Background: Patients hospitalized for acute decompensated heart failure (ADHF) are at high risk for early mortality and rehospitalization. Risk stratification of ADHF using clinically available data on admission is increasingly important to integrate with clinical pathways. Our goal was to create a simple method of screening patients upon admission to identify those with increased risk of future adverse events.
Methods: Using ASCEND-HF, a pragmatic clinical trial conducted in 398 sites globally, we developed and validated logistic regression risk models for (a) 30-day mortality/HF rehospitalization, (b) 30-day mortality/all-cause rehospitalization, (c) 30-day all-cause mortality, and (d) 180-day all-cause mortality. Fifty-one candidate variables were evaluated based on prior publications and clinical review. Final models were selected based on stepwise selection with entry and a staying criterion of P < .01. The 30-day mortality model was externally validated, and coefficients were converted to an additive risk score.
Results: Among 7,141 patients, the median age was 67 years, 34% were female, and 80% had a left ventricular ejection fraction <40%. The models had between 5 and 12 risk factors with c-indices ranging from 0.68 to 0.75. A simplified score, including age, systolic blood pressure, sodium, blood urea nitrogen, and dyspnea at rest, discriminated 30-day mortality risk from 0.5% (score 0) to 53% (score 10).
Conclusions: Commonly available clinical variables provide simple risk stratification for clinical outcomes among patients with ADHF, and these models may be considered for integration into routine clinical care.
Copyright © 2015. Published by Elsevier Inc.