Principal components and further possibilities with the PANSS.

At the end of the last century, Hughlings-Jackson suggested that positive and negative syndromes should be kept apart in psychotic disorders. When the concepts of dementia praecox and schizophrenia were introduced by Kraepelin and Bleuler, emphasis was laid on the negative symptoms, regarded as fundamental. After the introduction of the "first rank symptoms" by Schneider emphasis switched to the positive symptoms in schizophrenia and these symptoms were included in most diagnostic criteria. In the 1980s Andreasen and Crow suggested a dichotomy into positive and negative syndromes in schizophrenia. Kay and co-workers introduced a Positive And Negative Syndrome Scale (PANSS) for schizophrenia. In the original studies satisfactory construct validity and inter-rater reliability were demonstrated. However, in studies outside the USA a high construct validity was found for the negative scales but not for the positive and general psychopathology scales. Furthermore, the inter-rater reliability of the negative scale was a problem. After introduction of the Structured Clinical Interview for the PANSS (SCI-PANSS) the inter-rater reliability increased for all three scales. In an early study Kay and Sevy found seven factors in a principal component analysis of the PANSS and suggested a four factor pyramidical model. Later principal component analyses by Lepine, Peralta et al. and Kawasaki et al. suggested that the four factor model was an oversimplification and Lindström and von Knorring suggested a five factor pyramidical model. A similar model was later suggested by Bell et al. after a reanalysis of the original series of Kay and Sevy.(ABSTRACT TRUNCATED AT 250 WORDS)