Sustainable Development Goal Halftime Project: Benefit-Cost Analysis Using Methods from the Decade of Vaccine Economics Model

Abstract In 2023, the world will be at “halftime” with respect to the sustainable development goals (SDGs). This midline acts as an important milestone to review the progress of the SDGs and develop policies based on the most effective interventions. To estimate the remaining resources needed to achieve SDG targets for vaccines from 2023 to 2030 as well the resulting economic benefits, in this analysis, the incremental economic benefit-cost ratio (BCR) for immunization programs in 80 low- and middle-income countries targeted by the Global Vaccine Action Plan from 2023 to 2030 is calculated. Of these 80 countries, 27 are classified as low-income countries and 53 are classified as lower-middle-income countries (LMICs). The economic evaluation covers 9 vaccines employed against 10 antigens and delivered through both routine immunization programs and supplemental immunization activities. The vaccines covered in the analysis include pentavalent vaccine, human papillomavirus vaccine, Japanese encephalitis vaccine, measles vaccine, measles-rubella vaccine, meningococcal conjugate A vaccine, pneumococcal conjugate vaccine, rotavirus vaccine, and yellow fever vaccine, and correspond to the vaccines covered in the return-on-investment estimates presented in Sim et al., which covered 94 LMICs from 2011 to 2030. For these countries, we estimate program costs from the health system perspective, including vaccine costs such as costs to procure vaccines, which incorporate injection supplies and freight; and immunization delivery costs, which include nonvaccine commodity costs to deliver immunizations to target populations and incorporate labor, cold chain and storage, transportation, facilities, training, surveillance, and wastage. Economic benefits are calculated using a value of statistical life year (VSLY) approach applied to modeled cases, and deaths averted are converted into averted years of life lost using life expectancy data. BCRs are presented as the final output that compares incremental costs and benefits from the baseline of 2022 levels, assuming diminishing returns to scale. Overall, for this period, we estimate total costs of US$ 7,581,837,329.08 with VSLY benefits of US$ 762,172,371,553.54, resulting in a BCR of 100.53.


Introduction
In 2023, the world will be at "halftime" with respect to the sustainable development goals (SDGs).This midline acts as an important milestone to review the progress of the SDGs and develop policies based on the most effective interventions.As we advance toward 2030, both funders and governments will continue to face high demands for health and social investments in order to make progress toward the SDGs and the achievement of universal health coverage while dealing with new challenges such as emerging infectious diseases, humanitarian crises, and climate change.All of these concerns present a need for further political commitment and contributions to protect the hard-won gains achieved during the first half of the SDG timeline.
Building on the previous Decade of Vaccine Economics (DOVE) Return-on-Investment (ROI) study and the subsequent Vaccine Economics Research for Sustainability and Equity (VERSE) project (Sim et al., 2020), this analysis aims to provide insights on the economic benefits and costs of immunization programs.Pediatric immunization is largely considered one of the most cost-effective interventions, with previous studies estimating the ROI for common pediatric vaccines to be between US$ 15 and US$ 52 per every US$ 1 invested (Ozawa et al., 2016;Sim et al., 2020).In addition, while immunization directly impacts health, and therefore the SDGs, it has also been found to play an indirect role in contributing toward advancements in 14 out of the 17 SDGs (Decouttere et al., 2021).As such, it is important to understand the benefits and costs of immunization programs in a manner that allows comparison directly across both healthcare interventions as well as nonhealth interventions targeted at other SDGs.

Objective
The objective of this analysis is to provide estimates of the economic costs, benefits, and benefit-cost ratios (BCRs) for interventions to attain SDG targets within 80 low-income countries (LICs) and lower-middle-income countries (LMICs) in order to advocate for more funding to the most effective interventions and policies across all sectors over the next 7.5 years.This particular evaluation shines a light on pediatric immunization, estimating total and incremental BCRs for nine different vaccines in 80 LMICs (Sim et al., 2020).

