Introduction
Congenital syphilis (CS) is a fetal disease resulting from mother-to-child (‘vertical’) transmission of the bacterium Treponema pallidum during pregnancy.1
In 2007, the WHO launched a global CS elimination initiative.2 In 2014, it set a target of no more than 50 cases per 100 000 live births as an elimination criterion, then in 2016, it set a target of eliminating CS in 80% of countries by 2030.2 However, by 2016, the global CS rate remained at 473 per 100 000 live births,2 almost 10 times the elimination criterion. Globally, CS is the second leading cause of preventable stillbirth23 and a leading cause of fetal and infant mortality.2–5 Infants with CS face long-term health risks, including congenital deformities, unilateral hearing loss and neurosyphilis. From 2012 to 2016, CS case numbers decreased globally but increased in the Americas and the Eastern Mediterranean region.2 From 2020 to 2022 CS cases increased globally from 425 to 523 per 100 000 live births.6 The WHO considers global CS cases, mortality and morbidity to be high.7
CS is in principle preventable and treatable with existing means.7 One means of preventing maternal infections is to reduce risky sexual behaviours. However, this remains challenging.8 One obstacle is that most adult syphilis infections are asymptomatic or unrecognised,9 which prevents care-seeking and treatment. Globally, among those aged 15–49, syphilis prevalence grew from 27.8M in 1990 to 43.8M in 2019,10 and incidence grew from 7.1M in 2020 to 8M in 2022,6 suggesting a failure to prevent and treat syphilis in prime-aged adults.
Another means for CS prevention is antenatal-care (ANC)-based screening and treatment of pregnant infected mothers.10 However, ANC coverage and ANC-based screening and treatment rates remain suboptimal and supply shortages of WHO-recommended treatment have occurred.911
Given global CS elimination targets, persistently high CS burdens, and the failure of existing means to prevent and treat maternal syphilis, syphilis vaccination of women has been considered an alternative and complementary means of preventing CS.1213 The WHO and the US National Institute of Allergy and Infectious Diseases jointly formulated the Global Sexually Transmitted Infections (STIs) Vaccine Roadmap for accelerating research and development (R&D) on STI vaccines, including on syphilis vaccines (SVs).14 This roadmap affirms the importance of developing SV, in part because of its CS prevention value.14 Global leaders in STI vaccine R&D and service provision rank SV as a high priority for STI vaccine development, in part for its contribution to CS prevention.12 However, SV development remains in the preclinical stage.13–16
The Global STIs Roadmap also affirms the importance of modelling the impact and cost-effectiveness of future STI vaccines to support market demand for such vaccines.17 Economic evaluation of potential vaccines is essential to rationalising R&D spending, preparing vaccination programme payers for future coverage and funding decisions and assessing prospects for infectious disease control.
Given the importance of CS elimination, SV’s potential role in such elimination, and the need to rationalise SV development efforts and support payer planning, we economically evaluate SV’s potential contribution to preventing CS. Given the importance of vaccination to population health and socioeconomic well-being, we perform this evaluation from a societal perspective (as opposed to a health payer perspective). More specifically, we estimate country-specific and global willingness-to-pay (WTP) and net monetary benefits (NMB) for the CS-prevention value of a hypothetical SV, reflecting its impact on infants’ health and their lifetime direct and indirect costs.
Given the sensitivity of WTP to ability-to-pay (ATP), we compute both standard and equity-weighted WTP and NMB, where equity weights control for cross-country ATP differences.
WTP and NMB are standard value measures within cost-benefit analysis (CBA) and cost-utility analysis (CUA).18–20 They help inform and support decisions regarding vaccine R&D and vaccine inclusion in national vaccine schedules and help predict country-level demand. Because we value only SV impacts on CS, ignoring female health and sexual transmission impacts, our value estimates constitute floors to the full value of female SV.
Methods
This section summarises our methods and data. Further details are provided in online supplemental appendix 1.
