Spectral Power Distribution of Heart Rate Variability in Contiguous Short-Term Intervals

Heart rate variability (HRV) is determined by the variation of consecutive cardiac electrical excitations, usually from RR intervals of an EKG. The sequence of intervals is a time series that yields three HRV parameter categories: time domain, frequency domain, and nonlinear. Parameter estimates are based on widely different EKG sample times: short-term (~5-10 minutes), longer (24 hours), and ultra-short (<5 minutes). Five-minute intervals are useful to evaluate intervention effects that change HRV in a single session by comparing pre-to-post values. This approach relies on knowing the minimal detectible change (MDC) that indicates a real change in clinical and research studies. The specific aims of this pilot study were to (1) evaluate HRV power and its spectral distribution among contiguous five-minute intervals, (2) compare the power distribution in a five-minute interval with a full 45-minute assessment, and (3) provide data to aid estimation of the MDC between pre- and post-interventions during a single session.  Methods: Twelve self-reported healthy young adults participated after signing an approved consent. Participation required subjects who had no history of cardiovascular disease or were taking vasoactive substances. Persons with diabetes were not eligible. While subjects were supine, EKG leads were placed, and EKG was recorded for 45 minutes at 1000 samples/sec. The 45 minutes were divided into nine five-minute contiguous intervals, and the spectral density in each was determined. Total power and spectral percentages within each interval were assessed in the very low (VLF, 0.003-0.04 Hz), low (LF, 0.04-0.15 Hz), and high (HF, 0.15-0.4 Hz) frequency bands. These were compared among intervals and to the full 45-minute sample. The MDC was determined by comparing powers in five-minute intervals separated by 10 minutes. The standard error of the measurement (SEME) for each pair was calculated from the square root of the mean square error (√MSE). MSE was based on a two-factor analysis of variance, and MDC was 2×√2×SEME.

Differences in total power and spectral power distribution among intervals were not statistically significant. The total mean power±SD was 4561±1434 ms². The maximum difference in total power was 7.85%. The mean power for the VLF, LF, and HF bands was respectively 1713±1736 ms², 1574±1072 ms², and 1257±1016 ms². The maximum percentage difference in spectral power across all intervals for VLF, LF, and HF was respectively 3.75%, 8.5%, and 7.4%. The percentage of power in the VLF, LF, and HF bands was respectively 37.9%, 36.1%, and 25.9%. The ratios of spectral to total power for VLF, LF, and HF bands were respectively 0.80±0.07, 1.20±0.11, and 1.22±0.10. MDC percentage values were 21.0±4.9% for the HF band, 25.7±1.4% for the LF band, and 30.4±5.5% for the VLF band.

Results offer initial estimates of variations in HRV power in the VLF, LF, and HF bands in contiguous five-minute intervals and estimates of the minimum detectible "real" changes between intervals separated by 10 minutes. The pattern of variation and data are useful in experimental planning in which HRV spectral power changes are assessed subsequent to a short-duration intervention during a single session. MDC values (21.0% in the HF band to 30.4% in the VLF band) provide initial estimates useful for estimating the number of participants needed to evaluate the impact of an intervention on spectral components of HRV.