HPV vaccination among gay, bisexual, and other men who have sex with men aged 27–45 years in the United States is potentially Cost-saving

Abstract

Background:

The Advisory Committee on Immunization Practices (ACIP) recommends routine human papillomavirus (HPV) vaccination at age 11 or 12 years (vaccination can be given starting at age 9 years), with catch-up vaccination for all persons through age 26 years. ACIP recommends shared clinical decision-making for HPV vaccination of persons aged 27 through 45 years, although modeling indicated that broadly vaccinating persons in this age group would not be cost-effective. To help inform clinical decision-making, the objective of this study was to estimate the cost-effectiveness of HPV vaccination among gay, bisexual, and other men who have sex with men (GBMSM) aged 27 through 45 years in the United States.

Method:

We used an age-specific, compartmental, susceptible-infectious-susceptible (SIS) model of HPV transmission and disease among GBMSM to estimate changes in HPV, anogenital wart, and HPV-attributable cancer incidence when the upper age for vaccination among GBMSM was raised to age 30, 35, 40, or 45 years. We then estimated changes in quality adjusted life years (QALYs), costs (vaccination costs and direct medical costs of HPV disease), and the incremental cost-effectiveness ratio under each of these scenarios.

Results:

Our estimates showed that HPV vaccination of GBMSM was cost-saving through age 45 years, i.e., the costs of vaccination are more than offset by decreases in the direct medical costs of HPV disease. Vaccination of GBMSM through at least age 40 years was cost-saving across most sensitivity analyses.

Conclusions:

Vaccination of GBMSM through age 45 years is likely a cost-saving use of public health resources.

1. Introduction

The Advisory Committee on Immunization Practices (ACIP) recommends routine human papillomavirus (HPV) vaccination at age 11 or 12 years (vaccination can be given starting at age 9 years), with catch-up vaccination for all persons through age 26 years [1]. ACIP recommends shared clinical decision-making regarding potential HPV vaccination for persons aged 27–45 years. The ACIP recommendation for shared clinical decision-making for persons aged 27–45 years, rather than a routine catch-up vaccination recommendation, was based in part on prior health economic modeling work, most of which estimated limited public health benefits and a high cost per quality adjusted life year (QALY) gained ($300,000 to $1.5 million or more) of HPV vaccination for this age group ([2]; [3–6]). However, these prior cost-effectiveness estimates considered predominantly heterosexual populations, and thus may not apply to gay, bisexual, and other men who have sex with men (GBMSM).

Vaccination of GBMSM at ages older than 26 years may be more cost-effective than vaccination of heterosexual populations for three reasons. First, GBMSM, particularly GBMSM with HIV, have a higher incidence of HPV-associated anal cancer than non-GBMSM [7,8]. Second, GBMSM, on average, have a higher partner acquisition rate than non-GBMSM [9]. Third, while average partner acquisition rates of non-GBMSM decrease with age, average partner acquisition rates of GBMSM increase through age 45 years [9,10]. Thus, vaccinating older GBMSM may avert more HPV infections and HPV disease than vaccinating older non-GBMSM. In this study, we estimate the cost-effectiveness of incrementally raising the upper age limit of vaccination from 26 years of age to age 30, 35, 40, or 45 years of age in GBMSM in the United States to inform vaccination recommendations.

2. Methods

2.1. Study questions addressed

In this study, we estimated the cost-effectiveness of HPV vaccination for GBMSM older than 26 years in the United States. For our baseline scenario, we model vaccination coverage based on historic adolescent vaccination and historic catch-up vaccination in males through age 26. For our counterfactual scenarios, we model vaccination coverage based on historic adolescent vaccination and historical catch-up vaccination in males through age 26, with additional adult vaccination among GBMSM through 30, 35, 40, or 45 years of age. We included costs to the healthcare sector and thus measured the direct costs of vaccination, and the direct medical costs of anogenital warts (AGW) and HPV-associated cancers (i.e., anal, oropharyngeal, and penile) among GBMSM. We additionally measured the loss of QALYs due to diagnosed HPV disease (cancers and anogenital warts caused by HPV). Given that cancers caused by HPV are typically diagnosed decades after the causal HPV infection, all costs and QALY losses were summed over a 100-year timeline to ensure that we captured both the immediate costs and distant benefits of vaccination. We used a 3 % annual discount rate, meaning that future costs and benefits are valued less than current costs and benefits. All costs were reported in 2023 U.S. dollars.

