Predicting survival in heart failure: a risk score based on 39 372 patients from 30 studies

Abstract

Aims: Using a large international database from multiple cohort studies, the aim is to create a generalizable easily used risk score for mortality in patients with heart failure (HF).

Methods and results: The MAGGIC meta-analysis includes individual data on 39 372 patients with HF, both reduced and preserved left-ventricular ejection fraction (EF), from 30 cohort studies, six of which were clinical trials. 40.2% of patients died during a median follow-up of 2.5 years. Using multivariable piecewise Poisson regression methods with stepwise variable selection, a final model included 13 highly significant independent predictors of mortality in the following order of predictive strength: age, lower EF, NYHA class, serum creatinine, diabetes, not prescribed beta-blocker, lower systolic BP, lower body mass, time since diagnosis, current smoker, chronic obstructive pulmonary disease, male gender, and not prescribed ACE-inhibitor or angiotensin-receptor blockers. In preserved EF, age was more predictive and systolic BP was less predictive of mortality than in reduced EF. Conversion into an easy-to-use integer risk score identified a very marked gradient in risk, with 3-year mortality rates of 10 and 70% in the bottom quintile and top decile of risk, respectively.

Conclusion: In patients with HF of both reduced and preserved EF, the influences of readily available predictors of mortality can be quantified in an integer score accessible by an easy-to-use website www.heartfailurerisk.org. The score has the potential for widespread implementation in a clinical setting.

Keywords: Heart failure; Meta-analysis; Mortality; Prognostic model.