Data Source and Study Design:
This case control study design was analyzed based on 2016 Ethiopia Demographic and Health Survey (EDHS), which was designed as a nationally representative survey. It is the fourth Demographic and Health Survey (DHS) conducted in Ethiopia by the Central Statistical Agency (CSA) through the request of Federal Ministry of Health (FMoH) in collaboration with United States Agency for International Development(USAID) and ICF international as part of the International Demographic and Health Survey program known as MEASURE evaluation.
Sampling techniques and population
The 2016 EDHS sampling employed two-stage stratified cluster sampling technique to provide estimates for the health and demographic variables of interest for the country, structured into nine regional states and two City Administrations Councils (Addis Ababa and Dire Dawa).
The sampling frame was provided from the 2007 Ethiopian Population and Housing Census (PHC) conducted by the CSA. The census frame consisted of a total of 84,915 Enumeration Areas (EAs). The sample included 645 enumeration areas, (202 urban and 443 rural). An EA is a geographic area covering an average of 181 households. A nationally representative sample of 18,008 households and 10,752 children under age five years were eligible for height and weight measurements [12].
The ‘child record’ data set was downloaded from the MEASURE DHS website. All the surveys included boys and girls, and age ranged from 0 to 59 months. Child and maternal anthropometric data and various socio demographic variables were extracted from the data sets. The extracted and retrieved socio-demographic variables were: maternal age, educational status, place of residence, household wealth index (a composite measure of a household’s cumulative living standard), child’s age, sex, birth order, frequency of reading magazine, listening radio, watching television, maternal BMI and total children ever born.
After BMI Z-score of children was categorized in to normal weight (above minus two standard deviations (-2 SD) and below plus two standard deviations (+2 SD) from the median of the reference population), underweight (below minus two standard deviations (-2 SD) from the median of the reference population) and overweight and obese (more than two standard deviations (+2 SD) above the median of the reference population).
We retrieved children with underweight and those whose BMI z-score were missing or was recorded as “Height out of plausible limits” or “Age in days out of plausible limits” or “Flagged cases”, as their values were unusable since they were recorded in the database under special codes which corresponded either to responses that were considered inconsistent with other response in the questionnaire and thought to be probably an error, or to responses which value was “Don’t know”.
Finally, all eligible overweight and obese children (n=224) were included in to the analysis as cases and 448 normal weight children as controls were selected through systematic random sampling techniques from 7168 normal weight children living with their mothers. Outcome variable: According to WHO recommendations children who had BMI z-score over 2 were classified as overweight and obese children characterized as cases. And those who had BMI z-score ranged from -2 to 2 were classified as normal weight children, which are considered as controls [13].
Data analysis
Data analyses were carried out using SPSS 21TM – software. Cross tabulation was used to describe frequency or percentage of study participants. Bivariate regression analysis (with odds ratio and 95 percent confidence interval) was done to see the association of individual variables with the outcome of interest. Variables with p <0.25 at bivariate analysis were entered to multivariable logistic regression analysis using the forward likelihood ratio method. Finally, P<0.01 in multivariable analysis were considered to declare association between independent predictors and overweight and obese children.