2.2. Model overview

We used a deterministic, susceptible-infectious-susceptible (SIS), compartmental model of a GBMSM sexual network with HPV transmission; we describe this model briefly here with a complete description of model structure and parameter derivations in the appendix. We chose to use a compartmental model, rather than a more complex agent-based model, for three reasons. First, agent-based models are useful in separating one-time sexual partners from long-term sexual partners, as long-term sexual partners have a higher probability of STI transmission due to repeat exposures. However, estimates of HPV transmission probabilities per sexual contact are very high (up to 94 %, [11]). With high probabilities of transmission per sexual contact, the likelihoods of STI transmission from one-time and long-term sexual partners converge, and thus it is less important to distinguish between one-time and long-term sexual partners within a model. Second, each model involves a tradeoff between run-time and complexity. Using a compartmental model allowed for a greater number of sensitivity analyses. Finally, several research groups use long-standing agent-based models for HPV vaccine cost effectiveness analyses (e.g. [6,12]). Diversifying the model structures used to examine HPV vaccination is a greater contribution to the literature than replicating extant agent-based HPV models. Our model meets the HPV-FRAME reporting standards for cost-effectiveness vaccination models ([13], see appendix).

The model simulated all GBMSM ages 16–60 years in the United States, stratifying individuals into 5-year age groups (ages 16–20, 21–25, 26–30, 31–35, 36–40, 41–45, 46–50, 51–55, and 56–60 years). Individuals were further stratified into groups with relatively low, medium, or high sexual activity levels (defined by sexual partner acquisition rates), and into those with or without HIV. Based on historic adolescent vaccination data, 16-year-olds entered the simulated population with a year-specific probability of having been vaccinated (Table 1). Individuals had an age and HIV status-dependent mortality rate, and 16-year-olds entered the population at a rate such that the modeled population growth rate matched projected U.S. population growth rates. In addition to moving between age groups over time, individuals transitioned from the “HIV uninfected” category to the “HIV infected” category such that age-specific HIV prevalence in our model matched age-specific HIV prevalence among U.S. GBMSM, based on unpublished data produced by the Division of HIV Prevention of the Centers for Disease Control and Prevention using data from the National HIV Surveillance System. Finally, individuals moved between sexual activity groups, such that the age-specific distribution of partner acquisition rates matched that reported by Weiss et al. [10].

Table 1.

Model assumptions for HPV vaccine efficacy, cost, and coverage.

Vaccine Characteristic	Value (Range)	Source
Efficacy and cost
Type-specific per-partner vaccine efficacy	95 % (95 % – 100 %)	Model Assumption
Cost of 3-dose series including administration	$775.23 2020 USD	CDC Vaccine Price list, see appendix for derivation
Adolescent HPV vaccination probability
Probability that 16-year-old males entering sexually active population had received adolescent vaccination	0 % in 2010, linearly increasing to 78.1 % in 2022, and staying constant thereaftera	Pingali et al. 2023
Annual probability of adult HPV vaccination
GBMSM aged 16–21 years	9.0 % (4.5 % – 13.5 %)	[16], see appendix for derivation
GBMSM aged 22 years through the upper age for vaccination (26, 30, 35, 40, or 45 years)	2.5 % (1.3 % – 3.8 %)	[16], see appendix for derivation

GBMSM: gay, bisexual, and other men who have sex with men.

^aProbability of vaccination based on percent of 16-year-olds who had received at least one dose of the vaccine.

