PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of hviLink to Publisher's site
 
Hum Vaccin Immunother. 2013 May 1; 9(5): 1146–1152.
Published online 2013 January 16. doi:  10.4161/hv.23456
PMCID: PMC3899152

Using risk to target HPV vaccines in high-risk, low-resource organizations

Abstract

Organizations in developed countries with limited financial resources may find it difficult to determine whether it is preferable to use these resources for HPV vaccination, management of HPV-related diseases, or a “hybrid” strategy, such as vaccinating only the highest risk individuals. We determined the organizational costs and clinical impacts of three different organizational approaches to female HPV vaccination in a low-resource setting, including vaccinating everyone, vaccinating no one, or vaccinating only those considered high-risk. To determine patients at highest risk, HPV risk factors were identified using information routinely gathered at the annual preventive maintenance visit. The three vaccination strategies were then compared using a decision tree analysis. The three strategies demonstrated very little difference in cost. However, the least expensive strategy was to vaccinate no one. In contrast, the strategy with the best clinical outcomes was for the organization to vaccinate everyone. Organizations with limited resources must decide how to best allocate these funds to provide the greatest clinical benefits. This study showed little difference in costs but improved clinical outcomes when using the universal HPV vaccination strategy. Thus, the improvement in clinical outcomes when vaccinating everyone may be worth the relatively small increase in cost of vaccinating everyone.

Keywords: human papillomavirus vaccine, human papillomavirus, vaccines, decision tree

Introduction

The human papillomavirus is a common and costly virus causing cervical cancer and contributing to other HPV-related diseases such as vulvar cancer, vaginal cancer, anal cancer, oropharyngeal cancer and genital warts. Among 20 to 24 y olds, 45% of females carry the virus at any one time.1 The United States spends five billion dollars per year on the prevention and treatment of HPV, not including vaccination.2

Two vaccines are available for preventing infection with HPV. Gardasil® (Merck) is approved for the prevention of HPV-6, HPV-11, HPV-16 and HPV-18 in males and females ages 9 to 26 y old while Cervarix® (GlaxoSmithKline) is available for the prevention of HPV-16 and HPV-18 in females 10 to 26 y old. HPV-16 and HPV-18 cause 70% of all cases of cervical cancer3,4 and 50–60% of CIN 2 and 3 abnormal pap smears,5 while HPV-6 and HPV-11 cause 90% of genital wart cases.6 Therefore, HPV vaccines have the potential to substantially decrease the number of abnormal pap smears and cervical cancer cases in the United States and the world.

In 2006, the ACIP recommended that all females receive the HPV vaccine starting at age 11 to 12, with catch-up vaccination through age 267 and in 2011, the group extended the same recommendation to males. In clinical trials for Gardasil®, 27% of participants had already been exposed to a vaccine-type HPV strain, most with just one strain,7 supporting the recommendation for universal use. Five years after the recommendation for females was established, HPV vaccine uptake still remains low, especially among young adult females. Estimates suggest that series initiation is only 20% for the young adult population and series completion rates substantially lower.8

Organizations in developed countries with limited financial resources serving a mostly uninsured population, such as Planned Parenthood, often have difficulty supporting the cost of the HPV vaccines, which is the most expensive routinely recommended vaccine for adolescents and young adults. The private sector cost for Gardasil® is $135 and $128 for Cervarix®.9 Added to the cost is the additional fee for vaccine administration. In addition to the high vaccine cost is the problem that organizations with limited financial resources typically serve populations at high risk for HPV, such as young adults and those with multiple sexual partners, who are often un- or underinsured. Thus, organizations must often finance the cost burden associated with HPV-related diseases for their patients, such as abnormal Pap smear management or genital wart treatment.

For organizations with limited financial resources it can be difficult to determine if the best strategy is to use these resources for prevention, for example via HPV vaccination, management of HPV-related diseases, or some type of “hybrid” strategy, such as vaccinating only the highest risk individuals. Risk stratification to allocate limited resources is already being used by these organization for some services, for example paying for the cost of sexually transmitted infection testing or birth control methods, but only among certain high-risk subgroups such as adolescents or individuals with new sexual partners (for example, see www.ipp.jsi.com). It is unknown whether a similar risk-based strategy would be feasible or advantageous for an organization if applied to HPV vaccine administration, even though targeted vaccination is not a feasible strategy at the population level.10 To address this question, we used a decision tree analysis to assess the organizational costs and clinical impacts of three different organizational approaches to female HPV vaccination in a low-resource setting, including vaccinating everyone, vaccinating no one, or vaccinating only those considered high-risk.