Scope
This analysis is focused on the economic benefits and costs of immunization programs in 80 low-and middle-income countries targeted by the Global Vaccine Action Plan (GVAP) from 2023 to 2030.Of these 80 countries, 27 are classified as LICs and 53 are classified as LMICs.The economic evaluation covers 9 vaccines employed against 10 antigens and delivered through both routine immunization programs and supplemental immunization activities (SIAs).The vaccines covered in the analysis include pentavalent vaccine, human papillomavirus (HPV) vaccine, Japanese encephalitis (JE) vaccine, measles (MCV) vaccine, measles-rubella (MR) vaccine, meningococcal conjugate A (Men A) vaccine, pneumococcal conjugate (PCV) vaccine, rotavirus vaccine, and yellow fever (YF) vaccine and correspond to the vaccines covered in the return-on-investment estimates presented in Sim et al. (2020), which covered 94 LMICs from 2011 to 2030.Table 1 contains the full list of countries and detailed categorization of the countries according to the World Health (v) Other recurrent costs: program management, short-term training, information, education and communication (IEC)/social mobilization, disease surveillance, wastage management, and other recurrent costs.
The analysis was conducted from the health system perspective, and it does not factor in household costs such as transportation or lost productive time due to immunization sessions.
Vaccine cost.We generated demand forecasts for each type of routine and SIA vaccine.The number of doses procured is a function of the size of target population, vaccine coverage rate, the number of recommended doses for a fully immunized person, a wastage rate, and a buffer stock rate.The Vaccine Impact Modelling Consortium (VIMC) (n.d.) secretariat provided the demographic data based on the UN World Population Prospect 2019 as well as data for each antigen based on GAVI's operational forecast updated in 2018.For SIAs, we used separate data on target populations and the coverage rate provided by the VIMC.Vaccinespecific, time-invariant wastage rates are based on GAVI's Detailed Product Profile (World Health Organization, 2005).Based on consultations with the GAVI market-shaping team, uniform buffer stock rates (25% for routine immunization and 0% for SIAs) were applied to all vaccines (Public Price Forecast, 2021).
where i = vaccine, j = country, and k = year.Vaccine prices are from three different sources.The GAVI provided the public price forecast information (2023)(2024)(2025)(2026)(2027)(2028)(2029)(2030) for 73 GAVI countries (Pan American Health Organization (PAHO)/WHO, 2021).The other countries included both PAHO countries and non-GAVI/non-PAHO countries.Since PAHO and United Nations International Childrenʼs Emergency Fund (UNICEF) do not conduct price forecasts for future years, we generated price forecasts (2023)(2024)(2025)(2026)(2027)(2028)(2029)(2030) based on the same principle applied to the GAVI price forecasts, which takes the estimates from the latest year where data are available and assumes a constant price throughout the remaining years.This assumption is made due to difficulties associated with long-term forecasts of the market landscape and corresponding vaccine prices.The historical vaccine prices for PAHO countries were obtained from the PAHO Revolving Fund price list (Pan American Health Organization (PAHO)/ WHO, 2021).For the other non-GAVI and non-PAHO countries, the UNICEF vaccine price list was applied (UNICEF, 2018).
For PAHO, UNICEF, and GAVI's forecasted prices, we took an average price per dose for each vaccine across all listed products offered by multiple manufacturers, given the uncertainty in volume procured for each product type.GAVI's immunization supply costs (syringe, recon syringe, and safety box) and freight costs (as a percentage of the unloaded vaccine price) were applied to all 80 countries.
The number of doses was multiplied by price per dose for each vaccine, country, and year to estimate the total vaccine costs.

