Authors’ response

Sir,

We thank the author of the letter-to-Editor¹ for reading our article with great interest and providing the insightful comments on our article². The point raised is regarding the sample size calculation and the assumption of a 20 per cent dropout rate. The sample size was calculated based on the prediction of expected variance in the outcome measure using the reported findings of Trimurthula et al³, in a similar study conducted in South India. In any intervention study, the expected dropout rate is 20 per cent or less than that, so we also assumed a 20 per cent dropout rate⁴. We conducted a structured pattern of activities like interacting with the participants on a WhatsApp group, calling them once weekly, and weekly online therapy sessions to ensure the participants’ retention. In our study, the dropout rate was only 10 per cent, which is half of the assumed dropout rate. The reasons for those dropouts were also mentioned in our article⁴. Perhaps, there was no mention of either the sample size calculation or the dropouts in the study by Trimurthala et al³. In their study, the participants were randomly classified into experimental and control groups³.

We agree that providing a clearer rationale by the implementation of a blinding method could enhance the study’s validity. Double blindness is needed to avoid the problem of bias, especially in drug trials⁵. Therefore, double blinding is used for clinical trials with new drug interventions⁶. The participants and researcher were not blinded in our study, as the researcher, who is a trained psychologist, had to teach the therapy procedure to each participant individually in the intervention group. The person who knows the therapy details and scales can only do the assessment. The intervention group participants were instructed not to discuss the therapy procedure with anyone until the completion of the study. The participants were assigned to two study groups based on a randomisation procedure, and the control group was given standard care at baseline and was followed up after three months, whereas intervention group participants were asked to join a WhatsApp group to check the compliance and weekly therapy sessions were conducted to motivate them². There is no cross-contamination between the groups. Hence, there was no risk of performance bias.

Another important comment was on the confounding variables. The potential confounding variables in our study, such as diabetes medication, changes in diet, levels of physical activity, and other stress coping activities used by the participants that could reduce the accuracy of the isolation effect of the PMR therapy were pointed out. In any intervention study, the existence of confounding variables makes it difficult to establish a clear link between treatment and outcome measures. As reported by Skelly et al⁷, appropriate methods are to be used to adjust for the effect of the confounders. In our study, we took this into consideration while designing the study to ensure that participants in both the groups received the same treatment procedures, including diabetes medication, diet, and physical activity. The diabetes medication remained unchanged throughout the study. There is no possibility of confounding variable interference in the study because at baseline, the diet pattern, physical activity and treatment regimen were assessed for all the participants, and it remained the same throughout the study period. The study participants were provided with a prescription by the dietician, and they reviewed their diet pattern at follow up. Prior to enrolment, the participants were asked whether they were following any other stress reduction therapies; those who answered ‘yes’ were excluded from the study.

Additionally, bias on self-reported compliance through WhatsApp in our study and suggested methods like digital tracking were pointed out in the letter-to-editor⁸. Ioannidis and colleagues explained how the selective reporting methods of outcomes can bias clinical trials⁹. We followed the WhatsApp method to assess the compliance. Self-reported WhatsApp messages and weekly phone calls were done with the participants to ensure whether they were consistent in following the therapy, and this WhatsApp message from one participant in the morning gave motivation to the others to undertake the therapy and to share their experience on a regular basis. Digital tracking technology is primarily used for reading customer preferences in the marketing field, and it is difficult to implement in health care services due to the high cost of employing tracking software and obtaining consent from the participants. Furthermore, It is also difficult to manage the large data sets in accordance with strict privacy-preserving protocols¹⁰.

Another comment made was to assess the biochemical markers linked with stress relief. These biochemical markers can be effective for the assessment of depression. The markers include pro-inflammatory factors, brain-derived neurotrophic factor and substances indicating the level of oxidative stress¹¹. Although we agree upon that but we did not assess any biochemical markers linked with stress relief. We will plan to include these markers in our future studies.

Also highlighted was the importance of a longer follow up to assess the long-term effects of PMR therapy. Woolard et al¹², reported that longitudinal follow up of participants is essential for any intervention study, as the longer follow up could provide more information on the tenability of the intervention effect¹². We mentioned this as one of the limitations in our article and discussed the long-term sustainability of the PMR therapy effect on glycaemic control².

These modifications could indeed strengthen future investigations. Although there were a few limitations in our study, the findings highlight the positive effect of PMR therapy in the management of diabetes.