HPV Vaccination to Prevent Cervical Cancer

Published: October 2020

What is the problem?

Cervical cancer kills approximately 300,000 women per year and is thought to require infection of the cervix with specific types of human papillomavirus (HPV). The burden of cervical cancer is greatest in sub-Saharan Africa, but also substantial in the Pacific islands, Southeast Asia, Eastern Europe, the Caribbean, and parts of South America.

Disease background

HPV are a group of more than 150 viruses that infect the skin and mucosa.1 At least 14 HPV types can cause cancer.2 Expert opinion suggests that the most common cancer caused by HPV is cervical cancer, and nearly all cases of cervical cancer are caused by sexually transmitted HPV.3 This is supported by the fact that HPV infection of the cervix is nearly always found in cervical cancer4 and that vaccinating against high-risk HPV strains reduces the risk of developing cervical precancer,5 which is a causal precursor of cervical cancer.6

Expert opinion suggests that two HPV types (16 and 18) cause approximately 70 percent of cervical cancers,7 and this is consistent with the ability of vaccines against HPV 16 and 18 to substantially reduce the occurrence of precancerous cervical changes, as described below.8 The viruses can also cause cancers of the anus, vulva, vagina, penis, and throat.9

Estimates based on a cancer surveillance database suggest that, globally, cervical cancer is typically fatal, but it is more often fatal in low-income countries than high-income countries.10 These data also suggest that, globally, the average age of cervical cancer diagnosis is 53 and the average age of death is 59.11 We have not vetted these estimates.

Burden

Global Burden of Disease (GBD) estimates suggest that, globally, 259,671 women died of cervical cancer in 2017, implying that its mortality burden is similar to meningitis, drowning, protein-energy malnutrition, and alcohol use disorders.12 The World Health Organization (WHO) places its estimate at 311,000 deaths in 2018, and further estimates that more than 85% of these deaths occurred in low- and middle-income countries.13

Recent estimates based on a global cancer surveillance database suggest that sub-Saharan Africa has the largest cervical cancer death burden14 and that cervical cancer is the leading cause of cancer-related death in women in much of this region.15 According to these estimates, the ten countries with the greatest population-size-adjusted cervical cancer death burdens are Malawi, eSwatini, Burundi, Zimbabwe, Zambia, Tanzania, Uganda, Comoros, Guinea, and Burkina Faso.16 In eSwatini, the country with the highest estimated incidence of cervical cancer, these estimates suggest that 6.5 percent of women may develop cervical cancer before age 75.17 The same source of evidence suggests that cervical cancer also imposes a large burden on Melanesia, Micronesia, Southeast Asia, Eastern Europe, the Caribbean, and parts of South America.18 We have not vetted these estimates.

What is the program?

In HPV vaccination, health workers vaccinate girls against specific strains of HPV that cause cervical cancer. Three HPV vaccines are widely available:

• Bivalent. This vaccine targets two HPV strains that are estimated to cause 70 percent of all cervical cancer.
• Quadrivalent. In addition to the strains targeted by the bivalent vaccine, this vaccine targets two HPV strains that cause genital warts.
• Nonavalent. In addition to the strains targeted by the quadrivalent vaccine, this vaccine targets five cancer-causing HPV strains.19

According to a WHO market analysis, in 2019 the quadrivalent vaccine accounted for approximately 60 percent of global market share, while the nonavalent vaccine accounted for approximately 30 percent, and the bivalent vaccine approximately 10 percent.20 According to the same analysis, three additional HPV vaccines are currently in “advanced clinical development.”21

WHO recommends vaccinating girls 9 to 14 years old,22 in part because the vaccine is not effective against established HPV infections that become common once girls are sexually active.23 While it is also possible to vaccinate boys, WHO has determined that targeting girls with the vaccine is a more cost-effective measure against cervical cancer.24 Although we have not vetted this claim, it seems plausible.25 Accordingly, this report focuses on HPV vaccination for girls 9 to 14 years old.

WHO and the US Centers for Disease Control and Prevention (CDC) recommend two doses of HPV vaccine for girls 9 to 14 years old.26 We are currently unsure whether HPV-vaccine-related charities would typically employ bivalent, quadrivalent, or nonavalent vaccines, but since this choice is predicted to have a modest impact on effectiveness,27 it seems plausible that the decision may be based primarily on cost and availability.

What is the evidence of effectiveness?

We did not identify compelling direct evidence that the HPV vaccine reduces the risk of developing cervical cancer or dying of it; however, indirect evidence from randomized controlled trials (RCTs) provides strong evidence that HPV vaccination of women who are not already infected with high-risk HPV strains reduces the risk of developing precancerous changes of the cervix by approximately two thirds. A modeling study based on the effect of vaccination on HPV infection risk, which we draw from in our cost-effectiveness analysis, also suggests strong protection against cervical cancer. Given the link between HPV infection, precancerous changes of the cervix, and cervical cancer incidence and mortality, we believe this constitutes fairly strong evidence that HPV vaccination reduces the risk of developing cervical cancer and dying of it.

Direct evidence

We did not identify compelling direct evidence that the HPV vaccine reduces the risk of developing cervical cancer or dying of it.

We conducted a limited scientific literature search for direct evidence that HPV vaccination impacts cervical cancer incidence and/or mortality.28 We did not identify RCTs that report these endpoints.29 We identified one observational study, Guo, Cofie, and Berenson 2018, that reports the incidence of cervical cancer in US women between 2003 and 2014, a period that covers the 2006 introduction of the HPV vaccine.30

Guo, Cofie, and Berenson 2018 used nationwide cancer surveillance data to estimate changes in cervical cancer incidence after the introduction of the HPV vaccine.31 The authors report that the cervical cancer incidence rate was 29 percent lower in 2011-2014 than it was in 2003-2006 for women 15-24 years old, and 13 percent lower for women 25-34 years old.32 However, it appears that cervical cancer incidence may already have been trending downward prior to the introduction of the vaccine, so this finding is hard to interpret.33

It may be too soon to detect a substantial decline in cervical cancer incidence and mortality in response to widespread HPV vaccination. The CDC recommends vaccination at age 11-12,34 and cervical cancer typically develops in middle age,35 implying that the typical gap between vaccination and incidence in the US is approximately 40 years. Guo, Cofie, and Berenson 2018 considered a gap of 3-8 years, which may not represent enough time to measure the effect of vaccination. In addition, the data represent women 15 to 34 years old,36 which excludes the age range where cervical cancer diagnoses are most common. Considering these limitations, we do not believe this study provides compelling evidence for or against the effectiveness of the HPV vaccine.

Indirect evidence from randomized controlled trials

Indirect evidence from RCTs provides strong evidence that HPV vaccination of women who are not already infected with high-risk HPV strains reduces the risk of developing precancerous changes of the cervix by approximately 63 to 79 percent.

In 2003, WHO convened a meeting of experts to determine the best indirect measures of cervical cancer risk for HPV vaccine trials, due to the impracticality of using cancer itself as a trial endpoint.37 The participants concluded that precancerous changes in cervical tissue, specifically the more advanced CIN2 and CIN3 lesions, are the most informative indirect measure of cervical cancer risk.38 Reasons cited include:

• Cervical precancer is classified as CIN1, CIN2, or CIN3, in order of increasing proximity to cancer. Less advanced CIN1 lesions do not progress to more advanced lesions in approximately 90 percent of cases and are less strongly correlated with future cancer risk than CIN2, and particularly CIN3, lesions.39
• Screening for and removal of CIN2 and CIN3 lesions in high-income countries “substantially reduces the incidence of invasive cervical cancer.” Therefore, a vaccine that reduces the incidence of CIN2 and CIN3 lesions should be similarly effective.40

Many RCTs of HPV vaccination report data on CIN2 and CIN3 lesions. For these reasons, we focus on CIN2 and CIN3 as indirect markers of cervical cancer risk.

We performed a medium-depth scientific literature search for meta-analyses of RCTs on the impact of HPV vaccination on the risk of CIN2 and CIN3.41 The most recent, pertinent, and rigorous meta-analysis we identified is Arbyn et al. 2018.42 This meta-analysis reports a number of outcomes, but we focus on the risk of CIN2 and CIN3 in girls and women who were not infected with high-risk HPV strains at the time of vaccination.43 We include the latter criterion because many of the participants in the trials included in Arbyn et al. 2018 were adult women who were already sexually active and likely to be infected with high-risk HPV strains,44 while the target for community vaccination is 9- to 14-year-old girls who are unlikely to already be infected.

The results for our primary outcomes of interest are the following:

• Among women who were not infected with high-risk HPV strains at baseline, vaccination with the bivalent or quadrivalent HPV vaccine reduced the overall risk of developing CIN2 by 63 percent (95% confidence interval 45% to 75%, five trials, high evidence certainty, two- to six-year follow-up).45
• Among women who were not infected with high-risk HPV strains at baseline, vaccination with the bivalent or quadrivalent HPV vaccine tended to reduce the overall risk of developing CIN3 by 79 percent, but this was not statistically significant (95% confidence interval -10% to 96%, three trials, moderate evidence certainty, 3.5- to 4-year follow-up).46
• Among women who were not infected with high-risk strains at baseline, HPV vaccination reduced the risk of CIN2 and CIN3 associated with HPV strains 16 and 18 by 99 percent47 (three and two trials, high-certainty evidence).48 These are the two highest-risk HPV strains and they are targeted by all HPV vaccines.49 This suggests that the vaccine is highly effective against the specific strains it targets.

Although the second finding is not statistically significant, it suggests a trend toward protection, and in combination with the statistically significant result for CIN2, we believe this constitutes compelling evidence that HPV vaccination reduces the overall risk of advanced cervical precancer.