Vaccine costs
number of doses ijk × price per dose ijk À Á : Immunization delivery cost.Routine immunization.Estimates of routine delivery cost per dose were derived from the most recent empirical results estimated by Portnoy et al. (2020), which generated standardized delivery costs for 134 LMICs through a Bayesian metaregression model.The study used the Immunization Delivery Cost Catalogue (IDCC) to help predict future delivery cost per dose.For Kosovo, West Bank, and Gazawhere estimates are not available through the Portnoy et al. (2020) model, we used the estimates from the immunization costing study conducted by Sim et al. (2021).
number of doses ijk × delivery cost per dose ij À Á : Incremental cost for introducing new vaccines: The empirical studies from the IDCC provide unprecedented opportunities for estimating incremental cost for new vaccine introduction in addition to estimating total costs (Immunization Delivery Costs in Lowand Middle-Income Countries, 2020).Due to a lack of data for other vaccines, we estimated only the average incremental cost per dose for HPV, PCV, and rotavirus vaccines.We also assumed that, in the future, pentavalent and MR vaccines will slowly replace traditional vaccines against the same antigens (i.e., DTP and measles).Incremental costs include both introduction and startup costs for newly introduced vaccines, as well as recurrent costs.No distinction was made between HPV cost estimates from routine delivery via health facility and school delivery given a large degree of heterogeneity in costs of each method as well as decisions regarding HPV vaccine delivery strategies, even within countries.
Incremental delivery cost per percentage increase in coverage: Earlier modeling analyses took different perspectives on how routine immunization delivery cost per dose will change beyond baseline years.Gandhi et al. (2013) assumed a constant delivery cost per dose that is not linked to the coverage rate or additional doses.Portnoy et al. (2015) applied a marginal delivery cost for additional doses derived from a regression analysis of cMYP costing tools separately for countries with DTP3 coverage rates above and below 80%.Because it is increasingly important to understand the additional costs required to increase immunization coverage rates, we have used results from several recent studies (Batt et al., 2004;Pegurri et al., 2005;Ozawa et al., 2016).Ozawa et al. (2018) is an update to two systematic reviews (Batt et al., 2004;Pegurri et al., 2005) that aimed to summarize evidence in peer-reviewed or grey literature that examined the cost and effect of increasing the immunization coverage.Interventions used to increase coverage differs across studies, ranging from text message reminders to education, publicity, and incentives for healthcare personnel.Unlike these two reviews that focused on low-and middle-income countries, the new study by Ozawa et al. (2018) also included evidence from high-income countries and quantitatively examined the relationship between intervention cost per dose and coverage changes, which shows increasing intervention cost per dose for higher levels of coverage.We used the cost function derived from Ozawa et al. (2018) to estimate the incremental cost per dose for each annual coverage rate increase for each country.
We present side-by-side the results from a constant delivery cost per dose assumption ("baseline assumption") and from an increasing delivery cost per dose assumption ("diminishing returns to scale assumption").However, the results under the diminishing returns to scale assumption should be interpreted with caution.Underlying data from the systematic review have inherent limitations due to lack of standardized reporting, recall bias, and heterogeneity of study settings and designs.In addition, the cost function presented is based on data from both LMICs and high-income countries, presenting the possibility of overestimation.When excluding high-income settings from the analysis, a linear relationship between coverage increases and cost per dose cannot be rejected, and as a result, the assumption of increasing delivery cost per dose across all countries and baseline coverage rates remains a subject of debate.
SIAs.Immunization delivery costs for SIAs, often referred to as "operational costs" (Gandhi et al., 2013), consist of nonvaccine costs to deliver vaccines to the target population and manage SIA efforts that are targeted and time-limited.SIAs were conducted for six of the nine vaccines included in this analysis.Catch-up, follow-up, or past preventive campaigns were conducted for measles, measles-rubella, MenA, JE, and yellow fever vaccines.Multiage cohort (girls of age 10-14) for HPV is optional for countries that choose to immunize additional girls beyond the routine cohort and such efforts are also categorized as SIA.
To quantify the delivery cost per dose for SIAs, we used information from the IDCC, a systematic review by Gandhi et al. (2013), and budgeted amount per dose estimates from country proposals submitted to GAVI.We collected 52 estimates from these sources and calculated the average cost per dose for each vaccine type (see Table 3).These estimates were then applied to 80 countries.Sensitivity analysis.We conducted probabilistic sensitivity analysis (PSA) using Monte Carlo simulations to determine uncertainty ranges for each scenario.We varied five parameters simultaneously and performed 10,000 model runs to construct a 95% uncertainty range for total immunization program costs.We used a Gamma distribution for the cost per dose estimates from the compiled data mentioned above for three parameters - country-specific routine immunization delivery cost per dose, vaccine-specific SIA delivery cost per dose, and incremental delivery cost per dose for PCV, HPV, and RV vaccines.
A uniform distribution was used for the percent change in vaccine price per year (between ±15%) (Briggs et al., 2006).
Scenario analysis.Under the base-case scenario, we produced estimates with constant returns to scale for delivery costs at an 8% discounted rate per guidance from the Copenhagen Consensus Center.This scenario is presented as the primary result.We conducted additional scenario analyses by adopting a diminishing returns to scale assumption, using discount rates of 0 and 3% and adopting a wastage rate of 0% instead of the wastage rate based on GAVI's detailed product profile (GAVI, 2018) to demonstrate the impact of diseconomies of scale, vaccine wastage, and discounting on immunization program costs.
In addition, we estimated the incremental cost of achieving 2030 targets by comparing the total costs of achieving the 2030 coverage targets to the cost of immunization programs if the coverage level in 2022 was held constant over time.
Incremental to achieve 2030 target at halftime = Total costs 2030 target coverage À Total costs 2022 coverage : In summary, the scenarios evaluated included the following: Furthermore, due to limited data availability and no standardized vaccine impact models, we were unable to estimate comparable economic benefits for BCG and TCV vaccines.Therefore, these two vaccines were not included in the total immunization program costs or BCRs presented in the results.Instead, we generated cost estimates for both BCG and TCV vaccines and present these estimates separately in Section 5.

