Long-term spectral analysis of heart rate variability--an algorithm based on segmental frequency distributions of beat-to-beat intervals.

Reduced heart rate variability has been reported as a predictor of long-term mortality in recent myocardial infarction patients. However, it has not been systematically investigated whether the reduction in heart rate variability in those post myocardial infarction patients who later suffer death or severe arrhythmias is caused by a reduction of short-term variability of heart rate (such as respiratory arrhythmia) or whether the differences in long term variability (such as diurnal rhythm) are involved. In order to perform such an evaluation, a new algorithm has been developed which permits different wavelength components (including the long-term components due to diurnal rhythm) of heart rate variability to be approximated. In general, the method uses segmental frequency distributions of durations of intervals between successive normal cardiac beats. To assess the spectral components of heart rate variability, a scale of wavelength limits is used and for each limit of this scale, the algorithm excludes the rate changes of wavelength longer than the given bound. The method was applied to the analysis of electrocardiograms recorded in 14 post myocardial infarction patients who later suffered death or ventricular tachycardia, and in 14 other randomly selected patients with an uncomplicated course following acute myocardial infarction. The rate variability spectra obtained for both groups of patients were compared statistically and the results showed that the groups of positive and negative cases were most significantly distinguished when including both short- and long-term components of heart rate variability. Separate evaluation of different wavelength components showed that the very long-term components of heart rate variability were more powerful in distinguishing between positive and negative cases than the short term components.