Currently Viewing:
The American Journal of Managed Care March 2016
Understanding Vaccination Rates and Attitudes Among Patients With Rheumatoid Arthritis
Diana S. Sandler, MD; Eric M. Ruderman, MD; Tiffany Brown, MPH; Ji Young Lee, MS; Amanda Mixon, PA; David T. Liss, PhD; and David W. Baker, MD, MPH
Remembering the Strength of Weak Ties
Brian W. Powers, AB; Ashish K. Jha, MD, MPH; and Sachin H. Jain, MD, MBA
Prevalence, Effectiveness, and Characteristics of Pharmacy-Based Medication Synchronization Programs
Alexis A. Krumme, MS; Danielle L. Isaman, BS; Samuel F. Stolpe, PharmD; J. Samantha Dougherty, PhD; and Niteesh K. Choudhry, MD, PhD
Impact of Cost Sharing on Specialty Drug Utilization and Outcomes: A Review of the Evidence and Future Directions
Jalpa A. Doshi, PhD; Pengxiang Li, PhD; Vrushabh P. Ladage, BS; Amy R. Pettit, PhD; and Erin A. Taylor, PhD, MSPH
Trends in Hospital Ownership of Physician Practices and the Effect on Processes to Improve Quality
Tara F. Bishop, MD, MPH; Stephen M. Shortell, PhD, MPH, MBA; Patricia P. Ramsay, MPH; Kennon R. Copeland, PhD; and Lawrence P. Casalino, MD, PhD
Organizational Structure for Chronic Heart Failure and Chronic Obstructive Pulmonary Disease
Seppo T. Rinne, MD, PhD; Chuan-Fen Liu, PhD; Edwin S. Wong, PhD; Paul L. Hebert, PhD; Paul Heidenreich, MD; Lori A. Bastian, MD; and David H. Au, MD
Value of Primary Care Diabetes Management: Long-Term Cost Impacts
Daniel D. Maeng, PhD; Xiaowei Yan, PhD; Thomas R. Graf, MD; and Glenn D. Steele, Jr, MD, PhD
Currently Reading
The Budget Impact of Cervical Cancer Screening Using HPV Primary Screening
Thomas Wright, MD; Joice Huang, PharmD, MBA; Edward Baker, MD; Susan Garfield, DrPH; Deanna Hertz, MHEcon; and J. Thomas Cox, MD
LDL Cholesterol Response and Statin Adherence Among High-Risk Patients Initiating Treatment
Suma Vupputuri, PhD, MPH; Peter J. Joski, MS; Ryan Kilpatrick, PhD; J. Michael Woolley, PhD; Brandi E. Robinson, MPH; Michael E. Farkouh, MD, MSc; Huifeng Yun, PhD; Monika M. Safford, MD; and Paul Muntner, PhD

The Budget Impact of Cervical Cancer Screening Using HPV Primary Screening

Thomas Wright, MD; Joice Huang, PharmD, MBA; Edward Baker, MD; Susan Garfield, DrPH; Deanna Hertz, MHEcon; and J. Thomas Cox, MD
This study assesses the clinical and economic implications from a payer perspective of human papillomavirus genotyping for cervical cancer screening in comparison with existing practices.


Objectives: This study assessed the clinical and budgetary impacts of human papillomavirus (HPV) primary screening with HPV16/18 genotyping, in contrast to current cervical cancer screening strategies.

Study Design: A decision-tree framework and Markov model were used to model clinical and cost implications of screening and diagnosis of disease.

Methods: A model was developed to compare the annual clinical and budgetary impact of HPV screening with genotyping versus cytology, and co-testing with and without genotyping. Epidemiology and test performance inputs are from the literature and the Addressing THE Need for Advanced HPV Diagnostics (ATHENA) trial. Costs are from a US payer perspective. Clinical impact was measured as the resulting incidence of cervical cancer, and budget impact is reported as annual cost per screened woman. The model considered the impact of patient noncompliance (loss to follow-up) at both the initial screen and re-test.

Results: Cytology was found to be inferior to both co-testing and HPV primary screening. Co-testing was inferior to co-testing with genotyping. Co-testing with genotyping every 3 years (incidence = 5.5 per 100,000 women; annual investment = $61) or 5 years (incidence = 7.4 per 100,000 women; annual investment = $37) was slightly more effective, but more costly than HPV primary screening every 3 years (incidence = 6.2 per 100,000 women; annual investment = $48) or 5 years (incidence = 8.1 per 100,000 women; annual investment = $30). Genotyping strategies were relatively stable to the effects of patient noncompliance.

