*2.2. Analysis*

We conducted three analyses to characterize potential exemplars in reducing zero-dose children over time, as summarized below. R version 4.2.1 was used for data processing, analyses, and visualizations [42].

**Quantifying changes in zero-dose children across geographies.** We assessed changes in the percentage of under-one children without any doses of DTP (no-DTP) between 2000 and 2019 across two geographic dimensions: (1) national levels; (2) subnational gaps among second-level administrative units. For the latter—subnational gaps—we used the 5th and 95th percentile values of the prevalence of no-DTP children estimated across second-level administrative units and computed the difference for a given country–year. We opted to use the 5th and 95th percentiles rather than absolute minimum and maximum values of no-DTP prevalence to offset the potential for undue influence of outliers for a given subnational unit–year. Furthermore, how countries define second-level administrative units widely varies (e.g., 10 or fewer units in Comoros, São Tomé and Príncipe, and Lesotho to 774 local government areas (LGAs) in Nigeria); using percentiles to define subnational gaps may also help mitigate the degree to which having more (or fewer) administrative units could affect measures of subnational inequality.

**Identifying potential exemplars in reducing zero-dose children.** Second, countries with the largest declines for *both* no-DTP metrics between 2000 and 2019 were considered as potential exemplars in reducing zero-dose children. Prior research conducted under the EGH program has typically used one progress measure per geographic unit [32,34,35], and then benchmarked changes against indicators of sociodemographic development. Because many locations with the highest levels of unvaccinated children face compounding vulnerabilities [4], any investments in reaching zero-dose children should also correspond with action to address disparities in immunization rates. Our approach to operationalizing this pro-equity lens from a geographic perspective was equally weighting reductions at the national level and subnational differences for no-DTP. In other words, a country that achieved marked national reductions in no-DTP prevalence without corresponding declines in subnational gaps should not be considered a potential exemplar in reducing zero-dose children.

We ranked each country ordinally, 1 to 56, based on their national and subnational reductions in no-DTP prevalence from 2000 to 2019, with 1 being the largest reduction and 56 being the smallest reduction or, if applicable, the largest increase since 2000. We took the mean of those rankings to identify which countries had achieved the most progress across both geographic dimensions. We applied these rankings and calculations for absolute and relative progress separately: computing percentage point changes for absolute progress from 2000 to 2019 and then percentage change from 2000 to 2019 for relative progress. We opted to consider both progress metrics—absolute and relative progress—as they could better represent a range of successful approaches used to reduce no-DTP prevalence from different starting points (i.e., higher and lower absolute levels of no-DTP children in 2000), and thus likely mirror different stages of immunization delivery needs and strategies.

**Comparing divergent no-DTP trajectories since 2000 for select locations**. Third, we conducted so-called "neighborhood analyses" for select countries, comparing them to other countries that had similar levels for both no-DTP measures in 2000 but different trajectories through 2019. Such analyses are thought to be supportive of potential crosslocation learning and knowledge translation around what could work to address zero-dose challenges when starting from similar baseline levels of no-DTP prevalence. At the countrylevel, we focused on Nigeria, Mali, Uganda, and Bangladesh—the four countries selected for the Gavi Learning Hubs [43] and sought to match a "neighbor" exemplar to each country. Further detail on the Gavi's Learning Hub initiative is available elsewhere [43]; in brief, these four countries were selected on the basis of zero-dose metrics (i.e., high absolute numbers or prevalence of zero-dose children) as well as variations in zero-dose prevalence across geographic locations and among key populations that experience higher rates of no vaccination (i.e., rural, urban poor, refugeed, or conflict settings). A primary objective of the Learning Hubs is to support deeper assessment and engagement to improve monitoring and measurement systems, and to enable learning about what works programmatically to reach unvaccinated children and missed communities.
