Spatial Distributions and Risk Factors of Overweight and Obesity Among Women in Nigeria Using Structured Geo-Additive Regression Models: Analysis of 2018 Nigeria Demographic Health Survey

Abstract

Overweight and obesity which are known to pose serious health problems are becoming increasingly prevalent in Nigeria which is a sub-Saharan African country. This study utilized the 2018 Nigeria Demographic Health Survey to examine demographic and socio-economic risk factors of overweight and obesity among Nigerian women aged 15-49 years. Exploratory analysis was used to provide basic description of the data while a semiparametric structured additive models was used to describe the relationship between the presumed factors and overweight and obesity status while also accounting for spatial effects at state level. The national prevalence of overweight and obesity among Nigerian women was found to be 27.4%. Increased risk of overweight and obesity among Nigerian women was found to be strongly associated with being older, high educational level, being rich, living in an urban area, having many children, being pregnant, and residing in southern part of Nigeria. In respect to ethnicity and religion, the Fulani tribe and Islamic religion were associated with lower prevalence of overweight and obesity. Overweight and obesity were found to be significantly more prevalent in the Southern parts compared to the Northern parts of Nigeria. The highest and lowest prevalence of overweight and obesity were observed in Anambra and Yobe states respectively. Prevalence of overweight and obesity was higher among Muslim women compared to Christian women since most Northern women are Muslims and most Southern women are Christians. Random (unstructured) spatial effects were significant indicating that overweigh/obesity was influenced by unobserved state specific factors

Estimating Nonlinear Regression Parameters using Particle Swarm Optimization and Genetic Algorithm

“…Perhaps, the PSO or GA could be used to locate set of points that evolve to optimal global, then, a gradient based method that would accurately locate the deep local minimum. Future studies could also explore the usage of the PSO and GA techniques to solve optimization problems arising from other form of models such as nonlinear mixed effects models (Adeniyi et al, 2018), generalized mixed effects models (Ezenweke et al, 2022a), geo-spatial structured additive models (Ezenweke et al, 2022b)…”