*2.4. Ethical Approval and Reporting Guidelines*

Data were not obtained from subjects for the Global Burden of Diseases, Injuries, and Risk Factors Study or related analyses such as this study. Instead, we used preexisting, publicly available, de-identified datasets that include, but are not limited to, administrative and survey-based vaccine coverage reports. Data were identified through online searches, outreach to institutions that hold relevant data such as ministries of health, or individual collaborator references and identification. Most of the data used are publicly available. Therefore, informed consent was not required. This study was approved by the University of Washington's Human Subjects Division Study ID: STUDY00009060. Our

study follows the Guidelines for Accurate and Transparent Health Estimates Reporting (GATHER; Supplemental Table S4).

#### **3. Results**

*3.1. Mapping Overlap*

3.1.1. Country-Specific Overlap by Prevalence

Figure 3A shows an example bivariate map that illustrates the spatial overlap between the population-weighted quartile classifications for no-DTP and stunting in Nigeria, based on the prevalence of each indicator. In this example, when categorizing by prevalence within Nigeria, based on the available geospatial estimates for both no-DTP and stunting, higher-prevalence districts tended to be more widely distributed through the northern regions of the country, while the southern regions had a lower prevalence for both indicators. Overlap between no-DTP and stunting categorization was high; nearly two thirds of all districts in Nigeria (488 of 774 districts, or 63.0%) were designated to the same population-weighted quartile for both no-DTP and stunting. Figure 4A shows a simplified representation of the same analysis, restricting the mapped districts to only the high-quartile areas for each indicator. Of the 207 districts in the highest quartiles for either no-DTP or stunting, half of those districts (49.0%, or 100 of 207 total) were in the highest quartile for both indicators.

**Figure 3.** Country-specific Nigeria overlap for no-DTP and stunting, for all categorization quartiles. The top row (**A**) shows categorization based on prevalence, while the bottom row (**B**) shows categorization based on counts. Population-weighted quartile ranges for no-DTP and children with stunting are delineated in the bivariate color legends (center). District-level values are shown both as maps (left) and with scatterplots (right), with colors corresponding to quartile legend values. Point size in scatterplots reflects relative size of under-1 population in each district.

**Figure 4.** Country-specific overlap between no-DTP and stunting in Nigeria for highest category quartiles only. Districts in red are in the highest quartile for no-DTP only, blue are in the highest quartile for children with stunting only, and purple are in the highest quartile for both indicators. The map on the left (**A**) shows categorization based on prevalence, and the map on the right (**B**) shows categorization based on counts.

The spatial overlap between health indicators varies from indicator to indicator and country to country (Supplemental Figures S24–S69). In Ethiopia, for example, the locations with the highest no-DTP prevalence are located primarily in the east and south of the country (especially in the Afar and Somali regions) and are distinct from those with the lowest ITN coverage, which are located more centrally (for instance, in Amhara and Oromia) (Supplemental Figures S40a and S63a). In the Democratic Republic of the Congo, the geographic overlap between under-5 mortality and no-DTP prevalence is highly heterogeneous, with a mixture of high-prevalence areas for no-DTP, U5M, both, and neither indicator (Supplemental Figures S34a and S57a).

#### 3.1.2. Country-Specific Overlap by Counts

As expected, when these same analyses are repeated using an example categorization approach based on counts rather than prevalence, the results tend to emphasize areas of large populations—although this pattern is not universal across indicators and countries.

For the overlap between no-DTP and stunting in Nigeria, for example, when categorization is based on counts rather than prevalence, higher-quartile districts still tended to be in the northern regions of the country, while southerly districts tended to be in the lower quartiles (Figure 3B). Compared to the prevalence-based approach, there was more concordance between count-based classifications, with more than three fourths of all districts (597 of 774 districts, or 77.1%) being designated to the same quartiles for both no-DTP and stunting. Fewer districts were classified into the highest quartiles for either metric based on counts compared to prevalence (125 vs. 207 districts), but a greater proportion were in the highest quartile for no-DTP and stunting (78 of 125 districts, or 62.4%). There were 44 districts categorized in the highest quartile for both no-DTP and stunting according to both prevalence and counts (Figure 4).

However, these patterns again varied between countries and indicators (Supplemental Figures S24–S69). For Indonesia, for instance, locations that might be targeted for joint targeting based on spatial overlap between no-DTP and missed ORT would vary broadly depending on whether decisions were informed by analyses of prevalence or counts (Supplemental Figures S45 and S68). In Angola, prevalence-based analysis of the overlap between no-DTP and ITN use identifies broad areas of the country that is potentially amenable to joint targeting (Supplemental Figures S31a and S53a). Due to the population distribution in the country, however, count-based analysis suggests that joint targeting

opportunities might be focused upon relatively few locations (Supplemental Figures S31b and S53b).

#### 3.1.3. Overlap for Multiple Indicators

For some stakeholders, it may be of interest not only to understand the degree of geographic overlap between no-DTP and other health indicators individually, but also to identify locations that may be amenable to integrated intervention across many indicators. We, therefore, produced country-specific maps that show the number of health indicators in the highest quartile in each district, using our population-weighted classification approach. Here, we continue to show results from Nigeria as an example, although results for other countries can be found in Supplemental Figures S70–S73.

