Early empagliflozin initiation associated with lower mortality after heart failure hospitalization in observational cohort
In a US cohort of 99,738 patients hospitalized for heart failure (mean age 78 years), early empagliflozin initiation (during hospitalization or within 7 days post-discharge) was associated with lower 1-year all-cause mortality (RR 0.73) compared to delayed or no initiation, with absolute 1-year mortality of 24.7% versus 33.6%. Early initiation was also associated with reduced worsening HF events (RR 0.89), HF rehospitalizations (RR 0.86), ED visits (RR 0.87), and all-cause hospitalizations (RR 0.91).
What this changes
The finding has not established an approved change to our evidence assessment.
Source date: Sep 22, 2026 · Discovered: Sep 22, 2026 · Assessed: — · Research update
What it does not prove
- Limitations have not yet been verified. Read the primary source.
Extracted from supplied source material; model confidence is not a measured probability of correctness.
Study details
- Design
- Cohort
- Sample size
- 99,738
- Duration
- 12 months
- Population
- Adults hospitalised for heart failure between September 2021 and December 2023 from the Optum de-identified Clinformatics Data Mart (mean age 78.0 years; 54.2% female)
- Primary outcome
- All-cause mortality at 1 year post-discharge
- Effect size
- Early vs delayed/no initiation for all-cause mortality: RR 0.73 (95% CI: 0.69-0.78); Intermediate vs delayed/no initiation for all-cause mortality: RR 0.93 (95% CI: 0.89-0.97); Early vs delayed/no for worsening HF events: RR 0.89 (95% CI: 0.84-0.94); Early vs delayed/no for HF rehospitalisations: RR 0.86 (95% CI: 0.81-0.92); Early vs delayed/no for ED visits: RR 0.87 (95% CI: 0.82-0.93); Early vs delayed/no for all-cause hospitalisations: RR 0.91 (95% CI: 0.88-0.94)
- Significance
- Significant
- Journal
- ESC Heart Failure
- Published
- Sep 22, 2026
Sources
Primary source
- ESC heart failure ↗Peer reviewed
Assessment history
9/23/2026 · SUPERSEDED: Source updated; assessment withdrawn pending review.
Discussion (0)
One thread per scientific event, no matter how many outlets covered it.
No comments yet
Be the first to weigh in on this evidence.