This 3-month, 1-arm intervention study adopted a pre-post evaluation design. Baseline assessments, which consisted of paper-based questionnaires and physical measurements, were conducted in gymnasiums and other facilities arranged by the 7 municipalities in Kanagawa Prefecture and Kanagawa local government offices that participated in the study. The measurements were taken by the researchers or local municipal officials in Kanagawa Prefecture and Kanagawa local government officials trained by the researchers. Subsequently, the participants installed and set up a smartphone app called Minchalle (“Challenge yourself together with others” in Japanese) to commence the intervention.

Participants were required to send objective step count data for 1 week before the intervention as a baseline and during the 3-month intervention period via Minchalle. Objective step count data were measured using Google Fit (Google LLC) for Android users and Healthcare for iPhone users (Apple Inc), both capable of automatically counting steps while the phone is carried. Once Minchalle was connected to these apps, daily step count data were automatically transferred to and displayed in Minchalle. Following the 3-month intervention, participants were again required to complete paper-based questionnaires and undergo physical measurements as part of an endline survey. Endline measurements were performed in the same manner as the baseline assessments. In addition, to assess the validity of smartphone-measured data, step count measurements via an accelerometer (Active style Pro HJA-750C, OMRON Corporation) were conducted for 1 week before the intervention. Participants were required to wear the accelerometer around their waist while awake during this period. The study was conducted between December 2020 and June 2021.

This study was announced through email or online in the offices of 7 local municipalities in Kanagawa Prefecture and Kanagawa local government offices in Japan. Interested participants participated in briefing sessions to deepen their understanding of this study, and, if eligible, paper-based informed consent was obtained. Inclusion criteria were participants aged between 20 years and 75 years who owned and were accustomed to using a smartphone and could communicate in Japanese. Exclusion criteria were participants who were currently pregnant, had physical or medical problems that did not allow safe participation in the walking intervention, had experience using Minchalle over the past 6 months, or had undergone another face-to-face or digital interventional study for PA promotion. The recruitment was conducted between December 2020 and April 2021.

Minchalle was commercially developed by a health care company (A 10 Lab) to help users foster desirable habits. It is available on both Android and iPhone. Its development was guided by the social cognitive theory based on the need to increase self-efficacy, a key driver of behavior adoption and maintenance. To enhance self-efficacy, Minchalle adopted a group intervention to maximize the mechanisms of social support and comparison among group members, which thereby increased the retention rate and made substantial achievements. Based on this basic concept, the main function of Minchalle was to facilitate group interactions, limiting each group to a maximum of 5 members and a minimum of 3 members with similar goals. All users were required to share their daily achievements with other members in a chat box, accompanied by a photo taken on the same day as evidence of their activity. Users could also enjoy interactions with other members by commenting or chatting at any time, although this was not obligatory. This platform was designed to help users continue desired behaviors by receiving praise and encouragement from other members for their achievements and gaining inspiration from other members’ activities. Apart from group interaction, the examples of functions installed in Minchalle were goal setting, self-monitoring, reminders through push notifications, rewards for achievements such as exclusive stickers available within this app, and coins that could be used for donations to organizations that implement projects with social significance.

In this study, once the Minchalle app was installed, participants were anonymously assigned by researchers to a group that consisted of a maximum of 5 members. Each participant set their daily step goals based on their step counts from the past few days, without any instruction from the researchers. The intervention began with participants posting a photo of their total step count once a day in the group chat. To easily identify study participants, all the team members were composed exclusively of study participants. For their cooperation, participants were provided access to a premium version of Minchalle by A 10 Lab worth ¥500 (US $3.42) during the intervention period.

In a similar pilot study in Australia measuring the effectiveness of a social networking mobile app on improving PA, there were 55 participants. Therefore, we set a target of approximately 50 participants.

This study aimed to evaluate the feasibility of digital peer support interventions using retention rates. The secondary endpoints were the effectiveness of the intervention, measured via the change in step count between baseline and endline, and the association with the achievement rate of daily step goals.

Descriptive analyses were conducted to calculate the retention rate and rate of achieving daily step goals. The average daily step counts 1 week before and during the intervention period were calculated to measure the impact of the intervention, and the differences between the 2 were assessed. Pearson coefficients were used to assess the correlation between the rate of achieving daily step goals and the difference in average step counts before and during the intervention. Differences in weight, BMI, somatic fat rate, SBP, DBP, PA, lifestyle characteristics, and psychosocial factors between baseline and endline were analyzed via a 2-tailed t test. Furthermore, Pearson coefficients were examined between changes in average daily step counts and various variables, including physical measurements, lifestyle, and psychosocial factors, separately at baseline and endline. This enabled us to determine the variables to be used for subsequent multiregression analyses by omitting the possibility of multicollinearity among potential independent variables. Multiple regression analyses were conducted to demonstrate the impact of changes in daily step counts on health measurements, lifestyle, and psychosocial factors (Model 1), after adjusting for sex and age (Model 2), Model 2 + household income (Model 3), and Model 3 + average daily step count during the 1 week prior to the intervention (Model 4). Data analysis was performed using R software (version 4.2.2).