Economic benefits
Due to the scarcity of country-specific costs and epidemiologic data and the complexity of estimating the economic burden associated with the antigens modeled, the DOVE-COI models draw upon a variety of data sources.Health impact data are drawn from the focal models of the Goldstein et al. (2005Goldstein et al. ( , 2008)), Chen et al. (2012), Tartof et al. (2013), Walker et al. (2013), Garske et al. (2014), Vynnycky et al. (2019), Quan et al. (2020), and VIMC (n.d.).The modeler and modeling teams that produced these outcomes are listed in Table 4. Key input values that are uniform across the DOVE-COI models are described in Table 5.In addition to these uniform parameters, literature reviews were conducted to identify sources of information for all model inputs that vary by antigen (see Table 6).The use of these parameters in the DOVE-COI models is illustrated in Figure 1 and described in more detail in Section 4.1.4.
Additional input data not represented in the tables were drawn from validated, multilateral agency sources and include real gross domestic product (GDP) per capita, consumer price indices (CPI), US$ to local currency unit (LCU) exchange rates, and percentage of population living in urban areas (IMF, 2010;World Bank, 2013).Wherever possible, disease burden inputs (including the age of vaccination, age of infection, and age of death) were based on epidemiological data and assumptions provided by health impact modeling teams to ensure continuity by aligning the two sets of models as much as possible (VIMC, n.d.).

