Heart Failure with Preserved Ejection Fraction Trials Have Heterogeneous Control Groups: a Comparison of Kaplan-Meier Curves.

Abstract

Previous studies have evaluated intra-study heterogeneities of heart failure with preserved ejection fraction (HFpEF), but inter-study heterogeneities remain poorly understood. We investigate the heterogeneities of outcomes among control groups of HFpEF trials. We included randomized controlled trials recruiting HFpEF patients with ejection fraction ≥ 40% and reporting Kaplan-Meier curves for at least 36months. The Kaplan-Meier curves of control groups were extracted and calculated for hazard ratios and 95% confidence intervals. Two virtual trials were developed to validate the reliability and accuracy of our method. Of 4161 studies, we included six trials containing 7682 HFpEF patients in control groups. The DIG trial had the highest all-cause mortality, cardiovascular mortality, heart failure mortality, and composite endpoints of cardiovascular mortality and heart failure hospitalization (all p < 0.001). The TOPCAT trial had the lowest all-cause mortality, cardiovascular mortality, heart failure hospitalization, and composite of cardiovascular mortality and heart failure hospitalization (all p < 0.001). Adoption of different ejection fraction cut-off values for HFpEF diagnosis did not significantly change the outcomes of control groups in the DIG trial (45% vs. 50%: hazard ratio, 1.05, 95% confidence interval, 0.97-1.13, p = 0.271), or in the CHARM-Preserved trial (40% vs. 50%: hazard ratio, 1.01, 95% confidence interval, 0.93-1.09, p = 0.864) during 36-month follow-up. The control groups of HFpEF trials have heterogeneous outcomes. Future trials should consider these heterogeneities when designing protocols.