Diet Composition as a Candidate Mechanism Underlying the Depression-to-Obesity Association: The CARDIA Study

Elizabeth A Vrany, Purdue University


People with depression are at elevated risk for future obesity; however, little is known about the potential mechanistic role of diet composition in this association. The aims of the present study were: (1) to examine depressive symptom severity as a predictor of 13-year change in seven diet composition factors over time, (2) to test 13-year change in diet composition factors as mediators of the association between depressive symptom severity and 13-year change in adiposity, and (3) to explore whether the Aim 2 mediation models are moderated by sex and race. Participants were 2,449 non-Hispanic Black and White adults who participated in the 1990, 1992, and 2005 years of the CARDIA study (mean baseline age = 35 years, 54% women, 56% non-Hispanic Black, mean baseline waist circumference = 84.0 cm, mean change in waist circumference = 8.3 cm). Depressive symptoms were assessed in 1990 using the Center for Epidemiologic Studies-Depression Scale (CES-D). Waist circumference and seven diet composition factors were assessed in 1992 and 2005. Diet composition factors included intake of total energy, saturated fatty acid, monounsaturated fatty acid, polyunsaturated fatty acid, fiber, sugar, and protein. PROCESS bootstrapping analyses were used to test for mediation and moderated mediation (see Figure 1 for the conceptual model). All models included adjustment for age; sex; race; education; prevalent and incident cardiovascular disease, diabetes, and cancer; incident pregnancy; and 13-year change in physical activity. Greater baseline CES-D Total predicted greater 13-year increases in waist circumference (path c β = 0.032, p = .006). In separate models, there was a trend for greater baseline depressive symptoms predicting 13-year increases in total energy intake (path a β = 0.040, p = .054), and greater baseline depressive symptoms predicted 13-year increases in protein intake (path a β = 0.059, p = .004) and fiber intake (path a β = 0.040, p = .0496). Depressive symptoms were unrelated to 13-year changes in intake of the other diet composition factors of saturated fatty acid, monounsaturated fatty acid, polyunsaturated fatty acid, and sugar (all ps ≥ .129). Mediation models revealed that 13-year change in total energy intake and protein intake partially mediated the association between baseline depressive symptoms and 13-year change in waist circumference (total energy intake indirect effect = 0.001, 95% CI=0.0001-0.004; protein intake indirect effect = 0.002, 95% CI=0.0004-0.005); mediation was not observed for the other diet composition factors. Exploratory moderated mediation models detected a significant interaction for CES-D Total by sex predicting 13-year change protein intake only (Index of moderated mediation=0.004, 95%CI=0.001-0.010). All other tests of moderated mediation were non-significant (95% CIs included zero). Analyses stratified by sex revealed that CES-D Total was more strongly associated with 13-year increases protein intake in men (β = 0.011, p=.004) than in women (β = 0.027, p = .246). Findings from this larger, longitudinal, epidemiologic study suggest that the prospective relationship between depression and future obesity risk may be explained, in part, by intake of certain diet composition factors. Advancing the understanding of the relationship between depression and diet changes will inform clinical efforts to prevent or manage depression-related obesity.




Stewart, Purdue University.

Subject Area


Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server