3.1. Introduction of satellite-based products and applications to polio vaccination campaign planning

Cases began to increase once again in 2011, and a number of factors were identified that were hindering progress, including security issues and lack of strong government support (Nasir et al., Reference Nasir, Bandyopadhyay, Montagnani, Akite, Mungu, Uche and Ismaila2016). In addition to these challenges, there was a recognition that more fundamental challenges—including inaccurate maps and lack of tracking and accountability—were also contributing to these difficulties and resulting in children being chronically missed by vaccinators. In December 2011, the National Primary Health Care Development Agency of Nigeria, along with GPEI partners and the Bill and Melinda Gates Foundation, funded a project to use satellite data to create new maps for eight Nigerian states. Key aspects of the project are summarized here, and a full technical description can be found in Barau et al. (Reference Barau, Zubairu, Mwanza and Seaman2014).

Prior to this project, the vaccination teams had relied primarily on hand-drawn maps. This was necessary because official maps did not typically include boundaries for wards, the administrative level at which polio vaccination planning occurs. Unfortunately, these hand-drawn maps were often inaccurate and incomplete. They may be missing hamlets or the homes of nomadic individuals or have existing settlements placed inaccurately relative to each other. These missing settlements are particularly problematic, because the method used to validate the level of coverage achieved—lot-quality assurance sampling methodology (LQAS)—relied on the same maps as those used by vaccination teams. This means settlements missed in the initial campaign would also be missed when conducting post-campaign evaluations (Barau et al., Reference Barau, Zubairu, Mwanza and Seaman2014). Due to this situation, the vaccination coverage levels determined through this method would represent only vaccination coverage for the known population depicted on the hand-drawn maps and visited by vaccinators, and not the true vaccination coverage for the total population.

New geospatial information system (GIS) maps were developed using recent high-resolution satellite imagery collected by DigitalGlobe (now Maxar). Researchers used manual feature extraction—clear, pre-determined definitions of various types of settlements—to identify hamlets, small settlements, and built-up areas. Field teams using GPS-enabled phones added the settlement names and locations. The draft maps were then reviewed by ward leaders before being finalized. This resulted in more accurate, standardized, and geo-referenced maps. Initial use of the maps began in limited areas in mid-2012, and by December 2013, the GIS maps were distributed for use in all eight states, increasing efficiency and adding more than 3,000 previously unmapped settlements (Barau et al., Reference Barau, Zubairu, Mwanza and Seaman2014).

In addition to identifying previously missing settlements, the satellite-based GIS maps allowed for GPS tracking of the vaccination teams during the campaigns themselves. Each team carried a smartphone that recorded a GPS location with a date and time stamp every 2 min. At the end of each day, the resulting track was uploaded to a central computer and reviewed to determine whether the teams had successfully visited all assigned settlements. Any missed settlements were added to a list to be visited on a “mop up” day at the end of the four-day vaccination campaign. Visits on that day were also tracked, and additional mop-ups were added, if necessary (Barau et al., Reference Barau, Zubairu, Mwanza and Seaman2014). In 2011, before the first satellite-based maps were created, the vaccination coverage rate in Kano state, as determined by LQAS, was typically 60% or lower. Given that a vaccination coverage rate of 80% is needed to halt transmission, this is a significant concern.Footnote ³ In 2013, after the use of satellite maps and GPS tracking had been fully implemented, the average vaccination coverage rate in Kano measured through the LQAS process was about 83%, on average, and by 2014 it rose to 89% (Wannemueler, Reference Wannemueler2012).

A second satellite data-enabled project ran in parallel with these efforts. In addition to requiring knowledge of exactly where populations are located, officials planning vaccination campaigns must have an accurate estimate of the total population. This is necessary for ensuring adequate supplies, personnel, and time. When the goal is eradication, as is the case with the polio campaigns, accurate assessment of the total population is also essential for determining whether the percentage of the population that has been vaccinated is sufficient to provide herd immunity (Weber et al., Reference Weber, Seaman, Stewart, Bird, Tatem, McKee, Bhaduri, Moehl and Reith2018).

One traditional source for these estimates is data collected in the national census. In the case of Nigeria, the most recent national census data comes from 2006. Estimates of population since 2006 typically rely on the application of constant growth rates that do not take into account differential changes in different geographic areas (Weber et al., Reference Weber, Seaman, Stewart, Bird, Tatem, McKee, Bhaduri, Moehl and Reith2018).

