Aims of study
The primary objective of this study is to evaluate the effectiveness of a community-based diabetes prevention program in reducing the diabetes conversion rate at 3 years in overweight or obese men and women (BMI ≥23.0 kg/m2) with elevated risk of diabetes (defined as IFG and/or IGT). Secondary effectiveness objectives include the comparison of fasting plasma glucose (FPG), 2-hour plasma glucose, HbA1c, weight, waist circumference, BMI, physical activity level, health status, work impairment and diet.
Pre-DICTED is a randomized open-label, parallel arms, controlled trial that compares a diabetes prevention program consisting of lifestyle interventions with stepwise addition of metformin and added financial incentives (intervention) versus the current standard of care (control) in overweight or obese participants with isolated IFG (FPG 6.1–6.9 mmol/L), isolated IGT (2-hour plasma glucose in 75g oral glucose tolerance test (OGTT) 7.8–11.0 mmol/L) or both IFG and IGT. The definitions of IFG and IGT are based on the criteria of the Ministry of Health in Singapore . The study aims to recruit 846 participants across the two arms (423 participants per arm). The outline of the trial design is shown in Figure 1. This clinical trial follows the guidelines for randomized clinical trials (SPIRIT checklist (Additional file 1)).
The study participants are referred to the Pre-DICTED program from the following sources:
- Polyclinics/general practitioner clinics/hospitals
- Health Promotion Board (HPB) 
- Regional Health System Community Screening programs 
- Self-referral. A call for participants is made on HealthHub, an online portal and mobile application for Singaporeans to access health-related content and their public health records .
- Singapore citizens or permanent residents
- Age between 18 and 64 years
- BMI ≥23.0 kg/m2
- Laboratory tests completed within 3 months of enrolment:
- FPG 6.1–9 mmol/L (IFG) and/or
- 2-hour plasma glucose in 75g OGTT 7.8–0 mmol/L (IGT)
- Willing to sign informed consent
- Individuals with diabetes mellitus
- Health conditions impeding participation in lifestyle change program (e.g. active cancer, a recent myocardial event within 6 months, heart failure, chronic kidney disease)
- Current pregnancy or breastfeeding
- Intervention with medications known to alter glucose tolerance
- Known allergic reaction to metformin
- Active, or history of lactic acidosis
- Severe, acute or chronic hepatic disease
Randomization and blinding
Participants are randomized according to a 1:1 allocation ratio to the intervention and control arms using an internet-based computer program according to a randomization list. The randomization list was generated using the permuted blocks randomization technique. The block size was determined by a statistician generating the randomization list and will not be disclosed to clinical investigators and other study team members who have contact with study participants. For safety and practical reasons, participants and investigators involved in the stepwise addition of metformin are not blinded after study arm assignment. However, assessment of the primary and secondary outcomes is blinded.
All participants are given 2 study goals: weight loss of ≥5% of baseline body weight and moderate-intensity physical activity of ≥150 minutes per week. Both goals have been reported to be safe, feasible and effective based on previous clinical trials [1, 2].
Delivery of interventions
Participants in the intervention arm receive structured group-based lifestyle interventions with addition of metformin when needed. They are also eligible for a payment/rewards program.
The lifestyle intervention program is based on the Health Promotion Board’s (HPB) pre-diabetes intervention program , a public resource portal that includes dietary and physical activity recommendations. The lifestyle intervention program consists of (i) an intensive core intervention phase, delivered through in-person group sessions and worksheet-guided learning, and (ii) a maintenance phase, delivered through mobile phone text messages and telephone calls. Telephone delivery of lifestyle modification intervention has shown the potential to bring about weight loss in obese adults  and reduce diabetes incidence in men with impaired glucose tolerance , and has the advantage of being less resource intensive.
Core intervention phase (Month 1 to 3)
- Weeks 1 to 6: Participants attend 12 group sessions during the first 6 weeks (twice weekly) of the program. This consists of 2 nutrition workshops (2 hours per session), 9 exercise sessions (1 hour per session), and a goal setting workshop (a 2-hour session) (Table 1). Group sessions are delivered at community sites in Singapore by a fixed team of facilitators that include dietitians/nutritionists and fitness instructors. The dietitians/nutritionists follow the curriculum provided by HPB and the lessons are repeated for each cohort of participants. For Chinese participants who could not understand English, Mandarin sessions are conducted simultaneously. In the group exercise sessions, fitness instructions gradually increase the exercise intensity over the 9 sessions.
