Sensitivity of Nutrition Indicators to Measure the Impact of a Multi-Sectoral Intervention: Cross-Sectional, Household, and Individual Level Analysis

Interventions tackling multiple drivers of child malnutrition have potential, yet the evidence is limited and draws on different analysis and nutrition outcomes, reducing comparability. To better understand the advantages and disadvantages of three different analytical approaches on seven common nutrition indicators, we use panel data (2012, 2014, 2015) on 1420 households from a randomized control study of a multi-sectoral intervention in Chad. We compare program impact using three types of analysis: a cross-sectional analysis of non-matched children; a panel analysis on longitudinal outcomes following the worst-off child in the household; and a panel analysis on longitudinal outcomes of matched children. We find that the sensitivity of the nutrition outcomes to program impact increases with each subsequent analytical approach, despite the reduction in sample size, as the analysis is able to control for more non-measured child and household characteristics. In the matched child panel analysis, the odds of a child being severely wasted were 76% lower (CI: 0.59–0.86, p = 0.001), the odds of being underweight were 33% lower (CI: 0.15–0.48, p = 0.012), and weight-for-height z-score was 0.19 standard deviations higher (CI: 0.09–0.28, p = 0.022) in the treatment versus control group. The study provides evidence for multi-sectoral interventions to tackle acute malnutrition and recommends the best practice analytical approach.