- Email: [email protected]
Posttraumatic Stress Disorder in Peers of Adolescent Suicide Victims: Predisposing Factors and Phenomenology
Posttraumatic Stress Disorder in Peers of Adolescent Suicide Victims: Predisposing Factors and Phenomenology DAVID A. BRENT, M.D., JOSHUA A. PERPER, M.D., LL.B., GRACE MORITZ, A.C.S.W., LAURA LIOTUS, B.S., DOREEN RICHARDSON, M.Eo., REBECCA CANOBBIO, B.S., JOY SCHWEERS, M.Eo., AND CLAUDIA ROTH, M.S.W.
ABSTRACT Objective: To examine the factors predisposing to posttraumatic stress disorder (PTSD) in peers of adolescent suicide victims. Method: One hundred forty-six adolescents who were the friends of 26 suicide victims were studied. Five percent (n
=8) developed PTSD after exposure to suicide. These 8 subjects with PTSD were compared to the remainder
of the exposed subjects (n = 138). Results: Subjects with PTSD were more likely than those without PTSD to have had a history of substance abuse, agoraphobia, and suicide attempts. Subjects who developed PTSD were more likely to have developed a new-onset depression, to have more severe grief, and to have been closer to the suicide victim. Subjects with PTSD tended to have more severe exposure to suicide and came from discordant households with a history of disruptions in key relationships. The 8 subjects who developed PTSD were compared to 38 subjects who developed new-onset depression but not PTSD. Those with PTSD were more likely to have had past substance abuse, prior suicide attempts, family history of panic disorder, a history of parent-child disruption, and a history of loss. Symptoms of intrusive visual images, hypervigilance, and avoidance of reminders discriminated subjects who had PTSD from new-onset depressives without PTSD. Conclusions: PTSD is an expectable outcome in youth exposed to suicide. Further work is required to differentiate symptoms of depression from PTSD. J. Am. Acad. Child Ado/esc. Psychiatry, 1995, 34, 2:209-215. Key Words: adolescent, posttraumatic stress disorder, depression, suicide.
Posttraumatic stress disorder (PTSD) is now widely recognized as a response to intense overwhelming physicalor psychological trauma. Initial work by Terr (1991) and Pynoos et al. (1987) documented carefully that PTSD does occur in younger children and adolescents and that the phenomenology of the disorder does not differ greatly from the manifestation ofPTSD in adults. PTSD has been reported to occur in youthful Cambodian refugees who survived a concentration camp (Kinzie et aI., 1986), among witnesses and survivors of a homicide or other traumatic deaths (Green et aI., 1991;
AcceptedJanuary 19, 1994. Dr. Perper was with the Allegheny County Coroner's Office. He is now Medical Examiner, Brou/ard, FL. All other authors are with the Western Psychiatric Institute and Clinic, Piwburgh, PA. This work W(/S supported by NIMH grant MH44111, "Youth Exposed to Suicide." The authors thank Karen Rhinaman fOr her assistance in preparing the manuscript. Reprint requests to Dr. Brent, Western Psychiatric Institute and Clinic, 3811 O'Hartt Street, Pittsburgh, PA 15213. 0890-8567/95/3402-0209$03.00/0©I995 by the American Academy of Child and Adolescent Psychiatry.
Pynoos et al., 1987; Schwartz and Kowalski, 199 I), and in response to physical or sexual abuse (Famularo et al., 1993; Kiser et aI., 1988, 1991; McLeer et al., 1988). These reports have contributed important data to the validation of the diagnosis of PTSD. However, little is known about what factors may predispose to the development of PTSD in younger populations, other than the severity of life threat experienced and the closeness in relationship to the victim (Pynoos et al., 1987). In a recent case-control report of adolescents who witnessed a suicide on a school bus, the development of anxiety disorders, including PTSD, was predicted by a family history of suicide attempt and affective illness, and higher number of stressful life events in the previous year (Brent et aI., I993a). However, the number of exposed adolescents was relatively small (n = 28), and only four of this group developed PTSD, making it impossible to draw firm conclusions about risk factors for PTSD. In adults, as in younger populations, exposure to
]. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 34:2, FEBRUARY 1995
209
BRENT ET AI..
life-threatening situations is among the most salient predictor of the development of PTSD (Breslau et al., 1991; Goldberg et al., 1990; Green et al., 1991; Helzer et al., 1987). Moreover, while antisocial symptoms and substance abuse are associated with risk for exposure to traumatic violence (Breslau et al., 1991; Helzer et al., 1987), it has recently been shown that both family and personal history of anxiety disorder were most predictive of the development of PTSD (Breslau et al., 1991). Other factors, such as an "irritable family atmosphere" may also contribute to risk for PTSD (Green et al., 1991). In our ongoing study of the impact of exposure to suicide in the friends and acquaintances of suicide victims, we have reported an increased risk of PTSD subsequent to exposure to suicide (Brent et al., 1993b). This nonreferred sample, for whom characteristics of the exposure to suicide, previous psychiatric history, family history, and previous stressors were gathered, presents an opportunity to examine risk factors for the development of PTSD. Therefore, we compared those youth who developed PTSD to those who did not in the above-noted domains. Since a number of studies have noted a cooccurrence of PTSD, grief responses, and depression (Brent et al., 1993a; Davidson et al., 1991; Pynoos et al., 1987), we also attempted to identify risk factors and symptomatic components of PTSD that were independent from the potentially overlapping constructs of depression and grief.
METHOD Sample The sample consists of 146 adolescent friends and acquaintances of 26 adolescent suicide victims described previously (Brent et al., 1993b). The suicide victims were from 28 counties in western Pennsylvania. The families of 78% of all consecutive adolescent suicides agreed to a psychological autopsy, but only 67% provided a social network. The characteristics of suicide victims whose families agreed to the study were not different from the characteristics of those who refused with respect to age, race, gender, toxicology, or county of origin. However, acceptors were more often families of suicide victims who used a gun (80.8% versus 44.0%, X2 ~ 4.32, P ~ .04). Friends of the suicide victims were nominated by parents and siblings as being close to the victim and having had a mutually confiding relationship. Two friends nominated by both a parent and a sibling were chosen at random. Each of these two friends were asked to nominate other close friends, and of these peernominated friends, one additional friend was chosen. Acquaintances
210
J.
were chosen by asking the friends to nominate social contacts of theirs who knew the suicide victim, but not as well as the friends themselves did. Three acquaintances were chosen at random from this list. The rate of agreement for interview for friends and acquaintances was 61%. Refusers were younger than acceptors by an average of 10 months (p ~ .01) and were more likely to be male (76% versus 54%, X2 ~ 11.3, P ~ .0008). Exposed subjects were interviewed a median of 7 months after the death (mean ~ 7.3 months, SD ~ 1.7 months). The median network size was six (range two to seven). In a previous report (Brent et al., 1993b), we noted that the rate of new-onset PTSD was 8 (5%) of 146, compared with none in an equal number of nonexposed community controls (p < .05). Subsequent analyses presented herein will compare these 8 subjects with PTSD to the remaining 138 exposed subjects without the syndrome of PTSD.
Assessment Psychiatric assessment was made by use of the Schedule for Affective Disorders and Schizophrenia for School-Age Children, Epidemiologic and Present Episode versions (Chambers et al., 1985; Orvaschel er al., 1982). Interrater reliability for diagnoses (K ~ .98, SD ~ .13) and for symptoms (intraclass correlation coefficient ~ .96, SD ~ .7) was high. Symptoms of PTSD were assessed by the Post-Traumatic Stress Disorder Reaction Index (PTSDRI) (Pynoos et al., 1987). This is a 20-item inventory of PTSD symptomatology with good internal consistency and discriminant and convergent validity (Pynoos et al., 1987). Each item is rated on a Likert scale from I to 5, with 5 being the most problematic. Symptoms were rated both retrospectively at I month after exposure to the suicide, and at the time of the interview, about 7 months after the death. For this study, internal consistency of the PTSDRI was high (a ~ .83 to .88). Life events and family history were assessed by an interviewer blind to the proband's diagnostic status. Life events were assessed by a 29-item life events inventory with acceptable face and discriminant
validity (Brent et al., 1988, 1993b). Family history was assessed using a Family History-RDC with diagnoses modified to DSMIII (Andreasen et al., 1977). The closeness to the victim was assessed by the Adolescent Relationship Inventory (ARI) (Brent er al., 1992, 1993b), which covers social exchange, provision of social support, mutual admiration, and frequency of contact. In previous studies, we have shown high internal consistency and ability to discriminate friends from acquaintances with this measure. The Circumstances of Exposure to Death instrument was used to assess the degree of exposure to suicide (Brent et al., 1992, 1993b). This instrument covers circumstances of the death, direct exposure (if the suicide was witnessed), indirect exposure (discovering of the body, visiting the scene of death, etc.), and events following the death (wake, funeral, visitation). Interrater reliability with this instrument has been reported as quite high. Grief was measured by the Texas Grief Inventory (TGI) (Faschingbauer, 1977), a self-report questionnaire filled out by the exposed adolescents. Psychosocial functioning in school, with peers, siblings, and with parents was assessed by use of the Psychosocial Schedule (Lukens et al., 1983).
