- Email: [email protected]
Suicide in Adolescents With Disruptive Disorders
Suicide in Adolescents With Disruptive Disorders JOHANNE RENAUD, M.D., DAVID A. BRENT, M.D., BORIS BIRMAHER, M.D., LAUREL CHIAPPETTA, B.S., AND JEFF BRIDGE, B.S.
ABSTRACT Objective: To determine the psychiatric risk factors for suicide in adolescents with disruptive disorders. Method: Fifty-nine adolescent suicide completers and 18 community controls, both having a probable or definite current DSM-///diagnosis of disruptive disorders, were compared. Results: Adolescents with disruptive disorders who committed suicide had higher rates of current substance abuse, past suicide attempt, family history of substance abuse, and family history of mood disorder than disruptive community controls. Conclusions: Disruptive adolescents appear to be at risk for completed suicide when comorbid substance abuse and past history of suicide attempt are present. The risk increases if the adolescents have a past history of physical abuse and if they have parents with substance abuse and mood disorders. Clinicians should be aware of these risk factors and implement active interventions to prevent suicide. Treatment should focus on treating not only the adolescents, but also their family members.The findings of this study also highlight the need for future research in the prevention of suicide in adolescents with disruptive disorders and comorbid substance abuse. J. Am. Acad. Child Adolesc. Psychiatry, 1999, 38(7):846-851. Key Words: suicide, adolescents, disruptive disorders, conduct disorders, substance abuse.
Accepted February 5, 1999. From the Western Psychiatric Institute and Clinic, University of Pittsburgh Medical Center, Pittsburgh. This work wlzs supported by grant MH-55123from the ChildandAdolescent Developmental Psychopathology Research Centerfor Early Onset Affective and Anxiety Disorders (DavidA. Brent, M.D., principal investigator). Correspondenceto Dr. Brent, 3811 O’Hara Street, WesternPsychiatricInstitute and Clinic, University of Pittsburgh Medical Center, Pittsburgh, PA 15213. 0890-8567/99/3807-084601999 by the American Academy of Child and Adolescent Psychiatry.
managed through the juvenile justice, child welfare, or educational systems, where access to child mental health care is limited. From the large psychological autopsy study by ShafTer et al. (1996), one may infer that comorbidity with mood and substance abuse disorders increases the risk for suicide in adolescents with disruptive disorders, since so many of the older male adolescents showed this triple pattern of comorbidity. However, no comparison with living controls with disruptive disorders was reported. Within the disruptive disorders spectrum, it appears that C D is more consistently associated with suicide than is attention deficit disorder (ADD) (Brent et al., 1988, 1993a; Shaffer et al., 1996; Shafii et al., 1985). From developmental analyses, CD is most likely to be a risk factor for suicide in older male adolescents (Shaffer et al., 1996). Not surprisingly, we found that abuse and 1egaUdisciplinary problems were particularly closely associated with suicide and CD in adolescents (Brent et al., 1993b), although, as was the case in a similar study of Gould et al. (1996), no 3-way interactions between CD, stressor, and suicide were significant. For this purpose, we examine a subgroup of adolescent suicide completers and community controls with disruptive disorders (Brent et al., 1988, 1993a). We hypothesize that the suicide completers with disruptive
846
J . A M . A C A D . C H I L D A D O L E S C . P S Y C H I A T R Y , 3 8 : 7 , J U L Y 1999
Disruptive disorders have been reported to be a risk factor for suicide in at least 3 separate psychological autopsy studies (Brent et al., 1988, 1993a; Shaffer et al., 1996; Shafii et al., 1985). Moreover, antisocial behavior has been shown to be more characteristic of younger (i.e., age <30 years) than older suicides (Rich et al., 1986). Shaffer and colleagues (1996) found conduct disorder (CD) to be associated with suicide, particularly in older adolescent males, especially when comorbid with substance abuse and mood disorder. However, relatively little work has focused on which factors differentiate youths with disruptive disorders who commit suicide from those who do not. The results of such a comparison could help to identify and promptly intervene in those youths with disruptive disorders who are at higher risk for suicide. This is particularly important because many youths with disruptive disorders are primarily
SUICIDE AND DISRUPTIVE DISORDERS
disorders, when compared with a community sample with disruptive disorders, will show higher rates of CD and assaultive behavior, comorbid current major depression and current substance abuse, past suicide attempt, and lifetime sexual/physical abuse, and greater availability of guns in the home. In addition, we predict that the families of the completers with disruptive disorders will have higher rates of depression, substance abuse, and suicide attempt, when compared with the families of the Community controls. METHOD Sample A consecutive series of 140 adolescent suicide victims was reviewed, and all those with a probable or definite current DSM-ZII (American Psychiatric Association, 1980) diagnosis of disruptive disorders (ADD and/or CD) were selected (n = 59). The overall sample of 140 represents 72% of all available suicides during this time period. The 59 suicide victims with disruptive disorder were drawn from the following samples: 12 of 27 suicide victims with DSM-ZZZADD and/or C D first reported by Brent et al. (1988), 23 of 67 suicide victims first reported by Brent et al. (1993a), 22 of 37 suicide victims first reported by Brent et al. (1996), and 2 suicide victims not previously reported. There were no significant differences in the completers from these groups with respect to age, gender, race, family constellation, county of residence, and socioeconomic status. A sample of 131 community controls, previously described by Brent and colleagues (1993c), was obtained by geographic cluster sampling. We targeted communities with median income, population density, racial composition, and age distribution similar to those of the 140 suicide victims, but where an adolescent suicide had not occurred within 2 years. These controls were similar to the suicides on age, race, gender, socioeconomic status, and county of residence. From these 131 controls, we selected those who had a probable or definite current DSM-ZII diagnosis of disruptive disorders (ADD andlor CD), yielding 18 controls in total. All analyses described below refer to these subsamples, all with disruptive disorders.
Assessment Axis I psychiatric disorders were assessed by use of the Schedule for Mective Disorders and Schizophrenia for School-Age Children, Epidemiologic and Present Episode versions (Chambers et al., 1985; Orvaschel et al., 1982). For suicide completers, multiple informants were interviewed by using psychological autopsy format as previously described (Brent et al., 1988, 1993a). Definite disorders met DSMZZZ criteria, whereas probable disorders were one criterion short. In both orobable and definite disorders. evidence of functional imoairment was required. DSM-ZZZdiagnoses in the first- and second-degree relatives of subjects were obtained using the Family History Research Diagnostic Criteria (Andreasen et al., 1977), with diagnostic criteria modified from Research Diagnostic Criteria to DSM-ZZZ. An inventory of 29 life events was developed that covers the stressors known to be associated with suicide attempts and completion in adolescence (Brent et al.. 1988, 199313).This scale contained items such as interoersonal conflict, interpersonal loss, legalldisciplinary problems, sexuk abuse,
J. A M . ACAD. C H I L D A D O L E S C . PSYCHIATRY, 3 8 : 7 , JULY 1999
and physical abuse. The number, type, and accessibilityof all guns in the household at the time of the suicidal episode were ascertained using the methods described by Brent and colleagues (1988).
