Diabetes modifies the cross-sectional association between Healthy Eating Index-2015 and clinically relevant depressive symptoms: a NHANES analysis with an independent hospital-based replication cohort

Front Nutr. 2026 Jun 17;13:1844683. doi: 10.3389/fnut.2026.1844683. eCollection 2026.

ABSTRACT

BACKGROUND: Although dietary patterns have been increasingly linked to mental health, evidence regarding the cross-sectional association between overall diet quality and clinically relevant depressive symptoms remains inconsistent. We aimed to examine the association between the Healthy Eating Index-2015 (HEI-2015) and PHQ-9-defined clinically relevant depressive symptoms in a nationally representative U.S. population and to assess whether the findings were reproducible in an independent hospital-based replication cohort, with particular attention to the modifying role of diabetes.

METHODS: We analyzed cross-sectional data from 12,462 participants in the National Health and Nutrition Examination Survey (NHANES) 2007-March 2020. A replication analysis was performed in an independent cohort of 1,019 individuals who underwent routine health examinations at the Health Examination Center of the Second Affiliated Hospital of Shandong First Medical University between 2024 and 2025. Clinically relevant depressive symptoms were defined as a Patient Health Questionnaire-9 (PHQ-9) score ≥ 10, and dietary quality was assessed using HEI-2015. The hospital-based HEI-2015 score was derived from questionnaire-based dietary information and should be interpreted as a harmonized approximation rather than an identical replication of the NHANES 24-hour recall-based HEI-2015. Multivariable logistic and linear regression models were used to examine the associations of HEI-2015 with PHQ-9-defined clinically relevant depressive symptoms and with PHQ-9 score. Subgroup analyses and restricted cubic spline models were performed to assess effect modification and dose-response patterns.

RESULTS: In the NHANES cohort, higher HEI-2015 was associated with lower odds of PHQ-9-defined clinically relevant depressive symptoms in the fully adjusted model (OR per 1-point increase = 0.992, 95% CI: 0.984-1.000, P = 0.048; approximately OR per 10-point increase = 0.923) and with lower PHQ-9 scores (β per 1-point increase = -0.008, 95% CI: -0.014 to -0.002, P = 0.006). These findings were reproduced in the hospital-based replication cohort, in which higher HEI-2015 was also associated with lower odds of clinically relevant depressive symptoms (OR = 0.960, 95% CI: 0.925-0.997, P = 0.036; approximately OR per 10-point increase = 0.665) and lower PHQ-9 scores (β = -0.049, 95% CI: -0.077 to -0.022, P < 0.001). Diabetes significantly modified these associations in both cohorts. Among individuals without diabetes, higher HEI-2015 was associated with more favorable depression-related outcomes, whereas among those with diabetes, the association was directionally opposite. Restricted cubic spline analyses supported a significant overall association between HEI-2015 and clinically relevant depressive symptoms in both cohorts, without strong evidence of nonlinearity.

CONCLUSION: Higher dietary quality was cross-sectionally associated with lower odds of PHQ-9-defined clinically relevant depressive symptoms and lower depressive symptom burden, and these associations were reproduced in an independent hospital-based replication cohort. However, the association was consistently modified by diabetes, with an opposite-direction association observed among diabetic individuals. These findings should be interpreted cautiously because of the cross-sectional design, possible reverse causation, dietary measurement differences between cohorts, and residual confounding. Prospective and interventional studies are needed to clarify whether and how metabolic status influences the relationship between diet quality and depressive symptoms.

PMID:42389699 | PMC:PMC13318733 | DOI:10.3389/fnut.2026.1844683