increase in binding in nificant increase in cortical DHA binding suicide victims. It should be noted that ante with age was seen in our study (r = 0.60, N = morkm use of antidepressants would not ex- 19, p <.01) (38). plain the receptor alterations we observed. Another area that also represents a potential Data from animal studies indicates that problem in post mortem research is that of chronic antidepressant treatment causes a post mortem interval (PMI) that tine be- down-regulation of beta adtenergic ecep- tween death and the time the brain tissue is tors. Fmdings in suicide victims studied by removed and frown. The human post mor- Stanley et al. indicate alterations in receptor tem studies conducted by Stanley et al. (41) binding in the opposite direction from that assessed the influence of post mortem inter- which would be expected if drug effects had val on 5-HT and 5-HIAA levels. Their post been present. mortem interval was approximately 15 hours iiaving set forth the principal neurochemical with a range of 6 hours to 45 hours. They findings in suicide raearcb, it is important found that there was a significant positive also to examine some factors that may exert correlatfon between frontal cortex 5-HT an influence on some of the measures pre- levels and PML No significant findings were viously described in this review. noted for 5-HIAA levels with PML A preliminary analysis of 50 cases with an age Severson and colleagues (40) also found that range of 16 to 79yean revealed no significant PMI was related to significant changes in 5- relationship between age and 5-HT or 5- HT levels. However, their findings were in HIAA (2). However, imipramine WU posi- the opposite direction from that which Stan- tively correlated with age. Severson et al. ley and colleagues (41) observed- namtly, (40) recently published data on age effects they reported a significant decline in 5-HT and 5-11T and 5-111AA levels as well as im- with increasing PML One possible explana- ipramine binding in human post mortem tion for this discrepancy may be the dif- samples. They too noted that age did not ap- ference in the length of PMI between the two pear to influence 5-HT or 5-HIAA levels. studies. In Stanley's study, the PMI was ap-

2-218 257 f!,

10.StenierPost Modem Studies of Suicide.. prcadmately 15 hours, while in Severson's fashion for regional concentration differen- study PMI averaged 36 hours (in some cases ces of 5-HT and 5-HIAA levels (43). Addi- >72 hours). It may well be that while &HT tional significant variation was noted within levels appear to rise initially with a shorter the frontal cortex with a rostral to caudal in- PMI they subsequently fall with a more ex- crease in 5-HIAA levels; 5-HT levels wete tensive delay ( >1-1/2 days). In any case, pre- consistent. Therefore, differences are found vious research has Shown that amines, such not only among the general areas of the car- as DA and 5-HT, are more sensitive to PMI tes, Le,frontal, temporal, and occipital, but than are their acidic metabolite& Wilk and significant differences can also be found Stanley et al. (42) had previously published a within each are& study that assessed the influence of PMI and One of the potential criticisms of post mor- DA, DOPAC and HVA levels. 'They found tem studies is that their findings lack a proven that DA levelsnot DOPAC or HVA clinical utility, es no means of monitoring the were more hiely to be influenced by delay. alterations reported is provided. Thus, the Also, those same researchers published a clinical significance of post mortem findings similar study on the influence of PM on 5- must be inferred. (Both the ease and correct- HT and 5-HIAA levels. Again, in general, ness with which these inferences are drawn they found a significant change in 5-Hf, but remains largely untested.)Therefore, it not 5-HIAA, levels. would be useful to develop a method whh In contrast to variations in the concentra- clinical application that could be used in post tions of biogenic amines, it has been observed mortem studies. that most of the binding sites are In ante mortem studies, biogenic amine me- uninfluenced by post mortem interval (30). tabolites in CSF are generally regarded as the Thus, with the exception of QNB, vch dis- best indicator of neuronal function in the played a modest decrease in binding density brain. One way of testing the strength of this with increased post mortem delay, im- relationship is by simultaneously assessing ipramine, 54111, and beta adrenergic bind- the CSF and brain levels of the same meta- ing were not affected. bolite in the same individual. Another area of post mortem research that Stanley et al. (41) measured the acidic meta- can result both in variations within and be- bolites 5-HIAA and HVA, the principal me- tween studies, is nonspecific or 'regional* dis- tabolites of serotonin and dopamine, sections. In our animal studks (43), it had respectively, in the lumbar CSF and brains of been our piactice, and that of otherr, to the same individuals at autopsy. The post analyze samples taken from general areas, mortem lumbar punctures and samples of e.g., frontal cortex. It occurred to us that our frontal cortex corresponding to Brodmann's lack of precision in dissection might account Area 8-9 were obtained from 48 individuals for some of the variability we had observed (37 men and 11 women). The average age from time to time. was 37 (12.6, S.E) years with a range of 16 mpt to investigate possible regional to 78 yeats. The causes of ckath among the diffrences of MIT and 5-HIAA concentra- individuals in this study were generally sud- tions within the cortex, we dissected den in nature, e.g., homicides, auto accidents, homogenous samples corresponding to Iron- etc.The post mortem interval between tak temporal, and occipital cortex. In this ex- death and tissue collection for the individuals periment, the frontal cortex showed in this study ranged from 285 to 1,815 significantly higher concentrations of 5-1IT minutes and averaged 891 I 58 (S.E.) and 5-HIAA compared with temporal or oc- minutes. cipital samples.In r second experiment, Lumbar CSF samples were obtained. At three progressive 1 mm slices of the frontal autopsy, once the organs were removed from cortex were examined in a rostral to caudal

4k 2-219 258 Report of the Secretary's Task Force on Youth Suicide the chest and abdominal cavities, an 18 gauge dysfunction. With regard to youth suicide, it spinal needle was inserted in the L-3, L-4 should be emphasized that no post mortem inter space. A 10 cc syringe with a leur-lock biochemical studies have been conducted on sop cock was attached to the needle and this age group. We are, therefore, left to used to withdraw samples of CSF. (Figure 5) speculate whether the alterations reported in adult suicide studies wffi also be found when The results of this study indicate the presence appropriate youth suicide studies are con- of a significant correlation between CSF and ducted. As has been previously noted, some brain levels of 5-HIAA and HVA, r = 0.78, of the relevant neurochemical measures are p <.001; r = 0.35, p <.02, respectively (Figs. known to be influenced by age, e.g., im- 6 and 7). ipramine binding increases with age. There- In addition to the principal aim of the project fore, it will ptobably be necessary to conduct presented above (i.e., the assessment of the normative studies for many of these relationship between metabolite levels in measures, either separately or in parallel with (SF and brain), Stanley et al. (41) were also comparaCve studies of suicide victims. interested in determining the degree to One of the frequent criticisms of biochemi- which post mortem CSF measures agreed cal post mortem studies is that they fail to ob- with the CSF findings of ante mortem tain diagnostic information.This studies. In this regard, some of the findings information is critical if we hope to relate that point out similarities between these biochemical findings either directly to suicide results and those obtained from living in- behavior itself or to specific diagnostic dividuals are: (1) a significant gradient in me- groups. tabolite concentration in serial samples of CSF (Figs. 8 and 9); (2) the mean CSF con- It should be noted that this term "suicidal be- centrations of 5-HIAA (34.4 nem!) and havior" encompasses a complex array of HVA (71.6 ng/m1); (3) a significant correla- symptoms. Previous studies describing a link tion between the post mortem CSF con- between suicidal behavior and serotonin centrations of 5-H1AA and HVA (r = 0.69, have also reported an association between p <.001) (Fig. 10); and (4) an inverse correla- this neurotransmitter and other behaviors. tion between body height and (SF levels of Specifically, Brown and colleagues (44) have 5-H1AA. reported a significant inverse correlation between individuals' history of aggressive be- Thus, the relationship between metabolite havior and their CSF levels of 5-HIAA. levels in the brain and CSF provides direct Linnoila et al. (45) found lower levels of CSF evidence for the validity of using these CSF 5-11IAA in individuals who had engaged in measures as an index of brain metabolism in violent and impulsive acts. To the extent that the living. Further, this methodology could impulsivity and aggression can be regarded as be used to examine the interrelationship be- risk factors that have an identifiable tween a biogenic amine or its metabolite and biochemical substrate, it will be important for the status of the various receptors associated future studies of suicide in youth to assess the with the same neuronal system and to degree to which these behaviors are present provide a means for applying post mortem in this age group. Thus, it may be possible to finding to the clinical setting. systematically construct a behavioral and In summary, there are several lines of biochemical profile to aid the clinician in evidence suggesting that there may be a identifying individuals at high risk of commit- neurochemical component associated with ting suicide. the act of suicide. Thus far, the post mortem Recommendations for future studies of biochemical evidence tends to support the youth suicide should include projects that hypothesis that in Individuals who commit will integrate biochemical and behavioral suicide, there is some form of serotonergic factors. For post modem research, this will

2-220 25 9 `tfc' oz0.17-v:., ..

M.StenterPost Mottem Studies of Suicide...

necessitate interviewing next of kin to obtain Lowered moneemine oxidise arny in brains from alcoholic suicides. J Neurochem 1975; 2588773. the needed personality descriptors and diag- 9. Dome 11, Rpley Mit A study of suicide in the nostic information that can then be assessed Seattle snot Compr nqolday 1980; 1ray340.59. in the light of neurochemical findings. The 10. Barraciough 13, &inch J, Nelson B. st al: A hundred oases of suicide: Clinicel aspects. Br J Peyohlaby1974; priority for neurochemical studies should in- 125:35549. Wally parallel studies that have already been Robins E, Murphy GE. Wildman RH, et al: .Some clinical considerations in the prevention of suicide based conducted in adults. In an effort to maintain on a stAoat34 successful suicide& knJAM Health a link between post mortemfindings and the 1959: 19. Shaw DM, Camps FE, and Emission EG:5. clinical application of such findings, inves- hydroxytyptamine in the hind-brain of depressive tigators should obtain samples of post mor- suicides. J Psychisay 1987; 1131937-11. tem CSF where possible.Ante mortem 13. Bourne MR, Bunney WE, Jr., Coburn RN, Davis JM, Shaw DM, Coppen At hicrackenaline, 54trrAtt studies should follow the same basic ap- famine. and 5-11~dweft, acid in the proach as described for post mortem inves- of suicidal patients. lancet OM 8054 14. Pant CM13, Young 0P14, Pdoe K, and Stacey RS: 5- tigations. Thus, normative behavioral/ hydrontitto, noradrenalin , and dopaminein diagnostic and biochemical data should be badmen. hypothalamus, and caudate nucleus of controls and of patients committing suicide by coatgas collected together so investigators can iden- poisoning. Lancet 1959; 1334 tify behaviors or clusters of behaviorsthat 15, Lloyd KG, Fraley LI, Deck .114N. Homyldevria 0: Serotardn and 5-hydnioxylndoleacetic acid in discrete may correlate withbiochemical findings. areas of the brainstem of suicide vicars andoontrol Advances in Biochemical Based on the results of the studies proposed gti.rifiewYork Raven Press,19410-trannical°9/6 above, it may be important to explore the use 18. Ohm JC, Nelson J, Kleinman .1, et al: Studiesof the cholinergio stem in suicide and depression. of various pharmacologic probes in the treat- ol the 6krm York Aced of SOWN* °Worm* ment of suicidal behavior. Thebiochemical on psychobiology of Suicidal Behavior, 1965; 1920. findings in suicide to date seem to relate 17. Korpi ER, Kleinman JE, Goodman SJ, et at: Serotenin and Syiodollaostic acid oancenuation more to this specificbehavior itself, rather in different brain regions of suicide victimComparison in chronic schizopinnia patients with suicide as causeof than to any particular diagnostic group. death. Presented at the meeting of the International Therefore, while it may be necessary to treat Society for Neurochemistry, Vancouver. Canada, JelY 14. the symptoms associated with an individual's 1963. 18. Beskow J, Gantries CO, Roos OE, and %%bled 8: psychiatric syndrome, it may also be neces- Determination of monoamine and moncramine metabo- associated Otos in the human brain: Past modem studies in a group sary to separately treat symptoms of suicide. and in a °antral group. Acts Peyoklat &land with their suicidal behavior. 1978; 53:7-20. 19. OWsns F, Cress A/ Crow TJ, et el:Brain 54.1T2 receptors and Weide. Lancet 1953; 01258. 20. Crow TJ, Cross AJ, Cooper SJ, et al: REFERENCES Neurotransmitter receptors and monoemine metabolites in the brains of Wien* with al:helmet-type dementia and 1. Suicide Surveillance, Centers for Disease Control, depression end suicides.Neuropharmacology 1984; U.S. Dept. of Heeith and Human Unice& Summary 1970- 23(128)5514:11 1980; issued March 1985, 21. Canteen* E, Robins 0, and Grote 5: Regional 2. Stanley M, Mann .1.1, Ochen L Role of serotoner- serolonin levels in brain: A comparison of dg= la system in the post modem analysis of suicide. suicides and alcioholla suicides with controls. Psyshophsnn Bulletin 1901 (in press). Psychiatry 1978; 11(9)283.2114. 3. Asberg M. Thom P, Traskman L Berth:son, 22. Stanley M, Maintye I, and Gershon S: Post mor- Flingbergw V Serotonin depression: A biochemicalsub- tem serotonin metabolism in suloidevictims, presented at group within the affective disorders. Science1978, 1983 ACNP, Puerta moo. 191:479ea 23. Langer SF, Briley MS. Reisman R, et al:3H4m- 4. Treekman-Bendz L Depression and suicidalbe- ipremine Mndkv in human plidelets; influence of ageand havior: A biochemical and OrearmoW4oal study. Thesis, sax. Naunyn Wirniedeberge ArchPhermacol 1980: Stockholm, Sureden: Kemna ktelltule, 19* 319:18944. 5. van Pogo I CSF PIMA and suicide in non- 24. Andra I3J, Snyder Sk Regulation of&melanin depressed schizophrenic.. tame 1981k 2:9774 (514T2) receptors labeled with (311am= by chronic 8. Minn JJ, SiNiksy M NO modem monosmine treatment with .Pharmacci odds** enzyme idnetice in the frontal codes of suicide vic- Exp Ther 1981%21,11112.79. stint tims and control& Acta Psychist &lend 1984; 89:1354 25. Rainbow TC, Beigon, k Distribution ofimiprwnine 7. Grote SS, Moose SG, Robins E, et d: Astudy of bind Mee in the re brain studied by selecied cateoholemine metabolizing enzymes: A com- .NOM Sol Lett 1903;3713):=Istive parison of depressive suicides and alooholic suicideswith 28. Brunello N, Chuang Costa E: Different synap- controls. J Neurochem 1974; 23:791402. tic location of Miansedn and lmlpramtne bindingsites. & Makin CO, Greiend L. Wiberg A, laNnbland0: Science int 21511124

BEST COPY AWAKE 2-221 260 Report of the Secretary's Task Forceon Youth Suicide

27. Rhavi M, Utahif, Price KL, et al:2- =amine: A stadiakraersibb Nopricts= aphis and pit) imipramine binding in Ellochsm Sophys Ras Oamm 1981; 9904. 26. Paul SM, Rena Noe KC Does affinity lakniPronine binding .tabsi emolonin NOS in and pianist? Ufa ea 1001; 29. Langer SF, Rola/nen itBinding of OM im- Wanda adnicionZindea=hanalablzhgrical tools for studies in la 1903; 2240T- 0118461A WO* &fawn s:Tritiated im- OM** sites are demised in the frontal cortex of suicides. Science 1902 21813374. 31. Pad SM, IWiavi PA, Skolnick P. Goodwin Fle High affinity Waling of antidsprallata 10 a mine troisport allain human IMAM and studios in dapresslon. by Post RM, Baftw, ado of Mood Disolders. Baltimom: Mom and Nano MK 545.53. 32. Play EK, Marshall EP, lbassact G, Tomanson BE, Penyt Deonowed lamming Wiftg in the brains of patients with depressive Mmes. Br J ftydikay 1903; 141t 188412. 33. Meyerson LK %Venom,* LP. Mal MS: Human brain racepior alterations in suicide victims. Phaansool Bloohem Way 1902; 1715943. 34. Stanley PA, Mmn JJ: increased &tannin- 2 Bind- ing sites in toad cortex al suicide victims lancet 1903: 2144. 36. Stanley Pk Mame& bindng in the frontal cortex of suicide victims. Am J McNair 1984; 141:11. 33 Pony EK, Peery RH, Tomlinson BE:Orcadian variations in cholkagic anomie and muscarinic reor ex binding in human owls& codex. Named Lott 1977: 4: 155-9. 37. Kaufman CA, GM JO, OtaugNki 1, et at Plus- carinlo suicides. fn: New assamh Abshacts, i3fIth Annual of So American Paydilatricksools- don. Washington 1903. 36. Mann 11, Stantsy M Unpubiished data. 30. Zanko MT, Magoon k incroased adrawivic receptor binding in human frontal carter of suicide Oo- tirns. In: Abstract Paned Mating, Society for Noma- dism,. Bolan, MA, t9113. 40. Sone:ion Morwason JO, Ostarburg, HH: Seated density of OK Woman* binding in aged human brain. J of Pitiaidieni 1065; 45:13024. 41. ftmay PA, Traskman-Band: L, rovW.s IC: Correlations batmen saner& metabolites simultaneously obtained from ample' of CV and brain. Ufe Sciences 1005; 37:127941 42. Mk S, Stanley PA:Dopamine metabolites in human brain. Rn.atracology 1978; 57:17. 43. MWOA OA, &inlay PA: Pat madam and regional °hangs. of ftroiona, 5 FlyamiyindOpsaWo add and tryptophan hi brain. J of Neuroohem 1034; 4215110.02. 44. Brown GI., Goodwin RC, Ballenger JO, Joyst PF, Major LF:Aggression in humans oorralatss with cembroaphisi Sid amine metabolites. Pooh Rea 1979; 1:1314 45. Unnoila M. *Now M, Sdielnin M, Mufti A. Nmon 14, GOOdWIR FK:Low cerebrospinol add 5- hydroxyindoleacetio sold concentration Moridat =hem nonknpublie vfolont behavior. tife Scisnces 14.