Data Analysis The completers and controls were compared using standard parametric and, where appropriate, nonparametric statistics (i.e., MannWhitney U). Rates of disorder in the relatives of completers and controls were compared by Pearson’s or, when the expected value was less than 5, Fisher exact test (FET). The most parsimonious set of variables that were associated with suicide completion was selected by logistic regression.
x2
RESULTS Demographic Characteristics
The suicide completers were comparable with the control group with disruptive disorder with regard to gender (92% versus 78% male), race (92% versus 100% white), age (17.4 [ 1.91 versus 17.1 [ 1.21 years), socioeconomic status (3.1 [1.2] versus 2.9 [0.9]) (Hollingshead, 1975),and family constellation (living with both parents 25.4% versus 38.9%). Characteristics of the Suicide
Among the completers with disruptive disorders, about one third were toxicologically positive for substances (34.5%), with 26.8% positive for alcohol and 13.2% positive for other substances. The most common methods of completed suicide were firearms (62.7%), hanging (27.1%), and jumping (6.8%). Overdose and asphyxiation represented 1.7% each. Characteristics of the Disruptive Disorders
CD was more common in the suicide than the control group (59.3% versus 33.3%; x2 = 3.74 ;p = .05; odds ratio [OR] = 2.9; 95% confidence interval [CI] = 1.0, 8.8). While there were no differences between the 2 groups with respect to the type of antisocial behavior, completers showed a higher number of CD symptoms compared with the controls (median interquartile range = 3.0 [4.0] versus 1.0 [2.3]; Mann-Whitney U = 541.0; D = .02). There were no differences between the grouDs with respect to the prevalence of ADD. V
‘Omorbid
I
Psychiatric Disorders
The 59 completers with disruptive disorders had higher rates of current substance abuse than did community controls with disruptive disorders (46.6% versus
847
R E N A U D ET AL
x2
5.6%; = 9.92; p = .002; OR = 14.8; 95% CI = 1.9, 118.7), but not lifetime substance abuse (50.0% versus 44.4%; p = not significant [NS]) (Table 1 ) . More completers with disruptive disorders had a specific diagnosis of current alcohol abuse (43.1% versus 5.6%; = 8.61; p = .003; OR = 12.9; 95% CI = 1.6, 103.4) and current drug abuse (24.6% versus 0.0%; FET,p = .02) than controls with disruptive disorders. Counter to hypotheses, there were no differences between the 2 groups in the rates of comorbid major depressive disorder.
x2
death) (18.4% versus 0.0%; F E T p = NS), although the difference was in the predicted direction. Life Events
There were no differences between the groups with respect to frequencies of interpersonal conflict with parent, with boyfriend/girlfriend, or interpersonal loss of boyfriend/girlfriend (allp values = NS), within the year of suicide. There was a trend among the completers with disruptive disorders to have had legal difficulties in the previous year compared with the controls with disruptive disorders (63.0% versus 38.9%; = 3.07;p = .08). Within the subsample for whom assessment of lifetime abuse was available (suicides subsample, n = 29; controls subsample, n = 18), there was a higher rate of lifetime abuse among completers (58.6% versus 5.6%; = 13.24;p = .0003). This was primarily attributable to a higher rate of physical abuse in the completers (55.2% versus 0.0%; FET, p = .0001), with no difference in the rate of sexual abuse (8.0% versus 5 . 6 % ; = ~ NS).
x2
Past Suicidal BehaviodSuicidal and Homicidal Ideation Within the Previous Week
Completers with disruptive disorders had a much greater prevalence of past suicide attempts than controls with disruptive disorders (52.7% versus 5.6%; = 12.47;~ = .0004; OR = 19.0; 95% CI = 2.4, 152.6). In addition, completers were more likely to have shown evidence of suicidal ideation (defined as any ideation, gesture, or threat) than were controls during the week before interview (52.5% versus 16.7%; = 7.20; p = .007;OR= 5.5; 95%CI= 1.4,21.2).Therewerenosignificant differences in the manifestation of homicidal ideation (defined as a homicidal plan in the week before
x2
x2
x2
Availability of Firearms
Although all the differences were in the expected direction, completers did not differ from controls with
TABLE 1 Demographic and Clinical Risk Factors for Suicide in Disruptive Youths (%) Completer (n = 59)
Control ( n = 18)
59.3 3.0 [4.0]
33.3 1.0 [2.3]
.05 .02
2.9
Current substance abuse Lifetime substance abuse Current alcohol abuse Current drug abuse MDD
46.6 50.0 43.1 24.6 30.5
5.6 44.4 5.6 0.0 16.7
.002 NS .003 .02 NS
14.8 1.3 12.9 12.3 2.2
(1.8, 118.7) (0.4, 3.6) (1.6, 103.4) (0.7, 217.8) (0.6, 8.5)
Attempt Past suicidality Homicidal ideation
52.7 52.5 18.4
5.6 16.7 0.0
.0003 .007 NS
19.0 5.5 8.7
(2.4, 152.6) (1.4,21.2) (0.5, 157.2)
Lifetime abuse Lifetime physical abuse
58.6 (n = 29) 55.2 (n = 29)
5.6 0.0
.0003 .0001
24.1 45.2
(2.8, 206.3) (2.5, 821.6)
First-degree relatives Any affective disorder Substance abuse
27.6 (n = 214) 23.9 (n = 218)
14.0 (n = 50) 4.1 (n = 4 9 )
,046 .002
2.3 7.4
(1.00, 5.5) (1.7,31.4)
Second-degree relatives Any affective disorder Substance abuse
12.4 (n = 498) 16.8 (TI = 506)
4.0 (n = 126) 4.6 (n = 131)
,006 ,0004
3.4 4.2
(1.4, 8.7) (1.8,9.9)
Conduct disorder only No. of conduct symptoms: median [IQR]
Note: O R = odds ratio; CI = confidence interval; IQR = interquartile range; M D D significant. In cases in which a zero cell is present, 0.5 is added to all cells.
