Geographical variation and associated factors of childhood measles vaccination in Ethiopia: a spatial and multilevel analysis

Background In Ethiopia, despite considerable improvement of measles vaccination, measles outbreaks is occurring in most parts of the country. Understanding the neighborhood variation in childhood measles vaccination is crucial for evidence-based decision-making. However, the spatial pattern of measles-containing vaccine (MCV1) and its predictors are poorly understood. Hence, this study aimed to explore the spatial pattern and associated factors of childhood MCV1 coverage. Methods An in-depth analysis of the 2016 Ethiopia demographic and health survey data was conducted, and a total of 3722 children nested in 611 enumeration areas were included in the analysis. Global Moran's I statistic and Poisson-based purely spatial scan statistics were employed to explore spatial patterns and detect spatial clusters of childhood MCV1, respectively. Multilevel logistic regression models were fitted to identify factors associated with childhood MCV1. Results Spatial hetrogeniety of childhood MCV1 was observed (Global Moran's I = 0.13, p-value < 0.0001), and seven significant SaTScan clusters of areas with low MCV1 coverage were detected. The most likely primary SaTScan cluster was detected in the Afar Region, secondary cluster in Somali Region, and tertiary cluster in Gambella Region. In the final model of the multilevel analysis, individual and community level factors accounted for 82% of the variance in the odds of MCV1 vaccination. Child age (AOR = 1.53; 95%CI: 1.25-1.88), pentavalent vaccination first dose (AOR = 9.09; 95%CI: 6.86-12.03) and third dose (AOR = 7.12; 95%CI: 5.51-9.18, secondary and above maternal education (AOR = 1.62; 95%CI: 1.03-2.55) and media exposure were the factors that increased the odds of MCV1 vaccination at the individual level. Children with older maternal age had lower odds of receiving MCV1. Living in Afar, Oromia, Somali, Gambella and Harari regions were factors associated with lower odds of MCV1 from the community-level factors. Children far from health facilities had higher odds of receiving MCV1 (AOR = 1.31, 95%CI = 1.12-1.61). Conclusion A clustered pattern of areas with low childhood MCV1 coverage was observed in Ethiopia. Both individual and community level factors were significant predictors of childhood MCV1. Hence, it is good to give priority for the areas with low childhood MCV1 coverage, and to consider the identified factors for vaccination interventions.