In-hospital SGLT2 inhibitor initiation, prescribing gaps, and 30-day all-cause readmission in heart failure with reduced ejection fraction: a US post-guideline cohort study

BACKGROUND: Heart failure accounts for more than one million US hospitalizations annually, with 30-day all-cause readmission approaching 25% and triggering CMS Hospital Readmissions Reduction Program penalties. The 2022 ACC/AHA/HFSA guideline and the 2023 ESC focused update elevated SGLT2 inhibitors to Class I therapy for heart failure with reduced ejection fraction (HFrEF) [1, 2]. The DAPA ACT HF-TIMI 68 prespecified meta-analysis demonstrated reductions in cardiovascular death or worsening heart failure (HR 0.71) and all-cause mortality (HR 0.57). Real-world prescribing patterns and 30-day readmission outcomes in the post-guideline US era are not well characterized. The relative contribution of clinical stability variables versus co-prescribed guideline-directed medical therapy (GDMT) to confounding has not been directly quantified in this setting. METHODS: We conducted a retrospective cohort study at four NYU Langone Health hospitals from January 2023 to January 2026. Adults with a primary heart failure discharge diagnosis were included. The prespecified primary analysis was in the HFrEF subgroup (LVEF ≤ 40%). The primary outcome was 30-day all-cause readmission. Stabilized inverse probability of treatment weighting (IPTW) was the primary adjustment, with overlap weighting (ATO) as sensitivity analysis. Hierarchical logistic regression decomposed the confounding contribution of clinical stability parameters relative to GDMT. The E-value assessed robustness to unmeasured confounding. RESULTS: Among 438 patients, 122 (27.9%) received in-hospital SGLT2 inhibitor initiation. The HFrEF rate was 41.6%, a sixfold increase from 6.6% reported in INSIGHT-HF (2020-2021). Patients with prior heart failure hospitalization received SGLT2 inhibitors at 11.4% versus 29.7% in those without (p < 0.001). In HFrEF (n = 221), 30-day readmission was 12.1% versus 31.8% (crude OR 0.29, 95% CI 0.14-0.61). The primary IPTW estimate was OR 0.34 (95% CI 0.13-0.91, p = 0.032). Sensitivity analyses were directionally consistent. Clinical stability parameters contributed only 9.3% confounding attenuation; GDMT was the dominant confounder. CONCLUSIONS: In a contemporary US post-guideline cohort, in-hospital SGLT2 inhibitor initiation reached 41.6% in HFrEF but remained low in patients with recent heart failure hospitalization. In-hospital SGLT2 inhibitor initiation was associated with lower 30-day all-cause readmission, though initiation was strongly bundled with discharge GDMT optimization and cannot be distinguished from a GDMT optimization effect with this study design. These findings should be considered hypothesis-generating. Because short-term safety events and post-discharge persistence were not systematically captured, these findings should not be interpreted as establishing the benefit-risk profile of inpatient SGLT2 inhibitor initiation. The prescribing gap in high-risk patients is an actionable quality-improvement target.