Background: Acute heart failure exacerbations are a leading cause of morbidity and mortality and are associated with high rates of hospital admissions and healthcare expenditure. As part of a broader effort to improve heart failure care at our hospital, this project aims to characterize heart failure type prevalence, mortality, and readmission rates to benchmark against automated data and to provide a baseline to which to compare future intervention effects.
Methods: This is a retrospective chart review of 692 patients discharged in August 2024 from a Department of Medicine (DOM) service at TJUH. Patients were identified as a patient with heart failure if they received an intravenous diuretic and had a diagnosis of heart failure by manual chart review. Service type, mortality, and heart failure type were identified by manual chart review. Expected mortality was obtained from Vizient all-patient data cohort. Index-hospital 30-day readmission was obtained through automated export. Registry comparison data was sourced from our Vizient heart failure data cohort, not restricted to DOM. Analysis was performed in Excel.
Results: Of 692 patients, 73 (10.5%) were heart failure admissions with intravenous diuresis. Of these, 48 (66%) were cared for on a cardiology service and 25 (34%) were cared for on a hospital medicine service (Vizient data suggests 4:1 cardiology to hospital medicine ratio). The inpatient mortality rate was 5.5% compared to Vizient 3.3% (2 of 59). The observed to expected (O/E) mortality was 0.96 (Vizient 1.53). The 30-day readmission to the index hospital was 15% as compared to Vizient data of 14.8% (9 of 61). Regarding the type of heart failure, 38 (52%) had HFrEF, 25 (34%) had HFpEF, 7 (10%) had HFmrEF, and 2 (3%) had HFimpEF. One patient was not classified.
Conclusions: Despite likely missing some heart failure admissions with our intravenous diuretic and DOM discharge service restriction, we still captured more heart failure admissions than our Vizient database (73 vs 59-61) suggesting limited documentation, coding, database reliability. The majority of our patients are cared for on cardiology services, but a large minority are seen by hospital medicine. While our manual review mortality O/E was < 1, the Vizient O/E was 1.53 suggesting opportunities for both improved patient care and improved documentation, coding to ensure accurate denominators. The readmission rate was consistent; it is restricted to index-hospital readmission so cannot be compared to published Medicare data that captures OSH admissions but provides a base to explore interventions to drive down readmissions at our institution. The majority of patients had HFrEF/HFimpEF (40/73, 55%) compared to HFmrEF/HFpEF (32/73, 44%) which is a reversal of published community ratios (HFpEF > HFrEF) suggesting our HFrEF/HFimpEF population is at higher risk for admission and thus, likely, worthy of targeted intervention to improve care. Next steps include using this manual data to adjust and validate our automated process and outcome measures so that we can reliably assess impact of planned process and systems improvement interventions on key process and outcomes of interest. We hope it provides a template for future improvement projects to identify key baseline data and outsized opportunities for targeted intervention and validation of automated data sources prior to pre-post improvement impact analysis to ensure a methodical and evidenced-based approach to improving patient care.