As an alternative, researchers developed a novel census-independent method of estimating population using semi-automated analysis of satellite data. The full technical methodology is described by Weber et al. (Reference Weber, Seaman, Stewart, Bird, Tatem, McKee, Bhaduri, Moehl and Reith2018). The project used recent high-resolution satellite imagery of two Nigerian states—Kano and Kaduna—procured from DigitalGlobe (now Maxar) and Airbus. Analysts reviewed the imagery to identify settlements and classify settlement types. They trained a machine-learning model to analyze the imagery and validated the results with fieldwork (Weber et al., Reference Weber, Seaman, Stewart, Bird, Tatem, McKee, Bhaduri, Moehl and Reith2018).

Weber et al. (Reference Weber, Seaman, Stewart, Bird, Tatem, McKee, Bhaduri, Moehl and Reith2018) project demonstrated that the census-based population estimates had significantly underestimated the total population in the study areas. For example, in the Kano metropolitan area, the satellite-based estimates were higher than the census-based extrapolations by more than one million people. Comparing satellite data from 2006 to 2014 showed that the built-up area had increased at 4.37% per year, while UN census-based estimates had assumed annual population growth of just 2.1%.

These satellite-based products allowed critical improvements in both the accuracy of estimates of population size and location (and thus the accuracy of vaccination coverage) as well as the efficacy of campaigns. Satellite maps allowed experts to identify settlements that would otherwise have been missed, and satellite data enabled more accurate estimates of overall growth and population size in key areas relevant to the polio vaccination effort. In addition, GPS tracking made possible by the new GIS maps ensured that settlements were actually visited by vaccinators during the campaigns.

3.2. Focus on Kano state

Our analysis of population estimates and vaccination coverage focuses on Kano state. Kano is the historical epicenter of polio in Nigeria and has long been a focus of GPEI efforts (World Health Organization, 2014). In early 2014, the Nigerian state of Kano was considered a key battleground in the fight for eradication of polio in Nigeria. Many campaigns had been conducted in Kano in the prior years, but those campaigns had failed to interrupt transmission of poliovirus (Weber et al., Reference Weber, Seaman, Stewart, Bird, Tatem, McKee, Bhaduri, Moehl and Reith2018). Kano was one of the eight states for which satellite-based maps were first created, and it is one of the examples that Weber et al. (Reference Weber, Seaman, Stewart, Bird, Tatem, McKee, Bhaduri, Moehl and Reith2018) highlighted in their work.

Since the population estimates are associated with Kano state, the increase in the likelihood of halting transmission that we calculate is also specific to Kano state. However, we argue that because Kano state was so central to Nigerian eradication efforts, it is reasonable to use this percentage to represent the likelihood of halting transmission of polio in Nigeria as a whole. Thus, estimates of vaccination campaign costs and health benefits are computed for all of Nigeria, not just Kano state. Essentially, we are arguing that if you can successfully halt transmission in Kano, you can halt transmission for the country as a whole.

We recognize this is not a perfect assumption. In reality, after nearly 2 years with no polio cases detected, four children in the Nigerian state of Borno were determined to have wild poliovirus, indicating that the interruption of transmission was incomplete in 2014. The appearance of the disease in Borno was widely attributed to security challenges that prevented access to children in that region, which was controlled by the violent terrorist group Boko Haram (Roberts, Reference Roberts2016). However, following this event, no further cases were detected (in Kano, Borno, or anywhere else in Nigeria), and Nigeria was officially certified as free of wild polio in 2020 (Guglielmi, Reference Guglielmi2020). Also of note, strategies such as security-escorted vaccinator visits and rapid campaigns to enable vaccination efforts in terrorist-held areas also relied on satellite-based maps and GPS tracking to ensure efficient access to the target population (Adamu et al., Reference Adamu, Archer, Braka, Damisa, Siddique, Baig, Higgins, Sume, Banda, Korir, Waziri, Gidado, Bammeke, Edukugo, Nganda, Forbi, Burns, Liu, Jorba, Asekun, Franka, Wassilak and Bolu2019). Researchers developed new satellite-based methods to distinguish between inhabited settlements and those that had been abandoned due to violence, increasing the efficiency and safety of vaccination efforts in insecure areas (Higgins et al., Reference Higgins, Adamu, Adewara, Aladeshawe, Aregay, Barau, Berens, Bolu, Dutton, Iduma, Jones, Kaplan, Meleh, Musa, wa Nganda, Seaman, Sud, Vouillamoz and Wiesen2019).

