Background: Particulate matter below 2.5 mm diameter (PM_2.5) is a part of air pollution that has adverse effects on health. Data on exposure to ambient PM_2.5 is not well monitored in sub-Saharan Africa due to limited resources and skilled manpower. The effect of PM_2.5 on health is least explored in Ethiopia.

Objectives: The study has assessed the relationship between satellite-based ambient PM_2.5 pollution and under-five mortality in Ethiopia.

Methods: The study used the data from Ethiopian Demographic Health Surveys (EDHS) conducted in 2016, collected between January 18 and June 27. Under-five children with the child mortality information and coordinates of geographical location were included. Satellite-based ambient PM_2.5 concentration was extracted from the Atmospheric Composition Analysis Group website at Washington and Dalhousie University, in the United States and Canada, respectively. Datasets were downloaded from their respective websites. Annual pollution level and mortality datasets were matched by children’s geographical location, birth, death, and interview dates. The relationship between ambient PM_2.5 and under-five mortality was determined by multilevel multivariable logistic regression. The statistical analyses were two-sided at 95% confidence interval.

Results: The study addressed 10452 children with the proportion of under-five mortality being 5.4% (95% CI 5.0% - 6.8%). The estimated lifetime mean annual exposure of ambient PM_2.5 was 20.1 ± 3.3 mgm^-3. Significant clustering of mean annual PM_2.5 concentration and under-five mortality proportion were varied by region. A ten-unit increase in lifetime mean annual ambient PM_2.5 was associated with 2.40 [95% CI 1.51, 3.80] times more odds of under-five mortality after adjusting for other variables. In addition, children with food cooked inside a house but with no separate room, mothers without formal education, very large birth size, twins, born at home, and stunting prevalence were significantly positively.

Conclusions: Clustered spatial distribution of ambient PM_2.5 concentration and under-five mortality has existed. Satellite-based PM_2.5 is significantly associated with under-five mortality adjusted for other variables. Validating satellite-based PM data with ground-based measurements is advised. Additional ground-based PM_2.5 monitoring devices, particularly in Afar where PM_2.5 and under-five mortality were higher, is suggested.