Prediction and classification of cardiovascular disease risk in older adults with diabetes

We sought to derive and validate a cardiovascular disease (CVD) prediction algorithm for older adults with diabetes, and evaluate the incremental benefit of adding novel circulating biomarkers and measures of subclinical atherosclerosis. As part of the Cardiovascular Health Study (CHS), a population-based cohort of adults aged a parts per thousand yen65 years, we examined the 10 year risk of myocardial infarction, stroke and cardiovascular death in 782 older adults with diabetes, in whom 265 events occurred. We validated predictive models in 843 adults with diabetes, who were followed for 7 years in a second cohort, the Multi-Ethnic Study of Atherosclerosis (MESA); here 71 events occurred. The best fitting standard model included age, smoking, systolic blood pressure, total and HDL-cholesterol, creatinine and the use of glucose-lowering agents; however, this model had a C statistic of 0.64 and poorly classified risk in men. Novel biomarkers did not improve discrimination or classification. The addition of ankle-brachial index, electrocardiographic left ventricular hypertrophy and internal carotid intima-media thickness modestly improved discrimination (C statistic 0.68; p = 0.002) and classification (net reclassification improvement [NRI] 0.12; p = 0.01), mainly in those remaining free of CVD. Results were qualitatively similar in the MESA, with a change in C statistic from 0.65 to 0.68 and an NRI of 0.09 upon inclusion of subclinical disease measures. Standard clinical risk factors and novel biomarkers poorly discriminate and classify CVD risk in older adults with diabetes. The inclusion of subclinical atherosclerotic measures modestly improves these features, but to develop more robust risk prediction, a better understanding of the pathophysiology and determinants of CVD in this patient group is needed.