1University of Leuven, Leuven, BELGIUM2Istituto di Ricerche Farmacologiche Mario Negri, Milan, ITALY3Imperial College London, London, UNITED KINGDOM4University of Glasgow, Glasgow, UNITED KINGDOM5Stony Brook University, New York, USA6Inserm, Nancy, FRANCE
Checking for direct PDF access through Ovid
Objective:Heart failure (HF) is a syndrome that leads to a diminished quality of life and costly hospital admissions, affecting especially the elderly. The Heart ‘Omics’ in Ageing (HOMAGE) database (within the EU FP program 305507 www.homage-HF.eu) was used as a resource to conduct a subject-level meta-analysis to identify risk factors associated with new-onset HF and to develop a risk score for HF.Design and method:Two population-based studies (HealthABC and PREDICTOR) and two studies including patients at risk for cardiovascular disease (ASCOT and PROSPER) were included. Time-to-event analysis was conducted by using Cox proportional hazard models stratified by study. A point-based risk score for assessment of 5-year HF risk in each cohort was derived from the Cox regression model which was evaluated for discrimination (area under the receiver operating characteristic curve [AUC]) and calibration (Grønnesby-Borgan chisquare statistic).Results:In the population cohort, during a follow-up of 6.2 years, 589 participants out of 4438 subjects (mean age 73.4 ± 3.8 years, 51% women) developed HF. In the patient cohort 25059 patients at risk for cardiovascular disease were included (mean age 65.9 ± 9.2 years, 30% women) of which 526 developed HF. Sex (male), higher age, higher body mass index, higher systolic and lower diastolic blood pressure (only in the population cohort), higher heart rate, lower estimated glomerular filtration rate (eGFR), smoking, history of CV disease and use of antihypertensive medication were detected as significant predictors of incident HF. The 5-year HF risk associated with each point total, derived from the Cox model risk estimates of the population cohort, ranged from less than 1% risk for a point total of <= 0, to more than 30% risk for point totals above 13. In the patient cohort, a point total of 1 or less corresponds to a 5-year HF risk of < 1%, while the risk amounts to 30% or more for a point total of >=15.Conclusions:To estimate HF risk in a general and patient population, two risk score models were developed from clinically relevant parameters to aid in risk stratification and contribute to future prevention.