Arbyn et al. 2018 also provides evidence that the bivalent vaccine may be more effective than the quadrivalent vaccine,50 and that HPV vaccines do not increase the risk of serious harm.51

During our research we learned that Arbyn et al. 2018 has been criticised for not being sufficiently inclusive of relevant trials.52 After a light review of additional systematic reviews investigating the effectiveness of HPV vaccines, we believe it is unlikely that the effect size estimates reported in Arbyn et al. 2018 are substantially inaccurate.53

While Arbyn et al. 2018 provides strong evidence that HPV vaccines are effective at reducing precancer, the fact that CIN2 and CIN3 lesions are only indirectly related to cervical cancer risk introduces some uncertainty into our assessment of their ultimate impact on cancer risk. An additional source of uncertainty is that none of the trials represented in our outcomes of interest were conducted in Africa, which is a region we focus on in our cost-effectiveness analysis.54 However, we are fairly confident that HPV vaccination substantially reduces cervical cancer risk, for the following reasons:

• Cervical cancer is strongly linked to infection with high-risk HPV strains,55 and HPV vaccines generate robust and durable immunity to these strains.56
• Screening for and removing CIN2 and CIN3 lesions appears to substantially reduce cervical cancer risk,57 so a vaccine that prevents CIN2 and CIN3 lesions should also reduce cervical cancer risk.
• Arbyn et al. 2018 represents successful vaccine trials in racially, culturally, and economically diverse populations.58

Indirect evidence from a modeling study

The most up-to-date study to model the global impact of HPV vaccination on mortality is Abbas et al. 2020. This modeling study predicts that the intervention has a large impact on cervical cancer mortality rates, and we believe that the key assumption underlying the model is reasonable. We draw on this modeling study in our cost-effectiveness analysis (see below).

Abbas et al. 2020 is based on the PRIME model, which is used by WHO and Gavi to estimate vaccine impacts.59 The PRIME model incorporates country-specific estimates of cervical cancer burden, demographic data, and the regional prevalence of high-risk HPV strains.60 It uses the regional prevalence of specific HPV strains in cervical cancers to estimate the impact of vaccines targeting those strains,61 under the assumption that HPV vaccines provide complete protection against the strains they target.62 We believe this is a reasonable assumption because Arbyn et al. 2008 reports 99 percent protection against two HPV strains targeted by all HPV vaccines.63 The model does not incorporate data from RCTs of HPV vaccination.

The PRIME model suggests that at 90 percent coverage, HPV vaccination will reduce global deaths from cervical cancer by approximately two-thirds.64 This is roughly consistent with the effect size reported in Arbyn et al. 2018, although we are uncertain how directly precancer rates can be compared with cancer mortality rates.

In our preliminary cost-effectiveness analysis, we use Abbas et al. 2020’s estimates of illness and deaths averted due to vaccination. It estimates that per 1,000 12-year-old girls vaccinated with the bivalent or quadrivalent vaccine, HPV vaccination would avert the following number of deaths:65

• Africa: 22
• Southeast Asia: 11
• Americas: 8
• Western Pacific: 7
• Europe: 5
• Eastern Mediterranean: 4

It also provides estimates of impact at the country level, which suggest that the countries that would benefit most from HPV vaccination are in sub-Saharan Africa.66

Other benefits of HPV vaccination

Drolet et al. 2019 is the most recent meta-analysis of observational studies that compare the risk of HPV-related outcomes before vs. after widespread HPV vaccination.67 In addition to confirming that high-risk HPV strains and CIN2 precancer become substantially less common after the introduction of vaccines,68 it also reports that anogenital warts become less common69 and that protection against HPV-related outcomes extends to boys and men who have not been vaccinated.70

These results suggest that widespread HPV vaccination may have benefits beyond cervical cancer prevention and may cause “herd effects” that increase its effectiveness beyond what would be predicted from direct effects on individuals. However, we have not evaluated this possibility further.

Possible negative impacts of HPV vaccination

HPV vaccination can cause short-term pain, swelling, and redness at the injection site. There is weak evidence that HPV vaccination may increase the risk of death in women vaccinated between ages 24 and 45.

Our views on possible adverse effects of HPV vaccination are preliminary and based entirely on Arbyn et al. 2018, which reports pooled estimates for adverse events in HPV vaccine trials. Relative to placebo injection, HPV vaccination confers a higher risk of short-term pain, swelling, and redness at the injection site.71 The pooled estimates do not suggest that it increases the risk of serious adverse events or adverse pregnancy outcomes.72

All-cause mortality tended to be higher with HPV vaccination. Among the vaccine recipients as a whole, this was not statistically significant and could have been due to chance.73 However, when considering women vaccinated between ages 24 and 45 separately, all-cause mortality was significantly higher in vaccine recipients.74 Our preliminary view is that this finding is not likely to imply a higher mortality rate among vaccine recipients in typical charity settings, for the following reasons:

• The target age for vaccination is 9-14 years, which is 10-36 years younger than the age group in which vaccination may have increased mortality. Arbyn et al. 2018 does not report an increase in mortality in younger women.75
• Arbyn et al. 2018 judged the evidence on mortality to be low-quality.76
• The finding of increased mortality comes from a subgroup analysis, in which only a subset of participants are included. These are generally less reliable than analyses that include all participants, unless carefully designed in advance. In particular, they tend to yield a high rate of false positive findings.77
• The authors state that the causes and timing of deaths showed no clear pattern, and the causes of death were judged to be unrelated to vaccination.78 It seems unlikely that HPV vaccination would increase the risk of multiple unrelated causes of death.

We are not aware of other negative impacts of HPV vaccination but we have not investigated this beyond Arbyn et al. 2018.

How cost-effective is it?

Based on a cost-effectiveness model we put together in May 2020 and updated with our most recent moral weights as of February 2021, we estimate that, in some geographic contexts, HPV vaccination is within the range of cost-effectiveness of the opportunities that we expect to direct marginal donations to (about 10x cash or higher, as of 2021).79 We are currently uncertain about where the intervention would be implemented.

Note that our cost-effectiveness analyses are simplified models that do not take into account a number of factors. There are limitations to this kind of cost-effectiveness analysis, and we believe that cost-effectiveness estimates such as these should not be taken literally due to the significant uncertainty around them. We provide these estimates (a) for comparative purposes and (b) because working on them helps us ensure that we are thinking through as many of the relevant issues as possible. This cost-effectiveness analysis is in an early stage and therefore is not directly comparable to the cost-effectiveness analyses of our top charities. As a general rule, our estimates of a given program's cost-effectiveness tend to go down as we gain more information, and we would conduct a more thorough cost-effectiveness analysis before deciding to recommend funding to this program.

Our analysis estimates the cost-effectiveness of directly funding HPV vaccination, including materials and associated costs. It includes benefits from reducing cervical cancer illness and deaths. Estimates of these benefits are taken from Abbas et al. 2020, which relies on the PRIME model discussed above.

We use estimates from Abbas et al. 2020 rather than adapting estimates from the HPV vaccine trials for the following reasons:

• Abbas et al. 2020 incorporates detailed demographic information and regional estimates of cervical cancer burden, which we would have to separately incorporate into our CEA if we used the estimates from Arbyn et al. 2018.
• Abbas et al. 2020 incorporates estimates of the regional prevalence of specific cancer-causing HPV strains, which we would have to separately incorporate into our CEA if we used the estimates from Arbyn et al. 2018.
• Abbas et al. 2020 reports results by global region, age at vaccination, and vaccine type.
• The effect size estimate for HPV vaccination is similar between Abbas et al. 2020 (total cervical cancer mortality) and Arbyn et al. 2018 (total CIN2 and CIN3).80

However, we recognize that the estimates of vaccine effectiveness in Abbas et al. 2020 are less empirical than the estimates reported by vaccine trials and that this introduces additional uncertainty into our estimates.

Our cost-effectiveness estimate is very sensitive to our choice of certain inputs, including the burden of disease in the area where the intervention is implemented. Some regions have a higher burden of cervical cancer than others, and therefore benefit more from HPV vaccines. We are currently uncertain where this intervention would be implemented. We modeled the two highest-burden WHO regions, Africa and Southeast Asia. The model suggests that HPV vaccines are more than twice as cost-effective in Africa as in Southeast Asia and that other regions would be less cost-effective than Southeast Asia. We did not model individual countries, but in some cases, their burdens diverge significantly from regional averages.81

Under our default assumptions, HPV vaccination in the average country of the WHO African region is within the range of cost-effectiveness of programs we would consider recommending funding, but vaccination in other regions is below that range.82

As a sanity check of the PRIME model, which we do not fully understand, we performed a simple independent calculation based on the lifetime risk of dying from cervical cancer in sub-Saharan Africa and the protective effect of vaccination. This calculation yields an effect size consistent with the PRIME model.83

Additional uncertainties in our cost-effectiveness model include:

• Cost. Our understanding is that the cost of the intervention can vary substantially.84 The variation is driven in part by differential access to reduced vaccine prices. Gavi and the Pan American Health Organization (PAHO) negotiate lower prices for countries they support.85 However, those reduced prices are not available to countries that are neither eligible for Gavi support nor members of PAHO. The current model uses cost estimates for countries that are not supported by these organizations.86 We do so because Gavi provides vaccine implementation funding to countries it supports, so funding gaps are more likely to exist elsewhere. However, it is possible this over-estimates program costs.
• Number of vaccine doses. The model we use to estimate the impact of HPV vaccination assumes that the vaccines are administered exactly according to WHO guidelines,87 including that all girls receive two doses of vaccine. Realistically, some proportion of girls will only receive one dose.88 Although this is suboptimal from a public health perspective, we are uncertain of its impact on cost-effectiveness.89
• Vaccine type. Arbyn et al. 2018 provides some evidence that the bivalent HPV vaccine is more effective than the quadrivalent vaccine.90 Given that this finding is counterintuitive,91 we are unsure of how to interpret it. Abbas et al. 2020, which is the basis for our cost-effectiveness analysis, models the bivalent and quadrivalent vaccines identically and suggests that the nonavalent vaccine is somewhat more effective than the other two.92 We used estimates for bivalent and quadrivalent vaccines because they are more widely available and because the nonavalent vaccine is not represented in the Arbyn et al. 2018 meta-analysis of HPV vaccine trials.93 If the nonavalent vaccine is available for the same price, it may be more cost-effective than the bivalent and quadrivalent vaccines, but the difference would be modest.94 We are unsure of whether low- and lower-middle income countries would employ bivalent, quadrivalent, or nonavalent vaccines.
• Assumptions underlying the Abbas et al. 2020 model of HPV vaccination benefits. Our cost-effectiveness analysis relies on estimates of the benefit of HPV vaccination that originate from a modeling study, Abbas et al. 2020. This model relies on inputs that we have not evaluated, such as estimates of population age structure, cervical cancer burden, HPV strain prevalence, and sexual activity patterns in 177 countries. We would guess that the data underlying these estimates are not high-quality for certain countries, but we have not investigated this.
• Duration of protection. The Abbas et al. 2020 model assumes that HPV vaccination confers lifelong immunity against the targeted HPV strains.95 Our understanding is that the effectiveness of some vaccines wanes over time.96 Evidence suggests that HPV vaccines provide robust immunity that lasts more than ten years (which is the longest recipients have been followed) without declining significantly.97 Furthermore, protection matters most during the period of life when women engage with the most sexual partners, which we would guess is primarily at younger ages. However, we are uncertain how long immunity will last and whether waning immunity in middle or older age would significantly impact the effectiveness and cost-effectiveness of the vaccine.
• Changing access to cervical cancer screening and treatment services over time. The Abbas et al. 2020 model assumes that access to cervical cancer screening and treatment services does not change over time.98 Since screening and treatment appears to substantially reduce the burden of cervical cancer,99 increasing access over time could reduce the burden of cervical cancer deaths that HPV vaccination can prevent in low- and middle-income countries, and therefore reduce cost-effectiveness. Because an average of nearly five decades elapse between vaccination and the prevention of a cervical cancer death, it seems quite possible that some of the assumed cervical cancer deaths may not occur due to increasing access to preventive care over the next few decades.
• Herd effects and other benefits. By reducing the prevalence of HPV infection in populations, HPV vaccines may protect women and men who have not been vaccinated.100 In addition to the protection they offer against cervical cancer, HPV vaccines also protect against genital warts in women and men,101 and possibly against other HPV-associated cancers in women and men. Neither of these benefits are included in the Abbas et al. 2020 model,102 and we did not include them in our cost-effectiveness analysis. We would guess that including them would not have a large impact on cost-effectiveness.
• Immune response. Populations can respond differently to vaccines due to genetic
  or environmental factors.103 In theory, HPV vaccines could be less effective in some populations. We do not think this is likely to be a major concern, since the Arbyn et al. 2018 meta-analysis represents successful vaccine trials in racially, culturally, and economically diverse populations.104