2-222 261

BEST COPY AVAILABLE M.Stenley: Post Modem Studies of Suicide

Post Mortem Neurotransmitter andMetabolite Studies in Completed Studies

Shaw et al. (12) BraInstarn EMT

Bourne et al. (13) ,&einem 5-HIAA 1 Pam et al. (14) Brainstern &HT Ng change in brainstern BHIM

Uoyd at al. (15) Brainstern 5-HT No change In brainsiern &HIM Win et al. (16) 1 Hypothalamus 5-HT Nucleus Anumbons &HIM Komi at al. (17) Hypothalamus &HT Beskow at aL (15) 4 Brain 5-HIM Owens st al. (19) No change In 5-HIM levels in frontal cortex Crow et al. (20) No change In 5-HIM levels in frontal cortex Cochrane et al. (21) No change In brain &HT Stentey et al. (22) No r.hange in 5-HIM or 5-HT levels in frontal code(

Table 1.

Stanley et al. (30) 3H4nlprernIne binding In caw Paul et al (31) sig-knipramtne binding In brain Pony et al. (32) 3H4mipramine binding* In cortex Crow et rd. (20) 3H4miprarnine binding In cortex Meyerson et ai. (33)r/ I stfirtgprernine binding In cortex Stanley and Mann (34) t 54412 binding In codex Owen et rd. (19) 54412 binding kt cortex** Crow et al. (20) No change In 5HT2 binding In cortex

Stanley (35) No change In rnuecadnic cholinerglo receptor binding In cortex Kaufman et al. (37) No charge In muscadnio cholinargio receptor binding Meyerson et el. (33) muscarinic chollnergio receptor binding

Zanko and Wagon (39) In beta reoeptor binding Mann and Stanley (5) 1 In beta receptorbinding Meyerson et al. (33) No change in beta receptor binding

Depressed patients dying of natural causes Increased but not sipificantly

Table 2.

BEST COPY IMAMS

262 2-223 &spat of the Secietary's Task Fame on Youth Suicide

Characteristics of Suicide Victims (n =22) and Control Subjects (n =22) Whose Death was by Nonsuicidal Means

Time Between Number of Binding Acle Cause of Death and Singling Sites A????? Subject (years) Sox Math Autopsy (min. emax) (KM Suicide Victims 1 46 M Hangft 1,440 191 17 2 13 M Gunshotwound 1,140 197 15 3 15 M Hanging 665 677 13 4 25 M Gisehot wound 1,580 387 28 5 33 M Gunshot wound 1,020 773 15 6 55 ki Jumping from height 1.140 388 16 7 25 PA 1,365 450 10 s 30 M Hanging 1,320 640 11 9 34 M Jumping from height 1,320 454 13 10 22 M Gunshot wound 555 871 10 11 25 M Drowning 1,005 498 8 12 so PA Gunshot wound 1,335 555 24 13 18 M Gunshot wound 1,055 505 16 14 30 14 Jumping from height 1,260 553 9 15 37 M Drug overdose 795 605 11 16 64 M Drug overdose 1,110 574 16 17 43 M Gimshot wound 460 583 16 18 65 141 Jumping from height 1,110 10 19 30 F Gunshot wound 1,290 621 12 20 72 F Dm overdose 1,185 339 17 21 79 F Drug overdose 1,080 543 12 22 18 F Jumping from height 600 229 9 Control subjects 1 45 M Gunshot wound 1,650 232 2 21 FA Gunshot wound 1,205 242 17 3 22 M Cardiovascular disease 750 735 14 4 20 M Gunshot wound 1,570 411 23 5 31 M Cardlorascular disease 880 423 11 6 47 M Cardlovescidar disease 1,200 526 13 7 28 M Auto accident 735 439 8 18 PA Gunshot wound 805 649 19 9 39 M Falling in= height 1,245 510 11 10 30 144 Auto accident 460 527 10 11 26 m Failing from height 1,350 652 12 12 53 M Cardiovascular disease 860 393 36 13 24 M Cardiovascular disease 1,305 485 12 14 23 M Gunshot wound 1,035 563 10 15 39 M Knee wound 600 597 11 16 40 M Gunshot wound 735 608 16 17 33 M Gunshot wound 435 693 13 19 82 M Falling from height 770 382 8 19 23 F Gunshot wound 865 840 10 20 as F Knee wound 1,020 692 19 21 73 F Cardiovascular disease 1,440 279 10 22 50 F Auto accident 830 343 10

Table 3. 263 2-224 *Manley: Pos5,Alceent StudiesofSuicide

IComparison of maximal binding of (314 imipriminein samples of tantsi cortex from suicide vicitims and controlsubjects. Difference in `Pmax values is statistically significant (p < .01)

0 1000

900

800

142, 700

a) 600

500 03 400 db koc 300

2200

100

Suicides Controls

Figure 1.

264 2-225 Report of the Secretaries Task Fome on Youth Suicide

541612 binding parameters in post 'prism frontal cortex from suicide victims and control subjects. wmax and Kd values exPresirl as means ± S.E. p < 0.01 by Wilcoxon's test (bvo-talled) for max.

130 120 110 100 90 ao 70 60 50 40 30

Suicide Control Suicide Control

Figure 2.

2-226 21;5 1.;

M.Starder Post Modem Studies of Suicide

Number of binding sites(Bow) for suicide victims (n =22) and control subjects (n=22). 800

700 000 600 o80 0 500

400

300 8 200

100

Suicides Controls

.266 2-227 t Report of the Secretary's Tesk Force on Youth Suicide

1111=0111111111MIIP111111111011111W Receptor affinities (lCd) for suicide victims (n=22) and control subjects (n=22).

40 0

Suicides Controls

Figure 4.

267

2-228 A

M.Stanter Post Mortem Studies of Sutekle

re-

268 2-229 Report of the Secretary's Task Force on Youth Suicide

CSF Levels of SHIAA vs. SHIM In Cortex

0.3 p<0.001 r).78

A 0.2 e

0.E

444

0.0 90 10 150 20 250 CSF 51-11AA

Conflation of' the concentration of MP kw& of SAIAA (ng/m1) and brain ktfla of S-H1AA (Wins).

Agin I.

2-230 269 4

AtStanter Post Modem Studies of Suktde

CSF Levels of HVA vs. WA In Cortex

0.4 p<0.02

0.3

0.2

A 64 a A 44 Di- A 4441 4 a a 4 a

0.0 (1-0 0 50 150 2-10 CSF HVA nem!

Condition of CIF 1ekoIHVA (agtml) and brain lards oniVA (nen).

Figure 7.

270 2-231 Repott of the Secretary's Tesk Force on Youth &lode

Post Modem CSF 544IAA

p<0.01

of I..,--...... " i i 3 :. 3 6 Serial Mr Fractions (2 c.c.)

Graph display of prniressive coma tration gradient for S-HIAA.

Figure .

271 2-232 01. 1.=4,11. M.Stanier Post Modem Studies of Suicide

Post Morton CSF HVA

100 p<0.05

WINO

NNk

1 2 3 4 5 6

Serial CSF Fractions (2 c.c.)

C2-.7% display of progressive concentration gradient for HVA following serial sampling of CS? fractions (*ad).

Figure 9.

2 72 2-233 Report of the Secretary's Task Force on Youth Suicide

100-i A A

so- a

100 2110 250 HVA

Conelations between OF samples in post most= sampka (nen° of 5-HIAA and EWA.

Flinn 10.

273 2-234 ThE NEUROENDOCRINE SYSTEM AND SUICIDE

Herben Y. Meltzer, M.D., Bond Professor of Psychiaby, Department of Psychiatry, Case WatanReserveUniversity,Cleveland,Ohio

Martin T. Lony, Ph.D., Department of Psychiatiy, Case Western Reserve University, Cleveland, Ohio

SUMMARY Suicide is increased in frequency in Cushing's As will be reviewed, the thyroid stimulating syndrome which is characterized by increased hormone (TSH) response to thyrotropin- hypothalamic-pituitary-adrenal axis (HPA) releasing hormone (TRH) end the basal activity. Several types of studies suggest in- secretion of cortisol correlate with violent creased IDA axis activity in depression or suicide or suicidal ideation. Perhaps more stress may be related to increastAl suicidal be- importantly, cortisol as well as other hor- havior: increased serum cortisol, increased mones can influence the activity of 24 hour urinary free cortisol and an enhanced neumtransmitters such as serotonin (5-Hi) 5-hydroxytryptophan-induced increase in which may have a more direct causal effect in serum cortisol. However, these associations suicide. Given that no one factor is likely to are weak and some cortisol measures do not be the sole determinant of a complex be- relate to increased suicidal risk. The TRH- havior such as suicide, it is important to induced increase in TSH may be blunted in develop models of the etiology of suicide violent suicide but shows a positive relation which integrate a variety of influenrw such to suicidal ideation. There is some evidence as hormone secretion and neurotransmitter linking the HPA abnormalities and the chemistry. We will attempt to do this by con- blunted TSH response to serotonin, the sidering the hypothalamic- pituitary-adrenal neurotransmitter most closely linked to (WA) axis and the serotonergic system, con- suicide. Suicide in adolescence may occur in sidering the evidence relating both to suicide, the context of a rapidly changing and aroused and then the interaction between the two sys- neuroendocrine system. Hormonal markers tems. We will also briefly consider other hor- of suicide risk and the iole of hormones in al- mones that may contribute to suicide tering neurotransmitter function appear to potential and hormone challenge tests that be worthy of further study. may predict suicide. There is only minimal data concerning the INTRODUCTION endocrine status of adolescents who have made suicide attempts or putative biological The endocrine system is of interest in rela- markers of suicide in adolescents with thn to suicide for a variety of reasons. An in- psychiatric disorders. Therefore, this review creased likelihood of suicide is found in some will of necessity focus on studies in adult endocrine disorders, e.g. Cushing's populations. How applicable these results syndrome, and during corticosteroid therapy. are to adolescents remains to be determined. ,7

274 2-235 Report of the Secretary's Task Force on Youth Suicide

In the one instance where data exists for In addition to Cushing's syndrome, other en- adolescents and adults (the dexamethasone docrine disorders, various aspects of normal suppression test, INT), the findings are quite endocrine maturation, function and decline, comparable. Suicide in both adolescents and and hormones other than glucocorticoids, adults often occurs as a consequence of may be associated with profound affective, major depression or schizophrenia, compli- cognitive and psychomotor disturbances. cated by alcohol or drug abuse. Both adoles- Hypo- and hyper- thyroidism, hypocor- cents and adults may suicide impulsively and tisolism(Addisoniansyndrome), as a consequence of severe stress. There is hypopituitarism (Simmond's disease), hyper- no a priori reason to consider any of these pituitarism, pheochromocytoma, hypo- and factors would act differentially in adoles- hyperparathyroidism, hypo- and hyper- cents, so it is most likely relevant to the task glycemia, pancreatitis, pancreatic carcinoma, of this symposium on risk factors in adoles- androgen excess and deficiency and cent suicide to consider the data concerning estrogen-progesterone disorders associated suicide in adults. This is not to say that uni- with menarche, premenstrual syndrome, oral que biological, especially endocrine, factors contraceptives, pregnancy and the postpar- are inoperative in adolescents or that they tum period may produce highly disturbing are less important than those common to changes in mental status that could figure in suicide in both age groups. Rather, it is our an individual's ability to function adequately, belief that future studies on the biology of his sense of optimism concerning the future, adolescent suicide may well use findings in the desirability of continuing to live, thoughts older adults as appropriate guidelines for re- a. suicide and capacity to cony out a suicide search to determine what are the most impor- attempt. It is beyond the scope of this review tant influences on adolescent suicide. to conskler these conditions in detail. They are adequately described in textbooks of ENDOCRINOPATHIES, medicine and clinical endocrinology. Clear- CORTICOSTEROIDS AND ly, any adolescent pesenting with suicidal SUICIDE ideation or a suicidal attempt should have a thorough medical workup to evaluate the There are many aspects of the HPA axis possible presence of an endocrinopathy which point toward its importance for under- which might be causing symptoms that direct- standing the biological contribution to ly or indirectly compromise mental status. suicide. Six of 35 consecutive patients with Conversely, adolescents with serious en- Cushing's syndrome, a group of disordets as- docrinopathies such as juvenile diabetes, sociated with large increases in glucocor- hypothyroidism or Cushing's syndrome may ticoid output, were reported to have suicidal be at increased risk for suicide and a greater thoughts and two of these made suicide at- than usual index of suspicion concerning tempts (Starkman and Schteingart, 1981). In suicide might be advisable in such cases until another study, one of 29 cases of Cushing's adequate therapy was instituted. syndrome made a suicide attempt (Cohen, 1980). Lewis and Smith (1983) reviewed the TRH STIMULATION TEST AND literature on exogenous corticosteroid-in- SUICIDE duced psychiatric syndromes and found that 3 percent of the cases for whom outcome was A blunted TSH response to TRH ( <5 specified committed suicide.This data WM) has been reported in about 25 per- points towards a role of corticosteroids in in- cent of depressed patients (Loosen and creasing vulnerability to suicide. As will be Prange, 1980). Three studies have reported discussed, this might occur because of effects a relation between a blunted TSH response of corticosteroids on neurotransmitter or and violent suicide. Eight depressed patients neuromodulator physiology. with past, end three with recent, violent

2-236 275 H.Meitzer: The Neumendocrine System end Suicide suicide attempts were found to have a low positive relatior between the TSH response TSH response compared to patients who had to TRH and suicidal ideation are consistent made past (N=7) or current (N=7) non- with a relation between diminished brain 5- violent suicide attempts and 26 depressed HT and suicidal ideation. Krulich (1979) has patients without any suicide attempts reviewed the evidence from rodent studies (F=3.46, p< .005) (Linkowski et aL, 1983). which suggest that 5-HT may either inhibit or A two-way ANOVA showed a significant enhance 1511 secretion. Krulich et aL (1979) relationship of the TSII response to violence found that quipazine, a 5-1IT agonist, did not but not for recent vs. past attempts. Seven of affect the TRH-induced increase in TSH in 12 patients with an absent TSH response had the rat but quipazine and 5-HTdid inhibit en- a previous history of violent suicide titteinpts dogenous TSH secretion. The relationship compared to four of 39 patients with a maxi- of brain 5-HT, the TSH response to TRH, mum TSH response above 1 mUll (p <.001). and suicide requires further study. In light of During a five year followup period, three these findings, we carried out a retrospective committed suicide by violent means and one analysis of the relationship between suicide by overdose. All had an undetectable ISH ratings and TSH response in a group of newly response to TRH. Linkowski et aL (1984) admitted psychiatric patients to our Mental subsequently reported similar findings in a Health Clinical Research Center who had slightly expanded sample. In agreement with bad a TRH stimulation tat (500 ug in- this, Kjellman et aL (1985) reported that the travenously). ISH levels over the next 120 TSH response to TRH was significantly minute period was determined by radioim- lower in three deprened patients who made munoassay. The subjects were drug-free for violent suicides than in 27 who had made no at kart 7 days and consisted of 24 depressed attempt or a nonviolent attempt. van Praag patients (including six schizoaffective and Plutchick (1984) also report an associa- depiessed,mainlyaffective),13 tion between violent suicide and a blunted schizophrenic patients (including three TSH response to TRIL schizoaffective depressed,mainly schizophrenic patients), three manic patients The possibility that a blunted 'MR response and five with miscellaneous diagnoses. to TRH may have long term prognostic sig- Patients were diagnosed according to nificance for suicide is intriguing. This could Research Diagnostic Criteria. Suicklal be- be mediated by a relationship between the havior was assessed during the first week of blunted TSH response and specific hospitalization as part of the Schedule for neurotransmitter abnormalities such as Affective Disordets-Change (SADS-C) in- diminished activity in 5-HT pathways. There terview. Thirteen of the 45 patients admitted is some evidence relating the TSH response to slight or moderate suicidal ideation. Only to TRH to the serotonergic system. two patients had made suicide attempts, not Cyproheptadine, a 5-HT antagonist, was considered lethal in intent. There were no found to inlulit the TSH response to TRH in violent suicide attempts. We found r rig- two studies (Ferrari et al., 1976; Egge et aL, nificant positive correlation between maxi- 1977), but rot in another (Goldstein et al., mum TSHresponsetoTRH 1979). Cyproheptadine has multiple effects (peak-minus-baseline) and suicide ratings other than 5-HT antagonism and is not a par- (Spearman rho =0.30, N=45, p=0.014). The ticalarly potent 5-HT antagonist. Gold et aL mean TSH response to TRH in this group (1977) found a negative relationship be- was 9.8 ±S.D. 5.6 uIJ/ml. This is very similar tween the TSH response and CSF5-hydroxy- to that found in 19 norrral controls 8.9 ± S.D. indoleacetic acid (5-IIIAA), the major 72 utY/mL Ten of the 43 patients (22%) had metabolite of 5-1IT in depressed patients, a blunted TSH response (<5u1.71ml). Of which suggests that diminished serotonergic these, only three had any wicidal ideation, activity might be associated with a larger nil one mild, one slight, one moderate. Further response. Thus, our TSH results indicating a

