The fickleness of data: Estimating the effects of different aspects of acupuncture treatment on heart rate variability (HRV). Initial findings from three pilot studies
© Tony Steffert, Open University and David Mayor, University of Hertfordshire
Background. Heart rate variability (HRV) is a measure of the interplay between sympathetic and parasympathetic influences on heart rate. Higher HRV is usually associated with relaxation and health benefits, lower HRV with stress/pathology. HRV is used increasingly in acupuncture research. Electroacupuncture (EA) and transcutaneous electrical acupoint stimulation (TEAS) are frequently used modalities, variants of manual acupuncture (MA). This is the fourth of a series of conference posters from a study investigating the effects of EA and TEAS on HRV and the EEG (electroencephalograph).
Objectives. To assess how treatment factors – particularly stimulation frequency (Hz) – contribute to changes in HRV.
Methods. Three small pilot studies were conducted. All intervention and monitoring ‘segments’ lasted for 5 minutes. In Pilot 1 (N=7, 12 visits in all), 5-minute electrocardiograph (ECG) monitoring followed each intervention segment. In Pilot 2 (N=12, 48 visits) & Pilot 3 (N=4, 16 visits), 5-minute monitoring and stimulation were concurrent; ECG and then photoplethysmography (PPG) were used, and HRV (or pulse rate variability, PRG) derived from raw interbeat interval data following standard procedures, including artefact processing. Stimulation was at different combinations of the acupoints LI4 and ST36, and at either 2.5 Hz or 10 Hz. Eight HRV/PRV measures were selected for analysis. For each factor, numbers of significant differences in these measures were counted (N), and normalised percentage differences calculated (Diff%). In addition, coefficient of variance (CV), Cohen’s d (effect size) and correlation ratio eta were computed for the differences in measures induced by the various factors.
Results. Several methods of assessing differences suggested a small, non-significant difference in HRV measures in favour of 2.5 Hz. However, most of these could be explained by intrinsic variation (CV) of the measures rather than as a specific effect of stimulation frequency.
Further analysis. There were highly significant correlations between N, Diff%, d and eta for the treatment factor comparisons made (e.g. stimulation frequency, amplitude, location, visit, participant and baseline values of five main HRV measures). The sum of eta-squared for all factors considered was 0.678, suggesting that >2/3 of factors responsible for variance in outcomes were identified. This variance was mostly dependent on differences among participants, and least on stimulation frequency.
Conclusions. The analytical methods employed are accessible even to those with little statistical expertise. They offer a simple way of assessing the contribution of different experimental factors to outcomes when statistical significance is elusive and sample size is small. They are thus be appropriate for application in acupuncture research, which tends to involve a number of independent variables in small-scale studies. However, a mixed models approach and multivariate analysis should also be used to analyse new and existing results, with Bootstrap to ensure a sufficiently large sample size. In this study, the effects of stimulation frequency on HRV are likely to be masked by those of other treatment factors.
To Robert Kozarski and Tim Watson of the University of Hertfordshire for advice on statistics and study design, to Terje Alraek and Alan Watkins for sharing their considerable experience of HRV, to Harmony Medical for providing equipment (the Classic4 stimulator, needles and self-adhering electrodes), and of course to the acupuncturists and other health practitioners who gave up considerable amounts of their time to take part in our EEG/HRV study. Also to ourselves, for remaining patient with each other over a long period of collaboration. Any errors in this report remain, of course, our own responsibility.
If you have any problems with this web site, please contact David Mayor at +44 (0) 1707 320782 or