According to both prevalence and counts, more indicators were classified in the highest quartile in northern and northwestern Nigeria (Figure 5). Districts in southern Nigeria were largely only in the highest quartile for one to two indicators (missed ITNs and/or LF), whereas districts in northern and northwestern Nigeria had many cases of the overlapping classification for no-DTP, stunting, U5M, and missed ORT.

**Figure 5.** Country-specific multi-indicator overlap for Nigeria. Color given in each district reflects the number of indicators assigned to the highest quartile in that district. Districts outlined in white indicate those where no-DTP is among the indicators in the highest quartile. The map on the left (**A**) shows categorization based on prevalence, and the map on the right (**B**) shows categorization based on counts.

When classified by prevalence, high-quartile districts were relatively more concentrated across indicators in Nigeria compared to other countries (Figure 5A, Supplemental Figures S70–S73). More than two thirds of districts in Nigeria were categorized into the highest quartile for at least one of the six indicators analyzed (525 of 774, or 67.8%), but these proportions were even greater in all other countries, including 74.8% of districts in Indonesia (374 of 500), 81.1% of districts in DRC (194 of 239), 81.0% of districts in Ethiopia (64 of 79), and 87.7% of districts in Angola (143 of 163).

The opposite was true when categorizing the districts into population-weighted quartiles by counts. In this example, a much smaller proportion of districts—43.7%—were in the highest quartile for at least one indicator in Nigeria (338 of 774 districts; Figure 4B). This trend was consistent across other countries (Supplemental Figures S70–S73).

#### 3.1.4. Multinational Overlap by Prevalence

The analyses above focus on describing the spatial patterns of no-DTP and other indicators, based upon within-country classification for each indicator. For global or regional decision-makers, however, examination of the degree of spatial overlap across

countries may be of interest. We, therefore, repeated these analyses, but instead categorized districts as those with the highest prevalence or counts for each indicator across all five example countries included in these analyses, rather than within the countries separately. Given the limited number of countries and indicators used in this analysis, these example results are meant to be illustrative only, to demonstrate the magnitude of differences in the perceived overlap when looking across rather than between countries and are not meant as policy recommendations.

Categorizing by prevalence across our five focal countries combined, populationweighted quartile assignments for no-DTP and stunting were markedly similar (Figure 6). The quartile classifications for no-DTP and stunting exactly matched (i.e., districts in the lowest quartile for no-DTP were also in the lowest quartile for stunting, etc.) in nearly half of all districts (795 out of 1755 total; 45.3%). Districts in the highest quartile across all countries for no-DTP could be found in every country, as well as districts in the highest quartile for stunting (Figure 7). The districts where the highest-quartile categorization for no-DTP and stunting overlapped largely fell within Nigeria and Angola, with 27.0% of districts in Nigeria and 50.3% of districts in Angola being in the highest category for both indicators (209 of 774 in Nigeria and 82 of 163 districts in Angola). While significant portions of DRC and Ethiopia were in the highest quartile for one indicator or the other, there was little overlap between indicators in these countries, and none in Indonesia (Figure 7).

**Figure 6.** Multinational overlap between stunting and no-DTP for all categorization quartiles, based on prevalence. Ranges for population-weighted quartiles across the five example countries combined for no-DTP and children with stunting are delineated in the bivariate color legend (bottom right).

Different patterns were observed for other comparator indicators (Supplemental Figures S74–S81). For instance, when comparing categorization for no-DTP and ORT across all five countries to that for no-DTP and stunting, fewer districts in Nigeria and Angola were in the highest quartile for both indicators, whereas larger areas of Ethiopia and DRC were in the highest quartile for both (Supplemental Figure S80).

**Figure 7.** Multinational overlap between stunting and no-DTP for highest quartiles only, based on prevalence. Districts in red are in the highest quartile for no-DTP only, blue are in the highest quartile for children with stunting only, and purple are in the highest quartile for both indicators.

#### 3.1.5. Multinational Overlap by Counts

Categorization at the multinational scale was even more closely aligned between no-DTP and stunting when classifying districts according to counts rather than prevalence (Figure 8). When categorizing by counts, quartile assignment matched exactly between no-DTP and stunting in 71.1% of all districts (1248 of 1755 total). Far fewer districts were in the highest quartile when considered in terms of counts—only 5.8% of districts were in the highest quartile for either indicator using counts, compared to 35.5% of districts when considered by prevalence (102 vs. 623 out of 1755 districts, respectively; Figures 8 and 9). In addition, the highest-quartile districts for either indicator fell largely in Ethiopia and DRC. The highest-quartile districts were scarce in the other three countries, making up <3% each for districts in Nigeria, Angola, and Indonesia. Only 49 districts were in the highest quartile for both no-DTP and stunting (2.7% of all districts), and more than half (28 of 49) were found in Ethiopia. This trend was largely consistent across all indicators (Supplemental Figures S82–S89).
