Comparing four competing models of depressive symptomatology: A confirmatory factor analytic study of 986,647 US veterans

Background: Few rigorous studies have examined the factor structure of major depression symptoms as assessed by current diagnostic systems. This study evaluated four competing models of depressive symptomatology among a large, heterogeneous sample of U.S. veterans. Methods: To determine the best fitting model of major depressive symptoms among four competing models, this study conducted a series of confirmatory factor analyses on a national sample of 986,647 U.S. veterans. Results: A two-factor model first reported by Krause, Reed, and McArdle (2010) provided superior fit to symptom-level data compared to three other models. The optimal model consists of a somatic factor including anhedonia, sleep difficulties, fatigue, appetite changes, concentration difficulties, and psychomotor agitation; and a non-somatic factor including depressed mood, feelings of worthlessness, and thoughts of death. Factorial invariance testing found this model to be invariant by gender and major depression diagnosis. Limitations: A widely used self-report measure of depression was used and the sample consisted solely of veterans so further study is needed with clinician-administered measures and non-veteran samples. Conclusions: Together, these findings support separating symptoms of major depression into somatic and non-somatic factors which may have clinical relevance, and help clarify debates about the factor structure of depressive symptoms. Published by Elsevier B.V.