Type 2 diabetes mellitus (T2DM) is a metabolic endocrine disorder that is characterized by continually elevated blood glucose levels. In India, it is estimated that there are approximately 74.2 million people with diabetes, accounting for 13.8% of the global prevalence [1]. It is projected that by 2045, India will have 124.9 million adults with diabetes, corresponding to 16% of the global burden of the disease [1]. Even more alarming, the age of onset of diabetes in the Indian population is younger and with a considerably lower BMI as compared with those of other racial-ethnic groups [2,3]. This is typically classified as the “Asian Indian or South Asian phenotype,” represented by higher levels of belly fat, less muscle mass, and increased insulin resistance, even at low BMI [4].

As a chronic progressive disease, the medical management of T2DM often necessitates the intensification of medications during the course of treatment. However, there is also compelling evidence to support the role of therapeutic lifestyle changes such as weight loss, dietary restrictions, exercise, adequate sleep, and health coaching in the effective management of diabetes without requiring intensification of the medications [5-10]. The participation of patients by way of self-monitoring of blood glucose (SMBG), strict compliance to dietary restrictions, and exercise routine, along with adequate sleep to abate stress is essential in avoiding disease progression and the long-term complications associated with T2DM [11].

In recent years, digital technologies have attracted increasing attention to enable lifestyle modification by patients and help them achieve the goal of self-management of diabetes to reduce the disease burden [11-15]. As a mostly remote intervention form, digital therapeutic interventions (DTIs) may not be as impactful as in-person counseling and follow-ups, but have the advantage of ease of communication, anytime-anywhere accessibility, and availability of information. A DTI overcomes the barrier of physical transportation, particularly in restricted conditions such as during the COVID-19 pandemic [12-15]. The machine learning capabilities through cloud computing and interactive interfaces of smartphones have made behavior modification possible based on personalized nudges, information sharing, and communication [12-14]. The ease of intervention along with autonomy and constant reminders can keep a patient motivated in achieving the desired therapeutic goals. Several studies have demonstrated the role of digital technology such as digital therapeutic platforms, telehomecare systems, digitally enhanced diabetes self-management, education and support programs, as well as smartphone-based integrated online real-time diabetes care systems in the effective management of T2DM, without requiring escalation of existing medication [12-17]. However, there is also substantial variation in the reported intervention approaches and observed changes in outcome parameters. In view of these variations of the impact of DTIs, there is a requirement of measuring the effectiveness of a particular DTI approach before it can be scaled up for a larger target population.

The goal of our study was to assess the utility of a DTI in achieving glycemic control in Indian patients with T2DM. We hypothesized that lifestyle and behavior modification through the DTI would improve glycated hemoglobin (HbA_1c) of the participants from the preintervention level. Additional parameters to assess the effectiveness of the intervention included changes in fasting blood glucose (FBG) and postprandial blood glucose (PPBG) levels, BMI, and the Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) index. Moreover, we aimed to assess the potential relationships between user engagement and outcome results.

This was a 12-week, prospective, single-arm intervention study in Indian patients with T2DM receiving regular antidiabetic treatment from their physician. The sample size calculation was based on the study of Bollyky et al [18]. To detect a mean change in HbA_1c levels of –0.4% with a sample SD of 1.5, using a 5% level of significance, 80% power, and correlation coefficient of 0.5 with a two-tailed t test of paired mean difference, a sample size of 113 was needed. Assuming a dropout rate of 20%, a sample size of 136 was chosen for this study. The sample size was calculated using SAS software version 9.4 (SAS Institute Inc).

The eligibility criteria included male and female patients with T2DM, aged 18-65 years, HbA_1c≥6.5%, functional English literacy for use of the mobile app, and a smartphone capable of running the intervention app. Participants were also required to be on a stable dose of antidiabetic medications at the time of entry in the study, with the expectation to remain on the same stable dose during the study period. Subjects were excluded if they were unable or unwilling to provide informed consent and comply with the protocol procedures; had a previous diagnosis of gestational diabetes, myocardial infarction, or stroke; were pregnant or lactating; and had restricted physical movements as per the clinical judgment of the treating physician. The study was conducted between October 2021 and March 2022 at three metropolitan outpatient clinics catering specifically to patients with diabetes.

