Association between RR interval and high-frequency heart rate variability acquired during short-term, resting recordings with free and paced breathing

High-frequency (HF) oscillations in RR interval from 0.15-0.40 Hz are widely accepted as a measure of cardiac vagal outflow but the HF/RR relationship appears complex, particularly with longer RR intervals. The aim of this study was to evaluate the HF/RR interval relationship during free and paced breathing. HF power and mean RR interval length were recorded in 150 men and 120 women (mean age 34.5 ± 11.4) during 5 min of supine rest with either free or paced (12 cycles min-1) breathing. Linear and quadratic models were used to assess the relationship between RR interval and the natural logarithm of HF power (lnHF). The RR interval length at which there was no further increase in lnHF was determined as the deflection point. ANCOVA was used to determine differences in the linear regression slopes for lnHF/RR with paced or free breathing. With free breathing (n = 131), the adjusted R2 was similar between linear (15.3%) and quadratic (17.5%) fits and saturation of lnHF occurred within the recorded RR interval range (1326 ms). With paced breathing (n = 139), adjusted R2 values were again similar between linear (22.4%) and quadratic (23.2%) fits. The deflection point was outside the range of recorded RR intervals at 1458 ms. ANCOVA showed a significant difference in the slope of the lnHF/RR regression lines between free and paced breathing. The lnHF/RR relationship is weaker when derived from between-subject recordings than from repeated within-subject samples. lnHF/RR showed evidence of saturation at ∼45 bpm with free breathing. With paced breathing, a deflection in lnHF was found outside the recorded RR interval range (∼41 bpm). Paced breathing creates a stronger lnHF/RR relationship. The slope of the lnHF/RR regression line with paced breathing is significantly different from that observed with free breathing. It appears that lnHF is a valid index of vagal outflow, except in subjects with very low heart rates. Paced breathing data collection protocols appear preferable. © 2008 Institute of Physics and Engineering in Medicine