Statistical models to study the BMI of under five children in Ethopia.

Abstract

Maternal and child malnutrition has long and short-term consequences on the health status of the people and on the country’s economy. It is among the major public health problems in Ethiopia. Worldwide, maternal and child malnutrition is an underlying cause for more than 3.5 million deaths each year. About 35% of the global disease burden is in under five children. Such a heavy burden requires an understanding of the nutritional status of the people, especially children under the age of five years and associated factors. Therefore, this study attempted to use possible statistical methods to estimate the effects of the risks related to the nutritional status of children. It also tried to identify the socio-economic and demographic factors that are associated with the BMI of under five children in Ethiopia. The study employed the 2016 Ethiopian Demographic and Health Survey data. A nationally representative sample of children under the age of five years was used to get information on weight and height measures of under five children. The BMI of children under five years of age was used as a response variable to fit weighted quantile regression. The covariates, age of a child, sex and other relevant socio-economic and demographic factors were used in the study. Following the quantile regression, the generalized linear models such as logistic regression model was applied after categorizing the response variable, BMI of under five children, into two categories namely normal and malnourished. Following binary logistic regression, an attempt to fit ordinal logistic regression was made. That means nutritional status was considered as ordinal outcome with four categories namely underweight, normal, overweight and obese. The findings and comparison of estimates using these different statistical methods with and without complex survey design were presented. The results revealed that methods that take into account the complex nature of the design, perform better than those that do not take this into account. It has also been found that age of a child, weight of child at birth, mother’s BMI, educational attainment of mother, region and wealth index were significantly associated with under five children’s nutritional status. Furthermore, the results are discussed and then a conclusion is made in the context of policy implication for Ethiopia.