Spontaneous respiration should be avoided in frequency domain analysis of heart rate variability

To determine whether spontaneous respiration is suitable for frequency domain analysis of heart rate (R-R interval) variability, we studied 15 volunteers (5 men and 10 women, aged 22-34 years) and evaluated the reproducibility of the power spectrum. Electrocardiograms were recorded for 5 min each with spontaneous and rate-controlled respiration (15 breaths·min^-1), repeating the same protocol 1 week later. Fast Fourier transformation was performed using the digitized data of the R-R intervals. Mean heart rate, arterial pressure, and plasma catecholamines remained constant during the measurements. In spontaneous respiration, however, the respiratory rate was significantly lower during the second measurement (9.4±2.1 breaths·min^-1) than during the first measurement (10.9±2.6 breaths·min^-1), and the low-frequency power increased from 2.61±2.36 to 5.14±5.06 sec²·Hz^-1·10^-3. After deleting five data sets because the respiratory peak was inseparable from the low-frequency area, there was no correlation in power spectra in four out of ten subjects between the two measurements. Data were comparable for rate-controlled respiration. Since respiratory parameters strongly influenced the low- and the high-frequency R-R interval power spectra, spontaneous respiration should be avoided. A constant respiratory condition is required to interpret results of frequency domain analysis of R-R interval variability.