Results

A total of 678 participants were included in the health center sample used to parameterize the model. Most participants (72.8%) were uninsured, while 18.6% were privately insured and 8.6% had public insurance. Very few data points were missing from the sample. See Table 1 for complete sample data.

Table thumbnail
Table 1. Characteristics of participants used to assess association of risk factors with HPV-related outcomes (n = 678)

Assessment of risk factors

Risk factors described below (partner change in the past six months, years of sexual activity, condom use, forced sex, smoking, history of oral contraceptive use and family history of cervical cancer) were assessed for associations with two HPV-related outcomes—abnormal Pap smears and genital warts. Of these factors, only six or more years of sexual activity (x2 = 35.533, p = 0.000) and smoking (x2 = 5.349, p = 0.021) were significantly associated with a history of abnormal Pap smears. For a complete list of bivariate calculations, see Table 2. In a multivariable logistic regression model that included years of sexual activity and smoking, only years of sexual activity remained significantly associated with a history of abnormal Pap smear (p = 0.000). Thus, six or more years of sexual activity was used as the decision point in the risk-based vaccination strategy arm of the model. A cutoff of six or more years of sexual activity was chosen, as this was the point that showed the largest significance when testing bivariate associations of risk factors for HPV-related outcomes.

Table thumbnail
Table 2. Bivariate associations between risk factors and HPV-related outcomes

Decision tree analysis

There were only small variations in clinical costs and outcomes among the three strategies. The least expensive organizational strategy was to vaccinate no one ($123.42 per person). The organizational strategy to vaccinate everyone prevented the greatest number of abnormal Pap smears (see Table 3).

Table thumbnail
Table 3. Decision tree analysis outcomes for three vaccination strategies

One-way sensitivity analyses demonstrated that vaccinating based on risk factors was never the least expensive strategy, even when large ranges in costs and probabilities were considered. Instead, the model overall was most sensitive to varying the cost of the vaccine, the cost of an abnormal Pap smear, the probability of being vaccinated if the organization vaccinates no one (for example, if people receive the vaccine elsewhere) and the probability of unvaccinated patients having a normal Pap smear. For example, a threshold value of $81 was found for the cost of the vaccine, such that if the cost of the vaccine was $81 or less, the less expensive strategy from an organizational standpoint was to vaccinate everyone, whereas when the cost of the vaccine exceeded $81, the less expensive strategy was to vaccination no one. See Table 4 for all threshold values.

Table thumbnail
Table 4. Baseline values for decision tree model

Discussion

Key results

Prior to this study, targeting vaccination based on risk factors was not found to be useful in the general population as choosing females with certain risk factors excluded too many others who would have benefited.10 However, for this study, it was hypothesized that targeting vaccination may be a useful strategy for high-risk, low-resource organizations. This study found that hypothesis to be incorrect. Under no conditions assessed (either using base-case estimates or in any of the sensitivity analyses) in the model was targeting vaccination using risk factors better than vaccinating everyone or no one in terms of costs and clinical outcomes. Instead, under most conditions, vaccinating no one was found to be least expensive strategy and vaccinating everyone was found to prevent greatest numbers of abnormal Pap smears. However, there was surprisingly little difference in cost among the three strategies compared, indicating the improvement in clinical outcomes when vaccinating everyone may be worth the relatively small increase in cost for this strategy.

Variability in the cost of the HPV vaccine and abnormal Pap smears, as well as the probability of vaccination outside of the organization and the probability of having a normal Pap smear if unvaccinated, changed the optimal vaccination strategy chosen. Changes in these variables are clinically relevant. If the organization is able to reduce costs of vaccine administration through utilizing outside resources such as the Merck Vaccine Patient Assistance program, the benefits of vaccinating everyone clearly exceed vaccinating no one.

