Animal source food social and behavior change communication intervention among girinka livestock transfer beneficiaries in rwanda: A cluster randomized evaluation

Animal source foods (ASFs), including cow’s milk, contain essential nutrients and contribute to a healthy diet, but frequency of intake is low among children in low- and middle-income countries. We hypothesized that an ASF social and behavior change communication (SBCC) intervention implemented by community health workers (CHWs) would increase child milk consumption and dietary diversity in households that received a cow from the Government of Rwanda’s Girinka livestock transfer program. We tested the 9-month SBCC intervention among children aged 12-29 months at baseline in administrative cells randomly assigned to the intervention or control. Most mothers in the intervention group were exposed to CHWs’ home visits (90.7%) or community-level activities (82.8%). At endline, more mothers in the intervention group compared with the control group knew that cow’s milk was an ASF (90.1% vs. 81.7%, P=.03) and could be introduced to children at 12 months (41.7% vs. 18.7%, P<.001). More mothers in the intervention group compared with the control group knew they should feed their children ASFs (76.2% vs. 62.1%, P=.01) and give them 1 cup of cow's milk per day (20.6% vs. 7.8%, P<.001). Children's consumption of fresh cow's milk 2 or more times per week increased in the intervention group, although not significantly (8.0 percentage points, P=.17); minimum dietary diversity was unchanged. Children in the intervention group had increased odds of consuming cow's milk 2 or more times per week if their mothers recalled hearing that children should drink 1 cup of cow's milk per day during a CHW's home visit [odds ratio (OR) 2.1, 95% confidence interval (CI) (1.1, 3.9)] or a community activity [OR 2.0, 95% CI (1.2, 3.5)]. Approximately half of the children had no milk during the past week because their households produced too little or sold what was produced. In poor households receiving a livestock transfer, strategies to further tailor SBCC and increase cow's milk production may be needed to achieve larger increases in children's frequency of milk consumption. This cluster-randomized controlled trial was designed to test the impact of an SBCC intervention to promote the consumption of ASFs, especially cow’s milk, on maternal ASF knowledge and awareness and on child milk consumption and dietary diversity in households that had received a cow from the Girinka program. The trial was registered at ClinicalTrials.gov ({"type":"clinical-trial","attrs":{"text":"NCT03455647","term_id":"NCT03455647"}}NCT03455647). This study tested the impact of an SBCC intervention promoting consumption of ASFs on maternal ASF knowledge and on child milk consumption and dietary diversity. The study was conducted in Nyabihu and Ruhango Districts, Rwanda. The districts were selected in consultation with the Ministry of Local Government to include districts with a high prevalence of childhood stunting and poverty.31,32 Districts in Rwanda are subdivided administratively into sectors, which are further divided into cells. Cells typically contain 5–7 villages, but they can range from 4 to 12 villages. We randomly assigned administrative cells in the 2 districts to intervention or control. Nyabihu had nutrition programs in different parts of the district, whereas Ruhango had nutrition programs operating throughout the district. Therefore, randomization in Nyabihu was stratified by ongoing nutrition programs. The existing nutrition programs in the counties did not specifically promote ASF or cow’s milk consumption by young children. In both districts, the randomized cells were balanced on total population size. We obtained lists of households that had received a cow through the Girinka program from district and sector animal resources officers. Households were eligible for enrollment at baseline if they received a Girinka cow in 2017 or earlier or a Girinka calf in 2016 or earlier, the animal was still alive, the mother was 18–49 years of age and had a child who was 12–29 months of age, and the biological mother lived with the child. Our target was 4 households per cell. However, because we had challenges finding enough eligible households and many cells had fewer than 4 eligible households, we included up to 9 households per cell. If a cell contained more than 9 eligible households, the data collection team randomly selected from among those that were eligible. We calculated sample sizes for 2 child outcomes—minimum dietary diversity (consumption of ≥4 food groups in the past 24 hours) and milk consumption in the past 24 hours—based on a comparison of the changes in these parameters between baseline and endline. Minimum dietary diversity required a larger sample size (Supplement Table 1), so it was used as the sample size for the study. To detect a 15-percentage point difference between groups in the prevalence of minimum dietary diversity33 (i.e., at endline: control 29% and intervention 44%) with 80% power and alpha=0.05, required 208 households per group, assuming an average cluster size of 4 households per cell, an intracluster correlation of 0.10, and a design effect of 1.3. We added 10% to the sample to account for attrition, resulting in 229 households per group and a total baseline sample size of 458. The SBCC intervention was known as Gabura Amata Mubyeyi in Kinyarwanda, which translates to “Parents, Give Milk” in English. The intervention was developed based on formative research and guided by the theory of change shown in Figure 1. The theory of change posits that appropriate and effective SBCC on ASF consumption from CHWs reaches mothers and increases their knowledge. Mothers are concerned about child nutrition and are willing and able to adopt the recommended practices. They increase the child’s consumption of home-produced milk from their Girinka cow, which in turn increases child dietary diversity and may contribute over the long term to increases in child growth directly or through improved dietary diversity. In this analysis, we measured the effects of the intervention on the intermediate outcomes in the own-production pathway indicated in bold boxes in Figure 1. The theory of change also shows an alternate pathway to increased dietary diversity and growth through the purchase of ASFs. Theory of Change for the Gabura Amata Mubyeyi Social and Behavior Change Communication Intervention to Promote Consumption of Cow’s Milk Among Children, Rwanda Abbreviations: ASF, animal source food; HAZ, height-for-age z-score; SBCC, social and behavior change communication. Three