Study characteristics
Our study goal is to economically evaluate the CS-related benefits of female SV. We adopt a societal perspective, which is widely considered the ‘gold standard [and] preferable’ perspective for economic evaluation21 because it ‘optimize[s] efficiency and societal benefit and [avoids] potentially distorted clinical decisions and health policies resulting from adoption of narrower perspectives used by various healthcare system stakeholders’.22 Our analysis is compatible with both CBA and societal-perspective CUA.
Our evaluation includes 189 countries with available data, spanning all income levels and geographies (online supplemental table 3 lists study countries). The study population in each country is a single representative 15-year-old female. The policy comparison is between vaccinating and not vaccinating this female on turning 15.
We assume a single-dose vaccination with no boosters. We assume vaccine effectiveness is limited to reducing the female’s infection probability. Thus, we assume no herd effects or effects on the probability of birth outcomes conditional on infection status.
The study time horizon is from 2030, when we assume SV programmes begin, to 2080. This allows us to follow females throughout the reproductive years (ages 15–49) and avoids uncertainties associated with longer horizons.
Our impact measures are infant-related disability-adjusted life years (DALYs), direct costs and indirect costs. We therefore ignore non-CS-related impacts of female SV, including impacts on the female’s health and costs and on sexual transmission. Our value measures are WTP and NMB.
We value DALYs at per capita gross domestic product (PCGDP). We compute standard WTP and NMB that do not adjust for cross-country differences in ATP, and equity-weighted WTP and NMB that adjust for such differences. We measure standard and equity-weighted WTP and NMB per vaccinated female and in aggregate across all vaccinated females assuming 90% global uptake.
We discount future health and costs at 3% annually and adopt 2022 US dollars (USD) as our base year and currency. (Online supplemental appendix section 9 describes currency conversions.)
Health and cost impacts
For each representative female, we compute expected DALYs and costs with and without vaccination using the decision tree model depicted in figure 1 and described in more detail in online supplemental appendix section 1. All DALYs and costs are associated with the infants of each representative female throughout the reproductive years. DALYs consist of years of life lost and years lost to disability.
Costs consist of direct and indirect costs.
Each decision tree in figure 1, denoted Tva, represents a year in the life of a reproductive age female of age a=15,…,49 and vaccination status v=y,n (for ‘yes’ and ‘no’, respectively). The tree represents all sequences of CS-related events that we assume could occur during the year. The branches of the tree represent the following successive probabilities: dying from general mortality, syphilis infection, becoming pregnant, receiving ANC, syphilis screening within ANC, testing positive or negative from such syphilis screening, syphilis treatment after testing positive and one of seven birth outcomes.
These birth outcomes are low birth weight (LBW)/preterm, neonatal death, stillbirth, early symptomatic CS, late symptomatic CS resulting in neurosyphilis, late symptomatic CS resulting in unilateral hearing loss and non-adverse birth outcomes (ABOs; roughly speaking, a normal birth). Each sequence terminates in a state vector with two elements: the expected lifetime DALY losses faced by an infant with one of the above birth outcomes and the expected lifetime costs associated with such infant’s birth outcome.
A two-dimensional state vector equal to the expected present discounted value of the age-specific state vectors can thus summarise a female’s entire reproductive history:
where ca,v and Da,v are expected costs and DALY losses across all branches of Tva, respectively; cv and Dv are the expected present discounted value of costs and DALY losses, respectively, over the reproductive lifetime of the representative female exposed to vaccination policy v; s(a) is the survival probability; and r is the discount rate, which we set to 3%.
The impacts of female SV on CS-related health and costs are:
where -∆D§amp;gt;0 corresponds to DALY gains from vaccination and ∆c§amp;gt;0 corresponds to cost increases from vaccination.
Box 1 summarises analysis steps.
