Relations of lipid concentrations to heart failure incidence: the Framingham Heart Study.

Abstract

The relations of lipid concentrations to heart failure (HF) risk have not been elucidated comprehensively. In 6860 Framingham Heart Study participants (mean age, 44 years; 54% women) free of baseline coronary heart disease, we related high-density lipoprotein cholesterol (HDL-C) and non-HDL-C to HF incidence during long-term follow-up, adjusting for clinical covariates and myocardial infarction at baseline and updating these at follow-up examinations. We evaluated dyslipidemia-specific population burden of HF by calculating population attributable risks. During follow-up (mean of 26 years), 680 participants (49% women) developed HF. Unadjusted HF incidence in the low (<160 mg/dL) versus high (> or =190 mg/dL) non-HDL-C groups was 7.9% and 13.8%, respectively, whereas incidence in the high (> or =55 [men], > or =65 [women] mg/dL) versus low (<40 [men], <50 [women] mg/dL) HDL-C groups was 6.1% and 12.8%, respectively. In multivariable models, baseline non-HDL-C and HDL-C, modeled as continuous measures, carried HF hazards (confidence intervals) of 1.19 (1.11 to 1.27) and 0.82 (0.75 to 0.90), respectively, per SD increment. In models updating lipid concentrations every 8 years, the corresponding hazards (confidence intervals) were 1.23 (1.16 to 1.31) and 0.77 (0.70 to 0.85). Participants with high baseline non-HDL-C and those with low HDL-C experienced a 29% and 40% higher HF risk, respectively, compared with those in the desirable categories; the population attributable risks for high non-HDL-C and low HDL-C were 7.5% and 15%, respectively. Hazards associated with non-HDL-C and HDL-C remained statistically significant after additional adjustment for interim myocardial infarction. Dyslipidemia carries HF risk independent of its association with myocardial infarction, suggesting that lipid modification may be a means for reducing HF risk.