Does it have room for more funding?

After a limited investigation, we believe there are probably not substantial funding gaps in HPV vaccination in the next few years, mainly because the supply of HPV vaccine doses is not expected to meet demand for several years. However, since the intervention looks potentially promising, we expect to consider investigating room for more funding more deeply in the future when supply constraints are resolved.

After conducting a limited investigation of room for more funding, our main reasons to doubt there is space to productively employ funding in this area are:

• Supply constraints. A WHO market analysis concludes that HPV vaccine supply will be substantially lower than demand until 2023-2024.105 Since vaccine manufacturers require several years to scale up production to meet demand and have already committed to doing so,106 we do not believe additional funding will increase available vaccine doses at this time. The fact that demand currently exceeds supply implies that available vaccine doses are already being administered.
• Past research on vaccine funding. We have investigated vaccination programs in the past, and we have struggled to find opportunities that have additional room for funding.107 Vaccination is widely recognized as an effective intervention and our impression is that it tends to be well-funded.
• HPV vaccination does not appear to be a neglected intervention. Gavi is the primary organization that supports HPV vaccination globally, and it has committed substantial funding to HPV vaccination. Its publicly available funding schedule indicates a commitment of roughly $200 million between 2013 and 2019, which is expected to reach roughly $290 million by 2023.108 Our understanding is that, among low and lower middle income countries, there are only fifteen countries which (i) are not eligible for support from Gavi and (ii) have not implemented an HPV vaccine national program.109 However, a quick review suggests most of them have relatively small populations, or a low HPV burden per-person.110

Since the intervention is potentially promising and supply constraints may resolve after 2023 or 2024, we expect to consider investigating room for more funding further in the future.

Key questions for further investigation

Questions we would ask as part of further investigation include:

• How confident should we be in the claim that HPV is the cause of cervical cancer and in the percentage of cancers that are attributed to specific HPV strains? We find these claims credible based on the evidence we have encountered, but a deeper investigation into the underlying evidence may refine our confidence level.
• To what extent does the current supply shortage constrain room for more funding?
• How does vaccine effectiveness change across bivalent, quadrivalent, or nonavalent vaccines?
• What are the most promising contexts for HPV vaccination?
• Which organizations could use additional funding for HPV vaccination?
• How much confidence should we have in the HPV burden model published by Abbas et al. 2020, including its assumption that reduced HPV prevalence will lead to proportional declines in cervical cancer deaths?
• How will changing access to cervical cancer screening and treatment in low- and middle-income countries impact the cost-effectiveness of HPV vaccination?
• How do herd effects and other possible benefits of HPV vaccination excluded from our model impact the cost-effectiveness of HPV vaccination?

Our process

We began with a prior belief that vaccines in general can be cost-effective interventions. We chose to investigate HPV vaccination because we thought it could be a relatively straightforward, effective, and cost-effective intervention and because of the recent publication of the promising modeling study Abbas et al. 2020. For background on HPV vaccination, we reviewed information on HPV vaccination on the WHO and CDC websites, performed searches in GBD Compare, and consulted papers based on GLOBOCAN data on cervical cancer burden. To evaluate the effectiveness of HPV vaccination, we performed systematic scientific literature searches in PubMed for meta-analyses of HPV vaccine trials and studies directly linking HPV vaccination to changes in cervical cancer incidence and mortality. We had three conversations with HPV vaccine experts.111 We constructed a cost-effectiveness analysis, in part using data from Abbas et al. 2020.

Sources

Document	Source
Abbas et al. 2020	Source (archive)
Abbas et al. 2020, Supplementary appendix 1	Source (archive)
Abbas et al. 2020, Supplementary appendix 2	Source (archive)
Arbyn et al. 2011	Source (archive)
Arbyn et al. 2018	Source
Arbyn et al. 2019	Source (archive)
Arrossi et al. 2017	Source (archive)
Burke et al. 2015	Source (archive)
Campos et al. 2017	Source (archive)
CDC, Clinician FAQ, 2016	Source (archive)
CDC, HPV Vaccine Safety and Effectiveness	Source (archive)
CDC, HPV Vaccine Schedule and Dosing	Source (archive)
CDC, HPV Vaccines and Cervical Precancers	Source (archive)
CDC, Vaccinating Boys and Girls	Source (archive)
Deleré et al. 2014	Source (archive)
Doorbar et al. 2016	Source (archive)
Drolet et al. 2019	Source
Ekwunife et al. 2016	Source (archive)
Gallagher, LaMontagne, and Watson-Jones 2018	Source (archive)
Gavi, Disbursements and commitments	Source (archive)
Gavi, Eligibility	Source (archive)
Gavi, Human papillomavirus vaccine support	Source (archive)
GBD Results Tool search, 2017 deaths by cause	Source (archive)
Gheit et al. 2019	Source (archive)
GiveWell, 2020 Moral Weights	Source
GiveWell, Cervical cancer moral weights - Southeast Asia, 2021	Source
GiveWell, Cervical cancer moral weights - sub-Saharan Africa, 2021	Source
GiveWell, Cost-effectiveness analysis, 2020, Version 2	Source
GiveWell, HPV vaccination CEA, 2021	Source
Goodman 2018	Source (archive)
Guo, Cofie, and Berenson 2018	Source
IARC, Estimated number of deaths in 2018, cervix uteri, females, ages 10-79, by income	Source (archive)
IARC, Estimated number of deaths in 2018, cervix uteri, females, ages 15-79, by country	Source (archive)
Institute for Health Metrics and Evaluation, Global Burden of Disease, GBD Compare, Over 10 deaths due to cervical cancer, Southeast Asia (accessed February 19, 2021)	Source (archive)
Institute for Health Metrics and Evaluation, Global Burden of Disease, GBD Compare, Over 10 deaths due to cervical cancer, sub-Saharan Africa (accessed February 19, 2021)	Source (archive)
Jit et al. 2011	Source (archive)
Jit et al. 2014	Source (archive)
Jit et al. 2014, Supplementary appendix	Source (archive)
Jørgensen, Gøtzsche, and Jefferson 2018	Source
Lehtinen and Dillner 2013	Source (archive)
Li et al. 2010	Source
Lu et al. 2011	Source (archive)
Mawa et al. 2015	Source (archive)
Medeiros et al. 2009	Source (archive)
Mentzer et al. 2015	Source
Ng, Hutubessy, and Chaiyakunapruk 2017	Source (archive)
NIH, HPV and Cancer	Source (archive)
NIH, NCI Dictionary of Cancer Terms, Definition of CIN 2	Source (archive)
Pagliusi and Aguado 2004	Source
PAHO Revolving Fund	Source (archive)
Prendergast 2015	Source
Rambout et al. 2007	Source (archive)
Rey-Ares, Ciapponi, and Pichon-Riviere 2012	Source (archive)
Schiller, Castellsagué, and Garland 2012	Source
The Immunisation Advisory Centre, Efficacy and effectiveness	Source (archive)
The World Bank, World Bank Country and Lending Groups	Source (archive)
Tovey and Soares-Weiser 2018	Source (archive)
Vaccarella et al. 2014	Source (archive)
Walboomers et al. 1999	Source
WHO, Global Market Study, HPV, 2018	Source (archive)
WHO, Global Market Study, HPV, 2019	Source (archive)
WHO, Human papillomavirus (HPV) and cervical cancer, 2019	Source (archive)
WHO, Human papillomavirus vaccines: WHO position paper, 2017	Source (archive)
WHO, Life tables by Country, Bolivia (Plurinational State of)	Source (archive)
WHO, Life tables by WHO region, Africa	Source (archive)
WHO, Metrics: Disability-Adjusted Life Year (DALY)	Source (archive)
WHO, Summary of the WHO Position Paper on Vaccines against Human Papillomavirus (HPV), 2017	Source (archive)
WHO, V3P: Key Findings for HPV, 2018	Source (archive)
WHO, Vaccine in National Immunization Programme Update, 2019	Source
WHO, WHO estimates of Human papillomavirus immunization coverage 2010-2018	Source (archive)
World Bank, population estimates ages 0-14, female	Source (archive)

• 1
  • “Over 200 papillomaviruses have been identified and have been completely sequenced, including more than 150 HPV” Doorbar et al. 2016, Pg 4.
  • ”Papillomaviruses (PVs) infect basal keratinocytes of the mucosal and cutaneous epithelia of both animals (reptiles, birds, marsupials, and others) and humans.” Gheit et al. 2019, Pg 1.