Data Analyses The subjects who developed PTSD were compared to those who did not using standard parametric and nonparametric statistics. For categorical comparisons, a Pearson's X2 or Fisher's Exact Test
AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 34:2, FEBRUARY 1995
POSTTRAUMATIC STRESS DISORDER IN ADOLESCENTS
TABLE 1
Demographic Characteristics. The peers exposed to suicide who developed PTSD were more likely to be female (100% versus 42.8%, Fisher's Exact Test, p = .002), and older (20.1 [1.5J versus 18.2 [2.0J years, MWU = 248.5, P = .01) (Table 1). This demographic difference may have emerged in part because the most direct exposure to suicide in this study occurred among the female roommates of an older adolescent female who hung herself and was discovered hanging in her
Family History. There were no differences in the rates of disorder between the first-degree relatives of the PTSD and non-PTSD subjects, with respect to affective illness (14.0% versus 10.1 %), anxiety disorder (8.1% versus 5.8%), substance abuse (14.1% versus 12.1 %), conduct disorder (3.1 % versus 3.3%), or suicidal behavior (7.2% versus 3.9%). Exposure to the Suicide. PTSD subjects tended to be more likely to have seen the scene of death, witnessed the suicide, or discovered the body of the victim than were peers without PTSD (57.1 % versus 27.3%, X2 = 2.89, P = .09), to have known the victim's plans before death (42.9% versus 22.6%), to have believed the suicide was preventable (50.0% versus 17.1 %, X2 = 5.27, P = .02), and to have had a conversation with the victim on the day of the suicide (75% versus 44.9%, not significant). The total number of these exposure variables tended to be greater in those peers who developed PTSD (2.6 [2.0J versus 1.5 [1.5J, MWU = 367, P = .06). Distribution by Network. The PTSD cases were not randomly distributed by network (X 2 = 39.5, df = 25, P = .03). Three of the eight cases occurred in one network. In this network, exposed roommates of the suicide victim found the victim hanging in a closet. As reported previously, there was no association between network membership and new-onset depression (Brent et al., 1993b). Closeness to the Victim. Subjects who developed PTSD had a closer relationship to the suicide victim than did those without PTSD, as measured by the ARI (67.3 [33.6] versus 45.4 [25.2], MWU = 328.8, P = .02). Family Environment. As measured by the Psychosocial Schedule, the mother-subject relationship was more discordant in the PTSD group (17.5 [0.7] versus 12.0 [3.9], t = 1.99, df = 110, P = .05), as was the marital relationship between parents (33.0 [O.OJ versus 20.7 [5.9], MWU = 5.0, P = .04). Father-child and overall parent-child relations were in the same direction
closet.
as the ones noted above, although the differences
Past Psychiatric Disorder. Peers with PTSD were more likely to have had a history of substance abuse (62.5% versus 28.3%, X2 = 4.21, P = .04), agoraphobia (12.5% versus 0.7%, X2 = 7.76, P = .005), and suicide attempts (37.5% versus 2.9%, X2 = 19.8, P < .0001) (Table 2).
escaped statistical significance. Life Events. In the year before the assessment, subjects with PTSD were more likely than those without PTSD to have had interpersonal conflict with siblings (25.0% versus 5.9%, X2 = 4.27, P = .04) or with boyfriends or girlfriends (50.0% versus 19.1%, X2 = 4.37, P = .04), as well as to have experienced the death of a
Demographic Characteristics PTSO
Sex (% females) Age (yr) Mean SO Race (% white) SES (%) I II III IV V Lives with both biological parents (%) Note: PTSO nomic status.
=
No (n = 138)
Yes (n = 8)
p
42.8
100.0
.002
18.2 2.0 97.1
20.1 1.5 100.0
.01
5.8 27.7 32.8 28.5 5.1
0.0 25.0 50.0 12.5 12.5
48.9
25.0
posttraumatic stress disorder; SES
.001 =
socioeco-
was used. For comparisons on continuous data, either a t test or Mann-Whitney U (MWU) was used. To further differentiate depression and PTSD, subjects who developed PTSD were compared to those who developed depression, but not PTSO, using standard univariate statistics. Logistic regression was used to select those variables that were most closely associated with the diagnosis of PTSO. Multiple linear regression was used to examine the relationship of these variables to the symptomatic expression of PTSO. Multivariate analyses of covariance were used to examine the discriminant validity of specific items for the PTSORI while controlling for depression and for grief.
RESULTS
J. AM.
ACAD. CHILD ADOLESC. PSYCHIATRY. 34:2. FEBRUARY 1995
211
BRENT ET AI..
TABLE 2 Sym p to ms of PTS D on th e PT SD RI in Those w ith and witho ut DSM-III-R PT SD O ne Mo nt h PT SD
No
Yes (n = 8)
= 138)
Symp toms
(n
Scared whe n think o f death Im ages of death (volun ta ry) Im ages of death (in trusive) D reams/ n igh tmares related to th e death Fear that others or self mi ght d ie Difficulty enjoying thi ngs O ther people don 't und erstand C ur off from own feelings D ifficulty explaining feelings M ore likely to be startled Problems sleepi ng G uilt/ rcgrers abou t de at h Int erfering w ith memOlY Interfer ing wi th con centrat ion Staying away fro m reminde rs Rem ind ers creat e tensio n or stress W ant fami ly aro u nd more often Affected well-bein g More impulsive or diffi cult
1.4 2.1 1.4 0.7 0.7 1.4 1.4 0.8 2.3 0.7 2.8 1.6 0.5 2.0 1.2 2.0 0.3 0.6 0.2
At Ti me of Int erview PTSD
(1.4) ( 1.4) (1.3) (1.2) ( J.1) (1.5) (1.6) (1.4) (1.7) (1.3) (1.5) (1.6) (1.0) (1.8) (1.5) ( 1.6) (0.8) ( \.2) (0.7)
3.3 3.6 3.0 1.8 1.8 3.6 3.4 2.0 2.0 2.6 0.5 2.4 0.8 1.5 3.4 3.8 2.9 2.9 0.4
( 1.5)** (0.7)** (J .4)** (1.9) (1.6)* (0.7)*** (J .4)** (2.1) (1.9) ( 1.7)*** (0.8)*** ( 1.8) (J.4) ( 1.9) ( 1.4)*'(0.5)** (1.8)**(2.0)*** (0.7)
No ( n = 138)
0.7 0.8 0.5 0.2 0.5 \.2 0.9 0.4 2.5 0.4 3.6 0.7 0.1 2.4 0.6 1.0 0.1 0.1 0.1
(1.0) ( 1.2) (0.8) (0.6) (0.9) (1.6) (1.4) (0.9) (1.6) (1.0) ( 1.1) ( 1.1) (0.4) (1.8) ( 1.0) ( 1.2) (0.5) (0.4) (0.5)
Yes (n = 8)
1.8 2.0 1.9 0.8 1.0 1.3 2.8 0.8 3.0 2.5 2.6 1.9 0.3 1.9 2.8 2.4 2.0 0.9 0.3
( 1.6)* ( J.1)** (1.6)* * (\.2)* ( 1.2) (1.4) (J .5)** ( 1.4) (1.4) ( 1.7)*** (1.4)* * ( 1.5)* (0.5) ( 1.6) (1.8)-** (J .4)** ( 1.9)*** ( 1.5)*' (0.7)
N ote: Va lues represent mean (SD) . PT SD = posttra umatic stress d isorder; PTSDRI = Post-Traumati c Stress D isord er Reaction Ind ex. < .0 1; ' ''p < .00 1 (after adj ustment for cu rrent depression).
