Spatial prediction of childhood malnutrition across space in Nigeria based on point-referenced data: an SPDE approach

Malnutrition remains a leading cause of child mortality in Nigeria. The spatial analysis based on areal level approaches could, in reality, conceal variations at smaller units. Using point-referenced data from Nigeria Demographic and Health Survey, we quantify the prevalence of malnutrition among under-five children in Nigeria at 1.63 by 1.63 km spatial resolution, and compute the exceedance probability maps for stunting, wasting and underweight at 20% threshold level using the stochastic partial differential equation approach with Bayesian inference based on integrated nested Laplace approximation. Results show divergence prevalence of the malnutrition indicators among children living in neighbouring locations and that the prevalence of stunting and underweight increase with age. The prevalence of stunting was uneven among children living in Kebbi, Zamfara, Sokoto, Kaduna, Kano, Katsina, Bauchi, Gombe and Taraba states with more concentrations in the northern fringes of some of the states. Except for few locations in about three states, the probability is more than 90% that the prevalence of stunting in all parts of the country exceeds 20% but this was not the case for wasting. The findings can assist in location-specific policy formulation and implementations.