A comparison of three quantification methods for estimation of respiratory sinus arrhythmia

The empirical literature has shown that respiratory sinus arrhythmia is a sensitive noninvasive index of parasympathetic cardiac control. Nevertheless there has been no general agreement among investigators as to the most preferable quantification technique for assessing respiratory sinus arrhythmia, although there has been much speculation that specific estimation techniques are more or less reflective of vagal processes and could be more or less contaminated by other influences upon heart period variability unrelated to respiration. This study compared three quantification procedures for estimating respiratory sinus arrhythmia (RSA): (1) a spectral analytic technique, (2) a complex detrending approach removing periodic and aperiodic cardiac variations unrelated to respiration, and (3) a time-domain, peak-valley procedure employing inspiratory and expiratory periods as windows for determining range of cardiac-interval fluctuations associated with respiratory phase. Measures derived from these techniques were intra- and interindividually compared using three different samples of male subjects, including students, adult normotensives, and adult hypertensives. All interindividual correlations between measures yielded coefficients above .92 and the mean within-subject correlation across 42 individuals was .96, thus indicating a marked degree of comparability between measures. Additionally, given that much evidence indicates lawful within-individual relations between RSA amplitude and respiratory parameters, we employed respiratory period as an external criterion and compared intraindividual correlations between this variable and (2) and (3); results showed that (3) was significantly more highly associated with respiration than was (2), although the mean r's for the two measures did not diverge greatly (.91 vs. .84). Finally, inspection of the data and further regression analyses did not suggest that any of the RSA estimates were differentially contaminated by other components of cardiac variability. Our findings suggest that the three techniques are almost equivalent as indices of cardiac vagal tone and would appear to ease concerns about the inferiority of any of the procedures. Choice of a quantification procedure should therefore be tailored to the specific empirical needs of an investigation. The advantages and disadvantages of each method are discussed.