This model assumed everyone in the 19 to 26 y old age group receives a Pap smear every year. However, updated recommendations by the American College of Obstetricians and Gynecologists for cervical cancer screening advise no Pap smear testing until age 21, and then Pap smear testing every two years thereafter. Decreased frequency of Pap smears overall and decreased probability of abnormal Pap smears influence the probabilities and costs of Pap smears for an organization, which may influence their vaccination strategy decision. However, the model was not found to be sensitive to even large ranges in the cost of normal Pap smears, indicating the updated guidelines do not likely affect the results of this study.

Finally, the model was sensitive to the number of people vaccinated outside of the organization, which is clinically relevant. As this model suggests, placing the cost burden of vaccination on health care organizations serving young adults may be too great for low-resource organizations to bear. These findings further support the ACIP recommendation for all females to receive HPV vaccination at age 11 to 12 y old. Over the next decade, if the adolescent vaccination rate increases, vaccinated adolescents will become vaccinated young adults to help decrease the burden.

Limitations

One limitation of the study was that the decision tree variable considerations were limited to costs and probabilities of HPV vaccination and Pap smears. In the population studied, 4.5% of participants reported a history of genital warts and an additional 1.2% were diagnosed with genital warts at the annual exam. Because there were no statistically significant risk factors for genital warts in this population sample, genital warts was not included in the model. Only the risk factor “years of sexual activity” had a statistically significant association with the HPV outcome “abnormal Pap smear” so only that risk factor and outcome could be used. Also, other HPV-related health outcomes such as vulvar or vaginal neoplasia were not diagnosed at any of the annual exams, the additional organizational costs of these HPV-related diseases were not included in the model. However, Gardasil® helps protect against genital warts, as well as vulvar and vaginal cancer, and therefore, potential cost savings and clinical outcomes from vaccinating may have been underestimated.

Additional simplifying assumptions were made in the decision tree model to increase model utility in practice but which may have affected the results of the study. For example, it was assumed that once someone became vaccinated, they would not have an abnormal Pap smear, despite the possibility, especially in the 19 to 26 y old age group, that participants may have been infected with HPV prior to vaccination, and that vaccination reduces but does not eliminate abnormal Pap smears since many HPV types not included in the vaccine result in Pap smear abnormalities. Therefore, people who receive vaccination may still incur an abnormal Pap smear and the associated costs. This assumption may have resulted in an overestimation of the benefits of vaccination in the model.

It was assumed that the organization could receive free vaccine for all patients from either the Merck Vaccine Patient Assistance Program or insurance, and therefore would only support the cost of administration, despite knowing that some uninsured patients would not qualify for the Merck Vaccine Patient Assistance Program and some insurance companies would not pay for vaccination, thereby underestimating the cost of the vaccine for the organization.

A 100% vaccination rate was assumed in the “vaccinate everyone” strategy, despite the likelihood that some patients would refuse the vaccine. Additionally, vaccine series completion was assumed despite knowing that many people who start the vaccine series are lost to follow up. Both of these assumptions would likely have overestimated vaccine cost and benefit.

In addition, some parameter estimates were decided without reference to the literature or PPMSM data, including the probability of 100% vaccination if a strategy of vaccinating everyone is used, and the probability of having a normal Pap smear once a participant was vaccinated. To minimize the impact of these limitations, extensive sensitivity analyses were performed using a wide range of possible cost and probability estimates for all variables. Overall, we found our model to be generally robust in its conclusions.

A final limitation of the study was the use of cross-sectional data among a small population to define parameter estimates. The lack of longitudinal data prevented the predictive ability of HPV infection outcomes from risk factors and relied instead on statistical associations. Also, using just two reproductive health centers in the population studied and the data available there limited results. To try to decrease selection bias, the health centers were in large, racially diverse cities in southern Michigan, all female patients presenting for annual exams were included and two health centers were used instead of just one. However, because targeting was not found to be the preferred vaccination strategy, the significance of this restriction is limited.

Generalizability

This study was specific to a population with higher rates of abnormal Pap smears and a higher rate of uninsured individuals than the greater population. This may limit the generalizability of the study to other populations and organizations. However, sensitivity analyses provided useful threshold values for other organizations considering use of this model. Furthermore, ranges used in the sensitivity analyses were likely outside of plausible ranges for many parameters, but the overall model remained robust, which supports the generalizability of the model.

