Discharge risk scoring method for predicting mortality in hospitalized chronic heart failure patients with severe systolic dysfunction

Abstract

Objective Prognostic risk stratifi cation in heart failure is crucial to guide clinical decision-making. The aim of our study was to develop a prognostic discharge risk score model to predict all-cause mortality for chronic heart failure patients with multiple co-morbidities and severe systolic dysfunction.

Methods and results A multivariable logistic regression model was developed with the use of data on clinical, laboratory, imaging and therapeutic fi ndings of 630 patients with advanced systolic heart failure. A risk score model was developed based on multiplying the β-coeffi cient number of each variable in the multivariable model. The model performance was evaluated by concordance index and internally validated by the bootstrapping method. 313 patients (49.7%) of the cohort died during a median follow-up duration of 54 months. Median age was 66 years, 37% were female, 26% had atrial fi brillation and 40% had diabetes mellitus. The mean left ventricular ejection fraction (EF) was 25 ± 10% and 264 patients (42%) had left ventricular EF ≤ 20%. Independent predictors of mortality were older than 70 years, orthopnoea, previous hospitalisations, lack of renin-angiotensin system inhibitor therapy at discharge, hyperuricaemia (> 7 mg/dl) and haemoglobin level (< 10 g/dL). Discharge risk score identifi ed low-, intermediate- and high-risk individuals with 18%, 40% and 52% mortality rates, respectively. The risk score had a discrimination ability with a concordance index of 0.70.

Conclusions In a large heart failure cohort, including patients with severe systolic dysfunction and having multiple co-morbidities, a simple discharge risk score with non-invasive and easy-to-obtain variables during hospital admission represents a valuable tool for risk assessment.

Key Words:

• Heart failure
• mortality
• prognosis
• risk score.