This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR mhealth and uhealth, is properly cited. The complete bibliographic information, a link to the original publication on http://mhealth.jmir.org/, as well as this copyright and license information must be included.
Singapore’s current prevalence of diabetes exceeds 13.6%. Although lifestyle modification can be effective for reducing the risks for complications of type 2 diabetes mellitus (T2DM), traditional lifestyle interventions are often difficult to administer in the primary care setting due to limited resources. Mobile health apps can address these limitations by offering low-cost, adaptable, and accessible platforms for disseminating lifestyle management interventions.
Using the RE-AIM evaluation framework, this study assessed the potential effectiveness and feasibility of GlycoLeap, a mobile lifestyle management program for people with T2DM, as an add-on to standard care.
This single-arm feasibility study recruited 100 patients with T2DM and glycated hemoglobin (HbA1c) levels of ≥7.5% from a single community health care facility in Singapore. All participants were given access to a 6-month mobile lifestyle management program, GlycoLeap, comprising online lessons and the Glyco mobile phone app with a health coaching feature. The GlycoLeap program was evaluated using 4 relevant dimensions of the RE-AIM framework: (1) reach (percentage who consented to participate out of all patients approached), (2) effectiveness (percentage point change in HbA1c [primary outcome] and weight loss [secondary outcome]), (3) implementation (program engagement as assessed by various participatory metrics), and (4) maintenance (postintervention user satisfaction surveys to predict the sustainability of GlycoLeap). Participants were assessed at baseline and at follow-up (≥12 weeks after starting the intervention).
A total of 785 patients were approached of whom 104 consented to participate, placing the reach at 13.2%. Four were excluded after eligibility screening, and 100 patients were recruited. Program engagement (implementation) started out high but decreased with time for all evaluated components. Self-reported survey data suggest that participants monitored their blood glucose on more days in the past week at follow-up compared to baseline (
Although measures of program engagement decreased with time, clinically significant improvements in HbA1c were achieved with the potential for broader implementation. However, we cannot rule out that these improvements were due to factors unrelated to GlycoLeap. Therefore, we would recommend evaluating the effectiveness and cost effectiveness of GlycoLeap using a randomized controlled trial of at least 12 months.
ClinicalTrials.gov NCT03091517; https://clinicaltrials.gov/ct2/show/NCT03091517 (Archived by WebCite at http://www.webcitation.org/77rNqhwRn)
The worldwide prevalence of diabetes has almost doubled over the past three decades and continues to rise [
Lifestyle modification can be highly effective at reducing the risks for complications of T2DM [
Given the rising health and cost implications of diabetes in Singapore, a comprehensive program that incorporates the most effective mHealth strategies and personalized health coaching offers a potentially scalable model to address Singapore’s diabetes epidemic. One potential program is GlycoLeap (Holmusk), a proprietary lifestyle management program for adults with T2DM. The GlycoLeap program, which was originally developed for use in Singapore, comprises two components: a comprehensive T2DM educational curriculum delivered through online lessons and the Glyco mobile phone app with a health coaching feature. The Glyco app enables users to log and monitor their blood glucose levels, weight, meals, and physical activity, which is captured via the mobile phone’s built-in pedometer. The app also serves as a vehicle for accredited dietitians, known as health coaches, to provide personalized feedback to participants on their progress and to present opportunities for improvement.