Methodology
All model costs are presented in 2020 US$ and represent the net present value at year of vaccination, calculated using the discount rates applied in the costing scenarios.Costs    Rotavirus: Model accounts for regional variation in the proportion of severe diarrhea caused by rotavirus; only includes protection from complete vaccination (either 2-dose or 3-dose rotavirus vaccine).
e YF: Proportion of cases leading to severe disease and the case fatality ratio has been updated to 12 and 47%, respectively for model runs following 2015.This analysis applies the lower estimates for consistency with previous analyses, therefore generating a conservative estimate of the economic impact.were adjusted to US$ 2020 through an initial conversion of all nonlocal currency unit (LCU) data to LCU, followed by an application of Consumer Price Index (CPI) growth in LCU, and then a conversion between 2020 LCU and US$ 2020 using IMF (2010) exchange rates.Costs for antigens where disease onset occurred at or before age one were not discounted and antigens with disease onset occurring past 1 year were discounted accordingly.If information was not available for a country-specific model input, a WHO region and World Bank country group-specific1 average for the relevant parameter was calculated and applied.For parameters where cost estimates were abstracted from country-specific studies, these costs were extrapolated out to all model countries using WHO-CHOICE inpatient bed-day costs at a secondary facility as a weighting factor, as illustrated below:  Additional disease burden/epidemiological assumptions.To properly account for longterm disability and convalescence resulting from acute disease, some additional epidemiological assumptions and parameters were incorporated into the DOVE-COI models.These assumptions are listed in Table 7.
Short-term costs.Treatment costs: To measure treatment costs averted that are attributable to immunization, it was necessary to determine how many vaccine-averted cases would have sought care, from where, and how much it would have cost.The number of cases that would have sought care during an illness episode was calculated by applying country-and symptom-specific care-seeking rates to total cases averted estimates provided by the health impact modeling teams (UNICEF n.d.; World Bank, 2013).Parameters for the rate of hospital admittance based on disease severity and the percentage of outpatients seeking care from hospitals were then applied to the overall number of care-seeking cases to determine the facility level at which these cases would have received care.In order to reflect the differential costs of treatment at facilities located in different areas (rural vs. urban), the number of cases seeking outpatient, health center, or hospital care was further stratified by the percentage of the population living in rural versus urban areas (World Bank, 2013).Each estimate of care-seeking cases by location and facility level was then multiplied by WHO country-specific costs of care at each facility level to estimate treatment costs (World Health Organization, n.d.-a).A diagrammatic depiction of treatment cost calculation is provided in Figure 2. Due to wide-ranging uncertainty and a lack of available data on long-term treatment costs for the antigens modeled, only short-term acute and first-year disability treatment costs are estimated in the models.Care-seeking for children suffering from acute disease managed at  Measles infection is assumed to be independent of HIV status Mother-to-child (MTC) HIV transmission rate is assumed at a constant 25% The proportion of measles inclusion body encephalitis (MIBE) is assumed to be 50% of measles cases with HIV Men A Only long-term disability associated with deafness, vision impairment, motor impairment, and seizure disorder was modeled.Other vaccine preventable the outpatient level alone was allocated one outpatient visit, regardless of the antigen (Table 8).
Transportation costs: Acute illness transportation costs were estimated by applying a country-specific cost per trip to a healthcare facility (described in Table 5) to each acute outpatient visit and hospital stay (Kim et al., 2010).Long-term disability transportation costs in the first year of life were estimated using the same method, but it was assumed that these cases would require two round trips to a health facility.For antigens like hepatitis B, where disease outcomes occur later in life, transportation costs were discounted from discount rates varying from 0 to 8%, dependent on the scenario, from the year of care-seeking to the year of vaccination.
Caregiver wages: Caretaker productivity loss was calculated by multiplying an estimate of a caretaker's daily productivity by the number of days lost due to care-seeking (hospital bed days).Given that individuals responsible for caretaking in GVAP countries may be predominantly working either in the home or employed in an informal or low-wage sector of the economy, U.S. State Department estimates of the legal minimum or lowest wage in these countries were used to approximate the value of a lost day of work (Country Reports on Human Rights Practices, 2015).
The loss of caregiver wages was only calculated for individuals seeking treatment under the age of 15, as this was the maximum age at which care-seeking would require supervision/the presence of a guardian in GVAP countries.After this age, it was assumed that care would be sought independently with no associated caretaker wage loss.For each bout of illness, we estimated that caretakers would lose 50% of one day's wages for seeking outpatient care and 100% of their daily wage multiplied by the number of hospital bed-days per illness for hospitalized cases.Long-term costs.A human capital approach was used to determine the economic impact of lost productivity due to disability and death under the COI scenario.For this value, we take the discounted lifetime earnings of an individual, assuming that the individual is in full health (Johannesson, 1996).In the DOVE-COI models, GDP per capita was used as an analogue for the economic contribution of affected individuals in each year (Watts et al., 2021).We assumed that work/economic productivity began at age 15 and that labor participation was 100%.
Productivity loss due to disability: To estimate the number of productive life years lost due to disability, total cases of disability were multiplied by life expectancy at age 16 and discounted back to the year of vaccination.This discounted life expectancy was then multiplied by projected GDP per capita, calculated using the IMF's estimated GDP per capita for the years 2011-2018 and extrapolating these estimates out for the years 2019-2020 using projected GDP per capita growth based on data from the years 2011-2018.Disability weights representing the severity (estimated on a 0-1 scale, with 1 being equivalent to death and 0 being equivalent to perfect health) of each disease outcome were then applied to adjust for the impact of illness on productivity over the duration of an individual's life.
In cases of acute illness, the discounted duration of illness was used in place of discounted life expectancy and multiplied by the number of acute cases.Age-specific survival rates were incorporated in the calculation of productivity loss for antigens where disease onset occurred  First-year long-term disability costs were extracted from four studies (Ding et al., 2003;Liu et al., 2008;Touch et al., 2010;Yin et al., 2012) for three countries.These countries (China, Indonesia, and Cambodia) were used to represent treatment costs in each of the three World Bank income groups represented in the models: upper-middle-income (UMIC), lower-middleincome (LMIC), and low-income countries (LIC), respectively The WHO-CHOICE cost from each country in the model was multiplied by the ratio of treatment costs to WHO-CHOICE cost per bed-day for China, Indonesia, or Cambodia depending on World Bank income group Care was sought for 10% of JE cases suffering from long-term disabilities Measles Estimates of access to care were derived from Demographic and Health Survey (DHS) data regarding proportions seeking care for fever All cases taken to outpatient health facilities incurred the cost of a vitamin A supplement in addition to medication and diagnostic costs Men A All cases taken to a health facility were subsequently hospitalized Chronic-care costs could not be quantified and were not included Rotavirus Estimates of access were derived from Demographic and Health Survey (DHS) data regarding proportions seeking care for diarrhea Rubella Estimates of access to care were derived from UNICEF (n.d.) data regarding percent of children born in an institutional health facility For cases suffering from multiple CRS syndromes, the lowest estimate of care-seeking for the syndromes present was used to remain conservative.All care-seeking acute and long-term CRS cases are hospitalized Medication and diagnostic costs are equivalent to 50% of the WHO-CHOICE cost of a bed-day at a secondary hospital (Lanzieri et al., 2004) CRS long-term disability To determine the cost of treating CRS disability in the first year of life in each country, we multiplied each country's WHO CHOICE cost per bed-day estimate by the ratio of treatment costs gathered in Brazil (Lanzieri et al., 2004)  Productivity loss due to death: The same human capital approach used to estimate productivity loss due to disability was used in the estimation of productivity loss due to premature death.Total deaths for each country were initially multiplied by the probability of survival to age 15 because we do not have this probability of survival for age 16, and then this number was multiplied by the disease-specific life expectancy at death (discounted to year of vaccination) and finally by GDP per capita.
Value of statistical life and VSLY: As an alternative to COI, a value of statistical life (VSL) approach was also adopted to estimate the economic benefits of cases and deaths averted.For these calculations, we rely upon VSL averages for LICs and LMICs, as provided by the Copenhagen Consensus Center.The VSL, derived from the marginal rate of substitution between willingness-to-pay and mortality risk reduction, represents the average value to society of reducing mortality, without respect to wage or productivity (Klose, 1999;Viscusi, 2004).In the United States, VSL is derived from both willingnessto-pay surveys and wage-risk studies.In previous applications of the Decade of Vaccines Economics (DoVE) model, VSL was allowed to vary between country and was estimated using a value-transfer, or benefits-transfer, approach as given by the following equation (Robinson et al., 2019): GDP per capita LMIC GDP per capita U:S: 1:5 × VSL U:S: : This approach assumes an income elasticity of 1.5 and uses GDP per capita values for the USA and LMICs calculated using long-term growth forecasts modeled by the Institute of Health Metrics and Evaluation (IHME, 2022).
However, this report presents a VSL calculated using the standardized Copenhagen Consensus Center VSL for low-and lower-income settings (VSL LIC=LMIC CCC ð Þ Þ and applies it directly to all LMICs using the following formula:

