Women will be eligible for recruitment if they meet the following criteria: older than 18 years; diagnosed with GDM defined using fasting blood glucose >5.1 mmol/1 or 2-hour postprandial >7.8 mmol/1, which are the standard guidelines for diagnosis of GDM in Malaysia [32]; permanent resident in the state of Selangor or Putrajaya; registered in one of the study health clinics; owning a smartphone (iOS 11 or Android 7) with internet connection; and able to speak, read, and understand English or Malay.

Women will be excluded if they meet any of the following criteria: are having a twin pregnancy, have type 1 or 2 diabetes, have a severe physical disability that would prevent any increased uptake of physical exercise, have a severe mental health disorder (psychosis, bipolar, substance dependence, or active suicidal ideation), or are currently participating in a weight loss program or diabetes prevention intervention. For women who decline to participate, their age, parity, BMI, ethnic origin, education level, occupation, and household income will be recorded if permission is granted.

Date of birth, ethnicity, postcode of residence, education level, occupation, household income, gravida, parity, and number of previous GDM pregnancies.

Body weight will be measured in light clothing, without shoes, to 0.01 kg, and height to 0.1 cm using a stadiometer. Height will be measured without shoes and socks using a stadiometer (SECA 213) to the nearest 0.1 cm. The BMI will be calculated as weight (kg) / height² (m²). Body fat percentage will be assessed using bioimpedance analysis (Tanita RD-545 InnerScan Pro).

Systolic and diastolic blood pressure (mmHg) will be measured using a digital Omron HBP-1120 monitor using standardized procedures of the average of two readings taken 1 minute apart while seated. Blood venous samples will be collected via venipuncture. The lipid profile (total, low-density lipoprotein [LDL], and high-density lipoprotein [HDL] cholesterol) will be measured using an Advia 2400 (Siemens Diagnostics) analyzer, detection limit 0.1 mg/L and 0.01 mmol/L, respectively. Hemoglobin A_1c (HbA_1c) level (mmol/mol) and fasting blood glucose (mmol/l) will be measured by affinity chromatography using the Primus Ultra 2 analyzer (Primus Corporation) and an Advia 2120 analyzer (Siemens Diagnostics), respectively. For the OGTT, participants will be asked to fast for 8 hours, after which fasting blood glucose and fasting insulin will be measured, then the participant will be invited to drink 75 g of glucose in 200 ml of water, and 2 hours later the blood glucose will be measured. The Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) will be calculated by multiplying fasting plasma glucose (mmol/l) by fasting serum insulin (U/mL) divided by 22.5.

Self-efficacy will be measured using the Self-Efficacy for Exercise (SEE) Scale, which comprises 18 items [33]. Response options are presented on a Likert scale from 0 to 100, ranging in 10-unit intervals from 0 (cannot do) through intermediate degrees of assurance, 50 (moderately certain can do), to complete assurance, 100 (highly certain can do). The scale is scored by summing the numerical ratings for each response and dividing by the number of responses. Higher scores indicate higher self-efficacy for exercise. The mother-infant relationship will be assessed in the antenatal period using the Maternal Antenatal Attachment Scale, which comprises 19 items with response options presented on a 5-point Likert scale [34]. Higher scores are indicative of stronger attachment, but a specific cutoff is not provided [34]. Depressive symptoms will be measured using the Patient Health Questionnaire (PHQ-9), comprising 10 items scored on a 4-point Likert scale with a score range from 0 to 27 and higher scores representing worse depressive symptoms and a score 10 representing caseness for diagnosis of depressive disorder [35].

Dietary intake will be assessed using standardized multiple-pass 24-hour dietary recall for 3 days (2 weekdays and 1 weekend day) as it is considered an objective and reliable measure of change in intervention studies [36]. Clinical researchers will be trained to follow a standardized protocol, ask neutral probing questions to encourage recall of food items, and taught about different methods of food preparations and brands in different cultures. Physical activity will be measured using the Short-form International Physical Activity Questionnaire (SF-IPAQ), which comprises seven items to capture average daily time spent sitting, walking, and engaging in moderate and vigorous physical activity over the last 7 days [37]. Step count will be measured using inbuilt features of the patients’ own smartphone; the previous 7-day average will be used. Smoking status will be measured using a single item measure with response options including current smoker, previous smoker, and never smoked. Alcohol intake will be measured using the first question of the Alcohol Use Disorders Identification Test [38].

Women allocated to the control arm will receive standard care and no intervention. This includes self-monitoring blood glucose and lifestyle advice (diet, physical activity, optimal body weight) by a multidisciplinary team.

Women allocated to the intervention arm will receive access to a smartphone app called MyManis (the English meaning of MyManis is my sweet baby) and group support.

The development of the app was guided by the Information-Motivation-Behavioral Skills (IMB) model of behavior change [40]. The IMB model identifies three core determinants of the initiation and maintenance of health behaviors: accurate information that can be readily translated into health behavior performance (ie, knowledge), motivation both personal and social to act on such information, and behavioral skills to execute the health behavior effectively and confidently. The constructs of the model are supported in the literature to improve healthy lifestyle behaviors and have been tested and used successfully in diabetes management [41-43], obesity prevention [44], and improving dietary and physical activity behaviors [45,46].

