Spatial location, temperature and rainfall diversity affect the double burden of malnutrition among women in Kenya

Abstract

Studies have looked into how environmental and climate covariates affect under-and over-nutrition, but little is known about the spatial distribution of different forms of malnutrition in Kenya and whether there are locations that suffer from double-burden of malnutrition. This research quantifies spatial variations and estimates how climatic and environmental factors affect under-and over-nutrition among women in Kenya. This enables us to determine if the patterns in which these factors affect the malnutrition indicators are similar and whether there are overlaps in the spatial distributions. The study used data from the Demographic and Health Survey, which included cross-sectional data on malnutrition indicators as well as some climate and environmental variables. A multicategorical response variable that classified the women into one of four nutritional classes was generated from the body mass index (BMI) of the women, and a Bayesian geoadditive regression model with an estimate based on the Markov chain Monte Carlo simulation technique was adopted. Findings show that women in Turkana, Samburu, Isiolo, Baringo, Garissa, and West Pokot counties are more likely to be underweight than women in other counties while being overweight is prevalent in Kirinyag'a and Kitui counties. Obesity is prevalent in Kirinyag'a, Lamu, Kiambu, Murang'a, and Taita Taveta counties. The study further shows that as mean temperature and precipitation increase, the likelihood of being underweight reduces. The chances of being underweight are lower among literate women [OR: 0.614; 95% CrI: 0.513,0.739], married women [OR: 0.702; 95% CrI: 0.608,0.819] and those from rich households [OR: 0.617; 95% CrI: 0.489,0.772], which is not the case for overweight and obesity. The generated spatial maps identify hot spots of the double burden of malnutrition that can assist the government and donor agencies in channeling resources efficiently.

Keywords: Climate; Food security; Geoadditive regression model; Malnutrition; Spatial effects.

Fig. 1

Nonlinear estimates for underweight. Shown are the posterior means (represented by the center line) together with 95% and 80% pointwise credible intervals based on fully Bayesian approach estimated with MCMC techniques.

Fig. 2

Nonlinear estimates for overweight. Shown are the posterior means (represented by the center line) together with 95% and 80% pointwise credible intervals based on fully Bayesian approach estimated with MCMC techniques.

Fig. 3

Nonlinear estimates for obesity. Shown are the posterior means (represented by the center line) together with 95% and 80% pointwise credible intervals based on fully Bayesian approach estimated with MCMC techniques.

Fig. 4

Shown are the spatial effects for underweight level of malnutrition.

Fig. 5

Shown are the spatial effects for overweight level of malnutrition.

Fig. 6

Shown are the spatial effects for obesity level of malnutrition.