The rate of partnerships between individuals in each compartment was determined by individuals’ sexual activity levels and a sexual mixing matrix, with individuals being more likely to select sexual partners of a similar age and matching HIV status. The rate at which susceptible individuals become infected by a given HPV type was the product of (a) the rate at which individuals acquired partners from each age/HIV-status/activity-level group, (b) the infection prevalence of that HPV type within each of those groups, (c) the type-specific probability of HPV transmission per sexual partner, and (d) 1 - vaccine effectiveness, if the susceptible individual was vaccinated. Upon transmission, individuals had a probability of developing either a transient infection which would clear at an age-dependent rate, or a persistent, lifelong infection. Males generally do not acquire strong, lasting resistance after clearing infections [14,15], and so we assume that males re-enter the susceptible class after clearing transient infections.

Individuals enter our model with a probability of having received the 9-valent vaccine based on historical trends of adolescent vaccination, and have a rate of adult vaccination that is higher for 16–21 year olds than for older adults [16], and which is capped at age 26, 30, 35, 40, or 45 years. Following previous models of vaccination in individuals older than 26 years, we assume that vaccination reduces the per-partner probability of HPV transmission by 95 %, and gives lifelong protection [2,6].

We assumed that cancer developed from persistent HPV infections. Each persistent infection had a probability of causing cancer, depending on the HPV type, cancer site, and HIV status. If cancer developed, we drew the time until cancer diagnosis from a site and HIV status-specific normal distribution. While we modeled HPV transmission only through age 60 years, we kept track of cancer diagnosis through age 100 years, factoring in the probability of survival until diagnosis.

We calculated AGW diagnosis as an age-specific rate of diagnosis per prevalent HPV 6 and HPV 11 infection.

2.3. Fitting parameter values

Our model fitting process included three parts. First, we fit HPV transmission probabilities and the proportion of HPV infections that become persistent to pre-vaccine era HPV prevalence among GBMSM, stratified by age and HIV status. Second, we fit probabilities of developing cancer and the mean and standard deviation of time until cancer diagnosis to incident cancer diagnoses among GBMSM, stratified by site, age, and HIV status. Third, we fit rates of AGW diagnosis per prevalent HPV 6/11 infection to incident AGW diagnosis among men in the pre-vaccine period, stratified by age (Table 2, See appendix for details). We determined all other demographic, natural history, vaccine, cost, and QALY parameters using information from the literature. See appendix for parameter values.

Table 2.

Examples of cumulative disease incidence used for model fitting, and of lifetime costs and QALY losses applied to model outputs.^a

Health Outcome (GBMSM age range)	Cumulative incidence/100,000 person years among GBMSM without HIV	Cumulative incidence/100,000 person years among GBMSM with HIV	Lifetime cost in 2023 USD per diagnosis (range)	Lifetime QALYs lost per diagnosis (range)b,c
Anogenital warts (31–35 years)	289.0	289.0	$989 ($868 – $1110)	0.024 (0.008–0.1)
Anal cancer (61–65 years)	44.9	107.5	$132,000 ($123,000 – $141,000)	4.1 (3.2–5.6)
Oropharyngeal cancer (61–65 years)	52.1	124.6	$144,000 ($136,000 – $148,000)	3.9 (3.5–4.8)
Penile cancer (61–65 years)	2.9	7.0	$93,100 ($75,100 – $111,000)	3.7 (3.0–4.9)

GBMSM: gay, bisexual, and other men who have sex with men.

QALY: Quality adjusted life years.

USD: United States dollars.

^aFor cancer, incidence for fitting came from 2018. For anogenital warts, incidence from fitting came from data collected prior to 2011. See appendix for details.

^bTotal QALY loss varies by age at diagnosis based on discounted years of potential life lost in the case of mortality.

^cSee Figs. S2–S10 for all age-specific fitting targets, and Chesson et al. [3] for age-specific QALY loss estimates.

