A PERSPECTIVE ON SUBTYPES OF SCHIZOPHRENIA FROM CLUSTER ANALYSIS
 
Helmes, E., James Cook University and Landmark, J., Australia
 
Subtypes of schizophrenia have been identified since the recognition of the disorder. They remain in wide use in diagnostic and recording systems, despite the lack of substantial empirical evidence for them. There is little agreement in the existing literature, with the most substantial support for the paranoid versus non-paranoid distinction. Current and historical data for 55 symptoms of schizophrenia from 107 chronic cases of schizophrenia diagnosed using DSM-III criteria were analyzed using hierarchical cluster analysis based upon Euclidean distance and Ward's method. The average age of the 40 men and 67 women was 38.2 years (SD = 9.91), with a mean of 27.9 months of hospitalization (SD = 27.35). Several methods failed to agree on the number of clusters. Analysis of two- and six-cluster solutions seemed appropriate. Clusters were contrasted with DSM-III, -III-R, Crow, Farmer et al. (1983), Liddle (1987), and Carpenter et al. (1976) subtypes. There was very little support of any of these types with none of them replicating in the present sample, including the paranoid-non paranoid distinction. Results are interpreted in terms of stability of symptoms over the course of the illness.