Global burden and economic impact of vaccine-preventable cancer mortality

Abstract

Background

Infections are responsible for approximately 13% of cancer cases worldwide and many of these infections can be prevented by vaccination. Human papillomavirus (HPV) and hepatitis B virus (HBV) are among the most common infections that cause cancer deaths globally, despite effective prophylactic vaccines being available. This analysis aims to estimate the global burden and economic impact of vaccine-preventable cancer mortality across World Health Organization (WHO) regions.

Methods

The number of deaths and years of life lost (YLL) due to five different vaccine-preventable cancer forms (oral cavity, liver, laryngeal, cervical, and oropharyngeal cancer) in each of the WHO regions (African, Eastern Mediterranean, European, the Americas, South-East Asia Pacific, and Western Pacific) were obtained from the Institute for Health Metrics Evaluation global burden of disease dataset. Vaccine-preventable mortality was estimated considering the fraction attributable to infection, to estimate the number of deaths and YLL potentially preventable through vaccination. Data from the World Bank on GDP per capita were used to estimate the value of YLL (VYLL). The robustness of these results was explored with sensitivity analysis. Given that several Epstein-Barr virus (EBV) vaccines are in development, but not yet available, the impact of a potential vaccine for EBV was evaluated in a scenario analysis.

Results

In 2019, there were 465,740 potentially vaccine-preventable cancer deaths and 14,171,397 YLL across all WHO regions. The estimated economic impact due to this mortality was $106.3 billion globally. The sensitivity analysis calculated a range of 403,025–582,773 deaths and a range in productivity cost of $78.8–129.0 billion. In the scenario analysis EBV-related cancer mortality increased the global burden by 159,723 deaths and $32.4 billion.

Conclusion

Overall, the findings from this analysis illustrate the high economic impact of premature cancer mortality that could be potentially preventable by vaccination which may assist decision-makers in allocating limited resources among competing priorities. Improved implementation and increased vaccination coverage of HPV and HBV should be prioritized to decrease this burden.

Keywords:

• Cancer
• hepatitis B virus
• health economics
• human papillomavirus
• indirect costs
• vaccines

JEL CLASSIFICATION CODES:

• I10
• I1
• I
• I18

Introduction

Cancer is the second highest cause of death globally, with an estimated 19.3 million new cases and 10 million deaths in 2020 alone¹. The total annual cost of cancer in Europe has been estimated at €199 billion (∼$220 billion), with €50 billion (∼$55 billion) from premature mortality². In the United States (US), the estimated economic burden of cancer is estimated at 1.8% of gross domestic product (GDP)³. In 2017, cancer resulted in a productivity loss in the US alone equivalent to $30.3 billion. These large productivity losses that stem from cancer deaths, representing a significant percentage of total costs, emphasize the importance of considering the indirect costs when quantifying the economic burden associated with disease.

Infections are responsible for ∼13% of cancer cases worldwide, providing opportunities for prophylactic measures as some of these infections can be prevented by vaccination⁴. Yet, viruses such as human papillomavirus (HPV) and hepatitis B virus (HBV), with highly effective prophylactic vaccines available to prevent virus-associated cancers, still impose a major burden of disease by causing cancer.

Cervical cancer, mostly caused by persistent HPV infection, is the fourth most common cancer in women worldwide, with a high age-standardized mortality rate of 13.3/100,000 women which equated to ∼341,831 deaths globally in 2020⁵. The majority of deaths occurred in lower middle income countries (LMICs), with particularly high mortality rates in Eastern and Middle Africa. Effective vaccines are available that protect against strains of HPV responsible for >90% of cervical cancers⁶. The global vaccine coverage rate (VCR) for HPV was reported at ∼12%, and within High Income Countries (HICs) the average VCR was reported as 42%; highlighting a large disparity between country income groups⁷. The incidence of cervical cancer in HICs has approximately halved in the past 30 years⁸. Primary preventative measures (e.g. vaccination) and secondary preventative measures (e.g. screening) are responsible for the fall in the incidence of cervical cancer, and in combination with treatment advances have also reduced the mortality rate of cervical cancer cases^8,9. In addition to cervical cancer, HPV can also cause some anal, penile, vulva, vaginal, and head and neck cancers⁶.

HBV infection is a major cause of liver cancer, with more than a third of cases attributed to HBV¹⁰. The impact of liver cancer itself is significant; in 2020 alone, liver cancer accounted for 830,200 global deaths, with the highest burden of disease in Eastern Asia and Northern Africa¹¹. As the number of new cases rise annually, it is predicted that 1.3 million people could die from this disease in 2040.

