Heterogeneity in potential impact and cost-effectiveness of ETEC and Shigella vaccination in four sub-Saharan African countries

Diarrheal disease burden has become more heterogenous in low- and lower middle-income countries as access to clean water, sanitation and health care has increased in wealthier urban populations. Enterotoxigenic Escherichia coli (ETEC) and Shigella are among the top five causes of diarrheal mortality in children living in sub-Saharan Africa. Here, we explored how accounting for subnational and economic heterogeneity in ETEC and Shigella disease burden affects projected vaccine impact and cost-effectiveness of standalone ETEC and Shigella vaccines during the first decade after introduction in four sub-Saharan African countries. We developed dynamic models for provincial areas and socioeconomic subpopulations of children in the Democratic Republic of Congo (DRC), Kenya, Zambia, and Zimbabwe. We estimated deaths and morbidity due to ETEC and Shigella diarrhea plus additional deaths from other infectious diseases attributable to ETEC- and Shigella-induced stunting. We analyzed cost-effectiveness using Incremental Cost-Effectiveness Ratios (ICERs) with Disability-Adjusted Life Years (DALYs) and Moderate-and-Severe Diarrheal episodes (MSD) averted as denominators. Other infectious disease deaths due to induced stunting accounted for 9–28% and 9–32% of the total provincial level ETEC and Shigella disease burden, respectively, across these four countries from years 2025 to 2034. Our results indicated that the lowest and most cost-effective provincial DALYs averted ICERs were below $600 and $500/DALY averted for ETEC and Shigella vaccination, respectively in Zimbabwe. ICERs were the highest in Zambia and Kenya, where all provincial ICERs where above $2000/DALY. The highest national and provincial MSD averted ICERs were in DRC, while the lowest were in Kenya and Zimbabwe. Vaccinations were most cost-effective in averting DALYs in lower wealth subpopulations living in the highest burden provincial areas. Our approach focused on subnational heterogeneity in ETEC and Shigella burden and vaccination access found that impact and cost-effectiveness were more favorable if vaccinations reach the most vulnerable children in underserved provinces.


Introduction
While diarrheal mortality has declined 34% worldwide, it is still one of the leading causes of mortality for children under five [1]. However, diarrheal morbidity has declined at a much slower rate, only 10% from 2005 to 15 [1]. The majority of diarrheal disease burden is concentrated in underserved populations in low-and middle-income countries (LMICs) lacking access to maternal and child health services, safe water supply or sanitation facilities [2,3]. Enterotoxigenic Escherichia coli (ETEC) and Shigella are two of the leading enteropathogens causing diarrheal disease burden in children under 5 years of age globally and in sub-Saharan Africa [1,4].
Diarrheal disease severity varies by pathogen, causing a variety of symptoms beyond loose or watery stools that may result in long-term sequalae, such as childhood stunting [5,6], defined as height-for-age z-scores (HAZ) less than two standard deviations below median WHO standards [7]. A key finding from the Global Enteric Multicenter Study (GEMS) [4] was that childhood episodes of moderate-to-severe diarrhea (MSD) were associated with negative shifts in average in HAZ  from other infectious diseases due to diarrhea-induced stunting and found that mortality estimates increased by 24% (ETEC) and 28% (Shigella) over estimates of direct deaths due to these pathogens [8,9]. Another study of global diarrheal disease burden that included long-term sequalae and other infectious diseases deaths from diarrhea-induced stunting found that DALY burden increased 39% over acute diarrheal disease burden [10].
A recent study found that ETEC vaccination was most costeffective for children living in countries grouped into the World Health Organization (WHO) Regional Office for Africa, which included the four countries in this analysis: Kenya, the Democratic Republic of Congo (DRC), Zambia, and Zimbabwe [11]. In the same study, Shigella vaccination was also found to be cost-effective in Regional Office for Africa (AFRO) countries as compared to other countries but was most cost effective in Regional Office for the Eastern Mediterranean (EMRO) countries. However, national estimates of ETEC and Shigella vaccine impact and cost-effectiveness were estimated without considering disease heterogeneity at the sub-national level. Previous rotavirus vaccination impact studies suggest diarrheal mortality risk is heterogenous within countries due to disparities in exposure (e.g., access to safe or improved water and sanitation) and susceptibility (e.g., access to healthcare, nutritional status) across geographic and socioeconomic subpopulations [12][13][14].
Vaccines for ETEC and Shigella are currently under development and could be critical interventions to reduce geographic and socioeconomic disparities in disease burden. The United Nations' Sustainable Development Goals (SDGs) have prioritized addressing disparities by setting goals to provide universal and equitable access to healthcare and vaccines (SDG3). Improving equity in health outcomes requires models that produce estimates at levels that can optimize intervention delivery by reaching the most vulnerable children within countries [15]. Here, our objective was to estimate impact and cost-effectiveness of introducing ETEC and Shigella standalone vaccines in geographic and socioeconomic subpopulations within four sub-Saharan African countries with high estimated ETEC and Shigella burden [16]. We explore how heterogeneity in burden and delivery result in more refined impact and cost-effectiveness estimates, which may inform implementation strategies to better address related inequalities in these countries.

