The classification of depression: traditional versus mathematical approaches.

The authors examine the classification of depressive disorders with an empirically and mathematically based method, cluster analysis. In a sample of 228 patients from the NIMH Collaborative Study of the Psychobiology of Depression, clusters were generated by using crossectional symptoms; course of illness, family history, and treatment variables were used as independent variables to evaluate the validity of the clusters. The four clusters identified correspond roughly to severe depressions with endogenous features, less severe depressions, bipolar depression with cycling within the episode, and psychotic depressions. This study supports the Research Diagnostic Criteria and DSM-III approaches to the classification of depressive disorders.