Methods

Overall study design

Clinical and economic impacts were assessed using a decision tree analysis for three different HPV vaccination approaches of young adult females attending low-resource health centers serving a high-risk population. The three HPV vaccination approaches were (1) vaccinating all adult (19 to 26 y old) females in this health center population, (2) vaccinating none of the females in this population or (3) vaccinating only the proportion of this population identified as having risk factors associated with HPV infection and/or disease.

Risk factors

To increase applicability of the decision tree model to the high-risk population seen by low-resource organizations such as Planned Parenthood, we first assessed whether a variety of behavioral/historical factors identified in other studies11,12 were associated with the HPV-related outcomes of abnormal Pap smears and genital warts in this particular patient population. For the risk factor analyses, we considered only those variables that could be readily elicited during a typical clinical encounter and therefore potentially used for delivering a targeted vaccination strategy to this clinical population. The risk factors assessed included whether there had been a sex partner change in the past six months (yes/no), age at first intercourse, condom use (never or sometimes/usually or always), history of forced sex (yes/no), current smoking (yes/no), history of oral contraceptive use (yes/no) and family history of cervical cancer (yes/no). All risk factors analyzed were dichotomous except the continuous variable “age at first intercourse.” To facilitate inclusion in a decision tree analysis, age at first intercourse was transformed into a categorical variable by converting it to number of years of sexual activity and dividing it at the point where association with HPV infection (history of abnormal Pap smear specifically) was greatest (for example, zero to five years of sexual activity and six or more years of sexual activity).

Associations between these risk factors and HPV-related outcomes (abnormal Pap smears or genital warts) were evaluated based on cross-sectional data derived from two high-risk, low-resource reproductive health centers in the PPMSM affiliate. Data was collected between May 1, 2010 and October 31, 2010, representing the first six months of HPV vaccine availability at these centers. The “Notice of Health Information Privacy Practices” provided to patients at registration served as informed consent for the research, as no personally identifiable information was collected during the chart review. Participants were included in the convenience sample if they (1) attended an annual exam during the six months of data collection, a visit when the necessary history and laboratory results were obtained, (2) were female, as the ACIP recommendation for universal vaccination applied only to females at the time and (3) were between the ages of 19 and 26, as the reproductive health centers in the study limit vaccine availability to this age group. All patient history forms and laboratory results were identical between the two health centers and routinely used by the health centers for patient care.

To determine which risk factors might be useful for a targeted vaccination strategy, bivariate associations between the above individual risk factors and each of the two HPV-related outcomes (history of abnormal Pap smear and history of genital warts) were calculated using chi-square tests. Of the two HPV-related outcomes, only history of abnormal Pap smear was found to have significant risk factors. Therefore, history of genital warts was excluded from further analysis. Risk factors found in the bivariate analyses to be significant (p < 0.05) were then included in a multivariable model to determine independent predictors of this outcome.

Decision tree structure

Decision tree analysis utilizes an algorithm approach to assessing uncertainty in costs and probabilities, and quantifies the value of outcomes in different scenarios, allowing for objective decision-making. To assess the impact of the three different vaccination strategies, a decision tree analysis was used to evaluate resulting costs and clinical endpoints. When available, the model was parameterized using data derived directly from PPMSM data. The decision tree assumed annual costs, probabilities and projected outcomes over a one-year period. Thus, no discounting for future costs was considered. This time frame was selected as low-resource organizations tend to rely on fluctuating funding that is determined annually, and often have a somewhat transient patient population that may not use care in these settings longitudinally.

Probabilities

We used PPMSM data specifically to define the cost of the vaccine, the probability of being vaccinated with HPV previously (for risk-based vaccination and vaccinating no one strategies), and the probability of having a normal or abnormal Pap smear if unvaccinated. Parameters that could not be defined from PPMSM data directly (costs of normal and abnormal Pap smears) were developed using estimates from the literature (Insinga, Glass, and Rush, 2004). See Table 4 for baseline values for the model and ranges used in sensitivity analyses (described below).

