Changes in suicide numbers in psychiatric hospitals: an analysis using log-linear time-trend models

Abstract

We examined the hypothesis that there is an inerease in the number of suicides committed by psychiatric inpatients. Data were recorded in five psychiatric hospitals/departments in Baden-Württemberg (Germany) for the period between 1970 and 1992. Statistical analysis was performed by means of log-linear time-trend models that allowed us to test several hypotheses, to compare their goodness of fit, and to correct for related factors (number of admissions, number of suicides in the general population). Results showed that the model of a constant number of suicides and the model of a general increase could be rejected. The separation of the yearly frequencies into two parts (1970–1979 and 1980–1992) revealed an increase in the inpatient suicide rate during the years 1970–1979 and a slowly decreasing inpatient suicide rate during the years 1980–1992, especially when taking admission numbers into account. Diagnostic subgroups (schizophrenia and depression) did not seem to deviate from these trends. The correction for the suicide numbers in the general population of Baden-Württemberg, which exhibited a remarkable decrease in the 1980s, did not influence these results. We speculate that, after some starting difficulties, the effects of the more humane treatment procedures introduced in the 1970s are beginning to be felt.