2-237 276 Report of the Secretaty's Task Force on Youth Suicide study is needed to determine if current tween the TSH response to TRH and basal suicidal ideation is associated with a more cortisol is directly relevant to consideration robust 1SH response but still within the nor- of the importance of TSH as a marker for mal limits. suicide since there have been a variety of findings which indicate excessive activity of There is conflicting data concerning the the HPA axis in suicide. 'No early National linkage between increased HPA axis activity Institute of Mental Health studies reported and the blunted TSH response to TRH in elevated urinary 24 hour 17-hydroxycor- depression. Kirkegaard and Carroll (1980), ticosteriod (17-0HCS) levels in depressed Asnis et al. (1981), Agren and Wide (1982) patients who suicided. The enhanced cor- found no correlation between these two vari- tisol secretion preceded the suicide attempts able& We found no relationship between the by several weeks (Bunney and Fawcett, 1965; TSH and 8 A.M. response to TRH plasma Bunny et al, 1969). These findings were not cortisol obtained within the same drug-free replicated in subsequent studies of four evaluation period that included the TRH patients who suicided but who did not stimulation test (Spearman rho=0.061, evidence elevated 24 hr 17-OHCS prior to N=38, p=NS). Sixteen of the 45 patients suicide with the period of study ranging from had also undergone other neuroendocrine a day to eight months prior to suicide (Levy challenge test within the same three week and Hansen, 1969; Fink and Carpenter, drug-free period as the TRH test, e.g., the 5- hydroxytryptophan-induced increase in 1976). Agren and Wide (1982) foul tcl a negative correlation between Medical Lethality serum col Owl (Meltzer et aL, 1984). We also of Worst Ever Suicide Mtempt and 24 hr uri- er -nined the correlation between basal nary free cortisol in 76 patients with major mum cortisol (obtained 60 minutes after depression. Ostroff et aL (1982), however, catheter placement) and the mit response did find higher 24 hr urinary cortisol levels (as in these 16 subjects. We found a highly sig- well as lower urinary norepinephrine-to- nificant negative correlation between thesre epinephrine levels) in three of 22 subjects, variables (Spearman rho=-0.65, N=16, two of whom made lethal and one a near p=0.007). This is consistent with the studies lethal suicide attempt. of Loosen et aL (1978) who reported a negative correlation between the TSH response In addition to urinary cortisol, which is a good to TRH and basal plasma cortisol just before measure of adreno-corticoid output, other a TRH infusion. The significant correlation studies have reported elevated plasma cor- between the TSH response to TRH basal and tisol in suicide.Thus, Krieg..r (1974) basal serum cortisol leveb obtained during reported that plasma cortisol in 13 patients the 5-H77 study may be due to Ica set in- who suicided during a two-year followup fluence of stress in the catheter study and the period (21.1 ± 5.6 ug/dI) was significantly fact that both the TRH and 5-H17 studies higher than in 39 who did not (163 2.4 were carried out at 10 A.M. Basal cordsol ugfell).However, single plasma samples, levels obtained at 8 A.M. may reflect the noc- especially those obtained by venepuncture, turnal surge of cortisol secretion. However, may not be an accurate measure of cortisol Asnis et al. (1981) did not observe any output. Venepuncture may induce stress-re- relationship between the TM response to lated cortisol secretion. The results of TRH and multiple measures of cortisol Kreiger (19/4) could indicate that in- secretion, including plasma cortisol levels dividuals who are vulnerable to suicide may before and throughout the TRH infusion. be particularly prone to stress-induced cortisol secretion. HYPOTHALAMIC-PITUITARYAD We have recently examined plasma cortisol RENAL AXIS AND SUICIDE concentrations in relation to recent suicidal history (Meltzer et aL, in preparation). Basal The presence or absence of a relationship be-

2-238 277 H.M&tzer The Neumendocrine System and Suicide serum cortisol levels from unmedicated since cortisol secretion during this period ap- patients with major depression, mania or pears to best reflect 24 hour cortisol output schizophrenia and normal controls who were (Halbreich et al., 1982). In the 61 depressed part of a study of the cortisol response to 5- patients, To cortisol was significantly cone- hydroxyhyptophan (5-HTP) (Meltzer et al., laced with the Hamilton scale ratings of help- 1984) were analyzed. In this study, subjects leuness (rho =0.38, p=.003), depiessed were fasted overnight and an indwelling mood (rho=0.29, p=0.02), hopelessness venous catheter was inserted at 9 A.M. Thir- (rho=0.27, p=0.03), paranoid symptoms ty minutes later the first basal sample was (rhomm-^26, p=0.($) and work and activities withdrawn. Other samples were drawn 15 (rho= il25, p=0.06). These relationships and 30 minutes later. The 10 AIL sample are consistent with the conclusion that in- (To) was related to the Hamilton Depression creased /WA axis activity is a state marker for Rating scale (HDRS) suicide item (0-4) and severity of depression and as such could be the SADS-C suicide ftem (0-6). Serum cor- an indicator of suicidality. This may beof tisol (To) was significantly but weakly cone- some clinical value in patients who falsely lated with the HDRS suicide item for all deny suicidal ideation and intent. subjects (Spearman rho=0.20, N=107, We have also examined the relationship be- p =0.04). This included 61 depressives, 16 tween basal cortisol levels and suicide at- mania and 30 schizophrenics. The HDRS tempts, both violent and nonviolent, suicide rating was also correlated with To together and separately, in these patients. cortisol in the combined group of affective As can be seen in Table 1, the 10 A.M. serum disorders (rho =0.23, N=77, p=0.05) but not cortisol was not significantly different in any in the depressed patients alone (rho=0.18, of these groups although the highest levels N =61, p =0.16). SADS-C suicide ratings and were found in the violent attempters, next in To cortisol were not significantly correlated the nonviolent attemptets and the lowest in. in the affective disorder patients. The mag- the normal controls. No significant differen- nitude of the correlation between the HDRS ces in basal serum cortisol were found when suicide item and serum cortisol indicates only the two types of attempters were combined a small portion of the variance insuicide and compared to nonattempters and normal ratings can be attributed to elevated 10 A.M. controls (data not presented). The trends cortisol levels. However, it is possible that evident in Table 1 for all psychiatric patients this relationship is stronger at other times of were more prominent in just theaffective dis- the day. It would be of interest to examine orders (Table 2) but these differences were the relationship between serum cortisol be- not significant either. tween 1 P.M. - 4 P.M. and suicide ratings,

Base Serum Carlini In All Basal Serum Cortisol in Affective Psychiatric Patients In Relation to Disorders in Relation to Suielde Attempts Suicide Attempts Basal Serum Basal Serum Group N Cortisol M(di) Group N Cordsol (ugfcli) Normals 21 11.5 A. 5.0* Normals 21 11.5 ± 5.0* No Attempts 73 12.0 .±..5.1 No Attempts 55 11.7 4.8 filinVident 11 12.7 ± 7.7 Nonviolent 9 13.5 8.4 Violent 12 13.6.1.8.1 Violerd 9 15.1 .±.6.4

X .±. SD. X t S.D.

Table 1. Table 2.

,_.27s 2-239 Report of the Secretary's Task Force on Youth Suicide

We did find a trend for serum cortisol in the suppressors whereas three completets with depressed and manic patients who had made diagnoses other than major depression were an attempt (14.3 ± 7.3 ugitil, N=18) to be suppressois. higher than that of the nonattempters (11 6 They also noted eleven other suicide at- ± 4.8 ug/d1, N=76, p=0.08). We examined tempters who were nonsuppressors but whether serum cortisol levels above 20.0 failed to report how many other nonsuppres- ug/c11 might have some value as a means of sots were not suicidal They proposed that identifying suicide attempters but found no the relationship between suicide and non- indication that was the case. Nevertheless, suppression was restricted to melancholic& the trend in this data, considered in the con- Banki and Arato (1983) and Targum et aL text of the evidence for an association between excessive HPA axis activity and (1983) also found evidence that nonsuppression predicted suicidal activity. Robbins and suicide, sugge ts that it would be of interest Alessi (1985) studied 45 newly hospitalized to carry out a prospective study in which adolescents, with various psychiatric disor- serum cortisol was monitored in adolescents ders 23 of whom had attempted to commit at high risk for suicide, e.g., those adolescents suicide. Of the 39 suppressors, 17 (43.6%) presenting at a clinical setting because of the made suicide attempts, none considered to suspicion of being suicidal or adolescent patients placed on suicide precautions in have a lethal intent. HOWeVert all six nonsuppressors had made suicide attempts; of these, clinical or perhaps forensic settings. four were medically dangerous or lethal at- In addition to plasma cortisol studies, there tempt- The two adolescents who had made has been one study which related nonmedically srious attempts and who were cerebrospinal fluid ((SF) cortisol to suicide. nonsuppressors subsequently made medical- Traskman et al. (1980) reported no differen- ly serious attempts, one of which was fatal. ces in CSF cortisol levels in five suicidal Thus, this study strongly supports a highly sig- patients and 14 nonsuicidal depressives. nificant association of DST nonsuppression There was also no relationship between a his- with lethd or potentially lethal suicidal be- tory of ever making a suicide attempt and havior in adolescents. high CSF cortisol. CSF cortisol was sig- Zimmerman et aL (1986) recently reported nificantly cortelated with urinary free cor no relation between suicidal ideation, serious tisol (r=0.67, N=14, p.01) which provides staid& attempts and nonsuppression in 187 further evidence against a relationship be- major depressives. Brown et al (1986) found tween elevated urinary free cortisol and no differences in the incidence of nonsup- suicide. Nevertheless, further studies of CSF pression in 10 recent suicide attempters, 10 cortisol and suicide, especially violent past attempters and 37 nonattempters. suicide, would be of interest. Moreover, nonserious suicide attempts were There have been eight published studies of more common in suppressors than nonsup- the relationship betsnen dexamethasone pressors. Meltzer et al (in preparation) have suppression test (DST) status at admission to recently reported no relation between hospital in psythiatric patients and prior violent, nonviolent and no suicide attempts suicidal activity. These are summarized in and nonsuppression in 55 patients with major Table 3. Five of the eigAt found a significant affective disorder.However, there was a relationship between suicidal activity and trend for suicide attempters to be nonsup- nonsuppression.Coryell ano Schlesser pressors (10116, 62.5%) more commonly (1981) found that all four patients who than nonattempters (16139, 41%) (Fsher suicided out of a goup of 205 unipolar exact test, p=0.085). We also found that depressions had been nonsuppressors. Car- Hamilton Depressinn Scale suicide ratings roll et al. (1981) reported that all five were significantly higher in nonsuppressors melancholic suicide completers were non- (all diagnoses) than suppressors and that

2-240 279 H.Meftzer: The Neuroendocrine System and Suicide nonsuppressors hrd suicidal ideation sig- Although there appears to be some evidence nificantly more frequently than suppmssors. supporting a relationship between the DST Differences in the number of days between and suicide, it is important that these results the suicide attempt and the DST could ac- be lute% neted cautiously because of the count for some of the discrepencies between evidence that nonsuppression with the oral 1 studies. However, Brown et al. (1986) found mg test may be related to differences in no such correlation in recent attempters. It dexamethasone pharmacokinetics. There is possible that the association between non- are now several studies reporting lower supprenion and suicide may bc confined to dexamethasone levels in nonsuppressors patients with endogenous depression as than suppressors (Arana et al., 1984; Berger proposed by Carroll et al. (1981). et al., 1984; Holsboer et al., 1986; Johnson et al, 1985; Lowy et at, in press). Differences In addition to these studies of the relation- in dexamethasone levels might produce both ship between suicidal ideation or acts and false positives and false negatives.Thus, DST status, there are several other relevant there could be an association between low reports. Two studies describe five patients plasma dexamethasone levels and suicide who made suicide attempts within a few days rather than nonsuppression and suicide. In of receiving dexamethasone as part of the any event, it would appear prudent to assess DST (Beck- Freis et at, 1981; Asberg et al., DST status, including measurement of 1981). Other investigators did not confirm dexamethasone levels, in relation to suicidal this finding (Coryell, 1982; Kronfol et al., activity. Such studies should include post- 1982). Yerevanian et al. (1983) reported an treatment repeat DST testing and long term association between failure of the DST to followup to assess whether nonsuppression normalize and subsequent suicide. Greden does have prognostic value for suicidal risk. et al. (1980) also noted one such case.

Suicide and the Dexamethasone SuppressionTest StaloldW Suicide/ Patient Attempt Attempt Authors Population SuPPression NonSuppresslon Comment

Corlett and UP Dap W109 4196 0.06 1 suicidal Sohlesser(1961) neurotic NS. Carroll et MD021 (VO 16/19 5/9 NS oompleters. (1991) Mixed 3 Bonk! and Mixed 2/20 12/37 < .10 Arato(1963) Thom et al UP Dap 3/26 14/23 0.01 5 N$ vs 0 S made (1963) subsequent attempt. Robbins and Adolescent. 17/22 516 0.01 4 NS made war WU msepsetg Inpatients attempts. 5 made no serious attempts. Zimmerman MDD 12/12r 2/60 NS Nonserious mom et el (1996) 31/127+ 5/60 0.01 common In S. Brown et al MCD 15/37 W20 NS (1aee) MOWN et al Mixed 6/29 10/26 NS (In preparation)

000suppressor S suppressor *does said& attempt + nosserioa mid* attempt UP DepUnkefer depression MDD major depressive disorder.

Table 3.

2-241 1. '4' 280 Report of the Secretary's Task Force on Youth Suicide

It appears unlikely that nonsuppression at of the GCCR in lymphocytes was observed. admisskm has noteworthy significance for We have found decreased in vivo inhibition suicide but it is possilge that failure to nor- of the lymphocyte proliferative response to malize during treatment might. We have the mitogens conconavalin A (ConA) and demonstrated that depression and nonsup- phytohemaglutinin (PHA) following 1 mg pression may be related to glucocorticoid dexamethasone in depressed patients who receptor subsensitivity (Gormley et al., are nonsuppressors (Lowy et al., 1984). 1985). These results have nvently been Depressed patients, especially nonsuppres- replicatA (Whalley et al., 198). These sors, also had lower binding of 3H-triam- studies involved measurement of glucocor- cinalone (Gormley et al., 1985). The failure ticoid receptor concentration in lym- to suppress serum cortisol followiag phocytes. They could reflect similar changes dexamethasone was associated with a smaller in the RPA axis. Glucocorticoid influences decrease in GCCR content. We measured on serotonergic neurons might mediate the sesum dommethasone levels and found sig- relation between suicide and DST status. nificant negative correlations between the This will be discussed subsequently. change in the PHA response, but not in the In addition to abnormalities in glucocorticoid ConA response (Meltzer et al., 1984). We output, we have raised the possibility that ab- have also observed a subsensitivity of the normalities in glucocorticoid response may lymphocyta of patients who were nonsup- be a factor in major depression and other pressors to the inhibitory effect of 10-9 and psychiatric disorders (Lowy et al., 1984). 10-10M dexamethasone on the lym- Failure to suppress cortisol after phoproliferative response to ConA (Lowy et dexamethasone may be a special instance of al., in press).These concentrations of glucocorticoid receptor (GCCR) resistance. dexamethasone correspond to those present The lack of stigmata of Cushing's syndrome at 8 A.M. after a 1 mg dose. The difference in psychiatric patients with excessive between suppressors and nonsuppressors glueocorticoid output suggests some GCCR was observed only with ConA, not PHA. subsensitivity.Glucocorticoid receptor Because of this, we thought it would be im- number or affinity may change in response to portant to examine the relationship between changes in the availability of glucocorticoids, suicide and subsensitivity to glucocorticoids e.g., administration of dexamethasone (1-24 as indicated by the in vivo responses to CAmA tng) produces a decreased number of GCCR and GCCR content. The presence riGCCR in lymphocytes from normal volunteers subsensitivity might counteract the effect of which can be detected as little as 12 hours increased glu output. The com- after first administration (Bloomfield et aL, bination of the two trs might turn out to 1981; Schlechte et al., 1982). GCCR down be a better predictor of suicide than regulation following glucocorticoid ad- measures of glucocorticoid output such as ministration can also occur in a selective basal plasma cortisol or urinary free cortisol manner within the brain (Meany and Aitken, alone. We correlated Hamilton Depression 1985; Tornello et al., 1982). Alterations in Scale suicide ratings with these measures in the level of endogenous glucocorticoids can unmedicated patients, the majority of whom also modify GCCR number e.g., stress results met RDC for major depression. The results in the decreased number of GCCR in both are preliminary because the number of sub- brain and liver (Loeb and Rosner, 1979; jects for whom data is available is swill. We Sapolsky et al, 1984). Chrousos et at (1983) found no significant correlations between has reported a familial glucocorticoid resis- the change in the ConA and GCCR content tance in man characterized by a marked in- following dexamethasone or basal GCCR crease in serum cortisol levels, abnormal concentration and the suicide rating. Since D61' and no physical stigmata of glucocor- the group included only two subjects who bad ticoid excess. A decreased number or affinity

2-242 . 281 , .