Other infections known to cause cancer, without an effective vaccine available yet, include human immunodeficiency virus (HIV), human herpesvirus 8 (HHS-8), and Epstein-Barr virus (EBV). EBV has been linked with various malignant disease and tumor types including nasopharyngeal cancer, stomach cancer, Hodgkin’s lymphoma, and Burkitt lymphoma¹². While EBV vaccines are not currently available there are several vaccine candidates in development^13,14. The potential candidates for HIV and HHS-8 vaccines are still in the initial phases of development^15–17.

The aim of this analysis was to estimate the global burden and economic impact of cancer deaths potentially preventable through vaccination and to understand the indirect effect of premature deaths by estimating productivity losses due to HBV- and HPV-related cancer deaths across World Health Organization (WHO) regions.

Methods

Model structure

A health-economic model was developed to estimate indirect costs due to premature death from HPV- and HBV-related cancers, that could be prevented by currently available vaccines. The model adopted a societal perspective and only considered costs associated with productivity losses due to premature mortality. The human capital method was used as in previous studies^18,19. Productivity loss was defined as loss of earnings caused by premature mortality from cancer. Direct costs, such as treatment costs, were not considered as patients only entered the model at death. The model considered cancer-related deaths in a single year (2019), with costs evaluated from death due to cancer up to life expectancy.

Population

The population in the base-case included persons who died from five different types of cancer across six different WHO regions. The following five cancers were included to determine the mortality due to HPV and HBV: oral cavity, oropharyngeal, liver, laryngeal, and cervical cancer. Other HPV-related cancers such as penile or anal cancer were not included in the analysis as these were not available within the Institute for Health Metrics Evaluation (IHME) dataset. The cancers included in this analysis make up 90% of global cancer cases attributable to HPV infection²⁰. The six WHO regions were included: African region, Eastern Mediterranean region, European region, the region of the Americas, South-East Asia region, and Western Pacific region.

Model inputs

The number of deaths and years of life lost (YLL) in 2019 from included cancers were sourced directly from the IHME Global Burden of Disease (GBD)²¹. The GBD is a well-established dataset collected and analysed by a consortium of over 9,000 researchers across 162 countries and territories. The GBD study incorporates an Independent Advisory Committee that monitors the scientific rigour of methods to ensure validity of results. Data from the IHME was stratified by WHO region, age-group (<20, 20–54, 55–59, 60–64, 65–69, 70+), sex (male, female), and cancer type.

Not all cancer cases are caused by infection; therefore, the model includes attributable fractions (AFs), the proportion of deaths for a given cancer which are attributed to a specific infection (Table 1). The AFs for each cancer type were sourced according to the latest data available. Sources were prioritized which reported AFs most relevant to the cancer types studied (using ICD-10 codes where possible) and for geographical regions that could be mapped on to the WHO regions. These sources and the corresponding ICD-10 codes were reviewed and validated by a team of clinicians and epidemiologists.

Table 1. Base-case attributable fractions for HPV and HBV vaccine-preventable cancer mortality by region and cancer.

Region	Oral cavity cancer (HPV)^a	Oropharynx cancer (HPV)^a	Liver cancer (HBV)^b	Larynx cancer (HPV)	Cervical cancer (HPV)^c
African	0.013	0.076	1	0.011	0.9
Eastern Mediterranean	0.013	0.105	1	0.011	0.9
European	0.039	0.292	1	0.046	0.9
The Americas	0.025	0.258	1	0.027	0.9
South-East Asia	0.013	0.113	1	0.011	0.9
Western Pacific	0.013	0.097	1	0.011	0.9
Source	de Martel et al.^20	de Martel et al.^22	NA	de Martel et al.^22	de Martel et al.^20

Abbreviations. HBV, hepatitis B virus; HPV, human papillomavirus.

^a The AFs for oral cavity cancer and oropharyngeal have been weighted from the original source material in order to delineate oral cavity cancer from lip and oral cavity cancer, and oropharyngeal cancer from other pharynx cancer in the IHME mortality data.

^b No AF was applied as the IHME dataset reported the number of liver cancer deaths associated with HBV.

^c Although HPV has been reported to cause almost all cervical cancer, a conservative estimate has been used.

AFs for HPV-related cancers, stratified by WHO region and cancer indication, were sourced from de Martel et al.^20,22. Due to data limitations and to reduce model complexity, case-related AFs were used instead of mortality AFs. Although infection may have an impact on the prognosis of the cancer, this analysis assumes that the AF for case incidence is equivalent to the AF for mortality. IHME data on liver cancer (ICD-10 C22) only reported cancer deaths attributable to HBV infection and therefore no AF was applied (effectively equalling an AF of 1) for this cancer type across all WHO regions²¹. Data on larynx (ICD-10 C32) and cervical (ICD-10 53) cancer were taken straight from IHME and HPV-related AFs were applied. The number of deaths and YLL from larynx cancer were multiplied by an AF between 0.011 and 0.046, depending on WHO region, sourced from de Martel et al. (Table 1)²². An AF of 0.9 was used to estimate potentially vaccine-preventable cervical cancer mortality, this is a conservative estimate as the literature indicates that almost all of cervical cancer is caused by HPV infection (Table 1)²⁰.

