Homeopathy and Heart Rate Variability: Concerns and Issues

Dear Editor:

We note with some concern your recent publication ¹ and promotion ² of results that use heart rate variability (HRV) spectra to determine the response of the autonomic nervous system to homeopathic remedies. We would like to highlight several methodological problems with this study not raised in the editorial, ² which include several noteworthy and unjustified procedural deviations from typical experimental design and the evidence-based use of statistical and HRV analyses. For instance, it is not clear whether the study is single- or double-blind; the presentation of placebo versus “medicine” was not counterbalanced; and atypical analytical procedures were employed without appropriate statistical analysis. However, for the purposes of brevity, we shall concentrate only on a number of issues associated with a pre–post comparison of amplitude in peaks identified from 0.02 Hz to 0.5 Hz in HRV power spectral density (PSD).

This is not a commonly understood or accepted methodology. Individual peaks within HRV spectra cannot be considered significant unless they are provoked by a stimulus that is specifically time-locked to that frequency (for example, paced breathing at 0.25 Hz will produce a respiratory-related peak in HRV PSD at 0.25 Hz ³ ).

When breathing or other paced stimuli are not controlled, frequency bands are defined: typically very low frequency (VLF) variability (0.003 Hz–0.04 Hz), low-frequency variability (0.04 Hz–0.15 Hz), and high-frequency variability (0.15 Hz–0.4 Hz). ⁴ This allows PSD to be calculated over a range of values that encapsulate specific physiologic phenomena. As such, it is likely that the deviations produced in Mishra et al. ¹ are simply from random variation within the distribution of the peak amplitudes.

This is exacerbated by two related errors. First, the inclusion of VLF data (from 0.02 Hz to 0.04 Hz) is unjustified in this context. The Task Force reference ⁴ cited in Mishra et al. ¹ warns explicitly against the use of this spectrum during short-term recordings, as it is unlikely to contain meaningful information concerning the frequency of the heart. In addition, no procedure for artifact correction or detrending was identified. Regular artifacts crucially affect the frequency-based analysis of HRV, ⁵ and detrending procedures have a significant overall effect on the nonharmonic (VLF) component of variability. ⁶

In summary, this methodology on the whole differs substantially enough from the evidence-based procedures to the extent that there is little to suggest that the HRV analyses presented produce anything beyond the codification of random variations in HRV power spectral density. We submit that this is far from “evidence that even the most ardent of skeptics would struggle to deny.” ²