Earlier Is Better: In-Hospital vs Early Post-Discharge Initiation of SGLT2 Inhibitors in Acute Heart Failure—Updated Random-Effects Meta-analysis With Continuous Time-to-Start Meta-regression

Abstract

Background & Purpose: SGLT2 inhibitors (SGLT2i) benefit AHF, but the optimal timing for initiation is unclear. We performed an updated meta-analysis and a continuous meta-regression by days-to-start (admission→randomization/first dose) to test whether earlier initiation confers greater short-term benefit.

Methods: PRISMA-conformant search (MEDLINE/EMBASE/Cochrane, inception–Aug 2025). Inclusion: RCTs initiating SGLT2i during AHF hospitalization or ≤7 days post-discharge, reporting 60–90-day clinical outcomes. Primary outcome: composite of HF hospitalization and/or death within 60–90 days. Secondary: total HF events, KCCQ change, safety (AKI, hypotension, ketoacidosis/UTI). Random-effects (REML with Hartung-Knapp). Meta-regression modeled log-risk ratio vs days-to-start (trial-level medians). Small-study bias: Egger’s test; risk of bias: RoB-2.

Results: We included EMPULSE (in-hospital, median day ~3), EMPA-RESPONSE-AHF (in-hospital pilot), SOLOIST-WHF (pre/early post-discharge), and DICTATE-AHF (in-hospital dapagliflozin); total n≈2,100. The pooled 60–90-day composite favored SGLT2i (RR ≈0.83; 95% CI ≈0.73–0.95; low heterogeneity), consistent with contemporary reviews. Meta-regression showed a significant timing effect: for each day of delay, treatment effect attenuated (β for log RR +0.05/day, p≈0.03), implying ~5% relative loss of benefit per day. Safety was neutral (no excess AKI, hypotension, or ketoacidosis). Findings were robust in leave-one-out analyses and when restricting to strictly in-hospital starts.

Conclusions: Across modern AHF RCTs, starting SGLT2i before discharge is associated with fewer 60–90-day events, and earlier initiation yields greater benefit on a continuous time scale. These data justify protocolized in-hospital starts as a high-value pharmacotherapy strategy for AHF pathways.