WTP and NMB
Standard WTP for female SV is:
where WTPSV is WTP per vaccinated female and WTPDALY is WTP per DALY. Within a societal perspective, WTP per DALY is given by the beneficiary’s WTP for health.18
The value of such WTP is contested. The WHO Commission on Macroeconomics and Health provided influential benchmark values of 1–3 times PCGDP.23 While this benchmark has been criticised for lacking rigorous theoretical foundations, it is theoretically plausible for WTP for health to be at least as large as PCGDP since health contributes to earnings and PCGDP is a rough proxy for earnings.24 Subsequent empirical evidence also broadly supports the benchmark estimates that individual WTP per quality-adjusted life year (QALY) in the USA is 6.1 times PCGDP.25 The UK Treasury’s Green Book (2022 edition) estimated WTP per QALY as 70 000 2021/2022 GBPs,26 which was roughly twice PCGDP of 34 675 2021 GBPs.27 Individual WTP for a statistical life year (which is broadly analogous to a DALY) has been found to be at least 3.5 times PCGDP in Burkina Faso,28 4.5 times per capita income in Tanzania,29 and between 1 and 9 times PCGDP in Kenya.24 We adopt the conservative assumption of setting such WTP for health equal to PCGDP.
Standard NMB of female SV is:
where CSV is the cost of female SV, inclusive of dose and delivery costs.
Equity-weighting
WTP is constrained by ATP, so standard WTP analysis risks undervaluing the benefits to low- and middle-income countries (LMICs). To address this, we measure benefits using both standard WTP methodology and a utilitarian equity-weighted methodology that adjusts country WTP for global ATP disparities.30 The latter roughly measures what benefits would be if ATP globally were the same as in the USA in 2030. We adopted conservative and simple assumptions under which such equity-weighting involved multiplying standard WTP and NMB in a country by the ratio of our projection of PCGDP in the USA in 2030 to our projection of PCGDP in that country in 2030.
So, the equity weight for the USA equals 1 by definition, and a country that has one-tenth the PCGDP of the USA in 2030 has an equity weight of 10. (See online supplemental appendix section 12 for further details.)
National decision-makers can focus on unweighted results since they are constrained by the same ATP as national beneficiaries. Global decision-makers and other stakeholders adopting a global perspective can consider both unweighted and weighted results. For example, consider a non-profit global donor with high ATP (eg, a US-based donor) that spends $X in 2030 on LMIC SV vaccination programmes and that wants to estimate a benefit-cost ratio (BCR) for such spending.
Such donor may have a utilitarian ethical view. This view would lead them to prefer a BCR measure that reflects the impact of such spending on global well-being. It would also lead them to prefer a measure of global well-being impact that adjusts for the fact that lower-income countries derive larger well-being benefits from a given income increase than wealthier countries.
Such donor can compute a BCR using the ratio of our equity-weighted benefits to $X. (Costs to a donor should also be equity-weighted by the donor’s ATP. In our example, the donor has US-levels of ATP and the equity-weight for the USA in 2030 is, by definition 1.
Therefore, the equity-weighted costs incurred in the USA in 2030 equal non-equity-weighted costs.)
Population-level values
Given our assumption of no herd effects, we compute country-level standard and equity-weighted WTP by multiplying standard and equity-weighted WTP per representative female by the size of the country’s vaccinated cohort. We set this size equal to 90% of the country’s projected 15-year-old population in 2030. (See online supplemental appendix section 12 for further details.)
The 90% uptake assumption is informed by the parallelism between SV and human papillomavirus (HPV) vaccination. The WHO has set a goal of global cervical cancer elimination, and a target of 90% global HPV uptake among 15-year-old females by 2030.31 Such high uptake has proved feasible even in low-income settings: Rwanda and Bhutan achieved HPV vaccine uptake rates of 93% and 89%, respectively, through primary school programmes.32
Given that the WHO has also set a goal of CS elimination33 and given the same target vaccination population for both goals (15-year-old females), we assume that 90% global uptake of SV among 15-year goal by 2030 can be a significant contributor to SV elimination goals. We assume the schools-based vaccination infrastructure developed for HPV vaccination can also be used for SV, so that the high uptake achieved with the former can be achieved for the latter.
We assume countries in which NMB per vaccinated female would be negative under a female SV programme would not undertake such a programme. Therefore, negative NMBs are never realised. Given this assumption, we compute country-level standard and equity-weighted NMB as the product of standard or equity-weighted NMB per representative female and the size of the vaccinated cohort in that country, or zero if the product is negative.
