Objective: Relationships between stressors and psychiatric diagnoses were studied among 13- to 19-year-old adolescent suicide victims with alcohol abuse/dependence (N = 14), depressive disorders (N = 18), and the remainder (N = 21). Method: The study included all adolescent suicides (N = 53) during a nationwide psychological autopsy study of suicides in Finland during a 12-month period (N = 1397). The data were collected through interviews with the victims' parents and health care personnel and from official records. Results: Interpersonal separations and difficulties regarding discipline or the law were common recent stressors among the alcohol abuse victims, and interpersonal conflicts and somatic illness among those with depressive disorders. Compared with the depressive victims, an unstable earlier family history was more common among the alcohol abusers. The accumulation of stressors and weakened parental support during the previous year were also more frequently found in the alcohol abuse victims. Conclusions: The results indicate that specific psychosocial stressors may be critical for suicidal adolescents with different diagnoses. In the evaluation of suicidal adolescent substance abusers, particularly recent interpersonal separations and family support need to be carefully weighed. Disruptions in the adolescent's interpersonal relationships, excess accumulation of stress, and lacking support from the family may be warning signs of suicide potential and indications for additional social support, for more intensive treatment, or for a change in the treatment setting. J. Am. Acad. Child Adolesc. Psychiatry, 1994, 33, 4:490-497. Key Words: suicide, stress, adolescence, substance abuse.

Several studies have reported psychosocial stressors in connection with completed suicides (Dorpat and Ripley, 1960; Heikkinen et al., 1992; Rich et al., 1986). Interpersonal losses or conflicts often precede youth suicide (Apter et al., 1993; Brent et al., 1988a; Rich et al., 1990). Difficulties with discipline and the law, and problems related to school or work, have also been reported often in adolescent suicides (Brent et al., 1988a; Shaffer, 1974). In addition to acute stressors, psychosocial stress related to the adolescent's developmental history, current life circumstances, and situational factors may make him or her more vulnerable to suicidal behavior (Adam, 1986; Cohen-Sandler et al., 1982; Hirschfeld and Blumenthal, 1986).

Perceived low family support has been reported to be associated with suicidal behavior among children and young adolescents (Asarnow and Carlson, 1988). Adolescent suicides have often been under the influence of alcohol (Brent et al., 1987; Shaffer et al., 1988).

In unselected suicide populations, the vast majority of victims, adults as well as adolescents, have suffered mental problems, most often arising from affective disorders and alcohol or other substance abuse (Barraclough et al., 1974; Brent et al., 1993; Henriksson et al., 1993; Marttunen et al., 1991; Rich et al., 1986; Robins et al., 1959). Interview studies of unselected adult suicides have reported that alcohol-abusing suicides have more often experienced recent interpersonal losses or conflicts preceding suicide than have victims with depressive disorders (Duberstein et al., 1993; Murphy and Robins, 1967,, Rich et al., 1988). However, among suicides younger than 30 years, no difference between alcohol-abusing and depressive suicides in terms of recent interpersonal losses or conflicts was found (Rich et al., 1988). Poor social support in particular has been reported to characterize adult suicide victims with alcoholism (Murphy et al., 1992).

In psychological autopsy studies of adolescent suicides published thus far, little attention has been paid to interrelationships between stressors and other contributing factors such as mental disorders. The aim of this study was to determine whether there were differences between adolescent suicide completers with alcohol abuse or dependence and those with depressive disorders in terms of recent stressors, major stress and parental support during the year preceding suicide, and earlier family history.

METHOD

All suicides committed in Finland between April 1, 1987, and March 31, 1988 (N = 1,397), were comprehensively analyzed using the psychological autopsy method, i.e., by interviews of the next-of-kin and data collection from various records (Shneidman, 1981). Finland is a North European country with an ethnically homogenous population of 5 million. The subjects of the present Study were the 44 male and 9 female adolescent suicides (age range 13 to 19 years) committed in Finland over the study period. The mean age of the victims was 17.4 years (SD, 1.6 years), and the male-female ratio was 4.9:1. All victims were unmarried. The most common suicide methods were shooting and hanging. Parental socioeconomic status (Central Statistical Office of Finland, 1987) was low in most cases (manual worker in 70%, lower-level employee in 9%, upper-level employee in 8%, employer in 6%, other or unknown in 8%). The methodology has been described in detail in previous reports (Marttunen et al., 1991, 1992a).

