[Problems in case identification using a semi-structured psychiatric interview as exemplified by an epidemiological study in general practices (author's transl)].

From 1973 to 1974 research psychiatrists studied 1,274 patients in general practices in the county of Traunstein (FRG). On the basis of an interview developed by Goldberg et al. (1970), it could be shown that objective symptoms registered by the interviewers generally differentiate better between cases and non-cases than do subjective symptoms. According to the weighting suggested by the developers of the interview, singly weighting the subjective and doubly weighting the objective symptoms, a total score was compared with the psychiatric diagnosis made by the interviewer. At a cut-off of 20 points, the overall agreement reached an optimum of 83.1% and increased only slightly due to a-posterior weighting of the items. A cluster analysis of the symptoms showed that patients demonstrate significantly different combinations of symptoms according to the nature of their mental disorder.