2.4. Simulation timelines

We first simulated HPV transmission dynamics from 1960 to 2010 (i. e. the period before the 2011 ACIP sex-neutral vaccination recommendation, [17]) to fit HPV natural history parameters within our model. We then ran our fit model from 2011 to 2023 (from the beginning of sex-neutral vaccine recommendations to the current period, defined by the most recent year for which vaccination coverage data was available at time of analysis), implementing historic adolescent and catchup vaccination rates. We then further ran our fit model from years 2024 to 2124, either leaving adolescent and catchup vaccination rates unchanged (baseline scenario) or implementing vaccination among older GBMSM through age 30, 35, 40, or 45 years.

2.5. Cost-effectiveness ratios

We estimated the incremental cost-effectiveness ratio (ICER) of vaccinating GBMSM older than age 26 years, through ages 30, 35, 40, or 45 years. Specifically, we calculated the cost per QALY gained for each iterative increase in the upper age for vaccination from age 26 years. We calculated ICERs according to the expression:

$$\mathit{\text{ICER}}=\frac{C_{\mathit{\text{high}}\_age}-C_{low\_age}}{{\mathit{\text{QLoss}}}_{low\_age}-{\mathit{\text{QLoss}}}_{\mathit{\text{high}}\_age}}$$

where C was the cost of vaccination for GBMSM aged 16 years and above plus the cost of treating HPV disease, QLoss was the loss of QALYs due to HPV disease, and high_age and low_age refered to higher and lower ages of vaccination used for incremental comparison. We conduct 4 incremental comparisons of upper ages of vaccination: age 30 years compared to age 26 years, age 35 years compared to age 30 years, age 40 years compared to age 35 years, and age 45 years compared to age 40 years.

2.6. Sensitivity analyses

We conducted the following sensitivity analyses.

First, for our main analysis, we used an annual discount rate of 3 %. However, since vaccination incurs an upfront cost to avoid future QALY losses and treatment costs, higher discount rates will lead to higher ICER values. We thus calculated ICER values for annual discount rates of 0, 2, 3, and 5 %.

Second, for our main analysis, we used an analytical timeline of 100 years. Given that vaccination incurs upfront costs to prevent future disease, the cost-benefit ratio will decrease as we extend the analytic timeline. Thus, we ran additional analyses using timelines of 50 or 75 years.

Third, for our main analysis, we used median treatment cost and QALY loss estimates for HPV-associated cancers from Chesson et al. [3] and Clay et al. [18]. We ran additional analyses using lower and upper bounds of treatment cost and QALY loss estimates.

Fourth, pathogen transmission responds non-linearly to increases in vaccination coverage [19], meaning that ICER values will depend on the rate of vaccination. Thus, we estimated ICERs with adult vaccination rates increased or decreased by 50 %.

Fifth, due to uncertainty in vaccine efficacy estimates, we ran an analysis with 100 % vaccine efficacy.

Sixth, we assumed no anal cancer screening in our main analysis. However, in 2024, the International Anal Neoplasia Society (https://pubmed.ncbi.nlm.nih.gov/38297406/) as well as the U.S. Adult and Adolescent Opportunistic Infection Guidelines ([20]; “What’s New” 2024) published recommendations for routine anal cancer screening for groups with a high likelihood of anal cancer, in particular, GBMSM with HIV. Thus, we performed a sensitivity analysis reducing the number of anal cancer diagnoses, assuming widespread anal cancer screening (see appendix for details).

Seventh, we tested the sensitivity of our results to assumed interactions between HIV and HPV. Individuals with HIV have higher rates of HPV incidence than individuals without HIV. This pattern could be caused by (a) higher partner acquisition rates among individuals with HIV or (b) distinct HPV natural history among individuals with HIV [21]. Lacking data on partner acquisition rates for individuals with vs. without HIV, we assumed that higher HPV prevalence among GBMSM was caused by higher susceptibility of people with HIV to HPV infection and a higher likelihood that the infection would become persistent in individuals with HIV. To test whether our results are unduly influenced by assumptions about HIV and HPV interactions, we ran our simulations with future HIV incidence set to 0, minimizing the impact of HIV infection on our results.

