This 3-month single-arm intervention study used a pre–post evaluation design. According to the transtheoretical model [13], a 3-month period is widely regarded in many studies as sufficient for individuals to progress from the “preparation stage” to the “action stage” and, subsequently, to the “maintenance stage.” In addition, the 3-month period is both physiologically justified and conceptually appropriate for evaluating behavioral and metabolic changes [3,4], as HbA_1c reflects the average blood glucose levels over the preceding 2‐3 months, which corresponds to the approximately 120-day lifespan of red blood cells [14].

This study was conducted between December 2021 and June 2022 as an industry-government-academia collaboration among local governments, app-making companies, and universities. It was part of the “TRY! YAMANASHI! Pilot Experiment Support Project (FY2021, First Term)” in Yamanashi Prefecture. The program supports the social implementation of advanced technologies and services developed by startups and other innovators to address regional issues and create public value. Support also offered multifaceted assistance (eg, funding, expert consultation, partnerships with local resources, and public outreach) for projects that had advanced to the practical application stage through prior pilot testing.

Informed consent was obtained from all participants before their enrollment. This study adhered to the ethical principles outlined in the Declaration of Helsinki and was approved by the independent Ethics Review Committee of Healthcare Systems Co., Ltd. (Approval No. 2130). Participants received a financial reward of 5000 JPY (US $34.70) in the form of a gift card when the end-of-study data were delivered (ie, HbA_1c levels at both time points and patient-reported outcomes at baseline and at 3 months). All collected data were anonymized prior to analysis to ensure participant confidentiality and privacy.

Participants were recruited through convenience sampling and began using the digital peer-supported app upon recruitment. An explanation of the study and its objectives was provided by physicians when they visited the clinic and hospital. The referral process involved the following scenarios: (1) during a health consultation at a hospital; (2) as part of a diabetes program at a hospital; and (3) during a medical consultation at a clinic.

The inclusion criteria were (1) individuals who had been fully informed about the objectives and details of the study, possessed the capacity to consent, had understood the information thoroughly, and have freely agree to participate by providing written consent; (2) Japanese men and women aged 20-74 years at the time of consent; (3) patients with T2DM or prediabetes with an HbA_1c level of ≥6.0% and <8.5%; and (4) individuals who owned and could operate a smartphone (iOS or Android) capable of installing the app. The exclusion criteria were (1) diseases that could interfere with the study, as determined by the medical institution conducting the study, and (2) severe symptoms or disabilities due to lifestyle-related diseases, as determined by the medical institution.

The experimental setup is shown in Figure 1. The participants began using the digital peer-supported app after being introduced to the app by physicians and downloading it. During the initial login, they received tutorials within the app. Participants were encouraged to join diabetes management teams within the digital peer-supported app and use this app for 3 months while continuing their treatment according to the standards of care, including any necessary adjustments to their diabetes pharmacotherapy. They were instructed to complete questionnaires at the beginning and end of the program to evaluate changes in their lifestyle habits and health behaviors. These questionnaires were designed to confirm participant sex, age, and health behaviors. HbA_1c levels, blood pressure, and body weight were measured at the beginning and end of the study, using the most recent data from hospitals and clinics.

The study used “Minchalle,” a commercially available digital peer-supported app [11]. This digital peer-supported app was developed by A10 Lab Inc. in June 2015 with an initial release in November 2015. It is available for both Android and iPhone devices. The digital peer-supported app created group chats of up to five participants that aimed at controlling their HbA_1c levels, and participants were anonymously assigned to each group (Figure 2). Once daily, the participants posted a set of daily activities to control their HbA_1c levels, along with photos and comments in a group chat box. The main functions of the digital peer-supported app used in this study were (1) posting photos, step counts, and comments about the day; (2) posting approvals from group members; (3) setting step-count goals on a group basis; and (4) providing feedback from assistant robots. Participants could post comments or photos more than once a day and enjoy interacting with other members; however, daily participation was not obligatory, and the participants were allowed to freely use the app. Examples of the user interfaces of digital peer-supported apps are provided in the Multimedia Appendix 1.

We collected basic demographic information at baseline, including sex (male or female), age (in continuous years), alcohol consumption (categorized as nondrinker, former drinker, occasional drinker, or nearly daily drinker), and smoking status (categorized as nonsmoker, former smoker, or current smoker).

The primary outcome of this study was the difference in HbA_1c levels, which were evaluated using values measured at health care facilities immediately during the participants’ visits at baseline and at the 3-month follow up. All participants’ baseline and 3-month follow-up values were measured at the same health care facilities using the same reagents. The secondary outcomes included differences in body weight and blood pressure, which were also obtained from the health care facilities. At the same time points, physical activity (≥1 hour/day) was evaluated using a self-reported questionnaire: “Do you engage in walking or equivalent physical activity for >1 hour per day in your daily life?” Responses were recorded as either “yes” or “no.”

We performed the Wilcoxon signed-rank test to assess the overall significance of differences in HbA_1c levels. Similarly, changes in secondary outcomes such as blood pressure and body weight were evaluated using the Wilcoxon signed-rank test to compare baseline and 3-month follow-up values. Additionally, McNemar test was used to analyze paired binary data, specifically to compare baseline and 3-month follow up values for physical activity, ≥1 hour/day (%).