• 2
  • “Human papillomavirus (HPV) is a group of viruses that are extremely common worldwide.
  • There are more than 100 types of HPV, of which at least 14 are cancer-causing (also known as high risk type).
  • HPV is mainly transmitted through sexual contact and most people are infected with HPV shortly after the onset of sexual activity.
  • Cervical cancer is caused by sexually acquired infection with certain types of HPV.
  • Two HPV types (16 and 18) cause 70% of cervical cancers and pre-cancerous cervical lesions.
  • There is also evidence linking HPV with cancers of the anus, vulva, vagina, penis and oropharynx.” WHO, Human papillomavirus (HPV) and cervical cancer, 2019

• 3
  • “Ninety percent of all ICC [invasive cervical cancer] in our present study were positive for HPV DNA, and this percentage has increased over time. Given that HPV is accepted as a necessary cause of ICC, being found in more than 99% of ICC tested under the best conditions, this increase is expected to be related to general improvements in HPV DNA testing protocols and not to any real changes in the contribution of HPV to ICC aetiology.” Li et al. 2010, Pg 933.
  • “Cervical cancer is by far the most common HPV-related disease. Nearly all cases of cervical cancer can be attributable to HPV infection.” WHO, Human papillomavirus (HPV) and cervical cancer, 2019
  • “Cervical cancer is caused by sexually acquired infection with certain types of HPV.” WHO, Human papillomavirus (HPV) and cervical cancer, 2019

• 4

  “The presence of HPV in virtually all cervical cancers implies the highest worldwide attributable fraction so far reported for a specific cause of any major human cancer. The extreme rarity of HPV-negative cancers reinforces the rationale for HPV testing in addition to, or even instead of, cervical cytology in routine cervical screening.” Walboomers et al. 1999, abstract.

• 5

  “HPV vaccines reduce the risk of any CIN2+ from 287 to 106/10,000 (RR 0.37 (0.25 to 0.55), high certainty) and probably reduce any AIS lesions from 10 to 0/10,000 (RR 0.1 (0.01 to 0.76), moderate certainty).” Arbyn et al. 2018, abstract.

• 6
  • Screening and removal programs for cervical precancer in high-income countries are correlated with a substantial suppression of cervical cancer risk.
    • “In the absence of screening, projected incidence rates for 2006–2010 in Nordic countries would have been between 3 and 5 times higher than observed rates. Over 60 000 cases or between 41 and 49% of the expected cases of cervical cancer may have been prevented by the introduction of screening in the late 1960s and early 1970s.” Vaccarella et al. 2014, abstract.
    • “Indeed, the success of Pap-screening has demonstrated that a strategy that removes CIN 2 and CIN 3 lesions substantially reduces the incidence of invasive cervical cancer. Therefore, it is reasonable to assume that a vaccine that prevents the development of CIN 2/3 lesions will likewise result in substantial reductions in cervical cancer incidence.” Pagliusi and Aguado 2004, Pg 572.

• 7
  • “Two HPV types (16 and 18) cause 70% of cervical cancers and pre-cancerous cervical lesions.” WHO, Human papillomavirus (HPV) and cervical cancer, 2019
  • “Cervical cancer is caused by infection with high-risk genotypes of human papillomavirus (HPV). Three HPV vaccines (bivalent, quadrivalent, and nonavalent) are widely available in the global market, and are highly efficacious in preventing persistent infection and associated disease from the high-risk HPV 16/18 genotypes, which cause 70% of all cervical cancers.4–7 In addition to HPV 16/18, the nonavalent vaccine also protects against high-risk HPV types 31/33/45/52/58 which cause 18.5% of HPV-positive cervical cancers.” Abbas et al. 2020, Pg e536.

• 8

  “There is high‐certainty evidence that HPV vaccines protect against cervical precancer in adolescent girls and young women aged 15 to 26. The effect is higher for lesions associated with HPV16/18 than for lesions irrespective of HPV type.” Arbyn et al. 2018, abstract.

• 9

  “The infection with certain HPV types also causes a proportion of cancers of the anus, vulva, vagina, penis and oropharynx, which are preventable using similar primary prevention strategies as those for cervical cancer.” WHO, Human papillomavirus (HPV) and cervical cancer, 2019

• 10

  Abbas et al. 2020 provides estimates of the lifetime risk of cervical cancer incidence and mortality for 177 countries, based on GLOBOCAN data.

  • Figure 1 (Pg e538) suggests that the number of unvaccinated women who die of cervical cancer in these 177 countries is approximately three-quarters as large as the number of unvaccinated women who develop cervical cancer.
  • Abbas et al. 2020, Supplementary appendix 2 (“A7. Cervical cancer burden” tab) includes estimates for individual countries, which suggest that in low-income countries like eSwatini, Zambia, and Malawi, cervical cancer is typically fatal in more than three quarters of cases, while in high-income countries like the United States, France, and Germany, it is typically fatal in approximately half of cases. We speculate that this may relate to the higher quality of medical systems in high-income countries.

• 11

  “Globally, the average age at diagnosis of cervical cancer was 53 years, ranging from 44 years (Vanuatu) to 68 years (Singapore). The global average age at death from cervical cancer was 59 years, ranging from 45 years (Vanuatu) to 76 years (Martinique).” Arbyn et al. 2019, abstract.

• 12
  • GBD Results Tool search, 2017 deaths by cause
  • See a visual representation of proportions of death by cause with GBD Compare.

• 13

  “In 2018, approximately 311 000 women died from cervical cancer; more than 85% of these deaths occurring in low- and middle-income countries.” WHO, Human papillomavirus (HPV) and cervical cancer, 2019

• 14

  “The variations in rates are more striking when the focus is on subcontinents (figures 1, 2). Overall, the lowest incidence burden was observed in western Asia and the lowest mortality burden was observed in Australia–New Zealand (table). Rather modest incidences (ASIR < 10 per 100 000) were also noted in Australia– New Zealand, northern America, western Europe, northern Africa, southern Europe, and northern Europe. The highest burden was observed in southern Africa and eastern Africa. A very high burden of the disease (ASIR ≥15 per 100 000) was also observed in western Africa, Melanesia, middle Africa, Micronesia, southeastern Asia, eastern Europe, the Caribbean, and South America (table).” Arbyn et al. 2019, Pg e196.

• 15

  “Cervical cancer was the leading cause of cancer-related death in women in eastern, western, middle, and southern Africa.” Arbyn et al. 2019, abstract.

• 16

  Arbyn et al. 2019, Pg e195, Figure 3A.

• 17

  “The highest incidence was estimated in Eswatini, with approximately 6.5% of women developing cervical cancer before age 75 years.” Arbyn et al. 2019, abstract.

• 18

  “A very high burden of the disease (ASIR ≥15 per 100 000) was also observed in western Africa, Melanesia, middle Africa, Micronesia, southeastern Asia, eastern Europe, the Caribbean, and South America (table).” Arbyn et al. 2019, Pg e196.

• 19
  • “There are three prophylactic HPV vaccines approved and recommended in the United States, Europe, and many regions and countries: the bivalent (2vHPV; against HPV 16 and 18),9 quadrivalent (4vHPV; against HPV 6, 11, 16, and 18),10-12 and nine valent (9vHPV; against HPV 6, 11, 16, 18, 31, 33, 45, 52 and 58).13-15 These vaccines prevent (for those who are HPV naïve) and reduce the burden of infection of HPV types that are included in the vaccines (HPV vaccine types) overall.” Arrossi et al. 2017, Pg 615.
  • “Currently, three HPV vaccine subtypes are registered: GSK’s Cervarix (HPV2), using the proprietary AS04 adjuvant, and Merck’s Gardasil (HPV4) and Gardasil 9 (HPV9), both using aluminium-containing adjuvant.” WHO, Global Market Study, HPV, 2019, Pg 2.
  • ”Non-cancer causing types of HPV (especially types 6 and 11) can cause genital warts and respiratory papillomatosis (a disease in which tumours grow in the air passages leading from the nose and mouth into the lungs). Although these conditions very rarely result in death, they may cause significant occurrence of disease.” WHO, Human papillomavirus (HPV) and cervical cancer, 2019
  • “Cervical cancer is caused by infection with high-risk genotypes of human papillomavirus (HPV). Three HPV vaccines (bivalent, quadrivalent, and nonavalent) are widely available in the global market, and are highly efficacious in preventing persistent infection and associated disease from the high-risk HPV 16/18 genotypes, which cause 70% of all cervical cancers.4–7 In addition to HPV 16/18, the nonavalent vaccine also protects against high-risk HPV types 31/33/45/52/58 which cause 18.5% of HPV-positive cervical cancers.” Abbas et al. 2020, Pg e536.

• 20

  “Demand in 2019 is primarily concentrated on HPV4 with an estimated 60% market share by volume, followed by HPV9 (30%) and HPV2 (10%).” WHO, Global Market Study, HPV, 2019, Pg 2.

• 21

  “Three products are currently in advanced clinical development: two HPV2 vaccines from Innovax and Shanghai Zerun Biotech – the first currently undergoing registration and the second in Phase III – and one HPV4 vaccine from Serum Institute of India, also currently in Phase III. All use aluminium-containing adjuvants and should be licensed with an indication for girls 9 to 14 years old for both 2- and a 3-dose schedules. The success, timing and capacity of these pipeline vaccine efforts will have an important impact on the long-term outlook for HPV vaccine supply.” WHO, Global Market Study, HPV, 2019, Pg 4.

• 22

  “The first global recommendation on HPV vaccination was proposed by the WHO’s SAGE (Strategic Advisory Group of Experts) on Immunization in October 2008 [3], whereby HPV vaccination was recommended in all girls aged 9–13 years old. This recommendation was then updated in April 2014 [4], with the emphasis to include extended 2-dose HPV immunization for girls aged 9–14 years, who were not immunocompromised” Ng, Hutubessy, and Chaiyakunapruk 2017 Pg 2

• 23

  A review of randomized controlled trials on Gardasil (quadrivalent) and Cervarix (bivalent) suggests that these vaccines are not effective at preventing infection or genital disease among women infected with the targeted HPV subtypes at baseline.

  • “Although the clinical trials were primarily designed to evaluate immunoprophylaxis, the fact that women who had prevalent cervicovaginal infection or low grade disease were not excluded at entry provided a cohort to evaluate therapeutic efficacy. In the CVT, time to clearance of prevalent infection was evaluated. There was no difference in the rate of clearance of vaccine or non-vaccine types in Cervarix® vaccines and control [37]. For example, 48.9% and 49.8% of HPV16/18 infections were cleared after 12 months in vaccine recipients and controls, respectively. The therapeutic activity of Gardasil® was evaluated in FUTURE II [15]. No significant difference in the rate of progression of HPV16/18 infection to CIN2+ was observed in VLP vaccines versus controls, 11.1% and 11.9%, respectively. Thus the VLP vaccines do not appear to alter the course of established cervicovaginal HPV infection or disease.” Schiller, Castellsagué, and Garland 2012, Pg 12.

