A Bivariate Analysis of the Spatial Distributions of Stunting and Wasting Among Children Under-Five in Nigeria

Abstract

Stunting and wasting are major malnutrition issues among children under five years of age and have continued to remain unacceptably high in Nigeria leading to high rates of child morbidity and mortality. Evidence-based strategies are required by government and non-governmental agencies to mitigate the suffering of these children, and this could be realised when the association between the determinants and the geographical distributions are fully understood. Using data from four waves of the Nigerian Demographic and Health Survey, we employed a distributional bivariate probit model to examine the geographical distributions of the levels and linear association between acute and chronic malnutrition in Nigeria after accounting for possible observed determinants. Bayesian inference was based on Markov chain Monte Carlo simulation. The findings reveal substantial spatial variations in stunting and wasting among under-five children in Nigeria, indicating a north–south divide. The findings show negative linear association between the two malnutrition indicators among children in some northern fringe states but positive for Akwa Ibom, Ebonyi and Anambra. The correlation also peaks around age 20 months indicating that during the first 2 years of life, the children have an increasing likelihood of suffering from stunting and wasting.

Spatial trends and projections of chronic malnutrition among children under 5 years of age in Ethiopia from 2011 to 2019: a geographically weighted regression analysis

“…This is supported by previous spatial analysis of under-five stunting conducted previously in Ethiopia [ 24 , 25 , 27 – 30 ]. Furthermore, in the other settings, various pieces of evidence revealed the presence of geographical variation of under-five stunting such as studies conducted in Rwanda [ 44 ], Ghana [ 18 ], Zambia [ 19 ], Nigeria [ 20 , 21 ].…”

Spatial distribution of stunting and wasting in 6–59 months children in Nepal: analysis using a Bayesian distributional bivariate probit model

“…We employed a bivariate probit distribution regression model ( 17 , 24 ) to jointly fit the binary variables – stunting and wasting, accounting for their correlation and estimating their association with the considered covariates. Let, for an i th child, Y i 1 and Y i 2 denote stunting and wasting as binary response variables, and and refer to corresponding latent variables, defined such that, for j = 1,2 …”

“…The present study applies a distributional bivariate probit model to explore the association between acute and chronic malnutrition and other factors (16) . The inclusion of structured additive predictor into the model permits linking spatial and other covariates to the means and correlations between the undernutrition indicators (17,18) . The categorical predictors under the study were assumed to have a linear effect; the effects of metrical covariates were considered non-linear and modelled non-parametrically.…”