Targeting urban malnutrition: a multi-city analysis of the spatial distribution of childhood nutritional status

This paper assesses the degree to which childhood malnutrition and its contributing factors are clustered by neighborhood in seven different cities in Africa, Asia, and Latin America. The analysis is based on data from eight different national household surveys that used a two-stage sampling design (households within clusters). Spatial clustering was assessed using the intracluster correlation coefficient (ρ), which may be interpreted as the expected correlation between any randomly selected pair of observations drawn from the same cluster. In general, per capita expenditure and the share of the household budget spent on food showed relatively high spatial clustering across the seven cities, but the magnitude of this clustering varied markedly from city to city. Spatial clustering in the provision of basic services also varied greatly. There was consistently little evidence of spatial clustering of infectious disease, childhood mortality, or the weight-based nutrition indicators. Age-standardized height, on the other hand, showed a little more spatial clustering, with a median intracluster correlation of ρ=0.12. Some cities showed relatively higher levels of spatial clustering on several measures of deprivation simultaneously, while other cities showed consistently lower levels of clustering. Many nutrition interventions are intrinsically geographically targeted. While geographical targeting tends to be administratively simpler than individual targeting and can be politically convenient, the current analysis suggests that where nutrition interventions are focused on stunting (low height-for-age), targeting by neighborhood may often lead to unacceptably high rates of undercoverage and leakage of benefits to the non-needy.