Pre-school undernutrition is a global problem with life long adverse consequences. One form of undernutrition, chronic undernutrition or stunting, affects 171 million children under the age of 5 worldwide. 35% of these children live in Africa. In Ethiopia, the focus of this study, in 2014, 44.5% of children under 5 were stunted. Stunting is the consequence of several factors including low birth weights, sub-optimal infant and complementary feeding practices and repeated illness. In Ethiopia, complementary feeding is sub-optimal; only 4% of children aged 6-24 months met the minimum dietary diversity recommended by WHO. The investigators hypothesize four main reasons why many children and mothers in Ethiopia fall short of best practice in terms of meeting nutritional needs and providing appropriate childcare. (i) Lack of information on healthy eating and appropriate child-feeding practices; (ii) Limited affordability; (iii) Limited accessibility to markets and diverse food items; and (iv) Limited peer effects in spreading information and adopting new practices. This study will assess the efficacy of the interventions that address these four barriers to optimal complementary feeding practices in Ethiopia. Using a cluster randomized control design, mother-father-child pairs in two localities, Holeta and Ejere will be enrolled. Treatment will be randomized at the garee (village) level. There will be five treatment arms and a control group: T1, weekly maternal nutrition BCC sessions for four months; T2, weekly maternal nutrition BCC sessions for four months and weekly paternal nutrition BCC sessions for three months; T3, receipt of a food voucher for six months; T4, weekly maternal nutrition BCC sessions for four months and receipt of a food voucher for six months; T5 weekly maternal nutrition BCC sessions for four months and weekly paternal nutrition BCC sessions for three months and receipt of a food voucher for six months; and C, a control group. Within household, recipient of voucher (mother or father) will be randomly selected.
1. Specific Aim of the study Aim 1: How does maternal behavior change communication (BCC) on healthy eating affect knowledge on child feeding, child feeding behavior, and nutritional status of the child? o Does BCC lead to changes in knowledge about healthy child-feeding and child-feeding behaviors? o Does the changes in child-feeding behaviors due to BCC lead to improvements in nutritional status among young children measured by height-for-age and weight-for-height? Aim 2: How do food vouchers affect food consumption and nutritional status? * Do vouchers lead to increased food consumption and dietary diversity? * Do vouchers crowd out private spending on food? * Does the increased food intake due to food vouchers lead to improvements in nutritional status among young children measured by height-for-age and weight-for-height? Aim 3: How does increased accessibility affect food consumption, child-feeding behavior and nutritional status? * By sending petty traders of various food items to randomly selected markets with limited access to diverse food groups, the investigators hypothesize that increased accessibility to diverse food items will lead to increased food consumption and dietary diversity. * Is there complementarity between accessibility and BCC or vouchers? Aim 4: Is there complementarity between BCC and food voucher? • The investigators will test whether there is a complementarity between BCC and food vouchers, as those receiving the BCC would make more informed decisions on food consumption. Aim 5: Are there peer effects in nutrition education and corresponding behavior change? o Within treatment groups: As low food diversity is in part due to eating norms in the community, the investigators will test to what extent peer effects within the treatment group will play a role in reinforcing nutritional knowledge and in changing household's food consumption decisions. o Outside the treatment groups (spillovers): The investigators hypothesize that mothers receiving BCC will affect eating behaviors of other household members and/or friends. Aim 6: How do variations in the prices of healthy food impact demand? o The investigators hypothesize that offering lower price options of healthy but expensive food such as meat would increase the demand for that food. o What are the price elasticity of demand for important food items? What is the maximum price households are willing to pay-i.e., the stoppage rule-for healthy food items? Aim 7: Is there additional impact of paternal and maternal nutrition BCC sessions on complementary feeding practice compared to maternal nutrition BCC sessions alone? o There will be greater improvement in complementary feeding practices when fathers are engaged in nutrition BCC sessions. Increased fathers' support on purchasing expensive but nutritious food items and household labor will facilitate the changes. Aim 8: Is there differential impact of voucher transfer on complementary feeding practices depending on recipient's gender? o The investigators will compare the effect of giving vouchers to fathers with giving vouchers to mothers on complementary feeding practices, food consumption of household, and health-related outcomes. 