Background: New methods based on nonlinear theory have been developed to give more insight into complex heart rate (HR) dynamics. This study was designed to test the hypothesis that altered HR dynamics, as analyzed with complexity and fractal measures, may precede the spontaneous onset of paroxysmal atrial fibrillation (PAF). Secondly, the difference in the temporal change of these measurements between the different types of atrial fibrillation (AF) was assessed.
Methods and results: From 105 Holter tapes in which PAF was recorded, 44 PAF (>or=5 min) episodes in 33 patients (22 men, 58 +/- 12 years), preceded by sinus rhythm for more than 1 h, were selected and submitted to time-and frequency-domain HR variability analyses, along with detrended fluctuation analysis, approximate entropy (ApEn) and sample entropy (SampEn). The 60 min before the onset of AF were divided into 6 10-min periods and studied using repeated measures ANOVA. PAF episodes were divided into 3 subgroups: an increased HF component and decreased L/H ratio (vagal type, n=20); increased L/H ratio and decreased HF component (sympathetic type, n=14); and non-related type (n=10). None of the time- or frequency-domain parameters showed any significant change before AF in any type of AF. The alpha(1) showed a tendency to decrease before the onset of AF and the change in alpha(1) was divergent according to the AF type. The ApEn decreased before the onset of AF (1.005+/-0.046, 60-50 min before AF to 0.894+/-0.052, 10-0 min before AF; p=0.032). The SampEn also decreased progressively before the start of AF (1.165 +/- 0.085, 60-50 min before AF to 0.887 +/- 0.077, 10-0 min before AF, p=0.003). The decrease in both the ApEn and SampEn was irrespective of the AF type.
Conclusions: A reduction in the ApEn and SampEn, which reflects the nonlinear complexity of HR variability, is a hallmark of altered HR dynamics preceding the spontaneous onset of AF.