Longitudinal Metabolic Risk Factors Are Associated With Microvascular Complications

Objective Determine associations between summaries of longitudinal metabolic risk factors and microvascular complications. Methods We performed a secondary analysis of participants with and without diabetes enrolled within the Atherosclerosis Risk in Communities Study that completed metabolic risk factors assessments four times across 11 years. We calculated slope and area under the curve (AUC) as summaries of longitudinal data for fasting glucose, systolic blood pressure (SBP), high‐density lipoprotein, triglycerides and waist circumference. We assessed Chronic Kidney Disease (CKD) with estimated Glomerular Filtration Rate, Peripheral Neuropathy (PN) with monofilament testing and Retinopathy (RN) with the Early Treatment Diabetic Retinopathy Study scale. Area under the receiver operating characteristic curve (AUROC) determined discriminatory capability of models based on slope and AUC compared to models based on metabolic risk factors at baseline. Results We identified 4204, 2788 and 1792 persons with assessments of CKD, PN and RN, respectively. The mean age was 52.3, 57.1% were female, and 21.0% had diabetes. Logistic regression revealed fasting glucose AUC (1.09, 1.05–1.15) and SBP AUC (1.27, 1.19–1.36) associated with CKD, triglyceride AUC (1.14, 1.04–1.25) and waist AUC (1.11, 1.07–1.15) associated with PN and fasting glucose AUC (1.13, 1.04–1.22) associated with RN. Longitudinal metabolic summaries only improved AUROC by 0.001–0.01 versus models with cross‐sectional metabolic assessments. Conclusion AUCs of longitudinal metabolic risk factors, but not slopes, were associated with microvascular complications. This finding highlights the importance that sustained metabolic impairment has for the development of microvascular complications. Surprisingly, however, in this cohort, summaries of longitudinal metabolic risk factors did not meaningfully add discriminatory capability beyond models based on metabolic assessments at baseline.