Lip and oral cavity cancer (ICD-10 C00–C08) mortality was reported as aggregated figures in the IHME dataset. As lip cancer is not caused by HPV infection, deaths due to lip cancer were removed. Mortality due to lip versus oral cancer was not available in published literature and therefore the subsite distribution of lip and oral cancer incidence was used instead. Data from Shield et al., which sourced data from the International Agency for Research on Cancer and GLOBOCAN, was used to calculate that 91% of lip and oral cavity cancers were due to oral cancer²³. The AFs provided by de Martel et al.²⁰ on oral cavity cancer were multiplied by the 91% weighting to derive the AFs used in the model (estimated at between 0.013 and 0.039, depending on WHO region [Table 1]). In a similar manner, oropharynx cancer mortality was delineated from “Other pharynx cancers” (ICD-10 C09–10, C12–13), using a weighting between 0.424 and 0.831 depending on region as according to the WHO Cancer Today mortality database²⁴. The AFs sourced from de Martel et al. for each WHO region were multiplied by the oropharynx weighting²². This resulted in an AF of 0.076–0.292 (depending on WHO region) used in the model to estimate the potentially vaccine-preventable mortality from oropharynx cancer (Table 1).

To determine productivity loss associated with premature mortality, GDP per capita was sourced at the country level from the World Bank (reported in United States Dollars [$]) and averaged for each WHO region (Supplementary Table S1)²⁵.

The internal consistency of the model was evaluated by comparing the outputs with source data used for development of the model. This included inputting extreme values (e.g. for GDP per capita or discount rates) to determine if the model behaved as expected under these conditions.

Model calculations

The number of deaths and YLL attributable to cancer-causing infections were estimated by multiplying the mortality and YLL figures, taken directly from the IHME dataset, by the AF associated with each WHO region and cancer: $$\mathit{\text{Preventable deaths}}=\sum _{i=1}^I\left({\mathit{\text{deaths}}}_i\mathrm{*}\mathit{\text{AF}}\right)$$ $$\mathit{\text{Preventable YLL}}=\sum _{i=1}^I\left({\mathit{\text{YLL}}}_i\mathrm{*}\mathit{\text{AF}}\right)$$ where i = 1, 2, 3, … i refer to the population age-groups contained in the dataset.

These two outcomes were then used to calculate the average years of life lost (AYLL) per death by region and cancer indication: $$\mathit{\text{Average Years of Life Lost}}= \left(\mathit{\text{preventable YLL}}\right)/\left(\mathit{\text{total number of preventable deaths}}\right)$$

To calculate the value of years of life lost, the expected earnings had death been avoided (measured in GDP per capita [USD]) were applied to YLL: $$\mathit{\text{Value of YLL}}=\mathit{\text{preventable YLL}}\mathrm{*}\mathit{\text{GDP per capita}}$$

Value of years of life lost (VYLL) was discounted at a standard rate of 3% annually to obtain the present value of future costs²⁶.

Sensitivity and scenario analysis

Deterministic sensitivity analysis (DSA) was used to investigate the sensitivity of results, by varying inputs: number of deaths, YLL, AF, and GDP per capita. Mortality and YLL were varied by 95% confidence intervals. GDP per capita was varied by ±10% of the base-case value. AFs were varied by their 95% confidence intervals (Supplementary Table S2). GDP per capita and AFs were varied together.

Given that several EBV vaccines are in development, but not yet available, the impact of a potential vaccine for EBV was considered in a scenario analysis. The EBV vaccine was chosen as vaccines for HIV and HHS-8 are still in earlier stages of development^15–17. Furthermore, mortality data on key cancer types associated with these viral infections were not available from IHME, particularly for HIV-related cancers such as Kaposi sarcoma or aggressive B-cell non-Hodgkin lymphoma. The population in the scenario analysis included persons who died from the following: nasopharynx cancer (ICD-10 C11), stomach cancer (ICD-10 C16), Hodgkin’s lymphoma (ICD-10 C81), and Burkitt lymphoma (ICD-10 C83.7). These cancers account for more than 99% of global EBV-related cancer deaths²⁷.

