Prognostic value of nonlinear heart rate dynamics in hemodialysis patients with coronary artery disease

Background. Although altered nonlinear heart rate dynamics predicts death in patients with coronary artery disease (CAD), its prognostic value in chronic hemodialysis patients with CAD is unknown. Methods. We analyzed 24-hour electrocardiogram for nonlinear heart rate dynamics and heart rate variability in a retrospective cohort of 81 chronic hemodialysis patients with CAD. Results. During a follow-up period of 31 +/- 20 months, 19 cardiac and 8 noncardiac deaths were observed. Cox hazards model, including diabetes, left ventricular ejection fraction, and the number of diseased coronary arteries, revealed that abnormal alpha(2) (defined as both increase and decrease in alpha(2) because of its J curve relationship with cardiac mortality), decreased approximate entropy and decreased heart rate variability (triangular index and ultra-low frequency power) were significant and independent predictors of cardiac death. No significant and independent predictive power for noncardiac death was observed in either the heart rate dynamics or the heart rate variability measures. The predictive power of alpha(2) and approximate entropy was independent of that of triangular index and ultra-low frequency power. Combinations of two categories of measures improved the predictive accuracy; overall accuracy of approximate entropy + ultra-low frequency power for cardiac death was 87%. Conclusion. Altered nonlinear heart rate dynamics are independent predictors of cardiac death in chronic hemodialysis patients with CAD and their combinations with decreased heart rate variability provide clinically useful markers for risk stratification.