H.Meltzer: The Neuroendocrine System and Suicide made a severe suicide attempt, further study a variety of ways. 5-1IT has a well known is needed to rule out a relationship between stimulatory effect on the adrenocortical sys- GCCR resistance and suicide. tem (Meltzer et al., 1984). Glucocortkoids, in turn, have been shown to have a facilitoty We have reported that the 5-HTP-induced effect on 5-HT biosynthesis and turnover increase in serum cortisol showed a sig- (Rastogi and Signhal, 1978; deKloet et al., nificaut positive correlation with HDRS 1982).Glucocorticoids also modify 5-HT suicide ratings at admission in 24 depressed receptors (Bigeon et al., 1985) and various and manic patients and that seven patients serotonergic drup modify GCCR (Angeluc- who made violent attempts had a larger cor- ci et al., 1982; Patacchioli et aL, 1984). A tisol response to 5-HTP than 33 who had not decrease in glutmcorticoid-mediated 5-HT (Meltzer et al., 1984). We have continued synthesis due to a GCCR dysfunction could this investigation using the same methodol- contribute to the postulated decreased level mit with the etception that L-5-HTP 100 mg of 5-HT which occurs in some depressed has been substituted for D,L-5-HTP, 200 mg. patients. Recently, an association between We have now examined our data in relation the DST and platelet 5-HT uptake in to suicidal attempts at any time in life, with depressed patients has been reported information from the patient and informant& (Meltzer et al., 1983).In addition, As can be seen in Table 4, patients with af- daxamethasone has Own shown to directly fective disorders who made violent attempts modify cerebrospinal fluid levels of the 5-HT had the highest cortisol response, followeiS metabolite, 5 hydroxyindoleacetic acid, in by those who made no attempt, a nonviolent psychiatric patients (Banki et al, 1983). attempt, and normal controls. The violent attempters had a significantly greater cortisol Adrenalectomy increases 5-HTI receptor response than the nonviolent attempters binding in some regions of the hippocampus (p =0.003) and the normal controls (Bigeon et al., 1985). Mrenalectomy also (p=0.0008) but just failed to differ from the counteracted the stimulatory effect of nonattempters (p =0.08). vasoactive intestinal peptide (VLP) on 5-11Ti binding sites in the dorsal subiculum of the Glucocorticoids modulate the biosynthr.sis hippocampus, but not the inhibitory effect of and functionalactivity of raany VIP on the 5-HT1 binding sites in the neurotransmitters and neuromtesulators. suprachiasmatic nucleus. Other complex in- Thus, abnormalities in the GCCR could con- teractions between VIP and adrenal steroids tribute to some of the known biochemical on 5-HT1 binding were also reported (Ros- changes associated with depression. 5-HT, tene et al., 1285). Adrenalectomy also inin particular, interacts with glucocorticoids in creases brain 'H-imipramine binding (Arora and pleltzer, in press).Chronic car- Corded Response to ticosteroid administration appears to have 541ydroxylryptoohan In Patients with gigot AtiedIve Disorders mixee effects on 5-HT function (Dickson et al., 1985; Nausieda et al., 1982). Cortlsol Group N Response (AUC) As discussee Asewhere in this symposium, 5- HT is thought to play an integral role in the Normal Controls 22 1624 ± 540 biology of suicide, especially violent suicide, No Attempt 55 2112± 713 impulsivity or violence per se. In view of the NonViolent Attempt9 1776±1045 above mentioned interactions between 5-HT and glucocorticoids, it will be important to Violent Attempt 9 2772 ± 846 obtain bioloecal measures of both serotonin and glucocorticoid activity or function in in- F 610 At am 3,89, F. Ng 0.0008 diViduals who have made suicide attempts or ar e. considered high risks. Furtherbasic re- Table 4.

,282 2-243 Report of the Secretary's Task Force on Youth Suicide search on the interaction of these two sys- tifying duration and intensity of ideation tems is strongly indicated. might helP

CONCLUSION REFERENCES There is a possibility of neuroendocrine trait 1. Paton H, and L Patterns of &premien markers for violent or lethal suicide. A reflected pituitwy- KlitTold and Pito=onsi cloak* dunces. Psychonsuroendo- 7(4030111- blunted TSH response, abnormal DST or in- 327, 1032. creased UFC appears to be associated to 2. Beck-Rise VP6Kft D, Klemm 9, ticar Pederson J, Sara V. Siolin A, Mien F. an L past, current or future violent suicides. Suiddel behavior and the dexamothasone suppression Similarly, the 5-HTP-induced increase in Wet. Pm J Poph inIngoggi 1991. serum cortisol may be largest in patlents who 3.=rdht Pkto KM. Dow 12, Kook) JC, vonZenteen IX The of tier decomethasanssuoomesiors have made a violent suicide attempt anytime test for tin pieces, in psychiatry. EOM in life. Some, all or none of these neuroen- URI:3724M 994. 4. Sisson A. .Reinbow TC, MoEwen SS:Car- docrine abnormalities may relate to timensrone moaddon of neurotransmitter receptors in decreased brain serotonerec activity. If they rat hippooempum A quenetstIve autonenographio study. are related to decreased serotonergic ac- Brain Res 392:306.314, WM & Bloomfield DC, Smith KA, Palma% BA, Okil-Pse. tivity, then the results would be consistent nide K. Mune* AU:In vitro glucooratioold studies in human lymphoma: Clinical and biologic significance. J with a broad range of other biochemical Steroid Boehm 15:276284, 1081. studies summarized elsewhere in this report. 1 Bunny WE, and Fawcett ilkPoulbaity of a Together the neuroendocrine studies sug- biochemical WM for suicidal potential. Arch Gen Psychlat la232-23e. 191S. gest n concerted effort should be made to 7. &tram WE. Fawcett JA, David JM, Gifford & fur- identify the abnormalities of 5-HT and the ther evaluation of urinary_17-hydroxycerticositrold in endocrine system in anyone who has made a suicidal patients. Arch Gen PlInfolW 21:135.150, 1901 8. Carroll ELI, Oradea JF, Foinbrag M:Suicide, violent suicide attempt and survived. The neuroondocrine dysfunction and CSF MAA tencentra- blunted TRH response may be related to in- dons in depresakin. Recent Pdv. kt rads) B Art GDBurrtiviinsCrtri creased HPA activity. More sophisticated ,0 Sahel, PYammAty. Ammon Prase, Ox- ways of assessing increased HPA activity are , pp 307413, 1991. 9. Chrome GP, !crime Di. Brandon 0, Tomita IA, now available to be applied to suicide re- Idngerholds ACM, Merriam 0, Johnson E0 Upset. MS: conisol resistance: A famitialarcarome and on search.A DST with measurer of model. J Sword Boehm * -571 1983. dexamethasone levels and a measure of 10. Coryid1W: Suicidal behavior and the DST: Look of glucocorticoid receptor sensitivity might assoolation. Amer 61 Pe)ottiaby 139:1214, Ian prove a more sensitive indot of suicidal ac- 11, Corral W. end Seamy Mk Suicide and the dexamothasone suppression test in unipolar depression. tivity. The 1-4 P.M. cortisol output, AMTI kr.xJ Psychlat 138:1120.1121, 1981. and CRF stimulation tests, measures of 12. Cohen SkCushing's syndrome: A psydgalo serum cortisol binding globulin and circadian study of 29 patients. Br J Psychlat 13&120.124. 1900. 13. deKlost ER,_Kovacs CIL Scab° 0, Telegdy 0, rhythm disturbances are among the measures Bohm El, Westing OHM Decreased seratoran turnover in tint might be evaluated. The aim of these the dorsal hippocampus of rat brainshortly after dronalectomy:Selective normalisation site . oor- studies would be to identify possible biologi- doosterone substitution. Brain Res 239:85114963. 1992. cal markers and to further develop an in- 14. Dicidneon $I., Kennett GA, anon & Rscluced 5- dependent behaviour in rats tegrated neuroendocrine-neurotransmitter =1111toonfPrteill:steronsine.trammed. main Res34V26111 hypothesis of the etiology of suicide. For 1985. some biologkal factors that predispose to 15. Egg. AC, RegjoD",usVatedarma MM:Effect of cyptohiptadhie on and TSH suicide, suicidal ideation, nonviolent, non- feignin men. J Clin EndoodnolWar, lethal intent attempts may represent a con- 11 Rime C. Parecohl A, Rondo* kl, Book-Room P, Fegfls0: Effect of two seroionin antlgonists on prolactin tinuum For other factors there may be a 5:57arist secretion in man. CM Endocrinology unique asseniation with each level of suicidal 17. Rnk ES, and Carpenter WT: Further examination intent or in relation to violence. A uniform of a biochemical test for suicide potential. Cis New Sys! way of collecting and presenting data n this 37341-343. 1978. regard might be useful. Methods for quan- 18. Gold PW, Goodwin FK. Wohr T, Rsbar it Pituitary thyrotropin response to thyrotropin reieasing hormone in

BEST COPY AVAILABLE

Iht 2-244 S 3 1-1.114eitzer: The Neuroendocrine System and Suicide

affective ilinen: Relationehlp to spinal fluid amine meta- J Psychlat 135144248, 1975. Wines. Mi J Psycidat 134:1 -.M-1031, 1977. 38. 10411 PAT, Goner/ GJ, Roder AT, Hospelhom VD, 19. Golestein J. Vanharrist L, Brow OM Effect of Mid JP, Meltzer HY: Gluouorticoed receptor function in cypeohiptedine on thyrotropin and pokian secretion in tgranzion. in: Hormones and Depreulon (ern U normal man. Acta EndocAnol .W..M13, 1979. and R Roes, Raven Press, New York. pp 91-112. 1987. 20. Gormley GJ, Lowy MT, Rider AT, Kupelhom VD, Antal JP, Meltzer HY: Gitkviorticald reception In &pees- * Lorry PAT, Rieder AT, Mtel JP, Meltzer HY: sion:Relationship to the dexamethasone suppression Oluoacostioold insistence in depression:fielation be- test. Am J Peyohlat 1421278-12841 1985. tween the DST and lymphocyte senaltiviq to dexamelhasone. Mi J Psychiat 141:1386-1370, 1964. 21. &soden JF, Mats PA, Haskell RF, James NM, Goodman L. Siainer kt, Canall 131Nonnalization of 40. Milanw PAkand Aitken Oft tam deximelhasone dexamehasone supreselon test: A labcratory Index of binding in rat froMic cortex. Brain Res sat- 178-180, 1985. from endogenous depression. Sid Psyohlat 41. Meltzer HY, Mit RC, Tricou BA Fang VS: Ste- t5: 1900. epen uptake in blood platelets and the dexamethasons 22. Halbreich J. Zumoff B, Krum J. FoSueninis asogsZellgset In dePreseed Patients. Pr/01ft Cies The mean 13004000 hr plasma coaled concentration es test for hmrcortisatism. J Clin Enckidnol 42. Meltzer HY, Lowy MT, Koenig JI: The PA 55: 1282-1 254. 1 hypothaland,.pltuwy4drnna1LdstnssicnIn: 23. Holsbur F, Widmann K, Gerken A, Boil E: The dyeunction in nat disonlers plasma dexamethasone variable in depression: Test- I FK Goodwin, Riven Press, Now York, 105482, MT. retest Mass and early blophase kinetics. illsychlatry Res 43. Meltzer HY, Perlin* R, TriCOU 84, Lowy PA, 17:97-103, 1998. Robertson AM Effect of 5-hydroxytryptophen on serum 24. Johnson GE, Hunt 3, Kerr K, Caterson I: °ordeal levels in the melee elective disorders. P. Relation Deurnethasone suppression test asp and plasma to Waldo, psychosis end &peach" synclearne. *ahem dezernethesons 'sub in depressed patients. Psychiatry Psychial 41:379487, 1084. Res 13:305413, 1984. 44.Meltzer HY, Umbeekoman-W8te B, Robinson AG, 25. lOrkegaard C. and Carroll SJ: Dissociation of TSH dcou EU, Lowy MT, Peens ft Effect of and adrenocortical disturbances in endagenour depres- tophan on serum cortisol levee in thetnejor=t sion. Psychiatry Res 3153-264, 1980. orders I.Enhances response in tr,imftt and mania. Arch Gen Psychiat 41:366.374, 1954. 25. IWmen BF, tjungsren J-G, Beck-Fries J, Wetter- twg L WPM of TRH on TSH and prolamin levels in affec- 45. Naushrds PA, Carve,/ PM, Weiner WJ: Modification tive deciders P chiatry Res 14:353.303, 1905. of central serotonergio and clopemitwgic behaviors in the course of chronic corticosteroir administration. Eur 27. Krieger 3: The plasma level of costs& as a predic- Phannsool 72335443 1982. tor of suicide. Ms New Syst 35:237-240, 1974. 45. Ostroff R. Geller E. Bonus OC, Ebersole E Harkins Kronfol Z Grader' JF, Gardner R, Carroll BJ: L. Mason J: Neuroendocilne risk tre of suicidal be- Suicidal behavior and the DST: Lack of association. Amer havior. Arn J Psychlat 13213234325, 1902. J Psychiatry 1391214, 1982. 47. Pataechiall FR, deKloet ER, Chiappinl P, 29. Welch L Central neurotransmitters and the sure- Chierichettl C, Soaccianoce 8, Mgedual I.:Brain tion of 1,werWn, 1.1-1, end TSH. Ann Rev Physiol serotonergic innervation in the regulation of stress 41*M415, 1979. the rat kr Street The role of catecholamtnes 30. Krulich L, Mac:heti A, Coppings RJ, McCann SM. =nino::neurotransmitters. Vol 2. (ads) E Usclin, R KVet- MaAeld Mk On the role of oril serotonergic system nzitafeand Axelrod, Gordon and Breach Beene* in the regulation of the secretion of thyrotropin and probe- , NewYork, pp 787493, 1964. thr: hibiting and prolactin-refeasing effect 48. Restogi RB, and Singhal RL Mren000rtioolds of and quipazine. EndocAnology control 5-hydroxy. typtamine metabolism in rat blain. 105: 290. 1 . Neural Transmission 4203-71, 1978. 31. Levy B, and Hansen E: Failure of the urkwy test 49. Robbins DR. and Alessi NE:Suicide and the for suicide potential: Malys* of urinary 17-hydroxyoor- dexamethesone supassion test in adolescence. Bie tloosterolds in suicidal patients. Arch Gen l'sychlat 20:415- Paw:hist 209019, 1 418, 1989 50. Ratline WH, Rachete CT, Dussailient PA McEwen 32. Lewis DA, and Smith RE:Steroid-induced BS: Adrenal steroid modulation of vasoactive intestinal syndromes: A report of 14 cases and a review peptide effect on serotonln binding sites En If* rat brain ofPeCillirature.JAffect Dia 5:319432. 19* shunt by In vitro quantitative autoradiography. Neuman- 33. Linkowsid P, Van Witten JP, Keekhafs M, Brauman doodnology 40:129-134, 1985. MendiewiczThyrokoOktaue to thyreostimulin 51. %Pesky RM, Krey LC, MoEwen B& Stress down- in affecthely Ill women: avAtO to suicidal behaviour. regulate) cortionterane receptors in a site-speofflo mu- J Plychlat 10401-405, 1983. ner in the brain. Endocrinology 1141A7-292, 1904. 34. Linkowski P. Van Welters JP, Kerkhots U. Gregoire 52. SchischteJA, Ginsberg Bit Sherman BR lieges- F. Bauman H, Mandiewiaz 4: Violent suicidal beluirolor tion of the&woo/look! receptor in human lymphocytes. and the thyrotropin- releasing hormone.thytold-stimulat- J SteroidMaven18:69-74. Ma ktg hormone test A clinical outcome study. *crop- sychobiology 1219-22. 1904. 53. Stantman MN, and Schteingart DE:Neurop- sychlatria manifestations of patients with Cushing's 35. Loeb JN, and Rosner W. Fall In hepatic %skeet syndrome. Arch Intern Med 141215-219, 1901. uoocorticold receptor induced by stress and partial hipalectomy: Evidence for neural mechanisms. En- 54. Targum SD, Rosen L Cepodanno AE: The docrinology 10410034006, 1919. dexamithavane suppression test In suicidal patients with unipoisr depression. Mi J Payohlatr 10177-879, 1083. 36. Loosen PT rue Prange AL Jr: Thyrotropin refus- ing hormone (MI: A ussi7c1 tool far psychonsurcen- 55. Tomei° S, ti E, DeNloofa AF, Rainbow TC, Mc. donne* investiga ion. Psycho- neuroendocrinclogy Ewen BS: Regulation of gluccoortioold reel:state In brain 5:6340. 1900. cortZumwluteenzogy* treatment of edrinalectomized rats. 35:411417, 1982 37. Loosen PT, Prang Ai Jr, Wilson IC: Influence of conlsol on TFE.Induced TSH response in depression. Mr

,2S4 2-245 BESTtoPY WHAM Report of the Secretarys Task Force on Youth Suicide

Traskinen 4 1Wft Mon At, ellriZaLln Lando 0, &haling 0. in thecowat and suicidal petiole. Pah Oen Psychiet37701:W.107-110. 57. RAW ft Depression type Madam *hoe violent suicide aztmilesetanPsyclike Res 2333.331t M. Witairitak Dialtivolok N, Copotav D, Dick H. Chds tie JE, Rik Olueoeortioold recoplors and depression. Brit Med Ifit151)4181, 51Yersvalin el, Clatedattir H, Winne E, Russet. to 4, fAelion P. sMoj 0 Seal M Monetization at the dexametheesne simiession test at discharge: Its prow node yak*. 4 Mf s 5191-ifff, ioea ft Dimino M. (rys5W,PfoNB The validity of the destmetkuone suppression test as a meter for en. waysdepreeelen. Arch Oen Psychtat 41k3474815.

