Abstract
Research on identifying the relevant risk factors for suicides is faced with a multitude of methodological problems. The present study attempts to improve on some of these problems and to isolate those risk factors that are accessible in the early stages of the treatment of inpatients. A total of 3792 inpatients with monopolar or bipolar depression were treated during the period 1981-1992. Suicides (n=33) and controls (n=3759) were compared with respect to 77 sociodemographic and anamnestic variables and 195 standardised items of the admission summary. In addition to an analysis of contingency tables a discriminant analysis was performed. The suicide rate of patients with depressive psychosis was 2.7 times higher than the average rate of 0.324% for the entire clinic. Suicidal tendencies on admission proved to be the best predictor with a frequency of 91% in the suicide group and 40% in the control group, previous attempted suicide being the second best predictor. We conclude that the rate of inpatient suicide may have been underestimated for methodological reasons in the past decades. Many of the risk factors discussed in the literature may be of little predictive value at least in the initial stages of hospital treatment.