Inferential and descriptive influences on measures of respiratory sinus arrhythmia: sampling rate, R-wave trigger accuracy, and variance estimates.

The use of respiratory sinus arrhythmia (RSA) is dependent on several quantification factors. The statistical consequences of sampling rate, R-wave trigger accuracy, and the data structure of RSA estimates were evaluated via simulations and simulated modifications. Simulations demonstrated that sampling rate determined the amplitude of the smallest oscillation that could be accurately described. When the amplitude of the oscillation was smaller in milliseconds than the sampling interval, all smaller oscillations, regardless of amplitude, were indiscriminately assigned a default value defined as 50% of the sampling interval. Also, reduced R-wave peak detection accuracy was demonstrated to influence statistical analyses. The distributional characteristics of several common methods of structuring RSA estimates were investigated, and results indicated that several methods results in asymmetrical distributions. We recommend using rapid sampling rates (i.e., 1 kHz), accurate R-wave peak detectors, and avoiding statistical trouble.