Better risk assessment with glycated hemoglobin instead of cholesterol in CVD risk prediction charts

Abstract

Traditional risk charts for the prediction of cardiovascular disease (CVD) include cholesterol parameters. We evaluated how models predict fatal CVD when cholesterol is replaced by glucose parameters. We used data from NHANES III, a US survey conducted 1988-1994 (follow-up until 2006); 15,454 participants (1,716 CVD deaths) were included. Based on the ESC SCORE method, we used age, sex, blood pressure, smoking and either of the following: (1) total cholesterol, (2) total-to-HDL-cholesterol, (3) glucose, (4) glycated hemoglobin (A1C). Scaled Brier score (BS), Nagelkerke's R(2) (NR) and integrated discrimination improvement (IDI) were used for model comparison. The ranking (best to worst) was: A1C (BS=11.62%; NR=0.0865; IDI=0.0091), glucose (11.16%; 0.0734; 0.0067), total-to-HDL-cholesterol (9.97%; 0.0547; 0.0010), cholesterol (9.75%; 0.0484; 0, reference). Differences between models with cholesterol and glucose or A1C were statistically significant. This study suggests the use of A1C instead of cholesterol parameters in charts to assess CVD risk.