Antigen
Assumptions/Model notes As treatment options and access to care may be low in GVAP countries, we assumed that only 10% of children suffering CRS-caused cardiac difficulty and 20% of all other long-term disability cases would seek care in the first year of life No first-year treatment costs for CNS were modeled (only acute hospitalization and diagnostics) Diabetes treatment costs were not included in the analysis No long-term treatment costs for diabetes were included Yellow fever Estimates of access to care were derived from Demographic and Health Survey (DHS) data regarding proportions seeking care for fever In addition to the VSL approach, we also adopt a value of statistical life-year (VSLY) approach.VSLY is defined based on the marginal rate of substitution between willingnessto-pay and changes in life expectancy and therefore places a larger weight on the value of children's lives, who have a greater life expectancy as compared to older adults (Kniesner & Viscusi, 2019).In previous iterations of the DoVE model VSLY was calculated as: VSLY = VSL Discounted life years remaining : For the purposes of this report, the model was adjusted to compute VSLY based on the Copenhagen Consensus Center's standardized halftime estimates and so the takes on the formula: 0:5 × Life expectancy at birth LIC=LMIC : Similarly to the total VSL impact, that of VSLY is calculated by multiplying the VSLY for LMICs by the total number of life years averted: LMIC : Scenario analysis: Under the base-case scenario, we produced estimates for economic benefits using an 8% discount rate.This scenario is presented as the primary results.We also conducted additional analyses for discount rates of 0 and 3%.
In addition, we estimated the incremental benefits of achieving 2030 target by taking the difference between the total economic benefits of achieving 2030 targets and the benefits of immunization programs assuming the level of cases and deaths averted in 2022 were held constant over time.
In total, 12 benefit estimation scenarios were conducted: (i) The total COI of immunization programs (discounted at 8%).
(ii) The total COI of immunization programs (discounted at 3%).
(iii) The total COI of immunization programs (undiscounted).
(v) The total VSL of immunization programs (discounted at 3%).
(ix) The total VSLY of immunization programs (undiscounted).(x) Incremental benefit of achieving 2030 target at halftime compared to 2022 level through the COI approach.(xi) Incremental benefit of achieving 2030 target at halftime compared to 2022 level through the VSL approach.(xii) Incremental benefit of achieving 2030 target at halftime compared to 2022 level through the VSLY approach.

