Characterizing Pentavalent Vaccine Dropout Estimates in Ethiopia Using Administrative and Survey Data

Abstract

Introduction: Vaccination is an important and effective approach to reducing under-five mortality from preventable diseases. Low- and middle-income countries like Ethiopia continue to face challenges in ensuring vaccination for vulnerable populations, with vaccine coverage varying by place of residence. One approach to understanding the vaccination-related performance of the health system is through assessment of vaccine dropout, which occurs when an individual starts but does not finish the required doses of a vaccine series such as the pentavalent vaccine. Evaluating data sources for obtaining this information is important. Methods: The two data sources used for this study were the 2019 mini-Demographic and Health Survey (DHS) and the District Health Information System 2 (DHIS2). In the survey, dropout was defined as receiving penta1 but not penta3. Dropout was calculated as the percentage difference between the number of children receiving pentavalent dose 1 and the number receiving dose 3 in DHIS2. Both zonal (n=100) and regional (n=14) level data were examined. Negative dropout was used to assess data quality. Pearson and Spearman rank correlations were used to compare dropout based on the DHS and DHIS2, estimated at both the zonal and regional levels. Results: For DHIS2, negative dropout was identified in 27 of 816 valid zone-year observations, corresponding to an overall percentage of 3.31%, with the highest number of impacted zones observed in 2024 and 2025. Negative dropout was not identified in the survey data. For the 2019 DHS, dropout ranged from 5.3% to 78% across the zones, with Addis Ababa having the lowest estimate and Afar and Somali the highest. In the DHIS2 data, across regions, Addis Ababa City Administration had the lowest penta dropout estimates, while Afar, Somlai, Gambella, Dire Dawa, and Benishangul-Gumuz generally had higher estimates. Across zones, dropout estimates varied widely over time, with the highest values and several negative dropout estimates observed in specific zone-years. Weighted Pearson and Spearman correlation at the regional level were 0.77 and 0.89, respectively. At the zonal level, the weighted Pearson correlation was 0.41, and the weighted Spearman correlation was 0.42. Conclusion: The results indicated that agreement between DHIS2 and DHS-based dropout estimates was stronger at the regional level compared to the zonal level. Overall, DHIS2 underestimated dropout compared with DHS-based estimates. As countries like Ethiopia increase their use of digital health systems, it is essential to ensure these tools provide reliable information for decision-making and vaccine program planning. To assess data quality, data triangulation between different sources and auditing may be conducted.