Statistics

Permanent URI for this collection

http://etd.aau.edu.et/handle/123456789/32

Browse

Browsing Statistics by Subject ", Multilevel"

Now showing 1 - 1 of 1
  • No Thumbnail Available
    Item
    Identification of Risk Factors and Regional Differentials in Under-Five Mortality in Ethiopia Using Multilevel Count Model
    (Addis Abeba university, 2011-06) Getiye, Tibebu; G/ Yohannes, Emmanuel (PhD)
    Under-five death (U5CD) is used as a population health indicator. It remains a big issue for developing countries, especially as researchers attempt to distinguish what factors contribute to the high levels. However under five death is still an issue and most studies have not considered regional variations of mothers in their studies. Objective: The Objective of this study is to investigate and quantify the regional variation of under-five death per mother and explore the major risk factors of under-five child death taking into consideration various demographic, socio-economic and health and environmental factors. Method: Thus, the study analyzes responses mothers on the number of deaths of children aged less than 60 months that they have experienced in their lifetime. Single and multilevel count model are used to explore the major risk factors and regional differentials in under-five mortality in Ethiopia. Result and conclusion: Descriptive statistics results show that nationally approximately one third (36%) of mothers have lost at least one under-five child in their lifetime. This figure (percent) is almost similar in all regions except Addis Ababa (which is 16%). The single level Poisson regression (multiple regression) results showed higher incidence rate of under-five deaths for SNNP, Gambela and Ben-Gumuz, while lower rates were estimated for Somalia and Addis Ababa. It was observed that regional differences in under-five mortality were reduced in magnitude when controls were added in the regression model. The relationship between under-five mortality and the explanatory variables age at first birth, mother education level, religion, employment status of mother and economic status of mother are significant. The preliminary plot of estimated/predicted U5CD for all regions versus different predictors shows mortality varies between regions of Ethiopia. The multilevel analysis further showed that there are substantial under-five death variations per mother among regions in Ethiopia, and are significant (šˆ š’–šŸŽ šŸ = 0.14, se= 0.021 and P-value <0.0001). Key words: U5CD, Poisson, Extra-Poisson, Negative binomial, Multilevel, MLwiN