"P < .05; -'p
nonparental relative (50.0% versus 19.9%, X2 = 4.07, P = .04). For life events that ant edated the year before diagnostic assessment , subjects with PTSD had higher rates of parent-child disrup tions (62.5% versus 25.7% , X2 = 5.09, P = .02), sibling deaths (25.0% versus 3.7%, X2 = 7.43, P = .006), and death of a friend oth er th an the suicide victim (50% versus 19.4%, X2 = 4.07, P = .04). T here was no difference between the two groups as to the total number of life events experienced. Current Psychiatric Symptomatology. Subjects with PT SD were more likely than th ose without PT SD to have a concurrent diagnosis of major dep ression (62.5% versus 23 .9%, Xl = 5.85, P = .02). All five subjects with PTSD and concurre nt depression developed major depression simultaneously with onset of PT SD. For the thre e others with PT SD , a preexistin g depression antedated the PTSD. Subjects with PTSD had significantly more symptoms of depression than the non PTSD group (8.6 [2.0J versus 3.9 [3.4], MWU = 148.5, P = .0001 ). Subjects with PTSD had significan tly more severe ratin gs of sleep disturbance (p = .01) , guilt (p = .002), worthlessness (p = .007), somatic compl aint s (p = .005), and suicidality (p = .006).
212
J.
PTSD Symptomatology. At 1 month after the death in the index suicide victim, subjects with PT SD scored much highe r on th e PT SDRI than the non-PTSD peers (53.3 [10.3J versus 26.0 [12.9], t = 5.86, df = 131, P < .000 1) (Table 2). At the time of th e int erview, PTSDRI scores were still higher in subjects with PTSD (35. 5 [9.5J versus 15.8 [7.7J, t = 6.92, df = 138, P < .0001). There was a fall-off in severity of symptomatology in both group s, altho ugh the fall-off was more rapid in the PTSD group (group X time interaction, F [ I ,13 l] = 4.96, P = .03). Subjects with PT SD also showed more grief scored on the TGI (42.3 [11.4J versus 63.0 [17.0], t = 2.42, df = 115, P = .02). Symp toms of PTSD were highly correlated with both depression (r = .63, P < .001 ) and grief ( r = - .47, P < .00 1). At 1 month, 12 of 20 and at the time of inte rview, 10 of 20 symptoms of PTSD were more severe in subjects d iagnosed with PTSD compared to those who were not. These persistent sympto ms that differenti ated the PTSD and non -PTSD groups most shar ply were both voluntary and intrusive images of th e death , a feeling that other people did not understand the
AM . AC A D . C H IL D AD OL ES C. PSYCH IATR Y. 34 :2 . F E BRUAR Y 1995
POSTTRAUMATIC STRESS DISORDER IN ADOLESCENTS
experience, hypervigilance and an exaggerated startle, and avoidance of reminders of the death. Two other symptoms, guilt and regrets about the death, and difficulty sleeping also differentiated the two groups, but may represent general symptoms of grief and depression rather than specific symptoms of PTSD. After adjustment for severity of depressive symptomatology using analysis of covariance, all 10 symptoms noted on Table 2 still differentiated subjects with PTSD from those without PTSD. After adjusting for differences in the severity of grief, using the TGI, the PTSD group still scored higher on the PTSDRI than did the non-PTSD group (F[20,80] = 5.88, P < .0001). The following individual items still differentiated PTSD from non-PTSD subjects: scared when thinking of death (p = .05), intrusive images of death (p = .008), more likely to startle (p = .002), staying away from reminders (p < .0001), wanting family around more often (p < .0001), and feeling that symptoms may have affected well-being (p < .0001).
Comparison of Peers with PTSD and Those with New-Onset Depression. Because of the high rate of comorbidity of PTSD and depression in this sample, those who developed PTSD were compared to 38 subjects who developed new-onset depression, but not PTSD. Those with PTSD were more likely to be female (100% versus 50%, p = .01) and older (20.1 [1.5] versus 17.4 [2.0] years, MWU = 35.0, P = .001). Also, those with PTSD were more likely to have had previous substance abuse (62.5% versus 26.3%, X2 = 3.94, P = .05), and past suicide attempts (37.5% versus 2.6%, X 2 = 10.1, P = .002). PTSD subjects were more likely to have had a family history of panic disorder (40% versus 2.7%, X2 = 9.24, P = .002). Prior to the year before the exposute to suicide, the PTSD group was more likely to have experienced the death of a friend (50.0% versus 13.2%, X2 = 5.70, P = .02) and disciplinary problems (50.0% versus 15.8%, X2 = 4.55, P = .03). No differences were noted in exposure to suicide, closeness of relationship, family environment, symptoms of grief, severity of depressive symptomatology, or life events in the year before exposure to the suicide. The scores on the PTSDRI were higher in the PTSD group both within a month of exposure (53.3 [10.3] versus 36.5 [10.4], MWU = 30.0, P = .001) and at the time of the interview (35.5 [9.5] versus 20.8 [7.6], MWU = 34.0, P = .001). The following items
differentiated subjects with PTSD from those with new-onset depression: feeling scared when thinking of the death (p = .04), intrusive images of the death (p = .004), being more likely to startle (p = .005), feeling guilt and regrets about the death (p = .03), staying away from reminders (p = .005), wanting family around more (p = .0003), and symptoms having affected wellbeing (p = .05). Logistic Regression. The following variables were entered into a logistic regression equation to predict multivariate risk for PTSD: age, gender, history of alcohol abuse, history of agoraphobia, past suicide attempt, close relationship with victim, current depression, discordant mother-child relationship, discordant parental marital relationship, and the life stressors noted above to differentiate PTSD and non-PTSD subjects. A logistic regression equation of three variables fit the data well (X 2 = 27.4, df = 29, P = .55): age> 19 (odds ratio [OR] = 4.9, 95% confidence interval [CI] 1.2 to 19.1), history of a suicide attempt (OR = 7.9, 95% CI = 2.1 to 29.9), and close relationship with victim (ARI ~ 66, OR = 3.0, 95% CI = 1.2 to 7.9). If PTSD was treated as a continuous outcome variable (i.e., the PTSDRI), then the following three variables explained 27.2% of the variance: current major depression (~ R2 = .15), ARI ~ 66 (~ R2 = .09), and history of a suicide attempt (~ R2 = .04).
DISCUSSION
In this paper, we have identified risk factors associated with the development of PTSD subsequent to exposure to the suicide of a friend. Most prominent among these risk factors are a close relationship with the victim, concomitant major depression, and a history of a suicide attempt. Family/environmental discord and disruption of important social bonds also were associated with the development ofPTSD. The clinical and research implications of these findings will be discussed after reviewing the strengths and weaknesses of this study. The strengths of this study include the representativeness of the sample. Moreover, information on domains likely to contribute to vulnerability to PTSD was obtained in a structured and reproducible manner. With respect to weaknesses, the main limitation is the small number of subjects who actually developed PTSD. Another limitation is that, because of the nature
]. AM. ACAD. CHIl.D ADOl.FSC. PSYCHIATRY. 34:2. FEBRUARY 1995
213
BRENT ET AI..
