Author Reply to Letter to the Editor regarding ‘Heart rate variability in patients with dementia or neurocognitive disorders: A systematic review and meta-analysis’

To the Editor

We have read Santos-de-Araújo and Pontes-Silva’s letter about our meta-analysis. We appreciate their interest and positive feedback on our article. Santos-de-Araújo and Pontes-Silva mentioned six suggestions about the application of heart rate variability (HRV) in the psychiatric field. We agree that these are all important notifications. Although some key points about measuring or interpreting HRV have been discussed in the literature, it is not common to see systematic reminding. Therefore, a summary or checklist should be meaningful for clinicians or researchers working in this field.

To Santos-de-Araújo and Pontes-Silva’s suggestion 1, we agree that respiratory rate is important to HRV studies. The respiratory rate affects the main frequency band of parasympathetic activity (Shaffer and Ginsberg, 2017). Low-frequency power is affected by both sympathetic and parasympathetic modulations; but under a low respiratory rate, the main frequency band of parasympathetic activity may move from high frequency to low frequency. Therefore, recording respiratory rate could clarify the biological meaning of low-frequency power. This is also the reason we viewed respiratory sinus arrhythmia (or heart rate Max – heart rate Min) as the HRV index with the highest hierarchy in our meta-analysis; the effect of respiration is considered in this index.

To suggestions 2 and 5, a feature of HRV is its high sensitivity and can be influenced by many factors of subject and environment (Laborde et al., 2017). For reducing the impact of these potential confounders on psychiatric HRV studies, we suggest three steps. First, when the subject has a biological feature that may affect HRV, such as severe diabetes mellitus, arrhythmia, multiple system atrophy, taking tricyclic antidepressants or taking antipsychotics with high muscarinic affinity, HRV cannot be viewed as valid markers for reflecting psychiatric features. If these factors are not major targets of studies, individuals with these features should be excluded. Second, the controllable factors should be identical in a whole study, such as measuring in the daytime and fixed position, preventing exercise and substance exposure before measurement. If sex or sex hormone is an important target for investigation, the menstruation cycle of the female subjects could be controlled. Finally, the factors that existed in all subjects (such as age, body mass index, level of depression and level of anxiety) can be comprehensively collected and adjusted during the analysis.

To suggestions 3, 4 and 6, data acquisition, analysis and filtering are important in HRV studies. There are a lot of hardware and software for measuring HRV; some can calculate HRV in a real-time manner and some needs further analysis of the heart rate signals. The performance of distinct devices differs a lot. HRV is affected by the algorithm and filtering. The filtering is important for removing the artefacts and reducing the influence of outliers. The algorithm is involved with the correct interpretation of the R wave and the exclusion of ventricular premature complex. To examine whether calculated HRV is reliable, it is suggested to store the raw signal of electrocardiography. It is helpful for further checking by experts or re-calculating with another software. About these points, we agree with Santos-de-Araújo and Pontes-Silva’s suggestions.

Several of the above points were also discussed in another recent article (Huang et al., 2022). In addition, when applying HRV in studies about dementia or neurocognitive disorders, the main challenge should be the confounder issue. Older adults often have physical comorbidity and are more likely to take medications. A careful study design is needed to diminish the possible influence of these factors.

Thanks again for Santos-de-Araújo and Pontes-Silva’s letter and the possibility of discussing these issues.