This was a single-center, double-arm, open-label, 12-week randomized controlled trial with a 12-week extension study comparing two groups: 1) that used a smartphone PHR app (control group) and 2) that used the app and received individualized motivational text messages every week to increase daily step counts based on the mean number of steps collected per day by the app (intervention group). The trial was conducted at the Diabetes Center, Division of Endocrinology and Metabolism of the Samsung Medical Center (SMC), Republic of Korea, and participants were recruited from April 2018 to September 2019. All the participants provided written informed consent. The Institutional Review Board (IRB) of the SMC approved the study protocol (No. 2017–12-061), which was in accordance with the ethical guidelines of the Declaration of Helsinki and Korea Good Clinical Practice. This study was registered at ClinicalTrials.gov (NCT03407222).

Eligibility criteria included patients with type 2 diabetes aged 20–69 years, with a HbA1c of less than 8.5%, who have not taken anti-diabetes medication for the past 4 weeks or who have taken more than or equal to one oral hypoglycemic agent for more than 12 weeks using the same dosage, who had overweight or obesity (body mass index (BMI) ≥ 23 kg/m2), who were able to use an Android smartphone and wireless internet, and who voluntarily agreed to participate. The exclusion criteria were as follows: diabetes other than type 2 diabetes, including type 1 diabetes and gestational diabetes; use of insulin or a GLP-1 receptor agonist; presence of comorbidities such as uncontrolled chronic liver disease, acute kidney injury, and psychological disorders; use of a weight-lowering agent; presence of alcohol or drug addiction within the previous 3 years; use of systemic corticosteroids; pregnancy or lactation; no voluntary agreement to participate in the study; and unsuitable for participating in clinical research.

This trial consisted of the following three periods: a one-week run-in period, a 12-week randomized treatment period, and a 12-week extension period (Figure S1). At visit 1 (week -1), the participants who met the inclusion criteria without meeting any exclusion criteria were provided smartphone PHR app to record blood glucose, blood pressure, and body weight, developed by Samsung Medical Center. The step count was measured using the Samsung Health application, which was automatically linked to the PHR app. The number of steps taken by all participants was automatically uploaded to the app, and the weight measured using the Bluetooth scale provided to all participants was continuously linked to the app. The baseline number of steps was measured during the one-week run-in-period. At week 0, the participants were assigned randomly at 2:1 ratio to the intervention or control group for 12 weeks. During the 12-week period, both groups used the smartphone PHR app, and the intervention group received text messages every week, which encouraged step-by-step increments according to the mean daily number of steps per week monitored by the app, whereas the control group did not receive text messages. The participants of intervention groups were divided into six groups such as 1) basal activity (< 2,500 steps/day), 2) limited activity (2,500–4,999 steps/day), 3) low activity (5,000–7,499 steps/day), 4) somewhat active (7,500–9,999 steps/day), 5) active (10,000–12,499 steps/day), 6) highly active (≥ 12,500 steps/day) according to the average number of steps per day. [13]). The average number of steps taken by each participant per day during the week was calculated and divided them into five groups. We suggested a step goal based on participants’ average number of steps in the past week. If a participant took an average of 9,000 steps per day during the week, a message was sent: 'You walked less than 10,000 steps a day on average this week. Next week, aim to walk more than 10,000 steps a day! Making changes to your lifestyle can improve your health.' During the extension period, for 12 weeks from week 12 to week 24), encouraging text messages were discontinued in the intervention group to check the durability of the intervention.

Demographic, anthropometric, and laboratory data were collected from all participants. Body weight and height were measured, and the BMI was estimated as body weight (kg) divided by height squared (m2). Smoking status and alcohol consumption data were collected using a self-reported questionnaire and classified as never, ever, or current, and either less than a cup per day or not, respectively. The estimated glomerular filtration rate (eGFR) was estimated using the Chronic Kidney Disease Epidemiology Collaboration 2021 (CKD-EPI 2021) formula [14]. Medical history data, including duration of diabetes and type of medication, were collected. During the study period, the average step count per week, laboratory data, body weight, and physical activity were obtained at weeks 12 and 24 in both groups. Physical activity was measured using the 7-item International Physical Activity Questionnaire-Short Form (IPAQ-SF) designed by a team of specialists in physical activity for population-level tracking of adult physical activity [15, 16]. Physical activity was classified with the following three categories: walking, moderate physical activity, and vigorous physical activity [17]. We used the total Metabolic Equivalent of Task (MET)–min/week to express weekly metabolic engagement in walking and in both moderate and vigorous physical activity practice.

The primary outcome was the change in daily step count between the control and intervention groups after 12 weeks of intervention. Secondary outcomes included e mean daily step count at weeks 12 and 24, mean HbA1c levels at weeks 12 and 24, fasting glucose levels at weeks 12 and 24, body weight at weeks 12 and 24, physical activity at weeks 12 and 24, and lipid levels including total cholesterol, low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, and triglycerides at weeks 12 and 24.

Assuming a 1,000 step difference in daily step count increments between the study groups (standard deviation, 2,150 daily steps) according to previous studies [18,19,20], a sample size of 55 and 111 participants for control and intervention group, respectively was needed for a two-sided alpha threshold of 0.05 and 80% power. We aimed to recruit 200 participants to allow for a dropout rate of 17% during follow-up (control group, n = 66; intervention group, n = 134).

Continuous and categorical variables are summarized as mean and standard deviation (SD) and percentages, respectively. Statistical analysis of the primary outcome was performed according to the intention-to-treat principle in the full analysis set (FAS) (i.e., all randomized participants used the smartphone PHR app and had baseline measurements and at least one measurement during the study period). Two-sample t-tests for continuous variables and chi-square tests for categorical variables were used to compare the baseline characteristics between the intervention and control groups. For the primary outcome, we conducted a two-sample t-test for the difference in the mean change in the daily step count between the intervention and control groups after 12 weeks. For secondary outcomes, we conducted independent t-tests for the mean difference at 12 and 24 weeks and change in mean difference from baseline to 12 and 24 weeks in average steps per day, HbA1c, fasting glucose, body weight, and total MET-min/week. A paired t-test was used to assess the mean difference in the intervention and control groups from baseline to 12 and 24 weeks. Although it is not pre-specified, subgroup analyses were conducted in the FAS, stratified by the baseline HbA1c levels (< 7.5% and ≥ 7.5%) and baseline daily step counts (< 7,500 and ≥ 7,500 steps). The mean HbA1c level for the entire population was 7.1% and to differentiate between low and high HbA1c groups with appropriate sample sizes, a cut-off value of 7.5% was chosen (n = 152 vs n = 30). A previous study conducted on the Asian population found that taking more than 7,500 steps per day was associated with a decrease in BMI and body fat, compared to taking fewer than 7,500 steps [13]. To identify the differences in outcomes by group and time, we conducted a linear mixed effect model for step counts and hemoglobin A1c considering random slope and intercept. All statistical analyses were performed using R software, version 4.1.3 (R Foundation for Statistical Computing),. Two-sided P values < 0.05 were regarded as significant.