- Weeks 7 to 12: Participants receive worksheets containing information and recommendations on physical activity and nutrition. The worksheets are distributed weekly for 6 weeks of self-directed learning at home. The participants are also given physical activity trackers and digital weighing scales to facilitate self-monitoring of physical activity and weight.
Maintenance phase (Months 4 to 36). During this phase, the participants receive monthly mobile phone text messages on health tips as well as 6-monthly telephone calls from study coordinators to discuss the progress of their lifestyle modifications.
Stepwise addition of metformin
From month 6 onwards, intervention arm participants are prescribed metformin if they are considered to be at the highest risk of conversion to diabetes, defined as having IFG + HbA1c ≥6.0% or IFG + IGT. Metformin is prescribed at a starting dose of 250 mg twice daily followed by up-titration to 500 mg twice daily after 3 months if the participants do not experience any side effects.
In the Pre-DICTED program, participants randomized to the intervention arm are given financial incentives for program adherence and weight loss achievement. Intervention arm participants are given a financial incentive of S$10 (sports vouchers) for attending each group session during the core intervention phase (maximum amount for 12 sessions = S$120). They are also eligible for financial incentives (cash) if they meet the weight loss target (≥5% of baseline weight) at any of the study visits at 3, 6, 12, 18, 24, 30 and 36 months. The weight loss target is logged-in following the baseline visit and conveyed to the participants. Each participant who meets the weight loss target is given the option of choosing either guaranteed payment or lucky draw payment (1 in 10 chance) according to the schedule in Table 2.
The current standard of care for pre-diabetes includes referral and follow up by primary care physicians, and incorporates the following features:
- Education on pre-diabetes
- Importance of lifestyle modifications to prevent progression to diabetes
- Rescreening of glucose tolerance every 6 months (Table 3)
Apart from study visits, participants in the control group will not have additional contact with study staff.
Retention of participants
Appointment reminders have improved attendance in chronic disease follow-up , and monetary incentives have demonstrated effectiveness in improving retention in trials [29, 30]. To minimize dropouts, all participants are given text messaging reminders and telephone call reminders by the study coordinators prior to every study visit. All participants are also asked to make an upfront payment of S$20 as a form of deposit contract. This amount is forfeited if the participant drops out of the project before the end of 3 years or project closeout. The entire amount is quadrupled (i.e. S$80) and returned to the participant if he/she completes the last study visit at the end of the third year. In addition, 6-monthly lucky draws of S$50 sports vouchers are conducted for participants who complete their follow-up visits.
The primary endpoint is the proportion of participants who develop diabetes within 3 years from the baseline visit. Diabetes is diagnosed on the basis of a single FPG of ≥7.0 mmol/L or a 2-hour plasma glucose of ≥11.1 mmol/L, in accordance with the 2017 criteria of the American Diabetes Association . An FPG test is conducted every 6 months, and an OGTT is conducted annually.
Secondary endpoints include the change from baseline to end-of-study (3 years) in the following outcome measures:
- 2-hour plasma glucose
- Body weight
- Waist circumference
- Dietary intake (assessed by the Food Frequency Questionnaire (FFQ) )
- MET-minutes per week (assessed by the International Physical Activity Questionnaire (IPAQ) )
- Health status scores (assessed by the EQ-5D-5L ), and
- Work productivity loss and activity impairment (assessed by the Work Productivity and Activity Impairment Questionnaire: General Health V2.0 (WPAI:GH) ).
Secondary endpoints also include the proportion of participants who have engaged in the targeted level of physical activity (≥150 minutes/week of walking, moderate activity, or vigorous activity as defined by the IPAQ ).
Assessments and visit schedule
Baseline measurements (questionnaire completion, physiological and anthropometric measurement, retinal photography) are taken for all participants. Participants who are referred based on FPG alone also undergo a 75-gram OGTT at the baseline visit. Follow-up visits are conducted at 3 months and every 6 months from the date of randomization (6, 12, 18, 24, 30, and 36 months). The acceptable tolerance in the 6-monthly visits is +/- 1.5 months. All assessments are performed by trained staff following standard operational procedures and unaware of participants’ study group allocation. Study visits and assessments are summarized in Table 3. The assessments are described below.
