Application of Binomial and Beta-Binomial Regression Models on Incidence of Diarrhea Among children Aged 6-59 Months in Miesso District, Oromia, Ethiopia

Abstract

In binomial regression analysis model dependent variable follows a binomial distribution. β-binomial model is a regression model for outcomes following the β-binomial probability distribution. The aim of present study was to fit binomial regression and beta-binomial regression models for incidence of diarrhea, a prominent cause of health disorders/ deaths in under-five year age children. The study was conducted in the children aged 6-59 months in Miesso Wereda, Oromia, Ethiopia from June to September 2017. The data used was obtained from a cross-sectional study using well-structured questionnaires. The Predictors used for the current study were child sex, household food security status, sources of drinking water, and toilet ownership status. Binomial and β-binomial models were fitted to identify the significant predictors. The overdispersion status of the data was tested and revealed that there was no over-dispersion and the fitted models had similar Akaiki information criteria and Bayesian information Criteria with similar results. The source of water (pipe water) and having toilet facility significantly affected the incidence of diarrhea among the studied group of children. The study recommended that giving attention to environmental sanitation and creating pure water facility shall reduce diarrhea incidence in children.