Related blog posts
Stunting, or low height-for-age, remains a significant development challenge throughout much of Africa south of the Sahara. According to the World Health Organization (WHO) , childhood stunting can have significant long-term effects, including decreased cognitive and physical development, increased vulnerability to disease, and reduced productive capacity into adulthood.
In Zambia, 1 million children under the age of 5 (45 percent) are stunted as a result of chronic malnutrition caused by a lack of high-quality food and inadequate health and care practices in the first 1,000 days of life. Reducing this number and improving overall child nutrition has become a key development priority, according to the Realigning Agriculture to Improve Nutrition (RAIN) project , coordinated by IFPRI and Concern Worldwide .
Specifically, efforts have focused on integrating agricultural, health, and nutrition interventions for children from conception until 24 months of age. Such interventions aim to improve agricultural production at the household level, increase availability and access to high-quality foods for children, and improve health and nutrition practices. However, until recently, little rigorous evaluation of the actual effectiveness of these interventions has been done. To fill this knowledge gap, the RAIN project has released an impact evaluation report tracking the project’s impacts on a variety of indicators, including child stunting, infant and young child feeding (IYCF) practices, agricultural production, and health and nutrition knowledge among caregivers of young children.
The five-year project was carried out in Mumbwa District, where child stunting is even higher than the national average, at 59 percent. In addition to IFPRI and Concern Worldwide, project partners included the Mumbwa District Child Development Agency (MCDA), Women for Change (WFC), the Zambian Ministry of Agriculture and Livestock (MAL), the Zambian Ministry of Health (MoH), and the Zambian Ministry of Community Development Maternal and Child Health (MCDMCH).
The project included two intervention packages: one that provided agricultural training and inputs only and one that combined agricultural training and inputs with additional training to promote improved health and nutrition practices. The evaluation utilized a hybrid evaluation design combining a cluster randomized probability design that compares RAIN’s two intervention packages with a plausibility design that compares the RAIN intervention packages to a non-randomized control group. The sample size covered by the panel was approximately 3044 households with a child aged 24-59 months at the baseline.
The interventions were delivered through local women’s groups led by a Smallholder Model Farmer (SMFs) who was nominated by her group to receive agricultural training and inputs, which she passed along to the rest of the group during monthly meetings. The women’s groups were linked with existing Community Health Volunteers (CHVs) to conduct the nutrition and health intervention.
According to the impact evaluation report, results from the project have been mixed. Overall participation in the program was low – 31 percent in the agriculture-only intervention and 34 percent in the agriculture-nutrition intervention. The attendance of SMFs at RAIN groups was high, at approximately 90 percent, but attendance from CHVs was low, at only 38-45 percent. This meant that participating households had more opportunity to interact with agricultural trainers than health and nutrition trainers.
In terms of child stunting, the interventions had no discernible impacts. While stunting fell in all three study groups (the two intervention groups and the control group) between baseline and endline, the control group experienced a significantly greater decrease. Similarly, the evaluation found no overall attributable program impact on improving IYCF practices; the exception is the consumption of legumes/nuts, which was found to be higher in both intervention groups than in the control group.
Caregivers’ health and nutrition knowledge also saw little change as a result of the program. IYCF knowledge increased over time across all groups, with the exception of knowledge about breastfeeding based on child’s demand and continued breastfeeding if the mother is ill. Knowledge regarding breastfeeding in general was lower in the agriculture-only group than in the control group, but knowledge regarding when to introduce complementary foods was significantly higher in the agriculture-nutrition group than in the control group. Knowledge regarding hygiene (including hand washing practices, safe drinking water, and protection of children from worms) varied significantly across the different groups at endline, but the direction of these differences were not consistently attributable to a single study group or to the RAIN interventions.
The program’s impacts on agricultural production were more positive. Both the agriculture-only group and the agriculture-nutrition group had greater increases in the total number of foods produced, the total number of agricultural activities engaged in by households, and the number of months in which vitamin A-rich foods and dairy products were produced than the control group. These increases show that the interventions did have a positive impact on the availability of and access to year-round supplies of diverse, nutritious foods at the household level.
There are several possible explanations for this lack of consistent results. The program’s low coverage is the first such potential explanation. Local women’s groups formed the major channel through which RAIN interventions were conducted, but only only one-third of eligible households in the district reported participating in the program through such a group. In addition, home visits by SMFs and CHVs were low, resulting in a lack of one-to-one trainings that could have boosted the program’s success.
As noted, SMFs were generally more active in the program, both through women’s groups and through home visits. This could explain why the agriculture outcomes improved more than the health and nutrition outcomes. It is possible that SMFs were more actively involved because they were more incentivized by the program (through the provision of agricultural inputs) than CHVs, who did not receive similar incentives until later in the program. In addition, the position of SMF was created specifically for the RAIN project, meaning that the information they provided was new, while the CHVs had existed in the community previously and had likely communicated their health and nutrition messages before.
The program’s design also prevented the evaluation from identifying informal leakages of secondary intervention components, such as agricultural inputs and knowledge provided by non-program staff, among peers and relatives across the study groups. In addition, the evaluation is unable to account for the impact of improvements in government health services across the district, since these services were available to both intervention groups and to the control group and since improvements in IYCF knowledge improved across all groups during the study period. Increased access to and use of nutrition services provided through government health clinics could improve households’ health and nutrition knowledge and child feeding practices, thus preventing detection and attribution of RAIN interventions beyond general trends.
While the program’s results were mixed, however, the report indicates that the consistent impact of RAIN interventions on agricultural production in Zambia is worth noting and could inform the design and scaling up of future programs.