Published estimates of AFs for EBV-related cancers are scarce and estimates are not available by WHO region. The published regional estimates do not fully align with regional boundaries and were therefore mapped to WHO regions to estimate the proportion of cancer deaths that were caused by EBV infection (Supplementary Table S3)^28–31. Individuals across age categories were assumed to experience the same AF of cancer to disease. For Burkitt lymphoma, AF estimates for those aged 0–14 years were applied to the entire population. Results were then derived in the same way as in the base-case.

As a further scenario analysis, productivity losses were restricted to YLL that occurred prior to retirement as a more conservative estimate of the economic impact. In the base case GDP per capita considers the full population and therefore provides a financial value for any year of life, not just those when a person is working. The proportion of YLL that occurred when people would have been employed, had their death been prevented, was estimated using retirement age as an input (Supplementary Table S4). This was then applied to preventable YLL to determine years of productive life lost (YPLL): $$\mathit{\text{YPLL}}= \mathit{\text{labour force partipation rate}}\mathrm{*}\sum _{i=1}^I\left(\frac{{\mathit{\text{Expected productive life years remaining}}}_i}{{\mathit{\text{Expected life years remaining}}}_i}\mathrm{*}{\mathit{\text{YLL}}}_i\mathrm{*}\mathit{\text{AF}}\right)$$ where i = 1, 2, 3, … i are the population age groups used in the model. YPLL was discounted in a similar manner to YLL and the value of YPLL (VYPLL) was then calculated using the same method as VYLL.

Model assumptions

The model used real world data to provide the input parameters where possible, however, a number of assumptions (Table 2) were made to account for the granularity of data available and model structure. Assumptions made around age-groups followed existing model approaches^18,19.

Table 2. Parameter assumptions made in the model.

Model settings	Assumption details
Age groups	Individuals within each age group are assumed to be homogenous and will have the same outcomes.
Retirement ages	The retirement age of each WHO region is equal to the average retirement age of the countries that make up that region. Individuals in an age-group above the average retirement for that region have no years of productive life to lose. Individuals across age-groups can expect the same average retirement age for their sex and WHO region.
GPD per capita	The GDP per capita of each WHO region is equal to the average GDP per capita of the countries that make up that region.
Attributable fractions	Assumed to be homogenous across age-groups. As case-related AFs are used, infection is not assumed to impact the prognosis of cancer, this means that the proportion of cancer cases attributable to infection is proportional to the proportion of cancer deaths attributable to infection.

Results

All cancer results stratified by WHO region

In 2019, there were 465,740 cancer deaths and 14,171,397 YLL that could potentially have been prevented through vaccination across all WHO regions (Table 3). The estimated economic impact due to vaccine-preventable cancer deaths was $106.3 billion globally.

Table 3. Global humanistic and economic impact of HPV and HBV vaccine-preventable cancers, by WHO region, in 2019.

Region	Number of deaths	Number of deaths as % of global deaths	YLL	YLL in males/females as % of total	AYLL	VYLL ($)
Africa	56,625	12%	1,944,285	11%/89%	34	4,141,722,433
Eastern Mediterranean	15,659	3%	523,375	32%/68%	33	5,308,430,050
Europe	47,506	10%	1,261,207	29%/71%	27	32,960,000,102
The Americas	51,162	11%	1,493,230	13%/87%	29	34,552,900,575
South-East Asia	94,172	20%	2,998,611	26%/74%	32	21,726,028,370
Western Pacific	200,616	43%	5,950,689	60%/40%	30	7,565,159,206
Global	465,740	–	14,171,397	37%/63%	30	106,254,240,735

Abbreviations. HBV, hepatitis B virus; HPV, human papillomavirus.

Number of deaths and YLL is equal to the number that was preventable according to the AF for each WHO region and cancer indication. Global AYLL is equal to a weighted average of WHO region values, weighted by the respective number of deaths as percentage of global. VYLL has been discounted with a standard cumulative discount rate of 3%.

The Western Pacific region had the greatest number of vaccine-preventable deaths (200,616; 43%) and YLL (5,950,689; 42%). The Eastern Mediterranean region had the lowest number of deaths (15,659). The European region and the Americas experienced the highest economic impact due to vaccine-preventable cancer deaths, with a VYLL of $33.0 and $34.6 billion, respectively (64% of the global impact).

AYLL allows comparison between WHO regions and provides a measure of the average age at which people died of vaccine-preventable cancers (Figure 1). The African and Eastern Mediterranean regions had the highest AYLLs (34 and 33, respectively), indicating that, on average, deaths occurred at a younger age.

Figure 1. Average years of life lost per death, by WHO region.

Global results stratified by cancer

Cervical cancer had the highest burden of disease, with 251,846 deaths and a VYLL of $71.9 billion (Table 4). Cervical cancer and liver cancer combined represented a total of 95% and 96% of total deaths and YLL, respectively. Larynx cancer had the lowest burden of disease, with 2,472 deaths and an economic impact of $1.2 billion.