Methods
We developed dynamic subnational models projecting the impact and cost-effectiveness of ETEC and Shigella vaccines across different geographic areas and socioeconomic subpopulations of children in DRC, Kenya, Zambia, and Zimbabwe. We estimated deaths and morbidity due to ETEC and Shigella diarrhea as well as burden from other infectious disease deaths attributable to ETEC-and Shigella-induced stunting using methodology [9] that assumed moderately-to-severely stunted (MSS) children have higher risk of mortality due to other infectious diseases [17,18].

Geographic and socioeconomic subpopulations
Malaria Atlas Project subnational estimates were used to project 2025-2034 populations in each country [19]. In the case of Kenya and DRC, we aggregated Demographic and Health Surveys (DHS) data to previous provinces to correspond to DHS administrative levels, since each country has recently changed their first-level administration units. Thus, for all countries the subnational geographic unit of analysis were provinces (Fig. 1). We modeled socioeconomic subpopulations by grouping individuals from zero to four years of age (children) into five wealth quintiles within each provincial area. We created wealth quintiles based on house-hold asset index scores [20] with water and sanitation assets excluded to avoid confounding with exposure risk variables [21].

ETEC and Shigella etiological fractions
We used previously published methods to estimate the etiological fractions of diarrheal burden due to ETEC and Shigella. Briefly, we used bacterial culture-based estimates for AFRO countries [22] and transformed them using adjustment factors of 1Á5 (50% increase) for ETEC and 2Á0 (100% increase) for Shigella, consistent with the molecular-based re-analysis of GEMS data [23]. We assumed that etiological fractions were uniform within countries and remained static over time.

Diarrheal mortality and morbidity
We estimated diarrheal mortality for each country by using the mid-point of two diarrheal mortality estimates; one from the Institute for Health Measurement and Evaluation (IHME) [24] and the other from Maternal and Child Epidemiology Estimation Group (MCEE) [25] (Table 1). We used published annual regional diarrhea incidence rates [26] to produce national estimates of morbidity. We projected diarrheal mortality from 2015 to 2034 assuming continued trends in estimates of mortality from IHME and MCEE between 2000 and 2010 to account for pre-rotavirus vaccine mortality rates (Table 1). We projected diarrheal morbidity from 2015 to 2034 using estimated Years Lived with Disability (YLD) from diarrheal diseases between 2000 and 2010 from IHME [24].

Distribution of mortality and morbidity
We assumed mortality risk from ETEC and Shigella was heterogeneous across economic and geographic subpopulations. We attributed mortality to each subpopulation using aggregated individual child mortality risk index scores (R i ) [8,27]. We calculated R i as the product of relative risks associated with factors related to individual exposure (E i ) to enteropathogens and susceptibility (S i ) to diarrheal disease mortality (Eq. (1.1)). We calculated prevalence data for each risk factor using DHS estimates (Table 1) [28][29][30][31].
Our susceptibility index (S i ) was designed to be proportional to case-fatality rates based on three known risk factors (n) for diarrheal mortality: utilization of oral rehydration therapy (ORT) [32], nutritional status (weight-for-age, [WFA]) [18], and receiving vitamin A supplements [33] (Table 1). As shown in Eq. (1.2), S i is the product of the relative risk (s) associated with the jth level of susceptibility risk factor k and the status of each risk factor (F) for individual (i). As only a small subset of children in DHS experienced diarrhea two weeks prior to the survey, we used a logistic regression analysis with relevant household and demographic predictors to impute the probabilities of receiving ORT values (Supp. Table 1) [12].
We assessed established diarrheal disease relative risk estimates (e j;k ) associated with household levels of water and sanitation access (Table 1) [34] to develop an exposure index [27] (Eq. (1.3)), where G i;j;k represents the jth level of household water and sanitation risk factor level k for individual child i. Adapted from Bagamian et al. [8], we scaled the mean individual risk index for each provincial area quintile subpopulation by overall mean risk index for the country. We assumed that the likelihood of a child experiencing an ETEC or Shigella diarrheal episode was associated with exposure risk. We used the same method to distribute diarrheal incidence rates as we used for mortality rates, except scaled, mean exposure index scores for each subpopulation were used in place of the overall diarrheal mortality risk index.

