Social and behavioral change communication, maternal wellbeing and child development: Evidence from a digital intervention in Bangladesh

The Government of Bangladesh implements the Mother and Child Benefit Program (MCBP) to improve maternal health, nutrition, and child development in poor households through cash transfers and social and behavioral change communication (SBCC). SBCC messages providing guidance on prenatal and postnatal care, maternal mental health, nutrition, and nurturing caregiving practices are being planned for in-person delivery as courtyard sessions using government frontline workers in select upazilas. The government plans to scale and expand these courtyard sessions nationwide, which will require substantial investment in institutional capacity to hire, train, mobilize, and monitor frontline workers to deliver such sessions. The courtyard sessions are not tailored to women’s stages of pregnancy or childbirth. This study proposes to test the impact of using low-cost and low-tech digital solutions to deliver SBCC messages using the following approaches:

1. T1 (Digital delivery of SBCC): In T1, eligible participants will receive automated phone calls with pre-recorded SBCC audio content of approximately 1–2 minutes, accessible even on basic feature phones, every week for a period of 50 weeks. The SBCC messages are tailored based on the stage of pregnancy or childbirth at the time of enrollment. Each participant receives a pictorial guidebook to complement the audio content to provide an accessible reference for key messages.

To encourage engagement, an optional quiz segment is embedded within the automated audio call. A small subset of participants will be randomly selected every two weeks to receive nominal cash prizes via a lottery for correct quiz answers. Participants will also be eligible to enter a lottery for nominal cash prizes mid-program if they complete 25 audio lessons, and at the end of the intervention if they complete all 50 audio lessons. Additionally, weekly reminders via SMS or phone calls will be sent to encourage continued participation in the intervention.

Participants will have access to a free technical support helpline every day of the week to ask clarifying questions about the intervention services, obtain help with technical challenges, update their preferred time for receiving the automated calls, report the birth of their child, change the registered phone number linked to receiving the intervention services, or opt out of the program.

2. T2 (Digital delivery of SBCC and teleconsultation services): In addition to the intervention components described above, participants in T2 will have access to trained paraprofessionals via a teleconsultation service to privately discuss their child’s health, early childhood development, and other maternal and child-related concerns. These paraprofessionals will provide personalized guidance, respond to participant questions, and make referrals to local health services as appropriate and in line with local guidelines for telehealth services. Access to this service is free of charge and available seven days a week at designated times. This is intended to replicate the personalized consultations that would be available to participants during the planned in-person SBCC sessions, where they would have direct access to government frontline workers.

3. T3 (Control group): Participants in the control group will not receive either the digitally delivered SBCC content or teleconsultation services during the intervention period but will continue to have access to standard government services available in their area.

2026-02-15

2027-01-31

Based on the program’s theory of change, we organize our outcomes into two distinct groups: the impact on children’s and maternal outcomes, and the impact on mothers’ parenting abilities and the quality of the home environment.

Our first group of hypotheses (H1) relates to the direct impact on maternal health, child health, and developmental outcomes:
• Child cognitive and non-cognitive developmental outcomes
• Child health outcomes
• Maternal physical health outcomes
• Maternal mental wellbeing

Our second group of hypotheses (H2) relates to intermediate outcomes that can contribute to the impact on the outcomes described in the hypotheses group I (H1):
• Maternal knowledge of maternal and child health and nurturing caregiving practices
• Parental self-efficacy
• Parental stress
• Quality of the home environment as measured by family care indicators (FCI)
• Quality of parent-child interaction
• Use of disciplinary approaches
• Investments in health-seeking behavior (whether children receive the set of recommended preventive and curative health investments relevant to the local context and the quality of nutritional intake; whether mothers obtain the set of recommended prenatal and postnatal visits and the quality of nutritional intake)

• Paternal involvement in caregiving
• Broader measures of maternal wellbeing, including intimate partner violence and empowerment

We implement a clustered-randomized controlled trial in randomly selected villages and wards in the Mymensingh division, where child development and nutritional indicators are among the lowest nationally, to evaluate the digital delivery of SBCC on maternal and child outcomes. Participants in randomly selected villages and wards were screened for eligibility based on the following criteria: (i) Pregnant women aged 18 years or older, (ii) Permanent resident of the selected village/ward, (iii) Has access to a mobile phone within the household, (iv) At least three months or twelve weeks pregnant, and (v) Has two or fewer living children at the time of the survey. Recruitment to the intervention occurred approximately two months after baseline data collection. At the time of recruitment, participants were either in different stages of pregnancy or had already given birth. In addition to in-person recruitment, we conducted phone-based recruitment for the intervention for mothers who were away at their natal home for childbirth.

We use a two-stage stratified cluster sampling design with probability proportional to size (PPS) sampling to randomly select villages and wards with at least 250 households to ensure an adequate participant pool of pregnant women. We stratify the sample by district (four in Mymensingh division) and area type (urban vs. rural) into eight distinct strata, with each stratum receiving a proportional allocation of villages/wards based on population size. We adjust each stratum’s main allocation to a multiple of three to ensure balance in the number of villages/wards across the two treatments and one control group, and we systematically add 9 replacement villages/wards to each stratum following the same process to reach our intended sample size if we are unable to recruit 10 eligible participants per village/ward from the main allocation. We employ the PPS selection to simultaneously select both main and replacement villages/wards in a single, unified process, yielding a sampling frame of 522 villages/wards. The replacement villages/wards come from the same stratum as the main village/ward and have identical treatment assignments to preserve experimental balance. Within the treatment-matched backup pool for each stratum, villages/wards are utilized in predetermined rank order (1st priority, 2nd priority, 3rd priority) based on their random assignment during the original selection process.

We initially targeted a sample of 450 clusters (villages/wards), with a pre-specified sequential backup pool of 72 clusters. During baseline fieldwork, 31 backup clusters were activated from the sampling frame. Two clusters yielded zero eligible participants, resulting in a realized baseline sample of 479 clusters. In the realized baseline sample, we have 161 villages/wards in T1, 159 villages/wards in T2, and 159 villages/wards in the control group.

Not available

Randomization was done in Stata.

Village/ward

Yes

The planned number of clusters (villages/wards) was 450. During baseline fieldwork, 31 backup clusters were activated from the sampling frame. Two clusters yielded zero eligible participants, resulting in a realized sample of 479 clusters.

The study targeted 4,500 individual observations (10 per cluster). The realized baseline sample includes 4,676 individual mother-level observations.

The study planned for 150 clusters with 1,500 observations per arm (across 2 treatment arms and 1 control arm). The realized baseline sample includes: Treatment 1 with 161 clusters of 1,575 observations; Treatment 2 includes 159 clusters of 1,552 observations, and the control group includes 159 clusters with 1,549 observations.