Parameter construction and data sources
Disease-, health- and cost-related parameters
Key disease-, health- and cost-related parameters are summarised in table 1. The online supplemental appendix provides further details on the construction of these parameters. Online supplemental appendix section 2 describes the Global Burden of Disease Study 2021 (GBD 2021)34 data used in the analysis.
Online supplemental appendix section 3 describes the data used for all-cause mortality risks. Online supplemental appendix section 4 derives the probability of syphilis infection for females aged 15–49. Online supplemental appendix section 5 derives the probability of a pregnancy.
Online supplemental appendix section 6 derives the probabilities of receiving basic ANC, syphilis testing for ANC participants and syphilis treatment for ANC participants who test positive for syphilis. Online supplemental appendix section 7 constructs the probabilities of a ‘true positive’ and ‘false negative’ syphilis test result. Online supplemental appendix section 8 derives the probabilities of each of the seven birth outcomes in our model.
Online supplemental appendix section 9 constructs direct costs, online supplemental appendix section 10 constructs indirect costs and online supplemental appendix section 11 derives DALY losses.
Vaccine-related parameters
Vaccine efficacy
Since SV development remains in the preclinical stage, there is no trial-based or real-world SV-specific evidence of human immunogenicity or vaccine efficacy. We must therefore rely on assumptions. We follow Champredon et al35 in assuming baseline SV efficacy of 80%, with exponential waning at 5% annually.
We explore alternative efficacy and waning patterns in sensitivity analysis.
Vaccination cost
We estimate SV vaccination costs using those associated with HPV2 vaccines. (The WHO Position Paper on HPV vaccines declares bivalent, quadrivalent and non-avalent vaccines as providing ‘comparable … effectiveness [against] cervical cancer, which is caused mainly by HPV types 16 and 18’.36 Of these, the bivalent is the most affordable37 and hence the focal option for cervical cancer elimination.)
We take vaccination costs to equal dose costs plus delivery costs. We set the dose cost in low-income countries (LICs), middle-income countries (MICs) and high-income countries (HICs) to the Gavi/UNICEF HPV2 dose price of $5.18 for LICs, the median HPV2 dose price for self-procuring MICs of $12 and the median HPV2 dose price for self-producing HICs of $27.37 We set delivery costs in LICs and MICs equal to $8.33, which is the average HPV delivery costs estimated in LMICs.38 We set delivery costs in HICs equal to $15, which was the assumed HPV vaccine delivery cost in a US study.39 Thus, vaccination costs inclusive of dose and delivery costs are $13.51 in LICs, $20.33 in MICs and $42 in HICs.
One-way sensitivity analyses
We conduct one-way sensitivity analyses that explore alternative values for decision tree node probabilities and state vectors. Figure 2 describes these sensitivity analyses (see online supplemental appendix section 13 for further details).
Scenario analyses and budget impact analysis
We perform two scenario analyses. Our first analysis values a DALY at three times PCGDP.
Our second analysis adopts a health payer perspective. In this analysis, costs consist only of public-sector-borne direct costs. And to reflect recent evidence on supply-side cost-effectiveness thresholds given historical budgets,40 we set WTP per DALY to 60% of PCGDP.
We construct public-sector-borne direct costs by multiplying base-case direct costs by the country-specific share of current health expenditures funded from domestic public sources in the WHO Global Health Expenditure database.41 See online supplemental appendix section 14 for further details.
We perform a simplified budget impact analysis (BIA) that measures the impact of SV on the health payer’s budget over a 5-year time horizon from 2030 to 2034. This budget impact is measured per vaccinated female, and consists solely of the costs measured in the health payer perspective scenario analysis described above truncated to 2034. See online supplemental appendix section 15 for further details.
We conduct our analysis in Stata V.18.
Patient and public involvement
Neither patients nor the public were involved in this research.
Results
Decision tree model parameterisation
Columns (3)–(6) of table 1 present distributional statistics of our decision tree model parameters, which we computed across all observations in the constructed model input data.
Analytical results
Tables 2 and 3 summarise results per vaccinated female and table 4 presents results aggregated globally and by country income group. Online supplemental appendix table 3 gives country-specific estimates of the outcomes in tables 2 and 3. Table 5 summarises the results of our scenario analysis and figure 2 summarises one-way sensitivity analyses.