Finally, our model structure differed from the agent based [2,6,12] and highly simplified [3] model structures used previously to estimate the cost-effectiveness of HPV vaccination for all persons aged 27–45 years. These past studies showed that vaccinating all persons aged 27–45 years would not be cost-effective. Thus, if our results show that vaccinating GBMSM aged 27–45 years is cost-effective, this differing result may be due to either model structure or population characteristics. To control for differences in results due to model structure, we ran our model replacing GBMSM population characteristics (sexual behavior, HPV prevalence, and disease incidence) with characteristics of the entire population, including GBMSM (see appendix for details). Thus, we were able to compare the cost-effectiveness of vaccinating all persons aged 27–45 years with the cost-effectiveness of vaccinating GBMSM aged 27–45 years under a single model structure.

3. Results

3.1. Health impact of HPV vaccination

In our main analysis, comparing simulations with vaccination through age 45 years to simulations with vaccination through age 26 years, we estimated that HPV vaccination of GBMSM aged 27 through 45 years would avert a median of 24,525 additional AGW cases, 5407 additional anal cancer cases, 8433 additional oropharyngeal cancer cases, and 1232 additional penile cancer cases over the next 100 years among GBMSM (Table 3). Most of the additional AGW cases averted would be averted over the next 20 years, while most of the additional cancer cases averted would be averted 35 to 55 years from now, due to the time between HPV infection and cancer diagnosis (Fig. 1). Vaccination through age 45 years rather than 40 years would result in more vaccine administrations than the preceding incremental increases in the upper age for vaccination as fewer individuals in the 40 to 45 year age group would have received adolescent or catchup vaccination (Table 3), while vaccination through age 35 rather than 30 years would avert more cancers than other incremental increases in the upper age for vaccination (Table 3). Our estimates for the number of cancer and AGW cases averted assume an initial population size of 4,706,000 GBMSM, growing to a population size of 7,320,000 by 2125.

Table 3.

Number vaccinations and HPV disease cases among gay, bisexual, and other men who have sex with men (GBMSM) in the United States from 2025 to 2124 under different upper ages of vaccination. Bold rows represent the estimated number of gay, bisexual, and other men who have sex with men (GBMSM) who would be (a) vaccinated against HPV or (b) diagnosed with a disease caused by HPV under different upper ages of vaccination. Un-bolded rows represent differences in these estimates between a vaccination scenario and the status quo scenario (“vs. vaccination through age 26 years”) or through the next lowest upper age of vaccination (“vs. vaccination through previous upper age group”).^a

	Upper age for vaccination
	26 years	30 years	35 years	40 years	45 years
Vaccinated GBMSM aged ≥ 16 years (N)	1,466,476	1,639,535	1,868,747	2,111,839	2,369,675
vs. vaccination through age 26 years		173,058	402,270	645,362	903,199
vs. vaccination through previous upper age group		173,058	229,212	243,092	257,837
Anogenital wart cases diagnosed (N)	264,582	259,172	253,486	246,995	240,788
vs. vaccination through age 26 years		−5350	−11,175	−17,878	−24,525
vs. vaccination through previous upper age group		−5350	−5791	−6743	−6606
Anal cancer cases diagnosed (N)	32,253	30,984	29,562	28,148	26,933
vs. vaccination through age 26 years		−1275	−2755	−4207	−5407
vs. vaccination through previous upper age group		−1275	−1476	−1450	−1201
Oropharyngeal cancer cases diagnosed (N)	51,974	49,991	47,722	45,537	43,775
vs. vaccination through age 26 years		−2031	−4349	−6612	−8433
vs. vaccination through previous upper age group		−2031	−2318	−2220	−1808
Penile cancer cases diagnosed (N)	8088	7773	7434	7108	6891
vs. vaccination through age 26 years		−323	−662	−978	−1232
vs. vaccination through previous upper age group		−323	−342	−312	−236

^aValues are medians across posterior parameter sets. These values were not discounted.

Fig. 1.