The study was conducted according to the ethical principles stated in the latest version of the Helsinki Declaration and the applicable guidelines for good clinical practice. Ethics committee approval for this study was obtained from the Good Society for Ethical Research–Independent Ethics Committee for Biomedical Research, Delhi (approval numbers: GSER/2021/BMR-AP/035, GSER/2021/BMR-AP/037, and GSER/2021/BMR-AP/039 dated October 8, 2021). The outpatient subjects who were willing to participate in the study were asked to sign a written informed consent form to assess their eligibility. Those unwilling to participate continued to receive the usual standard of care at the clinics.

The outcome measures of the study were mean changes in HbA_1c, FBG, PPBG, BMI, and HOMA-IR index at the end of the 12-week intervention from the respective baseline values. The data pertaining to the above variables were collected from the laboratory reports and SMBG readings. The outcomes were measured based on a pre-post change in the values of these variables. Participants were classified as “responders” if they showed an improvement in the HbA_1c value at the end of 12 weeks and as “nonresponders” if the HbA_1c value had either remained unchanged or had increased at the end of 12 weeks.

Based on the level of participation, participants were also divided into three groups: low-engagement group (50%-60% engagement), medium-engagement group (61%-80% engagement), and high-engagement group (>80% engagement). Program engagement was assessed subjectively by the health coaches based on the participant’s response to communication, dietary changes, physical activities, and SMBG. Participation was measured quantitatively in terms of the frequency of engagement components as reported by the participants themselves. A monthly report of progress and participation was shared with participants based on these data. The engagement score for final analysis was calculated from the arithmetic sum of individual scores and then converted into percentages. A questionnaire was further administered to the participants at the end of the intervention to assess the impact of the program on diabetes self-management. The responses were summarized in simple arithmetic measures.

Continuous variables are summarized by the arithmetic mean (SD). A paired-sample t-test or repeated-measures ANOVA (as applicable) was used to compare the change in the mean values of parameters at the end of the study. Categorical variables are summarized using frequencies and percentages. Statistical analysis was performed using the Pearson χ² or Fisher exact test, as appropriate, to test the association between the variables. Two-sided P<.05 was considered statistically significant. Simple logistic regression was performed to evaluate the association of each baseline characteristic with the achievement of HbA_1c response. Statistical analysis was performed using SAS version 9.4 (SAS Institute Inc).

A total of 146 subjects were screened for eligibility and 136 meeting the eligibility criteria were included in the study (Figure 1). Of these 136 subjects, 8 withdrew consent during the study period for personal reasons. Final results were analyzed for the remaining 128 subjects who completed the intervention period of 12 weeks (Figure 1). The final study population had a mean age of 48.48 years, with a slight majority of males. Baseline characteristics of the participants are presented in Table 1.

The general characteristics of T2DM of the participants at the time of entry into the study are presented in Table 2. Over 90% of the subjects had diabetes for ≥1 year. The mean baseline HbA_1c for participants was 8.32% and the mean baseline FBG was 139.16 mg/dl.

This study has shown that patients with diabetes benefit from a component of lifestyle modification through digital means along with routine care with antidiabetic medication. Lifestyle modification through digital therapeutics resulted in a decrease in HbA_1c, thereby bringing diabetes under control. In our study, a significant reduction in HbA_1c of 0.84% (SD 1.36; P<.001), in FBG of 8.39 mg/dl (SD 40.65; P=.02), and in PPBG of 14.97 mg/dl (SD 46.11; P<.001) were observed for the overall study population at the end of 12 weeks. These results are consistent with similar findings from multiple studies. A previous study from India that evaluated glycemic control in 102 patients with T2DM of South Asian origin using a digital therapeutic platform found a significant change (–0.49%, 95% CI −0.73 to 0.25; P<.001) in the mean HbA_1c level after 16 weeks of the intervention [12]. Berman et al [11] also showed a mean change in HbA_1c of –0.8% (SD 1.3; P<.001) after a 12-week intervention. A meta-analysis of 18 randomized controlled trials showed a significant change in mean HbA_1c levels (–0.54%, 95% CI –0.75 to –0.34; P<.05) with a telecare intervention [14]. A retrospective study of a digital health intervention in adults with T2DM also demonstrated a significant reduction in HbA_1c levels (–0.81%; P<.001) after 3 months [16].

