Associations of diet composition and quality with continuous glucose monitor-derived glycemic metrics in a community-based cohort

Continuous glucose monitors (CGM) offer real-time assessment of glucose levels, providing an opportunity to understand the determinants of glycemic variability. We aimed to cross-sectionally evaluate the associations between carbohydrate substitution, indices of diet and carbohydrate quality with CGM-derived measures in individuals without diabetes. Participants from the Framingham Heart Study, with ≥3 days of CGM data and ≥2 days of dietary records, were included in this analysis. CGM-derived glycemic traits were calculated using the "cgmanalysis" R package. Multivariable linear regression models were used to derive β (SE) from the relationships of carbohydrates (% energy intake (EI)), carbohydrate quality (carb-to-fiber ratio), and indices of diet quality (healthy eating index (HEI), alternate healthy eating index (AHEI), dietary approaches to stop hypertension (DASH) and alternate Mediterranean diet score (aMED)) with CGM-derived measures among all participants and stratified by glycemic status. Least square means were estimated to visualize adjusted group differences of percent time spent above 140 mg/dL across quartiles of diet and carbohydrate quality metrics. We included 677 individuals in this analysis (56.9% with normoglycemia, 59.4% female, mean age 60.1y, mean glucose 117.7mg/dL). Overall, higher diet and carbohydrate quality were associated with favorable CGM-derived measures. Replacing 5% EI from protein with the equivalent from carbohydrate was associated with 0.97mg/dL higher CGM mean glucose (SE=0.47; p=0.04). The associations of diet quality with glycemic variability were typically more pronounced among those with normoglycemia, but for those with prediabetes, consuming a diet with >1g fiber for every ∼9g carbohydrates was associated with 7-10% lower time spent above 140mg/dL compared to higher carb-to-fiber ratios (p-trend<0.001). Our findings demonstrate that carbohydrate intake and quality, in addition to overall diet quality, is associated with dynamic fluctuations in glucose levels. Future prospective analysis should examine whether glycemic variability mediates the association between diet and incident diabetes.