UC Irvine

BACKGROUND: There is significant heterogeneity in cardiovascular disease (CVD) risk among patients with diabetes mellitus (DM). OBJECTIVES: The purpose of this study was to develop risk scores for total CVD and its components from a contemporary pooled, observational cohort of U.S. adults with DM. METHODS: CVD-free adults with DM aged 40 to 79 years were pooled from 4 U.S. population-based cohorts (CARDIA [Coronary Artery Risk Development in Young Adults], Framingham Offspring, Jackson Heart Study, and the MESA (Multiethnic Study of Atherosclerosis) studied since 2000. Baseline DM-specific and non-DM-specific CVD risk factors were evaluated as predictors. We developed 10-year DM Risk Scores (DMRS) for total CVD, atherosclerotic CVD (ASCVD), coronary heart disease (CHD), heart failure (HF) and stroke. Score performance was validated internally and externally. RESULTS: We included 2,174 adults with DM mean age 59.2 ± 10.5 years, 55.4% female and 47.5% Black followed up to 10 years. Age, sex, HbA1c, creatinine, systolic blood pressure, DM medication, and smoking were the most important predictors. The DMRS had good internal discrimination (c-statistics 0.72, 0.72, 0.72, 0.79 and 0.73 for CVD, ASCVD, CHD, HF, and stroke) and calibration (calibration slopes 0.93, 0.95, 0.93, 0.98, and 0.89 for CVD, ASCVD, CHD, HF, and stroke; Greenwood Nam-DAgostino calibration tests were significant for CHD (P < 0.01) and CVD (P < 0.05) but not for ASCVD, HF, and stroke). From external validation in 2 other cohorts, the DMRS outperformed current risk scores. CONCLUSIONS: Our U.S. pooled cohort DMRS for predicting CVD events demonstrated good predictive performance for assessing CVD risk in adults with DM.