A Study of Multiple Micronutrient Supplementation (MMS) to Increase Adherence With Digital Technology

This study is a nested (two-stage) cluster-randomized controlled trial conducted in Lombok Island, designed to evaluate strategies to improve adherence to Multiple Micronutrient Supplementation (MMS) among pregnant women, as well as to assess the feasibility of integrating calcium supplementation into the MMS program within routine health systems. The first stage of the study consists of a cluster randomized trial at the puskesmas (primary healthcare center) level. Puskesmas are randomized into four intervention arms to evaluate the effectiveness of different delivery approaches for MMS, specifically comparing combinations of digital and non-digital strategies, as well as packaging formats. The digital intervention includes mobile-based tools such as WhatsApp reminders, chatbot-assisted education, call center support, and a gamified alert system to encourage adherence. Artificial intelligence-based approaches, including adherence monitoring and prediction models, will also be explored. The non-digital group receives standard antenatal and postnatal care services, including routine health promotion and printed educational materials. The second stage involves a cluster randomized trial at the posyandu (community health post) level within selected puskesmas. This stage evaluates the addition of calcium supplementation to MMS compared with MMS alone. Additional pregnant women will be enrolled outside the main MMS cohort and distributed across study arms. This component aims to assess the feasibility, acceptability, and operational considerations of co-administering calcium alongside MMS within the existing health system. To support implementation across both study components, several operational and digital systems will be applied. Each supplement package will be labeled with a unique QR code to enable real-time tracking of supplement distribution and consumption through KoboToolbox. A Dynamic Worker Support (DWS) system will be implemented to assist Community Health Promoters in automated scheduling, task management, and field activity optimization. Sample Size and Statistical Considerations The sample size has been calculated based on the primary outcome of MMS adherence, accounting for clustering effects at both puskesmas and posyandu levels. Approximately 10,012 pregnant women will be enrolled in Stage 1 across 60 puskesmas clusters. In Stage 2, approximately 900 additional participants will be recruited to evaluate calcium co-supplementation. The sample size is designed to ensure adequate statistical power to detect meaningful differences between study groups. A comprehensive quality assurance framework will be implemented to ensure data integrity, accuracy, and consistency across all study sites. Electronic data capture systems (KoboToolbox) will incorporate built-in validation rules, including range checks, consistency checks, and logical constraints across related variables. In addition, data validation will be strengthened through a double data entry process, whereby selected records are entered independently by two data entry personnel and discrepancies are identified and resolved through predefined verification procedures. Routine centralized data monitoring will be conducted to assess data completeness, timeliness, and internal consistency. In addition, quality assurance (QA) activities will be implemented to assess field staff performance through direct verification. This includes periodic visits by QA staff to pregnant women to validate home visit activities conducted by field workers. QA may also be conducted through HealthComm by contacting participants to confirm services received and ensure consistency with reported data. These procedures aim to ensure adherence to study protocols, verify implementation fidelity, and maintain overall data quality. A standardized data management system will be implemented, including a comprehensive data dictionary that defines all study variables, formats, permissible values, coding structures, and data sources. Where applicable, standardized coding approaches will be applied to ensure consistency. All study procedures will be guided by Standard Operating Procedures (SOPs) and a detailed field manual to ensure standardized implementation across sites. The field manual provides step-by-step operational guidance for field staff, including participant recruitment, informed consent procedures, home visit protocols, supplement distribution, digital system utilization, and data collection processes. It also outlines procedures for data management, adverse event reporting, and communication workflows between field teams and supervisors. Field staff, including Community Health Promoters, will receive structured training and ongoing supervision based on the field manual to ensure consistent, accurate, and high-quality implementation throughout the study. Data quality checks during data collection will aim to minimize missingness. During analysis, appropriate statistical techniques, such as multiple imputation or sensitivity analyses, will be applied as specified in the statistical analysis plan. All analyses will follow the intention-to-treat principle. Primary analyses will assess adherence to MMS across intervention arms using mixed-effects regression models to account for cluster randomization and potential confounders. The factorial design allows estimation of the independent and interaction effects of digital intervention and packaging type. Secondary analyses will examine factors influencing adherence, maternal and neonatal outcomes, and the impact of calcium co-supplementation. Depending on the nature of the outcome variables, appropriate generalized linear (mixed-effects) models will be employed. Binary outcomes will be analyzed using mixed-effects logistic regression models, continuous outcomes using linear mixed-effects models, and count or proportion outcomes using Poisson or binomial-based models with appropriate link functions. All models will account for clustering at the study design level and adjust for relevant covariates. Subgroup and sensitivity analyses will be conducted to assess the robustness and generalizability of the findings.