Empirical validity of Leonhard's psychoses: A long-term follow-up study of first-episode psychosis patients.

The validation of nosological diagnoses in psychiatry remains a conundrum. Leonhard's (1979) nosology seems to be one of the few acceptable alternative categorical models to current DSM/ICD systems. We aimed to empirically validate Leonhard's four classes of psychoses: systematic schizophrenia (SSch), unsystematic (USch), cycloid psychosis (Cyclo), and manic-depressive illness (MDI) using a comprehensive set of explanatory validators. 243 patients with first-episode psychosis were followed between 10 and 31 years. A wide-ranging assessment was carried out by collecting data on antecedent, illness-related, concurrent, response to treatment, neuromotor abnormalities, and cognitive impairment variables. Compared with USch, Cyclo, and MDI, SSch displayed a pattern of impairments significantly larger across the seven blocks of explanatory variables. There were no significant differences between Cyclo and MDI in explanatory variables. Except for the majority of illness-onset features, USch displayed more substantial abnormalities in the explanatory variables than Cyclo and MDI. SSch and MDI showed higher percentages of correctly classified patients than USch and Cyclo in linear discriminant analyses. Partial validation of Leonhard's classification was found. SSch showed differences in explanatory variables with respect to Cyclo and MDI. USch showed also significant differences in explanatory variables regarding Cyclo and MDI, although with a lower strength than SSch. There was strong empirical evidence of the separation between both Leonhard's schizophrenia subtypes; however, the distinction between the Cyclo and MDI groups was not empirically supported. A mild to moderate discriminative ability between Leonhard's subtypes on the basis of explanatory blocks of variables was observed.