Additional number of cases of disease averted over time for different upper age for vaccination scenarios. Y-axis shows the cumulative number of HPV-associated cancers, or anogenital warts averted over time on the X-axis, compared to the baseline scenario where the upper age for vaccination is 26 years. Values are medians across posterior parameter sets. These values were not discounted.

3.2. Vaccination costs, medical costs averted, and QALYs gained due to vaccination

In our main analysis, we estimated that raising the upper age for vaccination from age 26 to age 45 would result in a median value of $347,420,733 in additional vaccination costs. Raising the upper age for vaccination would result in a median value of $455,391,408 in medical costs averted, and an additional 10,563 QALYs gained among GBMSM (Table 4).

Table 4.

Summed vaccine costs, treatment costs, and QALY losses under different upper ages for vaccination in the United States from 2025 to 2124 for gay, bisexual, and other men who have sex with men (GBMSM)^a. Bold rows represent the estimated summed costs or QALY losses among GBMSM under different upper ages of vaccination. Un-bolded rows represent differences in these estimates between a vaccination scenario and the status quo scenario (“vs. vaccination through age 26 years”) or through the next lowest upper age of vaccination (“vs. vaccination through previous upper age group”).

	Upper age for vaccination
	26 years	30 years	35 years	40 years	45 years
Vaccine cost (2023 USD)	381,307,348	436,374,738	519,830,262	617,831,514	728,728,188
vs. vaccination through age 26 years		55,067,272	138,522,808	236,524,059	347,420,733
vs. vaccination through previous upper age group		55,067,272	83,455,521	98,001,251	110,896,737
Treatment cost (2023 USD)	3,547,613,483	3,454,385,462	3,332,278,239	3,200,173,176	3,090,917,699
vs. vaccination through age 26 years		−94,619,635	−215,556,529	−343,477,002	−455,391,408
vs. vaccination through previous upper age group		−94,619,635	−120,867,086	−127,854,922	−111,272,266
QALYs lost	85,504	83,123	80,271	77,210	74,882
vs. vaccination through age 26 years		−2416	−5308	−8171	−10,563
vs. vaccination through previous upper age group		−2416	−2878	−2895	−2360
Cost per QALY gained (2023 USD) b
vs. vaccination through age 26 years		<$0	<$0	<$0	<$0
vs. vaccination through previous upper age group		<$0	<$0	<$0	<$0

^aValues are medians across posterior parameter sets. These values are summed over a 100-year timeline and discounted at an annual rate of 3 %.

^bA cost per QALY gained of < $0 indicates that the vaccination strategy was cost-saving relative to the comparison strategy.

3.3. Cost-effectiveness of vaccination

We estimated that raising the upper age for vaccination from 26 years to 30 years, 30 years to 35 years, and 35 years to 40 years would be cost-saving, as vaccination costs would be less than medical costs averted for each iterative increase in the upper age for vaccination (Table 4, Fig. 2). We estimated a cost-effectiveness ratio of <$0 [IQR: <$0 – <$0] per QALY gained for incrementally raising the upper age for vaccination to 30, 35, and 40 years, and < $0 [IQR: <$0 – $2086] dollars per QALY gained for raising the upper age for vaccination from 40 to 45 years (Fig. 2).

Fig. 2.

Y-axis shows estimated dollars spent per QALY gained (i.e. ICER) of raising the upper age for vaccination. Negative values indicate that the cost of vaccination was less than the treatment costs averted due to vaccination. Boxplots indicate median, interquartile, and full ranges of ICERs across all fit parameter sets at the end of our 100-year timeline.

3.4. Sensitivity analyses

We found that across most sensitivity analyses, vaccinating GBMSM would be cost-saving through at least age 40 years (Table 5).

Table 5.

Incremental cost-effectiveness ratios for raising the upper age for vaccination for gay, bisexual, and other men who have sex with men (GBMSM) under our main analysis and for each sensitivity analysis^a.