Table 4. Global humanistic and economic impact of HPV and HBV vaccine-preventable cancers, by cancer, in 2019.

Cancer	Number of deaths	Number of deaths as % of total deaths	YLL	YLL % occurring in Males/Females	AYLL	VYLL ($)
Oral cavity	3,634	1%	94,880	70%/30%	26	1,424,473,318
Oropharynx	17,291	4%	476,873	80%/20%	28	7,347,435,260
Liver	190,496	41%	5,713,246	83%/17%	30	24,425,947,715
Larynx	2,472	1%	61,460	87%/13%	25	1,154,315,583
Cervical	251,846	54%	7,824,938	0%/100%	31	71,902,068,859
All cancers	465,740	–	14,171,397	37%/63%	30	106,254,240,735

Abbreviations. AYLL, average years of life lost; HBV, hepatitis B virus; HPV, human papillomavirus; VYLL, values of years of life lost; YLL, years of life lost.

Larynx cancer mortality had the highest VYLL per death (Figure 2), despite having the lowest overall VYLL. Cervical cancer and liver cancer had the lowest VYLL per death ($196,259 and $256,181, respectively) despite having the highest AYLL per death (31 and 30, respectively) (Figure 2).

Figure 2. Global humanistic burden of HPV and HBV vaccine-preventable cancers, by cancer, in 2019.

Cervical cancer was the biggest contributor to productivity loss compared to other cancers in the African region (85%) and the Americas (82%) (Figure 3). The proportional impact of laryngeal cancer and oral cavity cancer compared to other cancer types was ≤2% across all regions.

Figure 3. Proportion of productivity loss (value of years of life lost) due to each cancer in 2019, by WHO region.

HPV-related cervical cancer results

In 2019, there was 251,846 global HPV-related cervical cancer deaths, with productivity loss valued at $71.9 billion (Table 4). The Western Pacific and South-East Asia regions had the highest HPV-related cervical cancer mortality burden (combined 49% of global deaths).

The largest proportion of HPV-related cervical cancer costs was in the Americas (39% of global VYLL) (Figure 4). Only 3% of the global economic impact of HPV-related cervical cancer was attributable to the Western Pacific region ($2,377,989,255) which had the lowest GDP per capita (Table 5).

Figure 4. Proportion of economic burden (VYLL) imposed HPV and HBV vaccine-preventable cancers, in 2019, by WHO region.

Table 5. Humanistic and economic impact of HPV-related cervical cancer, by WHO region, in 2019.

Region	Number of deaths	Number of deaths as % of total deaths	YLL	YLL % occurring in Males/Females	AYLL	VYLL ($)
Africa	48,057	19%	1,658,223	0%/100%	35	3,529,511,093
Eastern Mediterranean	8,500	3%	291,179	0%/100%	34	2,925,151,567
Europe	29,773	12%	796,945	0%/100%	27	20,626,100,782
The Americas	41,292	16%	1,231,963	0%/100%	30	28,228,834,147
South-East Asia	60,202	24%	1,985,161	0%/100%	33	14,214,482,015
Western Pacific	64,023	25%	1,861,468	0%/100%	29	2,377,989,255
Global	251,846	–	7,824,938	0%/100%	31	71,902,068,859

Abbreviations. AYLL, average years of life lost; HPV, human papillomavirus; VYPLL, values of years of productive life lost; YPLL, years of productive life lost; YLL, years of life lost.

HPV-related head and neck cancer results

In 2019, there was a total of 23,397 global HPV-related head and neck cancer deaths with an economic burden of $9.93 billion in productivity losses (Table 6). The South-East Asia and European regions combined had 71% of total deaths and YLL. The Americas and the Western Pacific region had the lowest AYLLs of the WHO regions considered (25).

Table 6. Humanistic and economic impact of HPV-related head and neck cancer, by WHO region, in 2019.

Region	Number of deaths	Number of deaths as % of total deaths	YLL	YLL % occurring in Males/Females	AYLL	VYLL ($)
Africa	387	2%	11,575	72%/28%	30	26,123,879
Eastern Mediterranean	998	4%	31,322	70%/30%	31	326,909,566
Europe	7,942	34%	208,026	86%/14%	26	5,562,730,370
The Americas	3,321	14%	83,681	82%/18%	25	2,068,783,868
South-East Asia	8,691	37%	247,645	73%/27%	28	1,872,738,991
Western Pacific	2,058	9%	50,963	84%/16%	25	68,937,487
Global	23,397	–	633,213	79%/21%	27	9,926,224,161

Abbreviations. AYLL, average years of life lost; HPV, human papillomavirus; VYPLL, values of years of productive life lost; YLL, years of life lost; YPLL, years of productive life lost.