The goal of this study was to assess the potential effectiveness and feasibility of the GlycoLeap program as an add-on to standard care using the RE-AIM evaluation framework [
This effort consisted of a 6-month (24-week), single-arm, preintervention (baseline) and follow-up evaluation. One hundred participants were recruited by a research coordinator from a single community health care facility, SingHealth Polyclinics–Tampines, in Singapore. All participants signed an informed consent to participate in a baseline and follow-up assessment that was conducted at least 12 weeks after starting the intervention and to allow for medical records abstraction. At the baseline assessment, participants completed a brief survey (
Participants were eligible to participate if they (1) were aged 21 to 70 years, (2) had been medically diagnosed with T2DM as listed in electronic health records, (3) had an HbA1c result of ≥7.5% within the past 2 months, (4) had a body mass index (BMI) of >23 kg/m2, (5) were not on insulin, and (6) owned and were able to use an iPhone or Android mobile phone. Participants were excluded if they (1) had cancer requiring treatment in the past 5 years, (2) had cardiovascular diseases (heart attack or cardiac procedure within the past 3 months), (3) had stroke or history of treatment for transient ischemic attacks in the past 3 months, (4) had chronic renal failure or were on dialysis, (5) had any amputation of lower limbs, (6) were using medication for weight loss, (7) had chronic treatment with systemic corticosteroids, (8) had bariatric surgery or extensive bowel resection, (9) were unable to converse in or read and write English, or (10) did not have a valid HbA1c blood test within the 2 months prior to the date of recruitment. Each participant was compensated SGD20 (approximately US $15) in vouchers at study completion. All study procedures were approved by the SingHealth Centralized Institutional Review Board (CIRB Ref: 2017/2013) and the study is registered at ClinicalTrials.gov [NCT03091517].
All eligible participants received 24 weeks of free, unlimited access to the GlycoLeap program. Participants downloaded the Glyco app onto their mobile phones upon recruitment. An Accu-Chek Performa (F. Hoffmann–La Roche Ltd) glucometer kit with lancets and test strips, a BodyTrace (BodyTrace Inc) wireless weighing scale, and a resistance band for strength training were provided at no cost to participants. Although the recommendation to switch glucometers may have been an inconvenience, this was not raised as a concern by any of the participants. Participants were also given printed instruction guides on operating the Glyco app and devices and two guidebooks educating them on how to make healthier food choices and achieve weight loss.
Description of GlycoLeap program components and recommendations for engagement.
Component | Description | Recommended frequency |
Online health lessons on diabetes and self-management | A total of 24 educational lessons on diabetes and self-management were delivered online. This curriculum was adapted for the local population and covers topics that take reference from the 7 healthy self-care behaviors as described by the American Association of Diabetes Educators. Quizzes tested knowledge on diabetes, obtained information about participants’ lifestyle habits, and were designed to keep participants engaged throughout each lesson. | Complete one lesson (lasting about 15 minutes) per week |
Blood glucose monitoring | Blood glucose measurements obtained using the Accu-Chek Performa glucometer kit were input manually by participants into their Glyco app accounts. | At least 4 blood glucose logs per week (preferably paired pre- and postmeal readings) |
Weight monitoring | Wireless weighing scale readings were automatically synced to participants’ accounts via cellular connectivity (3G). | At least one weight log per week |
Meal logging | Meal photos taken by participants were uploaded onto the app for health coach evaluation. Health coaches rate meals using a 1 to 5 linear scale. Meal scores are awarded based on the balance of nutrients, food quality, and nutritional content. The meal scores take reference from the Singapore Health Promotion Board’s national dietary guidelines. | No recommendation was provided. Participants were encouraged to log as often as they wanted to. |
Physical activity tracking | The Glyco app tracks the number of daily steps taken using the participants’ built-in phone pedometers. | 70,000 steps per week |
Health coach | Health coaches rate and respond to all meal logs and regularly send messages to participants to provide recommendations, encouragement, and personalized feedback on progress and answer participants’ questions ( |
No recommendation was provided. Participants were encouraged to engage as often as they wanted to. |
The GlycoLeap program was evaluated using relevant dimensions of the RE-AIM framework [
A proxy for Reach was used and defined as the percentage of those who gave informed consent to participate out of all patients approached.
