Revisiting the 5 dimensions of the positive and negative syndrome scale

BACKGROUND: The Positive and Negative Syndrome Scale (PANSS) is a widely used instrument to measure severe psychopathology in schizophrenia. Recently, its widely accepted 5-factor solution is questioned using confirmatory factor analysis (CFA). This article examines the appropriateness of applying CFA to PANSS data and the dimensionality of the PANSS in a large sample of persons with chronic schizophrenia, schizophreniform disorder, and schizoaffective disorder. MATERIALS AND METHODS: Data reduction was conducted on PANSS baseline assessments from 2 multicenter clinical trials consisting of 1284 persons' available PANSS data at baseline and Week 8. Supplementary analyses were conducted for separate trials and for noncompleters at baseline (n = 1872). RESULTS: Examination of the data indicated that the statistical requirements of CFA (eg, multivariate normality) were not met by the data. Multidimensional scaling and a scree plot of principal components superimposed on simulated random data converged to indicate that 5 dimensions are generally a parsimonious fit to the PANSS at baseline, whereas 4 dimensions emerged after 8 weeks of treatment. The components included negative symptoms, positive symptoms, disorganized (baseline only), anxiety/depression, and excited. CONCLUSIONS: The results demonstrate realistic alternatives to CFA when studying the PANSS, support a 5-component model of PANSS ratings at baseline, and indicate that the structure of symptoms change after 8 weeks of treatment. These results seem to be robust because they generally replicate within trials for completers and at baseline for the entire sample.