	Iterative increase in upper age for vaccination
	Vaccination through age 30 years (vs through age 26 years)	Vaccination through age 35 years (vs through age 30 years)	Vaccination through age 40 years (vs through age 35 years)	Vaccination through age 45 years (vs through age 26 years)
Main Analysis	<$0b	<$0	<$0	<$0
Sensitivity in annual discount rate:
Decreased to 0 %	<$0	<$0	<$0	<$0
Decreased to 2 %	<$0	<$0	<$0	<$0
Increased to 5 %	$4627	$14,221	$20,256	$41,357
Sensitivity in timeline:
Lowered to 50 years	<$0	$708	$1919	$13,265
Lowered to 75 years	<$0	<$0	<$0	<$0
Sensitivity in treatment costs and QALYc losses:
Lower bound estimates	<$0	<$0	<$0	$2553
Upper bound estimates	<$0	<$0	<$0	<$0
Sensitivity in adult vaccination rate:
Decreased by 50 %	<$0	<$0	<$0	<$0
Increased by 50 %	<$0	<$0	<$0	<$0
Per-partner vaccine efficacy raised to 100 %	<$0	<$0	<$0	$4719
Inclusion of future anal cancer screening	<$0	<$0	<$0	$8991
Future HIV incidence set to 0	<$0	<$0	<$0	$7842
Adult vaccination included for the entire population (including non-GBMSM)	$119,893	$134,574	$154,771	$167,075

^aValues are given as medians across posterior parameter sets.

^b<$0 indicates that the incremental cost-effectiveness ratio was negative, meaning that vaccination was estimated to be cost-saving.

^cQuality Adjusted Life Years.

First, we estimated that vaccinating GBMSM through age 45 years would remain cost-saving if we lower the annual discount rate to 0 or 2 %. However, raising the upper age for vaccination would never be cost-saving if we raise the annual discount rate to 5 %, as this heavily discounts the benefits of future cost savings from averted HPV disease (Fig. S17).

Second, we estimated that vaccinating GBMSM through age 45 years would remain cost-saving if we use an analytical timeline of 75 years, but vaccinating GBMSM would only remain cost-saving through age 30 years if we use an analytical timeline of 50 years (Fig. S18).

Third, we estimated that vaccinating GBMSM through age 45 years would remain cost-saving if we use upper bound estimates of cancer treatment costs and QALY losses due to cancer. If we use lower bound estimates, then vaccinating GBMSM would only remain cost-saving through age 40 years (Fig. S19).

Fourth, we estimated that vaccinating GBMSM through age 45 years would remain cost-saving regardless of the vaccination rate (Fig. S20).

For sensitivity analyses 5 through 7, vaccinating GBMSM would remain cost-saving only through age 40 years if we assume a vaccine efficacy of 100 %, if we include future anal cancer screening, or if we set future HIV incidence to 0 (Fig. S21–23).

Finally, when applying our model to all individuals (including non-GBMSM), we estimated median ICERs ranging from $119,893/QALY gained to $167,075/QALY gained, depending on the upper age for vaccination (Fig. S24). Thus, we estimated that raising the upper age for vaccination would be much more cost-effective among GBMSM than among all individuals.

4. Discussion

We estimated that vaccination for GBMSM through age 40 years compared to vaccination through age 26 years would be cost-saving, and thus highly cost-effective. This contrasts with previous modeling of vaccination in all individuals that showed that vaccination at ages older than 26 years was not cost-effective. The result that vaccination of GBMSM through age 40 years was cost-saving was robust across nearly all sensitivity analyses. These results derive in part from to how sexual behavior of GBMSM changes with age; while sexual partner recruitment peaks among non-GBMSM in their early 20’s, the rate at which GBMSM acquire new sexual partners increases on average through at least 45 years of age [10].

