Heart Rate Variability Standards
To the Editor:
The special report on heart rate variability by the European Society of Cardiology and North American Society of Pacing and Electrophysiology1 presented important standards of measurement for heart rate variability analysis. In the report, the frequency range of the total power was defined as 0.04 to 0.4 Hz, the low-frequency (LF) component as 0.04 to 0.15 Hz, and the high-frequency (HF) component as 0.15 to 0.4 Hz. There are several problems in the calculation of specific spectral powers using these standards of measurement in heart rate variability analysis.
As stated in the report, the HF component is respiration related, and the distribution of the power and the central frequency of LF and HF components are not fixed but vary in relation to changes in autonomic modulations of heart period in the short-term recordings. Therefore, integrating the HF power within the all-frequency range of 0.15 to 0.4 Hz might have inherent error, especially when the respiration rate does not fall within this range for patients who have tachypnea or are under controlled respiration. Because the maximum frequency in the spectrum is the Nyquist frequency (half the sampling frequency),2 3 it might be better to use the Nyquist frequency as the upper limit of both HF power and total power.
When the direct current is excluded by baseline or trend removal in the calculation of spectral powers, the nonharmonic components in the very-low-frequency (VLF) region (<0.04 Hz) can be removed. In this case, it is not necessary to set a cutoff limit (0.04 Hz in most instances) for LF power or total power. In addition, the purpose of normalization of the LF and HF powers by the total power is to minimize the effect of the changes in total power on the values of LF and HF components.1 The placement of a cutoff at 0.04 Hz to the lower limit of total power will result in incomplete normalization of the LF and HF components. Finally, if a lower limit (0.04 Hz in most instances) was set to the LF power, the VLF power (≤0.04 Hz in most instances) must be dealt with for the sake of completeness. However, the physiological explanation of the VLF component is much less defined than other components in the spectrum. The existence of a specific physiological process attributable to these heart period changes might even be questioned. The VLF is then a dubious measure and should be avoided.1
In spectral analysis used in sciences such as physics or chemistry, integration of the area under the peak rather than fixed-range integration is usually suggested to evaluate the relative contribution of a specific frequency to total power.4 5 Thus, a peak-related integration according to the respiration rate might be a better method of representing HF power. Fixed-range integration is justified only if no apparent peak can be identified in the HF range.
Because of the above considerations, it seems that the area of spectral peaks within the entire range of 0 Hz to the upper limit of HF peak or to the Nyquist frequency can be used as the total power, the area of spectral peaks within the range of 0 to the lower limit of HF peak as the LF power, and the area under the HF peak as the HF power. If no apparent peak related to respiration can be identified, the total power can be defined as the area of spectral peaks within the entire range of 0 Hz to the Nyquist frequency, the LF power as the area of spectral peaks within the range of 0 to 0.15 Hz, and the HF power as the area of spectral peaks within the range of 0.15 Hz to the Nyquist frequency.
- Copyright © 1998 by American Heart Association
Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Heart rate variability: standards of measurement, physiological interpretation, and clinical use. Circulation. 1996;93:1043–1065.
DeBoer RW, Karemaker JM, Strackee J. Comparing spectra of a series of point events particularly for heart rate variability data. IEEE Trans Biomed Eng. 198431:384–387.
Krauss TP, Shure L, Little JN. Signal Processing Toolbox for Use With MATLAB. Natick, Massachusetts: MathWorks Inc; 1992.
Braithwaite A, Smith FJ. Chromatographic Methods. 4th ed. New York, NY: Chapman & Hall; 1985:329.
Skoog DA. Principles of Instrumental Analysis. 3rd ed. New York, NY: Saunders College Publishing; 1985:749.
Drs Kuo and Chen suggest a substantial modification to the standards for computing low-frequency and high-frequency spectral components of heart rate variability. While I appreciate their theoretical and technical reasoning, I am afraid that Drs Kuo and Chen ignore many practical and physiological issues that make their proposal inappropriate. Actually, the discussion about proper definitions of spectral components of heart rate variability is not new. Similar proposals as the one by Drs Kuo and Chen were made and dismissed in the past.
There is very little experience with physiological interpretation of heart rate variability components if the respiration is forced, by a pathological process or by instruction, to an extreme frequency that falls outside the limits of 0.15 to 0.4 Hz. Extending the upper limit of the high-frequency component beyond 0.4 Hz would only be applicable to extreme tachypnea of >24 respiratory cycles per minute. This is linked to extreme sympathetic overdrive under which it is rather difficult to interpret the high-frequency component. Moreover, because the cardiac period signal is discrete rather than continuous, it is difficult to properly estimate respiratory arrhythmia under such conditions of very fast tachypnea.
Regular periodic bradypnea of <8 respiratory cycles per minute may appear with forced metronome breathing. In such a case, the recommendations made by the Task Force cannot be applied blindly. More importantly, however, forcing a subject into an extremely slow respiration rate is again sympathetically stimulating, which makes it difficult to compare the heart rate variability components with those obtained under different circumstances.
The very-low-frequency component seems indeed to be a dubious measure because its physiological background is not known. It is likely that even with short-term recordings, these components reflect nonstationarity of heart rate modulations. Consequently, extending the limits of low-frequency components below 0.04 Hz would pollute the measurement and make the physiological interpretation even more difficult, especially when the dominant spectral peak belongs to the very-low-frequency component.
It is important to understand that the proposals of frequency components made in our report are based on experience with existing physiological models that allow interpretation of individual components. This is quite different from making a proposal based merely on hypothetical speculations.
Our report clearly suggested that if parametric methods are used for the spectral analysis, the integration of the area under distinct peaks should be used, whereas fixed-range integration was proposed for a nonparametric spectral analysis, which is, in practice, much more frequently used. Drs Kuo and Chen seem to forget that even when parametric methods are used, total power is not a simple sum of high-, low-, and very-low-frequency components. Frequently, other components are present that cause the sum of normalized high-frequency and normalized low-frequency components not to be constant.
Finally, Drs Kuo and Chen forget that for practical reasons, the definitions of high- and low-frequency components must not depend on the duration of the analyzed RR-interval series. For instance, if their proposal were applied to 24-hour recording, the results would be completely meaningless.