During the period of this nationwide study, police investigation and medicolegal examinations, including toxicological analyses of all deaths suspected of suicide, were more systematic and detailed than usual. In addition, data collection included thorough interviews of the next of kin and attending health care personnel. The interview forms were planned for the project. Medical records, those of social agencies and the police, as well as any others were all included. All cases officially registered as suicides were included in the study (Lonnqvist, 1988).

A structured face-to-face interview of at least one family member was conducted in 50 cases. The interview form contained 234 items concerning the victim's everyday life and behavior, family factors, use of alcohol and drugs, previous suicidality, help seeking, and life events. Interviews of the relatives were usually conducted in the homes of the families; they took place approximately 2 months after the suicide, and the mean duration of each was 3 hours. The interviewers were mental health professionals trained in the technique used. Informed consent was obtained from the relatives before the interview.

After the interview, all available records were collected from the victim's last domicile. Medical records were available in 36 cases. Professionals were systematically interviewed in 29 cases. In 12 cases, health care professionals who had attended the deceased during the preceding year were interviewed about the victim's state of health, treatment received, psychosocial symptoms, stressors, and level of functioning. These face-to-face interviews were structured, and the form contained 113 items. In 28 cases, a semistructured interview containing 8 items was conducted either face-to-face or by telephone with the health or social care professional the victim had last consulted before death. In addition to the structured and semistructured interviews, supplementary unstructured interviews of nonprofessionals and professionals were made when needed, most often by telephone. A multidisciplinary team discussed all the cases, and case reports were written on the basis of all information available.

In addition to the life event questionnaire administered to the relatives (Heikkinen et al., 1992), all cases were thoroughly reviewed, and any other items documented during the data collection concerning life events causing psychological distress to the subjects were included as stressors. The stressors were originally classified into 18 categories (Marttunen et al., 1993). For this report, separations from parent, and from girlfriend or boyfriend, as well as other interpersonal separations, were combined as "interpersonal separations." Conflicts with parent, and with girlfriend or boyfriend, and other interpersonal conflicts, were combined as "interpersonal conflicts." In all, then, there were 14 stressor categories. Victims with more than one stressor in these combined stressor categories were counted only once for that category.

When it was considered evident that a stressor occurring during the final month had directly contributed to the suicide, it was classified as a precipitant. The assessment of stressors and precipitants was based on consensus between two investigators. When the same precipitant was reported more than once during different intervals before the suicide (last 24 hours, last week, last month), it was included only in the category closest to the suicide.

A victim was considered to have experienced major stress during the year preceding suicide if he or she had experienced one or more of the following major stressors: death of significant other, severe somatic or psychiatric illness, abortion, psychiatric hospitalization, imprisonment, parental divorce, severe parental somatic or psychiatric illness.

Parental support during the year preceding the suicide was considered weakened if the evidence indicated parental alcohol abuse, severe psychiatric or somatic illness, attempted suicide, or violence.

Psychiatric diagnoses were independently assessed by two psychiatrists in accordance with DSM-III-R criteria (American Psychiatric Association, 1987). Best-estimate diagnoses were generated by weighing and integrating all available information from different sources. Multiple diagnoses on Axes I, II, and III were allowed. After these assessments each suicide was thoroughly reanalyzed by three investigators conferring to achieve general consensus for the final diagnoses (Marttunen et al., 1991).

The "depressive disorder" group included all cases with major depression, dysthymia, and depressive disorder not otherwise specified but without alcohol abuse or dependence (3 females, 15 males). Victims with a diagnosis of alcohol abuse or dependence were assigned to the "alcohol abuse" group (4 females, 10 males). In two of these cases other additional psychoactive substance abuse was also found. Subjects with both alcohol abuse or dependence and depressive disorder (3 females, 6 males) were assigned to the "alcohol abuse" group. There was no significant difference in the sex distribution between the two alcohol-abusing groups. The abusers with depressive disorder were marginally older than those without (18.3 [+ or -] 0.7 years versus 17.2 [+ or -] 1.6 years, t = 1.43, df = 12, p = .09), but significantly older than the "depressive disorder" group 113.3 [+ or -] 0.7 years versus 16.9 [+ or -] 1.8 years, t = 2.25, df = 25, p = .03). All five alcohol abusers without depression, and 78% of the abusers with depression, had other nonaffective comorbid diagnoses. Of the victims in the "depressive disorder" group, only 39% had comorbid diagnoses not related with alcohol abuse.

