Evaluating the PANSS using item response theory in forensic psychiatric samples from five European nations.

Item Response Theory (IRT) describes a set of statistical models describing how individual items in a test or questionnaire relate to the underlying characteristic or trait that the test claims to measure. Until now IRT models have not been applied to the Positive and Negative Syndrome Scale (PANSS) in forensic and general psychiatric samples to establish its psychometric properties and explore the link between psychotic symptom severity and violent behavior in schizophrenia. This study investigated patients with schizophrenia spectrum disorders and a history of violence from forensic institutions and non-violent patients from general psychiatric settings in five European countries. A total of 398 participants were assessed using the PANSS. IRT analysis revealed a poor model fit for the Partial Credit Model (PCM) with considerably disordered thresholds for most items. Differential item functioning (DIF) revealed significant differences between the two groups, notably for items hypothetically linked to violence risk, such as delusions and hostility. These findings reveal potential limitations when trying to compare PANSS scores across these two clinical populations.