No description available.

Classification systems in psychiatry are a work in progress. Therefore, continued efforts to improve their validity are necessary. Epidemiology provides a scientific method to assess the extent of psychiatric morbidity in community populations. However, data from epidemiological surveys have also contributed, either directly or indirectly, to many changes in the classification systems. Recent reviews of the current state of psychiatric classification indicate four unresolved issues: 1) the presence of two differing classification systems, 2) the role of the clinical significance criterion in differentiating psychopathology from normality, 3) the relationship of the exclusion criteria to the co-occurrence of psychiatric disorder pairs, and 4) the relative validity of categorical versus dimensional conceptualizations of psychiatric disorders. The current thesis examines these four unresolved issues, using data from a large-scale epidemiological survey of psychiatric disorders. With regard to GAD, differences in diagnostic criteria between DSM-IV and ICD-10 resulted in different types of cases identified, despite similarities in prevalence. The DSM-IV diagnostic criterion for clinical significance impacted, albeit to different degrees, on the prevalence, health service use and impairment of five disorders. The exclusion criteria in both DSM-IV and ICD-10 were significantly related to the patterns of co-occurrence found in the data. Using the example of depression, symptoms were more consistent with a dimensional rather than a categorical structure. A specific research agenda is proposed, the aim of which is to provide possible avenues of research that may benefit revisions to classification systems and the conduct of epidemiological surveys. This research agenda contains a number of suggestions. Future revisions will benefit from an explicit understanding of the differences between the classification systems. Better definitions of the concepts of clinical significance and psychiatric disorder are required. The co-occurrence of disorder pairs in epidemiological data informs understanding of the exclusion criteria, but the validity of these criteria relies on different data. Dimensional models of classification may yield more information than categorical models, and methods for incorporating them in large-scale surveys are proposed. It is concluded that epidemiological data should continue to play a significant part in the refinement of psychiatric classification.