Bivariate Spatial Modeling of Malnutrition among Under Five Children in Nigeria. A Bayesian Hierarchical Modelling Approach

Abstract

Childhood malnutrition continues to be a serious public health concern in Nigeria, with persistent geographic disparities and limited understanding of the joint spatial pattern of malnutrition indicators. This study examined the spatial distribution and determinants of childhood malnutrition in Nigeria using a two-stage Bayesian hierarchical modeling approach. The data were extracted from the 2024 Nigeria Demographic and Health Survey (NDHS). Separate univariate spatial models were fitted for stunting, underweight and wasting, incorporating fixed effects, nonlinear age effects, cluster-specific random effects, and spatial dependence. Spatial effects were extracted to evaluate pairwise correlations and estimate covariate differential fixed effects utilizing a stacked bivariate model. All models were estimated within the Integrated Nested Laplace Approximation (INLA) techniques. The study revealed that the prevalence of stunting, underweight, and wasting was 36.14%, 25.13%, and 8.27%, respectively. Maternal education exhibited the strongest protective effect across all indicators, while household wealth index, child gender, and climatic factors are also identified as significant risk factors. Differential effects analysis showed that maternal education had stronger protective effects on underweight than stunting, while temperature and precipitation showed significant differential effects across malnutrition outcomes. The study also identified strong clustering of stunting and underweight in northern states while wasting exhibited a more dispersed pattern. The findings highlight the need for geographically targeted and outcome-specific interventions that prioritize poverty alleviation programmes, maternal education and climate adaptation to improve child nutrition in Nigeria.