4.1. Population estimates and vaccination coverage

As noted above, an accurate estimate of the population is a critical element to accurately determining the true vaccination coverage during a campaign. Since polio vaccination campaigns target children under 5 years old, the population estimates discussed here are for that group (Table 1).

Table 1. Definition of key variables and their values

4.2. Campaign Cost Savings

The socioeconomic benefits of halting transmission of polio in 2014 are directly related to the costs of maintaining the large-scale vaccination campaign efforts necessary to halt transmission in a country where polio is endemic—in this case, Nigeria. Therefore, in this section, we identify and enumerate those costs. GPEI organizes polio eradication costs into four categories: immunization activities (e.g., the vaccine campaigns themselves), surveillance and response capacity, containment and certification, and core functions and infrastructure.

The total cost of immunization activities is heavily influenced by the number and quality of vaccination campaigns (Global Polio Eradication Initiative, 2014). Each of these campaigns is complex and expensive. The goal of a polio vaccination campaign is to give two drops of oral polio vaccine to every child under 5 years old residing in the target areas over the course of the four-day campaign (Barau et al., Reference Barau, Zubairu, Mwanza and Seaman2014). To enable this, the program must purchase a sufficient number of oral polio vaccines and cover operational costs. They must also account for campaign-related social mobilization costs and communication efforts to ensure there are high levels of awareness and demand for the vaccine within the community (Global Polio Eradication Initiative, 2014).

Planning for these campaigns requires significant operational and managerial efforts to facilitate national, international, and local coordination, supervision, and performance evaluation. Carrying out the campaign requires a large number of vaccinators, engagement of local community-based educators, and technical assistance from experts. These individuals require access to adequate transportation and communication capabilities to reach inaccessible and sometimes insecure areas, as well as adequate vaccines and cold-chain storage (Nasir et al., Reference Nasir, Bandyopadhyay, Montagnani, Akite, Mungu, Uche and Ismaila2016; Vaz et al., Reference Vaz, Mkanda, Nsubuga, Ado and Etsano2016). In Nigeria, each campaign involves visiting hundreds of wards and managing thousands of vaccination teams operating during each campaign (Barau et al., Reference Barau, Zubairu, Mwanza and Seaman2014).

In addition to the campaigns themselves, funding is also required to support surveillance and emergency response. This includes a surveillance network to detect and investigate acute flaccid paralysis cases and maintain laboratories able to do this analysis. Emergency response efforts are undertaken in response to outbreaks, and thus are especially relevant in countries where polio is still endemic. Emergency response is also required when outbreaks occur in countries where polio is not endemic, but is imported from an endemic country (Global Polio Eradication Initiative, 2014). Accordingly, funds needed to support these activities will decrease when transmission of polio is halted in an endemic country like Nigeria.

Containment and certification of wild poliovirus require not only highly sensitive surveillance, but also full application of bio-containment requirements across the entire world. These efforts continue even in regions of the world certified to be polio-free. GPEI also supports countries with their core infrastructure, including technical assistance, efforts to improve the efficacy of immunization campaigns, community engagement, and research and development. In some high-priority countries, this also involves supporting efforts to improve routine immunization systems.

GPEI provides annual expenditure documents with high-level summaries of expenditures for the GPEI program as a whole. Table 2 shows total annual expenditures as well as the portion of expenditures spent directly on immunization activities. Immunization activities represent a very large proportion of overall GPEI expenditures, although spending in this area decreased notably after Nigeria halted transmission of wild poliovirus in 2014.

Table 2. GPEI annual expenditures (US$ millions)

^a The organization of costs in the annual expenditure report changed in 2016 compared to earlier years. This number represents the cost of “campaigns” and “other immunization activities.”

Financial Resource Requirements (FRR) reports (GPEI’s budget planning documents) from the period 2013–2018 were released approximately bi-annually from 2012 to 2014 and provide significantly greater detail, including a breakdown of costs in Nigeria. These documents show that costs associated with GPEI activities in endemic countries account for nearly half of the total costs for the program, with efforts in Nigeria being particularly costly. As of 2014, GPEI efforts in Nigeria alone were expected to account for more than a quarter of all GPEI costs from 2013 to 2018. Estimated total costs for GPEI activities in Nigeria based on these documents are shown in Table 3. As of 2017, GPEI moved to an interactive online tool to provide budget information and the categories used to break down funding areas changed (Global Polio Eradication Initiative, 2020a). Total budget estimates for Nigeria for 2017–2020 based on this tool are shown in Table 4.

