Abstract

Aims: We investigated the prevalence, clinical characteristics, and prognosis of patients with heart failure (HF) with improved ejection fraction (HFimpEF). Methods and results: We used data from BIOSTAT‐CHF including patients with a left ventricular ejection fraction (LVEF) ≤40% at baseline who had LVEF re‐assessed at 9 months. HFimpEF was defined as a LVEF >40% and a LVEF ≥10% increase from baseline at 9 months. We validated findings in the ASIAN‐HF registry. The primary outcome was a composite of time to HF rehospitalization or all‐cause mortality. In BIOSTAT‐CHF, about 20% of patients developed HFimpEF, that was associated with a lower primary event rate of all‐cause mortality (hazard ratio [HR] 0.52, 95% confidence interval [CI] 0.28–0.97, p = 0.040) and the composite endpoint (HR 0.46, 95% CI 0.30–0.70, p < 0.001) compared with patients who remained in persistent HF with reduced ejection fraction (HFrEF). The findings were similar in the ASIAN‐HF (HR 0.40, 95% CI 0.18–0.89, p = 0.024, and HR 0.29, 95% CI 0.17–0.48, p < 0.001). Five independently common predictors for HFimpEF in both BIOSTAT‐CHF and ASIAN‐HF were female sex, absence of ischaemic heart disease, higher LVEF, smaller left ventricular end‐diastolic and end‐systolic diameter at baseline. A predictive model combining only five predictors (absence of ischaemic heart disease and left bundle branch block, smaller left ventricular end‐systolic and left atrial diameter, and higher platelet count) for HFimpEF in the BIOSTAT‐CHF achieved an area under the curve of 0.772 and 0.688 in the ASIAN‐HF (due to missing left atrial diameter and platelet count). Conclusions: Approximately 20–30% of patients with HFrEF improved to HFimpEF within 1 year with better clinical outcomes. In addition, the predictive model with clinical predictors could more accurately predict HFimpEF in patients with HFrEF.