848
pValue
=
OR
95% CI" (1.0,S.S)
major depressive disorder; NS
=
not
J . A M . ACAD. C H I L D A D O L E S C . PSYCHIATRY, 3 8 : 7 , JULY 1999
SUICIDE AND DISRUPTIVE DISORDERS
respect to any gun availability variables: any guns (66.6% versus 50.0%;p = NS), long guns (56.0% versus 43.7%; p = NS), hand guns (33.9% versus 11.8%; p = NS), or loaded guns (11.1% versus 5.6%; p = NS). Family History of Psychiatric Disorder
The rates of any mood disorder (27.6% versus 14.0%;
x2 = 3.98; p < .05) and substance abuse (23.9% versus 4.1%; x2 = 9 . 6 9 ; =~ .002) among first-degree relatives of the completers were higher than those of controls. Much of the difference in the familial rates of substance abuse between suicides and controls was due to the higher rates of alcohol abuse in fathers of the completers (34.6% versus 0.0%; FET,p = .01). The differences with respect to the rates of suicide attempts did not reach significance (9.9% versus 3 . 6 % ; ~ = NS). Similarly, the rates of mood disorder (12.4% versus 4.0%; x2 = 7.60; p = .006) and substance abuse (16.8% versus 4.6%; x2 = 12.70; p = .0004) were also higher among the second-degree relatives of the suicide victims. Logistic Regression
After control for CD, the most parsimonious model consisted of the following risk factors: current substance abuse (OR = 16.0; 95% CI = 7.8, 32.9), past suicide attempt (OR = 14.0; 95% CI = 7.2, 27.4), family history of substance abuse (OR = 3.9; 95% CI = 1.7, 9.2), and family history of mood disorder (OR = 2.5; 95% CI = 1.2, 5.1). This model fit the data well (HosmerLemeshowX2=6.14;df=7;p= .52). DISCUSSION
To our knowledge, this is the first study to have examined risk factors for completed suicide in those with disruptive disorder. We found that several risk factors were associated with suicide above and beyond disruptive disorder, namely current substance abuse, past suicide attempt, history of physical abuse, and family history of substance abuse and mood disorders. However, we did not find differences compared with the control group with respect to rates of either ADD or comorbid mood disorder. Before discussing the above-noted findings, it is important to take into account that this study included a relatively small sample of suicide completers and controls. Nevertheless, both groups are representative of the population, and therefore the estimates of the risk factors are
J . AM. ACAD. C H I L D A D O L E S C . PSYCHIATRY, 3 8 : 7 , JULY 1999
not subject to the usual biases involved in studies of clinical populations. Finally, because of the nature of a psycho-
logical autopsy study, in-depth assessment of personality traits and family interaction variables that are likely to contribute to disruptive behavior were not available. Because the sample was almost entirely white and male, caution should be made in generalizing to females and to other ethnic groups. However, Shaffer and colleagues' (1996) more ethnically diverse report shows similar results and supports the view that C D is a risk factor for suicide in females as well. Cairns et al. (1988) found high rates of suicide attempt in delinquent females, also consistent with this view. However, in an analysis of all 140 suicides and 131 controls, we found that C D was a risk factor only for males (Brent et al., unpublished). The interaction of sex and aggression and its impact on risk for suicide is an important area of further inquiry. We found a high prevalence of current comorbid substance abuse in the suicide completers but no differences in lifetime rates of substance abuse. This suggests that it is not vulnerability to substance abuse per se that puts one at risk for suicide, but rather the acute and subacute effects of substance abuse that heighten the disruptive youth's risk for suicide. These mechanisms may include decreasing brain serotonin (Ballenger et al., 1979), impairing judgment and increasing impulsivity, and increasing the likelihood of using a gun (Brent et al., 1987, 1993~). In the literature, past suicide attempt was found as a strong predictor of future suicidal behavior across diagnostic groups, being reported consistently in several psychological autopsy studies (Brent et al., 1988, 1993a; Shaffer et al., 1996; Shafii et al., 1985). In this study, past suicide attempt was a major risk factor, and, while not as strong, suicidal ideation within a week of the completion was also a predictor. Past suicide attempt appears to be a strong risk factor for completed suicide, independent of diagnosis (Brent et al., 1993a; Shaffer et al., 1996). Although only data on a subsample were available in our study, physical abuse was strongly associated with suicide. This is consistent with other studies showing that physical and sexual abuse have been reported to be associated with suicide and suicidal behavior in several studies (Brent et al., 1993b, 1994; Fergusson et al., 1996; Gould et al., 1996; Hibbard et al., 1988; Kaplan et al., 1997; Romans et al., 1995). Abuse may set in motion a complex cascade of psychological and physiological alterations that increase the 849
R E N A U D E T AL.
regard to ADD alone (Brent et al., 1993a; Shaffer et al., 1996).Therefore, ADD per se does not appear to predispose to risk for suicide, although ADD may predispose to other difficulties such as C D and substance abuse, which in turn mediate poor outcome in youths with ADD (Brent et al., 1993a; Mannuzza et al., 1993).
risk of dysphoria and self-destructive behavior (De Bellis et al., 1994; Kaplan et al., 1997; Kaufman, 1991; Kaufman et al., 1998). We also found a trend toward more frequent 1egaVdisciplinary problems in the cornpleters. Therefore, youths in detention or in the midst of a disciplinary crisis may be at increased risk for suicide, due to both the disruptive disorder that precipitated the crisis and the stressor itself (Shaffer, 1974). In addition, family history of substance abuse and of mood disorders were also risk factors for suicide. As previously demonstrated in the literature using unselected samples of adolescent completers, family history conveys risk even with control for proband psychopathology (Brent et al., 1994; Gould et al., 1996). The relationship between family history of psychopathology and adolescent suicide may be mediated by both genetic and environmental mechanisms. The likelihood that environmental mechanisms play a role in this sample is increased by the persistence of family history variables in logistic regression even with control for suicide attempt and substance abuse in the individual. In our study, the specific association of suicide with paternal alcoholism may indicate that there was a male-to-male transmission of type I1 alcoholism (Cloninger, 1987). In turn, early-onset, malelimited alcoholism has been associated with impulsive violence and suicidal behavior (Buydens-Branchey et al., 1989). An alcoholic parent, in addition to increasing the genetic liability for alcoholism in hidher child, also increases the likelihood of family violence, discord, and instability (Chaffin et al., 1996). It is also interesting to note that in our samples of completers and controls, in the presence of disruptive disorders, comorbid major depressive disorder was not associated with suicide. Shaffer and colleagues (1996) found that a combination of dysthymic, conduct, and substance abuse disorders occurred as a "triple threat" in older male suicides. Alternatively, we previously reported that C D was at least as great a risk factor for suicide in the absence of mood disorder (Brent et al., 1993a). The results presented herein, from a sample that overlaps with the above-cited study, not surprisingly supports the latter point of view. Therefore, the combination of disruptive disorder and substance abuse should be regarded critically with respect to risk for suicide, even in the absence of mood disorder. Finally, similarly to other studies, we found that the completers had a higher rate of CD, whereas there were no differences between completers and controls with