• 24

  “Some countries have started to vaccinate boys as the vaccination prevents genital cancers in males as well as females, and two available vaccines also prevent genital warts in males and females. WHO recommends vaccination for girls aged between 9 and 14 years, as this is the most cost-effective public health measure against cervical cancer.” WHO, Human papillomavirus (HPV) and cervical cancer, 2019

• 25

  Women suffer a larger health burden from HPV infections than men, and since most sexual activity is between women and men, vaccinating girls also protects men.

• 26
  • The CDC recommends a two-dose schedule for young adults before 15 and a three-dose schedule for those receiving the vaccination after their 15th birthday, and for those suffering from certain immunocompromising conditions. WHO recommended a three-dose schedule until 2014, when it switched to a recommended schedule of two doses.
    • “For girls and boys starting the vaccination series before the 15th birthday, the recommended schedule is 2 doses of HPV vaccine. … CDC continues to recommend a 3-dose schedule for persons starting the HPV vaccination series on or after the 15th birthday, and for persons with certain immunocompromising conditions.” CDC, Clinician FAQ, 2016
  • “In addition, WHO’s decision to switch from a recommended schedule of three doses to two doses helps to facilitate country roll-outs and reduce costs.” Gavi, Human papillomavirus vaccine support
  • “The first global recommendation on HPV vaccination was proposed by the WHO’s SAGE (Strategic Advisory Group of Experts) on Immunization in October 2008 [3], whereby HPV vaccination was recommended in all girls aged 9–13 years old. This recommendation was then updated in April 2014 [4], with the emphasis to include extended 2-dose HPV immunization for girls aged 9–14 years, who were not immunocompromised” Ng, Hutubessy, and Chaiyakunapruk 2017 Pg 1

• 27
  • “Current evidence suggests that the 3 licensed HPV vaccines have relatively similar effectiveness in preventing cervical cancer.” WHO, Human papillomavirus vaccines: WHO position paper, 2017, Pg 256.
  • “With the updated model, the bivalent or quadrivalent HPV vaccine was estimated to avert 15 cases, 12 deaths, and 243 DALYs per 1000 vaccinated girls, and the nonavalent HPV vaccine was estimated to avert 19 cases, 14 deaths, and 306 DALYs per 1000 vaccinated girls.” Abbas et al. 2020, abstract.

• 28

  We conducted a search in PubMed on April 22, 2020, using the following search terms: papillomavirus AND vaccine AND cervical AND cancer AND (incidence OR mortality). Due to the large number of results (4,019), we narrowed the search parameters by activating the “humans” filter and restricting the search to the most recent five years. We believe studies on incidence and mortality would have been published recently because the vaccine was not recommended for routine use in US girls until 2006 and cervical cancer risk peaks in middle age. Due to the large number of results (1,236), we activated the “best match” setting and considered the first 40 results. To search specifically for RCT follow-up studies, we then conducted an additional search identical to the previous one, but with the “clinical trials” filter activated.

• 29

  In addition to the lack of pertinent results in our scientific literature search, a 2018 meta-analysis of RCTs of HPV vaccination concluded that “Studies were not large enough or of sufficient duration to evaluate cervical cancer outcomes.” Arbyn et al. 2018, abstract.

• 30

  ”Since 2006, human papillomavirus vaccine has been recommended for young females in the U.S.” Guo, Cofie, and Berenson 2018, abstract.

• 31

  “This cross-sectional study used data from the National Program for Cancer Registries and Surveillance, Epidemiology, and End Results Incidence–U.S. Cancer Statistics 2001–2014 database for U.S. females aged 15–34 years. This study compared the 4-year average annual incidence of invasive cervical cancer in the 4 years before human papillomavirus vaccine was introduced (2003–2006) and the 4 most recent years in the vaccine era (2011–2014).” Guo, Cofie, and Berenson 2018, abstract.

• 32

  “The 4-year average annual incidence rates for cervical cancer in 2011–2014 were 29% lower than that in 2003–2006 (6.0 vs 8.4 per 1,000,000 people, rate ratio=0.71, 95% CI=0.64, 0.80) among females aged 15–24 years, and 13.0% lower among females aged 25–34 years.” Guo, Cofie, and Berenson 2018, abstract.

• 33
  • Figure 1 suggests that the trend in incidence prior to 2006 looks similar to the trend after 2006. Guo, Cofie, and Berenson 2018, figure 1.
  • This downward trend may be related to other factors mentioned in the discussion of Guo, Cofie, and Berenson 2018, such as increased Pap testing. “Trends in cervical cancer incidence and mortality reduction in the U.S. are widely credited to regular Pap testing for precancerous lesions and subsequent follow-up for treatment.” Guo, Cofie, and Berenson 2018, Pg 6.

• 34

  “When Should My Child Get the HPV Vaccine? Dose #1: 11–12 years. Dose #2: 6–12 months after the first dose” CDC, Vaccinating Boys and Girls.

• 35

  “Globally, the average age at diagnosis of cervical cancer was 53 years, ranging from 44 years (Vanuatu) to 68 years (Singapore). The global average age at death from cervical cancer was 59 years, ranging from 45 years (Vanuatu) to 76 years (Martinique).” Arbyn et al. 2019, abstract.

• 36

  “This study included data on young females (aged 15–34 years) from USCS [U.S. Cancer Statistics].” Guo, Cofie, and Berenson 2018, Pg 2.

• 37
  • ”Last year, the World Health Organization (WHO) convened a gathering of experts, including scientists, national regulatory authorities, industry representatives, epidemiologists and government officials from both developed and developing countries1 to discuss appropriate endpoint measurements for HPV vaccine efficacy and effectiveness trials.” Pagliusi and Aguado 2004, abstract.
  • “An endpoint of invasive cancer provides the most valid endpoint, but also presents the largest feasibility problems. Requiring this clinical endpoint could delay vaccine availability by decades and would be ethically controversial, because in the context of accepted recommendations for treating high-grade cervical dysplasia [16], invasive cervical cancer should normally be encountered very rarely, if at all, in an experimental setting. The consultation generally agreed that surrogate markers of pre-cancer would be more appropriate endpoints for vaccine efficacy studies at present.” Pagliusi and Aguado 2004, Pg 572.

• 38

  “Evaluation of HPV vaccine efficacy using prevention of CIN2/3 dysplasias and cancer was recommended as the globally accepted endpoint for population-based studies.” Pagliusi and Aguado 2004, Pg 577.

• 39

  “CIN 1 (or low-grade dysplasia) is seen as the most common clinical manifestation of cervical HPV infections. CIN 1 lesions are accompanied by a high rate of clinical regression, as around 60% of low-grade dysplasias resolve without the need for treatment, and about 10% can progress to CIN 2 and 3 [17–19]. CIN 2 (moderate-grade dysplasia) also shows a high rate of regression, but women with CIN 2 are still at substantial risk for cervical cancer. A meta-analysis of cervical dysplasia natural history studies estimated that 22% of CIN 2 lesions progressed to CIN 3 without treatment [17].2 CIN 3 more often persists or progresses to cancer, regression being less common. Women with CIN 3 are at substantial risk for invasive cervical cancer [17]. CIN 3 includes high-grade pre-cancerous lesions and carcinoma in situ and is well accepted as the pathological state that immediately precedes invasive cervical cancer.” Pagliusi and Aguado 2004, Pg 571-572.

• 40

  “Indeed, the success of Pap-screening has demonstrated that a strategy that removes CIN 2 and CIN 3 lesions substantially reduces the incidence of invasive cervical cancer. Therefore, it is reasonable to assume that a vaccine that prevents the development of CIN 2/3 lesions will likewise result in substantial reductions in cervical cancer incidence.” Pagliusi and Aguado 2004, Pg 572.

• 41

  We performed a search in PubMed on May 4 using the following search terms: papillomavirus* AND cervi* AND cancer AND vaccin*. We activated the “meta-analysis” filters and only considered studies published within the last five years. This search returned 18 results, all of which we examined for relevance. We also examined the “similar articles” suggested by PubMed in the sidebar of the pages of the studies we identified.

• 42

  It is pertinent because it reports pooled outcomes for CIN2 and CIN3 risk in RCTs of HPV vaccination. We believe it is high-quality in part because it is a Cochrane Collaboration meta-analysis of RCTs that tend to have high-quality designs like placebo controls and double blinding. We created a checklist for lightly assessing the methodological quality of this meta-analysis. Overall, we judge the paper to provide reasonably strong evidence that HPV vaccine decreases rates of cervical precancer. Factors that increase our confidence include:

  • The review was pre-registered.
  • The review only included randomised controlled trials.
  • The review included a substantial number of trials, and the sample size was large overall.
  • The review assessed quality of evidence across a comprehensive set of criteria.

• 43
  • Arbyn et al. 2018 presented findings for three different categories of study participants:
    1. ”Adolescent girls and women who were negative for [high-risk] HPV DNA at baseline
    2. Adolescent girls and women who were negative for HPV 16/18 DNA at baseline
    3. Adolescent girls and women regardless of HPV DNA status at baseline” Pg 16.
  • We focus on the first category of participants.

• 44

  “Most randomised trials assessing vaccine efficacy enrolled younger women, in the age range 15 to 26 years (Table 7). Only three randomised controlled trials (RCTs) evaluated the efficacy of the vaccines (FUTURE III trial (ph3,4v), VIVIANE trial (ph3,2v); Chinese trial (ph3,2v)_mid-adult) in mid-adult women (aged 24 to 45 years).” Arbyn et al. 2018, Pg 36.

• 45
  • See Arbyn et al. 2018, Pg 5. Summary of findings table, “HPV vaccine effects on cervical lesions in adolescent girls and women negative for hrHPV DNA at baseline,” “Any CIN2+ irrespective of HPV type, bivalent or quadrivalent vaccine.”
  • We calculate the reduction in risk and 95% confidence interval by subtracting the risk ratio from 1. (e.g., 1 - 0.37 = 0.63)

• 46
  • See Arbyn et al. 2018, Pg 6. Summary of findings table, “HPV vaccine effects on cervical lesions in adolescent girls and women negative for hrHPV DNA at baseline,” “Any CIN3+ irrespective of HPV type, bivalent or quadrivalent vaccine.”
  • We calculate the reduction in risk and 95% confidence interval by subtracting the risk ratio from 1. (e.g., 1 - 0.21 = 0.79)

• 47

  “There is high-certainty evidence that vaccines lower CIN2+ from 164 to 2/10,000 (RR 0.01 (0 to 0.05)) and CIN3+ from 70 to 0/10,000 (RR 0.01 (0.00 to 0.10).” Arbyn et al. 2018, abstract.