The biggest difference in male and female mortality due to HPV-related head and neck cancers occurred in the European region (Table 6). This region had one of the highest number of deaths as a percentage of total deaths (34%).

The European region, which had the highest GDP per capita, contributed the largest proportion of the global economic impact of HPV-related head and neck cancer costs (56%; Figure 4). The Western Pacific region, despite accounting for 43% of global deaths across all cancers, only made up 1% ($68,937,487) of the global economic burden (Table 6).

HBV-related liver cancer results

In 2019, there were a total 190,496 global HBV-related liver cancer deaths with $24.4 billion estimated in productivity losses (Table 7). The Western Pacific region was the most impacted by HBV-related liver cancer and accounted for 71% of global deaths with a cost of $5.1 billion. The European, South-East Asia, and Western Pacific were responsible for similar proportions of global VYLL (Figure 4) whereas the African region had a significantly lower share (2%).

Table 7. Humanistic and economic impact of HBV-related liver cancer, by WHO region, in 2019.

Region	Number of deaths	Number of deaths as % of total deaths	YLL	YLL % occurring in Males/Females	AYLL	VYLL ($)
Africa	8,182	4%	274,486	74/26%	34	586,087,461
Eastern Mediterranean	6,161	3%	200,874	73/27%	33	2,056,368,917
Europe	9,791	5%	256,236	74/26%	26	6,771,168,950
The Americas	6,550	3%	177,586	71/29%	27	4,255,282,560
South-East Asia	25,278	13%	765,805	77/23%	30	5,638,807,363
Western Pacific	134,534	71%	4,038,258	87/13%	30	5,118,232,463
Global	190,496	–	5,713,246	83/17%	30	24,425,947,715

Abbreviations. AYLL, average years of life lost; HBV, hepatitis B virus; VYPLL, values of years of productive life lost; YLL, years of life lost; YPLL, years of productive life lost.

Deterministic sensitivity analyses

The DSA shows that varying GDP per capita and AFs together led to a range of 403,025–495,330 global deaths and $78.8–125.7 billion in global productivity costs (Supplementary Tables S5–S9). Variations in global deaths were entirely driven by AFs as preventable mortality was not dependent on GDP per capita.

When the uncertainty in mortality and YLL inputs were considered, this led to a range of 481,870–582,773 global deaths and a productivity cost of $113.0–129.0 billion due to premature mortality (Supplementary Tables S10–S13).

Scenario analyses

Vaccine-preventable EBV-related cancer mortality increased the global burden of vaccine-preventable cancers by 159,723 deaths and $32.4 billion in 2019. The Western Pacific region accounted for the greatest percentage of global deaths (48%) and the European region accounted for the greatest proportion (31%) of the global economic impact of EBV-related cancers. Stomach cancer had the greatest economic impact (45% of total EBV-related VYLL) and Burkitt lymphoma had the least (3%).

When limiting to VYPLL for HPV- and HBV-related deaths the global economic burden was estimated at $19.6 billion (82% reduction compared to VYLL). The Western Pacific region had the greatest number of vaccine-preventable deaths (200,616; 43%) and YLL (5,950,689; 42%). The Eastern Mediterranean region had the lowest number of deaths (15,659). The European region and the Americas experienced the highest economic impact due to vaccine-preventable cancer death, with a VYLL of $33.0 and $34.6 billion, respectively (64% of the global impact).

Discussion

There are high productivity losses resulting from premature cancer mortality that could potentially be prevented by vaccination. In 2019, there were 465,740 potentially vaccine-preventable global cancer deaths with an estimated economic impact of $106.3 billion. The Americas ($34.6 billion) and African region ($4.1 billion) had the highest and lowest VYLL, respectively. Cervical cancer had the highest economic burden ($71.9 billion) and the lowest was larynx cancer ($1.2 billion).

In 2019, 296 million person were living with chronic HBV infection worldwide, 39% resided in the Western Pacific region and 28% in the African region³². The high HBV prevalence in the Western Pacific region is reflected in the results, with 71% of global vaccine-preventable HBV-related cancer deaths occurring there. In the Western Pacific region, 82% of people infected with HBV remain undiagnosed and less than half of these are undergoing treatment³³. The low diagnosis and treatment rates might explain why the majority of HBV-related liver cancer deaths occurred in this region.

