Psychometric properties of the PANSS autism severity score (PAUSS) in young peoples at clinical high risk for psychosis: does it really measure autistic features?

The PANSS Autism Severity Score (PAUSS) is a commonly used measure for autistic features in early psychosis, including in adolescents at Clinical High Risk (CHR). However, no research examining its psychometric properties in this at-risk population has been published to date. Therefore, we explored reliability, validity, and longitudinal course of PAUSS scores in a sample of young CHR individuals treated in an early intervention service in Italy during 2 years of follow-up. Participants completed the Autism-spectrum Quotient (AQ) and the Positive And Negative Syndrome Scale (PANSS) at baseline and annually during the follow-up. Cronbach's α statistics were used for internal consistency, while Cohen's k statistics and Spearman's correlation coefficients (ρ) were used for convergent validity with AQ scores and longitudinal stability. In the 214 recruited CHR participants, Cronbach's value was 0.849, but the PANSS G5 item showed unacceptable inter-item and item-total correlations. Furthermore, Cohen's k value in relation to AQ scores was unacceptable (0.022), as were ρ (< 0.750) and k (< 0.600) values for longitudinal stability. The results of this research suggest that the PAUSS may not be a valid tool for assessing autistic characteristics in CHR individuals. In fact, it may detect attenuated psychotic symptoms and their transient psychopathological severity, rather than being a reliable indicator of autistic features. Alternatively, its poor discriminatory ability could also reflect contamination of autism measurement by concurrent psychopathology. Therefore, it should never replace careful clinical judgment and the use of appropriate instruments for the diagnosis of ASD.