Association of Obstructive Airway Disease and Lung Function with Incident Heart Failure, and Optimal Prediction of Heart Failure in Community Settings: The Atherosclerosis Risk in Communities Study

Background: This epidemiologic investigation of heart failure (HF) has two components, one focused on airway disease as a putative antecedent factor to HF and the other centered on the prediction of HF as a means toward reducing the growing burden of HF in the population. Methods: Forced Expiratory volume-1 second (FEV1) and covariates were measured for the ARIC cohort in 1987-89. Incident HF was ascertained annually from hospital records and death certificates. Cox proportional hazards models were used to derive a risk score to predict 10 years risk of HF. Area under curve (AUC) and Net Reclassification Improvement (NRI) were estimated as measures of discrimination. Results: Over an average follow-up of 14.9 years, 1369 (10%) ARIC participants free of HF at baseline had incident HF. The hazard ratios (HRs) for HF increased monotonically over descending quartiles of FEV1. The associations were seen in each of cigarette smoking strata, inclusive of never-smokers. After multivariable adjustment for traditional cardiovascular risk factors at baseline, the HRs of HF and their 95% confidence intervals (CI) comparing the lowest with the highest quartile of FEV1 were 3.91(2.40, 6.35) for white women, 3.03(2.12, 4.33) for white men, 2.11(1.33, 3.34) for black women and 2.23(1.37, 3.59) for black men. The multivariable adjusted hazards of HF were higher in those with FEV1/FVC < 70% vs. [greater than or equal to] 70%: HR 1.42 (95% CI 1.22, 1.68). The ARIC HF risk score included information easily available to the primary care physician including COPD. The estimated AUC of the ARIC HF risk score was 0.810(optimism-corrected = 0.808), 95% CI = 0.807, 0.813. It was higher than AUC estimated using variables from the Framingham risk score(0.762) and the ABC risk score(0.784). Overall classification using the ARIC HF risk score improved for 23.5% individuals relative to the Framingham, and 12.8% relative to the Health-ABC classification. Conclusions: In this population-based cohort, low FEV1, and obstructive respiratory illness were strongly and independently associated with incident HF. The underlying mechanisms may include diastolic-dysfunction, cor-pulmonale, silent-CHD, and require exploration. The ARIC HF risk score performs better than extant scores and may improve risk prediction of HF in the community.