The Western Pacific region has made substantial progress in controlling HBV infection through successful immunization programs for children³². By immunizing children, the prevalence of HBV infection is expected to steadily decline but the impact on mortality would not be seen for a number of years. The timeliness of implementing vaccination programmes is critical. Studies on the global uptake of HPV vaccination indicate that HICs introduced vaccination programmes at a fast rate (80% implemented in under a decade)³⁴. The introduction of vaccination programmes in LMICs has been slower, but recent trends indicate VCRs are starting to improve. In these countries, it may take time to see the impact of these programmes on the mortality burden. Evidence from Sweden shows that vaccinating girls before age 17 reduced the incidence of cervical cancer in this cohort by ∼90% over an 11-year follow-up³⁵. While adolescents may become exposed to HPV shortly after inoculation, cervical cancer can take decades to develop following infection, leading to a lag in incidence trends³⁶. If the effects of action today will not be seen for some time, then it is even more important that policy-makers implement changes now.

Globally in 2019, approximately 15% of girls and 4% of boys within the target age group were given a full course of the HPV vaccine³⁴. The results of this analysis indicate that 72–84% of head and neck HPV-related cancers occurred in males across regions (Table 6). The gap in vaccination coverage between males and females combined with head and neck HPV-related cancer disproportionally impacting the male population may lead to greater disparities in associated cancer deaths in the future. The effects of gender-neutral vaccination (ensuring that both boys and girls are vaccinated against HPV) on incidence will take several decades to be observed but early implementation could be important to ensure additional benefits of HPV vaccination coverage.

Previous research indicates that the disparity in mortality between regions is amplified by the greater access to secondary prevention measures and treatment associated with HICs⁷. The African region and the Western Pacific region had an AYLL due to vaccine-preventable cancers of 34 and 30, respectively, while the European region and the Americas had an AYLL of 27 and 29, respectively. These results indicate deaths are occurring in younger age-groups in the African and Western Pacific regions (compared to other regions). This may be driven by secondary preventative measures and access to advanced cancer treatments in regions with lower AYLLs. The main benefits of vaccination in HICs will be to reduce reliance on invasive screening to lower expenditure and morbidity of disease and its treatment⁷.

This model uses robust, publicly available datasets from sources including the IHME and the World Bank to develop quantifiable evidence of the indirect costs of vaccine-preventable cancer. The use of consistent data sources across the regions and cancers allowed comparison across regions and cancer, yielding greater insights from results.

The model adjusts for the proportion of the cancer cases that are caused by infection by applying AFs according to region and cancer type. To facilitate comparisons between WHO regions AYLL was estimated to standardize for population size. This is particularly important for identifying differences in impact between the developing world and the developed world. The model avoids reliance on crude death rates by using YLL as an input (and estimating YPLL) to give more weight to deaths that occurred in younger age-groups. As mortality naturally increases sharply in older age-groups, this approach provides a more valid estimate of overall disease burden. Lastly, the model incorporated sensitivity and scenario analysis to test variation in the results. The mortality and economic burden of disease estimated in this analysis therefore reflects what could potentially be prevented with vaccination, helping increase the utility in guiding policy decisions.

Despite the robust modelling approach, several assumptions remain in relation to the input parameters due to the complexity of real-world data. For instance, the model does not account for the impact of socio-demographic factors on disease incidence and burden. The use of GDP per capita introduces the assumption that infection-related cancer deaths are uniformly distributed across socio-economic groups. If there are more deaths in lower income groups, then the value of lost productivity would be overestimated in this analysis. However, devaluing the deaths that occur in different socioeconomic groups would have introduced problems from an ethical standpoint and was avoided in this analysis. GDP per capita is therefore considered an acceptable proxy for estimating productivity losses in this setting, as it considers the full population. In addition, the model does not account for the fact that patients diagnosed with cancer could be replaced in the workforce, potentially reducing the productivity loss resulting from the premature deaths.

The total economic burden of disease is made up of indirect and direct costs, some of which occur whilst a patient is still alive due to the morbidity associated with a cancer diagnosis reducing productivity, especially towards the end of life. In 2018 the cost of cancer morbidity accounted for €20 billion out of the €70 billion in total productivity losses in Europe. However, this model focused on indirect costs accumulated after death and the results should therefore be considered a conservative estimate of the total indirect cost². The overall annual direct medical cost of preventing and treating HPV-associated disease alone has been estimated at ∼$8.0 billion³⁷. The model therefore only captures part of the indirect cost, which in turn is only part of the total cost. Additionally, not all HPV-related cancers (e.g. anal or vaginal) were included, due to data limitations. Nor were other infections with potential vaccines in development such as HIV. It is also important to place the costs estimated by this study in the context of the treatment and disease landscape status prior to 2019. As the profile of available therapies change, this could impact mortality trends and associated costs in the future.