ETEC-and Shigella-induced stunting
We calculated the marginal effect of diarrheal episodes on increased likelihood of stunting by modeling the shift in z-scores from ETEC and Shigella episodes using GEMS results [4]. For each subpopulation, fractions of MSS children were estimated from DHS data [28][29][30][31] to generate normal HAZ distributions and corresponding counterfactuals of distributions without ETEC or Shigella diarrheal episodes [9].
We estimated the shift in probability of MSS in each subpopulation using methods described previously [9]. The number of stunting cases due to ETEC and Shigella diarrheal episodes is the product of expected MSD episodes and the marginal effect on the probability of MSS. The probability is generated from the proportion of children below thresholds of moderate (height < À2 standard deviations (SD) below mean HAZ) and severe (height < À3 SD below mean HAZ) stunting.
We used a previously published approach [9] to estimate the number of deaths that stunted children experience from infectious diseases for which stunting is a risk factor (e.g., pneumonia, malaria, measles, and diarrheal deaths) for each subpopulation. We used the mid-point of IHME and MCEE national mortality rates for lower respiratory, malaria, measles, and non-ETEC and Shigella diarrhea from 2000 to 2016 as our baseline for mortality projections ranging from 2017 to 2034 (Table 1). We estimated the burden of these diseases in each provincial area by distributing the projected other infectious diseases mortality rates using the DHS under-five child mortality rate estimates. We assumed these rates did not vary across quintiles.

Costs
We estimated country-specific direct medical costs of ETEC and Shigella illness associated with inpatient and outpatient care for a diarrhea episode [11]. We based our direct medical costs on WHO-CHOICE Service Delivery Unit Cost estimates and commodity costs [35]. We assumed the proportion of children who sought care at outpatient facilities based on DHS data on seeking care at a healthcare facility after a diarrhea episode [28][29][30][31]. We assumed one of eight outpatients were referred for inpatient care [36,37].
We distributed the direct medical costs per child [11] using the estimated relative cost per child in each quintile (Table 1) [12]. We calculated subpopulation relative medical costs based on combining utilization of public and private, inpatient, and outpatient  [52] services after child diarrheal episodes from DHS with WHO-CHOICE national cost estimates adjusted to 2016 US$ [35]. As described for the probability of receiving ORT above, we imputed estimates for each child using the results of a logistic regression that included demographic and household characteristics as predictors for medical costs of illness (Supp. Table 1) [12][13][14]. We calculated net costs (N) for each country (c) by summing the difference of vaccination and averted costs in each quintile population (Eq. (2)). We calculated vaccination costs (V) for each quintile (q) by summing vaccine administration costs [38], vaccine price, and quantity for each vaccinated child in the subpopulation assuming 10% wastage over the first 10 years (t) postintroduction. We calculated averted costs (A) in each subpopulation as the product of vaccine effectiveness and estimated direct medical costs in quintiles (q) of each provincial area (a) population. We assumed the same costs of vaccination for ETEC and Shigella standalone vaccines.
All costs were discounted at 3% per year over the 10 years postintroduction and presented in 2016 US$.

