Sensitivity of individual items of the Positive and Negative Syndrome Scale (PANSS) and items subgroups to differentiate between placebo and drug treatment in schizophrenia.

Lack of hard clinical endpoints is an essential problem in schizophrenia research. Disease state and treatment outcomes are measured using rating scales, e.g. Positive and Negative Syndrome Scale (PANSS). However, the PANSS score cannot always differentiate between placebo and drug treatment, even for established antipsychotics. The goal of this study was to identify the individual items of PANSS and subscales of selected items which are most sensitive to differentiate between placebo and drug effect. We analysed data from seven clinical trials of different antipsychotics. "Mini-PANSS" scales consisting of the most sensitive items were created and analysed statistically. The power of these scales to show a significant difference between placebo and drug treatment was compared with the power of total PANSS. Furthermore, pharmacokinetic-pharmacodynamic analysis was performed to determine which of these scales shows the highest drug effect on top of the placebo effect. The results reveal that all 30 items of the PANSS scale show a therapeutic drug effect. The magnitude of placebo effect was not predictive for the power to detect drug effect. Mini-PANSS scales consisting of items with the largest drug treatment response and the scale with the largest mean-to-SD ratio are somewhat better in differentiating between placebo and drug treatment than the total PANSS. However, the difference between these scales and total PANSS is small. Therefore, our analysis does not support replacement of the total PANSS by a reduced scale in the analysis of primary endpoints.