All other victims were assigned to the "other" group (2 female, 19 males). The most common Axis I diagnoses in this group were adjustment disorders (no females, 10 males), and three male victim, in the "other" group received no psychiatric diagnosis.

Subjects with alcohol abuse were compared with those with depressive, disorders using [chi.sup.2] test with Yates' correction, and Fisher's exact probability test.

RESULTS

Recent Stressors

Among subjects with stressors, the mean number of stressors during the month preceding suicide ([+ or -] SD) was 3.9 ([+ or -] 1-7) among the alcohol abusers, 3.0 ([+ or -] 1.5) among the depressives, and 2.7 [+ or -] 1.3) among the other victims. Interpersonal problems and family discord were the most common stressors (Table 1). Interpersonal separations and disciplinary or legal difficulties were more common among victims with alcohol abuse than those with depressive disorders. Compared with the alcohol abusers, interpersonal conflicts and somatic illness were more common (though not statistically significantly) among those with depressive disorders. In the "depressive disorder" group, four of the five subjects with somatic illness as a stressor had a chronic somatic illness bronchial asthma in three cases, and psoriasis in one case).

[TABULAR DATA OMITTED]

Analysis for male subjects only (10 males with alcohol abuse, 15 males with depression) did not materially alter these findings, although interpersonal conflicts were significantly less common among victims with alcohol abuse than among those with depressive disorders (1/10 alcohol abusers versus 8/15 depressives; Fisher exact, p < .05).

At least one precipitant was found in two thirds of the victims in each diagnostic category, with no differences between alcohol-abusing and depressive victims. Interpersonal separation was a more frequent precipitant among alcohol abusers than among those with depressive disorders (Table 2). No differences between the groups were found in frequency of precipitants occurring within the 24 hours before suicide (in 5/14 of the alcohol abusers, in 7/18 of those with a depressive disorder, in 10/21 of the other victims). In half of the cases in each diagnostic category (7/14 alcohol abusers, 9/18 depressives, 11/21 other victims) precipitants had occurred during the previous week.

[TABULAR DATA OMITTED]

All but one of the victims with alcohol abuse and almost half of those with depressive disorders were under the influence of alcohol at the time of suicide (Table 3).

[TABULAR DATA OMITTED]

Family History and Stressors during the Previous year

During the year preceding suicide, weakened parental support was twice as common (though not statistically significantly) among the alcohol-abusing at among the depressive victims (Table 3). In most cases with major stress within the year before suicide, there were multiple stressors during that time irrespective of diagnosis. Compared with victim's with depressive disorders, those with alcohol abuse were more likely to have had an earlier family history of parental problems.

DISCUSSION

Alcohol abuse and depressive disorders have consistently been the two most common diagnostic groups in diagnostic interview studies of unselected adolescent suicides (Brent et al., 1989a; Marttunen et al., 1991; Shafii et al., 1988). This study thus focused on the relationships between stressors and victims with alcohol abuse or depressive disorders. No previous reports comparing the frequency and quality of stressors between these diagnostic groups in adolescent suicide have been published.

This study shares the general methodological problems of psychological autopsy studies, which include the possibility of incomplete and biased information (Beskow et al., 1990; Brent et al., 1988b; Marttunen et al., 1992b; Shaffer et al., 1972). Informants may be unaware of important events in the victims' lives, and data in various records may be incomplete.

In retrospective studies, important sources of error in the measurement of life events include the informants' selective memory, denial of certain events, and overreporting to justify a current illness or suicide (Paykel, 1989; Rabkin and Struening, 1976). In the study of life events, perhaps particularly among adolescents, information obtained by interviewing parents may be incomplete. Supplementing interviews with information from other sources, such as complementary interviews and various records, probably gives a more comprehensive picture of the number and quality of stressors. On the other hand, this approach requires the researcher's clinical judgment in the integration of data from different sources.