ACKNOWLEDGMENT Supported in part by NIMH grants MH 41683 awl MH 41684 and by the Cleveland Foundation. Dr. Meltzer is recipient cf a Research Career Scientist Award, MH 47808 from NIMH.

2-248 285 A.Roy: Genetics and Suicidal Behavior

GENETICS AND SUICIDALBEHAVIOR

Alec Roy, M.B., Labotatosy of Clinical Studies,Division of Intmmural Clinical and Biological Research, National Institute on Alcohol Abuse and Alcoholism,Bethesda, Matyland

There are five lines of evidence about genetic of the hospital charts of 505 children and factors in suicide. This paper will review the adolescents who appeared at the emergency clinical, twin, Iowa-500, Amish, and Copen- room of the Hospital forSick Children in hagen adoption studies, all of which provide Toronto between January 1, 1970, and evidence about genetic factors in suicide. January 30, 1977, and bad *deliberately in- Most of these are studies of suicidal behavior flicted self-injury with a documented con- among adults but it is likelythat genetic fac- scious desire to die from the injuiy? A tors for suicide are similar in both adolescents control group was derived by exam:ning the and adults. charts of individuals of the same sez and of a similar age who did not have a history of attempting suicide but who had been admitted CLINCAL STUDIES at about the same time to the same emergen- A family history of suicide has been noted to cy room. Theresearchers found that sig- be associated with suicidal behavior at all nificantly more of the youthful suicidal stages of the life cycle. There arefive such attempters, than their controls, had a family studies among adolescents. In 1974 Shaffer historyof suicidal behavior (8.3 percent vs 1.1 (1) reported a comprehensive survey of al131 percent) (Table 1). suicides among children aged 14 years or during the Garfinkel et al. also used Weisman and younger in England and Wales Worden's Risk/Rescue Rating Scale (4) to seven years from 1962 to1968. Suicidal be- obtain ratings of severity for the adolescent's havior in a first degree relative had occurred suicide attempts. Interestingly, they found in seven of these 31 youth suicides (72.6 per- that significantly more of those who k cent). In four relatives this suicidal behavior made serious &sickle attempts had a family had occurred before the child's suicide, but histoiy of suicide (Table 2). in three other cases a first degree relative attempted suicide after the child had com- As found in Shaffer's London study, the mitted suicide. There was also a high suicide attempters in Garfinkel et al.'s study incidence of depression among the first de- also had a significant excess of relatives with gree rektives (20 percent). mental illness and they concluded: *A strong geikztic loading for affective disorder in In 1981 Uhler et al. (2) found that 22 per- families of individuals who attempt suicide is cent of a series of 108 adolescents seen at a supported by our findings of an eight times children's hospital emergency room after at- higher rate of suicide attempts or suicide in tempting sikide had a history that *at least ab- exhibited suicidal be- the families of the index group and an one family member had suicide in the families of havior in the past? sence of completed the controls? They also noted: *Attempted In 1982 Garfinkel et al. (3) reported areview suicide is GB a continuum with completed

4, 2-247 266 Report of the Secretzuys Task Force on Youth Suicide suicide, as demonstrated by the finding that methodolov as vim used with the families of Caere was no family history of completed the suicide victims. Shafii et aL also found suicide in the control group and that the fol- that significantly more of the youth suicide lowup mortality rate from suicide in the index victims, than controls, had a family history of group surpassed the death rate for the con- suicide (Table 3). Again, there was a sig- trols.* nificant difference between the families of In 1985 Shafii et aL (5) reported data derived suicide victims and controls for emotional from psychological autopsies carried out problem in the family. These workers concluded that *exposure to suicide or suicidal after lengthy home visits with the families of behavior of relatives and friends appears zo 20 children and adolescents who had con.- mined suicide in Jefferson County, Louis- be a significant factor in influencing a vul- ville, betwten January 1980 and June 1983. nerable young person to commit suicide: Friends and significant others were also con- In an ongoing study, Shaffer et aL (6) are per- tacted and three extensive questionnaires forming extensive psychological autopsies on Viele completed. Shari et aL were able to a consecutive series of youthful suicides match 17 of the suicide victims with controls under 19 yeats of age occurring in New York drawn from among the suicide victim's City. In a preliminary report of the first 52 closest friends. These controls, and their suicide victims studied, they noted that a sub- families, were interviewed using the same stantial number (20, 38 percent) had a rela-

Family History Differences Between Children and Adolescents Who Attempted Suicide and Controls

Suicide Attempts.* Controls Available WM Chariot. Available 1/Vith Cheroot. Signif Characteristic N N S x2dTP Pantry History of mental Ohms 442 228 51.6 452 74 16.4 122.3 1 .01 History of suicide 443 37 8.3 442 5 1.1 23.952.01 Suicide attempts 28 6.9 5 1.1 Completed suloide 11 2.5 0 Reproduced with permission of the knedoen Journal of Psychiatry.

Table 1.

Variables Discriminating the Severity of 804 Suicide Attempts by Children and Adolescents. Pement of Attempts Low Danger Moderate Danger Severe Danger Significance Variable (N .386) (N . 149) (N -70) X2 df P Famly history al suicide 4.6 12.7 No fandly history of suicide 95.4 87.3 63.9 13.09 4 .01

Reproduced with permission of the American Journal of Psychiatry

Table 2.

2S7 2-248 ARoy: Genetics and Suicidal Behavior tive who had either committed or attempted tempted suicide.Flinn and Leonard (10) suicide. Mthough cautioning that the family noted that among 480 young nonpsychiatric data are complex and are not yet fully ex- subjects, those reporting their own suicidal plored, Shaffer et al. (7) consider that this behavior also reported more knowledge aspect of these youthful suicides may repre- about suicidal behavior in others. sent an environmental rather than agenetic More recently, Murphy and Wetzel (11) sys- phenomenon; they may be imitative acts tematically interviemd a random sample of similar to the recentlywell publicized cluster- all persons seen and admitted during a one- ing of teenage suicides. year period at the St. LouisOaunty Hospital Among adults who exhibit suicidal behavior, following a suicide attempt. Of the 127 there are also surprisingly few studies about patients in the study, 14 peroznt gave a fami- the presence or absence of a family history of ly history of suicide, 24 percent a family his- :Weide.In an mrly study Farberow and tory of attempted suicide, and 6 percent a Simon (15) reported that among 100 suicide family history of suicide threats. One or victims in Vienna and Los Angeles, six had a more of these family suicidalbehaviors was parent who had killed himself, a rate more reported by 36 percent of the suicide at- than 86; times the expected rate. Robins et tempters. Among suicide attempterswith a al. (8) tbund that 11 percent of 109 suicide primary diagnosis of primaty affective disor- attempters had a family history of suicidalbe- der, 17 percent had a family history of suicide havior. Murphy et al. (9) reported that one and 17 percent a family history of suicide at- third of $5 canes to a suicide prevention cen- tempt (Table 4). As individualswith affec- ter had a family history of suicidalbehavior tive disorders comprise a larger proportion of and that this was sipificantly morehiely to suicides than individuals with personality dis- be the case if the caller had himself at- orders, Murphy and Wetzel predicted that

Significant Differences Between Children andAdolescents Aged 12-10 Who Committed Suicide and Matched-Pair Control Subjects.

Suicide Victims Control Subjects McNemar Toat N=20 N =17 Variable N S N % (df= 1) P Family and environmental variables Exposure to suicide 13 65 3 18 6.12<.008 Sibling's or Mend's (attempted or completed suicide) 12 61 2 12 6.12< .008 Parent's or adtit relative's (suicidal ideation, threats, or attempts of completed n.sa suicide) 6 30 2 12 2.25 Parent's emotional problems 12 60 4 24 5.14 < .02 4.3.3,df-1,Pc.04 Reproduced with permission of the .American Journalof Psychiatry Table 3.

2-249 . 2 88 Report of the Secretary's Task Force on Youth Suicide

more of their patients with affective disorder compared with the 5,602 inpatients without could be expected to present a significant such a history. A family history of suicide was suicide risk in the future. Therefore, they found to significantly increase the risk foran concluded that a 'systematic family history of attempt at suicide in a wide variety of diag- such behavior coupled with modern clinical nostic groups (Table 5). Almost half (48.6 diagncsis should prove useful in identifying percent) of the patients with a family histoty those attempters at increased risk for of suicide had themselves attemptet: suicide. suicide.' Mort than half (56.4 percent) of all the The first study providing data about how patients with a family history of suicide had a commonly a faro& history of suicide is found primary diagnosis of an affective disorder and among psychiatric patients was that of Pitts more than a third (34.6 percent) had a recur- and Winokur (12). They found that among rent unipolar or bipolar affective disorder. 748 consecutive patients admitted to a hospi- Recently Linkowski et at (14) investigated tal, 37 reported a possible or definite suicide pest suicidal behavior and family history of in a first degree relative (4.9 percent). In 25 suicide among a consecutive series of 713 (68 percent) of these 37 cases the diagnosis patieMs with affective disorder admitted was an affective disorder, and thaw inves- over several years to the psychiatric depart- tigators noted that the statistiail probability ment of the University of Brussels. They of this distaution occurring by ch. nce was found that 123 of the depressed patients (17 las than 0.02. When the probable diagnoses percent) had a first or second degree relative in the cases of the first degree relatives who who had committed suicide. They also found suicided were considered, in 24 of .te 37 that a family history of suicide significantly inpatient-relative pairings, both menibeis had creased the probability of a suicide attempt affective disorders.Pifts and Winokur es- among the depressed women, especially the timated that 79 percent of the suicides of the risk for a violent buicide attempt. Among the first degree relatives were associated with male depressives, a family history of suicide probabk affective disorder. significantly increased the risk only for a In 1983, Roy (13) reported a study of all the violent suicide attempt (Tables 6 and 7). 5,845 psychiatric patients admitted to the Linkowski et al. concluded that "a positive aark Institute of Psychiatry in Toronto be- family history for violent suicide should be tween isnuary 1974 and June 1981. There considered as a strong predictor of active were 243 inpatients with a family history of suicidal attempting behavior : imajor depres- suicide (4.2 percent), a percentage very shre similar to the 4.9 percent reported by Pitts A family history of suicide has also been and Winokur nineteen years earlier. The found significantly more among psychiatric patients with a family history of suicide were patients who commit suicide (i6). As manic-

Family History of Suicidal Behavior by Broad Diagnostic Groups % with Family History of Diagnostic Group Suicide Attempt Any

Personality disorders 56 20 34 46 Primly affective disorder 29 17 17 39 Other diagnoses and none 42 17 21 patients 127 14 24 36

*My suicidal bohemia includes suicide and attempted and threatened suicide. Reproduced with perntssion of Journal of Nervous and Mental Diseases 11111=11MP Table 4.

2-250 2 9 A.Roy: Genetics and Suicidal Behavior depression is the psychiatric diagnosis most noted to be igssociated with suicidal behavior commonly found among suicide victims, it is among individuals in the last stagesof the life not surprising to find that, across the various cycle. Batchelor and Napier (21) found that published series, approximately 10 percent of among 40 consecutive casesof attempted manic depressive patients have a family his- suicide admitted to a general hospital, aged tory of suicide (17-20). 60 years or over, a family history of suicide was present in 7 (17 percent) of the cases. A family history of suicide haa also been

A consecuthfe owlet of 5845 InpatleMs admitted tothe Clerks Institute of Psychiatry between January 1974 and Juno 1981. Patients, bydiagnostic group, who attempted suicide comparing those with a amity history of suicidewith those without such a history. No family Second- or fimt..degree history of relative suicided suicide

No. (S) No. of No. (%) Diagnostic Group Attempted Attempts Attempted

Schizophrenia 15/33 (45.4) 28 '40/1114(13.5) < .0001 .0001 Unkooinr 13/32 (41.6) 24 50/373(13.4) Bipolar 22/58 (31.9) 48 56/405(13.9) < .0001 Damask(*) neurosis 2W47 (55.3) 45 221/715 (30.9) < .0001 Personality disorder 33/48 es 328/1048(31.3) <4001 Alcohd 3/7 (42.9) 3 42/147(28.5) NS Others 6/18 (33.3) 16 378/1801 (21.0) NS Total 118/243 (48.8) 252 1225/5602 (21.8) < .0001 Reproduced whh permhsion of NOM* of General Psychiatry

Table 5.

Clinical characteristics of the major depressive (MD)patients

MD with MD with MD with past violent past non-violent no suicidal suicidal attem_e_ suicidal attempt attempt Males Females Males Females MalesFemales BP FH+* 3 0 7 14 14 FH 10 11 9 19 72 79

UP FH + 4 12 1 19 16 27 FH 18 15 23 84 92 160 Totail 33 44 33 129 194 280

FH + patients with familial Moly of suidde SP bipolar; UP m. unlpolar Fisproduowl with permission of Acta Psychishica Scandinavia

Table EL

-.290 2-251 Report of the Secretary's Task Force on Youth Suicide

THE IOWA-500 STUDY normal controls (5.5 percent vs 0.6 percent) (30). This finding is impressive evidence for The Iowa-500 study is a frglowup study of j ust the close association of suicide with over 500 psychiatric patmts consecutively psychiatric disorder. admitted to the Uniersity of Iowa The Iowa-500 study bas yielded other infor- Psychiatric Hospital between 1934 and 1944. mative data (31) and Tsuang has recently The 525 patients in the study were chosen because they met certain research criteria. reported the development of this study in They consisted of 200 schizophrenic, 100 another important direction. He and his as- manic, and 225 depressed patients. They sociates not only followed up the psychiatric were compared with a control group of 160 patients and controls but also their first de- psychiatrically normal individuals admitted gree relatives (32). The first degree relatives to the University of Iowa Hospital during the were interviewed, and Twang was intetested same period for appendectomy or herniog- in the answers to four questions which are raphy. Followup studies between 1972 and relevant to the possible role of genetic fac- 1976 revealed that 30 of these 685 subjects tors in suicide. These questions were: subsequently committed suicide; 29 of the 1. Are relatives of patients with suicides were found among the 525 schizophrenia and affective disorders si psychiatric patients and only 1 amung the 160 ject to higher risk of suicide than relatives

Significances end relative odds ratios for the comparisons tested in subgroups of depressive attempters end noo-etternp 'ere Comparison*

pA F Significance of main effect" Polarity n.a 0.045 n.s. n.s. n.s. 0.021

Family history (FH + ) n. s. 0.004 n.s. 0.0003 0.010 0.010 Age n.s. rya n.s. n.s. 0.050 n.s. Relative odds ratio*** Polarity (BP vs LIP) 0.65 2.43

Fairly Nato), (FH + vs PH-) 2.02 3.53 14.98 2.L.3 Age, years 3146 vs 16-30 0.32 46-60 vs 3145 4.92 >60 vs 46-60 2.19 >60 vs 15-30 3.46 No. of patients 260 453 260 453 66 173

*A) Presence versus absenoe of any suicidal attempts; B) viceent MMUS non-violent attempt plus no suiddal attempt C) violent versus non-violent attempt% M mew Ffemales "PS interactions between main effects (potartty x FH + . poiarity x Av. and FH + x Age) non-significant. ***Given only for statistically significant main effect,. Reproduced with permission of Acta Psychistrice Scandinavia'

Table 7.

