Hunachew Kibret Yohannis1, Moges Zerihun Fetene2, Habtamu Gebremariam Gebresilassie3. 1. Statistics Department, Under Natural and Computational Science College, at the University of Gondar, Gondar, Ethiopia. huneastat@gmail.com. 2. Statistics Department, Under Natural and Computational Science College, at the University of Gondar, Gondar, Ethiopia. 3. Statistics Department, Under Natural and Computational Science College, Dilla University, Dilla, Ethiopia.
Abstract
BACKGROUND: Mortality rate under the age of five is the proportion of deaths of children below the age of 5 years out of 1000 live births. It is related with the living standard of a population, and it is taken as one of the health and socioeconomic status deterioration index. Mortality rate under the age of five also indicates a poor quality life standards of a population. It is very significantly high in Sub-Saharan African countries. Ethiopia is one of these Sub-Saharan African countries where mortality rate under the age five is high. This research work aims to identify the determinants and associated factors of under-five mortality in Ethiopia. METHODS: The data for this paper were gathered from the EDHS 2016, collected by CSA. In this study, count family models such as Poisson, negative binomial, zero-inflated Poisson and zero-inflated negative binomial regression were applied for analyzing the data. Each of these count models were compared with different statistical tests like log-likelihood ratio test, Akaike information criteria, mean absolute difference, Vuong test and observed versus predicted probability plot. RESULTS: The study revealed that as mothers' age at first birth increased by one unit, the average number of under-five mortality rate decreased by 2.69%. In the same way the number of under-five mortality of Afar, Benishangul Gumuz and Dire Dawa were 1.3446, 1.6429 and 1.3320 times more likely to Tigray respectively. The risk of under-five mortality for primary and secondary education level of the mother was 28.31 and 40.96% less likely than to mothers who have no education respectively. CONCLUSION: From the result we found that, there were overabundance zeros and broad heterogeneity in the non-zero outcomes. Zero-inflated negative binomial regression model was found to best fit the data, and from the regression model, age of mothers at first birth, mother's education level, place of residence and region were statistically significant factors of under-five mortality per mother.
BACKGROUND:Mortality rate under the age of five is the proportion of deaths of children below the age of 5 years out of 1000 live births. It is related with the living standard of a population, and it is taken as one of the health and socioeconomic status deterioration index. Mortality rate under the age of five also indicates a poor quality life standards of a population. It is very significantly high in Sub-Saharan African countries. Ethiopia is one of these Sub-Saharan African countries where mortality rate under the age five is high. This research work aims to identify the determinants and associated factors of under-five mortality in Ethiopia. METHODS: The data for this paper were gathered from the EDHS 2016, collected by CSA. In this study, count family models such as Poisson, negative binomial, zero-inflated Poisson and zero-inflated negative binomial regression were applied for analyzing the data. Each of these count models were compared with different statistical tests like log-likelihood ratio test, Akaike information criteria, mean absolute difference, Vuong test and observed versus predicted probability plot. RESULTS: The study revealed that as mothers' age at first birth increased by one unit, the average number of under-five mortality rate decreased by 2.69%. In the same way the number of under-five mortality of Afar, Benishangul Gumuz and Dire Dawa were 1.3446, 1.6429 and 1.3320 times more likely to Tigray respectively. The risk of under-five mortality for primary and secondary education level of the mother was 28.31 and 40.96% less likely than to mothers who have no education respectively. CONCLUSION: From the result we found that, there were overabundance zeros and broad heterogeneity in the non-zero outcomes. Zero-inflated negative binomial regression model was found to best fit the data, and from the regression model, age of mothers at first birth, mother's education level, place of residence and region were statistically significant factors of under-five mortality per mother.