LAUSR

OBJECTIVE The accepted primary outcome measure for evaluating psychotic symptoms is decades old, long, and initially designed for adults. Surprisingly, the psychometric properties of primary outcome measures have never been reported for a pediatric sample using modern methods. The present study's aim is to use a pediatric sample to evaluate the psychometrics of the most used primary outcome measure in pediatric schizophrenia trials, the Positive and Negative Syndrome Scale (PANSS). METHOD To evaluate the factor structure, item characteristics, and treatment sensitivity of the PANSS in a pediatric sample, secondary analyses of PANSS data at baseline and weekly throughout an 8-week randomized double-blind study of three antipsychotic agents (registered and previously published) were conducted. Subjects were 118 youths receiving outpatient psychiatric treatment for schizophrenia spectrum disorders (Mage=14.26(2.41) years). RESULTS A 10-item short form, keeping two strongest items for each factor, had r=.89 with the full-length scale. Each of the 5 2-item subscales has alphas ranging from .66 to .84. Item Response Theory (IRT) found that the 10-item scale and 2-item subscores had high reliability across the severity range typical of clinical trials. Criterion validity was high, with equal sensitivity to clinical changes over time. CONCLUSION A 10-item PANSS version eliminates weaker items in the pediatric population while preserving coverage of five factors and similar sensitivity to clinical changes over time. It thus may be more appropriate for subsequent pediatric trials, and for clinical use when time and efficiency are paramount.