Health and cost results per vaccinated female
Table 2 presents our health and cost results per vaccinated female. Rows (1)–(5) show that the median across all countries in the representative female’s probability of syphilis infection during the reproductive years absent vaccination is 0.04. They show that such probability is much higher in LICs and MICs, and varies widely even across countries with similar incomes.
Rows (6)–(10) show that the median reduction in that probability from vaccination is 43.72% and such reduction is relatively uniform across all countries, reflecting our assumption of identical efficacy and waning patterns across countries. Rows (11)–(15) show that the median infant-related DALY gain is 0.0037, that such gains are much higher in LICs, and vary widely even across countries with similar incomes. Such variation mirrors that in syphilis infection risk absent vaccination shown in rows (1)–(5).
Rows (16)–(20) show a median cost increase from vaccination of $1 and that such cost increase does not significantly vary with country income (there is a less than $2 difference in the median increases in LICs and HICs). These small cost impacts result from two offsetting impacts. On the one hand, SV prevents the two fatal adverse birth outcomes (ABOs), stillbirths and neonatal deaths.
This raises lifetime costs because fatalities are associated with little to no lifetime costs, while non-fatalities are associated with significant lifetime costs. On the other hand, SV prevents nonfatal ABOs: LBW/preterm, early symptomatic CS, late symptomatic CS resulting in neurosyphilis and late symptomatic CS resulting in hearing loss. This reduces lifetime costs because non-ABOs have lower lifetime costs than nonfatal ABOs.
Positive entries in these rows reflect the former impact, while negative entries reflect the latter. Expected cost impacts tend to be uniformly small within income groups: the IQR in impacts is 1.16–0.08=1.08, 2.43–0.24=2.19 and 3.45–1.0=2.45 in LICs, MICs and HICs, respectively.
WTP and net monetary benefit results per vaccinated female
Table 3 presents our WTP and NMB results per vaccinated female. Rows (1)–(5) show a median equity weight of 11.64, though such weights are significantly higher for countries with lower incomes. Thus, median equity weights are 119 in LICs, 14 in MICs and 2 in HICs.
Equity-weighting will therefore have a powerful impact on WTP and NMB.
Rows (6)–(15) show our base-case WTP results. The median across all countries in standard and equity-weighted WTP is $27 and $280, respectively. The relatively small cost impacts in rows (16)–(20) in table 2 imply that the vast share of WTP is accounted for by DALY gains.
Cross-country variations in standard WTP are therefore accounted for by variations in DALY gains and in PCGDP, which is used to value such gains. Since equity-weighting controls for PCGDP variations, cross-country variations in equity-weighted WTP are driven by variations in DALY gains, which in turn are driven by baseline infection probabilities absent vaccination. Thus, equity-weighted WTP tends to be much higher in LICs and MICs (medians are $3343, $451 and $29 in LICs, MICs and HICs, respectively).
Our results, however, show significant variation even across countries with similar income levels, reflecting underlying variation in syphilis infection probabilities. Standard WTP tends to be higher in LICs and MICs (medians are $33, $30 and $19 in LICs, MICs and HICs, respectively), reflecting their higher infection risks, which their relative inability to pay does not fully counteract. Standard WTP also tends to vary even across countries with similar incomes, reflecting the underlying variation in infection risk.
The different percentile values of standard WTP trace out a demand curve for the CS-prevention benefits of SV. For example, row (6) tells us that at a vaccination cost (ie, inclusive of dose and delivery costs) of $5, about 90% of countries would demand the SV based on its CS-prevention benefits alone. At a vaccination cost of $11, about 75% of countries would; at $27, about half of countries would; at $56, about 25% would; and at $111, about 10% would.
Rows (8), (10) and (14) tell us that the median LIC, MIC and HIC would find female SV economically optimal for vaccination costs below $32.90, $30.46 and $18.57, respectively.
