Abstract Body (Do not enter title and authors here): Introduction: Although sodium-glucose cotransporter 2 inhibitors (SGLT2i) are guideline-directed medical therapy for heart failure, their safety in patients with acute decompensated heart failure (ADHF) remains uncertain. This study compared the risk of adverse events between early versus late initiation of SGLT2is following ADHF hospitalization.

Methods: We conducted a retrospective cohort study emulating a target trial using linked administrative data from Ontario, Canada. Adults aged ≥66 years discharged after hospitalization for ADHF between April 1, 2016, and March 31, 2021 were included. Patients with prior SGLT2i use or with prior heart failure hospitalizations were excluded. We compared two strategies for initiating SGLT2is: early initiation (within 30 days post-discharge), and late initiation (between days 31 and 365 after discharge). The primary outcome was a composite of eight prespecified adverse events requiring hospitalization or emergency department visit within one year of discharge (diabetic ketoacidosis, genitourinary infections, hypotension/syncope, dehydration, acute kidney injury, and falls/fractures). We applied a clone-censor-weight approach, using inverse probability weighted Cox regression models to estimate cause-specific hazard ratios (csHR) and cumulative incidence functions to account for the competing risk of death. Multiple imputation was used to deal with missing values of serum creatinine.

Results: The mean number of included patients across 32 imputed datasets was 58,744 (median age, 83 years [IQR 77–89]; 53.4% women). In the overall cohort, during the 1-year follow-up, 18.2% experienced at least one adverse event, most commonly genitourinary infections (5.9%) and acute kidney injury (4.9%). At one year, no difference in the risk of adverse events was observed between the early and late initiation strategies (cumulative incidence at 1 year was 14.5% in the early and 18.7% in the late initiation strategies; absolute risk difference: 4.1%; csHR: 0.96; 95% CI 0.90 to 1.01; p=0.13). Sensitivity analyses using alternative definitions of early and late initiation yielded similar results.

Conclusions: In this population-based target trial emulation study, there was no difference in the risk of one-year adverse events with early initiation of SGLT2i within 30 days post-discharge for ADHF compared with delayed initiation. These findings support the safety of early SGLT2i initiation in patients recently hospitalized for ADHF.