Background

Color groups of fruits and vegetables (FV) are part of a healthy diet, but evidence for an association with cardiometabolic outcomes is inconsistent.

Objective

To examine the association between intake of FV of different colors with incident diabetes and cardiometabolic risk biomarkers among U.S. Hispanics/Latinos.

Subjects/methods

We used data from 9206 adults ages 18-74 years who were free of diabetes at baseline (2008-2011) and had follow-up data at visit 2 (2014-2017) in the Hispanic Community Health Study/Study of Latinos (HCHS/SOL), a multicenter, prospective cohort study of self-identified Hispanics/Latinos. Dietary intake was assessed using two 24 h recalls at baseline. FV were categorized into five color groups: green, white, yellow/orange, red/purple, and uncategorized. Diabetes was defined based on laboratory measures and self-reported antihyperglycemic medication. We used survey logistic regression models to evaluate the association between FV color groups and incident diabetes and survey linear regression models to evaluate the association of FV color groups with cardiometabolic risk biomarkers at visit 2.

Results

During ~6 years of follow-up, 970 incident cases of diabetes were documented. The red/purple FV color group was the least consumed (0.21 servings/day), whereas white FV were the most consumed (0.92 servings/day). For each serving of total FV intake, body mass index (BMI) was lower by 0.24% (p = 0.03) and insulin by 0.69% (p = 0.03). For each serving of red/purple FV intake, HDL was 1.59% higher (p = 0.04). For each serving of white FV intake (with potato), post-OGTT was 0.83% lower (p = 0.04) and triglycerides 1.43% lower (p = 0.04). There was no association between FV intake and incident diabetes.

Conclusions

Specific FV colors were associated with cardiometabolic benefits though the associations were of relatively small magnitudes. Dietary recommendations could consider varying colors of FV intake, especially white and red/purple color groups, for a healthy diet.