of the design, the "risk factors" were assessed after the development of PTSD, and hence reports by subjects and their parents may have been colored by this. Finally, the risk factors associated with the development of PTSD after the loss of a friend due to suicide may not be generalizable to adolescents whose traumatic experiences differ substantially from this. Despite these caveats, the risk factors associated with the development of PTSD are consistent with previous reports. A history of agoraphobia, suicidal behavior, and concurrent depression were associated with PTSD in this study. This is consonant with other reports associating PTSD with a personal and family history of anxiety disorders, a history of suicide attempts, and with concurrent depression (Breslau et aI., 1991; Davidson et aI., 1991). The association between PTSD and a history of suicide attempts may reflect either an independent association or one mediated by a history of depression and/or anxiety. The association of PTSD and female gender has not been previously reported to our knowledge. Males may be more vulnerable to be exposed to violence (Breslau et aI., 1991), and PTSD has been studied most frequently in samples of male veterans (Davidson et aI., 1991; Helzer et aI., 1992). While the prevalence of PTSD among females may have been skewed by the occurrence of one case in which several college roommates found the suicide victim hanging in her closet, the associations between female gender, depression, and anxiety, and between anxiety, depression, and PTSD may have accounted for female excess in this sample. The overlap of PTSD and depression in this sample deserves further comment. Of eight subjects with PTSD, five simultaneously developed a new-onset depression after losing a friend to suicide, and the remaining three subjects had previously been depressed. There could be several explanations for this association between PTSD and depression. First, PTSD and depression share several symptoms in common: social withdrawal, difficulty concentrating, anhedonia, sleep difficulties, and guilt. This makes the likelihood of fulfilling criteria for both disorders quite high. Second, as has been observed by Pynoos et al. (1987), the experience of trauma often co-occurs with that of loss. While trauma may predispose to PTSD, loss may predispose to depression. In this sample, a close relationship to the victim and witnessing the suicide or
seeing the body were associated with the development of both PTSD and depression (Brent et aI., 1993b). Third, other factors that may predispose to depression may also predispose to PTSD, such as a family history of psychiatric disorder, a history of previous losses, and previous personal psychiatric history (Brent et aI., 1993a,b; Breslau et aI., 1991; Helzer et aI., 1987). Nevertheless, not all subjects in this sample who became depressed also developed PTSD, suggesting additional risk factors that predispose to PTSD exert their influence independently or interactively with risk factors for depression. Risk factors that are common for both depression and PTSD include a close relationship to the victim, history of previous losses, and possibly witnessing the suicide or finding the body (Brent et aI., 1993b). However, certain factors predispose to PTSD and not depression. These factors include personal history of substance abuse and suicide attempts and family history of panic disorder, similar to previous findings in adults (Breslau et aI., 1991; Davidson et aI., 1991; Helzer et aI., 1987). Relatively few symptoms on the PTSDRI distinguish PTSD from non-PTSD subjects after statistically controlling for the presence of depression. These symptoms are related to hypervigilance, intrusive imagery, and a sense of isolation. Many symptoms that are part of the current PTSD nosology may be more closely tied to depression than PTSD, such as difficulty sleeping, difficulty concentrating, and anhedonia. However, the core symptomatology of PTSD may be much narrower than current diagnostic nomenclature dictates. Future studies using larger samples and signal detection methodology may be helpful in determining which symptoms are most salient to the diagnosis of PTSD (Kraemer, 1988). Life stressors and family/environmental influences also may affect the risk of PTSD. In the present study, mother-child discord was associated with risk for PTSD. Other studies have correlated PTSD with a history of abuse or an "irritable family atmosphere" (Davidson et aI., 1991; Green et al., 1991; Helzer et aI., 1987). Moreover, like previous studies, a history of parent-child disruption, and exposure to a previous loss was also associated with an increased risk of PTSD (Breslau et aI., 1991; Davidson et al., 1991; Helzer et al., 1987). It is also interesting to speculate about how early separation may somehow render certain individuals more vulnerable to the effects of trauma,
214
ACAD. CHILD ADOLESC. PSYCHIATRY, 34:2, FEBRUARY 1995
J. AM.
POSTTRAUMATIC STRESS DISORDER IN ADOLESCENTS
possibly through upregulation of the catecholamine neurotransmitter system. There is evidence that a history of sexual abuse may lead to long-lasting changes in the adrenal-cortical response to corticotropin-releasing hormone as well as to increased urinary catecholamine secretion (De Bellis et aI., 1994a,b). Taken together, the above-noted findings reinforce the validity of PTSD as an entity in youthful populations. Future research should focus on risk factors for the development of PTSD under different traumatic conditions (e.g., exposure to homicide), biological correlates of PTSD, the expectable longitudinal course of those with PTSD, and the proper treatment of this condition. The close relationship between both anxiety and depression and PTSD suggests that treatment with antidepressants may bring about symptomatic relief to those suffering from PTSD. Since prominent symptoms associated with PTSD relate to intrusive imagery about the death and guilt about survivorship, cognitive-behavioral treatments may be effective in relieving these disturbing manifestations of PTSD. Moreover, since family discord may predispose to the development of PTSD and the associated sense of isolation, family interventions that augment family support may be a critical aspect of successful treatment. In conclusion, PTSD is a relatively rare, but clinically significant, complication of exposure to suicide, particularly affecting closer friends of the victim with a history of a suicide attempt and who are currently depressed. Youth exposed to suicide, especially those fitting this profile, should be carefully screened for the presence of PTSD.
Brent DA, Perper JA, Moritz G et al. (l993a), Adolescent witness to a peer suicide, JAm Acad Child Adolesc Psychiatry 32: 1184-1188 Brent DA, Perper JA, Moritz G ct al. (I 993b), Psychiatric sequelae to the loss of an adolescent peer to suicide. J Am Acad Child Adolesc
Psychiatry 32:509-517 Breslau N, Davis GC, Andreski P, Peterson E (1991), Traumatic events and post-traumatic stress disorder in an urban population of young adults. Arch Gen Psychiatry 48:216-222 Chambers WJ, Puig-Anrich J, Hirsch M er al. (1985), The assessment of affective disorders in children and adolescents by semi-structured interview: test-retest reliability of the Schedule for Affective Disorders and Schizophrenia for School-Age Children, Present Episode version. Arch Gen Psychiatry 42:696-702 Davidson JRT, Hughes D, Blazer DG, George LK (1991), Post-traumatic stress disorder in the community: an epidemiological study. Psycbol
Med21:713-721 De Bellis MD, Chrousos Gl', Darn LD cr al. (I 994a), Hypothalamicpituitary-adrenal axis dysregulation in sexually abused girls. ] Clin Endocrinol Metab 78:249-255 Dc Bellis MD, Leiter I., Trickett pK, Putnam FW Jr (l994b), Urinary catecholamine excretion in sexually abused girls. ] Am Acnd Child Adolesc Psychiatry 33:320-327 Famularo R, Fenton T, Kinscherff R (1993), Child maltreatment and the development of post-traumatic stress disorder. Am] Dis Child 147:755-760 Faschingbauer TR, Devaul RA, Zisook S (1977), Development of the Texas Grief Inventory. Am] Psychiatly 134:696-698 Goldberg J, True WR, Eisen SA, Henderson WG (1990), A twin study of the effects of the Vietnam War on post-traumatic stress disorder.
]AMA 263:1227-1232 Green BL, Korol M, Grace MC er al. (1991), Children and disaster: age, gender, and parental effects on pTSD symptoms. ] Am Acad Child
Adolesc Psychiatry 30:945-951 Helzer JE, Robins LN, McEvoy I. (1987), Post-traumatic stress disorder in the general population: findings of the epidemiologic catchment area survey. N Engl ] Med 317:1630-1634 Kinzie JD, Sack WH, Angell RH, Manson S, Rarh B (1986), The psychiatric effects of massive trauma on Cambodian children. I. The children. ]
Am Acad Child Psychiatry 25:370-376 Kiser LJ, Ackerman BJ, Brown E ct al, (1988), Post-traumatic stress disorder in young children: a reaction to purported sexual abuse. J Am Acad
Child Adolesc Psychiatry 27:645-649 Kiser LJ, Heston J, Millsap pA, Pruitt DB (1991), Physical and sexual abuse in childhood: relationship with post-traumatic stress disorder. ]
Am Acad Child Adolesc Psychiatry 30:776-783 Kraemer HC (1988), Assessment of 2 X 2 associations: generalization of signal detection methodology. Am Stat 42:37-49 Lukens E, Puig-Anrich J, Behn J, Goetz R, 'Fabrizi M, Davies M (1983), Reliability of the psychosocial schedule for school age children. ] Am
Acad Child Psychiatry 22:29-39 Mcl.cer SV, Deblinger E, Atkins MC, Faa EB, Ralphe DL (1988), Posttraumatic stress disorder in sexually abused children. ] Am Acad Child
REFERENCES Andreasen NC. Endicott J, Spitzer RL, Winokur G (1977), The family history method using diagnosric criteria: reliability and validity. Arch Gen Psychiatry 34: 1229-1235 Brent D, Pcrpcc ], Coldsrciu CE ct al. (1988), Risk factors (or adolescent suicide: a comparison of adolescent suicide victims with suicidal inpatients. Arch Gen Psychiatry 45:581-588 Brent DA, Pcrpcr JA, Moritz G cr al. (1992), Psychiatric effects of exposure to suicide among the friends and acquaintances of adolescent suicide victims. JAm Acad Child Adolesc Psychiatry 31 :629-640
]. AM. ACAD. CHIl.D ADOl.ESc:. PSYCHIATRY, 34:2, rEBRUARY 1995
Adolesc Psychiatry 27:650-654 Orvaschel H, Puig-Antich J, Chambers W, Tabrizi MA, Johnson R (1982), Retrospective assessment of prepubertal major depression with the Kiddie-SADS-E. JAm Acad Child Psychiatry 21 :392-397 Pynoos RS, Frederick C, Nader K et al. (1987), Life threat and posttraumatic stress in school-age children. Arch Gen Psychiatry 44:1057-1063 Schwartz ED, Kowalski 1M (1991), Post-traumatic stress disorder after a school shooting: effects of symptom threshold selection and diagnosis by DSM-III-R, or proposed DSM-IV Am J Psychiatry 148:592-597 Ten LC (1991), Childhood traumas: an outline and overview. Am]
Psychiatry 148:10-20
215
ABSTRACT Objective: To examine the factors predisposing to posttraumatic stress disorder (PTSD) in peers of adolescent suicide victims. Method: One hundred forty-six adolescents who were the friends of 26 suicide victims were studied. Five percent (n
=8) developed PTSD after exposure to suicide. These 8 subjects with PTSD were compared to the remainder
of the exposed subjects (n = 138). Results: Subjects with PTSD were more likely than those without PTSD to have had a history of substance abuse, agoraphobia, and suicide attempts. Subjects who developed PTSD were more likely to have developed a new-onset depression, to have more severe grief, and to have been closer to the suicide victim. Subjects with PTSD tended to have more severe exposure to suicide and came from discordant households with a history of disruptions in key relationships. The 8 subjects who developed PTSD were compared to 38 subjects who developed new-onset depression but not PTSD. Those with PTSD were more likely to have had past substance abuse, prior suicide attempts, family history of panic disorder, a history of parent-child disruption, and a history of loss. Symptoms of intrusive visual images, hypervigilance, and avoidance of reminders discriminated subjects who had PTSD from new-onset depressives without PTSD. Conclusions: PTSD is an expectable outcome in youth exposed to suicide. Further work is required to differentiate symptoms of depression from PTSD. J. Am. Acad. Child Ado/esc. Psychiatry, 1995, 34, 2:209-215. Key Words: adolescent, posttraumatic stress disorder, depression, suicide.