Vaccine effectiveness and benefit
We evaluated the impact of national ETEC and Shigella vaccine programs separately, assuming each vaccine would prevent 60% of MSD episodes and 60% of acute mortality from diarrhea. We assumed full protection after a child received a three-dose course and did not include partial protection for less than three doses in our estimates. We estimated vaccination coverage for each regional and socioeconomic subpopulation using DHS data on diphtheria, pertussis, and tetanus vaccines (DPT3) [28][29][30][31] as a proxy for ETEC and Shigella vaccine coverage.
As vaccine prices are uncertain until they are ready for market, we assumed that the vaccines in our study would cost US$3.30 per dose, as in our previous analysis [11], based in part on the Gavi Rotarix price of $2 per dose [39]. We used Disability-Adjusted Life Years (DALYs) to express burden from Years of Life Lost and YLD (Table 1) [40].
Vaccine benefits are calculated for each country (B c ) based on the sum of the product of quintile coverage (C q ), efficacy (Ef r;q ), and DALY burden, mortality or MSD episodes (D q ) in each provincial area (a) from 2025 to 2034 (t, Eq. (3)).
We cumulated benefits over the first five years of life in each annual birth cohort as our measure of vaccination benefit. We assumed vaccine effectiveness did not wane, there was no effectiveness beyond five years of age, and did not include the potential of herd protection in our estimates.
As reducing disparities in access to vaccination could prevent more deaths and improve cost-effectiveness [41], we explored a hypothetical scenario simulating the impacts and costeffectiveness of ETEC and Shigella vaccination if coverage in every socioeconomic subpopulation within each province was equal to the national average. We did not include costs of implementing interventions aimed at increasing coverage but did include costs of vaccinating additional children.

Cost-effectiveness
We calculated Incremental Cost-Effectiveness Ratios (ICERs) with net costs and outcomes cumulated over 2025-2034 projections for each subpopulation. We presented ICERs with two denominators, DALYs and MSD episodes averted. We included MSD episodes averted as a denominator to provide an alternative perspective on cost-effectiveness focused on preventing ETEC and Shigella episodes that induce stunting. Our comparator scenario is burden without ETEC or Shigella vaccination.

Uncertainty and sensitivity analyses
We conducted a series of analyses to assess the impact of uncertainty on our predicted outcomes using SimVoi [42] and Microsoft Excel Ò (version 16.27). We used probabilistic sensitivity analysis (PSA) to assess uncertainty in multiple input variables. We characterized key input variables as distributions based on our assumptions or uncertainty reported from original sources (Table 1) and conducted the PSA over 10,000 iterations to develop a distribution of estimated impact and cost-effectiveness by provincial area. We used the results from the uncertainty analysis to estimate upper (97.5%) and lower (2.5%) 95% uncertainty limits (UL) for key outputs and are reported along with point estimates.
We explored willingness-to-pay for vaccination introduction by conducting a threshold analysis on PSA results for each subpopulation. We present the results as the percentage of iterations from Monte Carlos simulations that produced ICERs below thresholds based on national GDP estimates. We compared ICERs to thresholds ranging from 0.25 to 4 times GDP [43][44][45]. We used a oneway sensitivity analyses to estimate the effect of changes in individual input variables. We explored ranges for key inputs (Table 1) and present results as tornado diagrams.

ETEC and Shigella burden
Morbidity and mortality rates declined in every country except in Zimbabwe, where both were projected to increase from year 2025 to 2034 (Table 1). Children from DRC had the greatest number of ETEC and Shigella MSD episodes, while children from Zimbabwe had the highest incidence rate of MSD episodes associated with ETEC and Shigella (Table 2).
Within countries, children from the provinces of Kasai-Occidental (DRC), North Eastern (Kenya), Northern (Zambia), and Mashonaland Central (Zimbabwe) had the highest predicted incidence rates of ETEC and Shigella MSD episodes from 2025 to 2034 (Table 2).
Children from these provinces were estimated to be among the highest rates of total mortality (ETEC and Shigella diarrheal mortality plus other infectious disease deaths due to ETEC-and Shigella-induced stunting). In DRC, Shigella total mortality rates were greatest in Sud-Kivu and ETEC total mortality rates were greatest and equivalent in Kasai-Occidental and Sud-Kivu (Table 2). In Kenya, total mortality rates were greatest in North Eastern, but equivalent to rates in Eastern for ETEC and Shigella. In Zimbabwe, children from Midlands province had the highest total mortality rates ( Table 2). In the highest total mortality provinces in all four countries, children in the lowest and lower quintiles are projected to experience the highest total mortality rates (Fig. 2, Supplemental Table 1). This is a reflection of higher percentages of stunted children and higher diarrheal mortality risk index scores for children in these provinces ( Fig. 3B and C).

