Homestead Food Production and Child Anemia in Burkina Faso: The Mediating Roles of Mother's Knowledge and Production of Micronutrient‐Rich Fruits and Vegetables

Abstract

BackgroundThere is sparse evidence on how integrated agriculture‐nutrition interventions can improve child nutrition. In complex programs it is important to investigate hypothesized pathways to impact, estimating the contributions of potential mediators to the impact on nutrition. We use baseline and endline data from a cluster‐randomised controlled trial (cRCT) in rural Burkina Faso evaluating a homestead food production (HFP) programme that previously found an improvement in child hemoglobin (Hb) levels over a 2‐year period in the treatment (15 villages) compared to the control (25 villages).ObjectiveThis study seeks to understand the pathways through which an HFP program achieved impact on child Hb. Specifically, we aim to disentangle the roles that maternal knowledge of complementary feeding and hygiene (an index), and production of micronutrient‐rich fruits and vegetables (binary) played in the impact found on Hb. Further, we test if the effects vary by market access, as found in previous studies.MethodsAnalyses of a panel of 902 children (aged 3–12 months at baseline; 24–39 months at endline) whose households participated in a cRCT from 2010–2012. A parametric approach to causal mediation analysis was applied – an extension of Robins' g‐computation formula to more than one mediator, implemented using Monte Carlo simulation – to estimate novel measures of direct and indirect effects known as “interventional” effects; these effects are identified under weaker assumptions than required in traditional mediation methods. The approach accounts for intermediate confounders, and accommodates nonlinearities in the observed data models, thereby overcoming some limitations of traditional approaches.ResultsIncreases in knowledge and production were found to be significant pathways to improved child Hb, with the increase in knowledge alone playing a large mediating role (Interventional Indirect Effect (IIE): 0.257 g/dL (CI 0.053, 0.460), Proportion Mediated (PM: IIE/Total Causal Effect (TCE)): 0.802). The IIE through production (a combination of the effects via production alone and with knowledge leading to production) was 0.223 g/dL (CI −0.011, 0.457) and the PM 0.698. Further, among those with market access, the TCE and both IIEs were insignificant. Conversely, among those without market access the TCE was greater (0.524 g/dL, CI 0.253, 0.789), the IIE via knowledge the same (0.258 g/dL, CI −0.060, 0.577), and that via production greater (0.303 g/dL, CI −0.076, 0.682) (PM 0.494 and 0.579 respectively).ConclusionOver 80% of the improvement in child Hb was mediated by maternal knowledge, while 70% was mediated by production of micronutrient‐rich fruits and vegetables (alone and with knowledge leading to production). This informs us about the respective roles of these two upstream mediators, confirming that they play large roles in improving child Hb in the context this HFP program. Additional possible mediators subsequent to knowledge and/or production (feeding and hygiene practices, vitamin A supplementation, chicken ownership, diarrhoea) may or may not have contributed to impact on Hb. Other known determinants of anemia (malaria, helminths) may have been affected by knowledge, but were not measured. Finally, the program was found to be most effective in villages without a market, with production playing a larger role relative to knowledge than before disaggregating.Support or Funding InformationThe EHFP program in Burkina Faso was supported by USAID through HKI; the Gender, Agriculture and Assets Project; and the CGIAR Research Program on Agriculture for Nutrition and Health led by IFPRI. Andrew Kennedy is funded by the Leverhulme Centre for Integrative Research on Agriculture and Health (LCIRAH).