American Psychiatric Association (1980), Diagnosticand Statistical Manual of Mental Disorders, 3rd edition (DSM-III). Washington, DC: American Psychiatric Association Andreasen N, Endicott J, Spitzer R, Winocur G (1977), The family history method using Research Diagnostic Criteria: reliability and validity. Arch Gen Pychiatry 34: 1229-1235 Ballenger JC, Goodwin FK, Major LF, Brown GL (1979), Alcohol and central serotonin metabolism in man. Arch Gen Pychiatry 36:224-247 Brent DA, Bridge J, Johnson BA, Conolly J (1996), Suicidal behavior runs in families: a controlled family study of adolescent suicide victims. Arch Gen Pychiatry 53: 114 5-1 152 Brent DA, Perper JA, Allman C (1987),Alcohol, firearms, and suicide among youth: temporal trends in Allegheny County, Pennsylvania, 1960-1983. JAMA 257:3369-3372 Brent DA, Perper JA, Goldstein CE et al. (1988), Risk factors for adolescent suicide. Arch Gen Pychiatry 45:581-588 Brent DA, Perper JA, Moritz G et al. (1993a), Psychiatric risk factors for adolescent suicide: a case-control study. J Am Acad Child Adolesc Pychiatry 32:521-529 Brent DA, Perper JA, Moritz G et al. (1993b), Stressful life events,, psychopathology, and adolescent suicide: a case control study. Suicide Lifi Threat Behav 23:179-187 Brent DA, Perper JA, Moritz G et al. (1993c), Psychiatric sequelae to the loss of an adolescent to suicide.J A m Acad ChildAdolescPsychiatry 32509-5 17 Brent DA, Perper JA, Moritz G et al. (1994), Familial risk factors for adolescent suicide: a case-control study. Acta Pychiatr Scand 89:52-58 Buydens-Branchey L, Branchey MH, Noumair D (1989), Age of alcoholism onset. Arch Gen Pychiatry 46:225-230 Cairns RB, Peterson G, Neckerman HJ (1988), Suicidal behavior in aggressive adolescents. J Clin Child Pychol17:298-309 Chaffin M, Kelleher K, Hollenberg J (1996), Onset of physical abuse and neglect: psychiatric, substance abuse, and social risk factors from prospective community data. ChildAbuse Negl20: 191-203 Chambers WJ, Puig-Antich J, Hirsch M et al. (1985), The assessment of affective disorders in children and adolescents by semistructured interview: test-retest reliability of the Schedule for Mective Disorders and
850
J . AM. ACAD. C H I L D A D O L E S C . PSYCHIATRY, 38:7, JULY 1999
Clinical Implications
In conclusion, adolescents with the combination of disruptive disorder and current substance abuse are at high risk of committing suicide. Proper assessment and management of substance abuse in disruptive youths is likely to be an important means of reducing the substantial suicide risk in this subgroup. The risk increases for adolescents who have had a previous suicide attempt, who have a history of physical abuse, and who have a family history of substance abuse and mood disorders. Clinicians should be aware of these risk factors and implement active interventions to prevent suicide. Treatment should focus not only on the adolescent, but also on family members. REFERENCES
SUICIDE A N D DISRUPTIVE DISORDERS
Schizophrenia for School-Age Children, Present Episode version. Arch Gen Psychiatry 42:696-702 Cloninger CR (1987), Neurogenic adaptive mechanisms in alcoholism. Science 236:4 10-4 16 De Bellis MD, Chrousos GI: Dorn LH et al. (1994), Hypothalamic-pituitaryadrenal axis dysregulation in sexually abused girls. / Clin Endocrinol Metab 78:249-255 Fergusson DM, Honvood LJ, Lynskey MT (1996), Childhood sexual abuse and psychiatric disorder in young adulthood, 11: psychiatric outcomes of childhood sexual abuse./Am Acad ChildAdolerc Psychiatiy 35: 1365-1374 Gould MS, Fisher P, Parides M, Flory M, Shaffer D (1996), Psychosocial risk factors of child and adolescent completed suicide. Arch Gen Psychiatry 53:1155-1162 Hibbard RA, Brack CJ, Rauch S, Orr DP (1988), Abuse, feelings, and health behaviors in a student population. Am/ Dis Child 142:326-330 Hollingshead AB (1975), Four-Factor Index of SocialStatur. New Haven, CT: Yale University Department of Sociology Kaplan S, Pelcovitz D, Salzinger S, Mandel F, Weiner M (1997). Adolescent physical abuse and suicide attempts. J A m Acad Child Adolesc Psychiatiy 36:799-808 Kaufman J (1991), Depressive disorders in maltreated children. / A m Acad ChildAdolesc Psychiatiy 30:257-265
Kaufman J, Birmaher B, Perel J et al. (1998), Serotonergic functioning in depressed abused children: clinical and familial correlates. Biol Psychiatry 44973-981 Mannuzza S, Klein RG, Bessler A, Malloy P, LaPadula M (1993), Adult outcome of hyperactive boys: educational achievement, occupational rank, and psychiatric status. Arch Gen Psychiatiy 50:565-576 Orvaschel H, Puig-Anrich J, Chambers W, Tabrizi MA, Johnson R (1982), Retrospective assessment of prepubertal major depression with the Kiddie-SADS-E. J A m Acad Child Psychiatry 21:392-397 Rich CL, Young D, Fowler RC (1986), San Diego suicide study, I: young vs old subjects. Arch Gen Psychiatiy 43:577-582 Romans SE, Martin JL, Anderson JC, Herbison GP, Mullen PE (1995), Sexual abuse in childhood and deliberate self-harm. Am J Prychiatry 152:1336-1342 Shaffer D (1974), Suicide in childhood and early adolescence./ ChildPsychol Psychiatry 15:275-291 Shaffer D, Gould MS, Fisher P et al. (1996), Psychiatric diagnosis in child and adolescent suicide. Arch Gen Pychiatiy 53:339-348 Shafii M, Carrigan S, Whittinghill JR, Derrick A (1985), Psychological autopsy of completed suicide in children and adolescents. Am/ Psychiany 142:1061-1064
Accidental and Suicidal Adolescent Poisoning Deaths in the United States, 1979-1994. Greene Shepherd, PharmD, Wendy Klein-Schwartz, PharmD, MPH
Objective:To describe the epidemiological features of poisoning deaths in adolescents in the United States. Design: Descriptive analysis of poisoning deaths in persons aged 10 to 19 years in the United States from January 1, 1979, to December 31, 1994, based on national mortality data. Stud’ Population: Adolescents whose cause of death was identified as poisoning using Znternational Chs2fication of Diseases, Ninth Revision codes. Main Outcome Measure: Nature of injury (accident vs suicide). Results: There were 4129 suicides and 3807 accidental deaths due to poisoning. Victims were most frequently male and white. However, poisoning was more often the method of suicide in adolescent girls than in boys (28.0% vs 8.7%). The number of deaths (7138 vs 798) and death rate (2.36 vs 0.28 per 100 000 population) were higher in 15- to 19-year-olds vs 10- to 14-year-olds.The distribution of substances involved was different for 10- to 14-year-olds compared with 15- to 19-year-olds and for suicides compared with accidents. Among 10- to 14-year-olds, drugs other than alcohol accounted for 232 (85.3%) of 272 suicides but only 118 (22.4%) of 526 accidental deaths. Gases and vapors played an important role in accidental deaths and suicides in 15- to 19-year-olds and in accidents in 10- to 14-year-olds. Conclwions:The rates of suicides and accidental poisoning deaths were lower in 10- to 14-year-olds compared with 15- to 19-year-olds. Areas where injury-prevention efforts might have an influence on adolescent fatalities include management of depression, substance abuse education, and use of carbon monoxide detectors or shutoff switches. Arch Pediatr Adolesc Med 1998; 152:1181-1 185.