• 48

  See Arbyn et al. 2018, Pg 5, Summary of findings table, “HPV vaccine effects on cervical lesions in adolescent girls and women negative for hrHPV DNA at baseline,” “CIN2+ associated with HPV16/18” and “CIN3+ associated with HPV16/18.”

• 49
  • “Two HPV types (16 and 18) cause 70% of cervical cancers and pre-cancerous cervical lesions.” WHO, Human papillomavirus (HPV) and cervical cancer, 2019
  • “There are three prophylactic HPV vaccines approved and recommended in the United States, Europe, and many regions and countries: the bivalent (2vHPV; against HPV 16 and 18),9 quadrivalent (4vHPV; against HPV 6, 11, 16, and 18),10-12 and nine valent (9vHPV; against HPV 6, 11, 16, 18, 31, 33, 45, 52 and 58).13-15 These vaccines prevent (for those who are HPV naïve) and reduce the burden of infection of HPV types that are included in the vaccines (HPV vaccine types) overall.” Arrossi et al. 2017, Pg 615.

• 50
  • “The size of reduction in CIN3+ with vaccines differed between bivalent and quadrivalent vaccines (bivalent: RR 0.08 (0.03 to 0.23), high certainty; quadrivalent: RR 0.54 (0.36 to 0.82), moderate certainty).” Arbyn et al. 2018, abstract.
  • This finding is somewhat counterintuitive since the quadrivalent vaccine protects against a greater number of cancer-causing HPV strains. We have not evaluated the evidence underlying this finding.

• 51

  “The risk of serious adverse effects was similar in those vaccinated and those who received placebo or control vaccine (RR 0.98, 95% CI 0.92 to 1.05; participants = 71,597; studies = 23; I2 = 6%)) high-quality evidence).” Arbyn et al. 2018, Pg 28

• 52

  “The Cochrane review conducted trial searches up until June 2017 and included 26 randomised trials with 73 428 women.1 In January 2018, we published an index of the study programmes of the HPV vaccines that included 206 comparative studies.3 As of June 2017, about one-third of the 206 studies were not published and half of the completed studies listed on ClinicalTrials.gov had no results posted.3 Although we sent our index to the Cochrane group handling the Cochrane review, the review stated that, ‘nearly all end-of-study reports have been published in the peer-reviewed literature’. When we applied the Cochrane review’s inclusion criteria to the 206 studies, we identified 46 completed and eligible trials. The number of randomised participants could be assessed for 42 of the 46 trials and was 121 704. With nearly half of the trials and half of the participants missing, the Cochrane authors’ conclusion, ‘that the risk of reporting bias may be small’, was inappropriate. Fifteen of the 20 additional trials were listed on ClinicalTrials.gov; the Cochrane authors would therefore have identified more trials if they had searched ClinicalTrials.gov in more depth and searched additional trial registers (we searched 45 trial registers).” Jørgensen, Gøtzsche, and Jefferson 2018, Pg 165.

• 53

  Arbyn et al. 2018 has been criticised on the basis that it only includes published papers; more details on the criticism can be found here. We have not assessed these concerns in depth. To further evaluate the reliability of the estimate from Arbyn et al. 2018, we have conducted a brief search for other systematic reviews estimating the effectiveness of HPV vaccines on Google Scholar and SumSearch. Details can be found in the ‘Vaccine efficacy’ tab in this spreadsheet. The findings of these systematic reviews are broadly similar to those of Arbyn et al. 2018 (see above). On the basis of this result, we would guess Arbyn et al. 2018 is unlikely to provide a substantially biased estimate for our preferred specification.

• 54

  “HPV vaccine effects on cervical lesions in adolescent girls and women who are hrHPV DNA negative at baseline” … “Setting: Europe, Asia Pacific countries, South & North America.” Arbyn et al. 2018, Summary of Findings table, Pgs 5-6.

• 55
  • “Ninety percent of all ICC [invasive cervical cancer] in our present study were positive for HPV DNA, and this percentage has increased over time. Given that HPV is accepted as a necessary cause of ICC, being found in more than 99% of ICC tested under the best conditions,12 this increase is expected to be related to general improvements in HPV DNA testing protocols and not to any real changes in the contribution of HPV to ICC aetiology.” Li et al. 2010, Pg 933.
  • “Cervical cancer is by far the most common HPV-related disease. Nearly all cases of cervical cancer can be attributable to HPV infection.” WHO, Human papillomavirus (HPV) and cervical cancer, 2019
  • “Cervical cancer is caused by sexually acquired infection with certain types of HPV.” WHO, Human papillomavirus (HPV) and cervical cancer, 2019

• 56
  • The meta-analysis of randomized controlled trials of HPV vaccination Arbyn et al. 2018 reports near-complete (99%) protection against precancer associated with HPV 16 and 18 in vaccinated women who were not infected with high-risk strains at baseline. These strains are targeted by all HPV vaccines included in the trials.
    • “There is high-certainty evidence that vaccines lower CIN2+ from 164 to 2/10,000 (RR 0.01 (0 to 0.05)) and CIN3+ from 70 to 0/10,000 (RR 0.01 (0.00 to 0.10).” Abstract.
  • “Studies suggest that HPV vaccines offer long-lasting protection against HPV infection and therefore disease caused by HPV infection. Studies of the bivalent and quadrivalent vaccines have followed vaccinated individuals for more than 10 years and have found no evidence of protection decreasing over time. Duration of protection provided by HPV vaccination will continue to be studied.” CDC, HPV Vaccine Safety and Effectiveness.

• 57
  • “In the absence of screening, projected incidence rates for 2006–2010 in Nordic countries would have been between 3 and 5 times higher than observed rates. Over 60 000 cases or between 41 and 49% of the expected cases of cervical cancer may have been prevented by the introduction of screening in the late 1960s and early 1970s.” Vaccarella et al. 2014, abstract.
  • “Indeed, the success of Pap-screening has demonstrated that a strategy that removes CIN 2 and CIN 3 lesions substantially reduces the incidence of invasive cervical cancer. Therefore, it is reasonable to assume that a vaccine that prevents the development of CIN 2/3 lesions will likewise result in substantial reductions in cervical cancer incidence.” Pagliusi and Aguado 2004, Pg 572.

• 58

  Arbyn et al. 2018 appendix 6 (Pgs 164-169) contains a list of trial locations, which includes North and South America, Oceania, Africa, South and East Asia, and Europe.

• 59

  “The Papillomavirus Rapid Interface for Modelling and Economics (PRIME) is a model to assess the health impact and cost-effectiveness of HPV vaccination of girls for prevention of cervical cancer.11,12 It was developed in 2014 in collaboration with WHO and has been validated against 17 published studies set in low-income and middle-income countries. It is endorsed by the WHO Immunization and Vaccines Implementation Research Advisory Committee to provide a conservative estimate of the health impact and cost-effectiveness of vaccinating girls before sexual debut.11,13 PRIME has been used to inform the impact of vaccine investments by Gavi, the Vaccine Alliance, in 97 countries, as well as at the national level.” Abbas et al. 2020, Pg e537.

• 60

  “PRIME is a static proportional impact model that estimates the impact of both single-age and multiple-age cohort vaccination. Vaccination impact is estimated in terms of reduction in age-dependent cervical cancer incidence, prevalence, and mortality in direct proportion to vaccine coverage, vaccine efficacy, and distribution of high-risk HPV types (HPV 16/18 for the bivalent and quadrivalent vaccines and HPV 16/18/31/33/45/52/58 for the nonavalent vaccine)... PRIME data inputs include country and age-specific cervical cancer incidence, prevalence, and mortality among females from the Global Cancer Incidence, Mortality and Prevalence (GLOBOCAN) database… PRIME uses country-specific life tables to model the annual population size of longitudinal birth cohorts.” Abbas et al. 2020, Pg e537-538.

• 61
  • “Vaccination impact is estimated in terms of reduction in age-dependent cervical cancer incidence, prevalence, and mortality in direct proportion to vaccine coverage, vaccine efficacy, and distribution of high-risk HPV types (HPV 16/18 for the bivalent and quadrivalent vaccines and HPV 16/18/31/33/45/52/58 for the nonavalent vaccine).” Abbas et al. 2020, Pg e537.
  • The passage above cites Li et al. 2010, which is a meta-analysis that reports the regional prevalence of HPV types in 30,848 invasive cervical cancers worldwide.

• 62

  “Vaccinating girls before sexual debut fully protects them from developing cervical cancer caused by high-risk HPV types, in accordance with the efficacy observed in vaccine trials.” Abbas et al. 2020, Pg e537.

• 63

  The meta-analysis of RCTs of HPV vaccination Arbyn et al. 2018 reports near-complete (99%) protection against precancer associated with HPV 16 and 18 in vaccinated women who were not infected with high-risk strains at baseline. These strains are targeted by all HPV vaccines included in the trials.

  • “There is high-certainty evidence that vaccines lower CIN2+ from 164 to 2/10,000 (RR 0.01 (0 to 0.05)) and CIN3+ from 70 to 0/10,000 (RR 0.01 (0.00 to 0.10).” Arbyn et al. 2018, abstract.

• 64
  • The reduction is vs. a counterfactual scenario of no vaccination. The two-thirds estimate comes from comparing mortality estimates for the pre-vaccination and vaccination scenarios in Abbas et al. 2020, Pg e538, Figure 1.
  • “We estimated the lifetime impact of routine vaccination of girls at 90% coverage (as recommended in the WHO cervical cancer elimination strategy) for the 10-year period of 2020–29 in comparison with the counterfactual scenario of no vaccination in 177 countries for bivalent or quadrivalent and nonavalent vaccination of 9-year-old and 12-year-old girls.” Abbas et al. 2020, Pg e539.

• 65

  Abbas et al. 2020, Supplementary appendix 1, Pg 141.

• 66

  Abbas et al. 2020, Supplementary appendix 2, tab “HPV vaccination impact.”

• 67
  • “[T]he aims of this systematic review and meta-analysis are to: (1) update and summarise the most recent evidence about the population-level impact of girls-only HPV vaccination on HPV infections and anogenital wart diagnoses among girls, boys, women, and men; (2) summarise new evidence about the population-level impact of girls-only HPV vaccination on CIN2+ occurrence among screened girls and women; and (3) compare the population level impact of HPV vaccination on anogenital wart diagnoses and CIN2+ occurrence between countries that have implemented either a single or a multiple age-cohort vaccination strategy.” Drolet et al. 2019, Pg 3.
  • “We identified 1702 potentially eligible articles for this systematic review and meta-analysis, and included 65 articles in 14 high-income countries: 23 for HPV infection, 29 for anogenital warts, and 13 for CIN2+.” Drolet et al. 2019, abstract.

