Modeling determinants of anemia among pre-school children in Ethiopia: application of marginal and subject-specific models

Anemia is a chronic disease that seriously affects young children and pregnant women. Knowledge of disease clustering is important because it may provide insights into the etiology of disease and risk factors operating within different levels of the clusters. In this study, we tried to identify determinants of anemia among pre-school children aged 6-59 months in the 11 regions of Ethiopia, with higher probability of occurrence of these determinant factors would be inferred to be most likely to experience anemia. To answer the objective of the research question, models that handle the complexities of correlated data were employed. Hence, both marginal and subject-specific models are employed. The models used were: Generalized Estimating Equations, Alternating Logistic Regression, Proportional Odds Model and Generalized Linear Mixed Model. Statistical findings revealed that the risk of being anemic reduced with age in girls, while boys showed higher risk. Children in rural were found to be less likely to be anemic. Children in large households were found to be more having a higher risk of anemia. Similarly, malaria occurrence was strongly correlated to anemia