The incidence of suicide among adolescents and young adults in the United States approximately tripled between 1952 and 1992 (1). Accordingly, the U.S. Public Health Service has targeted the reduction of suicide among adolescents and young adults as a goal of the Healthy People 2000 initiative (2). Identifying risk factors for suicidality will facilitate early detection and the targeting of treatment for those most likely to commit suicide.
Bereavement has been shown to elevate the risk of suicidality (3–5). While young adults who survive the death of someone close have been found to be at risk of suicide (5), the effects of traumatic grief symptoms on suicidal ideation among young adults have not been explored. The present study used follow-up survey data from 76 young adult friends of suicide victims to examine the main and interactive effects of traumatic grief and depression on suicidal intent.
Subjects. The initial group contained 146 friends of 26 consecutive adolescent suicide victims who died between December 1988 and March 1991 in metropolitan Pittsburgh (6). Family members nominated peers whom they considered closest to the victim. The initial participation rate was 64%. Subjects who refused to participate were younger than those who agreed to participate and were more likely to be male. Friends of the suicide victims who participated in the intake and 3-year follow-up interviews were identified (N=132), and questionnaires were sent to the 122 for whom current addresses were available. Of the 76 participants in our follow-up, 44 (58%) were women, and 70 (92%) were Caucasian. The mean age was 23.8 years (SD=1.8). The 76 participants came from 23 of the social networks represented in the original group. Subjects who did not participate in our follow-up (N=56) were more likely to be men and of lower socioeconomic status and had higher rates of current anxiety disorders. They did not differ from subjects who participated in the follow-up on age, severity of depressive or traumatic grief symptoms, rates of suicidal ideation, or rates of suicide attempts.
Procedure. The measures listed below were sent to subjects a mean of 6.3 years (SD=0.6) after their friend’s suicide. Written informed consent for investigation of human subjects was obtained.
Traumatic, or what we formerly called "complicated," grief was assessed by the Inventory of Complicated Grief, which has been shown to have sound psychometric properties in an elderly sample (7). Cronbach’s alpha was 0.94 in the current study group. Because subjects with Inventory of Complicated Grief scores in the upper 20th percentile have been shown to have impaired quality of life (7), those with scores in the upper 20% on this measure (N=15 of 76) were considered "cases."
The Beck Depression Inventory (8) demonstrated good internal consistency in the current group (Cronbach’s alpha=0.88). The inventory’s item assessing suicidal ideation was omitted from the summary score. On the basis of prior results (D.A. Brent, unpublished data), a Beck Depression Inventory score greater than 10 was selected as the cutoff point for "caseness." Fifteen of 76 subjects had scores higher than 10.
Comorbidity was determined by multiplying traumatic grief by depression caseness (i.e., yes=1 if met criteria for both and no=0 otherwise). Six of the 76 respondents had comorbid traumatic grief and depression; eight of the 15 who met criteria for traumatic grief did not meet criteria for depression (phi=0.34).
We modified the well-established Beck-Kovacs Scale for Suicidal Ideation (9) to be a 15-item self-report questionnaire. Cronbach’s alpha for the modified Scale for Suicidal Ideation in the present group was 0.81. Because the scale’s scores were highly skewed toward no suicidal ideation, a dichotomous variable was created in which 0 reflected no suicidal ideation (N=45) and 1 reflected suicidal ideation at or above one item endorsed (range=1–17 items endorsed; median=4) (N=31). Three of the original 132 subjects (2.3%) attempted suicide, but there were no suicides.
Analyses. Logistic regression models estimated the main and interactive effects of syndromal traumatic grief and depression on the likelihood of suicidal ideation, with control for gender and time since loss. Grouped logistic regression and mixed model analysis of variance were used to test for the effect of membership in one of the 23 social networks in predicting suicidal ideation. When each technique was used, there was no evidence of a network effect; therefore, we did not adjust for network status in subsequent analyses. Posttraumatic stress disorder (PTSD) was omitted from the final models because 1) preliminary stepwise regression models predicting suicidal ideation failed to retain PTSD in models that included traumatic grief, and 2) a phi coefficient of 0.34 between caseness for PTSD and traumatic grief suggested that collinearity was not an issue.
Bivariate analyses also revealed no significant associations among traumatic grief, depression, and suicidal ideation and measures of exposure to the suicide (e.g., seeing the body), age, race, education, income, social support, and religion. Consequently, these variables were not entered into the regression models.
Subjects with syndromal levels of traumatic grief were found to be approximately five times more likely to report suicidal ideation than were subjects with nonsyndromal levels of traumatic grief (t1). The significant effect of traumatic grief on suicidal ideation remained even after control for depression, gender, and time since loss. The interactive effect of comorbid traumatic grief and depression was not significantly associated with suicidal ideation over and above the main effects.
Syndromal levels of traumatic grief significantly increased the likelihood of suicidal ideation. The significant effect of traumatic grief on suicidal ideation remained even after depression was entered into the model, suggesting that it is an independent clinical correlate of suicidal ideation. The absence of a significant interactive effect of traumatic grief and depression on suicidality indicated that comorbidity did not heighten the risk of suicidal intent.
Although there were no suicides reported among the friends of suicide victims within the study observation period, prior suicidal thoughts and gestures have been identified as a leading predictor of subsequent suicide (10). The results of this study indicate that traumatic grief and depressive symptoms independently heighten the risk of suicidal ideation, which poses a risk for subsequent suicide.
Longitudinal data will be needed to determine whether traumatic grief and depression are preludes to suicidal ideation. Future research might identify confounding factors (e.g., fear of abandonment, aggression ) that both predispose survivors to be traumatized by the death of someone close and impel them to want to commit suicide. Further studies are needed to determine whether the effect generalizes to young adults bereaved by a death from other causes (nonsuicide) and to subjects with other relationships to the deceased (e.g., parents, siblings).
Received April 14, 1997; revisions received Jan. 28 and May 18, 1999; accepted May 26, 1999From the Department of Psychiatry and the Donaghue Women’s Health Investigator Program, Yale University School of Medicine, New Haven, Conn.; the Mental Illness Research, Education, and Clinical Center, VA Medical Center, West Haven, Conn.; and the Department of Psychiatry, University of Pittsburgh School of Medicine, Pittsburgh, Pa. Address reprint requests to Dr. Prigerson, Rm. 522, Connecticut Mental Health Center, 34 Park St., New Haven, CT 06519; Holly.Prigerson@Yale.edu (e-mail)Supported in part by NIMH grant MH-01100 and by grants from the American Suicide Foundation and the Claude D. Pepper Older Americans Independence Center (P60-AG-10498). The authors thank the project coordinator, Grace Moritz, A.C.S.W., for her assistance and Satish Iyengar, Ph.D., for his statistical support.