Phenotypes and Outcomes of Heart Failure First Diagnosed During Sepsis: A Single-Center Retrospective Cohort Study...

Background: Sepsis-associated cardiac dysfunction has historically been characterized as predominantly systolic, yet the burden of heart failure with the preserved ejection fraction (HFpEF) phenotype among patients with heart failure (HF) first documented during sepsis remains poorly defined. Methods: In a single-center retrospective cohort of adults (≥18 years) admitted to a tertiary-care hospital between 1 January 2022 and 31 December 2024, we identified patients whose first sepsis diagnosis (ICD-10) coincided with their first HF diagnosis (ICD-10 I50.x) during the same encounter, with no prior HF in the EHR. Patients were classified by ICD-10 subtype as HFpEF, HFrEF, combined, or unspecified; the unspecified group underwent echocardiographic reclassification when LVEF was documented. Demographics, selected comorbidities, peak inflammatory markers, and unadjusted in-hospital and subsequent-encounter outcomes were compared between HFpEF and HFrEF. Results: Of 924 patients with HF first documented during sepsis, 438 (47.4%) were ICD-10-coded HFpEF and 405 (43.8%) HFrEF. HFpEF patients were older (74.0 vs. 69.2 years; p < 0.001) and more often female (53.0% vs. 37.0%; p < 0.001), with higher prevalence of COPD exacerbation (21.5% vs. 13.1%; p = 0.004) and hypothyroidism (26.3% vs. 17.5%; p = 0.034). HFrEF patients exhibited higher peak lactate (2.88 vs. 2.48 mmol/L; p = 0.009) and procalcitonin (8.52 vs. 7.24 ng/mL; p = 0.011). Unadjusted in-hospital mortality (28.1% vs. 27.7%; p = 0.95), ICU admission (75.8% vs. 74.8%; p = 0.80), and length of stay did not differ. Subsequent encounters were descriptively more frequent after HFpEF (11.0% vs. 6.2%). Conclusions: ICD-10-coded HFpEF and HFrEF occurred at nearly equal frequency among adults with HF first documented during sepsis, challenging the systolic-centric paradigm. Despite distinct demographic and biochemical profiles, unadjusted in-hospital outcomes did not differ. These descriptive findings, limited by administrative-coding-based phenotype classification, the absence of multivariable adjustment, and unmeasured sepsis-severity and comorbidity variables, are hypothesis-generating and support phenotype-aware prospective study of HF arising during sepsis.