BCR
The BCR compares the present value of all benefits with that of the costs and investments in the immunization program.This is shown in the following equation: where PV benefits, present value of benefits and PV costs, present value of cost.Please note that while the DOVE programmatic costing model accommodates BCG and TCV vaccines, these vaccine antigens are absent from the benefits model as their health impacts have yet to be estimated.Therefore, the costs of BCG and TCV vaccination programs are presented separately in Section 5.

Economic benefits: COI
Through the COI approach, the total economic benefits of vaccines in 80 LICs and LMICs were projected to exceed US$ 254 billion from 2023 to 2030, assuming a discount rate of 8%.The largest share of economic benefits from vaccination is owed to productivity loss due to deaths averted, accounting for 93.7% of the total benefits.Productivity loss due to disability averted comprises the second most influential component, responsible for 4.5% of the estimated economic benefits (Tables 9 and 10).

Economic benefits: VSL/VSLY
Using a discount rate of 8%, total economic benefits of vaccination for all pathogens for 2023-2030 via the VSL approach for all 80 countries totals over US$ 2.8 trillion.When applying the same parameters for the VSLY method, the benefits of vaccination are nearly US$ 5.7 trillion (Tables 11 and 12).

Immunization program costs
Under the base assumption of an 8% discount rate, the total programmatic costs of vaccination in 80 LICs and LMICs from 2023 to 2030 were estimated to be US$ 20.9 billion (see Table 13).Immunization delivery costs accounted for the greatest proportion of future total immunization program costs at 56.6%, with vaccine costs comprising the remaining costs 43.4% of costs.
We estimated that under a diminishing returns to scale scenario, delivery costs increased by US$ 24.9 billion (19.2%) over the period of 2023-2030.Under the 0% wastage rate scenario, the total vaccine costs decreased by US$ 1.1 billion (9.5%).The results for the different discount rate scenarios are presented annually in Table 13.
Incremental cost calculations show that the costing gap of achieving 2030 target coverage rates for routine immunization compared to the 2022 coverage level is significant.Under constant returns to scale with an 8% discount rate, the incremental costs were estimated at US$ 2.3 billion for vaccines and US$ 1.3 billion for immunization delivery (see Table 14).In other words, it would cost a total of US$ 3.6 billion to reach the 2030 target.

BCG and TCV vaccine costs
Per Copenhagen Consensus Center request, we also estimated the vaccine-specific commodities and delivery costs for Bacille Calmette-Guérin vaccine (BCG) and typhoid conjugated vaccine (TCV).Note that the costs associated with BCG and TCV are omitted from the BCR calculation as benefits models for these two vaccines are still under production.Under the base-case scenario with an 8% discount rate, the cost of BCG and TCV programs would add an additional US$ 3.85 billion to the total vaccination costs between 2023 and 2030 (Tables 15 and 16).

