Comparing the factor structures and reliabilities of the EPDS and the PHQ-9 for screening antepartum and postpartum depression: a multigroup confirmatory factor analysis

Abstract

To evaluate and compare the factor structure and reliability of EPDS and PHQ in antepartum and postpartum samples. Parallel analysis and exploratory factor analysis were conducted to determine the structure of both scales in the entire sample as well as in the antepartum and postpartum groups. McDonald’s omega statistics examined the utility of treating items as a single scale versus multiple factors. Multigroup confirmatory factor analysis (MCFA) was utilized to test the measurement invariance between the antepartum and postpartum groups. Two-factor models fit best for the EPDS in both the antepartum and postpartum groups; however, the most reliable score variance was attributable to a general factor for each scale. MCFA provided evidence of weak invariance across groups regarding factor loadings and partial invariance regarding item thresholds. PHQ-9 showed a two-factor model in the antepartum group; however, the same model did not fit well in the postpartum group. EPDS should be preferred to PHQ-9 for measuring depressive symptoms in peripartum populations. Both scales should be used as a single-factor scale. Caution is required when comparing the antepartum and postpartum scores.