Participants randomized to the intervention will receive a brochure outlining the app features and a step-by-step guide on how to download the app and set up an account. A link to download the app via SMS text message will also be sent to participants for convenience. Once the app has been downloaded, participants will be invited to set up an account by inputting their name, date of birth, current week of gestation, height, and weight. Once completed, this will give them full access to the app. The app content is presented under six main tabs:

The app is available to women in both Malay and English

Women will be invited to join group peer support. This will consist of 1-hour sessions with a group of 10 women meeting weekly via a virtual group chat function and communication of SMS text messages supporting each other during the week. Group support will be facilitated by a dietician trained in motivational interviewing (MI).

Women will be notified via SMS text message if an update of the app is required. During follow-up visits to their health clinic, women will be provided with in-person app support if required. Women who have not accessed the app for over a 2-week period will be flagged, and a notification will be sent via the app to motivate engagement.

The primary feasibility outcomes are to estimate the proportion of women who were randomized out of those identified from the clinical registers (study population); the proportion of women who take up the intervention out of those in the intervention arm, to be defined by measures of digital analytics (data will be collected on date of log in, time spent on each component of the app, and total time spent on the app); and the proportion of women who are withdrawn or lost to follow-up out of those randomized (ie, those who provide no primary or secondary outcome data).

The secondary feasibility outcomes are to estimate the proportion of women identified from clinical registers (study population) who give consent for screening for eligibility; the proportion of women who were randomized out of those eligible to participate; the proportion of those who completed the secondary outcomes at follow-up out of those randomized; and the number of women randomized per month overall and per month per study site.

A mixed methods six-component process evaluation framework will be used. Data collection methods for each component of the process evaluation are described below.

The following maternal and neonatal variables will be collected to describe the sample and the potential attrition and generalizability of a larger trial in this population. These variables will be collected at 3 months post partum (time 2) from postnatal records by a research team member. While we will not statistically assess them as such in this feasibility study, these variables may also be potential predictors or modifiers of postpartum outcomes at the later time points, which we may need to consider for a larger trial. These include maternal medical data (gestational hypertension, gestational hypercholesterolemia, pre-eclampsia); maternal pregnancy and birth data (place of antenatal care, date of admission, date of delivery, place of delivery, date of discharge, delivery method, pregnancy complications, birth complications, or breastfeeding status on discharge); and neonatal data (birth weight; appearance, pulse, grimace, activity, and respiration [APGAR] score 5 minutes; congenital anomaly; congenital pneumonia/heart defects; admission to neonatal intensive care; special care or qualified on ward; length of stay).

The aim is to recruit 30 women per arm, as this should be a large enough sample to estimate parameters such as the SD, which in turn can be used to estimate the sample size for an appropriately powered full-scale RCT [48]. This sample should also allow us to estimate the 95% CI for the primary feasibility outcome of the proportion randomized of those eligible, which we estimate will be approximately 10%, to a CI of within +/-8% [49]. This was calculated as:

Data will be double entered into an SPSS (IBM Corp) data set, and data analysis for the study will be performed using SPSS. Statistical analysis and reporting will be presented in line with the CONSORT (Consolidated Standards of Reporting Trials) guidelines [50]. Statistical analyses will primarily be descriptive, aiming to provide estimates related to feasibility parameters and to help inform power calculations for a future definitive trial.

For the primary outcomes of feasibility, we will calculate proportions or rates and appropriate CIs. We will consider whether a larger phase III effectiveness trial is feasible based on three potential factors: ≥25% of eligible women agree to be randomized, ≥50% of women in the intervention arm take up the intervention, or there is less than <20% attrition (women withdrawn from the study or lost to follow-up).

The secondary outcomes will be summarized using appropriate summary statistics overall and by randomization arm at each outcome time point. Normally distributed continuous variables will be summarized using the mean and SD, skewed continuous variables using the median and IQR, and categorical variables using frequencies and proportions. As the majority of the secondary outcomes are continuous and repeatedly measured, we will estimate the mean difference between the arms using linear mixed effects models with a random intercept at the participant level. The repeatedly measured outcomes will be the dependent variables, with the independent variables being trial arm, time, and a trial arm by time interaction term to extract differences at different time points. We will use analysis of covariance models where possible (ie, also include the baseline measure of the outcome as another independent variable). Appropriate 95% CIs will be calculated for the overall group and by trial arm, but as this is a feasibility study, we will not compare these statistically nor provide a P value for comparisons between arms. We will also not impute missing data, although fitting the mixed effects models using maximum likelihood will help account for missing data under a missing at random assumption. Some variables (eg, PHQ-9 and conversion to T2D [HbA_1c of 48 mmol/mol]) will also be summarized (but not analyzed) as categorical variables based on their clinical cutoffs. Smoking status and the obstetric outcomes that will only be measured once will also be summarized rather than modeled.

In terms of the process evaluation, all quantitative data including checklists, logs, and feedback surveys will be analyzed using SPSS. Data will be analyzed descriptively by calculating total numbers and percentages and mean and SD or, if the data are not normally distributed, the median and range. Qualitative data will be analyzed using thematic analysis to identify, analyze, and report patterns (themes) within the data [51].

Overall, the proportion of missing data for individual items and measures will be examined to determine the suitability of instruments and the level of burden for a future full-scale trial. As this is a feasibility study, no subgroup analyses are planned.