Table 3. GPEI budget estimates for Nigeria 2012–2017 (US$ millions)

Table 4. GPEI budget estimates for Nigeria 2017–2020 (US$ millions)

Note. These budget estimates are based on the GPEI interactive online budgeting tool (Global Polio Eradication Initiative, 2020b).

As noted above, these costs are significantly impacted by the number and extent of vaccination campaigns required, which decreases when interruption of transmission is achieved. These charts show that while costs do not go down immediately, due to the need to continue intense immunization efforts until the elimination of wild poliovirus is confirmed, costs decrease significantly within a few years. Funding needed during these “ramp down years” would remain similar regardless of the year transmission was halted. Therefore, for the purpose of this analysis, we will use 2014 and 2020 as representative years.

If transmission of wild poliovirus had not been achieved in 2014, then GPEI could have expected to face an additional year with costs of approximately $281.18 million in place of an additional year with low-cost, post-certification costs similar to those faced in 2020, of approximately $110.331 million. Therefore, the cost of not halting transmission in 2014 is estimated at $170.85 million.

4.3. Improved health

Perhaps the most salient benefit of halting transmission of polio is the decrease in the number of children that become ill with the disease. We look at both wild polio cases in Nigeria as well as instances of outbreak polio—cases outside Nigeria caused by ongoing polio in Nigeria (Table 5).Footnote ⁵ Based on this data, we make rough estimates of the number of polio cases that may have occurred if satellite-based maps and population estimates had not been used. We also translate this reduction of disease into socioeconomic costs using widely-accepted values from the global health literature.

Table 5. Polio cases in Nigeria 2011–2020

Source: Progress Toward Poliomyelitis Eradication—Nigeria, January 2014–July 2015.

4.4. Cost of generating satellite-based estimates

Finally, we account for the costs of generating satellite-based estimates. Gates Foundation officials estimated the top-line number for this activity at approximately $5.75 million. This estimate includes not only the cost of the satellite data itself (about $1.5 million), but also the effort required to analyze the data and create the data products, as well as validation activities on the ground in Nigeria and training and capacity-building activities to enable local actors to use the products (Bill and Melinda Gates Foundation, 2021). While some additional funds have been spent to maintain, update, and improve these capabilities, the maps themselves as well as the approach developed can be, and has been, leveraged and applied in additional situations and products for significantly lower cost. The conclusion section identifies some of the broader applications that have come from this work.

5.1. Calculating likelihood of halting transmission

Transmission of a disease is halted when sufficient vaccination coverage—the number of people vaccinated divided by the total population eligible—is achieved. This is the threshold vaccination coverage (V _th) required. In practice, programs aim to achieve full coverage—vaccinating every eligible child. However, for a variety of reasons—for example, children that are not home, not feeling well, or otherwise unavailable when vaccinators visit—the actual vaccination rate is lower than the ideal. The vaccination coverage achieved by a campaign is determined through a statistical sampling method undertaken by a validation team after the campaign has been completed, and coverage levels measured by these teams can differ across vaccination campaigns over time (World Health Organization, 2013). To account for this, we include the vaccination coverage rate measured both before satellite-based maps were introduced V _{measured-nosat} and after satellite-based maps were introduced, V _measured-sat.

As noted previously, because the validation process relies on the same maps and estimates as the vaccination campaign itself, the vaccination coverage estimated using this method only applies to the known (or estimated) population, not the true population. Therefore, the number of people vaccinated as part of the campaign can be calculated with the following straightforward formula:

$$ \left(\mathrm{Number}\ \mathrm{of}\ \mathrm{People}\ \mathrm{Vaccinated}\right)=\left(\mathrm{Measured}\ \mathrm{Vaccination}\ \mathrm{Coverage}\right)\ast \left(\mathrm{Population}\ \mathrm{Estimate}\right) $$

This formula results in different estimates for the number of people vaccinated as part of the campaign depending on the estimate of the population. If the census-based population estimate (P _c) is used, then the number of people vaccinated would be P _c*V _{measured-nosat}. If the satellite-based population estimate (P _s) is used, then the number of people vaccinated would be P _s*V _measured-sat.

To determine how likely either of these efforts would be to halt the transmission of polio, we need to know how they compare to the true population (P _true) multiplied by the threshold vaccination coverage required to halt transmission (V _th). This can be determined by examining where the satellite and non-satellite cases fall along the probability distribution of the true number of people that needed to be vaccinated in order to halt transmission of polio in Nigeria (P _true*V _th). This is represented in Figure 1.