ETEC-and Shigella-induced stunting burden
ETEC and Shigella MSD diarrhea was projected to cause the highest number of stunted children in DRC, with the highest rates of stunted children in Zambia, followed closely by DRC (Table 2) from 2025 to 2034. Burden from other infectious disease deaths In the highest total mortality rate provinces, the highest rates of induced stunting burden were found in the lowest and lower wealth quintiles, with the exception of Midlands, Zimbabwe where the middle quintile was equivalent to the lowest quintile estimate (Fig. 2).
Disparities in rates of total mortality averted by wealth status were higher in DRC and Zimbabwe as compared with Zambia and Kenya. In Zimbabwe, total mortality rates averted in Manicaland and Mashonaland Central were over 5 times higher than in urban Bulawayo (Table 3). In DRC, total mortality rates averted in Kasai-Occidental were 4 (ETEC) and 3.5 (Shigella) times higher than those predicted for urban Kinshasa. The total deaths averted were greatest in the lowest and lower subpopulations, where there were larger fractions of stunted children (Fig. 1) and higher ETEC and Shigella induced stunting burden rates (Fig. 2). Rates of total mortality averted were higher in provincial areas where mortality rates were among the highest, despite lower vaccination effectiveness (% reduction), in provinces such as Kasai-Occidental (DRC), North Eastern (Kenya), and Manicaland (Zimbabwe) ( Table 2).

Cost-effectiveness
Both vaccines had ICERs below $1000 per DALY in all provinces except Bulawayo in Zimbabwe ( Subnational ICERs with MSD averted as the denominator showed a different pattern, with lower variation between countries than DALY ICERs. The highest provincial MSD ICER was found in Bulawayo and the lowest in Matabeleland North, both in Zimbabwe (    (Fig. 4). Generally, ETEC and Shigella vaccinations were most cost-effective in subpopulations with the greatest burden (Fig. 5).
Under our equitable scenario (vaccination coverage of each subpopulation being equal to the national average), the most significant gains in deaths averted were for children in the middle, Stunting prevalence (proportion of children in each wealth quintile and province with height-for-age z-scores less than À2 standard deviations below the median WHO reference height), modeled using DHS data on child nutritional status. (C) Subpopulation diarrheal mortality risk was calculated by averaging individual index scores for children in each subpopulation. Higher scores represent higher risk of mortality from ETEC or Shigella diarrhea.     lower, and lowest quintiles in Equateur, Kasai-Oriental, and Katanga provinces in DRC, North Eastern province in Kenya, and Manicaland provinces in Zimbabwe (Fig. 6). National ICERs declined by 20%, 15%, 8%, and 2% in Zimbabwe, Zambia, DRC, and Kenya, respectively (Table 3). Based on the threshold analysis, subpopulations within Zimbabwe met willingness-to-pay thresholds of ICERs being less than 1 times GDP in greater than 50% of our simulation iterations in seven of the ten provincial areas (Supplemental Fig. 1). ICERs were less than three times GDP in more than 90% of model iterations in all subpopulations in all regions with the exception of the wealthiest subpopulations in Bulaweyo. In DRC, Kenya, and Zambia, ICERs in some of the poorest subpopulations were less than two times GDP in over 50% of simulation iterations (Supplemental Figs. [2][3][4]. With the exceptions of Kinshasa (DRC) and Lusaka (Zambia) ICERs estimated for the poorest subpopulations were below three times GDP in 50% or higher of simulation iterations.
The model inputs tested in the sensitivity analysis that had the greatest impact on DALY ICERs in order from greatest to least impactful were price per dose, efficacy and etiological fractions for both ETEC and Shigella vaccination in all countries (Fig. 7). These three inputs accounted for the majority of ICER value swings with different patterns in the impact of variation in the fourth and fifth most impactful model inputs across countries. In Kenya, the next two most impactful inputs were administration cost followed by mortality change over time, while in Zambia and Zimbabwe the order was reversed with mortality change having more impact than administration cost. In DRC, ICER sensitivity to variation in model inputs was higher for estimates of administration costs followed by the fraction of episodes estimated to be MSD.