A new approach is useful if the majority of patients are benefitting from it and if the benefit is highly probable. In our study, 76.6% (98/128) of the subjects benefitted from the digital intervention by achieving a reduction in their HbA_1c of 1.24% (SD 1.30; P<.001) from preintervention levels. This study used an inclusion criterion of HbA_1c≥6.5% at baseline, thereby ensuring close to real-world representation of patients with diabetes in terms of HbA_1c levels, in contrast to similar effectiveness studies that have considered higher levels of baseline HbA_1c such as above 7.5% [15,19] or 8% [20] as one of the inclusion criteria. This may have reduced the effective postintervention mean change in HbA_1c for the total sample population. Yet, similar to studies by Wilson-Anumudu et al [15] and Krishnakumar et al [12], we also found that individuals with higher levels of baseline HbA_1c stand to benefit more from lifestyle intervention. This finding has significance by indicating that rather than escalating the medication dosage or type, such patients may be advised to undergo a lifestyle intervention to bring their HbA_1c levels under control or at manageable levels, without any significant side effects.

Comparison with other DTIs is difficult as these approaches differ with regard to the use of tools, frequency of interventions, and stakeholders involved, among other factors [21]. Hence, measurement of the impact of each DTI approach separately might be warranted. The DTI approach used in this study included a significant human component by way of involvement of the primary care physician of the patients, apart from dieticians and health coaches. The presence of a patient’s physician in the DTI care team may have created the psychological impact that the participant’s health is being monitored by their physician.

As seen in Table 3, there was a significant reduction in the mean BMI (–0.34 kg/m²; P=.007) in the responders subgroup. However, this reduction was relatively minimal compared with the observed change in HbA_1c levels (–1.24%; P<.001). Further subanalysis suggested the possibility of the impact of ongoing antidiabetic medications causing weight gain, as the majority (107/128) of the participants were on sulfonylureas, thiazolidinediones, alpha-glucosidase inhibitors, and insulin. Future studies may be needed to understand the confounding impacts of antidiabetic drugs causing weight gain and weight loss, and simultaneous lifestyle modification guided toward weight loss.

With regard to the program engagement level, we hypothesized that greater program engagement should result in greater improvement in postintervention blood glucose parameters. However, in this study, all three engagement groups showed statistically significant improvement in HbA_1c postintervention. In fact, the subjects in the medium-engagement group exhibited the smallest change in HbA_1c of –1.16% (SD 1.22), as compared to the low- and high-engagement groups with an HbA_1c reduction of –1.31% (SD 1.45) and –1.30% (SD 1.32), respectively. Although the linear reduction in HbA_1c in the medium- and high-engagement groups is in line with our hypothesis, the higher reduction in the low-engagement group could be attributed to the effect of a smaller sample size. This needs to be evaluated in future studies, as there is a difference between the mere arithmetic sum of engagement parameters vis-a-vis the compliance provided by the patient to engagement by the care team.

The main limitations of our study are the lack of a control group, a short duration of follow-up to evaluate certain parameters such as BMI and HOMA-IR, and lack of measures to evaluate the compliance with respect to the level of program engagement. The prospective design of the study under a controlled environment, statistically derived sample size, and low dropout rate can be considered as the main strengths of the study. Further, one of the inclusion criteria of our study was patients on a stable dose of antidiabetic medication at the time of entry to the study, and they were expected to remain on the same stable dose during the study period. This minimized the scope of any potential observation bias and the baseline values of each subject served as their own control, thereby attributing the difference observed at the end of study to the study intervention. Further studies in this area in the form of randomized controlled trials are warranted.

The use of a DTI as an adjunct therapy to conventional medications significantly reduced HbA_1c, FBG, and PPBG levels in patients with T2DM. This in turn may reduce the risk of cardiovascular complications as well as all-cause mortality associated with T2DM.