A Simple Clinical Risk Score to Predict Post-Discharge Mortality in Chinese Patients Hospitalized with Heart Failure

Abstract

Background

Cardiovascular diseases are the leading causes of death in China. However, present efforts to identify the risk factors for death in patients hospitalized with heart failure (HF) are primarily focused on in-hospital mortality and 30-day mortality in the United States. Thus, a model similar to the model used for predicting the risk in patients considered for cardiovascular surgical procedures is needed to evaluate the risk of the patients admitted with a diagnosis of HF.

Objective

To identify variables that can predict post-discharge one-year HF mortality and develop a risk score to assess the risk of dying within one year.

Methods

In the present study, 1,742 Chinese patients with HF were randomly divided into two groups: a derivation sample group and a test sample group. A Markov Chain Monte Carlo simulation method was used to identify variables that can predict the one-year post-discharge mortality. Variables with a frequency of >1% in the bivariate analysis and that were considered clinically meaningful were eligible for further modeling analyses. The posterior probability that a variable was statistically and significantly associated with the outcome was calculated as the total number of times that the variable’s 95% CI did not overlap with 1 (i.e., the reference point) divided by the total number of iterations. A variable with a probability of 0.9 or higher was considered a robust risk factor for predicting the outcome, and this was included in the final variable list. The level of statistical significance adopted was 5%.

Results

Five variables that could robustly predict the one-year post-discharge mortality were identified: age, female gender, New York Heart Association functional classification score >3, left atrial diameter, and body mass index. Both derivation and test models had a receiver operating curve area of 0.79. These selected variables were used to assess the one-year HF mortality risk score, and these were divided into three groups (low, moderate, and high). The high-risk group corresponds to nearly 86% of the deaths, while the moderate group corresponds to 12% of the deaths.

Conclusion

A simple 5-variable risk score can be used to assess the one-year post-discharge mortality of hospitalized Chinese patients with HF.

Heart Failure; Propensity Score; Mortality; Patient Discharge; Epidemiology