Stones International coordinator training community health workers. © 2019 Jean Claude Gasangwa/Three Stones International The SBCC intervention was known as Gabura Amata Mubyeyi in Kinyarwanda, which translates to “Parents, Give Milk” in English. The intervention and SBCC materials were designed in collaboration with the National Child Development Agency, which coordinates nutrition activities in Rwanda. The SBCC materials consisted of counseling cards, a poster, and a brochure translated into Kinyarwanda. The counseling cards were designed using the same style as the Rwanda maternal, infant, and young child nutrition counseling cards. The messages from the SBCC materials related to this analysis are shown in the Box. Rwanda does not have food-based dietary guidelines, so the recommendation in this study to give children 1 cup of milk per day was based on the Rwanda Agriculture Board’s One Cup of Milk per Child program.34 This quantity of milk is low compared with the U.S. Department of Agriculture dairy recommendations for children 12–23 months (1 2/3 to 2 cups) and 2–3 years (2 to 2 1/2 cups).35 The recommendation to introduce cow’s milk to the child’s diet at 12 months is based on evidence that cow’s milk can result in occult blood loss from infants’ gastrointestinal tracts36 and the inability of infants’ kidneys to handle the high levels of protein, sodium, and potassium in cow’s milk.37 Importance and benefits of animal source foods (ASFs) and milk consumption for children aged 1–3.5 years: Appropriate quantities of ASFs and cow’s milk to be consumed by children aged 1–3.5 years: Appropriate time to introduce cow’s milk and ASFs to young children: Community and environmental health officers, who supervise CHWs, were trained to train CHWs to use the SBCC materials and conduct household and community SBCC sessions. The household visits were specifically targeted at households included in the intervention arm of the study. The community sessions were offered to all community members in the intervention cells. The intervention was implemented from February to October 2019 and was designed as an addition to CHWs’ usual activities. CHWs were asked to visit households in the SBCC intervention group monthly and conduct community SBCC sessions monthly. At the time of this study, SBCC materials specifically promoting ASF consumption were not available to CHWs through the government or its implementing partners. In the 2 study districts, only CHWs in the intervention group had copies of the Gabura Amata Mubyeyi SBCC materials. CHWs work within their own administrative cells, so the possibility of the intervention being inadvertently implemented outside the target cells was very low. Experienced enumerators were trained to conduct the baseline and endline surveys. The training covered screening and enrollment, consent procedures, review of the questionnaire on paper and in Open Data Kit (ODK), and a pilot. Enumerators collected the data at the participants’ households using tablets with the questionnaire programmed in ODK. Completed interviews were reviewed by the field supervisor and uploaded to a secure server. The baseline survey was conducted in batches in April–May, July–August, and October–November 2018 as the lists of Girinka participants were received. The bulk of the endline survey was conducted from January–March 2020; 6 participants had their interviews in July 2020 because of travel restrictions related to COVID-19. The questionnaires were developed in English then translated into Kinyarwanda. They included questions on child diet and feeding practices from the World Health Organization (WHO) infant and young child feeding questionnaire,38 including the types of fluids and foods the child consumed in the past 24 hours (24-hour recall). The questionnaire also collected information on the frequency of the child’s consumption of cow’s milk and other ASFs in the past 7 days (7-day recall), maternal knowledge and awareness related to milk, participation in nutrition activities conducted by CHWs, household food insecurity, livestock ownership, household milk production, and socioeconomic characteristics. Maternal ASF knowledge questions were asked without providing response options, whereas maternal awareness was gauged by asking women if they had ever heard about specific practices. Questions on general exposure to home visits and community activities conducted by CHWs were asked to participants in both study groups at baseline and endline. The endline questionnaire also included questions on intervention exposure for participants in the intervention group only. Intervention exposure questions were posed in a yes/no format, except for questions about the numbers of home visits or community activities attended. Several variables in this study were calculated or derived from the data. Child dietary diversity was calculated using the WHO infant and young child feeding indicator guidelines.38 We did not use the updated dietary diversity indicator that includes breast milk because part of our study population was ≥24 months at baseline and most children were ≥24 months at endline and no longer breastfeeding. Household food insecurity access categories were calculated using guidelines from the FANTA project.39 The household domestic asset index was calculated for all movable assets including livestock, using guidelines by Njuki et al.40 Each of the assets was assigned a weight, which was then adjusted for the age of the asset. Higher asset scores indicate higher socioeconomic status. The household land asset was calculated as total agricultural land parcels owned by the household in square meters. A CASHPOR housing index that captures the quality of housing in terms of roof, wall, and floor materials was used as a proxy for measuring poverty.40 CASHPOR scores below 5 indicate very poor housing and scores from 5 to 9 indicate poor housing. We used longitudinal random effects regression models with robust standard errors in Stata (MP, version 16.0) to account for clustering at the level of the cell and estimate difference-in-difference for the impact of the SBCC intervention on child milk consumption (24-hour recall and 7-day recall) and minimum dietary diversity. We calculated unadjusted difference-in-difference estimates and performed an analysis adjusted for factors that could influence the outcomes (child’s age, child’s sex, current breastfeeding status, mother’s educational status, and mother’s marital status). We calculated the average means or percentages across districts by study group for outcome, socioeconomic, and other variables and used regression models to evaluate the difference in means.