Abstract

The traditional methods of analysing heart rate variability based on means and variance are unable to detect subtle but potentially important changes in interbeat heart rate behaviour. This research was designed to evaluate the clinical applicability and prognostic significance of new dynamical methods of analysing heart rate behaviour derived from nonlinear mathematics.

The study covered four different patient populations, their controls and one general population of elderly people. The first patient group consisted of 38 patients with coronary artery disease without previous myocardial infarction, the second of 40 coronary artery disease patients with a prior Q-wave myocardial infarction, and the third of 45 patients with a history of ventricular tachyarrhythmia. The fourth group comprised 10 patients with a previous myocardial infarction who had experienced ventricular fibrillation during electrocardiographic recordings. The fifth group comprised a random sample of 347 community-living elderly people invited for a follow-up of 10 years after electrocardiographic recordings.

Heart rate variability was analysed by traditional time and frequency domain methods. The new dynamical measures derived from nonlinear dynamics were: 1) approximate entropy, which reflects the complexity of the data, 2) detrended fluctuation analysis, which describes the presence or absence of fractal correlation properties of time series data, and 3) power-law relationship analysis, which demonstrates the distribution of spectral characteristics of RR intervals, but does not reflect the magnitude of spectral power in different spectral bands.

Approximate entropy was higher in postinfarction patients (1.17 ± 0.22), but lower in coronary artery disease patients without myocardial infarction (0.93 ± 0.17) than in healthy controls (1.03 ± 014, p < 0.01, p < 0.05 respectively). It did not differ between patients with and without ventricular arrhythmia. The short term fractal-like scaling exponent of the detrended fluctuation analysis was higher in coronary artery disease patients without myocardial infarction (1.34 ± 0.15, p < 0.001), but not in postinfarction patients without arrhythmia (1.06 ± 0.13) compared with healthy controls (1.09 ± 0.13). The short term exponent was markedly reduced in patients with life-threatening arrhythmia (0.85 ± 0.25 ventricular tachycardia patients, 0.68 ± 0.18 ventricular fibrillation patients, p < 0.001 for both). The long term power-law slope of the power-law scaling analysis was lower in the ventricular fibrillation group than in postinfarction controls without arrhythmia risk (-1.63 ± 0.24 vs. -1.33 ± 0.23, p < 0.01) and predicted mortality in a general elderly population with an adjusted relative risk of 1.74 (95% CI 1.42–2.13).

The present observations demonstrate that dynamic analysis of heart rate behaviour gives new insight into analysis of heart rate dynamics in various cardiovascular disorders. The breakdown of the normal fractal-like organising principle of heart rate variability is associated with an increased risk of mortality and vulnerability to life-threatening arrhythmias.