Finally, there are limitations in the AFs used to estimate potentially preventable cancer deaths from the IHME mortality data. The vaccination status of the population is not recorded, instead the model utilizes case-related AFs (rather than mortality-related AFs). This therefore assumes that any impact from ongoing vaccination would be captured within the AF specific to that cancer type, as those who are vaccinated would have a decreased risk of infection-related cancer and therefore the AF would be reduced. However, this introduces the assumption that cancer caused by viral infection has the same prognosis as a cancer not caused by viral infection. The literature indicates that head and neck cancers caused by HPV, especially oropharyngeal cancers, may in fact have a better prognosis than other etiologies^38,39. This effect may be limited to 5-year survival rates rather than fully curative rates, meaning HPV-related mortality may be postponed but not necessarily permanently reduced. Furthermore, the impact of recent vaccination programmes on 2019 deaths is minimal due to the length of time it takes for viral load to lead to cancer, and ultimately death. It should also be noted that over 90% of the deaths and productivity losses estimated by this model across cancers are due to cervical cancer and liver cancer. The overwhelming majority of the attributable mortality in these cancer types is only due to the causative virus, and the overall effect of this assumption on our combined results is therefore not substantial.

It should be noted, however, that AFs were only region- and cancer-specific. Individuals across age-groups were assumed to face the same AF as each other according to their region and cancer type. This is likely not reflective of real world cases, as evidence indicates the prognosis of EBV-related Hodgkin’s lymphoma is better in lower age group compared to older ages, but this difference is not reflected in the model³¹. Similarly, the AF used for Burkitt lymphoma was the same across age-groups, though the incidence of EBV-related Burkitt lymphoma is lower in older age groups⁴⁰. It is therefore difficult to argue that the full humanistic and economic impact estimated by this analysis could have been prevented with complete vaccination coverage. Limited data was available to adjust for these potential considerations, and therefore the best available data was utilized with careful consideration of the underlying populations when applying averaged parameters across a group.

Future research should include consideration and projection of productivity losses in a future perspective to help global and national policy-makers guide strategies to prevent cancer mortality. Vaccination should be considered a longer-term investment, and such a prediction may be of more use to policy-makers to plan for future changes. Additional research to facilitate cost-benefit analysis in which the potential gains of productivity are compared to the overall costs of vaccination programmes will also be useful.

Conclusion

This analysis illustrates the high humanistic and economic burden associated with vaccine-preventable cancers through providing insights into the indirect cost burden imposed by HPV- and HBV-related cancer mortality. The results can inform policy stakeholders and decision-makers on how the cost of increasing vaccination efforts could one day be offset by the potential avoidance of indirect productivity losses due to early mortality across different WHO regions and vaccine-preventable cancers. Moreover, these results provide an economic measure of the cancer burden which may assist decision-makers in allocating limited resources among competing priorities.

Transparency

Declaration of funding

This work was funded by Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA. GB, and EM are employees of Merck & Co and contributed to model validation and suggested revisions to the manuscript.

Declaration of financial/other interests

GB and EM are employees of MSD subsidiaries of Merck & Co., Inc., Rahway, NJ, USA and may own stocks and/or stock options in Merck & Co., Inc., Rahway, NJ, USA. KS has received research grants and consultancy fees to her affiliating institution as part of a strategic collaboration between Karolinska Institutet and Merck/MSD.

EO, AM, RH, and GW are employees of Adelphi Values (PROVE), paid consultants to MSD.

Author contributions

GB and GW conceptualized and designed the study. AM, GW, and KS supervised the study. GW, RH, AM, and EO conducted the data analysis, visualized and interpreted the data, and reviewed the literature. EO drafted the manuscript with supervision from GW and AM. All authors contributed to the interpretation of the results and commented on the manuscript. All authors read and approved the final version of the manuscript. All authors had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Reviewer disclosures

Peer reviewers on this manuscript have received an honorarium from JME for their review work but have no other relevant financial relationships to disclose.

Previous presentations

The results of this analysis were previously presented at the 2023 ASCO Annual Meeting (Goran Bencina, Edward Oliver, Anne Meiwald, Robert Hughes, Edith Morais, Julia A Schillinger, and Georgie Weston. Global burden and economic impact of vaccine-preventable cancer mortality. Journal of Clinical Oncology 2023 41:16_suppl, 10542-10542).

Acknowledgements

We thank Elizabeth Goodman and Julia A. Schillinger (Merck & Co) for providing constructive comments during review of manuscript drafts.

Data availability statement

The data that support the findings of this study are available from the Institute for Health Metrics Evaluation (IHME) Global Burden of Disease. Restrictions apply to the availability of these data, which were used under license for this study. Data are available at https://www.healthdata.org/research-analysis/gbd with the permission of IHME.