Suicide Attempters’ Reaction to Survival as a Risk Factor for Eventual Suicide

The identification of patients who are most likely to kill themselves following a suicide attempt presents a serious challenge to clinicians (1). To examine risk factors for eventual suicide, a prospective study that involved 499 suicide attempters who were recruited from a large public hospital was conducted between 1970 and 1975 (2). Beck and Steer (3) examined 413 of these attempters, 20 (4.8%) of whom committed suicide during the follow-up period. Of several clinical and demographic variables considered (i.e., age, gender, race, marital status, employment status, presence of a depressive diagnosis, presence of a schizophrenia diagnosis, history of drug abuse, presence of an alcoholism diagnosis, previous suicide attempt, Beck Depression Inventory [4] score, Beck Hopelessness Scale [5] score, and Suicide Intent Scale [6] score), only a diagnosis of alcoholism, unemployment, and score on the precautions subscale of the Suicide Intent Scale accounted for significant variance in predicting eventual suicide.

It was surprising that hopelessness did not predict eventual suicide because Beck et al. (7) found that hopelessness was one of the most robust predictors of suicide in hospitalized patients with suicidal ideation. Beck and Steer (3) noted that hopelessness was assessed after the attempt rather than before and that “its predictive validity may be obscured by the psychological aftermath of the attempt” (p. 208). This interpretation suggests that characteristics of the aftermath of a suicide attempt might have relevance in predicting ultimate suicide. Beautrais (1) found that reactions to the attempt (e.g., hoping to die before making the attempt, failing to be relieved by survival) contributed modestly in predicting eventual suicide in 302 suicide attempters.

The purpose of the current study was to reanalyze the suicide attempters studied by Beck and Steer (3) to determine if their reaction to surviving the suicide attempt predicted eventual suicide. We hypothesized that individuals who wished they were dead following the suicide attempt would be at greater risk for killing themselves than those who were glad to be alive or ambivalent.

Method

The present analysis was based on 393 patients drawn from a group of 499 patients who had been admitted for recent suicide attempts to Philadelphia General Hospital between 1970 and 1975. Following the strategy of Beck and Steer (3), we identified 413 patients who had completed the Beck Depression Inventory, Beck Hopelessness Scale, and Suicide Intent Scale and eliminated 20 of them from analysis because participants failed to respond to the reaction to suicide item. There were 167 men (42.5%) and 226 women (57.5%). The racial composition was 51.4% Caucasian (N=202), 47.1% African American (N=185), and 1.5% from other backgrounds (N=6). The mean age was 30.12 years (SD=10.93). Of the 393 attempters, 20 (5.1%) eventually committed suicide.

Beck and colleagues obtained the initial study group by screening the admissions of suicide attempters to Philadelphia General Hospital from 1970 to 1975. Within 24–48 hours of admission, eligible patients were asked for their consent to participate in a longitudinal study of suicidal behavior, were interviewed by a clinical psychologist or psychiatrist to establish psychiatric diagnoses, and were administered standardized assessment instruments. Attempts were made to follow these patients through 1980, and 95% were successfully followed. Whenever possible, patients themselves were contacted semiannually for the first 2 years and then annually by telephone. Other means of contact included collateral informants, first-class or certified mail, and home visits.

If the follow-up indicated that the patient was deceased, appropriate agencies were contacted to verify cause of death, mode of death, and circumstances surrounding death. In addition, the death records of the Philadelphia Medical Examiner’s Office were scanned daily to determine whether any of the patients were on the list. When a patient died, the relevant medical examiner’s or coroner’s records were requested so that the cause of death could be verified.

The variable of interest for the present study is an item that asked about the individual’s reaction to the suicide attempt. Reactions were classified as to whether respondents endorsed that they regretted the attempt and were glad to be alive, were ambivalent about the attempt, or regretted the failure of the attempt and wished they had died. All analyses compared respondents who said they wished they were dead with a combined group of those who were glad to be alive and those who were ambivalent because we viewed regretting the failure of the attempt as having the most clinical relevance.

Results

One hundred forty attempters (35.6%) were classified as wishing they had not made the attempt and being glad to be alive, 168 (42.7%) were classified as ambivalent, and 85 (21.6%) were classified as wishing the attempt had succeeded. There were substantial differences among those who wished they had died and those who were glad to be alive or ambivalent on each of the following measures: Beck Depression Inventory score (t=7.64, df=382, p<0.03), Beck Hopelessness Scale score (t=11.01, df=379, p<0.001), and Suicide Intent Scale score (t=9.05, df=391, p<0.001). In all cases, subjects in the group who wished they had died scored higher on these measures than those in the other two groups.

We performed a Cox regression survival analysis to examine whether the two groups of attempters differed in the likelihood of eventually committing suicide. Results indicated that attempters who wished they had died were more than twice as likely to commit suicide than those who were glad to be alive or who were ambivalent (hazard ratio=2.56, 95% CI=1.06–6.34, Wald χ²=4.34, df=1, N=393, p<0.04). Forty percent of those who committed suicide had endorsed that they wished they had died in their previous attempt (N=8), compared with 21% of those who did not eventually complete suicide (N=77). However, results from a logistic regression analysis indicated that this variable did not predict eventual suicide better than the factors identified by Beck and Steer (3).

Discussion

Consistent with findings reported by Beautrais (1), results from this investigation suggest that a suicide attempter’s reaction to surviving the attempt may be an important risk factor in predicting eventual suicide. Those individuals who wished they had died following their suicide attempt were significantly more likely to kill themselves eventually than those who were glad to be alive or ambivalent about the suicide attempt. In addition, individuals who wished they were dead following the attempt were more depressed and hopeless and had more suicidal intent than those who were glad to be alive or ambivalent about the attempt.

These findings are of particular import because an individual’s reaction to having survived a suicide attempt may be assessed easily and documented during a psychiatric evaluation. Although this variable did not predict eventual suicide beyond other established risk factors, it is valuable in that it is easier to determine in an emergency setting than it is to administer an entire self-report inventory, such as the precautions subscale of the Suicide Intent Scale. Moreover, unlike demographic variables that also predict eventual suicide (e.g., employment status), this variable is one that has been documented as modifiable with appropriate intervention, such as cognitive therapy (8). Thus, individuals who indicate that they wished they had died after a suicide attempt may be monitored and receive interventions to reduce the likelihood that they will attempt suicide again.

Received June 23, 2004; revision received Aug. 26, 2004; accepted Sept. 21, 2004. From the Department of Psychology, James Madison University, Harrisonburg, Va.; and the Department of Psychiatry, University of Pennsylvania Medical School. Address reprint requests to Dr. Beck, Psychopathology Research Unit, Department of Psychiatry, University of Pennsylvania, 3535 Market St., Rm. 2032, Philadelphia, PA 19104; [email protected] (e-mail). Supported by NIMH grants MH-47383 and RO1 MH-60915 (Dr. Beck).