AHA Scientific Sessions 2026 Abstract · Nov 2, 2026
Evaluating the Performance of PREVENT-HF Risk Equations Across HFpEF vs HFrEF
By Mounsey L, Ribeiro P, Wang D et al.
Introduction: The American Heart Association’s PREVENT-HF risk equations estimate 10-year risk of overall heart failure (HF), without distinguishing preserved vs reduced ejection fraction (HFpEF; HFrEF). While HF subtypes share certain risk factors, they have unique pathophysiology and treatment. Accurate estimation of subtype-specific risk may better inform tailored prevention and treatment strategies. Research Question: How does the performance of the PREVENT-HF risk equations compare in forecasting 10-year incident HFpEF and HFrEF? Methods: We examined two longitudinal primary-care electronic health record-based samples across two independent hospital networks (Beth Israel Deaconess Medical Center [BIDMC], Mass General Brigham [MGB]). We included adults age 30-79 years without prevalent cardiovascular disease or HF. We assessed performance of the PREVENT-HF equations in predicting incident HFrEF and HFpEF, including model discrimination (Harrell’s c-statistic) and calibration (Integrated Calibration Index [ICI], with a value of 0 indicating perfect calibration). Results: The BIDMC sample included 29,815 individuals (55% women, 50 ± 13 years), of whom 259 developed HF (51% HFpEF) over a median of 7.6 ± 3.4 years of follow up; MGB included 138,999 (62% women, 52 ± 13 years), of whom 1050 developed HF (56% HFpEF) over a median of 6.3 ± 3.5 years of follow-up. The PREVENT-HF model demonstrated better discrimination for HFpEF than HFrEF across both samples ( Figure ). In BIDMC, the c-statistic for HFpEF was 0.894 (95% CI 0.871, 0.916) versus 0.819 (95% CI 0.779, 0.859) for HFrEF; in MGB, the c-statistic for HFpEF was 0.850 (95% CI 0.834, 0.866), compared 0.773 (95% CI 0.747, 0.800) for HFrEF. Calibration was slightly better for HFpEF than HFrEF in MGB (ICI 0.025 [95% CI 0.024, 0.025] vs. 0.028 [95% CI 0.027, 0.029]), yet similar between subtypes in BIDMC (HFpEF: 0.025 [95% CI 0.023, 0.025]; HFrEF: 0.026 [95% CI 0.025, 0.027]). Conclusions: The PREVENT-HF model discriminates risk for HFpEF better than HFrEF, with similar calibration between subtypes. Development of subtype-specific models may improve risk prediction and inform the use of emerging preventive and therapeutic strategies.