Probability of vaccination

When vaccinating everyone, the probability variable for becoming vaccinated was valued at 1.0, recognizing that a 100% vaccination rate may not be replicated in reality, but was used as a best-case scenario for the model. The probability variable for being vaccinated when the organization vaccinated no one was 0.18, as PPMSM data indicated that 18% of the population studied had previously initiated or completed vaccinated elsewhere. When utilizing a risk-based vaccination strategy, the percentage of participants in the two reproductive health centers with the particular risk factor being tested was used as the probability of vaccination variable.

Probability of normal and abnormal Pap smears

To determine the probability variables for normal and abnormal Pap smears if a participant was vaccinated, a best-case scenario was applied such that all vaccinated individuals were assumed to have normal Pap smears. For unvaccinated participants, it was calculated that 12% would have an abnormal Pap smear, based on Pap smear results from the annual exam during the time period studied at the two reproductive health centers.

Costs

The cost of the vaccine was considered to be $90, derived from the cost the two reproductive health centers charged patients in administration fees for receiving the full three dose series. Costs of the purchase of Gardasil® were assumed to be covered by either the patient’s insurance, or more commonly, the Merck Vaccine Patient Assistance Program, which can be utilized by the mostly uninsured, low-income patient population. GlaxoSmithKline has the GSK Vaccines Access Program that would cover the cost of Cervarix® for low-income patients in organizations using Cervarix®.

Costs for normal and abnormal Pap smears were derived from a study using Kaiser Permanente Northwest data13 (see Table 4). In this study, the researchers analyzed the costs of cervical HPV from 103,476 health plan participants between 1997 to 2002. The results were applicable to this model as they provided a per person average cost of normal and abnormal Pap smears, although 10% of the overall costs were attributable to invasive cervical cancer which low-resource organizations rarely support due to referral of services.

Sensitivity analyses

A one-way sensitivity analysis on all cost and probability variables was run to determine the impact of parameter uncertainties on the model. A wide range of possible probabilities and costs were applied to increase model utility beyond the specific organization used to develop the model parameters. See Table 4 for sensitivity analysis ranges. For the cost of the vaccine, the range included no cost as well as the total cost of the vaccine plus administrative fees charged by the two reproductive health care organizations for patients without insurance or the Merck Vaccine Patient Assistance Program ($540).

Statistics

For the initial analysis to determine risk factors, we used PASWStatistics 18.0 to calculate chi-square results for bivariate associations between risk variables and HPV-related outcomes. We then performed logistic regression from the statistically significant bivariate associations to establish independent predictors of HPV-related outcomes, and calculated demographic and descriptive data. To develop and test the three vaccination strategies, TreeAge Pro 2011 was utilized for decision tree analysis. Sensitivity analyses were then performed on all cost and probability variables to test the base case parameter values described above. Participants with missing data for a variable were excluded.

Approvals

The University of Michigan Institutional Review Board, Planned Parenthood Mid and South Michigan, and Planned Parenthood Federation of America, Inc. approved all study activities. The findings and conclusions in this article are those of the authors and do not necessarily represent the views of Planned Parenthood Federation of America, Inc.

Conclusion

Many health care organizations are able to rely on insurance companies for reimbursement of the cost of HPV vaccines and abnormal Pap smear follow up. However, some organizations supporting a largely uninsured population have to make decisions on how to best allocate limited funds to provide the greatest good. This study showed little difference in cost outcomes whether or not the organization supported the cost of HPV vaccination for everyone, no one, or those at high risk, although vaccinating no one was still the least expensive option and vaccinating everyone produced the least number of abnormal Pap smears. Thus, the improvement in clinical outcomes when vaccinating everyone may be worth the relatively small increase in cost of vaccinating everyone.

Acknowledgments

The Nurse Practitioner Healthcare Foundation/sanofi Pasteur Health Through Immunizations Award provided investigator support for this study. The authors would like to thank Maris Vinovskis, PhD for input into earlier versions of the manuscript.

Financial Disclosure

The Nurse Practitioner Healthcare Foundation/sanofi Pasteur Health Through Immunizations Award provided investigator support for this study.

Glossary

Abbreviations:

ACIP
Advisory Committee on Immunization Practices
HPV
human papillomavirus
PPMSM
Planned Parenthood Mid and South Michigan

Submitted

Submitted

10/19/12

Revised

Revised

12/15/12

Accepted

Accepted

12/26/12

Disclosure of Potential Conflicts of Interest

Disclosure of Potential Conflicts of Interest

Amanda Dempsey serves on an Advisory Board for Merck related to HPV vaccination and for Pfizer related to meningococcal vaccines. Neither company had any role in this project and are unaware of the study’s results. Dr Dempsey does not receive research funding from these companies.

