Statistical predictions of suicide in depressives

The purpose of the present study was to determine how accurately suicide potential could be identified among patients with primary affective disorder by means of multivariate statistical procedures. Among patients with this disorder, suicide has been estimated to be about 30 times more prevalent than in the general population 1 and to be the cause of death in about 15% of patients. 2 However, many patients with depressive disorder never attempt suicide or become preoccupied with suicidal thoughts. The present study is concerned with identifying some of the factors that distinguish the suicidal from the nonsuicidal depressive. A number of studies to derive predictions of suicide potential by multivariate statistical means have appeared recently. Some that have applied vague predictor variables and met with little success have been reviewed elsewhere. 3 Others have been concerned with a diagnostically heterogeneous group of patients with a history of suicide attempts in order to determine prognostic factors associated with repeated attempts. 4,5 In contrast, the present study was concerned with a diagnostically homogeneous group that is heterogeneous with regard to suicidal behavior. These patients were classified as attemptors, contemplators, or nonsuicidals according to their responses in a structured interview, and the ability of predictor variables to identify membership in one of these groups was assessed by means of discriminant function analysis. The predictor variables were age, sex, marital status, diagnostic subtype, family history of suicidal behavior, and scaled ratings of work impairment and social isolation.