Posttraumatic stress disorder (PTSD) is now widely recognized as a response to intense overwhelming physicalor psychological trauma. Initial work by Terr (1991) and Pynoos et al. (1987) documented carefully that PTSD does occur in younger children and adolescents and that the phenomenology of the disorder does not differ greatly from the manifestation ofPTSD in adults. PTSD has been reported to occur in youthful Cambodian refugees who survived a concentration camp (Kinzie et aI., 1986), among witnesses and survivors of a homicide or other traumatic deaths (Green et aI., 1991;
AcceptedJanuary 19, 1994. Dr. Perper was with the Allegheny County Coroner's Office. He is now Medical Examiner, Brou/ard, FL. All other authors are with the Western Psychiatric Institute and Clinic, Piwburgh, PA. This work W(/S supported by NIMH grant MH44111, "Youth Exposed to Suicide." The authors thank Karen Rhinaman fOr her assistance in preparing the manuscript. Reprint requests to Dr. Brent, Western Psychiatric Institute and Clinic, 3811 O'Hartt Street, Pittsburgh, PA 15213. 0890-8567/95/3402-0209$03.00/0©I995 by the American Academy of Child and Adolescent Psychiatry.
Pynoos et al., 1987; Schwartz and Kowalski, 199 I), and in response to physical or sexual abuse (Famularo et al., 1993; Kiser et aI., 1988, 1991; McLeer et al., 1988). These reports have contributed important data to the validation of the diagnosis of PTSD. However, little is known about what factors may predispose to the development of PTSD in younger populations, other than the severity of life threat experienced and the closeness in relationship to the victim (Pynoos et al., 1987). In a recent case-control report of adolescents who witnessed a suicide on a school bus, the development of anxiety disorders, including PTSD, was predicted by a family history of suicide attempt and affective illness, and higher number of stressful life events in the previous year (Brent et aI., I993a). However, the number of exposed adolescents was relatively small (n = 28), and only four of this group developed PTSD, making it impossible to draw firm conclusions about risk factors for PTSD. In adults, as in younger populations, exposure to
]. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 34:2, FEBRUARY 1995
209
BRENT ET AI..
life-threatening situations is among the most salient predictor of the development of PTSD (Breslau et al., 1991; Goldberg et al., 1990; Green et al., 1991; Helzer et al., 1987). Moreover, while antisocial symptoms and substance abuse are associated with risk for exposure to traumatic violence (Breslau et al., 1991; Helzer et al., 1987), it has recently been shown that both family and personal history of anxiety disorder were most predictive of the development of PTSD (Breslau et al., 1991). Other factors, such as an "irritable family atmosphere" may also contribute to risk for PTSD (Green et al., 1991). In our ongoing study of the impact of exposure to suicide in the friends and acquaintances of suicide victims, we have reported an increased risk of PTSD subsequent to exposure to suicide (Brent et al., 1993b). This nonreferred sample, for whom characteristics of the exposure to suicide, previous psychiatric history, family history, and previous stressors were gathered, presents an opportunity to examine risk factors for the development of PTSD. Therefore, we compared those youth who developed PTSD to those who did not in the above-noted domains. Since a number of studies have noted a cooccurrence of PTSD, grief responses, and depression (Brent et al., 1993a; Davidson et al., 1991; Pynoos et al., 1987), we also attempted to identify risk factors and symptomatic components of PTSD that were independent from the potentially overlapping constructs of depression and grief.
METHOD Sample The sample consists of 146 adolescent friends and acquaintances of 26 adolescent suicide victims described previously (Brent et al., 1993b). The suicide victims were from 28 counties in western Pennsylvania. The families of 78% of all consecutive adolescent suicides agreed to a psychological autopsy, but only 67% provided a social network. The characteristics of suicide victims whose families agreed to the study were not different from the characteristics of those who refused with respect to age, race, gender, toxicology, or county of origin. However, acceptors were more often families of suicide victims who used a gun (80.8% versus 44.0%, X2 ~ 4.32, P ~ .04). Friends of the suicide victims were nominated by parents and siblings as being close to the victim and having had a mutually confiding relationship. Two friends nominated by both a parent and a sibling were chosen at random. Each of these two friends were asked to nominate other close friends, and of these peernominated friends, one additional friend was chosen. Acquaintances
210
J.
were chosen by asking the friends to nominate social contacts of theirs who knew the suicide victim, but not as well as the friends themselves did. Three acquaintances were chosen at random from this list. The rate of agreement for interview for friends and acquaintances was 61%. Refusers were younger than acceptors by an average of 10 months (p ~ .01) and were more likely to be male (76% versus 54%, X2 ~ 11.3, P ~ .0008). Exposed subjects were interviewed a median of 7 months after the death (mean ~ 7.3 months, SD ~ 1.7 months). The median network size was six (range two to seven). In a previous report (Brent et al., 1993b), we noted that the rate of new-onset PTSD was 8 (5%) of 146, compared with none in an equal number of nonexposed community controls (p < .05). Subsequent analyses presented herein will compare these 8 subjects with PTSD to the remaining 138 exposed subjects without the syndrome of PTSD.
Assessment Psychiatric assessment was made by use of the Schedule for Affective Disorders and Schizophrenia for School-Age Children, Epidemiologic and Present Episode versions (Chambers et al., 1985; Orvaschel er al., 1982). Interrater reliability for diagnoses (K ~ .98, SD ~ .13) and for symptoms (intraclass correlation coefficient ~ .96, SD ~ .7) was high. Symptoms of PTSD were assessed by the Post-Traumatic Stress Disorder Reaction Index (PTSDRI) (Pynoos et al., 1987). This is a 20-item inventory of PTSD symptomatology with good internal consistency and discriminant and convergent validity (Pynoos et al., 1987). Each item is rated on a Likert scale from I to 5, with 5 being the most problematic. Symptoms were rated both retrospectively at I month after exposure to the suicide, and at the time of the interview, about 7 months after the death. For this study, internal consistency of the PTSDRI was high (a ~ .83 to .88). Life events and family history were assessed by an interviewer blind to the proband's diagnostic status. Life events were assessed by a 29-item life events inventory with acceptable face and discriminant
validity (Brent et al., 1988, 1993b). Family history was assessed using a Family History-RDC with diagnoses modified to DSMIII (Andreasen et al., 1977). The closeness to the victim was assessed by the Adolescent Relationship Inventory (ARI) (Brent er al., 1992, 1993b), which covers social exchange, provision of social support, mutual admiration, and frequency of contact. In previous studies, we have shown high internal consistency and ability to discriminate friends from acquaintances with this measure. The Circumstances of Exposure to Death instrument was used to assess the degree of exposure to suicide (Brent et al., 1992, 1993b). This instrument covers circumstances of the death, direct exposure (if the suicide was witnessed), indirect exposure (discovering of the body, visiting the scene of death, etc.), and events following the death (wake, funeral, visitation). Interrater reliability with this instrument has been reported as quite high. Grief was measured by the Texas Grief Inventory (TGI) (Faschingbauer, 1977), a self-report questionnaire filled out by the exposed adolescents. Psychosocial functioning in school, with peers, siblings, and with parents was assessed by use of the Psychosocial Schedule (Lukens et al., 1983).
