[On the importance of the "decision-making model" view of diagnosis as a clinical framework in psychiatry].

After the advent of DSM-III, operational diagnostic criteria, along with the classification of disorders using such criteria, received considerable attention, and many studies on the reliability and validity of psychiatric diagnosis were conducted worldwide. Operational methodology was applied to diagnosis and classification, especially, in the area of research, and has contributed greatly to advances in reliable and refined clinical research. Such methodology, however, has not necessarily been accepted as a guiding principle in the area of clinical practice by all psychiatrists. Rather, some psychiatrists, especially more experienced psychiatrists, took a somewhat negative attitude toward the use of operational methodology. The author contends that one of the causes for the relatively poor acceptance of operational methodology in the area of clinical practice lies in the "classification model" view of diagnosis that forms the implicit background for the methodology. From a clinical perspective, it is not from the "classification model" basis but rather, from the "decision-making model" basis that the actual process of clinical diagnosis in psychiatry is explained properly. This is a very important point, because the latter model is potentially more useful both to psychiatric patients and to researchers in psychiatry than the former model. There have been however, few reports in psychiatry that highlight the importance of this model as the clinical framework. The author analyzes the limitations of the "classification model" view, and then, based on this analysis, lists prerequisites that a model for the framework of clinical practice should possess. The prerequisites listed are: that clinical information not sufficient to meet the disease criteria should be used as effectively as possible, that diseases low in probability but high in seriousness should be considered by clinicians in the differential diagnoses; that diagnosis should be readily changed when necessary; that the component of benefit to the patient should be included in the model; and that the relationship between nosological definition of diseases and practical diagnostic criteria should be explained explicitly. In addition, the author emphasizes that psychiatric diagnosis involves "decision-making under the condition of incomplete information", because most psychiatric diseases are still of unknown etiology. With all these conditions in mind, the author contends that the Bayesian statistical model of decision-making is suited to the frame of reference, and that it should replace the classification model as a guiding principle. This model can integrate various aspects of psychiatric clinical activities systematically and explicitly. A case is presented to illustrate the model in a clinical context. The author points out problems to be solved in current psychiatry from the perspective of the Bayesian model of decision-making. Finally, limitations of the Bayesian model view are discussed.