Mapping geographical inequalities of incomplete immunization in Ethiopia: a spatial with multilevel analysis

Background Immunization is one of the most cost-effective interventions, averting 3.5–5 million deaths every year worldwide. However, incomplete immunization remains a major public health concern, particularly in Ethiopia. The objective of this study is to investigate the geographical inequalities and determinants of incomplete immunization in Ethiopia. Methods A secondary analysis of the mini-Ethiopian Demographic Health Survey (EDHS 2019) was performed, utilizing a weighted sample of 3,865 children aged 12–23 months. A spatial auto-correlation (Global Moran's I) statistic was computed using ArcGIS version 10.7.1 to assess the geographical distribution of incomplete immunization. Hot-spot (areas with a high proportion of incomplete immunization), and cold spot areas were identified through Getis-Ord Gi * hot spot analysis. Additionally, a Bernoulli probability-based spatial scan statistics was conducted in SaTScan version 9.6 software to determine purely statistically significant clusters of incomplete immunization. Finally, a multilevel fixed-effects logistic regression model was employed to identify factors determining the status of incomplete immunization. Results Overall, in Ethiopia, more than half (54%, 95% CI: 48–58%) of children aged 12–23 months were not fully immunized. The spatial analysis revealed that the distribution of incomplete immunization was highly clustered in certain areas of Ethiopia (Z-score value = 8.379419, p -value 0.001). Hotspot areas of incomplete immunization were observed in the Afar, Somali, and southwestern parts of Ethiopia. The SaTScan spatial analysis detected a total of 55 statistically significant clusters of incomplete immunization, with the primary SaTScan cluster found in the Afar region (zones 1, 3, and 4), and the most likely secondary clusters detected in Jarar, Doola, Korahe, Shabelle, Nogob, and Afdar administrative zones of the Somali region of Ethiopia. Indeed, in the multilevel mixed-effect logistic regression analysis, the respondent's age (AOR: 0.92; 95% CI: 0.86–0.98), residence (AOR: 3.11, 95% CI: 1.36–7.14), living in a pastoralist region (AOR: 3.41; 95% CI: 1.29–9.00), educational status (AOR: 0.26; 95% CI: 0.08–0.88), place of delivery (AOR: 2.44; 95% CI: 1.15–5.16), and having PNC utilization status (AOR: 2.70; 95% CI: 1.4–5.29) were identified as significant predictors of incomplete immunization. Conclusion and recommendation In Ethiopia, incomplete immunization is not randomly distributed. Various factors at both individual and community levels significantly influence childhood immunization status in the country. It is crucial to reduce disparities in socio-demographic status through enhanced collaboration across multiple sectors and by bolstering the utilization of maternal health care services. This requires concerted efforts from stakeholders.