Abstract
Childhood undernutrition is a major public health challenge in sub-Saharan Africa, particularly Nigeria. Determinants of child malnutrition may have substantial spatial heterogeneity. Failure to account for these small area spatial variations may cause child malnutrition intervention programs and policies to exclude some sub-populations and reduce the effectiveness of such interventions. This study uses the Composite Index of Anthropometric Failure (CIAF) and a geo-additive regression model to investigate Nigeria's prevalence and risk factors of childhood undernutrition. The geo-additive model permits a flexible, joint estimation of linear, non-linear, and spatial effects of some risk factors on the nutritional status of under-five children in Nigeria. We draw on data from the most recent Nigeria Demographic and Health Survey (2018). While the socioeconomic and environmental determinants generally support literature findings, distinct spatial patterns were observed. In particular, we found CIAF hotspots in the northwestern and northeastern districts. Some child-related factors (Male gender: OR = 1.315; 95% Credible Interval (CrI): 1.205, 1.437) and having diarrhoea: OR = 1.256; 95% CrI: 1.098, 1.431) were associated with higher odds of CIAF. Regarding household and maternal characteristics, media exposure was associated with lower odds of CIAF (OR = 0.858; 95% CrI: 0.777, 0.946). Obese maternal BMI was associated with lower odds of CIAF (OR = 0.691; 95% CrI: 0.621, 0.772), whereas, mothers classified as thin were associated with higher odds of CIAF (OR = 1.216; 95% CrI: 1.055, 1.411). Anthropometric failure is highly prevalent in Nigeria and spatially distributed. Therefore, localised interventions that aim to improve the nutritional status of under-five children should be considered to avoid the under-coverage of the regions that deserve more attention.