2-252 29 ARor Genetics end Suicidal Behavior

of nonpsychiatric control patients? deceased relatives were considered, the relatives of psychiatric patients were found to 2. Are relatives of patientswith have a risk of suicide almost six times greater schizophtenia and affective disordets who than the risk among the deceased relatives of committed suicide subject to higher risks the controls (Table 8). of suicide than relatives of patients who did not commit suicide? Among the first degree relatives of the of psychiatric patients, those who were the rela- 3. Is the risk of suicide among relatives tive of one of the 29 patients in the Iowa-500 schizophrenics different from that for study who committed suicide themselves had relatives of mania and depressives? a four times greaterrisk of committing 4. Are the suicide risks different for maleand suicide compared with the relatives of the female relatives of patients from different patients who di.1 not commit suicide. Among diagnostic categories? the deceased relatives, the suicide risk was three times greater (Table 9). The first degree relatives of the psychiatric patients were, found to have a risk of suicide Next, the individual psychiatric diagnoses almost eight times greater than the risk in the were examined.The risk of suicide was sig- relatives of the normal controls. When only

Risk ot Suicide among Relatives of Patientsand Control* Relatives' Suicides Subjects N N % BZ MR(%) SE Patients (N =510) relatives 3941 55 1.4 2348 2.3 .±. 0.3° Deceased relatives 2294 55 2.4 1338 Controls(N-153) All relatives 1403 2 0.1 672 0.3 .1.0.2 Deceased relatives 569 2 0.3 305 0.7 0.5

Morbidity risk. a. 111,2 Eferugsziffer (age-adjusted size of the urnple). MR b. P< .01 (comparison of patients withcontrols). Rapmduced with permission of Journal of MalPsychisky Table &

Risk of Suicide Among Relatives ofPatients with and without Suicide Roitiatives" Suicides Subjects N N % BZ MR(%) SE Suicide (N 29) 7.9 2.5° Ail relatives 193 9 4.7 114 Deceased relatives 136 9 6.6 78 11.5 .±.3.68 No suicide (N -481) 2.1 _±,.0.3 Ali relatives 3754 46 12 2234 Deceased relatives 2158 46 2.1 1259

a. P.c.*, (comparisons of suicidewith no suicide). Roproduced with permission of Journal of ClinicalPsychiatry

Table 9.

2-253 t Report of the Secretary's Task Force on Youth Suicide nificantly greater among the first degree rela- The suicide risk was also examined separate- tiva of depressed patients than it was among ly for the male and female relating of the the relativa of either schizophrenic or manic psychiatric patients. In general, the suicide patient& When the relatives of patients who risk s as higher for male first degree [datives committed suicide were compared, the than it was for females (Table 11). suicide risk was even higher, but it was equal- Thus, this followup of the first degree rela- ly high among the relatives of both depressed tives of the subjects in the Iowa-500 study is and manic patients (Table 10).

Riskof Suicide Among Living and Deed Relatives of Schizopinnial Monica, and Depressivn with Suicide and without Suicide.

Suicides Relatives Patients N N % OZ MR(%) ± SE Schizophrenia (S) Suicide (N a, 8) 41 0 0.0 23 0.0 0.0 No Suicide (N = 187) 1159 9 0.8 723 12±0.4 Total (N. 195) 1200 9 0.8 746 1.2 ± 0.4 Mania (M) Suickie (N -6) 53 3 5.7 32 9.4 5.2 No Suickle (N =86) 748 4 0.5 42C 0.5 Total (N=92) 801 7 0.9 458 1.5 ± 0.6 Depression (D) Suicide (N airt 15) 99 6 6.1 59 10.2 3.9 No Suicide (N =208) 1847 33 1.8 1085 3.0 0.5 Taal (N=223) 1946 39 2.0 1144 4 0.5

Significant oompadsons are as follows: Suicide, S vs M (P < .10).vs D P.< .01): no suicide. S v$ D (Pc M 0 (P<.01): and Total S vs(P< M vs D (Pc .05). Reproduced with permiulon of Journal of ainical Psychiatry

Table 10.

WC of Stlickle Among Relatives of Schizophrenics. Monks, and Depressives by Sex

Suicides Relatives Diagnostic Group N N EIZ MR(%) ± SE Schizophrenia Male relatives 9 1.5 359 2.5 ±0.8a Femalerelatives 590 0 0.0 394 0.0 ± 0.0 Mania Male relatives 395 1.3 221 2.3 1.0 Female relatives 390 2 0.5 238 0.8 0.6 Depression Male relatives 994 30 3.0 587 5.1 0.911 Female relatives 921 9 1.0 568 1.6 0.5

a. P<.01 (comparison of males and females). Reproduced with permission of Journal of Clinical Psychiatry

Table 11. 2"3

2-254 ARoy: Genetics and Suicidal Behavior an important studyand demonstrates that twin pairs where one twin was known to have there are genetic factors in suicide. The main committed suicide. Three of these 11 twin findings are summarized in Table 12. pairs were monozygotic and 8 dizygotic. Ii none of these 11 twin pairshad the other twin committed suicide (24). This negative find- MIN STUDIES ing led him to conclude that "thereis no Ompelling evidence for the genetic trans- statistical evidence for the popular notion mission of manic-depression and that the tendency to commit suicide recursin schizophrenia is that the concordance rate certain families as the result of a special for these psychiatric disorders is substantial- hereditary trait or of a particular type of ly higher among identical twins, who share genetically determined personality devia- the same genes, than it is among fraternal tion.* twins who share only 50 percent of their However, 20 years later in 1967, Haberlant genes (22). Thus, if thepropensity to com- (26) pooled the accumulated data from mit suicide was genetically transmitted, con- studies from different countries. By cordance for suicide should be found more tw then, 149 sets of twins had ircen reportedin frequently among identical than fraternal which one twin was known to haveCOM- twins. This was well stated by Kaftan (23); mined suicide. Among these twin pairs there *If hereditary factors play a decisive role we where both twins had suicide were nine sets of twins should find a concordant tendency to committed suicide. All of these nine twin than in two-egg more frequently in one-egg pairs were identical twins; there was no setof pairs regardless of ordinary differences in en- fraternal twins concordant for suicide (Table vironment. If the main emphasis is placed on certain constellations of nongenetic factors, 13). concordance should be expected in some Foar of these nine monozygotic twin sets twin pairs of either type, who shared the same cancordant for suicide came from the Danish environment and responded to a similar de- Psychiatric lvin Register and their case his- gree of distress with the sametype of tories revealed that in three of them thetwins psychosis.* were also concordantfor manic-depressive In a nother of these nine Kaltman had collected Z500 twin index cases disorder (27). monozygotic hvin sets, the twins were also from mental institutions, TB hospitals.old 1-Nares, and other parts of the population concordant for schizophrenia. Since Haberlandt's review, Zair (28) has reported of New York State. In 1947, he reportedthat a tenth pair ofidentical twins who both corn- among this clinicalmaterial, there were 11 Summary of MogbIclity Risks of Suicide inPatients and Relatives Relatives Patients Relatives of Suicides Diagnostic Group N BZ MR% N BZ MR% N BZ MRS Schizophrenia 8 125 6.4 9 746 1.2 0 23 0.0 9.4 Mania 6 62 9.7 7 458 1.5 3 32 10.2 Depression 15 173 8.7 39 1144 3.4 6 59 0.3 0 7 0.0 Control 1 97 1.0 2 672

az -Benauguiffer (me-adjusted size of the *amok) MR Morbtdfty risk Rsproduced with permission of Journal of Clinical Psychiatry

Table 12.

2-255 294 Report of the Secretary's Task Forceon Youth Suicide

milled suicide. Again therewas an associa- for several reasons. Theyarc an Anabaptist, tion with affective disorderas both twins had nonviolent, pacifist society where themeare killed themselves during a depressive no violent crimes and where there has been episode and both their parents anda no known murder. Alcohol is prohibited and grandmother had also been treated for there is no alcohdirm. Also, the Amishare depression. a wealthy farming community among whom Approximately 1 in 250 live births isan iden- Mere is no unemployment. Their strong tical twin and between 0.5 and 1 percent of religious beliefs foster a tightly knitcom- all deaths among the general populationare munity, and three generations commonly live due to suicide. Thus, it is somewhat surpris- together under the same roof. Family life is ing that only 10 pairs of monozygotic twins valued and divorce precluded. Social isola- concordant for suicide have been reported in tion is rare and the cohesive nature of their the 173 years since the first report of suicide community offers social support for inin twins (29). Also, in 5 of these 10 twin pairs, dividuals who encounter stress or adverse life the twins were also concordant for either events. Thus, several of the important social depression or schizophrenia. Thus, although risk factors for suicide among individuals in twin data provide evidence for the genetic the general population such as unemploy- transmission of suicide, this evidencemay be ment, divorced or separated marital status, partly confounded by the issue of the genetic social isolation, and alcoholismare risk fac- transmission of psychiatric disorders them- tors not commonly found among these selves. An3ish (35-38). This means that genetic factors for suicide may play a larger part in suicides occurring among the Amish. Suicid in Twins Not surprisingly, suicide is a relatively Parmber of twin pairs rare Type of Number of where both twine event among this group of Amish. In fact twins twin palm committed suicide (S) Egeland and Sussex were only able to find 26 Identical 51 9* (17.7tia) suicides over the 100years from 1880 to 1980. Over these 100 years the suicide rates Fraternal 95 0 %) among these Amish have consistently been substan- P <0.0001 tially lower than the rates for the rest of the * Four of these 9 seta of twins have been United States. reported twice. They are Included in Habotiont% 1267 roview(211) end hove atso Egeland and Sussex's team used the boon roportod fri detail In 1970 by Juoi-Moison and Videtwoh.(0) Schedule for Affective Disorders and Riprodeood with ponnission of Climes of the Schizophrenia-Lifetime Version (SADS-L). Nervous System(22) They conducted an average of six interviews, with various family members, for each of the Table 13. 26 suicide victims. A five-member psychiatric board used the Research Diag- nostic Criteria (RDC) to make psychiatric THE AMISH STUDY diagnoses based on these interviews andsup- In 1985 Egeland and Sussex (33) made their plemented by information from othersour- rust report on the suicide data obtained from ces. The first important finding of the study the study of affective disordersamong the was that 24 of the 26 suicide victims met RDC Old Order Amish community of Lancaster criteria for a major affective disorder. Eight County in southeastern Pennsylvania. This had bipolar I, four bipolar II, and 12 unipolar is a continuing study into the genetics and affective disorder. A further case met Cle course of illness of the affective disorders dial,. mile criteria for a minor depression. among this population (34).Suicide re- Furthermore, most of the suicide victims had search among the Amish is of great interest a heavy family loading for affective disorders.

2-256 295 ARoy: Genetics and Suicidal Behavior

For example, among the eight bipolar I order. suicide victims the morbidity risk for affective Interestingly, the converse was not true as disorders among their 110 first degree rela- there were other family pedigrees with heavy tives was 29 percent compared with the 1 to loadings for affective disorder but without 4 percent found among the general popula- suickks. It is ako of note that the morbidity don. risk for affective disorders among 170 first The second finding of the study was that al- degree relatives in other bipolar I pedigrea; most three quarters of the 26 suicide victims without suicide was, similar to that found in were found to cluster in four family bipolar pedigees with suicide, also in the 20 pedigrees, each of which contained a heavy percent range. Thus, in this study, a familial loading for affective disorders and suicide. loading for affective disorders was not in it- Figure 1 shows a vety heavy loading for affec- self a predictor for suicide. tive disorden in one family where there have The third finding of the study was that onl) been seven suicides. AU seven suicide vic- six of the 26 suicide victims (23 percent) had tims were found among individuals with received any psychiatric treatment despite definite affective disorder. Figure 2 shows a the fact that 24 of them had severe affective second pedigree, also with a heavy loading disordets whose natural history is usually that for affective disorders, where there have of recurrent episoda. The other 20 suicide been six suicides, five of which were found victims had either never received any medi- among individuals with definite affectivedis-

Flgum 1.

29 6 2-257 Report oftheocietery. s Task Fomeon Youth Sukide cal treatment for their psychiatric disorder, illness in the kinship and suggests the role of or they were seeing a Emily doctor at the inheritance.' time they annmitted suicide, or they were planning to seek help for themselves. DANISH-AMERICAN Egeland and Stases =winded: 'Our study ADOPTION STUDIES teplicates findinp that indicate an inczeased suicidal risk for patients with a diagnosis of The strasgest evidence that we have for the major affective disorder and a strong family presence of poetic factors in aidicide comes from the adoption studies carried out in Den- historyofsuicide." They also noted: "Bipolar and unipoiar illness conveying a high risk as mark by Schubinger, Kay, Wender, and a diagnostic pattern in pedivees. The num- Rosenthal (39-41). The strength of the ber not recebing adequate treataent for adoption strategy is that it is one of the best manic-depressive illness (among the ways to tease apart *nature from "nurture issues. This is because indMduals separated suicides) supports the =mon belief that in- at birth, or shortly afterwards, share their tervention for these patients at risk is recom- genes, but no subsequent environmental ex- mended. It appears most warranted in those periences, with their biological relatives. In families in which there is a family history of contrast, adoptees share their environmental suicide. The clustering of suicides in Amish experiences through childhood and adoles- pedigrees follow the distribution ofaffective cence with their adopting relatives but they

Flow 2.

2-258 297 ARoy: Genetics and Suicidal Behavior share no genes with them. alcoholismas these are chronic disorders with frequent relapses usually requiring The Psykologisk Institut has a register of the psychiatric hospitalization. Schulsinger et al. 5,483 adoptions that occurred in greater (39) therefore proposed that there may be a Copenhagen between 1924 and 1947. A genetic predisposition for suicide inde- screening of the registers causes of death pendent of, or additive to, the major revealed that $7 of these adoptees eventual- psychiatric disorders associated with suicide. ly committed suicide. They were matched with adopted controls for sex, age, social class Worrler et al. (41,42) went on to study of the adopting parents, and time spent both another group of the Danish adoptees. with their biological relatives and in institu- These were the 71 adoptees identified by the tions before being adopted. Searches of the psychiatxic case register as having suffered causes of death revealed that 12 of the 269 from an affective disorder. They were biological relatives of these $7 adopted matched with 71 control adoptees without af- suicides had themselves committed suicide fective disorder. The results of this study compared with only 2 of the 269 biological showed that significantly more of these adop- relatives of the 57 adopted controls. This tees Ali affective disorder, than their con- a highly significant difference for suicidebe- trols, had committed suicide. Thus this study, tween the two groups of relatives (Table 14). too, demonstrates that there is a genetic com- Mile of the adopting relatives of either the ponent to suicide (Table 15). suicide or control group had committed Of further interest in this study was the ex- suicide. amination of the adoptee suicide victims and Also, these striking results are of additional their biological relatives by the type of affec- interest because the suicides were largely in- tive disorder suffered by the suicide victim. It dependent of the presence of psychiat ric dis- was particularly adoptee suicide victims with order.Schulsinger and coworke3 also the diagnosis of *affczt reaction* who had sig- investigated whether or not the names of the nificantly more biological relatives who had 12 biological relatives who c..mmitte4 suicide committed suicide than controls. The diag- appeared on the psychiatric case registers. nosis of "affect reaction* is %zed in Denmark They found that 6 of these biological suicide to describe an individual who has affective relatives had had no contact with the symptoms accompanying a situational crisis- psychiatric services and thus presumably did -often an impulsive suicide attempt (Table not suffer from one of the major psychiatric 16). These findings led Kety (42) to suggest disorders commonly found among suicide that a genetic factor in suicide may be an in- victimsmanie depression, schizophrenia, or ability to control impulsive behavior which

Ircidence of Suicide in the Relatives incidence of Suicide In the Relatives of Adoptees Who Committed Suicide of Adoptm Who Have Suffered a and Their Controls Depressive Illness and Their Controls

Biological Adoptive Biological Adoptive Adopts** Relatives Relatives Adoptees Relatives Relatives

5r Wooten t2 (4.5%) (0%) 71 adopts** _31 (It%) (0.5%) died by suicide 269 2148 with deprossioh 407 187 57 matched g(0"2(0%) 71 matched 1 (0.316) (0%) control edoptees 209 150 control adopted* 360 171

P,c0.01 P<0.01 Reproduced with permission of Marne and Publithed with permiselon of Williams end Wiuns %thins 411111. Table 14. Table 15.

a 2-259 298 Report of the Secretary's Task Force on Youth Suicide has its effect independently of, or additively SUMMARY to, psychiatric disorder. Psychiatric disorder, or environmental stess, may serve *as poten- Suicide, like so much else in psychiatry, tends tiating mechanisms which foster or trigger to run in familia. The question is what is the impulsive behavior, directing it toward a being transmitted. No doubt in some youth- suicidal outcome (42)." ful suicide victims what is transmitted is not a genetic factor but a psychological factor. Kety (42) also noted that there has been The family member who has committed much recent work on the biology of impul- suicide may serve as a role model to identify sivity and that disturbances in central with, and the option of committing suicide serotonin systems have been described in becomes one possible "solution" to in- relation to suicidal behavior in personality tolerable psychological pain. However, the disordered individuals and in patients with family, twin, and adoption studies reviewed various other psyuhiatric disorders (reviewed here show that there are genedc factors in in this volume by Asberg). In this regard it is suicide. In many suicide victims, these will be noteworthy that Buchsbaum et al. (44) found genetic factors involved in the genetic trans- that significantly more college students with mission of manic deprasion, schizophrenia, los 'eves of the enzyme monoamine oxidase and alcoholism--the psychiatric disorders (MAO) in their blood platelets had a family most commonly associated with suicide. history of suicide! behavior compared with However, the Copenhagen adoption studies students with high platelet MAO levels. This strongly suggest there may be a gene* fac- enzyme is involved with the metabolise of tor for suicide independent3or additive to, serotonin.Furthermore, as there is some the genetic transmission of psychiatric disor- evidence that lithium may be useful in impul- der. Interestingly, support for this possibility sive and aggressive individuals (43), Kety comes from the recent Amish studies, which (42) also suggested that controlled trials of showed that suicide was much more Rely to drugs acting on central serotonin system occur when an individual had genetic vul- might be informative among patients who ex- nerabilities to both suicide and to affective hibit suicidal behaviors. Incidence of suicide in the biological relatives of depressive end control adopts**

Incidence of suicide in Diegnvsis In Adopts. biological relatives Significance Affective reaction 6 (7.6%) P <0.0004* 66 Neurotic depression 3 (2.4%) P <0.056 127 Bipolar depression 4 (5.3%) P <0.0036 76 Unipolar depression 3 (2.2%) P <0.06/ 139 No mental illness (0.3%) 360

* oompirod wIth biological relatives of control adoptooswith no known history of mentei Onou Reproduced with permission of Mims end Mins

Table 16.