Data Analyses The subjects who developed PTSD were compared to those who did not using standard parametric and nonparametric statistics. For categorical comparisons, a Pearson's X2 or Fisher's Exact Test
AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 34:2, FEBRUARY 1995
POSTTRAUMATIC STRESS DISORDER IN ADOLESCENTS
TABLE 1
Demographic Characteristics. The peers exposed to suicide who developed PTSD were more likely to be female (100% versus 42.8%, Fisher's Exact Test, p = .002), and older (20.1 [1.5J versus 18.2 [2.0J years, MWU = 248.5, P = .01) (Table 1). This demographic difference may have emerged in part because the most direct exposure to suicide in this study occurred among the female roommates of an older adolescent female who hung herself and was discovered hanging in her
Family History. There were no differences in the rates of disorder between the first-degree relatives of the PTSD and non-PTSD subjects, with respect to affective illness (14.0% versus 10.1 %), anxiety disorder (8.1% versus 5.8%), substance abuse (14.1% versus 12.1 %), conduct disorder (3.1 % versus 3.3%), or suicidal behavior (7.2% versus 3.9%). Exposure to the Suicide. PTSD subjects tended to be more likely to have seen the scene of death, witnessed the suicide, or discovered the body of the victim than were peers without PTSD (57.1 % versus 27.3%, X2 = 2.89, P = .09), to have known the victim's plans before death (42.9% versus 22.6%), to have believed the suicide was preventable (50.0% versus 17.1 %, X2 = 5.27, P = .02), and to have had a conversation with the victim on the day of the suicide (75% versus 44.9%, not significant). The total number of these exposure variables tended to be greater in those peers who developed PTSD (2.6 [2.0J versus 1.5 [1.5J, MWU = 367, P = .06). Distribution by Network. The PTSD cases were not randomly distributed by network (X 2 = 39.5, df = 25, P = .03). Three of the eight cases occurred in one network. In this network, exposed roommates of the suicide victim found the victim hanging in a closet. As reported previously, there was no association between network membership and new-onset depression (Brent et al., 1993b). Closeness to the Victim. Subjects who developed PTSD had a closer relationship to the suicide victim than did those without PTSD, as measured by the ARI (67.3 [33.6] versus 45.4 [25.2], MWU = 328.8, P = .02). Family Environment. As measured by the Psychosocial Schedule, the mother-subject relationship was more discordant in the PTSD group (17.5 [0.7] versus 12.0 [3.9], t = 1.99, df = 110, P = .05), as was the marital relationship between parents (33.0 [O.OJ versus 20.7 [5.9], MWU = 5.0, P = .04). Father-child and overall parent-child relations were in the same direction
closet.
as the ones noted above, although the differences
Past Psychiatric Disorder. Peers with PTSD were more likely to have had a history of substance abuse (62.5% versus 28.3%, X2 = 4.21, P = .04), agoraphobia (12.5% versus 0.7%, X2 = 7.76, P = .005), and suicide attempts (37.5% versus 2.9%, X2 = 19.8, P < .0001) (Table 2).
escaped statistical significance. Life Events. In the year before the assessment, subjects with PTSD were more likely than those without PTSD to have had interpersonal conflict with siblings (25.0% versus 5.9%, X2 = 4.27, P = .04) or with boyfriends or girlfriends (50.0% versus 19.1%, X2 = 4.37, P = .04), as well as to have experienced the death of a
Demographic Characteristics PTSO
Sex (% females) Age (yr) Mean SO Race (% white) SES (%) I II III IV V Lives with both biological parents (%) Note: PTSO nomic status.
=
No (n = 138)
Yes (n = 8)
p
42.8
100.0
.002
18.2 2.0 97.1
20.1 1.5 100.0
.01
5.8 27.7 32.8 28.5 5.1
0.0 25.0 50.0 12.5 12.5
48.9
25.0
posttraumatic stress disorder; SES
.001 =
socioeco-
was used. For comparisons on continuous data, either a t test or Mann-Whitney U (MWU) was used. To further differentiate depression and PTSD, subjects who developed PTSD were compared to those who developed depression, but not PTSO, using standard univariate statistics. Logistic regression was used to select those variables that were most closely associated with the diagnosis of PTSO. Multiple linear regression was used to examine the relationship of these variables to the symptomatic expression of PTSO. Multivariate analyses of covariance were used to examine the discriminant validity of specific items for the PTSORI while controlling for depression and for grief.
RESULTS
J. AM.
ACAD. CHILD ADOLESC. PSYCHIATRY. 34:2. FEBRUARY 1995
211
BRENT ET AI..
TABLE 2 Sym p to ms of PTS D on th e PT SD RI in Those w ith and witho ut DSM-III-R PT SD O ne Mo nt h PT SD
No
Yes (n = 8)
= 138)
Symp toms
(n
Scared whe n think o f death Im ages of death (volun ta ry) Im ages of death (in trusive) D reams/ n igh tmares related to th e death Fear that others or self mi ght d ie Difficulty enjoying thi ngs O ther people don 't und erstand C ur off from own feelings D ifficulty explaining feelings M ore likely to be startled Problems sleepi ng G uilt/ rcgrers abou t de at h Int erfering w ith memOlY Interfer ing wi th con centrat ion Staying away fro m reminde rs Rem ind ers creat e tensio n or stress W ant fami ly aro u nd more often Affected well-bein g More impulsive or diffi cult
1.4 2.1 1.4 0.7 0.7 1.4 1.4 0.8 2.3 0.7 2.8 1.6 0.5 2.0 1.2 2.0 0.3 0.6 0.2
At Ti me of Int erview PTSD
(1.4) ( 1.4) (1.3) (1.2) ( J.1) (1.5) (1.6) (1.4) (1.7) (1.3) (1.5) (1.6) (1.0) (1.8) (1.5) ( 1.6) (0.8) ( \.2) (0.7)
3.3 3.6 3.0 1.8 1.8 3.6 3.4 2.0 2.0 2.6 0.5 2.4 0.8 1.5 3.4 3.8 2.9 2.9 0.4
( 1.5)** (0.7)** (J .4)** (1.9) (1.6)* (0.7)*** (J .4)** (2.1) (1.9) ( 1.7)*** (0.8)*** ( 1.8) (J.4) ( 1.9) ( 1.4)*'(0.5)** (1.8)**(2.0)*** (0.7)
No ( n = 138)
0.7 0.8 0.5 0.2 0.5 \.2 0.9 0.4 2.5 0.4 3.6 0.7 0.1 2.4 0.6 1.0 0.1 0.1 0.1
(1.0) ( 1.2) (0.8) (0.6) (0.9) (1.6) (1.4) (0.9) (1.6) (1.0) ( 1.1) ( 1.1) (0.4) (1.8) ( 1.0) ( 1.2) (0.5) (0.4) (0.5)
Yes (n = 8)
1.8 2.0 1.9 0.8 1.0 1.3 2.8 0.8 3.0 2.5 2.6 1.9 0.3 1.9 2.8 2.4 2.0 0.9 0.3
( 1.6)* ( J.1)** (1.6)* * (\.2)* ( 1.2) (1.4) (J .5)** ( 1.4) (1.4) ( 1.7)*** (1.4)* * ( 1.5)* (0.5) ( 1.6) (1.8)-** (J .4)** ( 1.9)*** ( 1.5)*' (0.7)
N ote: Va lues represent mean (SD) . PT SD = posttra umatic stress d isorder; PTSDRI = Post-Traumati c Stress D isord er Reaction Ind ex. < .0 1; ' ''p < .00 1 (after adj ustment for cu rrent depression).