Figure 1. Using the population probability distribution to determine likelihood of eradicating disease.

Figure 1 shows the number of people that would have been vaccinated under a census-based estimate (P _c) and a measured vaccination coverage rate associated with a no-satellite data scenario (V _{measured-nosat}). The area under the curve to the left of P _c*V _{measured-nosat}, x%, represents the likelihood that the true population was sufficiently small that disease spread would have been interrupted even if a census-based estimate had been used and a no-satellite vaccination coverage rate was achieved.

Figure 1 also shows the number of people that would have been vaccinated if a satellite-based estimate (P _s) was used and the measured vaccination coverage rate aligned with that seen in the satellite data scenario (V _measured-sat). The area under the curve to the left of P _s*V _measured-sat, y%, represents the likelihood that disease spread would have been interrupted under the scenario in which satellite data was used to plan and carry out the campaign. While we know in reality, the spread of polio was halted under the scenario in which satellite data was used, this allows us to acknowledge that this result was not guaranteed. Residual uncertainty in the satellite-based estimates of population could have meant that even under this scenario, the true population could have been even larger than estimated, such that the vaccination threshold still would not have been met.

Once we have these two estimates, we can determine the increase in the likelihood of interrupting transmission of polio by using satellite-based maps and population estimates rather than hand-drawn maps and census-based estimates as simply y%–x%. We can then multiply this estimate of the percentage increase in the likelihood of halting transmission of polio associated with the use of satellite data compared to census data to determine the proportion by the socioeconomic benefits associated with halting transmission of polio that can be attributed to satellite data, giving an expected value of the use of satellite data in this case.

5.2. Calculation of net socioeconomic benefits

The socioeconomic benefits fall into two categories. The first is cost-savings to the health system. Intensive vaccination and surveillance efforts are expensive in terms of human effort and materials, and if these can be completed in a shorter period of time, significant savings are possible. The second outcome is better health for individuals within the population. If disease transmission is halted earlier, fewer people get infected. In the case of polio, this means fewer children are paralyzed. The exact values for these variables are discussed in the data section above.

The socioeconomic benefits of using satellite-based population estimates (or costs of using traditional census-based estimates) would continue to grow with each additional year that transmission continues. In reality, program officials would likely adapt their approach and adopt new methods for planning, so we calculate a conservative estimate of benefits based on just 1 year of additional disease transmission.

Therefore, the expected value of socioeconomic benefits associated with the use of satellite data in this case is calculated as the cost savings and health impacts associated with 1 year of continued transmission, adjusted to account for the possibility that transmission would have been interrupted in 2015, even using traditional methods (our counterfactual).

We also take into account the costs associated with the development of the satellite-based maps and estimates of population. Thus the net socioeconomic benefit of satellite data in the Nigeria polio eradication effort is given by the following formula:

$$ {\displaystyle \begin{array}{c}\left(\mathrm{Net}\ \mathrm{socioeconomic}\ \mathrm{b}\mathrm{e}\mathrm{nefits}\right)=\left(y\%-x\%\right) \ast \left(\mathrm{total}\ \mathrm{economic}\ \mathrm{b}\mathrm{e}\mathrm{nefits}\right)\\ {}\hskip26.7em -\left(\mathrm{cost}\ \mathrm{of}\ \mathrm{generating}\ \mathrm{satellit}\mathrm{e}\hbox{-} \mathrm{b}\mathrm{ased}\ \mathrm{products}\right)\end{array}} $$

6.1. Probability of halting transmission

Using the probability distribution of the under-five population of Kano state provided by GRID3, we examine the probability of halting transmission of polio (i.e., achieving 80% vaccination coverage or more) under various scenarios. Using the mean satellite population estimate and the measured vaccination coverage achieved in 2014 of 89%, the probability of halting transmission is 91%—this percentage corresponds to y% in Figure 1. Using either the census-based population estimate or the estimate based on hand-drawn maps, if the measured vaccination coverage was 60%, the distribution suggested there was zero chance of halting transmission. This is not surprising, as both of these population estimates likely underestimate the true population and the vaccination coverage rate is lower than that needed to halt transmission. If we assume that even without the satellite-based maps, vaccine campaigns would have achieved 89% coverage, then the probability of halting transmission with census-based estimates would have been 61%—this percentage corresponds to x% in Figure 1. Overall, this suggests that the satellite-based maps and population estimates increased the likelihood of halting transmission in the range of 30% to 91%, a range generated by comparing the difference between satellite-based population estimates and any of the other potential counterfactuals (Table 6).Footnote ⁶

