Factor analysis of two COMPASS psychopathology symptom scales for first-episode psychosis

Improving outcomes for individuals with first-episode psychosis (FEP) benefits from rapid and accurate symptom assessment, yet few measures exist that can be used in 'real world' clinical settings. We evaluated the psychometric properties of a clinician-administered symptom measure, the COMPASS-10, and a briefer conceptually-derived 6-item version (the COMPASS-6), including factor structure, convergent validity, and internal consistency.

Data were gathered from individuals with FEP treated through coordinated specialty care (CSC) clinics as part of the Early Psychosis Intervention Network (EPINET) (N = 3187; age M = 21.4, SD = 4.02; 54.7% male; 43.5% White, 30% Black/African American) in the United States. We used confirmatory factor analysis to evaluate three-factor (positive, negative, and affective symptoms) and two-factor models (positive and negative symptoms) for the COMPASS-10 and COMPASS-6, respectively. Internal consistencies were calculated using McDonald's omega statistic. Convergent validity was assessed via correlations with the Modified Colorado Symptom Index (MCSI).

Fit statistics showed that the three-factor models for the COMPASS-10 and the two-factor model for the COMPASS-6 had generally acceptable fit parameters, with the COMPASS-6 having the best fit indices. Both measures had good internal consistency, and both measures had small to medium correlations with the MCSI.

Findings from a large clinical sample of individuals with FEP supported the psychometric properties of the COMPASS-10 and COMPASS-6 scales with individuals with FEP in largely community settings, demonstrating adequate support for the hypothesized factor structures, as well as good internal consistency and convergent validity with another measure of psychopathology.