Background: Heart failure (HF) affects a large number of Americans with an estimated lifetime risk of over 20% among individuals over the age of 40 years. Heart failure is associated with significant morbidity, mortality and health care cost. Epidemiological data suggests that HF disproportionately affects African Americans. Existing risk prediction models may not accurately predict the risk of incident HF in this population limiting the implementation of screening and preventive strategies. We sought to examine the predictors of incident heart failure in a community-based African American cohort.
Methods: The study population consisted of 2244 African American participants from the Jackson cohort of the Atherosclerosis Risk in Communities (ARIC) study. Participants with prevalent HF and missing data at the baseline were excluded. Cox proportional regression with backward elimination was used to assess the association of risk markers with first clinical HF event as follows: Model 1 - standard cardiovascular risk factors (age, gender, body mass index [BMI], systolic blood pressure, use of anti-hypertensive medication, diabetes status, current smoking, total cholesterol-HDL ratio [TC:HDL], logarithmically transformed-triglyceride); Model 2 - significant standard risk factors and biomarkers (logarithmically transformed-brain natriuretic peptide [BNP], logarithmically transformed-troponin, logarithmically transformed-albumin-creatinine ratio); Model 3 - significant standard risk factors and subclinical disease markers (estimated glomerular filtration rate [eGFR], ankle-brachial index, forced expiratory volume in one second [FEV1], forced vital capacity, left ventricular ejection fraction < 50%, left ventricular mass indexed to height2.7); Model 4 - significant standard risk factors, significant biomarkers, and significant subclinical disease markers.
Results: The mean age and BMI were 58 years and 30 kg/m2 respectively. During a median follow-up of 18.1 years (maximum 19.8), there were 268 incident HF events. Triglyceride from Model 1, Troponin and Albumin-Creatinine ratio from Model 2, and eGFR from Model 3 were eliminated. In the final model, diabetes status (Hazard Ratio [95% Confidence Interval]: 1.76 [1.07, 2.89]), use of anti-hypertensive medication (2.38 [1.40, 4.06]), TC:HDL (1.34 [1.18, 1.52]) and BNP (1.28 [1.14, 1.44]) were significant predictors of incident HF. Higher FEV1 was associated with reduced risk of HF in this study (0.29 [0.11, 0.76]).
Conclusion: In this community-dwelling African American population, diabetes status, use of anti-hypertensive medication, TC:HDL ratio, BNP and FEV1 were independent predictors of incident HF. This information may be useful to identify and target high risk individuals for aggressive preventive interventions.