Morbid risks of subgroups of affective disorders: some methodological and empirical results

Based on two independently ascertained family studies, the problem of analysing disorders with variable age of onset has been investigated by means of survival analysis. This method of approach results in empirical risk functions which can be used directly as weights for age correction in the analysis of pedigrees. The main interest of the present investigation has been focused on the reproducibility of such risk functions and on their specificity with respect to clinical diagnosis. For this purpose, one family study was referred to as a calibration sample for estimating empirical risk functions, whereas, the second family study served as a test sample in order to test the hypotheses under discussion. As it turned out, the empirically derived functions from both the calibration and test samples are indeed comparable, thus suggesting that, for sufficiently representative calibration samples, the differences between the two populations of (1) affected and (2) susceptible individuals may be less important than expected. As to the specificity of these risk functions, the underlying diagnostic structure of the calibration samples (as represented by four diagnostic subgroups) could be characterized by distinct types of risk functions, each of which, in addition, reflects the severity of the respective affective disorder.