Beyond HRV

David Mayor,a[1] Deepak Panday,a Tony Steffertb and Hari Kala Kandelc

a. University of Hertfordshire; b. MindSpire (CTO) & The Open University; c. Hertfordshire International College

Background information (PDF)

The poster (PDF)

Abstract for a poster presentation, 22nd International Acupuncture Research Symposium, Virtual event, 27 March 2021

Background. Heart rate variability (HRV) is useful in acupuncture research as a measure of autonomic response. However, although it accurately reflects parasympathetic activity, there is less agreement on whether and how HRV can assess sympathetic activity. This study explores some potential autonomic measures beyond HRV that may do so, and investigates the effects of different frequencies and amplitudes of transcutaneous electroacupuncture (TEAS) on these measures.  

Objectives. To develop two open-access MATLAB graphical user interface (GUI) packages to facilitate accurate extraction of autonomic measures and their interrelationships from amplitudes and time intervals in up to four channels of simultaneously recorded ECG and other time series data. (2) To explore how such measures correlate with established HRV indices. (3) To assess to what extent these measures reflect the effects of differences in the frequency and/or amplitude of applied TEAS, age, gender and the passage of time. (4) To revisit the results of our previous research using these new data, incorporating corrections to data previously used.

Methods. Data were gathered in a single-blind, semi-randomised cross-over study (N = 66), as reported at the 2019 ARRC Symposium. Our first GUI was created in MATLAB using standard methods, tested and revised until considered suited to extracting measures from raw ECG, BVP (blood volume pulse) and respiration data in a semi-automated manner. CEPS, our second GUI, is described elsewhere (link). More than 70 HRV, nonlinearity (HRNL) and other measures were explored on the basis of a literature review and allocated iteratively to PNS-like, SNS-like and two other groupings using correlations (Spearman’s rho), and then a more formal factor analysis. 75 complexity and entropy measures from CEPS were also investigated. (See link to Background information, above).

Results (2020). The following allocations of ‘new’ (non-HRV) measures were consistent both pre- and post-TEAS:




1. Finger temperature variability

(TEMP coefficient of variation, or ‘CV’)

2. Outbreath-to-inbreath interval (O/I) ratio 

3. T-wave amplitude variability

4. Finger temperature (TEMP)

5. Cardiac coherence ratio

6. Outbreath CV

7. O/I ratio CV

8. Respiratory amplitude CV

9. R-wave amplitude CV

10. Pulse transit time (PTT) CV

Most of these measures increased or decreased consistently, regardless of stimulation frequency. As for conventional HRV, effect sizes (Cohen’s d) for pre-post differences were at best small, 0.2-0.4 (*).

Results (2021). On the basis of strong correlations at baseline (rho > 0.8), Tone-Entropy (T-E) ‘Entropy’ and extended Poincaré Plot measures SD1 and SD2 at multiple lags could be considered PNS-like, Slope entropy and T-E ‘Tone’ at multiple lags SNS-like. These allocations remain to be confirmed. For T-E, pre-post effect sizes could be large (>0.8), or even ‘huge’ (>1.0).

Greatest numbers of large/huge effect sizes were found in the QTi and systolic interval (Si) data for differences between high and low amplitude stimulation, and for age (younger/older) in the PTT and pulse wave amplitude data. Using conventional RRi HRV measures for the same comparisons, far fewer large effect sizes were found.

We again observed that stimulation at 10 pps may be experienced as less stressful than at 2.5 or 80 pps, for both linear and nonlinear non-HRV measures. We also found that most PNS-like measures were higher in younger than older individuals, and in women rather than men, but most SNS-like measures were higher in men than women, and in older than younger participants.

Conclusions. Future research into the autonomic effects of acupuncture could explore using nonlinear measures of complexity and entropy, applying them not only to the usual ECG RR or BVP peak-to-peak interval data, but also to interval and amplitude data such as QTi, Si, RSa and PWA. Future research should also use these measures alongside accepted methods of assessing sympathetic activity such as skin conductance or salivary amylase.

[1] Corresponding author