An Innovative Care Model Integrating Mobile Health Support to Promote Weight Management in Adults With Overweight or Obesity

Background: Obesity leads to cardiovascular diseases, diabetes, musculoskeletal disorders, and many types of cancers. Insufficient physical activity and unhealthy diet (including high intake of saturated fatty acids, fat and inadequate consumption of dietary fibre) are the major causes of obesity but they are preventable and reversible.

The Population Health Survey 2020-22 showed that the overall prevalence of overweight and obese (level 1 and 2) in Hong Kong is 22.0% and 32.6%, respectively. The age-standardized prevalence of physical inactivity has risen from 12.4% (crude prevalence: 13.0%) in 2014-15 to 24.8% in 2020-22. Still 1,486,000 HK residents aged 18 years or above had not met the recommended physical activities level by World Health Organization (WHO). About 4.6% of the population have raised blood glucose or diabetes in 2020-22, compared to 3.9% in 2014-15. The age-standardized prevalence of raised total cholesterol has risen from 38.4% in 2014-15 to 46.5% in 2020-22. About 39.1% of the population eat red meat at least once every day, but only 1.5% of the population intake 2 standard bowls of fruits and 3 bowls of vegetables.

The government targets to reduce a relative 10% of prevalence of physical inactivity by 2025 (compared to 2018) and halt the rise in obesity. The Hong Kong government has been promoting physical activity and healthy diet through public education and social marketing campaigns, but local routine primary healthcare service providers, including Hospital Authority (HA), District Health Centres (DHCs) and non-governmental organizations (NGOs), do not have standard protocols for assessing behavioral risk factors and delivering effective intervention for physical activity and healthy diet.

Mobile Health (mHealth) intervention is a novel and cost-effective approach to increase physical activity, which was highly supported by the WHO. A meta-analysis showed that mHealth intervention was associated with an increase of physical activity at short-(≤ 3 months, by 506 steps per day, 95%CI= -80 to 1092) and long-term (≥6 months, by 753 steps per day, 95%CI= -147 to 1652) follow-up. The meta-analysis also showed that intervention can reduce sedentary time at short-term (≤ 3 months, standardized mean differences (SMDs)=-0.49, 95%CI= -1.02 to 0.03) follow-up but the result was not statistically significant. However, most apps cannot provide personalized intervention and their intervention do not involve behavior change techniques. A qualitative study showed 45.4%of users of smartphone apps disagreed or strongly disagreed that text messages promote them to do more physical activity since those messages did not match their motivation.

To capture the facilitators, barriers, and behaviors in naturalistic settings, ecological momentary assessment (EMA) using mobile devices was developed in recent years. During EMA, users are repeatedly prompted to provide simple information about their experiences and/or behaviors at predetermined and/or random intervals throughout the day. The advantage of EMA includes lower recall bias and providing information at different times of the day. Based on EMA summaries, healthcare professionals could design more personalized interventions. EMA information also enables the provision of ecological momentary intervention (EMI), which the software collecting EMA automatically analyze EMA data and provide tailored messages, potentially improving user engagement and compliance to the intervention. EMA and EMI have been adopted in improving mental health, smoking cessation, and reducing substance use with promising results, but interventions for weight management were still under development

Sai Kung, Southern and Wong Tai Sin have set up DHCs or DHC express to deliver health promotion, health assessment, chronic disease screening and educational programs to the residents in these districts. According to the FAMILY Cohort 2011-14, obesity and overweight is the highest reported chronic diseases (Southern: 66.0%; WTS: 63.0%; Sai Kung: 69.0%) in these 3 districts for both male and female. Each DHC has a multidisciplinary healthcare team to offer consultation and health education. As the population ages, there is an increasing demand for intervention, but manpower is limited. Hence a systematic and efficient system in screening, risk stratification, and delivering interventions for the overweight is urgently needed. Digital and remote intervention is also valuable to increase DHC clients' uptake of lifestyle intervention as most clients without disease diagnosis normally have lower intention to visit DHC for health coaching. A primary care model with shorter intervention time but sufficient human interaction would need to be developed in the DHC setting.

The stepped-wedge cluster randomized controlled trial (SWT) design has become increasingly popular for evaluating interventions implemented at the cluster level. In this design, all clusters start under the control condition. Clusters are randomized into sequences that determine when each cluster transitions to the intervention condition, which continues until the study concludes. In the final phase, new participants in all clusters receive the intervention. Compared to individually randomized trials, SWT offers several advantages: (1) reduced within-cluster contamination; (2) lower risk of staff delivering the wrong treatment; (3) increased fairness and acceptance among participants; and (4) reduced resource demands, as clusters are not required to simultaneously prepare for intervention delivery. As a randomized trial, SWT is more robust than quasi-experimental designs. Compared to conventional parallel cluster trials, SWT provides additional benefits: (1) it evaluates interventions across all clusters; (2) facilitates cluster recruitment when intervention effects are promising; (3) accommodates odd or very small numbers of clusters, unlike cluster RCTs that require balanced intervention and control groups; (4) allows for both within-cluster and between-cluster comparisons; and (5) requires fewer clusters to achieve adequate statistical power.