Our results showed that raising the upper age for vaccination among GBMSM from age 26 through age 45 years was considerably more cost-effective than raising the upper age for vaccination among all individuals from age 26 through age 45 years. Most prior studies have estimated that raising the upper age for vaccination among all individuals to 45 years would cost more than $300,000/QALY, and up to $1.5 million/QALY or more. [2,3,6]. It is possible that our ICER estimates for GBMSM were lower than these prior ICER estimates for all individuals due to differences in model structure; Laprise et al. [2] and Kim et al. [6] used an agent-based transmission model, while Chesson et al. [3] did not explicitly model HPV transmission, but rather assumed a linear decrease in HPV incidence with increasing vaccine coverage. When we applied our modeling structure to all individuals, we estimated significantly higher ICER values than for the GBMSM population, but lower ICERs than these prior studies that estimated the cost-effectiveness of raising the upper age for vaccination in all individuals, even when the prior agent based model focused on sub-populations that were more likely to benefit from adult vaccination (e.g. recently divorced/separated individuals, [12]). One possible reason for this difference is that we did not include natural immunity; if individuals in our model had developed short-term natural immunity to reinfection by the same HPV type after clearing an infection, the cost-effectiveness of vaccination would likely be less favorable, though we note that HPV antibodies show no sign of conferring resistance to infection among GBMSM [14]. Thus, while we are confident that vaccination of GBMSM aged 27–45 years will be much more cost-effective than vaccination of all adults aged 27–45 years, other modeling approaches should be used to validate whether vaccination of GBMSM aged 27–45 years would be cost-saving.

Our results were consistent across all sensitivity analyses except for when we excluded or devalued the future health benefits of vaccination. Increasing the rate of vaccination among older GBMSM will impose immediate costs to the healthcare system. The annual costs of vaccination will decrease over time as individuals who received adolescent vaccination age, reducing the proportion of the adult population who were not vaccinated as adolescents and thus eligible for adult vaccination. On the other hand, the health benefits of adult vaccination will not occur for many years, because it usually takes decades for a persistent HPV infection to progress to diagnosed cancer [22]. Thus, our analyses showed that vaccinating GBMSM older than age 26 years was not cost-saving when we increased the annual discount rate from 3 % to 5 %, which further reduced the importance of future costs and benefits compared to current costs and benefits. Similarly, vaccinating GBMSM older than age 30 years was not cost-saving when we cut off our analytic timeline at 50 years (i.e. 2075). However, even in these sensitivity analyses, our estimated ICERs were still significantly lower (<$14,000) than past estimates for vaccinating individuals older than age 26 years in all individuals [2,3,6].

Our analysis applies primarily to sexually active GBMSM. HPV transmission in our model was fit to data from studies that measured HPV prevalence among convenience samples from the general GBMSM population, as well data from studies that measured HPV prevalence among GBMSM in sexually transmitted infection (STI) clinics, GBMSM in anal-cancer-screening clinics, and GBMSM receiving HIV services [23–30]. Assuming that GBMSM in STI/HIV clinics are more sexually active on average than the general GBMSM population, our model is skewed towards a subset of the GBMSM population with increased sexual activity. Thus, our findings are most suitable for informing vaccination strategies for GBMSM who are sexually active and not in a monogamous partnership.

We assumed that GBMSM aged 27–45 years who opt for HPV vaccination would receive 3 doses, consistent with current ACIP recommendations. However, data have shown high efficacy with fewer doses [31–33]; in 2022, the World Health Organization recommended two doses for persons aged 9 years and older with an option for one dose for those aged 9 through 20 years. HPV vaccination of GBMSM through age 45 years would be even more cost-effective than we estimated if fewer than 3 doses were required.

Despite model uncertainty, we consistently found that vaccination of GBMSM aged 27–45 years was more cost-effective than vaccination of all adults aged 27–45 years and was cost-saving in our main analysis. ACIP, as well as the HIV opportunistic infection guidelines and the HIV medicine association Primary Care for HIV guidelines all currently recommend shared decision making for HPV vaccination among GBMSM aged 27–45 years, and these results could help to inform those shared decisions and inform future recommendations for HPV vaccination of GBMSM.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.vaccine.2025.127798.

Data availability

Data and code will be accessible at github.com/patrickaclay at time of publication