Trends in dietary macronutrient composition and diet quality among US adults with diagnosed and undiagnosed elevated glycemic status: a serial cross-sectional study

Background Knowledge of glycemic status may affect dietary intake for subjects with diabetes and prediabetes. Objectives We aimed to determine whether trends of macronutrient intake and dietary quality differed by diagnosis of glycemic status among US adults with diabetes and prediabetes. Methods Data from NHANES 1999-2016 were analyzed. Diagnosed cases were established by self-report, and undiagnosed cases were defined by laboratory criteria (glycated hemoglobin >= 5.7%, fasting plasma glucose >= 100 mg/dL, or 2-h oral-glucose-tolerance test >= 140 mg/dL). A difference-in-differences model was used to compare the temporal trends between the 2 groups. Results There were 7502 diagnosed and 12,974 undiagnosed cases with elevated glycemic status included in the study. During 1999-2016, the intake of low-quality carbohydrates was lower, whereas intakes of high-quality carbohydrates, animal protein, plant protein, and unsaturated fat and the Healthy Eating Index 2015 (HEI-2015) score were higher, among the diagnosed cases than among the undiagnosed cases (P < 0.001 for all). However, in the trend analyses from 1999 to 2016, the increase in intake of high-quality carbohydrates was smaller among the diagnosed cases than among the undiagnosed cases (difference: -1.16%; 95% CI: -1.82%, -0.50%; P = 0.001). Moreover, the decrease in low-quality carbohydrate intake was smaller (difference: 0.79%; 95% CI: 0.01%, 1.57%; P = 0.05) and the increase in saturated fat intake was larger (difference: 0.44%; 95% CI: 0.08%, 0.79%; P = 0.02) among the diagnosed cases than among the undiagnosed cases. A significant difference of temporal trends in the HEI-2015 score between the diagnosed and undiagnosed cases was also observed, in favor of the undiagnosed cases (difference: -2.56; 95% CI: -3.71, -1.41; P < 0.001). Conclusions Although dietary habits among adults with diagnosed diabetes and prediabetes were better than those among the undiagnosed cases, these advantages have been diminishing during the past 2 decades.