Measurement of body weight, height, waist circumference and blood pressure is performed at all visits. Height and body weight are measured in light indoor clothing and without shoes, using a portable stadiometer (Model Seca 213, Gmbh & Co.KG, Hamburg, Germany) and a digital weight scale (Model Seca 803, Gmbh & Co.KG, Hamburg, Germany) respectively. Waist circumference is measured at the midpoint between the lower margin of the last palpable rib and the top of the iliac crest, using a stretch‐resistant tape. Two waist circumference measurements are made, and the mean value is used. If the two measurements differ by more than 1 cm, both measurements are repeated. Blood pressure and heart rate are measured with the participant in sitting position after at least 5 minutes of rest, using a digital blood pressure monitor (Model HEM-7322, Omron Healthcare Co. Ltd, Kyoto, Japan). The first heart rate measurement is used, while the measurement is repeated for blood pressure. If the two measurements differ by more than 5 mmHg for diastolic blood pressure, or 10 mmHg for systolic blood pressure, a third measurement is made. The mean value of the closest two blood pressure measurements is used.
For HbA1c, FPG, 2-hour plasma glucose in OGTT and lipids profile, participants are instructed to fast overnight for at least 10 hours. Blood samples are collected from participants using standard phlebotomy procedures. Plasma glucose is assessed by photometry (ADVIA Chemistry, Siemens, Germany), and HbA1c is assessed by high-performance liquid chromatography (D100 System, Bio-Rad, USA). Serum total cholesterol, HDL-cholesterol and triglycerides are measured by means of enzymatic techniques (ADVIA Chemistry, Siemens, Germany).
Retinal photographs are captured using a handheld non-mydriatic fundus camera (Visuscout 100, Zeiss, Germany) by trained technicians. Two fundus images are captured from each eye. The images are subsequently graded for the presence or absence of diabetic retinopathy by trained non-physician graders at a central ocular grading centre .
Dietary intake is assessed using a locally-validated 163-item semi-quantitative FFQ . The questionnaire consists of a list of 163 food/beverage items, and participants are asked how often they consumed one serving of each item. Data on the responses will be combined with local data on the nutrient content of each FFQ item to calculate the daily intake of major food groups (e.g. vegetables), energy, macronutrients as a percentage of energy intake, and fibre and micronutrient intakes as amount per 1000 kcal.
Physical activity is assessed using the short form of the IPAQ , a 7-item questionnaire with a recall period of 7 days. It records physical activity at four intensity levels: (i) vigorous activity such as aerobics, (ii) moderate activity such as leisure cycling, (iii) walking, and (iv) sitting. Responses will be used to calculate the total time spent per week on walking, moderate or vigorous activities, and the total MET-minutes per week .
Health status is assessed using the EQ-5D-5L , a questionnaire consisting of 5 questions with Likert response options (descriptive system) and the EQ visual analog scale (VAS). The EQ-VAS asks participants to rate their own health from 0 to 100 (the worst and best imaginable health, respectively). The descriptive system asks participants to assess the degree of difficulty experienced in the following dimensions: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Responses in these 5 dimensions will be used in conjunction with the EQ-5D-3L value set for Singapore  to derive the EQ-5D index value. The index value reflects how good a health state is according to the preferences of the general population. A mapping algorithm will be used to link the 5L dimension responses to the 3L value set .
Work productivity and activity impairment
Work productivity and activity impairment is assessed using the WPAI:GH, a 6-item questionnaire with a recall period of 7 days . The work productivity loss score combines (i) the work time missed due to health problems, and (ii) the productivity loss during work time due to health problems (self-reported using a 0 to 10 numerical rating scale). The activity impairment score reflects the degree to which health problems affect regular activities (0 to 10 numerical rating scale).
Baseline sociodemographic information is collected to assess baseline imbalance, adjust for potential confounding in the statistical analyses, and stratify participants for subgroup analyses. The information collected include age, ethnicity, education, family history of diabetes and health history.
Safety and data monitoring
Internal reviews of adverse events related to metformin are conducted quarterly. In addition, an independent Data Safety and Monitoring Board (DSMB) has been established to review the safety and effectiveness of the study. This Board consists of two clinicians and one statistician, all independent from the trial. The primary objective of the DSMB is to monitor the safety of the interventions and the validity and integrity of the data. The DSMB also evaluates the pace of recruitment and will make recommendations regarding the continuation, modification, or termination of any or all arms of the study.
A planned interim analysis was performed at 2.5 years from the first participant recruitment. Adverse events related to metformin were reported using pre-defined categories: diarrhea, nausea/vomiting, bloatedness, indigestion, weakness, abdominal discomfort, headache, and others. Intervention effectiveness as measured by the primary endpoint (proportion of participants with diabetes) was evaluated using a group sequential method with the O’Brien-Fleming alpha-spending function (overall type I error rate pre-specified at 0.05). An additional interim analysis for safety and effectiveness will be performed in February 2021.