• 68
  • “After 5-8 years of vaccination, the prevalence of HPV 16 and 18 decreased significantly by 83% (RR 0.17, 95% CI 0.11-0.25) among girls aged 13-19 years, and decreased significantly by 66% (RR 0.34, 95% CI 0.23-0.49) among women aged 20-24 years. The prevalence of HPV 31, 33, and 45 decreased significantly by 54% (RR 0.46, 95% CI 0.33-0.66) among girls aged 13-19 years.” Drolet et al. 2019, abstract.
  • “After 5-9 years of vaccination, CIN2+ decreased significantly by 51% (RR 0.49, 95% CI 0.42-0.58) among screened girls aged 15-19 years and decreased significantly by 31% (RR 0.69, 95% CI 0.57-0.84) among women aged 20-24 years.” Drolet et al. 2019, abstract.

• 69

  “Anogenital wart diagnoses decreased significantly by 67% (RR 0.33, 95% CI 0.24-0.46) among girls aged 15-19 years, decreased significantly by 54% (RR 0.46, 95% CI 0.36-0.60) among women aged 20-24 years, and decreased significantly by 31% (RR 0.69, 95% CI 0.53-0.89) among women aged 25-29 years. Among boys aged 15-19 years anogenital wart diagnoses decreased significantly by 48% (RR 0.52, 95% CI 0.37-0.75) and among men aged 20-24 years they decreased significantly by 32% (RR 0.68, 95% CI 0.47-0.98).” Drolet et al. 2019, abstract.

• 70

  Drolet et al. 2019, Pg 24, Figure 4B.

• 71

  “All the vaccines were consistently associated with short-term local adverse effects (RR 1.18, 95% CI 1.16 to 1.20; participants = 18,113; studies = 8; I2 = 93%; moderate-quality evidence; Analysis 7.1), such as pain at the injection site (RR 1.35, 95% CI 1.23 to 1.49; participants = 25,691; studies = 13; I2 = 98%; moderate-quality evidence; Analysis 7.2), local swelling (RR 1.73, 95% CI 1.32 to 2.27; participants = 22,106; studies = 9; I2 = 95%; moderate-quality evidence; Analysis 7.3) and redness (RR 1.72, 95% CI 1.50 to 1.97; participants = 19,996; studies = 6; I2 = 82%; moderate-quality evidence; Analysis 7.4).” Arbyn et al. 2018, Pg 31.

• 72
  • “The risk of serious adverse effects was similar in those vaccinated and those who received placebo or control vaccine (RR 0.98, 95% CI 0.92 to 1.05; participants = 71,597; studies = 23; I2 = 6%)) high-quality evidence).” Arbyn et al. 2018, Pg 31.
  • “Similar rates of normal term deliveries of a healthy infant were noted (RR 1.00, 95% CI 0.97 to 1.02; participants = 8782; studies = 8; I2 = 0%; Analysis 8.1). The risk of miscarriage also was similar between HPV vaccinees and control vaccinees (RR 0.88, 95% CI 0.68 to 1.14; participants = 8,618; studies = 9; I2 = 78%; Analysis 8.2), as was elective termination of pregnancy (RR 0.90, 95% CI 0.80 to 1.02; participants = 10,909; studies = 9; I2 = 0%; Analysis 8.3). Analyses of still births and abnormal infants lack sufficient power to rule out small increases or decreases in risk. The observed risk of stillbirth of 70 per 10,000 translates to a rate of 78 per 10,000 (48 to 128) based on the RR of 1.12 (95% CI 0.68 to 1.83; Analysis 8.4). The observed risk of an abnormal infant in the control groups was 205 per 10,000 and in the vaccination arms was 250 per 10,000 (180 to 346) based on the RR of 1.22 (0.88 to 1.69) (Analysis 8.5). We downgraded the quality of evidence for both of these outcomes to moderate due to imprecision.” Arbyn et al. 2018, Pg 35.

• 73

  “Mortality during the study follow-up period in HPV vaccine recipients and control or placebo groups was reported in 23 trials (Analysis 7.7). We could not exclude an increased risk of mortality among vaccinated women (RR 1.29, 95% CI 0.85 to 1.98; participants = 71,176; studies = 23; I2 = 0%). In absolute terms the rate of deaths in the control groups was 11 per 10,000 whereas in HPV vaccinated women the rate observed was between 9 and 22 women per 10,000. The difference between the bi- and quadrivalent vaccine was not significant (P = 0.62, I2 = 0%). Again, results were very similar when data extraction was restricted to peer-reviewed published reports (RR 1.31, 95% CI 0.84 to 2.05; participants = 71,452; studies = 23; I2 = 0%, Figure 11). We downgraded the quality of evidence for mortality to low. This was due to imprecision from the wide confidence interval and inconsistency due to a statistically different risk between the two age cohorts, with a higher risk of mortality in older women.” Arbyn et al. 2018, Pg 33.

• 74
  • “Only three randomised controlled trials (RCTs) evaluated the efficacy of the vaccines (FUTURE III trial (ph3,4v), VIVIANE trial (ph3,2v); Chinese trial (ph3,2v)_mid-adult) in mid-adult women (aged 24 to 45 years).” Arbyn et al. 2018, Pg 35.
  • “When all the deaths among mid-adult women enrolled in the three trials are pooled, a higher case fatality rate was observed among those who received HPV vaccine compared to those who received placebo: (RR 2.36, 95% CI 1.10 to 5.03; participants = 10,737; studies = 3; I2 = 0%), with no differences between different HPV vaccines (P = 0.73).” Arbyn et al. 2018, Pg 33.

• 75

  We believe there was not a statistically significant increase in younger women because it was not mentioned alongside the subgroup finding in mid-adult women, but this is not stated explicitly. See Arbyn et al. 2018, “Summary of findings 3” table, “Deaths,” as well as the discussion on pages 33 and 35.

• 76

  “We downgraded the quality of evidence for mortality to low. This was due to imprecision from the wide confidence interval and inconsistency due to a statistically different risk between the two age cohorts, with a higher risk of mortality in older women.” Arbyn et al. 2018, Pg 33.

• 77

  “Limitations of subgroup analyses are well established—false positives due to multiple comparisons, false negatives due to inadequate power, and limited ability to inform individual treatment decisions because patients have multiple characteristics that vary simultaneously. It remains uncertain when subgroup analyses should influence clinical practice.” Burke et al. 2015, Pg 1.

• 78

  “The higher number of deaths in the vaccine arms among mid-adult women may be a chance occurrence, since there was no pattern either in the causes of death, or in the timing of the occurrence of death (period between vaccine administration and date of death). In the study reports, none of the deaths were deemed to be related to vaccination.” Arbyn et al. 2018, Pg 33.

• 79

  See our cost-effectiveness analysis of HPV vaccination, “HPV Africa” sheet, “HPV vaccination vs cash” row.

• 80
  • Abbas et al. 2020 estimates a reduction in total cervical cancer mortality of about two-thirds.
    • Modeling estimates based on the vaccine’s effect on HPV infection suggest that at 90 percent coverage, HPV vaccination will reduce global deaths from cervical cancer by approximately two-thirds. The reduction is vs. a counterfactual scenario of no vaccination. The two-thirds estimate comes from comparing mortality estimates for the pre-vaccination and vaccination scenarios in Figure 1 on Pg e538.
    • “We estimated the lifetime impact of routine vaccination of girls at 90% coverage (as recommended in the WHO cervical cancer elimination strategy) for the 10-year period of 2020–29 in comparison with the counterfactual scenario of no vaccination in 177 countries for bivalent or quadrivalent and nonavalent vaccination of 9-year-old and 12-year-old girls.” Pg e539.
  • Arbyn et al. 2018 estimates reducations in CIN2 and CIN3 lesions of 63% and 79%, respectively.
    • Among women who were not infected with high-risk HPV strains at baseline, vaccination with the bivalent or quadrivalent HPV vaccine reduced the risk of developing CIN2 by 63 percent (95% confidence interval 45% to 75%). Five trials, high evidence certainty, two- to six-year follow-up. See Pg 5, “Summary of findings” table, “HPV vaccine effects on cervical lesions in adolescent girls and women negative for hrHPV DNA at baseline,” “Any CIN2+ irrespective of HPV type, bivalent or quadrivalent vaccine.”
    • Among women who were not infected with high-risk HPV strains at baseline, vaccination with the bivalent or quadrivalent HPV vaccine tended to reduce the risk of developing CIN3 by 79 percent, but this was not statistically significant (95% confidence interval -10% to 96%). Three trials, moderate evidence certainty, 3.5- to four-year follow-up. See Pg 6, “Summary of findings” table, “HPV vaccine effects on cervical lesions in adolescent girls and women negative for hrHPV DNA at baseline,” “Any CIN3+ irrespective of HPV type, bivalent or quadrivalent vaccine.”

• 81

  Arbyn et al. 2019, Pg e194, Figure 2.

• 82

  For example, compare the “Total units of value per $100,000 donated to HPV vaccination program” on the “HPV Africa” tab to that of the “HPV SE Asia” tab.

• 83

  See our back-of-the-envelope calculation here. It suggests that HPV vaccination of a cohort of 10,000 girls in sub-Saharan Africa would avert approximately 189 deaths, compared with the PRIME figure of 220 deaths.

  • See Abbas et al. 2020, Supplementary appendix 1, Pg 141. The PRIME model predicts that HPV vaccination would avert 22 deaths per 1,000 12-year-old girls vaccinated with the bivalent or quadrivalent vaccine in Africa. (We have multiplied the number by 10 to represent a 10,000-person cohort.)

• 84

  Estimates we have identified in a brief review range from $1.80 to $1233 per person. This range comes from Ekwunife et al. 2016, Table 1, Pgs 70-71. Other estimates we have found fall within this range:

  • Based on data in Jit et al. 2014, we calculated an estimate of about $20 per person for low-income countries. (See Table1, Pg e409. Divide $190m (USD) vaccination cost by 9.7m vaccinated girls.)
    • The Jit et al. 2014, Supplementary appendix mentions that costs include both procurement and delivery cost: “Vaccination costs… Total of vaccine procurement costs and delivery costs.” Pg 4.
  • Campos et al. 2017, a study modelling the intervention in LIC and LMICs, assumes costs between $12 and $16 (Pg 52, Table1).