2-260 299 ARoy: Genetics and Suicidal Behavior

Mness. rbype,Rsk factors for suicide In griatric patients. Nth Gen Psychiatry, 19823121009-1 There is a possible practical implication for 17. Roy k Genetics of suicide. Psychobiology of suicidal behavior. Moak New York Academy of Science, the prevention of youth suicide arising from 1985 in press. this review.It is that an adolescent who t& _Roy k Fru* Wary of suicide in affective disor- develops a depressive episode, or wb1 ex- der Poem nynimoir, 19954e317419. 19. airy k Family history of suicide In manic-doom- hibits suicidal behavior, and who has a tami- eke patients. J Med Olson:fees, 19854187489. ly history of suicide might be considered to be SA Roy k Ganglia factors hi suicide. Psychopharm at risk of committing suicide. Such an in- Buil, 1969; in WM, 21. Batchelor I. Napier Pat:Attempted suicide in old dividual might, therefore, be more closely as- age. Br MsdJ 1953211854190. sessed and followed, particularly with a view 22. TsuPel ilkGenetic factocs In sutide, Dis Nom to determining whether he is developing a Syst, 1977;3b:498-1501. 23. Kalknan F. Anastasio M:TVAn studies on the recurrent affeztive disorder for which psychopathology of suicide. J Nem Mont Die, psychopharmacological intervention might 1947;10640-55. be appropriate. 24. Kaltman F. DePorte 4, DePorte E. Feingold L Suicide in twins and **children. Am,' Human Genetics, 19492 113-125. 25. Habsdandt W Der sullid ale grenetisches REFERENCES tr.dlings und *teen *Julys.). Anthrop Me, 1955;2965-

I.Shaffer 0: Suicide in childhood and early adoles- 25. Habertandt W:Apcwtecion is la genetic* del cence. J Chlid Psycho, Psychiatry, 1974;5175-291. suloido. Foils Clin int, 196717:319422. 2. Tishier C, ttAdolescent 27. Just-Nielsen N, Videbech T:A twin study of suicide Mem" Some &Adds Lift suicide. Acta Genet Med Gemelloi, 1970;12307-3l0. Threat Betsy, 1981;11: 28. Zaw K:A suicidal family. Br J Psychiatry, & Gerflnicela Posse A, Hood J: Suicide attempts in children anti adolescents. Am J Psychiatry, 1901:159:68.19. 1982131k12574281. 29. WUiamsS Ot Lcreenberg, 1941;1918. 4. Weissman A, %cordon J:Mk-rescue rail% in 30. Tr.cang PAT:Suicide in schizophrenia, mentos. suicide assessment Arch Gen Psychistry. 197Z20.553- depressives. and surgical wings: A comparison with 530. rg451 !Vern suicide modality. Arch Gen Psychlehy. 5. Shad! M, Caffigan R, Derrick k Psychological autopsy of completed suicides in children 31. Tsuang kfis Woolson RF:Excess matallty in and adolseoents. Am J Psychiatry, 1955 142108140114. schizophrenia and effective disorders: DO suicides and accidental deaths solely account for this excess? kah Gen 5. Shaffer 0, Gould M, Traubman P:Suicidal be- Psychiatry, 19783511814 185. Milky in children Piper presented at the Ganferemo on Ay of Suicidal Behavior. 32. Tsuang MT:Risk of suicide In the refidives of - New York Academy Sciences, New Yost, September ==manalcostaresives, and controls. 4 CHn 198& ,n11;s8t,44: 7. SheSfer D:Ouott a In Mut Psychiatry News, 33. Egeland 4, Sussex J: Suicide and burgh, loading 1985. for affective disorders. JAMA; 1985254:915-918. & Robins E, Schmidt E, O'Neal P: Some intermit 34. Egeland A. Hostetter AM: knish study I. Affec- dons of social factors and clinical diagnosis In attempted dye disorders among th. Amish, 1976- 19$0. Am J suicide. Pm J Psychiatry, 1957;114:221-231. Psychiatry 1983; 140:N4 9. Wilily I3, Wetzel R. Swallow C, McClure J: IAMo Robins E. MorphyG, Wikinson R, Gusn.r St KAWS calls the suicide prevention center: A study of 55 Jr: Some Onkel observations in the prevention of suicide calling on their own behalf. Am JPsycrifest7 based on e study of 134 successful suicides. Mi J Public 1989125:314-324. Health, 1959:49:016489. " Flinn D, 1.onardC : Prevalence of suicidal ideation 31 Carpet Ts noisy It A study of suicide in the Seat- rm. boa 'savior among basic trainees and college students. tle area. Compr Psychiatry, 1980.1:340459. 1972137317-321 37. Berreciough B, Bunch J, Nelson B, Sainsbury P: 11. zrphyO,tzslR Family history o( suicidal be- A hundred oases of suicide. Clinical Aspects Br J havior among suicide attempters. J New Mont OIL Psychiatry, 1974;125:355.373. 198217t28540. 38. kfurphy_13, Robins E:Social factors In suicide. 12. Pitts F, Wnokur F: Ailective disorder. Pad 3 (Diag- AMA, losrosemaaot nostic correlates and Widows of suicide).Nov Mont 39. fichuleInger A, KeV Wender P. A Dies t984; t311: 170481. family a* of suicide. In Prevention and Treat- 13. Roy k Family history of suicide. Arch Gen ment of ~ye Disorders. & E. Stromgren PcorhinitY, 198340971474. (ida). 277-287 Academic) Press Ino. New Yost. 14. Unkowski Ps Mascieleer de V. MendlevAcx 40. Sohuleinger F, Kety S. Rosenthal 0, Wender P: Sukkial behavior in 7r34.2drstelve Moss. Acta 1981. A fcm* study of suicide. Paper presented at the Nychiat Stand, 1989. Third Wok, of Biological Psychiatry, Stock- mSwedenMivress 15. Farberm. N, Simon * Suicide in Los Angeles and hol. Vienna: An Intercultural study of two Mies. Public Health 41. Wender P. Kety 5, elhulsIngsf F:Arch Gen Rep, 1951104310403. Psycfgat (In press).

300 2-261

BEST COPY AVAILABLE Report of the Secretary's Task Force on Youth Suicide

42. Kety & Genetic Fact= in Suicide. Chapter in Suicide. A Roy We. 101103, Miami end %Was. Bal- timore. 43. Shard M, IMiW J Ilddipte C. 'ham 5: The effect et lithium on Inhaler in man. kn J Psychiatry,lialt140.14t3. 44. Buchsbaum Al, Ceursey R. Murphy D:The biochemical Mett-dsk paradigm: Behaviend and famMal correlates of bus platelet moneamine exklase activity. Science, 11)7&330341.

301 SUMMARY AND OVERVIEW OF RISKFACTORS IN SUICIDE

Maeda K Goodwin, M.D., Scientific inr-aor, National Institute of Mental Health, NIH Clinical Center, Bethesda, Maryland

Gerald L. Bmwn, MD., Senior Investigator, Biological Psychiatry Branch, Intramural Research Program, Nadonal Institute of Mental Health, Bethesda, Maryland

INTRODUCTION This coliSerence brings together two inde- have committed suicide; neurocndocrine pendent traditions in the study of suicide correlates; and genetics. Each of these sub- the psychosocial and the psychiatric- jects has been well reviewed by the previous biomedical. Although we focus here primari- authors, Asberg (1), Stanley (2), Meltzer (3), ly on potential biological risk factors, we also and Roy and Kety (4). Rather than covering highlight opportunities for interdisciplinary the same ground again, we will briefly sum- cooperation that could enhance scientific un- marize these excellent reviews and then add derstanding and lead to improvements in some comments of our own. treatment and prevention. The educational effiirts that we hope will grow from this con- CSF BIOCIMMICAL STUDIES ference should, wherever possible, be based on firmly established knowledge, rather than Asberg reports that several studies have simply belief systemshowever compelling found an increased rate of suicide attempts they may seem. Much of the apparent dis- in psychiatric patients with low CSF con- parity among the different schools of thought centrations of the serotonin metabolite, 5- represented at this conference might be due hydroxyindoleacetic acid (5-141AA). Many to the fact that different approaches focus on of the study populations were depressed different populations. By and large, patients, but similar studies involving popula- psychosocial studies have focused on in- tions with other psychiatric diagnoses, such dividuals with suicidal ideation who contact as personality disorders (5,6) and suicide prevention programs, whereas the schizophrenia (7,8), suggest that the associa- psychiatric-biomedical studies have focused tion between low levels of CSF 5-HIAA and on actual suicides or major suicide attempts, suicide is not confined to depression. The as- largely among individuals with a major sociation may not, however, be present in psychiatric diagnosis. manic-depressive patients. Biological investigations of suicidal behavior Asberg notes that evidence to date also sug- have been most active in the following areas gests that low concentrations of CSF of study:brain chemistry as measured homovanillic acid (WA), a metabolite of through cerebrospinal fluid ((SF); postmor- dopamine, are associated with suicide attem analyses of tissues from individuals who tempts in depressed patients. This associa-

30 2 2-263 Report of the Sece.lary's Task Force on Youth Suicide

tion has not been founil t other psychiatric ments of behavior that biological systems populations. remain stable over time. Asberg points out that CSF metabolites art Asberg characterizes low (SF 5-HIAA as an only indirect measures of amine turnover in indicator of vulnerabnity rather than as a the brain. She notes some confounding in- marker of the state of depression. She cites fluences on such measuresage, sex, body longitudinal CSF studies as well as genetic height, concentration gradient, circadian and evidence in support of this conclusion. seasonal rhythms, drug effects, medical dis- The mechanisms by which serotonin binction eases, diet, and physical activity. influences suicidal behavior are unknown, Methodological problems include lumbar- puncture location and subject position, but Asberg points out that some evidence links impubivity and difficulties in handling amount of CSF drawn, handling and storage, aggression to suicide attempts. Formula- assay methods, and within-subject stability of tions of such a link come from animal studies, CSF 5-HIAA1 Despite these problems, the classical psychoanalytic observations, and be- biochemical methadoloty is elegant compared to measures of suicidal behavior. havioral and biochemical studies in clinical populations. Among problematic methodological factors Asberg notes, is the lack of clear definitions Asberg suggests that including lumbar punc- of sukIde attempt in most studies and the as- tures to obtain CSF is a reasonable part of sumption made in 'life history* measure- the clinical assessment in some psychiatric patients.In conclusion, she proposes a

Self-Rated Childhood Problems and CSF 5-Hydroxy Indoleacetic Acid (5-HIM) Levels of 12 Male Patients with Borderline Personality Disorder°

20 - No Suicide Attempts 15 - x Suicide Attempts as Late Adolescents 10 -

1 0.0 1.0 2.0 3.0 4.0 Mean Self-Rated Childhood Problem Score

a The selt-rated scores were negatively correlated with CSF 5-HIAA (r=-0.63. p

bMeasured by gas chromotography mass spectrometry

Figure 1.

2-264 303 F.ICGoodwin: &misty and Overview of Risk Factors..

model of interaction between biological vul- depression per se.He also cautions that nerability and psphiatric symptom (e.g., none of the postmortem studies he reviews

-depression), psychological factors, adverse focus on youth. environmental occurrences, and childhood Postmortem enzyme studies of suicide reveal history. no consistent patterns. Stanley's own group Of most interest to us is the factor that im- found no differences between suicidal sub- plies a trait, a relatively stable characteristic jects and controls in either the A or B forms that changes little, if at all, with clinical con- of monoamine orddase (MAO). Previous dition, in contrast to the factor that changes findings of lowered brain MAO activity with clinical state. Our data (5,6,9) show that among suicide victims may have been related young adults with a history of aggressive and to alcoholism. impulsive behavior (including a childhood Earlier studies of serotonin and its metabo- history) have low levels of C3F 5-HIAA. lite 5-HIAA generally found decreased levels Such evidence suggests that these behaviors in subcortical nuclei, including the reflect, in part, certain trait characteristics, hypothalamus and parts of the brain stem which may be relatively independent of Some of the variability in these studio; is an- changing environments and personal doubtedly due to confounding variables, such relationships. More ambiguous are data as manner of death (e.g., drug overdose), and linking CSF 5-HIAA levels to Asberg's the extent of delay between death and autop- categories of nonviolent and violent suicide, sy (*postmortem interval"). the latter usually associated with the lowat levels of CSF 5-HIAA. Is the violent be- Receptor-binding studies may be less in- havior a reflection of an isolated state or is it fluenced by these variables. Stanley's group one episode in a long-time history of similar controlled for age, sex, and postmortem in- behavior, not all ofwhich may have been self- terval, as well as the manner of death, which destructive or suicidal? in each case was sudden and violent. They found significantly fewer imipramine-bind- Some studies indicate that young children ing sites in the frontal cortex of suicide vic- have higher !Nets of CSF 5-HIAA than do tims, a finding consistent with the association adults. It is unclear whether this difference between suicide and low central serotonin relates to the observations that serious function since imipramine binding is a suicidal behaviors sire more common in marker for presynaptic serotonin nerve ter- adults and adolescents than in young minals. Of the four other imipramine-bind- children.In our replication study, young ing studies published, three have replicated adults and late adolescents with the lowest the findings of Stanley and his colleagues. levels of CSF 5-HIAA had childhood his- However, generalizing from these results in tories that included many affective-impulsive adults to suicide in the young is risky because symptoms, but no reported incidence of some relevant neurochemiad measures, such childhood suicidal attempts. (Figure I) as imipramine binding, are known to be influenced by age.Thus, Stanley's call for POSTMORTEM STUDIES studies of young people, both normal con- Stanley reviews postmortem studies of trols and suicide victims, is well taken. He biogenic amines, their metabolites and en- points to the need for better diagnostic infor- zymes, and remptor binding. He points out mation and greater specificity in reporting re- that available information indicates that lated aggressive, violent, and impulsive suicide victims are diagnostically behavior. He dons by suggesting a be- heterogenous, so that differences in havioral-biochemical profile to help neurochemistry found at autopsy may be clinicians identify patients at high risk for more related to suicidal behavior than to suicide..

