Risk factors for mortality in acute heart failure. A classification and regression tree analysis

Abstract

Introduction: The early estimation of risk in acute heart failure may help in the taking of clinical decisions. Objective: To identify the risk factors associated with in-hospital mortality in patients with acute heart failure. Methods: An analysis was performed on a retrospective cohort of patients greater than 18 years admitted to a tertiary hospital due acute heart failure between the years 2010 and 2016. Results: A total of 247 patients were included, with a mean age of 62.8 years, and of which 60% were male. The large majority (84%) of the patients had heart failure with a reduced ejection fraction (median 25%). The in-hospital mortality was 9.3%, and the accumulated rate at 30 days and 6 months after hospital discharge was 10.9% and 14.1%, respectively. The two predictors associated with in-hospital death was a blood urea nitrogen (BUN) > 37 mg/dL (OR: 10.8; 95% CI: 4.10-28.8) and a systolic blood pressure (SBP) ≤ 125 mmHg (OR: 3.42; 95% CI: 1.15-10.0). The classification and regression tree (CART) model identified elevated levels of as the best predictor of mortality, followed by a decreased systolic pressure (< 97 mmHg), and finally due to elevated creatinine levels (≥ 1,75 mg/dL). Conclusion: The analysis using the classification and regression tree (CART) model can provide an early classification of the probability of death by a risk tree that includes BUN ≥ 32.5 mg/dL, systolic pressure < 97 mm Hg, and creatinine levels ≥ 1.75 mg/dL.