"P < .05; -'p
nonparental relative (50.0% versus 19.9%, X2 = 4.07, P = .04). For life events that ant edated the year before diagnostic assessment , subjects with PTSD had higher rates of parent-child disrup tions (62.5% versus 25.7% , X2 = 5.09, P = .02), sibling deaths (25.0% versus 3.7%, X2 = 7.43, P = .006), and death of a friend oth er th an the suicide victim (50% versus 19.4%, X2 = 4.07, P = .04). T here was no difference between the two groups as to the total number of life events experienced. Current Psychiatric Symptomatology. Subjects with PT SD were more likely than th ose without PT SD to have a concurrent diagnosis of major dep ression (62.5% versus 23 .9%, Xl = 5.85, P = .02). All five subjects with PTSD and concurre nt depression developed major depression simultaneously with onset of PT SD. For the thre e others with PT SD , a preexistin g depression antedated the PTSD. Subjects with PTSD had significantly more symptoms of depression than the non PTSD group (8.6 [2.0J versus 3.9 [3.4], MWU = 148.5, P = .0001 ). Subjects with PTSD had significan tly more severe ratin gs of sleep disturbance (p = .01) , guilt (p = .002), worthlessness (p = .007), somatic compl aint s (p = .005), and suicidality (p = .006).
212
J.
PTSD Symptomatology. At 1 month after the death in the index suicide victim, subjects with PT SD scored much highe r on th e PT SDRI than the non-PTSD peers (53.3 [10.3J versus 26.0 [12.9], t = 5.86, df = 131, P < .000 1) (Table 2). At the time of th e int erview, PTSDRI scores were still higher in subjects with PTSD (35. 5 [9.5J versus 15.8 [7.7J, t = 6.92, df = 138, P < .0001). There was a fall-off in severity of symptomatology in both group s, altho ugh the fall-off was more rapid in the PTSD group (group X time interaction, F [ I ,13 l] = 4.96, P = .03). Subjects with PT SD also showed more grief scored on the TGI (42.3 [11.4J versus 63.0 [17.0], t = 2.42, df = 115, P = .02). Symp toms of PTSD were highly correlated with both depression (r = .63, P < .001 ) and grief ( r = - .47, P < .00 1). At 1 month, 12 of 20 and at the time of inte rview, 10 of 20 symptoms of PTSD were more severe in subjects d iagnosed with PTSD compared to those who were not. These persistent sympto ms that differenti ated the PTSD and non -PTSD groups most shar ply were both voluntary and intrusive images of th e death , a feeling that other people did not understand the
AM . AC A D . C H IL D AD OL ES C. PSYCH IATR Y. 34 :2 . F E BRUAR Y 1995
POSTTRAUMATIC STRESS DISORDER IN ADOLESCENTS
experience, hypervigilance and an exaggerated startle, and avoidance of reminders of the death. Two other symptoms, guilt and regrets about the death, and difficulty sleeping also differentiated the two groups, but may represent general symptoms of grief and depression rather than specific symptoms of PTSD. After adjustment for severity of depressive symptomatology using analysis of covariance, all 10 symptoms noted on Table 2 still differentiated subjects with PTSD from those without PTSD. After adjusting for differences in the severity of grief, using the TGI, the PTSD group still scored higher on the PTSDRI than did the non-PTSD group (F[20,80] = 5.88, P < .0001). The following individual items still differentiated PTSD from non-PTSD subjects: scared when thinking of death (p = .05), intrusive images of death (p = .008), more likely to startle (p = .002), staying away from reminders (p < .0001), wanting family around more often (p < .0001), and feeling that symptoms may have affected well-being (p < .0001).
Comparison of Peers with PTSD and Those with New-Onset Depression. Because of the high rate of comorbidity of PTSD and depression in this sample, those who developed PTSD were compared to 38 subjects who developed new-onset depression, but not PTSD. Those with PTSD were more likely to be female (100% versus 50%, p = .01) and older (20.1 [1.5] versus 17.4 [2.0] years, MWU = 35.0, P = .001). Also, those with PTSD were more likely to have had previous substance abuse (62.5% versus 26.3%, X2 = 3.94, P = .05), and past suicide attempts (37.5% versus 2.6%, X 2 = 10.1, P = .002). PTSD subjects were more likely to have had a family history of panic disorder (40% versus 2.7%, X2 = 9.24, P = .002). Prior to the year before the exposute to suicide, the PTSD group was more likely to have experienced the death of a friend (50.0% versus 13.2%, X2 = 5.70, P = .02) and disciplinary problems (50.0% versus 15.8%, X2 = 4.55, P = .03). No differences were noted in exposure to suicide, closeness of relationship, family environment, symptoms of grief, severity of depressive symptomatology, or life events in the year before exposure to the suicide. The scores on the PTSDRI were higher in the PTSD group both within a month of exposure (53.3 [10.3] versus 36.5 [10.4], MWU = 30.0, P = .001) and at the time of the interview (35.5 [9.5] versus 20.8 [7.6], MWU = 34.0, P = .001). The following items
differentiated subjects with PTSD from those with new-onset depression: feeling scared when thinking of the death (p = .04), intrusive images of the death (p = .004), being more likely to startle (p = .005), feeling guilt and regrets about the death (p = .03), staying away from reminders (p = .005), wanting family around more (p = .0003), and symptoms having affected wellbeing (p = .05). Logistic Regression. The following variables were entered into a logistic regression equation to predict multivariate risk for PTSD: age, gender, history of alcohol abuse, history of agoraphobia, past suicide attempt, close relationship with victim, current depression, discordant mother-child relationship, discordant parental marital relationship, and the life stressors noted above to differentiate PTSD and non-PTSD subjects. A logistic regression equation of three variables fit the data well (X 2 = 27.4, df = 29, P = .55): age> 19 (odds ratio [OR] = 4.9, 95% confidence interval [CI] 1.2 to 19.1), history of a suicide attempt (OR = 7.9, 95% CI = 2.1 to 29.9), and close relationship with victim (ARI ~ 66, OR = 3.0, 95% CI = 1.2 to 7.9). If PTSD was treated as a continuous outcome variable (i.e., the PTSDRI), then the following three variables explained 27.2% of the variance: current major depression (~ R2 = .15), ARI ~ 66 (~ R2 = .09), and history of a suicide attempt (~ R2 = .04).
DISCUSSION
In this paper, we have identified risk factors associated with the development of PTSD subsequent to exposure to the suicide of a friend. Most prominent among these risk factors are a close relationship with the victim, concomitant major depression, and a history of a suicide attempt. Family/environmental discord and disruption of important social bonds also were associated with the development ofPTSD. The clinical and research implications of these findings will be discussed after reviewing the strengths and weaknesses of this study. The strengths of this study include the representativeness of the sample. Moreover, information on domains likely to contribute to vulnerability to PTSD was obtained in a structured and reproducible manner. With respect to weaknesses, the main limitation is the small number of subjects who actually developed PTSD. Another limitation is that, because of the nature
]. AM. ACAD. CHIl.D ADOl.FSC. PSYCHIATRY. 34:2. FEBRUARY 1995
213
BRENT ET AI..
