OBJECTIVES: To quantify risks of all-cause readmission associated with differential timing of post-discharge sodium-glucose cotransporter-2 (SGLT2) inhibitor initiation in acute heart failure with type 2 diabetes.

METHODS: This retrospective cohort study utilized administrative databases in Ontario, Canada, housed at ICES. Individuals ≥66 years with type 2 diabetes who were discharged from acute care hospitals for heart failure as a major responsible diagnosis of an admission between July 1st, 2016, and March 31st, 2020, were selected. To identify new users, individuals dispensed an SGLT2 inhibitor in the year before admission were excluded. Individuals with contraindication(s) for SGLT2 inhibitors were excluded. We investigated 4 post-discharge exposures: SGLT2 inhibitor initiation ≤3 days, 4-90 days, 91-180 days, or delayed for ≥180 days. The “clone-censor-weight” approach, with a target trial emulation framework, was used to address time-related biases. Selection bias due to artificial censoring was adjusted by inverse probability weighting. The outcome was unplanned readmission for any cause within 1 year from discharge. A weighted pooled logistic regression conditional on baseline covariates was used to estimate cumulative incidence functions, and percentile confidence intervals (CIs) for risk differences (RDs) were derived from 200 non-parametric bootstraps.

RESULTS: There were 9,641 eligible individuals. After cloning and artificial censoring, there were 38,564 clones, 9,872 person years, and 13,964 events. Compared to delayed initiation for ≥180 days, initiation within 3 days was associated with an 27.0% absolute risk reduction in readmission risk over 1 year (RD [95% CI] = -27.0% [-31.7%, -21.2%]). Initiation 4-90 days post-discharge (RD [95% CI] = -1.2% [-2.9%, 0.5%]) and initiation 91-180 days post-discharge showed no significant difference (RD [95% CI] = 0.0% [-1.6%, 1.8%]).

CONCLUSIONS: This real-world evidence suggests that early SGLT2 inhibitor initiation post-discharge for acute heart failure was associated with reduced readmission risk over 1 year in people with type 2 diabetes.