Modelling child anaemia and co-existing infections using log-linear models

Abstract

Background Uganda grapples with a considerable anaemia-malaria-fever burden, reporting approximate prevalence rates as high as 33%, 34%, and 37% in specific regions. In recent years, attempts have been made by the Ministry of Health to address the combined burden of the characterized conditions of these illnesses. However, the relation- ship between malaria, fever, and anaemia has not been well characterized among young children living in many communities. By employing log-linear models, this study aims to examine patterns and associations between malaria, fever, and child anaemia in Uganda while controlling for maternal anaemia. Methods Utilizing secondary data from the 2018–2019 Uganda Malaria Indicator Survey (MIS), the study focused on children aged 0–60 months. The sample included 7,124 children selected through a two-stage sampling process involving clusters and households. Five log linear models, namely; saturated, mutual independence, joint independ- ence, conditional independence and homogenous models were fitted. The saturated model was used as the refer- ence model. Results The G2 statistics and p-values for each model were as follows: saturated model (G2 = 0.00, p = 1.00), mutual independence model (G2 = 321.45, p < 0.001), joint independence model (G2 = 214, p < 0.001), conditional inde- pendence model (G2 = 109.53, p < 0.001), and homogeneous model (G2 = 10.24, p = 0.76). The homogeneous model adequately fit the data, showing the smallest G2 statistic and the largest p-value, indicating no significant lack of fit. Additionally, children who tested positive for malaria were found to be two times more likely to have anaemia than those who tested negative. Conclusion This study underscores the interconnectedness of malaria, fever, and anaemia in Ugandan children, with maternal anaemia serving as a critical contextual factor. Using log-linear modelling, uncovered patterns and interactions that highlight how these conditions influence one another, emphasizing the value of integrated interventions. Targeted approaches that address maternal health, enhance malaria prevention, and provide nutritional support are essential to reducing the syndemic burden of these conditions in Uganda. Keywords Association, Homogenous, Independent, Saturated, Uganda