2. Detailed intervention component 2.1 Maternal behavior change communication BCC will be delivered in groups of six through weekly sessions for a duration of 4 months. Key gaps in knowledge and barriers identified through formative studies are also addressed in the BCC modules. 2.2 Paternal behavior change communication Materials will be delivered in groups of 6 through weekly sessions for the duration of 3 months. 2.3 Food vouchers Food vouchers are transferred monthly at the Project Office in Holeta. Eligible participants will receive 200 ETB, estimated around US$10. The length of the voucher scheme will be 6 months. Voucher recipient's gender will be randomly selected within group 3, 4 and 5 to assess whether there is any differential impact on complementary feeding practices depending on the gender of voucher recipients. 2.4 Access to healthy food For a random half of the treated and control groups, petty traders with diverse food items to their localities will be sent to increase accessibility. This is taking into consideration that some mothers, especially rural residents, have access a limited set of food items in the markets that they usually visit, lacking dairy products, meat, green leafy vegetables, and diverse fruits. To bridge this accessibility gap, petty traders will be sent to some selected markets with dairy products, meat, and various fruits and vegetables. 2.5 Price variation of healthy foods The investigators will introduce variation in the prices of healthy foods that are sold through petty traders to estimate the impact of price on demand. Price variations will take the form of 10%, 20%, or 30% discounts. While price will be varied by each petty trader every time, average food prices over the 6-month intervention period will be the same among all petty traders. 3. Quality assurance plan o Baseline and follow up data will be collected through mobile-survey to minimize errors. Commcare software will be used to build mobile survey. Data collected through mobile survey will be checked for errors and missing data on a daily basis. * For some variables, predefined rule will be applied within Commcare software using data validation tools as the data are collected to prevent avoidable errors. * Principle investigator and co-investigators will take turn and monitor the field site during data collection stage (both baseline and follow up) and during early intervention stage to ensure data quality. * Baseline survey data will be compared to demographic and health survey data to assess the accuracy, completeness and representativeness of the data. However, for most variables, there are no external data sources to compare. * Data dictionary not relevant * For variable with missing data above 5%, the variable won't be used as control variables. For main outcome of interest, data will be checked for errors and missing data on a daily basis. The investigators will revisit the study participant to minimize missing data. 4. Standard Operating Procedures to address registry operations and analysis activities 4.1 Patient recruitment The selection of the households that will be eligible to receive interventions will take place using the census data of Holeta and Ejere that was collected in 2016. To minimize potential contamination, control and treatment groups will be randomized at garee (village) level. Upon consent to participate in the study, mothers living in same garee will form a group to receive BCC education for mothers, and eligible spouse/partner of the participating mother will be recruited to form a group separately from mother to receive BCC for fathers. 4.2 Consent process Women will be recruited from home visits by AFF's enumerators assigned to this study. Women will be approached by an AFF's enumerator from the study to see if they are interested in participating. All women in their reproductive age (18-40) will be informed about the details of the survey and asked to participate in the survey with the consent form. Enumerators will read all out the consent form before asking for signature. If they accept, enumerators will conduct the baseline survey, which consists of demographic and socioeconomic information. 4.3 Survey data collection When eligible participants are identified, AFF enumerators will meet the participants at the health posts. 4.4 Sample size assessment Minimum detectable difference for the cluster-randomized controlled trial was calculated. Intra-cluster coefficient was calculated using 2011 DHS data for HAZ, WHZ and child dietary diversity score (DDS). Minimum detectable difference was estimated with statistical significance of 0.05 and power of 0.8. Six mother-child pair per cluster was decided on the study area's estimates of birth rates, 50 numbers of clusters available for each arm. 10% attrition was assumed. Details of power calculation for main outcomes of interest are shown below. Detectable difference • DDS: 0.30 (Mean 1.5, SD 1.05, ICC 0.10) * HAZ: 0.51 (Mean -1.56, SD 1.75, ICC 0.12) * WHZ: 0.42 (Mean -0.80, SD 1.44, ICC 0.11) 4.5 Statistical analysis • Estimation of treatment effects In this section, the investigators outline the basic estimation approach to measuring the effect