2-265 Report of the Secretwys Task Force on Youth Suicide

Issues that seem important to us include the dexamethasone suppression test (DST) non- fact that subjects in most postmortem studies suppression and suicidal behavior seemed to of serotonin (5-HT) and its metabolite (5- be posiave in earlier studies, but more recent HIAA) are predominantly depressed studie. bring this association into question. patient& None of the reports indicate any at- Some earlier indications that the administra- tempt to detennine a histoiyof aggremive be- tion of the test dose of dexamethasone might haviors.This absence of data on increase the risk of suicidal behavior seem aggressive-impulsive behavior in completed not to be borne out in the more recent suicides seems especially to highlight the analyses. portance of collecting careful psychological Meltzer discusses new work on glucocor- autopsy data in conjunction with the biologi- ticoid-receptor-resistance (GCCR) studies. cal assessments in postmortem studies. In an earlier report, he indicated that the A curious discrepancy is apparent in this plasma 5-hydroxytryptophan (5-HTP)-in- literature. The lowered 5-11T functioning in duced cortisol rmponse was negatively corre- the frontal cortex suggested by receptor lated with CSF 5-HIAA; such a correlation studies has not been subjected to pmtmor- might link the urinary and plasma cortisol tem analysis, which is so aptly suited to findings with CSF 5-HIAA in suicidal anatomical localization. Older studies have, patients. In ongoing studies of patients with however, shown lower levels of 5-HT and/or major affective disorders, the elevated car- 5-111AA in the midbrain-brainstem area, the tisol response to 5-HTP seems particularly nucleus acumbens, and hypothalamic area, linked to a history of violent suicide attempts. the latter of which has also been shown to Early studies in patients with histories of have increased 5-HT presynaptic reccptor violent suicidal attempts showed a blunted changes (10,11,12). These apprent incon- thyroid-stimulated-hormone (TSH) sistencies may be clarified as postmortem and response to thyrotropin-releasinghormone receptor studies move beyond the prelimi- (TRH); more recent work, including some of nary stage. Catecholamine neurotransmit- Meltzer's, does not always confirm the early ter-metabolite studies are not consistent and reports, however. Evidence that the TSH relevant receptor data are lacking; the same response is negatively related toCSF 5- can be said for the cholinergic systems. HIAA is consistent with a relationship between low levels of brain serotonin and NEUROENDOCRINE STUDIES suicidal ideation, which had been shown to A link between suicide and hormonalbe positively related to the TSH response. functioning is strongly suggested by the It seems particularly important to us that con- suicidal behavior of individuals with en- tinued efforts be made to integrate the docrinopatbies particularly Cushing's neuroendocrine and the 5-HT data. Animal syndrome. In his remiew of endocrine abnor- studies indicate that increased plasma car- malities in suicide, Meltzer suggests that they tisol decreases 5-HT synthesis in the CNS via may serve as weak markers for suicide, but activation of liver tryptophan hydrcacylase, further research is more likely to show their which shifts peripheral tiyptophan to the role in influencing neurotransmission, which kynurenine pathway from its availability to in turn may be more directly associated with the CNS for 5-HT synthesis (14). Humans causal factors. with carcinoid syndrome (5-HT-secreting Although urinary corticosteroids were the tumors in the small intestine) are often depressed, insomniac, and irritable, if not original biological variable associated specifi- overtly aggessive; they may have a decreased cally with suicidal behavior (13), plasma and CSF studies are still too few to warrant any 5-HT synthesis in the CNS secondary to conclusion. An association between decreased availability of the precursor, tryptophan. If so, the treatment of these patients

2-266 3r5 F.K9oodwin: Summary and Overview of Risk Factors.. . with para-chlorophenylalanine (PCPA), a 5- suicide, half had a major psychiatric illness HTsynthesis inhibitor that aosses the blood- among their biological first-degree relatives; brain barrier (12), might be expected to the remainder had family histories of suicide worsen their ONS symptoms, as has been without a major psychiatric diagnosis but reported (15,16). often with a history of aggressiverimpubive symptoms. Psychiatric and/or suicidal his- GENETIC STUDIES tories were virtually absent among the adoptive families or their relatives. The Amish Roy and Kety point out that suicidal behavior study has shown that a heavy loading for af- and depression are common in first-degree fective disorder and for suicide can be some- relatives of children and adolescents who what independent of each other. commit suicideboth before and after the suicide. A significantly higher rate of It seems to us particularly important to pur- psychiatric illness has also been found. In his sue the identification or clarification or pos- ongoing study, Shaffer has found that 38 per- sible biological differences within special cent of suicidal adolescents have a family his- families or populations with known, tory of suicide.Increased incidence of epidemiologically characterized behavioral suicidal behavior has also been shown in rela- vulnerabilities. An example of such work is tives of suicidal adults; a recent, controlled the ongoing attempt to clarify how study found that among 127 patients ad- chromosome 11 may be related to the occur- mitted to a hospital following suicide at- rence of affective disorders (and possibly to tempts, 24 percent gave a family history of suicidal behavior) in the Amish population, suicide attempts. Psychiatric diagnoses com- and whether this finding can be characterized monly associated with a family history of in other populations. Clearly, genetic factors suicide are personality disorders and affec- are involved in suicidal behavior, but that is tive illness, especially manic-depressive ill- not to say that environmental (learning) fac- ness. tors are not operating. The familial association with suicidal be- RESEARCH STRATEGIES havior is clear enough, but it begs the question of its source--in psychosocial or Before proposing research strategies, we will biomedical-genetic fsctors or some combina- briefly summarize the five categories of risk tion of them. First one must know what is factors that ought to be considered. transmitted: is it suicide per se, a vulnerable personality (i.e., aggressive/impulsive), or an Behavior illness (i.e., affective disorder)? Secondly, one must know whether proneness to suicide Longitudinal history from childhood to the is transmitted genetically or environmentally present. As noted above, not only does a his- through learn*. tory of aggressive and impulsive behavior seem to be associated with vulnerability to Twin and adoption studies address both of suicide, but affective or impulsive symptoms these questions. Evidence from all the twin in childhood are related to low levels of CSF studies together appears to show a genetic 5-HIAA in adolescence, which in turn is as- vulnerability both toward suicidal behavior sociated with increased suicidal behavior. In and severe mental illness. addition, differences in drug states may be The Danish adoption study provides espe- relevant to behavioral predispositions. The cially strong evidence for a genetic factor in natural history of an illness should also be suicide, and further, the data indicate that taken into account; for example, manic- this genetic vulnerability to suicide can be in- depressive illnessa condition with a high herited independently of overt psychiatric ill- risk for suicidehas been hypothesized to ness. Among the adoptees who committed begin when a genetic vulnerability interacts

2-267 13 I) 6 Report of the Secretary's Task Force on Youth Suicide with an environmental stressor, and once disoider, and indeed we believe that this will such an interaction begins, it takes on a life turn out to be the single best approach to of its own, eventually requiring little environ- preventing actual suicides. mental stress to produce recurrences. Aggression history. The single best predic- Medical History tor for aggrasive behavior is a history of such The presence of a major medical illness is a behavior, just as the best single predictor for risk factor for suicide.In addition, some suicidal behavior is a history of it (17,18). specific medical disorders have been associated with a higher incidence of suicidal or Family History aggressive behavior.Most are associated with disturbances in corticosteroid or The data reviewed here make a strong case serotonin metabolism. They include: for the importance of assessing family history of both suicidal and aggressive-impulsive- Endocrinopathies, particularly Cushing's violent behaviors as well as of psychiatric ill- syndrome (corticosteroids). nesses. Metabolic disorders, i.e., carcinoid syndrome (serotonin), Lesch-Nyhan Psychiatric History and Diagnoses syndrome (serotonin alteration and ag- The evidence further argues for determining gressive behavior (20), incidence of whether a psychiatric illness is present in insuicide is unknown). dividuals who are prone to self-destructive Neurological disorders, ie., Parkinson's behavior. Indeed, the number of individuals disease (21,22) and epilepsy (low CSF who actually kill themselves in the absence of HIAA) (23,24) and Gilles de la Tourette a psychiatric illness would appear quite low. syndrome (sometimes associated with As noted earlier, the psychosocial focus of low (SF 5-H1AA and, often, aggre&sive the suicide prevention movement has behavior (25), but incidence of suicide in primarily derived its experience from work- unknown). ing with people who threaten or attempt suicide, while the clinical-medical focus on suicide as an integral part of major psychiatric Biological Parameters illness has primarily derived its experience In the future, we should do studies to clarify from working with patients with completed the seasonality of suicide (26), receptor or nearly completed suicide. Although the studies in vivo, controlled pharmacological domains of attempters and completers do studies of suicidal and aggressive behaviors, overlap somewhat, by and large they repre- and seriously consider doing lumbar punc- sent different populations (19). The major tures as part of the osychiatric assessment in affective illnesses, particularly manic-depres- some patients. Very few of such studies are sive illness, alcoholism, schizophrenia, and being done now. The development of a clini- other psychoses all have substantially higher cal-risk profile for suicidal behavior might be associations with suicidal behavior than one very useful to clinicians. An important and would expect for a normal population. In- unanswered clinical-scientific question re- dividuals suffering from these major lated to the biological factors that may con- psychiatric disturbances make up the tribute to the high incidence of suicide in majority of completed suicides. To this we manic-depressive illnen, given the fact that can now add that the risk is increased when the association between low levels of the one of these disorders occurs in an individual semtonin metabolite and a history of suicide, with aggressive/impulsive *personality" traits. consistent across a wide variety of disorders, On a practical level, a proper diagnosis can does not hold for manic-depressive patients. lead to appropriate treatment for the specific Nevertheless, the tripartite relationship between suicide, aggression, and affective ill-

2-268 3C.7 F.KGoodwin: Summary and Overview of Risk Factors.. .

ness remains an important cornerstone on sonality" dimensions as we do to formal which to build further understanding. psychiatric diagnoses. In this regard, we (Figure 2) should evaluate the relationship between formal psychiatric profilese.g., -One of our stated goals in the introduction sensation-seeking (27)and relevant was to promote and foster interdisciplinary biological measures. collaboration and mutual cooperation in the pursuit of scientific knowledge in the service Longitudinal studies are particularly im- of public health. Some of the areas of re- portant if we are to tease apart whether search most likely to support such a goal in- some behaviors are strongly dependent clude: upon an environmental stimulus or whether some individuals tend to repeat Genetic research, especially of special populations (e.g., Amish, Indians, the same kinds of behaviors independent of a particular environment. The state Greenlanders, Hungarians), offers op- vs. trait issues cannot be confidently portunities for further dissection of elaborated without such studies. genetic and environmental factors. Powerful statistical methods should fol- Much could be gained from pursuing low well thoughtout hypotheses rather psychological-autopsy data with the than blindly applied in the hope of un- same methodological vigor that has been covering significant associations among applied to the biological ;lostmortem large samples. When such findings do studies. In such studies we need to assign occur, they must be replicated with a as much importance to measures of "per- fresh sample.

15% OF AFFECTIVE DISORDER PATIENTS DIE BY SUICIDE

WM% OF SUICIDES HAVE AFFECTIVE DISORDER

Rpm 2.

308 2-269 Report of the Secmtary's Task Force on Youth Suicide

Few studies have examined the impr. - 8, Roy A, Ninan P, Matonson A, Picker D, van Kam- men D. Unnoila M, Paul & CSF monoamine metabolites tan= of drug states as predispositions to in chronic suicidal behavior. Psycho!Med.°14:= Irtris wh° attimPt sukIlde. 9. Brown CR., One WJ, Gayer PF, Mnichiello MD Kreusi MJP, Goodwin FK:Relationship of childhood characteristics to cerebrospinal NIG 8-hydroxyin. SUMMARY dolomitic acid kt aggressive adults. In: Shagass et al (eds) INorld Congress of Biology and Psychiatry, El- Certain principles should underlie our at- wrier Press, pp. 177.179, 1998. tenipts to make a significant and meaningful 10. Paul SM, Rehavi M, Skolnick P. Goodwin Ft High affinity binding of antidepressents to blog.nlc amine difference in dealing with the major public transport sites in human brain and platsiei Studies in depression. Poet PM, BeisAillf JC (Ws) NeuroNn of health problem of suicidal behavior, par- Mood Mordent, pp. PM% tkill*ms & Wilkins, Bal- ticularly in youth. Recommendations receiv- timore, tta34. ing the highest priority should be based on 11. Komi ER, Kiskanan JE, Goodman SJ et al: Savior*, and &hplroxylndoleaostic acid concentration sound, scientific data. Second priority should kt different brain regions of suicide victims: Comparison be assigned to activities that will promote the In chronic schizophrenic patients with suicide as cause of death.Presented at the 'rowing of the Int Soc. for development of needed data (e.g., well-con- Neurochemistry, Vancouver, Canada, My 14, 1983. trolled epidemiological studies).For ser- 12.Giffin JC, Nolan J. Kleinman J et al: Studies of the in suicide and depression. New York vices, the highest priority should go to zrdierSoTtecinosConference on Psychobiology of programs that provide direct treatment to Suicidal Behavior, Sept 1940, 1985 13. Bunney WE Jr. Fawcett JA:Possibility of a high-risk individuals-those with a psychiatric biochemical test for suicidal potential.ArM Oen or medical disorder known to be associated Psychiatry 13:232.239, 1983. with a high rate of completed suicide (as dis- 14. Curzon 0: Effects of adrenal hormones and stress tinct from suicidal ideation), those who have on brain serotonin. Am J Pin Nutt 24:830134, 19/1. 15. Major I.E. Brown GI., Wilson WP: Carcinoki and made previous suicidal attempts, and those psychiatric symptom South Med J 88.787-790, 1973. with a strong family history of suicidal be- 18. Slordama A, Lovenbero M. Engelman K, Carpenter WT, WpW RJ, Gesso GL: Serolonin Plow. Clinical Implies- havior. Obviously, preference must be given tions of inhibiting_ Itssynthesis wit,pars- to treatments for which efficacy is based on chlorophenytalinine ( A).Cmbined Ctinft) Staff Conference at the National Institutes of Health. Ann Intern actual data, rather than impressions or Med 73:907429, 19* hopes. 17. Pokorny AD: Prediction of suicide in psychiatric pettents: Report of a prospective study.Arch Gen Piwdliatry 40:249-257, 'MU.< 18. Robins IN:Deviant Children Grown Up: A REFERENCES Sociological and Pilatrio Study of Smiopaihic Per- sonality. & WM* Baldmore, 1988. 1. Asberg PA: Neurotransmitter metabolites in CSF. 19. Clayton PJ: Suicide, In: Sylfigrium In:The Report of the Secreterfe Task Force on Youth on Self-Destructive Behavior, The fetid CI Ice of Suicide (DHHS).Government Printing Office, :203414, 1985, Washington, D.C. 1987. North American, WB Saunders Co, 8 20. Claranello RB, Mders IF, Birches JD, Sergi( PA. 2. ftuft kt:Poet.mortern studies of suicide.In: Com HM: The use of 5.1tydroxylryptophan In a child with The of ffue Secretary's Task Force on Youth Suicide Lesch.Nyhan syndrome.Child Psychiatry Hum Day Government Printing Office, Washington, D.C. 7:127.133, 1978. 21. Bunney WE Jr, Ana.vekti.DS, Goodwin Fit, Davis 3. Meltzer HY, Lowry MT: The neuroendocrine sys. at Brodie Hifff, Murphy PL, Moil TN: Effects of 1.- tern and suicide. In. The Aspoit of the Secretary's Tisk DOPA on depression. Lancet 1:898, 1989. Force on Youth Suicide Government Printing Of- fice, Washington, D.C. 1087. 22. &CAM OL Wilson 1AP, Green RL: Mental aspects of PaddftsonIsm and their management. In: Parkinson's 4. Rov&ICetvS: Genetics and suicidal behavior. In: Disease: Rigidity, Aklnesia, Behavior, Selected Com- Th.R,poñof the 's Task Force on Youth Suicide munications on Topic. Val 2, Siegfried, J (ed). Wig Government Printing Office, Washington, D.C. Hans Huber, Bern. pp. .M..278, 1973. 1987. 23. Matthews WS, Rarebits 0: ruicide and tOidw & Brown CIL, Goodwin FK, Ballenger JC, Goys' PF, A review of the literature. Psych mates =415-524, Major LF:ftgresslon in humans correittes with Mid amino metabolites.Psychiatry Res 1991. 411:111:13r179. 24. Chadwick D. Jenner P. Reynolds Eit Amines, an. doonvulsents, and epilepsy. Lancet 1:473.478. 1978. & Brown G., Ebert ME, 0_oyer PF, Jimerson DC, Mein WJ. BtemeyWE.Jr,_Goodwin He Aggreesion, eulcideland 25. Cohen DJ, Shaywitz BA, Capepulo at, Young JG, serotonln: memarwto CSF amine metabolites. AM J Bowen* MB Jr:Chronic, multiple tics of Min de Is Psychlaby 139:741-748, 1982. Tourettes' disease.Arch Gen Psychiatry 35:248-200, 7. Van Praag H: CSF &HIM and suicide In non. 197& depressed schizophrenic*. lancet Z977-978, 1993.

2-270 3 6 9 DTPAPV AVllAlF FAGoodwirr &ammo end Overview of Risk Factors.

2B. Dedandet Alk Needle adokto stadellos. In: Wal- denWarntarragon (TaMme=. (Maki/ and Atlempled Suicide. Boldtandalna Farng, WITIL V. Elkin R, Lanast DK Snowy W, Madam Glin JC,4=ad Min TP, R Wive due Olsordsr. Pai J Payohadry lat5454684,

3f0 2-271 u. s.Pd19a943.5 DEPARTMENT OF HEALTH & HUMAN SERVICES

Public Health Sen ice Akohol, Drug Abuse, and Mental Health Administration Rockville MD 20857

BEST COPY AVAILABLE

DHHS Publication No. (A DM)89-1622 Alcohol. Drug Abuse. and Mental Health Administration Printed 1989