• 85
  • “Thanks to the Vaccine Alliance, the lowest-income countries now have access to HPV vaccines for as little as US$ 4.50 per dose. The same vaccines can cost more than US$ 100 in high-income countries, and the previous lowest public sector price was US$ 13 per dose.” Gavi, Human papillomavirus vaccine support
  • “For 35 years, the Revolving Fund of the Pan American Health Organization has helped the countries of the Americas protect people against some of the world's worst diseases, including polio, measles, yellow fever, rotavirus and HPV. Through the fund, Member States pool their national resources to procure high-quality life-saving vaccines and related products at the lowest price" PAHO Revolving Fund
  • “Reported price per dose14 of HPV vaccines is tiered by procurement method and income group, with Gavi (UNICEF Supply Division [SD]) and PAHO Revolving Fund (RF) paying the lowest median prices, at $4.55 and $9.15, respectively. The non-Gavi MIC (UNICEF- and self-procuring) median prices for both HPV2 and HPV4 are ~3X the Gavi price while HICs pay ~7X the Gavi price” WHO, Global Market Study, HPV, 2018, Pg 3.
  • “Average non-Gavi UNICEF and self-procuring MICs prices 3X Gavi and ~1.5X PAHO” Goodman 2018

• 86

  See the values we use for intervention costs here.

• 87

  “The model assumes a two-dose schedule with perfect timeliness to the target ages given in the coverage estimates.” Abbas et al. 2020, Pg e537.

• 88
  • In countries with sufficient data (such as Germany, Spain, and Fiji), WHO estimates show that the percentage of girls who have received at least one dose by age 15 is higher than the percentage of girls who have received two doses by age 15.
  • See WHO, WHO estimates of Human papillomavirus immunization coverage 2010-2018. Compare values for “15HPV1_F” (Population turning 15 in the reporting year that received any time between age 9- 14 at least one dose of HPV vaccine) and “15HPVc_F” (Population turning 15 in the reporting year that received any time between age 9- 14 the full recommended schedule of HPV vaccine).

• 89

  It may increase or decrease cost-effectiveness, depending on whether the benefit from one dose is greater or smaller than half the benefit from two doses. We have not investigated this.

• 90
  • “The size of reduction in CIN3+ with vaccines differed between bivalent and quadrivalent vaccines (bivalent: RR 0.08 (0.03 to 0.23), high certainty; quadrivalent: RR 0.54 (0.36 to 0.82), moderate certainty).” Arbyn et al. 2018, abstract.
  • This finding is somewhat counterintuitive since the quadrivalent vaccine protects against a greater number of cancer-causing HPV strains. We have not evaluated the evidence underlying this finding.

• 91

  The quadrivalent vaccine protects against four cancer-causing HPV strains, including the two targeted by the bivalent vaccine.

• 92
  • “Vaccination impact is estimated in terms of reduction in age-dependent cervical cancer incidence, prevalence, and mortality in direct proportion to vaccine coverage, vaccine efficacy, and distribution of high-risk HPV types (HPV 16/18 for the bivalent and quadrivalent vaccines and HPV 16/18/31/33/45/52/58 for the nonavalent vaccine).” Abbas et al. 2020, Pg e537.
  • “With the updated model, the bivalent or quadrivalent HPV vaccine was estimated to avert 15 cases, 12 deaths, and 243 DALYs per 1000 vaccinated girls, and the nonavalent HPV vaccine was estimated to avert 19 cases, 14 deaths, and 306 DALYs per 1000 vaccinated girls.” Abbas et al. 2020, Pg e536.

• 93

  “Recently, a nona-valent vaccine targeting nine HPV types (HPV types 6, 11, 16, 18, 31, 33, 35, 45, 52 and 58) has been developed by Merck. We did not include the nona-valent vaccine in the current review, since the randomised trials assessing the efficacy of the nona-valent vaccine did not incorporate an arm with a non-HPV vaccine control.” Arbyn et al. 2018, Pg 14.

• 94
  • “Current evidence suggests that the 3 licensed HPV vaccines have relatively similar effectiveness in preventing cervical cancer.” WHO, Human papillomavirus vaccines: WHO position paper, 2017, Pg 256.
  • “With the updated model, the bivalent or quadrivalent HPV vaccine was estimated to avert 15 cases, 12 deaths, and 243 DALYs per 1000 vaccinated girls, and the nonavalent HPV vaccine was estimated to avert 19 cases, 14 deaths, and 306 DALYs per 1000 vaccinated girls.” Abbas et al. 2020, abstract.

• 95

  “We further assumed that vaccines provide lifelong protection, as suggested by the absence of vaccine failures in long-term follow-up of vaccinated cohorts, and statistical extrapolation of immunogenicity data.” Jit et al. 2014, Pg e407.

• 96

  The Immunisation Advisory Centre, Efficacy and effectiveness, Table “Duration of protection by vaccine”

• 97

  “Studies suggest that HPV vaccines offer long-lasting protection against HPV infection and therefore disease caused by HPV infection. Studies of the bivalent and quadrivalent vaccines have followed vaccinated individuals for more than 10 years and have found no evidence of protection decreasing over time. Duration of protection provided by HPV vaccination will continue to be studied.” CDC, HPV Vaccine Safety and Effectiveness

• 98

  “We assumed that no changes to methods of cervical cancer screening or uptake occur during the time horizon of the model—ie, the period during which model results are followed (in most cases, the lifetime of the vaccinated cohort).” Jit et al. 2014, Pg e407.

• 99

  “In the absence of screening, projected incidence rates for 2006–2010 in Nordic countries would have been between 3 and 5 times higher than observed rates. Over 60 000 cases or between 41 and 49% of the expected cases of cervical cancer may have been prevented by the introduction of screening in the late 1960s and early 1970s.” Vaccarella et al. 2014, abstract.

• 100
  • “After 5–8 years of vaccination, the prevalence of HPV 16 and 18 decreased significantly by 83% (RR 0.17, 95% CI 0.11–0.25) among girls aged 13–19 years, and decreased significantly by 66% (RR 0.34, 95% CI 0.23–0.49) among women aged 20–24 years. The prevalence of HPV 31, 33, and 45 decreased significantly by 54% (RR 0.46, 95% CI 0.33–0.66) among girls aged 13–19 years.” Drolet et al. 2019, abstract.
  • Figure 4 (Pg 24) of Drolet et al. 2019 illustrates data suggesting that the incidence of genital warts in unvaccinated boys and men declines after routine HPV vaccination is implemented in girls and women.

• 101

  “Anogenital wart diagnoses decreased significantly by 67% (RR 0.33, 95% CI 0.24–0.46) among girls aged 15–19 years, decreased significantly by 54% (RR 0.46, 95% CI 0.36–0.60) among women aged 20–24 years, and decreased significantly by 31% (RR 0.69, 95% CI 0.53–0.89) among women aged 25–29 years. Among boys aged 15–19 years anogenital wart diagnoses decreased significantly by 48% (RR 0.52, 95% CI 0.37–0.75) and among men aged 20–24 years they decreased significantly by 32% (RR 0.68, 95% CI 0.47–0.98).” Drolet et al. 2019, abstract.

• 102

  “Vaccination impact is estimated in terms of reduction in age-dependent cervical cancer incidence, prevalence, and mortality in direct proportion to vaccine coverage, vaccine efficacy, and distribution of high-risk HPV types (HPV 16/18 for the bivalent and quadrivalent vaccines and HPV 16/18/31/33/45/52/58 for the nonavalent vaccine). Herd effects and cross-protection are not considered; therefore, the estimated health benefits of HPV vaccination of 9–14-year-old girls are conservative.” Abbas et al. 2020, Pg e537.

• 103
  • “The effectiveness of some vaccines is limited by the variation in response observed between individuals and across populations. There is compelling evidence that a significant proportion of this variability can be attributed to human genetic variation...” Mentzer et al. 2015, abstract.
  • Mawa et al. 2015 reports that children born to mothers with latent tuberculosis infections tend to have a poor immune response to the tuberculosis vaccine. “In conclusion, maternal LTBI was associated with lower infant anti-mycobacterial T-cell responses immediately following BCG immunization.” Abstract.
  • Prendergast 2015 reports that “...the majority of malnourished children can mount a protective immune response following vaccination, although the timing, quality and duration of responses may be impaired.” Abstract.

• 104

  Arbyn et al. 2018 appendix 6 (Pgs 164-169) contains a list of trial locations, which includes North and South America, Oceania, Africa, South and East Asia, and Europe.

• 105

  “Currently, supply is insufficient to fully meet existing demand.20 As per the supply and demand base case scenarios, this imbalance is forecasted to remain for the short-/mid-term up to 2023/24, especially for LICs and MICs, due to the increased number of countries planning introductions and wishing to conduct MAC campaigns. The inability to serve this growing demand is estimated to be the cause of delayed planned introductions or delayed implementation of planned MAC campaigns in the period 2020–2023 in up to 29 countries.” WHO, Global Market Study, HPV, 2019, Pg 4.

• 106

  “The existing manufacturers are currently investing in increased capacity and supply has increased in the past few years; however, the required lead time for those new investments to translate into functioning new production lines will delay the availability of any additional doses to 2022 at the earliest.” WHO, Global Market Study, HPV, 2019, Pg 4.

• 107

  See, for example, our blog post on why we don’t currently recommend charities focused on vaccine distribution and our charity review on Gavi, the Vaccine Alliance

• 108
  • See our calculation of Gavi’s spending on HPV vaccination in this spreadsheet, “Gavi” tab.
  • Data comes from Gavi, Disbursements and commitments, “Data for commitments, approvals & disbursements in XLS format,” “Commitments” tab.

• 109
  • Gavi only supports countries which have an average Gross National Income (GNI) per capita that has been less than or equal to US$ 1,580 over the past three years. Gavi, Eligibility
  • As of 2019, fifteen low and lower middle income countries (as defined by the World Bank) are not eligible for support from Gavi and have not implemented a HPV vaccine national program.
    • See ‘HPV vaccination in LIC and LMIC’ tab, ‘Countries With No Hpv Vaccine Which Are Not Eligible For Gavi Funding’ section in this spreadsheet for details.
    • See The World Bank, World Bank Country and Lending Groups

• 110

  See ‘HPV vaccination in LIC and LMIC’ tab, ‘Countries With No Hpv Vaccine Which Are Not Eligible For Gavi Funding, population and per person burden’ section in this spreadsheet for details.

• 111

  We are unable to release these conversations publicly.