Sample size justification
A study in Chinese pre-diabetes population showed an 11% annual incidence of diabetes for participants who did not receive any intervention , which is equivalent to a 30% incidence of diabetes over 3 years. The study also showed that the RR for lifestyle interventions was 0.64. In addition, the Indian Diabetes Prevention Program study showed that the relative risk (RR) for lifestyle interventions and metformin was 0.72 and 0.74 respectively . Risk ratios for Malays are assumed to fall between those of Indian and Chinese ethnicity. We assumed that the relative risk of converting to diabetes for those receiving intervention in multi-ethnic Singapore (composed primarily of Chinese, Malay, and Indian ethnicities) would be between the estimated effect sizes in the two cited studies, RR of 0.70 for the intervention.
Given the above assumptions, we aim to recruit up to 846 participants across the two arms (423 participants per intervention arm and control arm). This will allow us to achieve 80% power with 5% two-sided type I error rate and 20% drop-out rate to detect a difference in intervention effect of 10% between the 2 study arms, assuming 33% and 23% incidence of diabetes by 3 years in the control and intervention arm respectively (RR = 0.70).
Data forms are coded with a study number (starting with P001). Data is entered and stored on a standalone computer and the information is password protected. Only the principal investigator, co-investigators and study coordinators have access to the data for analysis. Participants are identified by their study number, and only the principal investigator has access to identifiers that link the data to the individual participant. De-identified data is collected and analyzed as specified in the protocol. Participant data will be kept for seven years after the research is completed and all data (electronic and hard copy) will be destroyed after the storage period.
The primary effectiveness analysis will be carried out on the modified intention-to-treat (ITT) population, consisting of all randomized participants who are present for at least one follow-up visit. The risk of prediabetes-to-diabetes conversion at 3 years will be modelled using a generalized linear model (GLM) with an identity link function and binomial variance function. Potential confounders such as age, sex, family history of diabetes and FPG at baseline will be included as covariates. The risk difference between the intervention and control groups and its 95% confidence interval will be reported. The risk of diabetes conversion at months 6, 12, 18, 24 and 30 will also be analyzed using the modified ITT set.
Sensitivity analyses will be performed on the ITT population. Multiple imputation datasets will be generated using chained equations , and the imputed datasets will be analyzed using the same GLM and covariates as the primary effectiveness analysis. In addition, two supportive analyses will be conducted: (i) the Kaplan-Meier estimator will be used to compare the cumulative incidence of diabetes in intervention and control participants, and (ii) the proportional hazards model for interval-censored data will be used to analyze the time to diabetes onset. The hazard ratio between the two study arms and its 95% confidence interval will be reported. The same analyses will be performed using the per-protocol population (all randomized participants who (i) have provided FPG and OGTT results at the study conclusion, and (ii) if they are in the intervention arm, have attended the goal-setting workshop, at least one nutrition workshop and at least one exercise session).
Secondary outcomes (engagement in ≥150 minutes of physical activity per week; change from baseline in the following outcomes: HbA1c, FPG, 2-hour plasma glucose, body weight, BMI, waist circumference, health status scores, MET-minutes per week and activity impairment score) will be analyzed using generalized linear models. Subgroup analyses will also be performed, with participants stratified by baseline age (<50 years, ≥50 years), sex, pre-diabetes type (isolated IFG, isolated IGT, IFG + IGT), HbA1c (<5.7%, 5.7%–6.4%, ≥6.5%), family history of diabetes, BMI (23.0 to 27.4 kg/m2, ≥27.5 kg/m2), physical activity level (<150 minutes/week, ≥150 minutes/week of physical activity), ethnicity (Chinese, Malay, Indian, Others) and average monthly income (<S$5000, ≥S$5000). Baseline characteristics, intervention adherence and adverse events will be summarized using descriptive statistics.
Trial management and dissemination policy
The study team includes medical practitioners with expertise in diabetes prevention and treatment, as well as academics with expertise in trial methodology, community-based interventions, health economics and public health nutrition. The study team discusses recruitment, withdrawal, study progress and adverse events, and will advise on protocol amendments where necessary. Day-to-day running of the trial is managed by the study coordinators. The initial informed consent is taken by the principal investigator, co-investigators or study coordinators. Additional informed consent to receive stepwise addition of metformin are taken by the principal investigator or co-investigators who are qualified medical practitioners. The study team will communicate trial results to participants and health professionals via mainstream media and journal publications at the study conclusion.