Psychometric validation of the physician global assessment scale for assessing severity of psoriasis disease activity

Abstract

The Physician Global Assessment (PGA) is a key measure of psoriasis frequently used in clinical trials. A psychometric validation of a three-item (erythema, induration, and scaling) PGA scale was performed using Phase 2 data. Confirmatory factor analysis (CFA) tested the PGA measurement model and appropriateness of equal weighting of the items. PGA test-retest reliability was assessed by estimating the intraclass correlation coefficient (ICC). Internal consistency reliability was gauged by calculating Cronbach's coefficient α (CCα). Clinically important difference (CID) was defined using the repeated measures model to estimate the relationship between PGA and Patient Global Assessment (PtGA). Known-group, convergent, and divergent validity for the PGA were also assessed. 197 patients with psoriasis were randomized to tofacitinib 2, 5, 15mg twice daily, or placebo. CFA demonstrated that the PGA measurement model fitted the data using equal weighting of the PGA items. The PGA scale demonstrated good test-retest reliability (ICC>0.7) and internal consistency reliability (CCα>0.8). The CID for PGA was estimated at 0.52 (95% confidence interval: 0.47, 0.56). A robust monotonic relationship between PGA and Psoriasis Area and Severity Index (PASI) data substantiated known-group validity. Relatively high correlations of PGA with PASI and PtGA data (all correlations >0.5 except at baseline) supported convergent validity; relatively low correlations of PGA with the Pain/Discomfort Assessment and the Ocular Comfort Index supported divergent validity. The three-item PGA scale has sound psychometric properties with respect to reliability and validity, with equal weighting of the items being appropriate.