As this was a feasibility study without a control group, measures of potential effectiveness were assessed as changes in HbA1c levels (primary outcome) and weight (secondary outcome) between baseline and the follow-up. The analysis was conducted separately on the total sample with baseline values carried forward for those with missing data at follow-up (intention-to-treat [ITT] analysis) and on those who completed the study and did not initiate insulin (per-protocol analysis). All HbA1c tests were conducted using the polyclinic’s protocols and approved laboratories, and weight was measured using validated weighing scales at the polyclinic. The window for eligible tests was defined as within 2 months before the scheduled baseline assessment and from 12 weeks to a maximum of 38 weeks after the intervention start date for the follow-up. In addition to the above indicators of potential effectiveness, the association between changes in health outcomes and measures of program engagement (as defined in the Implementation section) were evaluated using the per-protocol approach with the expectation that participants with greater levels of program engagement will show greater improvements in health outcomes.
Adoption tends to focus more on system level factors and was not captured as part of this feasibility study.
Implementation was assessed on the total sample by exploring engagement with key components of the GlycoLeap program as shown in
Within-trial maintenance is included in the Implementation domain. Here, we assess sustainability at the setting level by exploring user satisfaction at follow-up (
Power calculations were not performed because this was a feasibility study. The target sample size of 100 was selected taking into consideration the aims of the study (testing acceptability and potential for effectiveness rather than efficacy) and practical feasibility. Studies on diabetes self-management mobile app interventions often involved a sample size of less than 100, including controlled trials [
Paired or one-sample
Between June and November 2017, a total of 785 SingHealth Polyclinics–Tampines patients were approached by the research coordinator. Of these patients, 681 declined to participate or were deemed ineligible based on a prescreen assessment, placing the Reach at 13.2% (
The average age of participants was 54 years old, and 50 were male. Of the 100 participants, 45 were Chinese, 29 were Malay, 18 were Indian, and 8 were of other ethnicity. Sixty-one participants had high school-equivalent or lower education, and 69 were employed. At baseline, the mean HbA1c and weight were 8.8% and 79.7 kg, respectively. On average, participants were diagnosed with T2DM 9.3 years ago. Other baseline characteristics can be found in
Thirteen of the 100 participants withdrew from the study, either due to insulin initiation or free will (
Participant recruitment and retention Consolidated Standards of Reporting Trials flow diagram.
Baseline characteristics of participants in the single-arm GlycoLeap feasibility study.
Characteristics | Total (n=100) | Completes (n=83) | ||
Age (years), mean (SD) | 53.5 (9.6) | 53.5 (9.7) | .98 | |
Male, n (%) | 50 (50) | 44 (53) | .67 | |
Weight (kg), mean (SD) | 79.7 (16.8) | 79.2 (15.6) | .87 | |
Height (cm), mean (SD) | 163.1 (8.9) | 163.5 (9.3) | .75 | |
Body mass index (kg/m2), mean (SD) | 29.8 (5.0) | 29.5 (4.6) | .68 | |
Systolic blood pressure (mm Hg), mean (SD) | 132.1 (11.6) | 132.5 (11.8) | .83 | |
Diastolic blood pressure (mm Hg), mean (SD) | 74.7 (10.6) | 74.9 (10.5) | .88 | |
HbA1ca (%), mean (SD) | 8.8 (1.6) | 8.9 (1.7) | .96 | |
Years diagnosed with diabetes, mean (SD) | 9.3 (7.3) | 8.8 (6.3) | .56 | |
On oral medication for diabetes treatment, n (%) | 98 (98) | 81 (98) | .85 | |
Chinese | 45 (45) | 38 (46) | — | |
Malay | 29 (29) | 22 (27) | — | |
Indian | 18 (18) | 15 (18) | — | |
Other | 8 (8) | 8 (10) | — | |
High school or lower | 61 (61) | 49 (59) | — | |
Precollege (A-levels/polytechnic diploma) | 21 (21) | 17 (20) | — | |
College graduate/postgraduate | 18 (18) | 17 (20) | — | |
Never married | 12 (12) | 10 (12) | — | |
Married | 82 (82) | 68 (82) | — | |
Other | 6 (6) | 5 (6) | — | |
Less than SGD 5000 | 10 (10) | 6 (7) | — | |
SGD 5000-SGD 9999 | 2 (2) | 2 (2) | — | |
≥SGD 10,000 | 1 (1) | 1 (1) | — | |
Prefer not to say | 87 (87) | 74 (89) | — | |
Working (full-/part-time) | 69 (69) | 59 (71) | — | |
Homemakerb | 22 (22) | 17 (20) | — | |
Retired and not workingc | 9 (9) | 7 (8) | — |
aHbA1c: glycated hemoglobin.
bIndividuals who are full-time housekeepers, regardless of prior employment status.
cIndividuals who were previously employed until retirement and are typically past retirement age.