of the treatments on various outcomes. Our basic treatment effects specification estimates the following equation: y= β₀ + β₁BCC + β₂Voucher+ β₃BCC\&Voucher+ β₄X+ ε where y is the outcome of interest. BCC, Voucher, and BCC\&Voucher are dummy variables equal to 1 if the participant was randomly assigned to the BCC, voucher group, or the BCC and voucher villages, respectively, and 0 otherwise. β₁, β₂ and β₃ represent the effect of being assigned to the specific treatment arm. X is a vector of individual's characteristics including demographic variables (e.g., age, marital status, birth order, and household size) and socioeconomic status (e.g., level of education, employment status/history, and household income and asset). ε is an error term. The outcome variable, y, include mother's nutritional knowledge score, household food and non-food expenditures, child's dietary diversity score and FCS, and child's height-for-age (HAZ) and weight-for-height Z scores (WHZ). To assess the gender-related impacts of engaging fathers in the BCC program, the investigators estimate the following equation: y= β₀ + β₁MotherBCC+ β₂BothBCC+ β₃MotherBCC\&Voucher+ β₄BothBCC\&Voucher+ β₅X+ ε where MotherBCC, BothBCC, MotherBCC\&Voucher and BothBCC\&Voucher are dummy variables equal to 1 if the household was randomly assigned to the mother BCC, mother and father BCC, mother BCC and voucher, or mother and father BCC and voucher villages, respectively, and 0 otherwise. β₁, β₂, β₃ and β₄ represent the treatment effects. To assess the differential impacts of randomly giving vouchers to mothers or fathers, a dummy variable representing the sex of the voucher recipient will be interacted with the voucher treatment assignment variable. Further, the investigators will implement the same analysis with a restricted sample by excluding BCC only villages. The investigators also have random variation in the accessibility to certain food items in the markets, as a random half of the participants are provided access to additional goods in the markets. To analyse the effect of increased accessibility on household expenditures and nutritional outcomes, the investigators estimate the following equation: y =α₀+α₁Access+α₂X+α₃P+ ε Where Access is a dummy variable equal to 1 if petty traders are randomly assigned to markets in village provide lacking food items such as meat, dairy products, and fruits and vegetables. P is a vector of food prices in village. Finally, the investigators use network data to estimate whether the treatment also influenced the outcomes of peers of the participants. The extent of such peer effects or information spillovers can be estimated with the following specification: y =α₀+α₁Peer+α₂X+ ε Where y is the outcome of interest for individual i with child less than four months of age or in pregnancy at the time of baseline survey. Peer is a dummy variable equal to 1 if respondent has a peer in cluster who is assigned to BCC or BCC and voucher treatment group.
Study Type
INTERVENTIONAL
Allocation
RANDOMIZED
Purpose
TREATMENT
Masking
NONE
Enrollment
1,243
Weekly behavior change communication (BCC) to mothers for the duration of 16 weeks Weekly behavior change communication (BCC) to fathers for the duration of 12 weeks
Monthly voucher worth 200 birr(\~$10) every month for the duration of 6 month
Africa Future Foundation Ethiopia
B/m/h/395/98, Goro Kerensa, Holeta Town, Oromiya, Ethiopia
Nutrition knowledge
Mean difference of nutrition knowledge score
Time frame: Measured at baseline and endline. Time frame between baseline and endline will be approximately 6 months.
Child dietary diversity score
Mean difference in child dietary diversity score defined by consumption of number of food group consumed by a child.
Time frame: Measured at baseline and endline. Time frame between baseline and endline will be approximately 6 months.
Food consumption score
Mean difference in food consumption score calculated using the frequency of consumption of different food groups consumed by a child.
Time frame: Measured at baseline and endline. Time frame between baseline and endline will be approximately 6 months.
Change in length-for-age Z scores
Difference between the treatment and control arms in the mean change in length-for-age Z scores
Time frame: Measured at baseline and endline. Time frame between baseline and endline will be approximately 6 months.
Change in weight-for-height Z scores
Difference between the treatment and control arms in the mean change in weight-for-height Z scores
Time frame: Measured at baseline and endline. Time frame between baseline and endline will be approximately 6 months.
Mid-Upper Arm Circumference
Difference between the treatment and control arms in the mean change in Mid-Upper Arm Circumference
Time frame: Measured at baseline and endline. Time frame between baseline and endline will be approximately 6 months.
Social support actions of father
Difference between the treatment and control arms in the mean score of social support actions of father as assessed in the follow-up questionnaire
Time frame: Measured at baseline and endline. Time frame between baseline and endline will be approximately 6 months.
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.