Predictors of mortality and morbidity in patients with chronic heart failure

Eur Heart J. 2006 Jan;27(1):65-75. doi: 10.1093/eurheartj/ehi555. Epub 2005 Oct 11.


Aims: We aimed to develop prognostic models for patients with chronic heart failure (CHF).

Methods and results: We evaluated data from 7599 patients in the CHARM programme with CHF with and without left ventricular systolic dysfunction. Multi-variable Cox regression models were developed using baseline candidate variables to predict all-cause mortality (n=1831 deaths) and the composite of cardiovascular (CV) death and heart failure (HF) hospitalization (n=2460 patients with events). Final models included 21 predictor variables for CV death/HF hospitalization and for death. The three most powerful predictors were older age (beginning >60 years), diabetes, and lower left ventricular ejection fraction (EF) (beginning <45%). Other independent predictors that increased risk included higher NYHA class, cardiomegaly, prior HF hospitalization, male sex, lower body mass index, and lower diastolic blood pressure. The model accurately stratified actual 2-year mortality from 2.5 to 44% for the lowest to highest deciles of predicted risk.

Conclusion: In a large contemporary CHF population, including patients with preserved and decreased left ventricular systolic function, routine clinical variables can discriminate risk regardless of EF. Diabetes was found to be a surprisingly strong independent predictor. These models can stratify risk and help define how patient characteristics relate to clinical course.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adult
  • Age Distribution
  • Aged
  • Cause of Death
  • Chronic Disease
  • Heart Failure / mortality*
  • Hospitalization / statistics & numerical data
  • Humans
  • Middle Aged
  • Morbidity
  • Prognosis
  • Proportional Hazards Models
  • Risk Factors
  • Survival Analysis
  • Sweden / epidemiology
  • Ventricular Dysfunction, Left / mortality