In the present study, family members were interviewed in all but three cases, and several informants and additional information from various records were available in most. All collected data were integrated, and the items concerning life events weighed. According to Paykel (1983), the fall-off in reported mean number of events increases with their distance in the Past. The focus of this study was on stressors during the final month, so it can be assumed that the interviewees' recall error did not strongly affect reporting of these events.

However, because of the potential incompleteness of information, it remains possible that the prevalences of reported stressors, of alcohol abuse, and of depressive disorders in this study are underestimates. Important for the present study is the unlikelihood that the reliability of the data would differ and thus cause bias between victims with different diagnoses. The relatively small number of cases in the diagnostic groups compared may leave some differences undetected.

The "alcohol abuse" group included all cases with alcohol abuse or dependence regardless of whether or not they had a depressive disorder. Affective disorders are common among alcohol and other substance abusers, both among adults and Adolescents (Bukstein et al., 1989; Regier et al., 1990). Deykin et al. (1992) found only few differences between chemically dependent adolescents with and without depression. Rich and his coworkers (1988) suggest that depression to some degree is nearly constantly present among substance abusers who commit suicide. Previous comparable studies of adult suicides have also considered victims with substance abuse together regardless of whether they had affective disorders or not (Duberstein et al., 1993; Murphy et al., 1979; Rich et al., 1988). In this study, there were no significant differences between alcohol abusers with and without depressive disorders in terms of age, gender, and the presence of nonaffective comorbid diagnoses. The decision to consider all alcohol abusers with or without depressive disorders together in this study seems to be justified.

In previous interview studies virtually all adolescent suicide victims had experienced recent stressors, and more often than adult suicides within a comparable time period (Brent et al., 1988a; Heikkinen et al., 1992; Rich et al., 1988). Such events are probably also more common among adolescents than older age groups in the general population. Consistent with previous reports of adolescent suicides (Poteet, 1987; Rich et al., 1990 Thompson, 1987), the most common stressors were interpersonal problems, regardless of psychiatric diagnosis.

The prevalence of interpersonal separations within 1 month before suicide among adolescent victims with alcohol abuse in this study was equal to the frequency within 6 weeks before suicide of the somewhat broader loss/conflict category among substance-abusing victims aged under 30 years in the, study by Rich and coworkers (1988).

Consistent with studies of adult suicide, recent interpersonal loss was more common among victims with alcohol abuse than among depressives (Murphy and Robins, 1967; Rich et al., 1988). However, in the study by Rich et al. (1988), this specific difference in the prevalence of separation or conflict between victims with substance abuse or with affective disorders was not found for subjects younger than age 30. In that study the younger suicides had more interpersonal losses/conflicts before death compared with suicides aged 30 years or more, irrespective of diagnosis. Furthermore, more subjects in the young group had a stressor within 1 week before death. In another study of suicides aged 15 to 29 years, separation from a partner was even more common (though not statistically significantly) among victims with no substance use disorder than among those with substance use disorder (Runeson, 1990).

In the present study, which reports on stressors among suicides with different diagnoses specifically in the adolescent age group, interpersonal separations were more common but interpersonal conflicts less common (for males) in adolescent suicides with alcohol abuse than with depressive disorders. The discrepancies between the findings of the present study and others which have also included young adult suicides are probably a consequence of differences in study populations and methodology. First, and most critically, the present study consisted solely of adolescent victims aged 13 to 19 years, whereas in the studies of Rich et al. (1988) and Runeson (1990) only a minority were adolescents (14 of 133 in Rich et al. [1988, 1990] and 9 of 58 in Runeson [1989, 1990]). The second distinction is that abuse of substances other than alcohol was rare in the present study population. Third, the definition of the stressor category "interpersonal separation" in this study was narrower than that of "separation/conflict" in the study of Rich et al. (1988), but broader than the variable "separation from Partner" in Runeson's (1990). Furthermore, in the study by Runeson (1990) substance-abusing suicides were compared with non-substance-abusing suicides, not specifically with depressive victims. Finally, the time period covered was 6 weeks in the two previous studies, but 1 month in the present one.