This study was approved by the Institutional Review Board of Kanagawa University of Human Services (30-018). The study was registered in the UMIN Clinical Trials Registry (UMIN000042520). All participants provided written informed consent. Data were collected in a face-to-face setting, and all user data were deidentified before analysis.

Figure 1 shows the study flowchart. Table 1 presents the baseline characteristics of the 62 participants. Participants’ ages ranged from 24 years to 63 years, with a mean age of 43.6 (SD 10.72) years. More than one-half (32/62, 52%) were men, and most (45/62, 73%) had a university degree or higher.

Among the 69 participants who signed the consent forms, 4 withdrew shortly after the intervention began due to personal conditions or smartphone defects, 2 did not meet the eligibility criteria, and 1 dropped out in the middle. Therefore, 63 participants installed the app and started the intervention, and the retention rate was 98% (62/63).

Data on daily step count were received from 54 (87%) of the 62 participants at the 3-month endline, and 8 participants failed to share step count data at endline due to misunderstanding the procedures of the survey when completing the intervention and endline questionnaires. Thus, data from 54 participants were used for step count analysis.

The average daily step counts during the week before the intervention began were 5811 (SD 2477) steps, as measured by Healthcare for iPhone or Google Fit by Android, and 7073 (SD 2301) steps, as measured by the accelerometer. Comparing the average daily step count from the smartphone app between baseline and during the intervention, an 1182-step increase was observed. The changes in average daily step count according to sex, age, and BMI at baseline are shown in Tables S1 and S2 in Multimedia Appendix 1.

The mean daily step goal at baseline was 6808 (SD 1668) steps. The mean difference between the daily step goal and step count before the intervention began was 997 steps (P=.002). The rate of achieving the daily step count during the intervention was 53.5% (SD 26.2%). A significant correlation (r=0.27, P=.05) was observed between achieving daily step goals and an increase in daily step count. There was a statistically negative correlation between the difference in the daily step goal and pre-intervention step count and an increase in daily step count (r=–0.77, P<.001).

Table 2 shows the comparative results of changes in health measurements, lifestyle characteristics, and psychosocial factors. Weight, BMI, somatic fat rate, SBP, and DBP were significantly different. Similarly, the daily amount of PA and engaging in hard labor significantly improved. However, no significant differences were observed in lifestyle characteristics nor psychosocial factors. The results of the groupwise analyses between those who maintained or increased their step count and those with a decreased step count during the intervention are summarized in Table S3 in Multimedia Appendix 1.

Table 3 shows the correlations between changes in step count, health measurements, and lifestyle characteristics measured at baseline and endline. A significant difference was observed between weight and BMI (r=–0.41, P=.002 for both variables). The correlations between the changes in step count and psychosocial factors are shown in Table S4 in Multimedia Appendix 1. Table 4 shows the results of the linear regression analysis. The increment in step count was significantly associated with weight, BMI, and SBP, even after adjusting for sex and age (Model 2: P=.02 for weight and BMI; P=.008 for SBP), household income (Model 3: P=.04 for weight; P=.03 for BMI; P=.02 for SBP), and all covariates (Model 4: P=.02 for weight and BMI; P=.01 for SBP). The results of the linear regression analysis to assess the association between changes in average daily step count and psychosocial factors are summarized in Table S5 in Multimedia Appendix 1.

This study demonstrated the feasibility, as indicated by the retention rate, of digital peer support. The finding that 98% of the participants continued to use the smartphone app even after the 3-month intervention shows that the retention rate in this study was higher than that of any other smartphone app equipped with the function of group interaction for promoting PA. A meta-analysis reported that engagement was generally low, while the retention rate of online social networks varies depending on the number of participants and the duration of the intervention [41]. A systematic review that examined the impact of social networks reported retention rates ranging from 60% to 80% [42]. An RCT assessing the efficacy of a smartphone app that facilitates interaction among users regarding walking over a 100-day intervention period reported that retention rates in the 60th percentile were higher than those in previous studies [16]. In a similar study evaluating a single-arm, 6-month intervention targeted at university students, in which user interaction was one of the components, the retention rate was 82% [21]. Considering these figures, digital peer support was effective at maintaining a high retention rate and can, therefore, be a promising tool for affordable and accessible health promotion. A reason may be that daily interaction, including social support and competition among participants and mutual comparison, serves as a positive factor for the prolonged use of a smartphone app, as shown in previous studies [18]. Another reason may be that the intervention period of this study was relatively short. Previous research has pointed out that the retention rate of smartphone apps generally remains high in short-term interventions that last from a few weeks to several months but tends to decrease after approximately 6 months from the beginning of the intervention [3]. We have conducted an RCT with a 6-month intervention period (UMIN000046936) to assess the effectiveness of digital peer support, including its long-term effects.

