Methodological issues in the spectral analysis of the heart rate variability: Application in patients with epilepsy

Purpose: Spectral analysis of heart rate variability (HRV) constitutes a useful tool for the evaluation of autonomic function. However, it is difficult to compare the published data because different mathematical approaches for the calculation of the frequency bands are applied. Our aim was to compare the HRV frequency domain parameters obtained by application of 2 parametric and 2 non-parametric spectral methods in a group of patients with chronic epilepsy. Methods: Sixty-eight patients and 69 healthy controls underwent a 5-min recording of RR signal, which was analyzed off-line in time and in frequency domains. Results: The time domain parameters – variation RR ratio, standard deviation of normal-to-normal RR and coefficient of variation – were significantly lower in patients than in controls. In spectral analysis of the patient group deviation toward opposite directions of Low Frequency band (p = 0.034) and Total Power (p = 0.013) measures was detected depending on the method used. The results of Burg’s and YuleWalker’s parametric methods fitted best to those of time domain estimates for both control and patient groups. Conclusions: Epilepsy-related abnormalities of HRV were disclosed by time as well as by frequency domain analysis. In the present setting, the parametric methods proved to be superior to the non-parametric ones in matching time domain parameters of patients and healthy subjects and at the same time in detecting abnormalities of the frequency domain measures of patients with epilepsy. © 2014 Elsevier Ltd. All rights reserved.