Equity-weighted WTP is very large in LICs (with a minimum value of $405.55) and vastly exceeds LIC vaccination costs ($13.51). This provides normative justification for global donors with US-levels of ATP to finance those vaccination costs, so that when both vaccination benefits and costs are calculated with the equity weights of those that respectively bear them, the resulting NMB=$405.55–$13.51=$392.04 is very large.
Rows (16)–(25) show NMB results. Median standard and equity-weighted NMB are 0.04 and 0.62, respectively, which are positive, suggesting the optimality of SV programmes, but small. However, this global median hides more significant positive results in LICs and MICs with median standard NMB of $19 and $10, respectively.
Given LIC and MIC vaccination costs of $13.51 and $20.33, respectively, these correspond to standard rates of return (RoR) of (19.39/13.51)×100=143% for LICs, and (10.13/20.33)×100=50% for MICs. Median equity-weighted NMB in LICs is very large at $1728, and the median in MICs is large at $159. Standard and equity-weighted NMB are negative for at least 75% of HICs, but positive for at least 10% of HICs, where a vaccination cost of $42 corresponds to a standard rate of return of at least (54.43/42)×100=130%.
Rows (18), (20) and (24) tell us that at expected vaccination costs, the median LIC and MIC would find female SV economically optimal, while the median HIC would find it sub-optimal.
Population-level results
Table 4 shows that global standard and equity-weighted WTP are $2.7B and $91.7B per vaccinated cohort, respectively, and that global standard and equity-weighted NMB are $1.6B and $60.7B per vaccinated cohort, respectively. Most of the global value of female SV is realised in LMICs, and within LMICs, mostly in MICs. The global standard and equity-weighted NMB are significant.
The average cost of bringing a novel vaccine to the US market has been estimated at $886.8M.42 Vaccinating a single cohort results in a global standard NMB exceeding expected development costs, and in global equity-weighted NMB approximately 68 times development costs. Vaccinating multiple cohorts further strengthens the value proposition of SV development.
Scenario analyses
Table 5 presents the results of our scenario analyses. Rows (1)–(4) show results of our scenario analysis valuing DALYs at three times PCGDP. Given the relative unimportance of cost impacts, this roughly triples median standard and equity-weighted WTP.
It also raises standard and equity-weighted NMB from near-zero values in the base case to $62.90 and $644.03, respectively. At least 75% of countries would now find female SV economically optimal based on CS-related value alone.
Rows (5) and (6) show results of our scenario analysis adopting a health payer perspective. Adopting this perspective reduces median WTP by 44% of the base case value from $27.16 to $15.21, which largely reflects the 40% reduction in the DALY value. It also reduces median NMB from near-zero but positive to −$8.14, so that the median country will no longer find female SV economically optimal based on CS-related value alone.
Row (7) shows the results of our simplified BIA, which ignores vaccination costs and focuses only on short-term direct costs borne by the payer. Again, reflecting our general result that vaccination impacts on costs are relatively small, row (7) shows that the impact of SV on short-term payer-borne costs is small. This further implies that in a fuller BIA that does account for vaccination costs, the vaccination impact on payer-borne direct costs can be ignored, and total budget impact roughly equals the product of the vaccination costs and cohort population size.
One-way sensitivity analyses
Figure 2 illustrates the results of our one-way sensitivity analyses. These results show the percentage impact on median standard WTP relative to the base-case analysis due to the parameter change described for each scenario. They also show the relative quantitative importance of different determinants of WTP.
Figure 2 oneD shows that results are most sensitive to discounting assumptions (base case assumes 3%). Discounting health and costs at 0% raises median WTP by 120%, while discounting health and costs at 6% reduces it by 50%. The discounting of health is more consequential than the discounting of costs given the relative unimportance of SV’s cost impacts.
The second most influential choice is which GBD 2021 estimates we use. Replacing the GBD 2021 mean estimates used in the base case with its upper-bound estimates raises median WTP by 91%, while using its lower-bound estimates reduces it by 59%.
The next most influential determinants are vaccine effectiveness assumptions (base case assumes vaccine efficacy is 80% and protection wanes exponentially 5% annually thereafter). Faster waning or a lower initial effectiveness reduce median WTP by about 38%, while faster waning combined with a lower initial effectiveness reduce it by about 62%.
