Despite progress in reducing high levels of undernutrition in Bangladesh, gaps in progress persist. They are particularly acute between rural and urban areas, and between the lowest wealth quintile and highest. According to the 2016 Bangladesh DHS report, 38% of rural children under five were stunted compared to 31% of urban children. Forty-nine percent of children in the lowest wealth quintile were stunted compared to 19% in the highest.

To address these discrepancies and lower the overall level of stunting, research is being conducted to assist the government of Bangladesh (GoB) in determining the most effective ways to reduce levels of stunting. In particular, positive correlations between household production and consumption of nutritious food have been widely documented by development organizations in Bangladesh. However, information on how to optimize the delivery of household food production programs is needed.

The primary objective of this study is to compare the effectiveness of current standard practice with two multisectoral intervention packages focused on homestead food production:

1. Homestead food production (HFP) supported by community farmers, Social and Behavior Change Communication (SBCC), strengthened health services, and referrals to health and other services
2. HFP supported by retailers, SBCC, strengthened health services, and referrals to health and other services

The study's primary outcome is the percentage of children 6-23 months old receiving a minimum acceptable diet (MAD), as a proximate determinant for stunting. MAD is defined as the proportion of children 6-23 months old who receive both the minimum feeding frequency and minimum dietary diversity for their age group and breastfeeding status. It will be assessed based on the mother/caregiver report. Secondary outcomes include assessing the knowledge, attitudes, and practices around breastfeeding, complementary feeding, water sanitation and hygiene, health services and gender norms. Quantitative surveys, in depth interviews, focus group discussions, report reviews and process documentation will be used to assess intervention strengths, weakness, and cost effectiveness.

According to the 2014 Bangladesh Demographic and Health Survey (BDHS), 36% of children under the age of five were stunted, 14% were wasted and 33% were underweight. These results reflect positive trends in stunting and underweight since 2004, though the rate of decline in undernutrition slowed from 2011 to 2014. And, despite positive trends, there remain gaps in key indicators between rural and urban areas and between those in the highest and lowest wealth quintiles. For instance, according to the BDHS, 38% of rural children under five were stunted compared to 31% of urban children. The wealth discrepancies are even greater; 49% of children under five in the lowest wealth quintile were stunted compared to 19% in the highest quintile.

In 2017, the GoB approved the second National Plan of Action for Nutrition (NPAN 2) 2016-2025. The plan aims to improve nutrition and eliminate malnutrition, with a focus on children, adolescent girls, and pregnant and lactating women. Specific targets of NPAN 2 include reducing stunting to 25% among children under 5; reducing wasting to less than 8% and reducing underweight to less than 15%. A significant acceleration in the annual rate of reduction to 3.3% needs to occur in order to achieve the ambitious Targets by 2025. This acceleration requires high-level political commitment, a strong policy framework, effective coordinating mechanisms, adequate resourcing, strong involvement of local civil society groups, and high impact, cost-effective, multisectoral nutrition interventions.

In 2017, USAID awarded FHI 360 the Strengthening Multisectoral Nutrition Programming through Implementation Science Activity (hereafter referred to as "the Project") to test and refine multisectoral nutrition approaches in high stunting areas of Bangladesh. Under the Project, research is being conducted to assess the effect of different multisectoral nutrition intervention packages aimed at improving nutrition outcomes that are known to contribute to overall healthy nutritional status of children under two in Bangladesh. One of the intervention packages to be studied by the Project focuses on integrated agricultural and livelihood activities, known as homestead food production (HFP). This protocol describes a cluster randomized controlled trial (cRCT) to compare two different HFP interventions to the current standard of practice. The two interventions:

1. Homestead food production (HFP) supported by community farmers, Social and Behavior Change Communication (SBCC), strengthened health services, and referrals to health and other services
2. HFP supported by retailers, SBCC, strengthened health services, and referrals to health and other services

A cluster-randomized, controlled trial (cRCT) design will be used to evaluate effectiveness. The interventions will be delivered at the level of the union, which is a geo-political unit with an average population of 25,000 people. A total of 45 unions in Khulna and Barishal Divisions of Bangladesh will be randomly allocated to one of the three study arms: Control (Current Practice), HFP intervention through community farmers (plus SBCC and strengthened health services), or HFP intervention through retailers (plus SBCC and strengthened health services) . Outcome data will be collected through face-to-face interviews using structured questionnaires with independently selected random samples of mothers/caregivers of children ages 6 to 23 months at baseline (pre-intervention) and again at endline. At both timepoints, participants will be chosen from a sub-sample of the general population who meet the eligibility criteria. Baseline data will be conducted prior to initiation of study activities. Endline data collection will be conducted after two years.

A process evaluation will be completed between baseline and endline to understand how well the interventions were implemented, their costs, and ways they may be improved.

The final evaluation of the cRCT to be done at endline, and will focus on comparing the effect of the intervention on the study outcomes. The study's primary outcome is the percentage of children 6-23 months old receiving a minimum acceptable diet (MAD), as a proximate determinant for stunting. MAD is defined as the proportion of children 6-23 months old who receive both the minimum feeding frequency and minimum dietary diversity for their age group and breastfeeding status. It will be assessed based on the mother/caregiver report. Secondary outcomes include assessing the knowledge, attitudes, and practices around breastfeeding, complementary feeding, water sanitation and hygiene, health services and gender norms.

It is anticipated an analysis of covariance (ANCOVA) approach will be used for a post-only comparison of study arms with possible adjustment for baseline levels in an aggregate manner (note: aggregation for baseline adjustment will be needed given the independent samples selected at each time point). Generalized mixed models will be used to compare the study groups; the will be adjustment for clustering at the Union level. A logit link will be used for the primary outcome as it is a dichotomous outcome, while other link functions will be used for other outcomes as appropriate. The comparison of each intervention to the standard of practice is the primary evaluation focus, while the comparison between intervention groups secondary, as the interventions' effects are not expected to be very different from each other. Adjustment for multiple comparisons is not anticipated. The main analysis will use an intention-to-treat (ITT) approach, where study participants' outcomes are analyzed in the study arm the Union was randomized to, regardless of whether or how much they have been exposed to the intervention.

Inclusion Criteria:

• mother/caregiver of child 6-23 months of age
• Child 6-23 months is mothers 1st or second (living) child
• resides in an extreme poor or poor household, which is defined in Barishal as less than BDT 2056/month on household expenditures or in Khulna as less than BDT 2019/month on household expenditures

Note: study inclusion criteria is different from intervention enrollee criteria

De-identified data will be shared in accordance with USAID's open data policy.

Approximately January 2022

Data will be made available upon request, submitted with a brief concept describing intended use of data.