We developed an integrated transmission dynamic model and economic evaluation to inform HPV vaccine policy recommendations and decisions. We gained valuable insights by comparing various vaccination strategies. In general, the results suggest that a quadrivalent HPV vaccine program that targets female adolescents and women, ages 12–24 years, can be cost-effective ($4,666/QALY) when compared with other commonly accepted medical interventions (32
). These findings are consistent with other cohort-based cost-effectiveness analyses, which generally show that vaccination of 12-year-old girls can be cost-effective but also illustrate the substantial herd immunity benefits provided by vaccination.
Some results from this model were qualitatively similar to the results of other studies with respect to the finding that male vaccination was more attractive the lower the coverage among girls and women (9
). However, the results of our base case differ qualitatively from that of Taira et al. (9
) regarding the conclusion that vaccinating males and females would not be cost-effective. This difference in results may be explained as follows. First, unlike Taira et al., we accounted for the additional benefits conferred by protecting against HPV 6/11 infection among adolescent boys and girls, women, and men. Second, we were able to account for all the benefits and costs of vaccination realized by both those vaccinated and not vaccinated. Third, we assumed lower weights for the quality of life of women with CIN. However, the comparison is not perfect because our model tracks a population, whereas the model of Taira et al. follows a cohort. Hence, the composition of the numerators and denominators used in the ICERs differs between models. Finally, other methodologic differences occur between the 2 approaches that may explain the differences in results. For example, Taira et al. used steady-state values of HPV infection rates as inputs in their cost-effectiveness model, whereas we measured all outcomes over time, thereby capturing all the effects of transient dynamics generated from widespread vaccination. We also note that the results of the sensitivity analysis, when the effects of herd immunity and benefits of prevention of HPV 6/11 were removed, suggest that the ICER of the female vaccination strategy was $21,404/QALY, which is close to the value of $22,755/QALY reported in another study by Sanders and Taira (33
An important finding from this analysis was that catch-up vaccination can substantially reduce disease in the short term. As a result, the female and male strategy that did not include a catch-up program was less effective and more costly.
One of the influential inputs was vaccine coverage. As female coverage rates decreased, male vaccination became more efficient. Another influential input in the analysis was the quality-of-life weights. The less HPV disease affected quality of life, the more the ICERs increased.
Duration of protection was also an influential parameter. Decreasing duration of vaccine protection to 10 years increased ICERs. However, the impact of this decrease may be mitigated by introducing a booster program. A reasonable approximation for how this program might fare would be to look at the sensitivity of ICERs to changes in vaccination cost. Thus, increasing the cost of the HPV vaccine series to $500 increased ICERs (). However, all nondominated (i.e., either are less costly or have lower ICERs than more effective strategies) female strategies remained cost-effective. Another influential parameter was the age vaccination was begun. Earlier vaccination resulted in greater benefits. F&M12+CUF&M was cost-effective ($42,697/QALY). However, vaccination by age 12 became less efficient, the higher the vaccination coverage was among older age groups.
Vaccination shifted the age of infection and disease to older age groups. For example, the age of peak cervical cancer incidence increased after introducing vaccination. The upward shifting of age of infection is a common feature of many vaccination programs (11
We believe our modeling approach has several strengths. First, we did extensive validation with existing data. The model is also flexible enough to incorporate better data as they become available. Second, this model accounts for actual screening practices in the United States. Third, because output from this model is population based, the comparison with national registry data is better aligned than comparison of cohort model output with population data (6
). Finally, all equations and inputs for this model are available to facilitate replication of findings and independent review of the model.
Several enhancements and extensions are desired. First, more relevant data on the natural history of type-specific HPV infection and disease (e.g., HPV transmission probability per sexual contact) are needed. Also, given the influence utility weights have on ICERs, more studies are needed to collect health utilities data on HPV disease states.
Second, we modeled only 4 HPV types and their associated diseases and assumed that HPV types have independent natural histories with no interaction among them. If cross-immunity exists between HPV types, a vaccine that reduces the prevalence of 1 type may promote the prevalence of other types through a process of competitive release. If, however, current or prior infection with 1 HPV type facilitates concurrent or subsequent infection with another HPV type, or if the vaccine provides cross-protection against other types, HPV vaccination could have the additional benefit of reducing the prevalence of HPV infection of types not covered by the vaccine (34
). The evidence on interaction among HPV types to date is mixed and inconclusive (35
Third, we modeled neither coinfection after disease developed in a person nor the coexistence of CIN lesions due to multiple HPV types in the cervix. By accounting for all the cost of vaccinating persons with undetected disease and no benefits for them as a result of the protection against the type that did not cause the disease, our results are biased against the catch-up program.
Fourth, the model assumed that all persons have equal access to healthcare, be it vaccination, screening, or treatment. However, this assumption may not be realistic and may overestimate the benefits of vaccination if women who have limited access to screening are also less likely to get vaccinated. Further studies are required to determine whether those who do not get vaccinated are also likely not to get screened.
Fifth, the current version of the model focused on heterosexual transmission of HPV and did not incorporate transmission between homosexual and heterosexual persons. Sixth, the scope of the model has been limited to cervical diseases and genital warts. HPV infection has also been associated with recurrent respiratory papillomatoses and cancers of the anus, penis, vagina, vulva, and head and neck. As evidence becomes available, the scope of the model will be broadened to incorporate the potential effects of vaccination on these other HPV conditions. Including these diseases in the model would render more favorable ICERs for vaccination.
Seventh, we did not include death and productivity costs (lost wages), as was done in other analyses (40
). Including these costs would further reduce ICERs.
Finally, we did not consider vaccination strategies that include infants or mid-adults because current data available on vaccine safety and efficacy are limited to ages 9–26 years (18
). As data for these other age groups become available, the model can examine these strategies.
In summary, the results from this model suggest that in a setting of organized cervical cancer screening, a prophylactic quadrivalent HPV (16/18/6/11) vaccine can 1) substantially reduce genital warts, CIN, and cervical cancer, 2) improve quality of life and survival, 3) be cost-effective (across a reasonably wide range of assumptions) when administered to girls before age 12 years (with or without a catch-up program), and 4) have a cost-effectiveness ratio near or below (depending on the underlying assumptions of the model) that of several other recommended vaccines, when implemented as a strategy that combines vaccination of both girls and boys before age 12 with a 12–24 years of age catch-up program.