BCR
Using the economic benefits and costing scenarios generated above, we calculated 3 BCR estimates through the COI, VSL, and VSLY approaches.At baseline, with an 8% discount rate, the BCR for attaining 2030 target coverage was estimated at 13.12 (8.20-16.40)through the COI approach, 143.27 (89.60-179.12)through the VSL approach, and 286.12(178.95-357.72)through the VSLY approach.The incremental BCR of attaining 2030 targets was under an assumption of diminishing returns was 3.58, 48.91, and 100.53 for the COI, VSL, and VSLY approaches, respectively (Tables 17-19).

target coverage
Incremental costs/ benefits to achieve 2030 target coverage accounts for differences in the age of mortality impact thereby making the assumption that all life years are treated equally.
The global BCR estimates from this study are large ranging from 12.18 to 273.79 and can inform decision-makers of funding agencies as they prioritize investments across the SDGs as well as contribute to resource mobilization efforts for immunization programs in order to reach the goals set by the global community as part of SDGs.
et al. (2013) Goldie et al. (2008) Quan et al. (2020) Chen et al. (2012) Tartof et al. (2013) (see Hib) (see Hib) Vynnycky et al. (2019) Garske et al. (2014) a Hib/PCV: Only includes impact on children under 5 years.Model estimates deaths averted using residual deaths after accounting for existing interventions, thus reducing the risk of double counting deaths averted from other (nonvaccine interventions); coverage of other interventions (sanitation, antibiotic treatment) held constant.b HPV: Vaccine provides protection against vaccine-type (HPV 16 and 18), no cross-protection.cMenA:Vaccination is assumed to be superior to natural immunity. d

Figure 2 .
Figure 2. Decision tree model for treatment costs.

Table 3 .
Summary table for immunization delivery cost per dose estimates.

Table 5 .
Sources of key input values used across DOVE-COI models.
Kim et al. (2010)estimated the price of transportation (one-time, roundtrip) to health facilities by extracting cost information from 14 studies, identified and narrowed down from a total of 1300 articlesKim et al. (2010)Journal of Benefit-Cost Analysis

Table 5 .
Continued Disability weights were estimated based on responses from household surveys of adults (Bangladesh, Indonesia, Peru, Tanzania, and the USA) and open-access, web-based surveys conducted between Oct. 28, 2009, and May 16, 2011.The surveys used paired Salomon et al. (2012)

Table 6 .
DOVE-COI model/antigen-specific sources of key input values.

Table 7 .
Continued Only estimated treatment costs for the first year of life were included For cases with multiple syndromes, the lowest estimate of care-seeking for the syndromes present was used CRS cases of cardiac abnormality will not go on to develop diabetes since age of death is 1 Yellow FeverOnly cases and deaths due to the most severe form of yellow fever, involving hepatitis, oliguric renal insufficiency, and thrombocytopenia are included Only epidemic disease is modeled All severe disease survivors enter a convalescent-phase following acute infection (LaBeaud et al., 2011) The transmission dynamics of the yellow fever vector, Aedes aegypti, is not captured in the modeling approach used
over the WHO CHOICE cost per-bed day in Brazil before age 15.Due to a lack of data for 15-16 year old children in many countries, we use age 15 data as a proxy for age 16 in order to calculate the number of children that would have reached productive age due to competing risks (WHO, n.d.-b).

Table 10 .
Incremental COI (2020 US$) averted from vaccination programs for 2023-2030, comparing estimates from Table8to base case COI assuming constant VIMC health impact estimates from 2022 for all years.

Table 11 .
Total economic benefits (2020 US$) using VSL and VSLY from vaccination programs for 2023-2030, using VIMC health impact estimates.

Table 12 .
Incremental economic benefits (2020 USD) from VSL and VSLY from vaccination programs for 2023-2030, comparing estimates from Table9to base case COI assuming constant VIMC death impact estimates from 2022 for all years.

Table 14 .
Incremental cost (2020 US$) of immunization programs for 2023-2030 to achieve 2030 target coverage under constant and diminishing returns to scale scenario.