BAYESIAN MODEL FOR SPATIAL PREDICTION OF POINT REFERENCED DATA WITH APPLICATION TO CHILDHOOD MALNUTRITION IN NIGERIA

Abstract

Malnutrition is the leading cause of child morbidity and mortality in Nigeria. Children are said to be stunted, wasted, and underweight if their height-for-age, weight-for-height, and weightfor- age z-score falls below the threshold of the World Health Organisation median growth standards respectively. Predicting the prevalence of childhood malnutrition has been by region which in the real sense could be misleading as variation occurs even at smaller units. This study makes use of point referenced data from the 2013 Nigeria Demographic and Health Survey to predict the prevalence of malnutrition in the forms of stunting, wasting, and underweight for every location in Nigeria, thus, avoiding issues of generalisation. The computational cost of obtaining model parameters was reduced by employing the stochastic partial differential equation approach, which allows the continuously indexed Gaussian field of the underlying spatial process to be represented by a Gaussian Markov random field. Inferences was based on integrated nested Laplace approximation method of Bayesian computing. Results show that children that resides in urban area offer all forms of malnutrition than their counterparts in rural areas, though, the estimates for stunting and wasting are insignificant. Female children are 20.4%, 13%, and 19.9% less likely to be stunted, wasted, and underweight respectively compared to male children. Results also show that children are prone to malnutrition at the early months of their life and older mothers have better nourished children. The predicted maps showing the prevalence and spatial distribution of malnutrition indicate that there exists huge variation in the nutritional status of children that live within the same geographical settings. In particular, there is uneven distribution of stunting among children living in Ondo state, while wasting and underweight were unevenly distributed among children from Borno and Nasarawa states respectively.