J . AM. ACAD. C H I L D A D O L E S C . PSYCHIATRY, 38:7, JULY 1999
85 1
ABSTRACT Objective: To determine the psychiatric risk factors for suicide in adolescents with disruptive disorders. Method: Fifty-nine adolescent suicide completers and 18 community controls, both having a probable or definite current DSM-///diagnosis of disruptive disorders, were compared. Results: Adolescents with disruptive disorders who committed suicide had higher rates of current substance abuse, past suicide attempt, family history of substance abuse, and family history of mood disorder than disruptive community controls. Conclusions: Disruptive adolescents appear to be at risk for completed suicide when comorbid substance abuse and past history of suicide attempt are present. The risk increases if the adolescents have a past history of physical abuse and if they have parents with substance abuse and mood disorders. Clinicians should be aware of these risk factors and implement active interventions to prevent suicide. Treatment should focus on treating not only the adolescents, but also their family members.The findings of this study also highlight the need for future research in the prevention of suicide in adolescents with disruptive disorders and comorbid substance abuse. J. Am. Acad. Child Adolesc. Psychiatry, 1999, 38(7):846-851. Key Words: suicide, adolescents, disruptive disorders, conduct disorders, substance abuse.
Accepted February 5, 1999. From the Western Psychiatric Institute and Clinic, University of Pittsburgh Medical Center, Pittsburgh. This work wlzs supported by grant MH-55123from the ChildandAdolescent Developmental Psychopathology Research Centerfor Early Onset Affective and Anxiety Disorders (DavidA. Brent, M.D., principal investigator). Correspondenceto Dr. Brent, 3811 O’Hara Street, WesternPsychiatricInstitute and Clinic, University of Pittsburgh Medical Center, Pittsburgh, PA 15213. 0890-8567/99/3807-084601999 by the American Academy of Child and Adolescent Psychiatry.
managed through the juvenile justice, child welfare, or educational systems, where access to child mental health care is limited. From the large psychological autopsy study by ShafTer et al. (1996), one may infer that comorbidity with mood and substance abuse disorders increases the risk for suicide in adolescents with disruptive disorders, since so many of the older male adolescents showed this triple pattern of comorbidity. However, no comparison with living controls with disruptive disorders was reported. Within the disruptive disorders spectrum, it appears that C D is more consistently associated with suicide than is attention deficit disorder (ADD) (Brent et al., 1988, 1993a; Shaffer et al., 1996; Shafii et al., 1985). From developmental analyses, CD is most likely to be a risk factor for suicide in older male adolescents (Shaffer et al., 1996). Not surprisingly, we found that abuse and 1egaUdisciplinary problems were particularly closely associated with suicide and CD in adolescents (Brent et al., 1993b), although, as was the case in a similar study of Gould et al. (1996), no 3-way interactions between CD, stressor, and suicide were significant. For this purpose, we examine a subgroup of adolescent suicide completers and community controls with disruptive disorders (Brent et al., 1988, 1993a). We hypothesize that the suicide completers with disruptive
846
J . A M . A C A D . C H I L D A D O L E S C . P S Y C H I A T R Y , 3 8 : 7 , J U L Y 1999
Disruptive disorders have been reported to be a risk factor for suicide in at least 3 separate psychological autopsy studies (Brent et al., 1988, 1993a; Shaffer et al., 1996; Shafii et al., 1985). Moreover, antisocial behavior has been shown to be more characteristic of younger (i.e., age <30 years) than older suicides (Rich et al., 1986). Shaffer and colleagues (1996) found conduct disorder (CD) to be associated with suicide, particularly in older adolescent males, especially when comorbid with substance abuse and mood disorder. However, relatively little work has focused on which factors differentiate youths with disruptive disorders who commit suicide from those who do not. The results of such a comparison could help to identify and promptly intervene in those youths with disruptive disorders who are at higher risk for suicide. This is particularly important because many youths with disruptive disorders are primarily
SUICIDE AND DISRUPTIVE DISORDERS
disorders, when compared with a community sample with disruptive disorders, will show higher rates of CD and assaultive behavior, comorbid current major depression and current substance abuse, past suicide attempt, and lifetime sexual/physical abuse, and greater availability of guns in the home. In addition, we predict that the families of the completers with disruptive disorders will have higher rates of depression, substance abuse, and suicide attempt, when compared with the families of the Community controls. METHOD Sample A consecutive series of 140 adolescent suicide victims was reviewed, and all those with a probable or definite current DSM-ZII (American Psychiatric Association, 1980) diagnosis of disruptive disorders (ADD and/or CD) were selected (n = 59). The overall sample of 140 represents 72% of all available suicides during this time period. The 59 suicide victims with disruptive disorder were drawn from the following samples: 12 of 27 suicide victims with DSM-ZZZADD and/or C D first reported by Brent et al. (1988), 23 of 67 suicide victims first reported by Brent et al. (1993a), 22 of 37 suicide victims first reported by Brent et al. (1996), and 2 suicide victims not previously reported. There were no significant differences in the completers from these groups with respect to age, gender, race, family constellation, county of residence, and socioeconomic status. A sample of 131 community controls, previously described by Brent and colleagues (1993c), was obtained by geographic cluster sampling. We targeted communities with median income, population density, racial composition, and age distribution similar to those of the 140 suicide victims, but where an adolescent suicide had not occurred within 2 years. These controls were similar to the suicides on age, race, gender, socioeconomic status, and county of residence. From these 131 controls, we selected those who had a probable or definite current DSM-ZII diagnosis of disruptive disorders (ADD andlor CD), yielding 18 controls in total. All analyses described below refer to these subsamples, all with disruptive disorders.
Assessment Axis I psychiatric disorders were assessed by use of the Schedule for Mective Disorders and Schizophrenia for School-Age Children, Epidemiologic and Present Episode versions (Chambers et al., 1985; Orvaschel et al., 1982). For suicide completers, multiple informants were interviewed by using psychological autopsy format as previously described (Brent et al., 1988, 1993a). Definite disorders met DSMZZZ criteria, whereas probable disorders were one criterion short. In both orobable and definite disorders. evidence of functional imoairment was required. DSM-ZZZdiagnoses in the first- and second-degree relatives of subjects were obtained using the Family History Research Diagnostic Criteria (Andreasen et al., 1977), with diagnostic criteria modified from Research Diagnostic Criteria to DSM-ZZZ. An inventory of 29 life events was developed that covers the stressors known to be associated with suicide attempts and completion in adolescence (Brent et al.. 1988, 199313).This scale contained items such as interoersonal conflict, interpersonal loss, legalldisciplinary problems, sexuk abuse,
J. A M . ACAD. C H I L D A D O L E S C . PSYCHIATRY, 3 8 : 7 , JULY 1999
and physical abuse. The number, type, and accessibilityof all guns in the household at the time of the suicidal episode were ascertained using the methods described by Brent and colleagues (1988).
