Heart failure with preserved ejection fraction (HFpEF) is a growing clinical challenge with limited treatment options. Hypoalbuminemia is common in HFpEF and associated with worse outcomes, yet the benefits of albumin infusion remain unclear. Due to practical limitations of randomized controlled trials (RCTs), we used Target Trial Emulation (TTE) to evaluate the effect of early albumin administration in this population. We conducted a retrospective cohort study using the MIMIC-IV database, emulating an RCT to assess the effect of early albumin infusion (≥ 12.5 g within 24 h of Intensive Care Unit (ICU) admission) on 365-day mortality among ICU patients with HFpEF. A cloning-censoring-weighting approach was applied to simulate randomization and adjust for confounding and selection bias using inverse probability of treatment (IPTW) and censoring (IPCW) weights. A marginal weighted Cox model was fitted, and multiple sensitivity, subgroup, dose–response, and negative-control analyses were performed to assess robustness. A total of 7,269 ICU patients with HFpEF were included, of whom 807 (11.1%) received ≥ 12.5 g of albumin within 24 h. After weighting, early albumin administration was not associated with improved 365-day survival (HR 1.07, 95% CI 0.87–1.32). Exploratory subgroup analyses suggested a possible association between early albumin infusion and reduced mortality among patients with albumin > 3.5 g/dL (HR 0.52, 95% CI 0.28–0.95), with no corresponding signal in those ≤ 3.5 g/dL. Reinforcement analyses within the > 3.5 g/dL subgroup—including alternative exposure definitions, dose–response modeling, E-value assessment (E-value 3.27), and a negative-control outcome (ΔCreatinine)—supported that this subgroup-specific association was consistent, though exploratory. Early albumin infusion does not improve long-term survival in the overall ICU HFpEF population. A subgroup association was observed among patients with serum albumin > 3.5 g/dL; however, this finding is hypothesis-generating only and requires prospective validation. Not applicable. This study is a retrospective observational study based on MIMIC-IV database and is not considered a clinical trial according to ICMJE criteria.
Wang et al. (Fri,) studied this question.