Study design: This study will employ a stepped-wedge cluster randomized controlled trial design. The study period will consist of four consecutive four-month phases. Three non-governmental social service organizations (NGOs) will serve as clusters for randomization. In the first phase, all participants across the three clusters will be allocated to the control group. In the second phase, newly recruited participants in Cluster 1, determined randomly, will receive the intervention, while the other two clusters remain in the control group. In the third phase, newly recruited participants in Clusters 1 and 2 will receive the intervention, while Cluster 3 remains in the control condition. In the fourth phase, all newly recruited participants across the three clusters will be allocated to the intervention condition. A web-based portal system will be developed to assist NGO staff in delivering the weight management intervention. During the study, NGO staff will undergo capacity training on study procedures, risk assessment, and using the web-based portal system.

Procedure: Recruitment will be conducted both offline (e.g., posters) and online (e.g., Facebook and Instagram). Interested participants will complete an online form to determine eligibility. Subsequently, recruitment staff (e.g., nurses, fitness instructors, dietitians, or social workers) will contact eligible participants to invite them to an in-person enrollment session. During this session, participants will be briefed about the study, provide informed consent, and undergo baseline health assessments, including anthropometric measurements (e.g., height, weight, waist circumference), body composition analysis (e.g., body fat and muscle mass using the Tanita MC-580 Dual Frequency Segmental Body Composition Analyzer), and self-reported physical activity levels. An optional 5cc blood sample collection will be offered if prior lipid and glucose profiles are outdated by more than 28 days, with all samples destroyed after the study. Participants will also complete the Physical Activity Readiness Questionnaire (PAR-Q) and the American College of Sports Medicine (ACSM) Risk Stratification Screening Questionnaire to determine their appropriate exercise intensity.

The project coordinator will provide biweekly updates on the number of screenings and enrolled participants. Recruitment efficiency will be reviewed and discussed in biweekly and monthly team meetings to identify and implement improvements in recruitment strategies. The three NGOs will be randomly assigned to clusters 1, 2, or 3 using computerized randomization, which will take place one month before recruitment begins. This timeline will allow sufficient time to complete staff training. While healthcare workers and participants will not be blinded to their group allocation, data analysts will remain blinded to minimize potential bias during data analysis. Once randomized, participants will remain in their assigned treatment group throughout the study. Any deviations from the protocol will be documented, along with explanations, in future reports.

Intervention Group:

Participants in the intervention group will engage in an mHealth weight management program comprising multiple components: ecological momentary assessment (EMA) using a mobile app, wearable activity tracking, counseling sessions, and risk-based coaching.

Participants will complete four EMA sessions, each lasting 7 days. During these sessions, they will report their diet for 3 days and physical activity for 4 days, completing five brief daily surveys (under 2 minutes each) via a mobile app developed for both iOS and Android platforms. To protect privacy, participants will use a 5-digit case number instead of personal information. The app will retrieve data from Fitbit devices (e.g., step count, moderate-to-vigorous physical activity, sedentary behavior, heart rate) and prompt participants to complete surveys spaced approximately 4 hours apart. If participants miss a prompt, they will have a 30-minute window to respond, with reminders sent 15 and 5 minutes before the deadline. On 3 selected days, participants will document their food intake using text, audio recognition, and photos.

Participants will utilize Fitbit devices to monitor various physical activity metrics, including step count, moderate-to-vigorous physical activity (MVPA), sedentary behavior, and physiological data such as heart rate and heart rate variability. To complement Fitbit data, participants will also install tracking apps such as Google Fit (for Android) or Apple Health (for iOS) to ensure comprehensive monitoring.

Within one week of completing the first EMA session, participants will attend a counseling session delivered by nurses using the 5A model (Assess, Advise, Agree, Assist, Arrange). This session will provide tailored feedback based on EMA data and address barriers to behavior change.

Following this, risk-based coaching will be conducted by a dietitian and fitness instructor, who will provide individualized dietary and physical activity plans, including targeted meal recommendations and home-based exercise programs.

After the first EMA session and counseling, participants will proceed to the second EMA session, during which they will receive additional support via instant messaging. The third EMA session will occur one week before the 3-month follow-up, and the fourth EMA session will take place one week before the 6-month follow-up.

Control Group:

Participants in the control group will receive a brief telephone counseling session following their EMA session. This session will provide basic information on physical activity and diet, referencing resources such as the Centre for Health Protection's guides on diet and nutrition and physical activity. After the 6-month follow-up, control group participants will receive the same intervention as the intervention group to ensure equitable access to the program.