Abstract
BACKGROUND: Psychosis symptom assessment instruments are rarely used in US mental health community clinical practice despite the advantages of measurement-based care. Barriers include the length of typical scales and that data for scale evaluation often come from researcher and not clinician raters. STUDY DESIGN: The 12-item COMPASS symptom assessment was designed for the RAISE-ETP study of early phase psychosis. Ratings were done by community facility clinicians. COMPASS psychometric properties were examined with a Mokken scale analysis. Subsequently, Mokken analyses were done on 10-item COMPASS data from the ESPRITO learning health system. STUDY RESULTS: 3600 RAISE-ETP COMPASS assessments were examined. The COMPASS 12-item version fulfilled Mokken scale criteria for unidimensionality (H = 0.329 (SE:0.007)) as did a derived reduced 10-item version (H = 0.359 (SE:0.007)) and 5-item version (H = 0.396 (SE = 0.009)). The 12-item version showed one significant monotonicity violation; no significant violations were found in the reduced item versions. Of the reduced item versions, clinicians preferred the 10-item over the 5-item version. In the ESPRITO data, both the 10-item and 5-item versions met unidimensionality criteria (H = 0.458 (SE:0.030) and 0.478 (0.034) respectively) with no monotonicity violations. CONCLUSIONS: The COMPASS scale offers clinicians versions with varying lengths for assessment of individuals with first episode psychotic disorders in community settings. COMPASS can also facilitate data collection for large scale initiatives; the 10-item version is a symptom assessment option in the US national EPINET project.