Background/aims: Cluster analysis has had limited success in establishing whether there are subtypes of schizophrenia. Grade of membership (GoM) analysis is a multivariate statistical technique which has advantages when, as in schizophrenia, individuals conforming to pure types are uncommon and mixed forms are frequent.
Methods: GoM analysis was applied to 118 chronic schizophrenic patients. The patients were of all clinical subtypes, including 13 with simple schizophrenia. Both current and 'lifetime' symptoms were assessed, and two different rating systems were used.
Results: Specifying 3 pure types resulted in robust findings across analyses. One pure type corresponded to paranoid schizophrenia, one to simple schizophrenia and the third combined elements of hebephrenic and catatonic schizophrenia. Specifying 4 pure types split the original 3 pure types in ways which were not clinically intuitive.
Conclusion: GoM analysis divides schizophrenia into subtypes along conventional lines, with the proviso that hebephrenic and catatonic schizophrenic patients are not separable, at least in the chronic stage of the illness.
Copyright 2009 S. Karger AG, Basel.