Response to: Reproducibility of 24-h ambulatory blood pressure and measures of autonomic function
The aim of our study was to assess the reproducibility of 24-h ambulatory blood pressure, 24-h heart rate variability (HRV) and short-term resting measurements of both HRV and blood pressure variability. We reported the magnitude of between-day variability in these measures and the appropriateness of familiarisation sessions before experimental data collection. Findings revealed that ambulatory blood pressure became more reproducible following repeated recordings. In addition, 24-h HRV was more reproducible than resting measures. Consecutive pairs of measurements (days 1–2, 2–3, 3–4) were analysed for the determination of the typical error and coefficient of variation as described by Hopkins 2. These statistical methods provide a point estimate of absolute and relative reliability and reflect the ability of a measure to detect systematic changes between repeated measurements within a participant 3. These methods also quantify the degree of uncertainty surrounding this point estimate and can be used by researchers to evaluate the likelihood that observed changes following an intervention in a sample population are attributable to the experimental conditions and not a manifestation of error inherent within the test. The effect of a therapeutic intervention on autonomic function and blood pressure is a fertile area of contemporary research 4, and the data presented in our manuscript are appropriate to inform practitioners making population-wide health recommendations based on the outcomes of experimental research.
Dr Sabour suggests that the conclusions of our study are misleading because of the use of a ‘global average’ approach to assess reproducibility, recommending instead the use of an individual-based approach in the form of an intraclass correlation coefficient (ICC) single measure. We agree that the ICC single measure can be a valuable indicator of reproducibility; however, limitations to its use for repeated HRV measurements have been widely reported in the literature 5. First, although a high ICC between repeated measures is typically accepted to indicate a high level of reliability, it could also be an artefact of high between-subject variability 3. Although we recruited a homogenous population, measures of HRV are typically known to be heterogeneous 3. Therefore, a high between-trial ICC for HRV could be misleading. Second, although the ICC measures the proportion of variance in a dataset that is attributable to error, it does not explicitly quantify the noise in the test. Therefore, we remain confident that our use of typical error and coefficient of variation were most appropriate for the current dataset.
Nevertheless, we thank Dr Sabour for drawing the attention of Blood Pressure Monitoring readers to the importance of using appropriate statistical methods when interpreting data pertaining to the consistency of measurements in experimental research.