Scatter Comparison of Heart Rate Variability Parameters

Abstract

Heart rate variability (HRV) is a simple and non-invasive method for autonomous nervous system activity and probability of cardiovascular morbidity and mortality monitoring. However, the high scatter of HRV parameter values, found in both intraindividual and homogeneous intergroup measurements, inhibits the process of determining normal values or the ones indicating certain diseases. Such a process could be useful in clinical practice.

The aim of this work was to compare the scatter of a selection of HRV parameters from linear, frequency and nonlinear domains. They were compared using three different methods: between persons, between consecutive nights and between 45 min fragments of RR time intervals.

The largest mean scatter of HRV parameters was found in method 2 and the lowest in method 1. In all scales, correlation dimension, Lyapunov exponent, and unnormalized frequency parameters were the most scattered among the methods. Shannon entropy, Hurst exponent, DFA method’s $\alpha$ exponent and mean RR interval had the lowest scatter. Low scatter of HRV parameters may be useful for determining the normal values of a parameter, however the quality of information carried by HRV parameter should be considered too.

The questionnaire revealed that the subjects formed a uniform group regarding the daytime events. Most participants did not have any highly stressful or exciting events. The sleep quality, however, was distributed almost equally over the possible answers.

The large dispersion of most parameters in different time scales indicates a high uncertainty in the interpretation of single measurements of the series of RR time intervals, because they may deviate from the typical values for the patient. The allostatic regulation of the organism should be considered using the averaged results to minimize the impact of intraindividual and interindividual variability.