Generally high participant engagement was observed for all components in the first week which then decreased progressively over time. Trends in program engagement for the total sample (n=100) are shown in
In total, two of the 100 participants did not engage with any of the 5 evaluated components, and 14 engaged with at least one component every week throughout the intervention period. Thirteen participants engaged with the same component(s) every week throughout the intervention period: health lessons (2 participants), blood glucose monitoring (3 participants), weight monitoring (6 participants), meal logging (7 participants), and health coach messaging (5 participants).
Self-reported changes in diabetes self-care activities and lifestyle behaviors at follow-up from baseline are show in
Proportion of program engagement by week for the total sample (n=100). Percentage of participants who (A) completed at least one lesson, (B) logged 1, 2, 3, or ≥4 glucose measurements, (C) logged at least one weight log, (D) logged at least one meal log, or (E) sent at least one message to their health coach a week.
Self-reported diabetes self-care activities and lifestyle behaviors at baseline and follow-up (n=80).
Category and behavior or activity | Baseline | Follow-up | Difference | ||
Days that blood glucose was monitored in past week | 0.6 (0.2 to 1.0) | 2.3 (1.9 to 2.7) | 1.7 (1.3 to 2.1) | <.001 | |
Positive change in diet due to app engagement, n (%) | —a | 68 (85) | N/Ab | <.001 | |
Days with fruit and vegetable consumption as per recommended servings in past week, mean (95% CI) | 1.3 (0.8 to 1.8) | 3.7 (3.1 to 4.2) | 2.4 (1.6 to 3.1) | <.001 | |
Days with high fat food consumption in past week, mean (95% CI) | 2.3 (1.9 to 2.7) | 1.6 (1.2 to 2.0) | –0.7 (–1.1 to –0.2) | .003 | |
Days with at least 30 minutes of continuous activity including walking in past week | 3.4 (2.7 to 4.0) | 3.9 (3.4 to 4.4) | 0.5 (–0.2 to 1.2) | .14 | |
Increased | — | 30 (38) | N/A | ||
Decreased | — | 0 (0) | N/A | ||
Stayed the same | — | 50 (63) | N/A |
aQuestion was not present in the baseline survey as it asks for self-reported change due to app engagement.
bN/A: not applicable as question was not present in the baseline survey.
Changes in HbA1c levels and weight are shown in
Using the per-protocol approach, linear regression showed that HbA1c decreased by an average of 1.0 percentage point more among those who logged their weight more (
Distributions of health outcomes at baseline and follow-up: (A) HbA1c for total sample (n=100), (B) HbA1c for completes (n=83), (C) weight for total sample (n=100), and (D) weight for completes (n=83).
Changes in health outcomes at follow-up compared to baseline for participants who reached follow-up.
Health outcomes and measures | Total (n=100) | Completes (n=83) | |||
Percentage point change in HbA1c, mean (95% CI) | –1.1 (–1.4 to –0.7) | <.001 | –1.3 (–1.7 to –0.8) | <.001 | |
Participants with ≥1 percentage point reduction, n (%) | 49 (49) | 49 (59) | |||
Weight change expressed as a percentage of baseline weight (%), mean (95% CI) | –2.0 (–2.8 to –1.2) | <.001 | –2.3 (–3.3 to –1.4) | <.001 | |
Participants with loss of ≥5% of initial baseline weight, n (%) | 17 (17) | 17 (21) |
aHbA1c: glycated hemoglobin.
