Risk Prediction in Patients With Heart Failure: A Systematic Review and Analysis

Study Questions:

What are characteristics and performance of contemporary multivariable statistical models for prediction of death, hospitalization, or both?

Methods:

This was a systematic review. Studies were eligible if they reported multivariable models for prediction of risk of death, hospitalization, or both in individuals with heart failure. For eligible studies, the following information was extracted: study and patient characteristics, candidate variables, final model variables, model discrimination (the ability of a statistical model to distinguish those subjects who experience the outcome from those who do not), calibration (how closely observed estimates of absolute risk agree with expected estimates from the risk prediction model), and internal and external validation.

Results:

The authors identified 43 main models for prediction of death, 10 main models for prediction of hospitalization, and 11 main models for prediction of death or hospitalization. Most studies (65%) either did not specify left ventricular systolic function or included all patients with heart failure. In models for prediction of risk of death, the following variables emerged consistently: age, renal function, blood pressure, blood sodium level, left ventricular ejection fraction, sex, B-type natriuretic peptide (or N-terminal pro-BNP) level, New York Heart Association functional class, diabetes, weight/body mass index, and exercise capacity. The following predictors appeared consistently in models for prediction of risk of hospitalization: age, sex, renal function, cardiovascular disease, and heart rate. Models for death had greater discriminatory ability than did models for hospitalization alone or both death and hospitalization.

Conclusions:

There are multiple models in the literature for predicting death or hospitalization in patients with heart failure. Although these models are heterogeneous, common predictors of outcomes emerged in this systematic analysis.

Perspective:

In this well-conducted systematic analysis, the authors draw attention to the substantial number of multivariable models in the literature that could be used for prediction of death, hospitalization, or both in patients with heart failure. Of note, risk discrimination was greater for models predicting death; as the authors opine, ‘hospitalization [may be] genuinely more difficult to predict than death.’ Also noteworthy is that only 36% of the identified models had been validated in an independent cohort. Nonetheless, with more than 60 multivariable risk prediction models, future efforts should help inform how these tools can be incorporated into informed decision making.

Clinical Topics: Heart Failure and Cardiomyopathies, Acute Heart Failure, Heart Failure and Cardiac Biomarkers

Keywords: Body Mass Index, Sodium, Heart Failure, Stroke Volume, Blood Pressure, Heart Rate, Hospitalization, Diabetes Mellitus, Natriuretic Peptide, Brain


< Back to Listings