Age-based preventive targeting of food assistance and behaviour change and communication for reduction of childhood undernutrition in Haiti: a cluster randomised trial. uri icon


  • Background Food-assisted maternal and child health and nutrition programmes usually target underweight children younger than 5 years of age. Previous evidence suggests that targeting nutrition interventions earlier in life, before children become undernourished, might be more effective for reduction of childhood undernutrition.
  • Findings There were no differences between programme groups at baseline. At follow-up, stunting, underweight, and wasting (using WHO 2006 reference data) were 4-6 percentage points lower in preventive than in recuperative communities; and mean anthropometric indicators were higher by +0 .14 Z scores (height for age; p=0. 07), and +0. 24 Z scores (weight for age and weight for height; p<0. 0001). The effect was greater in children exposed to the preventive programme for the full span between 6 and 23 months of age than in children exposed for shorter durations during this period. The quality of implementation did not differ between the two programmes; nor did use of services for maternal and child health and nutrition.
  • Interpretation The preventive programme was more effective for the reduction of childhood undernutrition than the traditional recuperative model.
  • Methods We used a cluster randomised trial to compare two World Vision programmes for maternal and child health and nutrition, which included a behaviour change and communication component: a preventive model, targeting all children aged 6-23 months; and a recuperative model, targeting underweight (weight-for-age Z score <-2) children aged 6-60 months. Both models also targeted pregnant and lactating women. Clusters of communities (n=20) were paired on access to services and other factors and were randomly assigned to each model. Using two cross-sectional surveys (at baseline and 3 years later), we tested differences in undernutrition in children aged 12-41 months (roughly 1500 children per survey). Analyses were by intention to treat, both by pair-wise community-level comparisons and by child-level analyses adjusting for the clustering effect and child age and sex. This study is registered with, number NCT00210418.

publication date

  • 2008
  • 2008
  • 2008