Discussion
We are the first to explore how geographic heterogeneity and socioeconomic status can influence the potential impact and cost-effectiveness of ETEC and Shigella vaccine introduction within countries. Previous studies have demonstrated the importance of considering heterogeneity in ETEC and Shigella diarrheal burden and the burden other infectious disease deaths due to ETEC and Shigella induced stunting [8,9] and vaccination for these pathogens in LMICs [11]. Both vaccines were most cost-effective (lower ICERs) in lower and lowest quintiles of higher burden subnational areas in all countries, with few exceptions where middle and higher wealth subpopulation estimates were similar to the lowest two quintiles. Stunting burden from other infectious disease deaths attributed to ETEC and Shigella diarrhea contributed between 8-19% and 8-22%, respectively, of the national total burden in four countries. In every country, the vaccines were projected to avert the most deaths in the populations with the highest burden of disease, in some cases, even despite lower vaccination coverage.
Our analysis provides useful information to national and local policymakers as they consider future interventions to reduce diarrheal disease burden. ETEC and Shigella vaccines are most costeffective when the most vulnerable and impoverished populations are vaccinated. Subnational vaccination via the Expanded Programme on Immunization (EPI) is a relatively new concept, but as new vaccines are developed, there is the potential to increase reductions in burden and cost-effectiveness by prioritizing vaccination for the most vulnerable children. Many countries already implement interventions in select areas or high-risk populations [46,47].
Improving vaccination coverage for all subpopulations to the national average by 2025 would decrease total deaths 16% to 62% in our four study countries. Strategies that show promise for reducing inequity in health care in other settings that could be relevant within these countries are: increasing the number of lay health workers; implementing vaccination campaigns targeting highrisk populations; increasing the number of health facilities where vaccination is given; and investing in transportation and storage capacity [48]. Vaccines can take advantage of existing infrastructure to lower enteric infections in the near term as countries move towards universal coverage in healthcare and in water and sanitation to meet long term goals of reducing diarrheal disease burden [49].
As the relationship between morbidity and long-term consequences are better understood, the cost per MSD episode may be a useful metric to consider when assessing cost-effectiveness. Childhood stunting in childhood has been associated with poor cognitive development that can lead to poorer educational outcomes, reduced wages and increased risk of noncommunicable diseases as adults [50]. An important next step is to include estimates of these longer-term consequences of ETEC and Shigella episodes to understand the expanded value of vaccination.

Limitations
As with any scientific study, there were limitations to our approach. First, we assumed that etiological fractions of disease burden from ETEC and Shigella infection were the same across countries and static over the study period due to limited data available on variation in estimates both across and within countries. Variation in estimates of etiological fractions were the third most important on DALY ICERs for both vaccines in all four countries. Second, we were limited to using regional estimates for WHO African Region countries as our incidence rate in morbidity estimates in the absence of published rates estimated at the national or provincial levels for each country. As a result, the differences in ranges of MSD episode ICERs were much smaller between countries and subpopulations than in DALY ICERs. Third, our projections of mortality for diarrheal and other infectious diseases are based on MCEE and IHME trends starting in 2000 or 1990. With vaccination introduction starting in 2025, it is likely that mortality trends could change before or over the 10-year time window. Fourth, our assumptions about shifts in the distribution of stunting due to MSD episodes were based on GEMS study results that included a site in Kenya and sites in three other countries in sub-Saharan Africa. However, our extrapolation of the relationship between diarrheal episodes and stunting to children from countries in this study would benefit by future surveillance and enteric etiology studies to validate or refine our approach. Finally, ETEC and Shigella candidate vaccines are still in development and while they are being tested in clinical trials, there is substantial uncertainty around their efficacy and potential pricing. These two parameters were major contributors to uncertainty around predicted outcomes. Another assumption about vaccine efficacy in this model is that efficacy is uniform across geographic and socioeconomic subpopulations within countries. However, there is evidence of lower rotavirus vaccine efficacy in lower income as compared to higher income regions [51]. In the absence of efficacy studies of enteric vaccines with sufficient power to account for variation across subnational or geographic or socioeconomic populations, it is difficult to address in the current model, but should be addressed by future research.

Conclusion
According to our analyses, ETEC and Shigella vaccines show promise for significantly reducing morbidity, mortality, and long-term impacts of ETEC and Shigella infection. Our results indicate that considering the subnational variation in ETEC and Shigella burden and access to vaccination when designing vaccination programs may lead to higher impact and improved cost-effectiveness of vaccine deployment programs. Childhood ETEC and Shigella vaccines will be critical tools to reducing disease burden and inequality, but funding and effective programs aimed at improving access to EPI vaccination for the underserved would vastly improve overall impact of these vaccines in all four countries. As vaccination coverage is variable across the geographic and socioeconomic populations within countries, achieving universal coverage of ETEC and Shigella and other EPI vaccines will require strategies tailored to the needs of each population.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.