These results suggest that the difference in the prevalence of interpersonal loss reported between adult substance-abusing or depressive suicides (Murphy and Robins, 1967; Rich et al., 1988) holds true among adolescent suicide victims, too. Consequently, if there is a difference between the way suicidal individuals with substance abuse or affective disorders respond to interpersonal loss as hypothesized in previous studies (Murphy and Robins, 1967; Rich et al., 1988), this probably already exists in adolescence. The higher frequency of interpersonal loss preceding suicide among alcohol-abusing than depressive adolescent victims may partly reflect the alcohol-abusing adolescents' vulnerability and difficulty in coping with interpersonal separations. Alcoholic adolescents often alienate natural social support systems (Flavin et al., 1990), and the result suggest that the alcohol-abusing youngsters in this study did not have adequate social support in their use

Additional research on the relationships between specific stressors and specific mental disorders in adolescent suicides is obviously required. Studies should also address the possible differences between adolescent an young adult suicides in this respect.

The unexpected finding that the depressive victims more frequently had although statistically significantly for males only) interpersonal conflicts than those with alcohol abuse suggests that specific (interpersonal) stressors may have different relationships with suicide. The small number of cases in the diagnostic groups compared in this study limits the generalizability of this finding. Nevertheless, in studying the relationships between stressors and suicide it is helpful to aim for as detailed analysis as possible of the stressors as well as detailed enough age-grouping of suicide victims. For example, when interpersonal separations, conflicts, and other interpersonal problems in this study were combined into "any interpersonal problem," the difference in frequencies of interpersonal stressors between victims in different diagnostic groups disappeared (Table 1).

It has been reported that antisocial behavior is common among alcoholics with early onset of heavy drinking (Cloninger et al., 1981; Irwin et al., 1990). In this study, difficulties regarding discipline or the law were relatively common recent stressors among the adolescent alcohol-abusing victims. From the methodological viewpoint comes the question of whether these difficulties were independent in the sense of not being caused by alcohol abuse preceding suicide. Whatever the source of the difficulties, such events may nevertheless be pathogenic in their own right (Paykel, 1983).

In spite of the young age of the victims in this study, unemployment and financial problems were also rather common among the alcohol-abusing victims. The finding that all but one of the suicides among the alcohol abusers and almost half among the depressives occurred under the influence of alcohol underscores the significance of alcohol as a situational factor in adolescent suicide, the influence of alcohol possibly altering reactions to stress.

The occurrence of recent stressors in all but 2 of the 53 victims shows the importance of stressful life events in adolescent suicide regardless of psychiatric diagnosis. However, there were differences between the diagnostic categories in the frequencies of specific stressors in the 4 weeks preceding the suicide. The findings that long-term stressful processes and excessive accumulation of stress characterized adolescent suicide victims with alcohol abuse are consonant with the report among adults by Murphy and coworkers (1992) that multiple risk factors predict suicide among alcoholics. However, the relationships between adverse life events and the individual's later maladjustment or psychopathology are complicated. Understanding the long-term effects of life events requires the concurrent investigation of other developmental processes acting to modulate the impact of the environment (Goodyer, 1993).

As far as we know, this is the first published study of adolescent suicide reporting differences in the frequency of stressors preceding suicide between alcoholic and depressive victims. This study further extends previous findings by indicating that, in addition to having experienced recent interpersonal separations, victims with alcohol abuse were more likely than the depressive subjects to have had a disturbed family history and tended to have had weakened parental support during the year preceding the suicide. The findings also raise the methodological suggestion that in future studies of life events in suicide it is important to aim for specific grouping of interpersonal stressors and detailed enough age-grouping of the victims.

Implications

Additional research is needed to explore the similarities and dissimilarities of suicides in different mental disorders at various specific phases of the life cycle. Future studies should also focus on mechanisms that mediate between risk factors during an individual's development and those occurring close to the suicide.

The assessment of stressful life events in general should be one element in the evaluation of suicide risk among adolescents. Results of the present study indicate that specific recent psychosocial stressors may be critical for suicidal adolescents with different psychiatric disorders. The "diagnosis-specificity" of certain stressors can guide the clinician in decisions regarding the treatment of adolescents with risk for suicide. Among depressive suicidal adolescents, recent interpersonal conflict or somatic illness may increase the likelihood of suicide. In the clinical evaluation of suicidal adolescent substance abusers, particularly recent interpersonal separations and family support need to be weighed carefully.