Among the 80 participants who completed the follow-up survey, the average user-friendliness rating for various program components was 3.6 to 3.8 out of 5, where 5 = very easy (
A reduction in HbA1c levels of 1 percentage point has been shown to be associated with a 21%, 14%, 37%, and 21% decrease in risk of any end point related to diabetes, myocardial infarction, microvascular complications, and diabetes-related death, respectively [
In spite of relatively high levels of engagement across Glyco app components in the first week, the decreased usage over time advocates that more should be done to improve and sustain engagement. Nevertheless, despite low completion rates for the online lessons, self-reported increases in blood glucose monitoring frequency and improvements in dietary habits argue that a mobile intervention like GlycoLeap may be a viable strategy for patient education and behavior change. As the online lessons were administered on a different platform and required separate email access, this may have presented a barrier to access. A greater completion rate may have been achieved if the lessons were made available directly on the mobile app.
Unlike conventional T2DM lifestyle management programs administered in the primary care setting through in-person sessions, mHealth interventions comprising mobile phone apps like Glyco are highly scalable, requiring comparatively fewer manpower resources. Although we were unable to determine the true reach as defined by the RE-AIM framework [
This study also suggests that GlycoLeap might be a scalable intervention. The app received good ratings with 21% (17/80) of participants claiming they were willing to purchase unlimited access to use the app and health coaches at a modest fee. In addition, the GlycoLeap program received relatively high user-friendliness and user-satisfaction scores compared to other similar mobile phone apps for diabetes management [
These results suggest that GlycoLeap may be an effective strategy for helping some adults with T2DM attain better diabetic control and that it is feasible to integrate it within the primary care setting. However, given the study design, we cannot exclude the possibility that any health improvements were due to factors unrelated to the GlycoLeap program. Therefore, future efforts should assess the effectiveness and cost effectiveness of GlycoLeap using a randomized controlled trial of at least 12 months to evaluate longer term outcomes.
Baseline survey.
Follow-up survey.
Glyco app home screen.
Glyco app blood glucose monitoring.
Glyco app weight monitoring.
Glyco app meal log.
Glyco app health coach correspondence.
Output of linear regression with change in HbA1c regressed on measures of program engagement.
Average participant rating of the user-friendliness of different GlycoLeap program components, where 1=very difficult and 5=very easy.
Average participant rating of the perceived usefulness of different GlycoLeap program components, where 1=not useful at all and 5=very useful.
Distribution of overall rating of the Glyco app, where 1=very poor, 2=poor, 3=average, 4=good, and 5=very good.
body mass index
hemoglobin A1c (glycated hemoglobin)
intention-to-treat
mobile health
type 2 diabetes mellitus
This work was supported by KKT Technology Pte Ltd (Holmusk). The sponsor funded the study, provided participants with the GlycoLeap program, collected program engagement data, and provided the program engagement data for analysis. The sponsor had involvement in the study design and the decision to submit the report for publication. The sponsor had no involvement in the analysis and interpretation of data, or writing of the report. We acknowledge Uma Rani, daughter of Singaram Samidurai, for recruiting patients and conducting follow-up assessments, Sankha Mukherjee for collating the program engagement data, and Mijie Li for contributing to manuscript writing.
EAF was the principal investigator and is the corresponding author. DK, PSCG, TYY, NCT, and EAF were responsible for the study design; DK, RSML, TYY, and EAF for data gathering; RSML, YT, and EAF for data analysis; DK, RSML, YT, and EAF for interpretation of the results and drafting the manuscript. All authors read, edited, and approved the submitted version. As the corresponding author, EAF had full access to all study data and takes responsibility for data integrity and the accuracy of the data analysis.
EAF formerly served on the management committee for KKT Technology Pte Ltd (Holmusk), a company that developed and owns digital lifestyle change programs including GlycoLeap. TYY is the medical director at KKT Technology Pte Ltd (Holmusk) and receives a salary and stock options. TYY has no direct ties to the recruitment site or patients. The other authors have no conflicts.