OBJECTIVES: Despite marked heterogeneity among patients with schizophrenia in their level of functioning, little is known what 'good' 'moderate' or 'poor' levels of functioning look like on various functional measures. This study used an empirical approach to identify and then validate these functional definitions.
METHODS: We used baseline data of a multicenter, effectiveness study comparing antipsychotics in the treatment of outpatients with schizophrenia (n=524; NCT00320489), as this study included several functional measures. A cluster analysis used the Heinrich's Carpenter Quality of Life Scale (QLS), the 12-item Short Form Health Survey (SF-12) mental composite score, and a previously studied productivity measure, to classify patients into functional groups. A three cluster solution was chosen to maximize simplicity, explanatory power and separation among the groups. Clusters were validated using two other functional measures and two previously published definitions of functional levels: an empirical definition that incorporated functioning and symptom severity, and another, using theoretically-driven definitions. Classification and regression tree (CART) analysis was used to establish the criteria for classifying functioning as 'good' 'moderate' or 'poor' with the QLS.
RESULTS: The three clusters consistently differentiated patients on the QLS, SF-12 and productivity measures, reflecting 'good' 'moderate' and 'poor' functional levels. The clusters similarly differed on other functional measures (the Schizophrenia Outcomes Functioning Interview [SOFI] and the Euro-QOL-5D scale), and were concordant with two previously published functional classifications. The CART analysis identified 'good' functioning as QLS total score >84.5, whereas 'moderate' and 'poor' functioning were separated by a cut-off score of 15.5 on the QLS intrapsychic foundation domain. Sensitivity ranged from 86% to 93% and specificity from 89% to 99%.
CONCLUSIONS: The substantial heterogeneity among schizophrenia patients in their level of functioning can be reliably classified in an empirical manner, using specific cut-off scores on commonly used functional measures. Findings have utility for schizophrenia research.