The following determinants are of roughly equal importance: using Qin et al43 estimates of ABO probabilities (see online supplemental table 4) reduces median WTP by 51%; achieving WHO ANC-related targets44 reduces median WTP by 34%; raising the age at which the cohort is vaccinated raises median WTP by 25%–37% (base-case vaccine administration age is 15 years old); and using upper-bound fertility estimates raises median WTP by 24%, while using lower-bound estimates reduces it by 27%.
The least important determinants are assumptions regarding the level of direct and indirect costs: raising or lowering these costs by 20% results in no more than a 3% change in median WTP.
Discussion
To summarise our results, the median DALY gain per vaccinated female is 0.0037, and the median standard and equity-weighted WTP per vaccinated female are $27 and $280, respectively, which largely reflects DALY, as opposed to cost, impacts. DALY gains and standard and equity-weighted WTP tend to be higher in LICs and MICs, mirroring variations in syphilis infection risk absent vaccination. Equity-weighted WTP in LICs is substantially larger than vaccination costs, suggesting a normative justification for global donors to fund all LIC vaccination costs to realise SV benefits.
Median standard and equity-weighted NMB per vaccinated female are positive but small globally. However, this conceals significant values in LMICs. Standard and equity-weighted NMB are negative for at least 75% of HICs, but positive and substantial for at least 10% of HICs.
Global standard and equity-weighted WTP are $2.7B and $91.7B per vaccinated cohort, respectively. Global standard and equity-weighted NMB are $1.6B and $60.7B per vaccinated cohort, respectively. These global values are mostly realised in LMICs, and especially MICs.
They compare extremely favourably with the expected costs of bringing a novel vaccine to market.
Scenario analysis shows that policy conclusions are sensitive to the choice of perspective and the value per DALY. Budget impact largely reflects vaccination costs. Sensitivity analysis shows that results are most sensitive to health discounting and the choice of mean as opposed to upper- or lower-bound GBD 2021 estimates.
Results are least sensitive to cost assumptions.
Equity weighting has quantitatively significant impacts on per female and population-level WTP and NMB. Such weighting departs from standard health economic evaluation methods. It reflects the view that benefit measures should control for ATP differences, a view that is fundamentally ethical rather than empirical in nature.
We assume representative females in each country, so ATP differences within our analysis exist across countries but not within them. Thus, national decision-makers can focus on the standard WTP and NMB estimates reflecting standard methods. Global decision-makers and other stakeholders adopting a global perspective and sharing the above ethical view can consider both unweighted and weighted results.
The following are study limitations. First, and most importantly, SV development remains in the preclinical stage. There exist no SV-specific trial-based or real-world evidence regarding efficacy, waning, number of doses and dose and delivery costs.
We must therefore rely on assumptions regarding these quantities. Our assumed values may deviate from their future realisations, and such possible deviation is the central source of uncertainty regarding our results.
Second, we analyse only the CS-preventive value of female SV. In particular, we ignore its impact on the vaccinated female’s health and direct and indirect costs, and on sexual transmission. For this reason, our WTP can be interpreted as a floor to the full WTP and value of female SV.
Third, and for simplicity, we use rule-of-thumb values such as one or three times PCGDP for the WTP per DALY from a societal perspective. A more fully developed approach would replace rule-of-thumb values with values better informed by utility theory and evidence (eg, following the methods described in Sevilla25). Fourth, for simplicity, we do not address and take no stand on the value of male SV.
Some further study limitations involve data scarcity. Our study population consists of a single representative female in each country. This simplification abstracts away from within-cohort epidemiological, demographic or socioeconomic heterogeneity.
However, it was necessary due to the lack of easily accessible standardised cross-country data on such heterogeneity.
We are unable to model a broader range of pregnancy, birth and long-term outcomes including miscarriages and disfigurement. Data limitations also require us to use cerebral palsy as a proxy disease for deriving the direct and indirect costs of neurosyphilis. GBD 2021 provides no forecasts of early symptomatic CS and late symptomatic CS with either neurosyphilis or unilateral hearing loss.