This study also showed that digital peer support had the potential to increase PA. Participants' daily average step count (+1182 steps), subjective daily amount of PA (+4.08 METs), and daily amount of engagement in hard labor (+2.28 METs) increased during the intervention period compared with 1 week before the intervention. Although step count was lower than those in a meta-analysis [4], which showed that an average 13-week intervention increased the number of steps per day by 1850, these results were consistent with those of other studies that showed increased objective PA [3,6]. The high retention rate in this study may have led to a significant increase in the amount of PA. These results indicate that digital peer support improved PA and daily step count. Additionally, this study demonstrated that the percentage of goal attainment was 53.5% (SD 26.2%). Furthermore, the statistical correlation between the rate of achieving daily step goals and the amount of increase in daily step counts was significant (r=0.27, P=.05), which was consistent with a previous study that reported that self-regulating, such as goal setting and self-monitoring, was a factor that influenced an increase in step count. Apart from self-regulation, previous studies revealed that other behavior change techniques, such as feedback, were also effective at improving PA [43]. Therefore, future studies should examine the impact of other behavior change techniques on digital peer support.

This study also suggested that digital peer support had the potential to improve health outcomes. Pre- and postintervention comparisons showed reduced weight (–1.26 kg), BMI (–0.47 kg/m²), somatic fat rate (–1.92%), SBP (–8.42 mm Hg), and DBP (–5.32 mm Hg). These results were consistent with meta-analyses that examined the effectiveness on health outcomes of smartphone apps that engage individuals [44]. Weight, BMI, and SBP could have decreased owing to the increased step count that resulted from smartphone app use. Participants’ mean age in this study was 43.8 years, which was similar to that in a previous study that found significant weight loss in participants with a mean age of 45 years or older [6]. However, it remains unclear in which age group digital peer support is particularly effective for promoting health outcomes. Changes in health outcomes are currently being investigated in our ongoing RCT (UMIN000046936), which will also examine which demographic backgrounds are likely to contribute to improved health outcomes.

This study demonstrated the possibility that digital peer support played an important role in increasing step count and thereby improved health outcomes with a high retention rate. However, several challenges remain unaddressed. First, it was a single-arm intervention. Since this study used a before-and-after comparison, other factors that may have affected step count and health outcomes could not be excluded. Furthermore, the effectiveness could not be verified in detail. Therefore, we are currently conducting an RCT (UMIN000046936). Second, the daily step count measured by smartphone apps may be underestimated compared with their true value. Our findings revealed a difference of approximately 18% in daily step count during the week before the intervention between the smartphone app (5811 steps) and accelerometer (7073 steps). This finding was consistent with those of previous studies that found that the number of steps measured by smartphones was 12% to 20% lower than the true value depending on where on the body the phone was being carried or due to intermittent bouts of activity [25,26]. Therefore, our daily step count results during the intervention should be interpreted based on these limitations. Third, a previous study reported that the retention rate for apps with social features is related to personal preferences. For example, users’ intent to use smartphone apps and a positive attitude toward technology, which are influenced by the simplicity or user-friendliness of the apps, were identified as positive factors for engagement [45,46]. The stage of behavioral change also affected the continued use of apps [18]. As this study did not collect information on personal preferences, future research should examine the effects of personal traits on the continued use of digital peer support. Fourth, the study did not elucidate participants’ specific backgrounds and lifestyles, examples of which were age, BMI [16], and being less physically active [21]. Although there was a slight increase in daily step count among middle-aged or older individuals or those who were overweight before the intervention, these were not statistically significant. However, future research is required to verify the groups or demographics more likely to increase their step count using digital peer support. Fifth, the intervention was conducted between the winter and spring seasons, and the impact of seasonality on PA was not considered, while previous research demonstrated PA level variations across the seasons, with a higher PA level in the summer compared with other seasons [47]. The possibility of seasonal and environmental effects should be considered when designing interventions in future research. Finally, the sample size in this study was 69 participants, which might have been underpowered to measure changes in health outcomes. The sample size was not calculated in this study because there were no prior studies examining the effectiveness of digital peer support on health outcomes. Future studies are needed to assess the effectiveness of digital peer support on both step count and health outcomes, so that effect sizes can be calculated appropriately.

This study demonstrates that digital peer support is feasible for maintaining high retention rates and can, therefore, effectively promote PA. Digital peer support can be a promising tool from a clinical perspective for improving daily step count, subjective PA, and health outcomes, such as weight, BMI, somatic fat rate, and blood pressure.