Table 6. Probability of halting transmission under different scenarios

6.2. Socioeconomic benefits

The estimated benefits of halting transmission of polio in 2014 associated with vaccine campaign and other programmatic cost savings were determined to be $170,850,000. The socioeconomic benefits associated with the decrease in wild poliovirus cases resulting from halting transmission in 2014 were estimated to fall in the range of $162,000 to $3,294,000, and those associated with the decrease in outbreak polio were $1,350,000. Added together, this suggests a total socioeconomic benefit of between $172,362,000 and $175,494,000.

Cost savings related to campaigns are significantly higher than the benefits from reduced health impacts. As noted above, this is due to the fact that there are typically very few Polio cases in the years immediately preceding a halt in transmission. Expending large sums to continue vaccination campaigns when only a handful of cases occur each year may seem counterintuitive. However, other research shows that the long-term benefits of Polio eradication make this expenditure worthwhile. While efforts to control polio would be less expensive than eradication in the first few years, the cumulative costs would quickly overtake those of eradication (Global Polio Eradication Initiative, 2013a).

6.3. Total socioeconomic value of using satellite data for polio eradication in Nigeria

For this analysis, we want to take into account the fact that the use of satellite maps was not associated with certainty in halting transmission and that there was a possibility that the transmission of wild poliovirus would have been halted in 2014 even if satellite data had not been used. Accordingly, the value attributed to the use of satellite data is between 30% and 91% of these values, resulting in an expected value for these benefits between $51.7 million and $159.7 million. Lastly, we factor in the cost of generating and using the satellite-based maps of $5.75 million. Considering all of these factors, we estimate that the net socioeconomic benefit associated with the use of satellite data to assist in halting the transmission of polio in Nigeria in 2014 is between $46.0 million and $153.9 million.

6.4. Robustness of results: map-based population estimate

We also undertake a second approach to calculating the difference between the satellite-based population estimates and census-based population estimates, due to the specific decision-making processes involved in the polio vaccination processes in Nigeria. Specifically, although census-based estimates may be used for high-level planning, for the planning of individual campaigns—determining how many doses of vaccine to purchase and how many vaccination teams to employ—the GPEI program often uses estimates based on experience from previous years, specifically, the number of children that were reported to have been encountered by vaccination teams in the previous year adjusted to assume some growth. Over time, it appears that these estimates became significantly inflated, so that the population estimates used for planning were significantly larger than the true population. For example, in January 2014, the tally sheet summary stated that 4,270,792 children under five were vaccinated as part of that month’s campaign in Kano state (SIPDs, 2014). Eventually, due to a rigorous supervised enumeration completed by March 2014, combined with the satellite-based estimates, the tally-sheet estimates decreased (Seaman, Reference Seamann.d.). The tally-sheet for the December 2014 campaign suggests that 2,570,295 children under five were vaccinated during that effort (GPEI Nigeria TallySheet Data Summary by State, 2014).

Paradoxically, this means that while the satellite-based maps and population estimates resulted in new settlements being discovered and more individuals being vaccinated, as well as significant increases in total population estimates compared to census-based estimates, they also resulted in a reduction of the population estimates used most directly in vaccination campaign planning (Seaman, Reference Seaman2021a).Footnote ⁷ To account for this situation—that an artificially inflated over-estimate of population was used for campaign planning while at the same time that the population being served during the campaigns was lower than the true population (because whole settlements were not included in planning maps and thus not being visited)—we test one additional method for estimating the total population included in polio vaccination campaigns prior to the creation of the satellite-data maps. This method takes into account an estimate of the proportion of the total population that was present in the missing settlements.Footnote ⁸ Researchers estimated that approximately 5% of all settlements were chronically missed before the satellite-based maps and GPS-tracking systems were implemented (Seaman, Reference Seaman2021b). This would suggest a total population for Kano State of 13,038,041 and a total for the under-5 population of 2,325,919.Footnote ⁹ If we use the assumption that the 89% coverage rate would have been achieved even without the use of satellite data and maps, then the likelihood of halting transmission in this scenario (x% in Figure 1) is 30%. This use of satellite data provides a 61% increase in the likelihood of halting transmission compared to this scenario, and the net economic benefit is at least $105.1 million. This falls within the range produced by our original estimate.