of the design, the "risk factors" were assessed after the development of PTSD, and hence reports by subjects and their parents may have been colored by this. Finally, the risk factors associated with the development of PTSD after the loss of a friend due to suicide may not be generalizable to adolescents whose traumatic experiences differ substantially from this. Despite these caveats, the risk factors associated with the development of PTSD are consistent with previous reports. A history of agoraphobia, suicidal behavior, and concurrent depression were associated with PTSD in this study. This is consonant with other reports associating PTSD with a personal and family history of anxiety disorders, a history of suicide attempts, and with concurrent depression (Breslau et aI., 1991; Davidson et aI., 1991). The association between PTSD and a history of suicide attempts may reflect either an independent association or one mediated by a history of depression and/or anxiety. The association of PTSD and female gender has not been previously reported to our knowledge. Males may be more vulnerable to be exposed to violence (Breslau et aI., 1991), and PTSD has been studied most frequently in samples of male veterans (Davidson et aI., 1991; Helzer et aI., 1992). While the prevalence of PTSD among females may have been skewed by the occurrence of one case in which several college roommates found the suicide victim hanging in her closet, the associations between female gender, depression, and anxiety, and between anxiety, depression, and PTSD may have accounted for female excess in this sample. The overlap of PTSD and depression in this sample deserves further comment. Of eight subjects with PTSD, five simultaneously developed a new-onset depression after losing a friend to suicide, and the remaining three subjects had previously been depressed. There could be several explanations for this association between PTSD and depression. First, PTSD and depression share several symptoms in common: social withdrawal, difficulty concentrating, anhedonia, sleep difficulties, and guilt. This makes the likelihood of fulfilling criteria for both disorders quite high. Second, as has been observed by Pynoos et al. (1987), the experience of trauma often co-occurs with that of loss. While trauma may predispose to PTSD, loss may predispose to depression. In this sample, a close relationship to the victim and witnessing the suicide or
seeing the body were associated with the development of both PTSD and depression (Brent et aI., 1993b). Third, other factors that may predispose to depression may also predispose to PTSD, such as a family history of psychiatric disorder, a history of previous losses, and previous personal psychiatric history (Brent et aI., 1993a,b; Breslau et aI., 1991; Helzer et aI., 1987). Nevertheless, not all subjects in this sample who became depressed also developed PTSD, suggesting additional risk factors that predispose to PTSD exert their influence independently or interactively with risk factors for depression. Risk factors that are common for both depression and PTSD include a close relationship to the victim, history of previous losses, and possibly witnessing the suicide or finding the body (Brent et aI., 1993b). However, certain factors predispose to PTSD and not depression. These factors include personal history of substance abuse and suicide attempts and family history of panic disorder, similar to previous findings in adults (Breslau et aI., 1991; Davidson et aI., 1991; Helzer et aI., 1987). Relatively few symptoms on the PTSDRI distinguish PTSD from non-PTSD subjects after statistically controlling for the presence of depression. These symptoms are related to hypervigilance, intrusive imagery, and a sense of isolation. Many symptoms that are part of the current PTSD nosology may be more closely tied to depression than PTSD, such as difficulty sleeping, difficulty concentrating, and anhedonia. However, the core symptomatology of PTSD may be much narrower than current diagnostic nomenclature dictates. Future studies using larger samples and signal detection methodology may be helpful in determining which symptoms are most salient to the diagnosis of PTSD (Kraemer, 1988). Life stressors and family/environmental influences also may affect the risk of PTSD. In the present study, mother-child discord was associated with risk for PTSD. Other studies have correlated PTSD with a history of abuse or an "irritable family atmosphere" (Davidson et aI., 1991; Green et al., 1991; Helzer et aI., 1987). Moreover, like previous studies, a history of parent-child disruption, and exposure to a previous loss was also associated with an increased risk of PTSD (Breslau et aI., 1991; Davidson et al., 1991; Helzer et al., 1987). It is also interesting to speculate about how early separation may somehow render certain individuals more vulnerable to the effects of trauma,
214
ACAD. CHILD ADOLESC. PSYCHIATRY, 34:2, FEBRUARY 1995
J. AM.
POSTTRAUMATIC STRESS DISORDER IN ADOLESCENTS
possibly through upregulation of the catecholamine neurotransmitter system. There is evidence that a history of sexual abuse may lead to long-lasting changes in the adrenal-cortical response to corticotropin-releasing hormone as well as to increased urinary catecholamine secretion (De Bellis et aI., 1994a,b). Taken together, the above-noted findings reinforce the validity of PTSD as an entity in youthful populations. Future research should focus on risk factors for the development of PTSD under different traumatic conditions (e.g., exposure to homicide), biological correlates of PTSD, the expectable longitudinal course of those with PTSD, and the proper treatment of this condition. The close relationship between both anxiety and depression and PTSD suggests that treatment with antidepressants may bring about symptomatic relief to those suffering from PTSD. Since prominent symptoms associated with PTSD relate to intrusive imagery about the death and guilt about survivorship, cognitive-behavioral treatments may be effective in relieving these disturbing manifestations of PTSD. Moreover, since family discord may predispose to the development of PTSD and the associated sense of isolation, family interventions that augment family support may be a critical aspect of successful treatment. In conclusion, PTSD is a relatively rare, but clinically significant, complication of exposure to suicide, particularly affecting closer friends of the victim with a history of a suicide attempt and who are currently depressed. Youth exposed to suicide, especially those fitting this profile, should be carefully screened for the presence of PTSD.
Brent DA, Perper JA, Moritz G et al. (l993a), Adolescent witness to a peer suicide, JAm Acad Child Adolesc Psychiatry 32: 1184-1188 Brent DA, Perper JA, Moritz G ct al. (I 993b), Psychiatric sequelae to the loss of an adolescent peer to suicide. J Am Acad Child Adolesc
Psychiatry 32:509-517 Breslau N, Davis GC, Andreski P, Peterson E (1991), Traumatic events and post-traumatic stress disorder in an urban population of young adults. Arch Gen Psychiatry 48:216-222 Chambers WJ, Puig-Anrich J, Hirsch M er al. (1985), The assessment of affective disorders in children and adolescents by semi-structured interview: test-retest reliability of the Schedule for Affective Disorders and Schizophrenia for School-Age Children, Present Episode version. Arch Gen Psychiatry 42:696-702 Davidson JRT, Hughes D, Blazer DG, George LK (1991), Post-traumatic stress disorder in the community: an epidemiological study. Psycbol
Med21:713-721 De Bellis MD, Chrousos Gl', Darn LD cr al. (I 994a), Hypothalamicpituitary-adrenal axis dysregulation in sexually abused girls. ] Clin Endocrinol Metab 78:249-255 Dc Bellis MD, Leiter I., Trickett pK, Putnam FW Jr (l994b), Urinary catecholamine excretion in sexually abused girls. ] Am Acnd Child Adolesc Psychiatry 33:320-327 Famularo R, Fenton T, Kinscherff R (1993), Child maltreatment and the development of post-traumatic stress disorder. Am] Dis Child 147:755-760 Faschingbauer TR, Devaul RA, Zisook S (1977), Development of the Texas Grief Inventory. Am] Psychiatly 134:696-698 Goldberg J, True WR, Eisen SA, Henderson WG (1990), A twin study of the effects of the Vietnam War on post-traumatic stress disorder.
]AMA 263:1227-1232 Green BL, Korol M, Grace MC er al. (1991), Children and disaster: age, gender, and parental effects on pTSD symptoms. ] Am Acad Child
Adolesc Psychiatry 30:945-951 Helzer JE, Robins LN, McEvoy I. (1987), Post-traumatic stress disorder in the general population: findings of the epidemiologic catchment area survey. N Engl ] Med 317:1630-1634 Kinzie JD, Sack WH, Angell RH, Manson S, Rarh B (1986), The psychiatric effects of massive trauma on Cambodian children. I. The children. ]
Am Acad Child Psychiatry 25:370-376 Kiser LJ, Ackerman BJ, Brown E ct al, (1988), Post-traumatic stress disorder in young children: a reaction to purported sexual abuse. J Am Acad
Child Adolesc Psychiatry 27:645-649 Kiser LJ, Heston J, Millsap pA, Pruitt DB (1991), Physical and sexual abuse in childhood: relationship with post-traumatic stress disorder. ]
Am Acad Child Adolesc Psychiatry 30:776-783 Kraemer HC (1988), Assessment of 2 X 2 associations: generalization of signal detection methodology. Am Stat 42:37-49 Lukens E, Puig-Anrich J, Behn J, Goetz R, 'Fabrizi M, Davies M (1983), Reliability of the psychosocial schedule for school age children. ] Am
Acad Child Psychiatry 22:29-39 Mcl.cer SV, Deblinger E, Atkins MC, Faa EB, Ralphe DL (1988), Posttraumatic stress disorder in sexually abused children. ] Am Acad Child
REFERENCES Andreasen NC. Endicott J, Spitzer RL, Winokur G (1977), The family history method using diagnosric criteria: reliability and validity. Arch Gen Psychiatry 34: 1229-1235 Brent D, Pcrpcc ], Coldsrciu CE ct al. (1988), Risk factors (or adolescent suicide: a comparison of adolescent suicide victims with suicidal inpatients. Arch Gen Psychiatry 45:581-588 Brent DA, Pcrpcr JA, Moritz G cr al. (1992), Psychiatric effects of exposure to suicide among the friends and acquaintances of adolescent suicide victims. JAm Acad Child Adolesc Psychiatry 31 :629-640
]. AM. ACAD. CHIl.D ADOl.ESc:. PSYCHIATRY, 34:2, rEBRUARY 1995
Adolesc Psychiatry 27:650-654 Orvaschel H, Puig-Antich J, Chambers W, Tabrizi MA, Johnson R (1982), Retrospective assessment of prepubertal major depression with the Kiddie-SADS-E. JAm Acad Child Psychiatry 21 :392-397 Pynoos RS, Frederick C, Nader K et al. (1987), Life threat and posttraumatic stress in school-age children. Arch Gen Psychiatry 44:1057-1063 Schwartz ED, Kowalski 1M (1991), Post-traumatic stress disorder after a school shooting: effects of symptom threshold selection and diagnosis by DSM-III-R, or proposed DSM-IV Am J Psychiatry 148:592-597 Ten LC (1991), Childhood traumas: an outline and overview. Am]
Psychiatry 148:10-20
215