We therefore projected 2022 estimates for these ABOs to all future births. As a result, unlike our projections for the other ABOs, those for early and late symptomatic CS do not reflect the continuation of any recent trends in CS risk factors.
We are unable to model treatment of infants with CS, so our results may fail to represent outcomes when treatment access and provision improve. Our estimates of the post-acute lifetime direct and indirect costs of late symptomatic CS neurosyphilis and of post-acute direct costs of LBW/pre-term births are based wholly on HIC data, extrapolated globally assuming these are constant shares of PCGDP. Such extrapolations may fail to adequately reflect LMIC-specific patterns in such costs.
Other limitations involve the simplifications of our modelling approach. The first of these is that we ignore the health burdens and costs of mothers and caregivers. Second, our use of a simple decision tree prevents us from investigating complex dynamics, such as the impact of past pregnancy, infection and vaccination on future reproductive decisions and other relevant behaviours.
Third, we limit vaccine effectiveness to preventing infections among the vaccinated, ignoring the possibility of transmission effects, herd effects or impacts on disease severity. Fourth, for simplicity and feasibility, we conducted one-way sensitivity analyses and scenario analyses rather than a probabilistic sensitivity analysis (PSA). A PSA could provide more information on the impact of parameter uncertainty on model outcomes.
However, a PSA would be very complex given our parameter-rich model and scarce information regarding probability distributions of model parameters. A final limitation of our modelling approach is that we ignore a broader range of long-term CS costs, such as increased mortality, special education and social care costs, and productivity losses.
We can compare our health impact estimates with those for related interventions, though differences in study design and data necessarily render cross-study comparisons inexact. Our global median health impact estimates (0.0037 DALYs) are mostly smaller than those found in modelling of related interventions. HPV vaccines have been found to yield 0.306 DALYs per vaccinated girl (Abbas et al).45 Screening and treatment of syphilis in ANC to reduce CS was found to yield between 5765/1 000 000=0.005765 and 93 484/1 000 000=0.093484 DALYs per pregnant woman in a variety of country settings (Kahn et al)46 and 2.6/23.5=0.11 in sub-Saharan Africa (Kuznik et al).47 Maternal vaccination with RSVPreF to protect against infant respiratory syncytial virus disease was found to produce 2264/3 710000=0.0006102 and 2061/342 110=0.00602 QALYs per vaccinated pregnant persons in the USA and Argentina, respectively (Averin et al48; Rey-Ares et al49).
Maternal group B Streptococcus vaccination to protect against infant invasive group B Streptococcus was found in modelling to yield 4/99.8=0.04 DALYs per pregnant woman (Procter et al).50
HPV vaccination is a close comparator to SV and sheds light on how to promote high SV uptake in young girls globally. Both involve vaccinating young girls (and boys) against sexually transmitted infections. Both correspond to global elimination targets set by the WHO for cervical cancer51 and CS.33 And both elimination targets may require very high uptake (eg, 90% of girls vaccinated with HPV vaccine by age 15 years51).
The HPV experience shows that 90% uptake is feasible, even in low-income settings, through school-based vaccination.32 Uptake can be facilitated by single-dose vaccines; vaccinating both genders in part to reduce stigma and hesitancy; comprehensive communication and education efforts that address schools (including teachers), parents and caregivers; community outreach to address out-of-school youth; facilitating access to vaccination without requiring a vaccination-specific health appointment52; and using presumptive language that assumes vaccination is the default.53 Future SV vaccination programmes should consider building on pre-existing schools- and clinic-based HPV vaccination programmes.
Conclusion
Global standard and equity-weighted NMB for the CS benefits of female SV strongly justify the development of SV. At expected vaccination costs, the median LIC and median MIC, but not the median HIC, will find female SV economically optimal based on its CS benefits alone. Strong normative justification exists for global donors to fund vaccination costs in LICs.
Four possible extensions of our analysis are to value SVs’ impact on sexual transmission, conduct a PSA of model outcomes to parameter uncertainty, estimate the value of SV assuming the vaccine is also administered to males, and to replace rule-of-thumb values for DALYs with values more strongly grounded in theory and evidence.
Supplementary material