Footnotes

References

1. Dunne EF, Unger ER, Sternberg M, McQuillan G, Swan DC, Patel SS, et al. Prevalence of HPV infection among females in the United States. JAMA. 2007;297:813–9. doi: 10.1001/jama.297.8.813. [PubMed] [Cross Ref]
2. Insinga RP, Dasbach EJ, Elbasha EH. Assessing the annual economic burden of preventing and treating anogenital human papillomavirus-related disease in the US: analytic framework and review of the literature. Pharmacoeconomics. 2005;23:1107–22. doi: 10.2165/00019053-200523110-00004. [PubMed] [Cross Ref]
3. Muñoz N, Bosch FX, de Sanjosé S, Herrero R, Castellsagué X, Shah KV, et al. International Agency for Research on Cancer Multicenter Cervical Cancer Study Group Epidemiologic classification of human papillomavirus types associated with cervical cancer. N Engl J Med. 2003;348:518–27. doi: 10.1056/NEJMoa021641. [PubMed] [Cross Ref]
4. Bosch FX, Manos MM, Muñoz N, Sherman M, Jansen AM, Peto J, et al. Prevalence of human papillomavirus in cervical cancer: a worldwide perspective. International biological study on cervical cancer (IBSCC) Study Group. J Natl Cancer Inst. 1995;87:796–802. doi: 10.1093/jnci/87.11.796. [PubMed] [Cross Ref]
5. Clifford GM, Smith JS, Aguado T, Franceschi S. Comparison of HPV type distribution in high-grade cervical lesions and cervical cancer: a meta-analysis. Br J Cancer. 2003;89:101–5. doi: 10.1038/sj.bjc.6601024. [PMC free article] [PubMed] [Cross Ref]
6. Greer CE, Wheeler CM, Ladner MB, Beutner K, Coyne MY, Liang H, et al. Human papillomavirus (HPV) type distribution and serological response to HPV type 6 virus-like particles in patients with genital warts. J Clin Microbiol. 1995;33:2058–63. [PMC free article] [PubMed]
7. Markowitz LE, Dunne EF, Saraiya M, Lawson HW, Chesson H, Unger ER, Centers for Disease Control and Prevention (CDC) Advisory Committee on Immunization Practices (ACIP) Quadrivalent Human Papillomavirus Vaccine: Recommendations of the Advisory Committee on Immunization Practices (ACIP) MMWR Recomm Rep. 2007;56(RR-2):1–24. [PubMed]
8. Centers for Disease Control and Prevention [Internet]. Atlanta: Adult Vaccination Coverage, Reported via NHIS. [cited 2012]. Available from www.cdc.gov/vaccines/stats-surv/nhis/default.htm
9. Centers for Disease Control and Prevention [Internet]. Atlanta: CDC Vaccine Price List. [cited 2012]. Available from http://www.cdc.gov/vaccines/programs/vfc/cdc-vac-price-list.htm
10. Dempsey AF, Gebremariam A, Koutsky LA, Manhart L. Using risk factors to predict human papillomavirus infection: implications for targeted vaccination strategies in young adult women. Vaccine. 2008;26:1111–7. doi: 10.1016/j.vaccine.2007.11.088. [PubMed] [Cross Ref]
11. National Cancer Institute [Internet]. Cervical Cancer Prevention (PDQ®). [cited 2010]. Available from http://www.cancer.gov/cancertopics/pdq/prevention/cervical/healthProfessional
12. National Cancer Institute [Internet]. Cervical Cancer Treatment (PDQ®). [cited 2011]. Available from http://www.cancer.gov/cancertopics/pdq/treatment/cervical/HealthProfessional
13. Insinga RP, Glass AG, Rush BB. The health care costs of cervical human papillomavirus--related disease. Am J Obstet Gynecol. 2004;191:114–20. doi: 10.1016/j.ajog.2004.01.042. [PubMed] [Cross Ref]

Articles from Human Vaccines & Immunotherapeutics are provided here courtesy of Landes Bioscience