Data Analysis The completers and controls were compared using standard parametric and, where appropriate, nonparametric statistics (i.e., MannWhitney U). Rates of disorder in the relatives of completers and controls were compared by Pearson’s or, when the expected value was less than 5, Fisher exact test (FET). The most parsimonious set of variables that were associated with suicide completion was selected by logistic regression.
x2
RESULTS Demographic Characteristics
The suicide completers were comparable with the control group with disruptive disorder with regard to gender (92% versus 78% male), race (92% versus 100% white), age (17.4 [ 1.91 versus 17.1 [ 1.21 years), socioeconomic status (3.1 [1.2] versus 2.9 [0.9]) (Hollingshead, 1975),and family constellation (living with both parents 25.4% versus 38.9%). Characteristics of the Suicide
Among the completers with disruptive disorders, about one third were toxicologically positive for substances (34.5%), with 26.8% positive for alcohol and 13.2% positive for other substances. The most common methods of completed suicide were firearms (62.7%), hanging (27.1%), and jumping (6.8%). Overdose and asphyxiation represented 1.7% each. Characteristics of the Disruptive Disorders
CD was more common in the suicide than the control group (59.3% versus 33.3%; x2 = 3.74 ;p = .05; odds ratio [OR] = 2.9; 95% confidence interval [CI] = 1.0, 8.8). While there were no differences between the 2 groups with respect to the type of antisocial behavior, completers showed a higher number of CD symptoms compared with the controls (median interquartile range = 3.0 [4.0] versus 1.0 [2.3]; Mann-Whitney U = 541.0; D = .02). There were no differences between the grouDs with respect to the prevalence of ADD. V
‘Omorbid
I
Psychiatric Disorders
The 59 completers with disruptive disorders had higher rates of current substance abuse than did community controls with disruptive disorders (46.6% versus
847
R E N A U D ET AL
x2
5.6%; = 9.92; p = .002; OR = 14.8; 95% CI = 1.9, 118.7), but not lifetime substance abuse (50.0% versus 44.4%; p = not significant [NS]) (Table 1 ) . More completers with disruptive disorders had a specific diagnosis of current alcohol abuse (43.1% versus 5.6%; = 8.61; p = .003; OR = 12.9; 95% CI = 1.6, 103.4) and current drug abuse (24.6% versus 0.0%; FET,p = .02) than controls with disruptive disorders. Counter to hypotheses, there were no differences between the 2 groups in the rates of comorbid major depressive disorder.
x2
death) (18.4% versus 0.0%; F E T p = NS), although the difference was in the predicted direction. Life Events
There were no differences between the groups with respect to frequencies of interpersonal conflict with parent, with boyfriend/girlfriend, or interpersonal loss of boyfriend/girlfriend (allp values = NS), within the year of suicide. There was a trend among the completers with disruptive disorders to have had legal difficulties in the previous year compared with the controls with disruptive disorders (63.0% versus 38.9%; = 3.07;p = .08). Within the subsample for whom assessment of lifetime abuse was available (suicides subsample, n = 29; controls subsample, n = 18), there was a higher rate of lifetime abuse among completers (58.6% versus 5.6%; = 13.24;p = .0003). This was primarily attributable to a higher rate of physical abuse in the completers (55.2% versus 0.0%; FET, p = .0001), with no difference in the rate of sexual abuse (8.0% versus 5 . 6 % ; = ~ NS).
x2
Past Suicidal BehaviodSuicidal and Homicidal Ideation Within the Previous Week
Completers with disruptive disorders had a much greater prevalence of past suicide attempts than controls with disruptive disorders (52.7% versus 5.6%; = 12.47;~ = .0004; OR = 19.0; 95% CI = 2.4, 152.6). In addition, completers were more likely to have shown evidence of suicidal ideation (defined as any ideation, gesture, or threat) than were controls during the week before interview (52.5% versus 16.7%; = 7.20; p = .007;OR= 5.5; 95%CI= 1.4,21.2).Therewerenosignificant differences in the manifestation of homicidal ideation (defined as a homicidal plan in the week before
x2
x2
x2
Availability of Firearms
Although all the differences were in the expected direction, completers did not differ from controls with
TABLE 1 Demographic and Clinical Risk Factors for Suicide in Disruptive Youths (%) Completer (n = 59)
Control ( n = 18)
59.3 3.0 [4.0]
33.3 1.0 [2.3]
.05 .02
2.9
Current substance abuse Lifetime substance abuse Current alcohol abuse Current drug abuse MDD
46.6 50.0 43.1 24.6 30.5
5.6 44.4 5.6 0.0 16.7
.002 NS .003 .02 NS
14.8 1.3 12.9 12.3 2.2
(1.8, 118.7) (0.4, 3.6) (1.6, 103.4) (0.7, 217.8) (0.6, 8.5)
Attempt Past suicidality Homicidal ideation
52.7 52.5 18.4
5.6 16.7 0.0
.0003 .007 NS
19.0 5.5 8.7
(2.4, 152.6) (1.4,21.2) (0.5, 157.2)
Lifetime abuse Lifetime physical abuse
58.6 (n = 29) 55.2 (n = 29)
5.6 0.0
.0003 .0001
24.1 45.2
(2.8, 206.3) (2.5, 821.6)
First-degree relatives Any affective disorder Substance abuse
27.6 (n = 214) 23.9 (n = 218)
14.0 (n = 50) 4.1 (n = 4 9 )
,046 .002
2.3 7.4
(1.00, 5.5) (1.7,31.4)
Second-degree relatives Any affective disorder Substance abuse
12.4 (n = 498) 16.8 (TI = 506)
4.0 (n = 126) 4.6 (n = 131)
,006 ,0004
3.4 4.2
(1.4, 8.7) (1.8,9.9)
Conduct disorder only No. of conduct symptoms: median [IQR]
Note: O R = odds ratio; CI = confidence interval; IQR = interquartile range; M D D significant. In cases in which a zero cell is present, 0.5 is added to all cells.
