Heterogeneity and symptom structure of schizophrenia.

Previous studies failed to identify a consistent factor structure of the BPRS-24 in schizophrenia. Our aims were to examine the fit of all previously published factor models and then to explore unobserved population heterogeneity and identify salient latent classes. Two hundred thirty-nine patients with ICD-10 schizophrenia admitted to a random sample of all Italian public and private acute inpatient units during an index period were administered the BPRS-24. Confirmatory factor analysis (CFA) was used to test all factor models derived in previous studies. Then, factor mixture analysis (FMA) with heteroscedastic components was carried out to explore unobserved population heterogeneity. No previously reported factor solution showed adequate fit in CFA. FMA indicated the presence of three heterogeneous groups and yielded a 5-factor solution (Depression, Positive Symptoms, Disorganization, Negative Symptoms, Activation). Group 1 was characterized by higher Disorganization, lower Activation, lower psychosocial functioning, greater lifetime number of admissions, more frequent history of compulsory admission. Group 2 displayed lower Disorganization. Group 3 showed higher Activation and more frequent history of recent self-harming behavior. Our finding that a reliable factor structure for the BPRS-24 could be obtained only after assuming population heterogeneity suggests that the difficulty in identifying a consistent factor structure may be ascribed to the clinical heterogeneity of schizophrenia. As compared with clinical subtypes, the psychopathological dimensions displayed much greater discriminatory power between groups identified by FMA. Though preliminary, our findings corroborate that a dimensional approach to psychopathology can facilitate the assessment of the clinical heterogeneity of schizophrenia.