Subtypes of schizophrenia: a cluster analytic approach.

OBJECTIVES To identify empirical subtypes of schizophrenia, based upon the symptoms recorded over the duration of the illness, and to validate the resulting clusters against other systems that are used for subtyping schizophrenia. METHOD Data for 55 symptoms of schizophrenia over the history of the illness from 107 chronic schizophrenia patients were analyzed using hierarchical cluster analysis with Euclidean distance and Ward's method. Except for 1 patient, all met DSM-III criteria. There were 40 men and 67 women, average (SD) age of 38.2 (9.91) years, with a mean (SD) hospitalization of 27.9 (27.35) months. RESULTS No clear and unambiguous solution for the number of clusters was evident. Examination of the clusters led to further analysis of 2- and 6-cluster solutions. These were contrasted with DSM-III, DSM-III-R, and DSM-IV criteria and with the subtypes taken from the literature. There was limited support for any of these types, with none replicating, including the paranoid-nonparanoid distinction. CONCLUSIONS Empirical clusters derived from lifetime symptom data failed to agree with either the established DSM or other empirically derived subtypes. Subtypes may have little utility when the variability of symptoms over the longitudinal course of the illness is considered.