Disruptions in the adolescent's interpersonal relationships, accumulation of stress, and lacking support from the family may be warning signs of suicide potential and indications for additional social support, for more intensive treatment, or for a change in the treatment setting.

REFERENCES

Adam KS (1986), Early family influences on suicidal behavior. Ann N Y Acad Sci 487:63-76 American Psychiatric Association (1987), Diagnostic and Statistical Manual of Mental Disorders, 3rd edition-revised (DSM-III-R). Washington, DC: American Psychiatric Association Apter A, Bleich A, King RA et al. (1993), Death without warning? A clinical postmortem study of suicide in 43 Israeli adolescent males. Arch Gen Psychiatry 50:138-142 Asarnow JA, Carlson G (1988), Suicide attempts in preadolescent child psychiatry inpatients. Suicide Life Threat Behav 18:129-136 Barraclough BM, Bunch J, Nelson B, Sainsbury P (1974), A hundred cases of suicide: clinical aspects. Br J Psychiatry 125:355-373 Beskow J, Runeson B, [Angstrom]sgard U (1990), Psychological autopsies: methods and ethics. Suicide Life Threat Behav 20:307-323 Brent DA, Perper JA, Allman CJ (1987), Alcohol, firearms and suicide among youth. JAMA 257:3369-3372 Brent DA, Perper JA, Goldstein CE et al. (1988a), Risk factors for adolescent suicide: a comparison of adolescent suicide victims with suicidal inpatients. Arch Gen Psychiatry 45:581-588 Brent DA, Perper JA, Kolko DJ, Zelenak JP (1988b), The psychological autopsy: methodological considerations for the study of adolescent suicide. J Am Acad Child Adolesc Psychiatry 27:362-366 Brent DA, Perper JA, Moritz G et al. (1993), Psychiatric risk factors for adolescent suicide: a case-control study. J Am Acad Child Adolesc Psychiatry 32:521-529 Bukstein OG, Brent DA, Kaminer Y (1989), Comorbidity of substance abuse and other psychiatric disorders in adolescents. Am J Psychiatry 146:1131-1141 Central Statistical Office of Finland (1987), Classification of Occupations. Helsinki: Central Statistical Office of Finland Cloninger CR, Bohman M, Sigvardsson S (19881), Inheritance of alcohol abuse: crossfostering analysis of alcoholic men. Arch Gen Psychiatry 38:861-868 Cohen-Sandler R, Berman AL, King RA (1982), Life stress and symptomatology: determinants of suicidal behavior in children. J Am Acad Child Psychiatry 21:178-186 Deykin EY, Buka SL, Zeena TH (1992), Depressive illness among chemically dependent adolescents. Am J Psychiatry 149:1341-1347 Dorpat TL, Ripley HS (1960), A study of suicide in the Seattle area. Compr Psychiatry 1:349-359 Duberstein PR, Conwell Y, Caine ED (1993), Interpersonal stressors, substance abuse, and suicide. J Nerv Ment Dis 181:80-85 Flavin DK, Franklin JE, Frances RJ (1990), Substance abuse and suicidal behavior. In: Suicide Over the Life Cycle, Blumenthal SJ, Kupfer DJ, eds. Washington, DC: American Psychiatric Press, pp 177-204 Goodyer IM (1993), Recent stressful life events: their long-term effects European Child and Adolescent Psychiatry 2:1-9 Heikkinen M, Aro H, Lonnqvist J (1992), The partners' views on precipitant stressors in suicide. Acta Psychiatr Scand 85:380-384 Henriksson M, Aro H, Marttunen M et al. (1993), Mental disorders and comorbidity in suicide. Am J Psychiatry 150:935-940 Hirschfeld RM, Blumenthal