All analyses were performed using the Stata software (version 18.0; StataCorp LLC). Statistical significance was set at P<.05 (2-tailed).

Our findings showed that a 3-month intervention using a digital peer-supported app led to a modest but statistically significant reduction in HbA_1c levels among patients with T2DM, indicating improved glycemic control. Additionally, self-reported frequency of physical activity improved significantly, highlighting the potential of this app to promote healthier behaviors. The findings of this study suggest that digital, peer-supported apps may serve as promising practical tools for the management of T2DM.

Moreover, the intervention using a digital peer-supported app significantly reduced HbA_1c levels among patients with T2DM, which is consistent with findings from previous studies that used nongroup-based, single-user apps for blood glucose control [15-17]. Peer support, grounded in the social cognitive theory, could effectively aid in blood glucose control when implemented through a digital peer-supported app [18].

Our results may be attributed to specific app design features, including reminder prompts and chat functions. One possible explanation is that communication via chats may also impact individual decision-making and behavior, shaping online community norms to maintain healthy behaviors [19]. These community norms may offer promising ways to improve adherence to diet and exercise recommendations by providing real-time reminders and emotional support from peers [20].

Previous studies on diabetes apps have highlighted the importance of chat-based communication in promoting healthy behavioral change, particularly by reinforcing positive actions such as initiating exercise, which can be more effective than restrictive measures [21]. Alternatively, real-time chats and reminders may help users remain frequently aware of their goals and consistently track their behaviors, which may facilitate effective acquisition, learning, and app of health information [15,22]. Although digital peer support platforms carry the risk of misinformation or inappropriate advice, no such incidents were reported during the study period. This may be attributed to the implementation of a safety-monitoring system, which included content oversight by the app provider and a dedicated research contact point. Future interventions should incorporate risk-management protocols to ensure safe and effective peer communication.

Our results showed a statistically significant change in HbA_1c levels, although the reduction did not reach a threshold considered meaningful in a clinical context [23]. One possible explanation for this finding is that the participants may have had low baseline HbA_1c levels. In a recent meta-analysis of online medicine, higher baseline HbA_1c levels (>7.5% or >8.0%), younger age (<55 years), and shorter duration of diabetes (<8.5 years and <7 years) were found to be associated with increased benefits [23]. Our study participants had lower baseline HbA_1c levels than those in other studies [15,16,24]. Future studies should aim to identify demographic or behavioral subgroups more likely to achieve HbA_1c reduction using digital peer-supported apps.

This study reported that a 3-month intervention using a digital peer-supported app led to a modest but statistically significant reduction in HbA_1c levels, which indicates improved glycemic control among patients with T2DM. Additionally, self-reported frequency of physical activity improved significantly, highlighting the potential of this app in promoting healthier behaviors. By minimizing interference from study personnel and addressing only technical aspects, the app was evaluated under real-life conditions to generate realistic and practical evidence. Therefore, digital peer-supported apps may offer cost-effective and scalable tools for managing diabetes and supporting lifestyle changes.

However, this study had several limitations that should be noted when interpreting our findings. First, as this was a single-arm interventional study with a small sample size, the analysis relied on classical statistical methods and a pre-post comparison, which makes it difficult to exclude the influence of confounding factors, such as baseline HbA_1c, age, or sex, through statistical adjustment. Although the reductions in HbA_1c and body weight were statistically significant, their relatively small magnitude along with the absence of detailed data on app usage, user engagement, and acceptability limit our ability to attribute these improvements solely to digital peer support interventions. These improvements may also have been mediated by increased adherence to treatment rather than by direct effects. To enhance the validity and generalizability of these findings, a multicenter randomized controlled trial with a larger and more diverse sample size and a parallel control group is currently underway (UMIN000056609). Second, this study used only HbA_1c level as the primary outcome measure. Although HbA_1c is a reliable indicator of long-term glycemic control and reflects average blood glucose levels over the past 2‐3 months, it is subject to a time lag between behavioral changes and measurable outcomes. In contrast, continuous glucose monitoring devices can capture interstitial glucose levels in near real time, allowing immediate feedback and short-term trend analysis [25]. Future studies may benefit from combining long-term markers such as HbA_1c with continuous glucose monitoring to investigate the effects of interventions more comprehensively. Third, unmeasured factors such as medication adjustments and dietary habits during the study period may have influenced changes in HbA_1c levels. Although the duration since the diagnosis of diabetes may influence a participants’ engagement with peer support and intervention outcomes, this information was not collected in the present study and should be considered in future studies. Fourth, this study did not collect information on personal or behavioral traits. As previous studies have shown that the stage of behavioral change may influence continued app usage [26], future studies should examine how individual characteristics affect long-term engagement in digital peer support. Finally, the sample size of 21 participants was relatively small and may have been insufficient to adequately detect significant changes in health outcomes. A formal sample size calculation was not conducted due to the lack of prior studies on the effectiveness of digital peer support on health outcomes. Future studies should assess the impact of digital peer-supported apps on HbA_1c levels, enabling appropriate calculation of effect sizes.

The use of a digital peer-supported app intervention may significantly lower HbA_1c levels in patients with T2DM after 3 months. In T2DM populations capable of using the app, this intervention could promote healthier behaviors, in addition to regular clinical care, and may help reduce HbA_1c levels.