848
pValue
=
OR
95% CI" (1.0,S.S)
major depressive disorder; NS
=
not
J . A M . ACAD. C H I L D A D O L E S C . PSYCHIATRY, 3 8 : 7 , JULY 1999
SUICIDE AND DISRUPTIVE DISORDERS
respect to any gun availability variables: any guns (66.6% versus 50.0%;p = NS), long guns (56.0% versus 43.7%; p = NS), hand guns (33.9% versus 11.8%; p = NS), or loaded guns (11.1% versus 5.6%; p = NS). Family History of Psychiatric Disorder
The rates of any mood disorder (27.6% versus 14.0%;
x2 = 3.98; p < .05) and substance abuse (23.9% versus 4.1%; x2 = 9 . 6 9 ; =~ .002) among first-degree relatives of the completers were higher than those of controls. Much of the difference in the familial rates of substance abuse between suicides and controls was due to the higher rates of alcohol abuse in fathers of the completers (34.6% versus 0.0%; FET,p = .01). The differences with respect to the rates of suicide attempts did not reach significance (9.9% versus 3 . 6 % ; ~ = NS). Similarly, the rates of mood disorder (12.4% versus 4.0%; x2 = 7.60; p = .006) and substance abuse (16.8% versus 4.6%; x2 = 12.70; p = .0004) were also higher among the second-degree relatives of the suicide victims. Logistic Regression
After control for CD, the most parsimonious model consisted of the following risk factors: current substance abuse (OR = 16.0; 95% CI = 7.8, 32.9), past suicide attempt (OR = 14.0; 95% CI = 7.2, 27.4), family history of substance abuse (OR = 3.9; 95% CI = 1.7, 9.2), and family history of mood disorder (OR = 2.5; 95% CI = 1.2, 5.1). This model fit the data well (HosmerLemeshowX2=6.14;df=7;p= .52). DISCUSSION
To our knowledge, this is the first study to have examined risk factors for completed suicide in those with disruptive disorder. We found that several risk factors were associated with suicide above and beyond disruptive disorder, namely current substance abuse, past suicide attempt, history of physical abuse, and family history of substance abuse and mood disorders. However, we did not find differences compared with the control group with respect to rates of either ADD or comorbid mood disorder. Before discussing the above-noted findings, it is important to take into account that this study included a relatively small sample of suicide completers and controls. Nevertheless, both groups are representative of the population, and therefore the estimates of the risk factors are
J . AM. ACAD. C H I L D A D O L E S C . PSYCHIATRY, 3 8 : 7 , JULY 1999
not subject to the usual biases involved in studies of clinical populations. Finally, because of the nature of a psycho-
logical autopsy study, in-depth assessment of personality traits and family interaction variables that are likely to contribute to disruptive behavior were not available. Because the sample was almost entirely white and male, caution should be made in generalizing to females and to other ethnic groups. However, Shaffer and colleagues' (1996) more ethnically diverse report shows similar results and supports the view that C D is a risk factor for suicide in females as well. Cairns et al. (1988) found high rates of suicide attempt in delinquent females, also consistent with this view. However, in an analysis of all 140 suicides and 131 controls, we found that C D was a risk factor only for males (Brent et al., unpublished). The interaction of sex and aggression and its impact on risk for suicide is an important area of further inquiry. We found a high prevalence of current comorbid substance abuse in the suicide completers but no differences in lifetime rates of substance abuse. This suggests that it is not vulnerability to substance abuse per se that puts one at risk for suicide, but rather the acute and subacute effects of substance abuse that heighten the disruptive youth's risk for suicide. These mechanisms may include decreasing brain serotonin (Ballenger et al., 1979), impairing judgment and increasing impulsivity, and increasing the likelihood of using a gun (Brent et al., 1987, 1993~). In the literature, past suicide attempt was found as a strong predictor of future suicidal behavior across diagnostic groups, being reported consistently in several psychological autopsy studies (Brent et al., 1988, 1993a; Shaffer et al., 1996; Shafii et al., 1985). In this study, past suicide attempt was a major risk factor, and, while not as strong, suicidal ideation within a week of the completion was also a predictor. Past suicide attempt appears to be a strong risk factor for completed suicide, independent of diagnosis (Brent et al., 1993a; Shaffer et al., 1996). Although only data on a subsample were available in our study, physical abuse was strongly associated with suicide. This is consistent with other studies showing that physical and sexual abuse have been reported to be associated with suicide and suicidal behavior in several studies (Brent et al., 1993b, 1994; Fergusson et al., 1996; Gould et al., 1996; Hibbard et al., 1988; Kaplan et al., 1997; Romans et al., 1995). Abuse may set in motion a complex cascade of psychological and physiological alterations that increase the 849
R E N A U D E T AL.
regard to ADD alone (Brent et al., 1993a; Shaffer et al., 1996).Therefore, ADD per se does not appear to predispose to risk for suicide, although ADD may predispose to other difficulties such as C D and substance abuse, which in turn mediate poor outcome in youths with ADD (Brent et al., 1993a; Mannuzza et al., 1993).
risk of dysphoria and self-destructive behavior (De Bellis et al., 1994; Kaplan et al., 1997; Kaufman, 1991; Kaufman et al., 1998). We also found a trend toward more frequent 1egaVdisciplinary problems in the cornpleters. Therefore, youths in detention or in the midst of a disciplinary crisis may be at increased risk for suicide, due to both the disruptive disorder that precipitated the crisis and the stressor itself (Shaffer, 1974). In addition, family history of substance abuse and of mood disorders were also risk factors for suicide. As previously demonstrated in the literature using unselected samples of adolescent completers, family history conveys risk even with control for proband psychopathology (Brent et al., 1994; Gould et al., 1996). The relationship between family history of psychopathology and adolescent suicide may be mediated by both genetic and environmental mechanisms. The likelihood that environmental mechanisms play a role in this sample is increased by the persistence of family history variables in logistic regression even with control for suicide attempt and substance abuse in the individual. In our study, the specific association of suicide with paternal alcoholism may indicate that there was a male-to-male transmission of type I1 alcoholism (Cloninger, 1987). In turn, early-onset, malelimited alcoholism has been associated with impulsive violence and suicidal behavior (Buydens-Branchey et al., 1989). An alcoholic parent, in addition to increasing the genetic liability for alcoholism in hidher child, also increases the likelihood of family violence, discord, and instability (Chaffin et al., 1996). It is also interesting to note that in our samples of completers and controls, in the presence of disruptive disorders, comorbid major depressive disorder was not associated with suicide. Shaffer and colleagues (1996) found that a combination of dysthymic, conduct, and substance abuse disorders occurred as a "triple threat" in older male suicides. Alternatively, we previously reported that C D was at least as great a risk factor for suicide in the absence of mood disorder (Brent et al., 1993a). The results presented herein, from a sample that overlaps with the above-cited study, not surprisingly supports the latter point of view. Therefore, the combination of disruptive disorder and substance abuse should be regarded critically with respect to risk for suicide, even in the absence of mood disorder. Finally, similarly to other studies, we found that the completers had a higher rate of CD, whereas there were no differences between completers and controls with