SJ (1986), Personality, life events, and other psychosocial factors in adolescent depression and suicide. In: Suicide and Depression Among Adolescents and Young Adults, Klerman GL, ed. Washington, DC: American Psychiatric Press, pp 215-253 Irwin M, Schuckit M, Smith TL (1990), Clinical importance of age at onset in type 1 and type 2 primary alcoholics. Arch Gen Psychiatry 47:320-324 Lonnqvist J (1988), National suicide prevention project in Finland: a research phase of the project. Psychiatria Fennica 19:125-132 Marttunen M, Aro H, Henriksson M, Lonnqvist J (1991), Mental disorder in adolescent suicide: DSM-III-R axes I and II among 13 to 19 year olds. Arch Gen Psychiatry 48:834-839 Marttunen M, Aro H, Lonnqvist J (1992a), Adolescent suicide: endpoint of long-term difficulties. J Am Acad Child Adolesc Psychiatry 31:649-654 Marttunen M, Aro H, Lonnqvist J (1992b), Research on mental disorders in adolescent suicide. In: Suicidal Behaviour in the European Region: Research Findings, Ferrari G, Crepet P, Platt S, Bellini M, eds. Rome: John Libbey, pp 253-259 Marttunen M, Aro H, Lonnqvist J (1993), Precipitant stressors in adolescent suicide. J Am Acad Child Adolesc Psychiatry 32:1178-1183 Murphy GE, Armstrong JW, Hermele SL, Fisher JR, Clendenin WW (1979), Suicide in alcoholism. Arch Gen Psychiatry 36:65-69 Murphy GE, Robins E (1967), Social factors in suicide. JAMA 199:303-308 Murphy GE, Wetzel RD, Robins E, McEvoy L (1992), Multiple risk factors predict suicide in alcoholism. Arch Gen Psychiatry 49:459-463 Paykel ES (1983), Methodological aspects of life events research. J Psychosom Res 27:341-352 Paykel ES (1989), Stress and life events. In: Report of the Secretary's Task Force on Youth Suicide, Vol 2: Risk Factors for Youth Suicide, Davidson L, Linnoila M, eds. Washington, DC: Alcohol, Drug Abuse, and Mental Health Administration, US Government Printing Office, pp 110-130 Poteet DJ (1987), Adolescent suicide: a review of 87 cases of completed suicide in Shelby County, Tennessee. Am J Forensic Med Pathol 18:12-17 Rabkin JE, Struening EL (1976), Life events, stress, and illness. Science 194:1013-1020 Regier DA, Farmer ME, Rae DS et al. (1990), Comorbidity of mental disorders with alcohol and other drug abuse. JAMA 264:2511-2518 Rich CL, Fowler RC, Fogarty LA, Young D (1988), San Diego suicide study II: relationships between diagnoses and stressors. Arch Gen Psychiatry 45:589-592 Rich CL, Sherman M, Fowler RC (1990), San Diego suicide study: the adolescents. Adolescence 25:855-865 Rich CL, Young D, Fowler RC (1986), San Diego suicide study I: young vs old subjects. Arch Gen psychiatry 43:577-582 Robins E, Murphy GE, Wilkinson RH Jr, Gassner S, Kays J (1959), Some clinical considerations in the prevention of suicide based on a study of 134 successful suicides. Am J Public Health 49:888-899 Runeson B (1989), Mental disorders in youth suicide. DSM-III-R Axes I and II. Acta Psychiatr Scand 79:190-497 Runeson B (1990), Psychoactive substance use disorder in youth suicide. Alcohol 25:561-568 Shaffer D (1974), Suicide in childhood and adolescence. J Child Psycbol Psychiatry 15:275-291 Shaffer D, Garland A, Gould M, Fisher P, Trautman P (1988), Preventing teenage suicide: a critical review. J Am Acad Child Adolesc Psychiatry 27:675-687 Shaffer JW, Perlin S, Schmidt CW, Himelfarb M (1972), Assessment in absentia: new directions in the Psychological autopsy. Johns Hopkins Medical Journal 130:308-316 Shafii, Steltz-Lenarsky J, Derrik AM, Beckner C, Whittinghill JR (1988), Comorbidity of mental disorders in the post-mortem diagnosis of completed suicide in children and adolescents. J Affect Disord 15:277-233 Shneidman ES (1981), The psychological autopsy. Suicide Life Threat Behav 11:325-340 Thompson TR (1987), Childhood and adolescent suicide in Manitoba: a demographic study. Can J Psychiatry 32:264-269

COPYRIGHT 1994 Lippincott/Williams & Wilkins