Associations of temporal protein patterns with diabetes and glycemic measures

This study aimed to examine temporal protein patterns (i.e., protein intake timing across the day) in American adults and their associations with diabetes, BMI and glycaemic measures.

Total protein intake at eating occasions was estimated from the National Health and Nutrition Examination Survey 2017-2020 one-day dietary recall data (≥ 20 y; n = 7625). Latent variable mixture models were used to identify temporal patterns based on hourly protein intake (g). Adults with fasting glucose ≥ 7.0 mmol/L, poor glycaemic control (HbA1c ≥ 6.5%), diabetic medications, or previous diabetes diagnosis were classified as having diabetes. Regression models were used to examine associations for temporal protein patterns with diabetes prevalence, BMI and glycaemic measures (fasting glucose, HbA1c, insulin, Homeostatic Model Assessment of Insulin Resistance), adjusted for multiple confounders.

Three latent classes for men's and women's temporal protein patterns were identified. Class 1 had higher conditional probabilities of eating protein at 18:00 h, while Class 2 tended to eat protein 1-h later than Class 1 and had higher overall protein intake (p < 0.001). Class 3 was characterised by variable protein intake timing and had lower overall protein intake than other classes (p < 0.001). Men's Class 1 had lower probabilities for poor glycaemic control (5.9%), but associations were attenuated after adjusting for BMI. There were no associations between temporal protein patterns with diabetes, BMI and other glycaemic measures.

Temporal protein patterns were not associated with diabetes and glycaemic measures. Future prospective studies may better examine the effects of temporal protein patterns on diabetes by considering variations in protein sources.