American Psychiatric Association (1980), Diagnosticand Statistical Manual of Mental Disorders, 3rd edition (DSM-III). Washington, DC: American Psychiatric Association Andreasen N, Endicott J, Spitzer R, Winocur G (1977), The family history method using Research Diagnostic Criteria: reliability and validity. Arch Gen Pychiatry 34: 1229-1235 Ballenger JC, Goodwin FK, Major LF, Brown GL (1979), Alcohol and central serotonin metabolism in man. Arch Gen Pychiatry 36:224-247 Brent DA, Bridge J, Johnson BA, Conolly J (1996), Suicidal behavior runs in families: a controlled family study of adolescent suicide victims. Arch Gen Pychiatry 53: 114 5-1 152 Brent DA, Perper JA, Allman C (1987),Alcohol, firearms, and suicide among youth: temporal trends in Allegheny County, Pennsylvania, 1960-1983. JAMA 257:3369-3372 Brent DA, Perper JA, Goldstein CE et al. (1988), Risk factors for adolescent suicide. Arch Gen Pychiatry 45:581-588 Brent DA, Perper JA, Moritz G et al. (1993a), Psychiatric risk factors for adolescent suicide: a case-control study. J Am Acad Child Adolesc Pychiatry 32:521-529 Brent DA, Perper JA, Moritz G et al. (1993b), Stressful life events,, psychopathology, and adolescent suicide: a case control study. Suicide Lifi Threat Behav 23:179-187 Brent DA, Perper JA, Moritz G et al. (1993c), Psychiatric sequelae to the loss of an adolescent to suicide.J A m Acad ChildAdolescPsychiatry 32509-5 17 Brent DA, Perper JA, Moritz G et al. (1994), Familial risk factors for adolescent suicide: a case-control study. Acta Pychiatr Scand 89:52-58 Buydens-Branchey L, Branchey MH, Noumair D (1989), Age of alcoholism onset. Arch Gen Pychiatry 46:225-230 Cairns RB, Peterson G, Neckerman HJ (1988), Suicidal behavior in aggressive adolescents. J Clin Child Pychol17:298-309 Chaffin M, Kelleher K, Hollenberg J (1996), Onset of physical abuse and neglect: psychiatric, substance abuse, and social risk factors from prospective community data. ChildAbuse Negl20: 191-203 Chambers WJ, Puig-Antich J, Hirsch M et al. (1985), The assessment of affective disorders in children and adolescents by semistructured interview: test-retest reliability of the Schedule for Mective Disorders and
850
J . AM. ACAD. C H I L D A D O L E S C . PSYCHIATRY, 38:7, JULY 1999
Clinical Implications
In conclusion, adolescents with the combination of disruptive disorder and current substance abuse are at high risk of committing suicide. Proper assessment and management of substance abuse in disruptive youths is likely to be an important means of reducing the substantial suicide risk in this subgroup. The risk increases for adolescents who have had a previous suicide attempt, who have a history of physical abuse, and who have a family history of substance abuse and mood disorders. Clinicians should be aware of these risk factors and implement active interventions to prevent suicide. Treatment should focus not only on the adolescent, but also on family members. REFERENCES
SUICIDE A N D DISRUPTIVE DISORDERS
Schizophrenia for School-Age Children, Present Episode version. Arch Gen Psychiatry 42:696-702 Cloninger CR (1987), Neurogenic adaptive mechanisms in alcoholism. Science 236:4 10-4 16 De Bellis MD, Chrousos GI: Dorn LH et al. (1994), Hypothalamic-pituitaryadrenal axis dysregulation in sexually abused girls. / Clin Endocrinol Metab 78:249-255 Fergusson DM, Honvood LJ, Lynskey MT (1996), Childhood sexual abuse and psychiatric disorder in young adulthood, 11: psychiatric outcomes of childhood sexual abuse./Am Acad ChildAdolerc Psychiatiy 35: 1365-1374 Gould MS, Fisher P, Parides M, Flory M, Shaffer D (1996), Psychosocial risk factors of child and adolescent completed suicide. Arch Gen Psychiatry 53:1155-1162 Hibbard RA, Brack CJ, Rauch S, Orr DP (1988), Abuse, feelings, and health behaviors in a student population. Am/ Dis Child 142:326-330 Hollingshead AB (1975), Four-Factor Index of SocialStatur. New Haven, CT: Yale University Department of Sociology Kaplan S, Pelcovitz D, Salzinger S, Mandel F, Weiner M (1997). Adolescent physical abuse and suicide attempts. J A m Acad Child Adolesc Psychiatiy 36:799-808 Kaufman J (1991), Depressive disorders in maltreated children. / A m Acad ChildAdolesc Psychiatiy 30:257-265
Kaufman J, Birmaher B, Perel J et al. (1998), Serotonergic functioning in depressed abused children: clinical and familial correlates. Biol Psychiatry 44973-981 Mannuzza S, Klein RG, Bessler A, Malloy P, LaPadula M (1993), Adult outcome of hyperactive boys: educational achievement, occupational rank, and psychiatric status. Arch Gen Psychiatiy 50:565-576 Orvaschel H, Puig-Anrich J, Chambers W, Tabrizi MA, Johnson R (1982), Retrospective assessment of prepubertal major depression with the Kiddie-SADS-E. J A m Acad Child Psychiatry 21:392-397 Rich CL, Young D, Fowler RC (1986), San Diego suicide study, I: young vs old subjects. Arch Gen Psychiatiy 43:577-582 Romans SE, Martin JL, Anderson JC, Herbison GP, Mullen PE (1995), Sexual abuse in childhood and deliberate self-harm. Am J Prychiatry 152:1336-1342 Shaffer D (1974), Suicide in childhood and early adolescence./ ChildPsychol Psychiatry 15:275-291 Shaffer D, Gould MS, Fisher P et al. (1996), Psychiatric diagnosis in child and adolescent suicide. Arch Gen Pychiatiy 53:339-348 Shafii M, Carrigan S, Whittinghill JR, Derrick A (1985), Psychological autopsy of completed suicide in children and adolescents. Am/ Psychiany 142:1061-1064
Accidental and Suicidal Adolescent Poisoning Deaths in the United States, 1979-1994. Greene Shepherd, PharmD, Wendy Klein-Schwartz, PharmD, MPH
Objective:To describe the epidemiological features of poisoning deaths in adolescents in the United States. Design: Descriptive analysis of poisoning deaths in persons aged 10 to 19 years in the United States from January 1, 1979, to December 31, 1994, based on national mortality data. Stud’ Population: Adolescents whose cause of death was identified as poisoning using Znternational Chs2fication of Diseases, Ninth Revision codes. Main Outcome Measure: Nature of injury (accident vs suicide). Results: There were 4129 suicides and 3807 accidental deaths due to poisoning. Victims were most frequently male and white. However, poisoning was more often the method of suicide in adolescent girls than in boys (28.0% vs 8.7%). The number of deaths (7138 vs 798) and death rate (2.36 vs 0.28 per 100 000 population) were higher in 15- to 19-year-olds vs 10- to 14-year-olds.The distribution of substances involved was different for 10- to 14-year-olds compared with 15- to 19-year-olds and for suicides compared with accidents. Among 10- to 14-year-olds, drugs other than alcohol accounted for 232 (85.3%) of 272 suicides but only 118 (22.4%) of 526 accidental deaths. Gases and vapors played an important role in accidental deaths and suicides in 15- to 19-year-olds and in accidents in 10- to 14-year-olds. Conclwions:The rates of suicides and accidental poisoning deaths were lower in 10- to 14-year-olds compared with 15- to 19-year-olds. Areas where injury-prevention efforts might have an influence on adolescent fatalities include management of depression, substance abuse education, and use of carbon monoxide detectors or shutoff switches. Arch Pediatr Adolesc Med 1998; 152:1181-1 185.
J . AM. ACAD. C H I L D A D O L E S C . PSYCHIATRY, 38:7, JULY 1999
85 1