The Stability of Psychiatric Diagnoses

Abstract

Numerous studies of the reliability of psychiatric diagnosis have been published in the last twenty years, but most have been based either on a comparison of the diagnoses made by two or more raters at a single joint interview, or separate interviews a few days apart (the observer agreement model), or on a comparison of the overall spectrum of diagnoses assigned to two comparable series of patients (the frequency agreement model). Comparisons between the diagnoses assigned to patients on successive admissions, or at other widely separated points in time (the consistency or stability model), have been carried out less often, in spite of the fact that the usefulness of our diagnostic categories is just as dependent on temporal stability as on reliability in an observer agreement situation.