Conclusions: Primary HPV screening with genotyping represents a sensible combination of clinical effectiveness and costs, while reducing the risks associated with patient noncompliance.

Am J Manag Care. 2016;22(3):e95-e105

Take-Away Points
Human papillomavirus (HPV) screening with genotyping represents a sensible combination of clinical effectiveness and costs.
  • Recent FDA approval and an interim clinical guidance have resulted in HPV testing as an option for primary screening of cervical cancer.
  • HPV screening with genotyping every 3 years leads to a lower incidence of cervical cancer than either of the 2 current guideline-recommended strategies—cytology every 3 years or co-testing every 5 years—with 6.2 of cervical cancer cases per 100,000 women versus 11.7 and 7.4, respectively. There is also lower cost per disease detected ($32,123 vs $36,876 and $36,196, respectively).
  • Incorporating genotyping into screening is especially important as the screening interval increases or when patient compliance is a concern. 
In 2014, a human papillomavirus (HPV) test that detects high-risk types and individual genotypes HPV 16 and 18 utilizing amplification of target DNA (the cobas HPV Test) was approved by the FDA for primary screening in cervical cancer. HPV types 16 and 18 have been found to cause more than 70% of cervical cancers1; women who are positive for HPV 16 and/or 18 are at an increased risk of high-grade cervical intraepithelial neoplasia (CIN), even if they have normal cytology.2,3 CIN is a dysplastic change beginning at the squamocolumnar junction in the uterine cervix that may be a precursor of cervical cancer: grade 1 (CIN1), mild dysplasia involving the lower one-third or less of the epithelial thickness; grade 2 (CIN2), moderate dysplasia with one-third to two-thirds involvement; grade 3 (CIN3), severe dysplasia or carcinoma in situ, with two-thirds to full-thickness involvement. Targeting detection of these high-risk HPV types allows clinicians to properly manage patients at highest risk for developing cervical cancer.

In 2015, a panel represented by multiple societies issued new interim guidance recommending HPV primary screening as an alternative to current cytology-based screening strategies.4 This provides clinicians and patients with another option for routine screening—options which now include cytology alone, cytology in conjunction with HPV testing (co-testing) with or without genotyping, or HPV primary screening with genotyping.5,6 Likewise, payers now have the opportunity to consider an expanded range of screening options.

This study was undertaken to estimate, from a US payer perspective, the near-term clinical and budgetary impacts of adopting HPV primary screening with HPV 16/18 genotyping compared with current cervical cancer screening strategies derived from established clinical guidelines.


A decision-tree framework was used to model the screening and diagnosis of disease ≥CIN2; a Markov transition model was constructed to simulate the natural history of HPV, CIN, and cervical cancer. Women enter the decision tree with the probability of initial disease representative of a US cervical cancer-screened population of individuals 30 years or older (mean age = 45 years).

The model compares the screening strategies currently recommended by the United States Preventive Services Task Force (USPSTF)/American Cancer Society (ACS) for women aged 30 to 65 years with strategies that incorporate HPV screening with genotyping to identify high-risk strains 16 and 18, resulting in comparison of 7 screening strategies in total. The screening strategies include: 1) cytology every 3 years, 2) co-testing every 3 years, 3) co-testing every 5 years, 4) co-testing with genotyping every 3 years, 5) co-testing with genotyping every 5 years, 6) HPV primary screening with genotyping every 3 years, and 7) HPV primary screening with genotyping every 5 years.5,6 The decision-tree diagrams are represented in Figure 1. A diagnosis of ≥CIN2 incurs treatment cost and exits the model.

Screening Algorithms

The screening algorithms are described as follows:

Cytology every 3 years. Cytology is the primary screening method. Women with indeterminate cytology results—referred to as atypical squamous cells of undetermined significance (ASC-US)—are triaged using HPV testing. A positive HPV result or cytology worse than ASC-US leads to colposcopy. Women with negative results return for routine cervical cancer screening in 3 years (see Figure 1A).

Co-testing every 3 or 5 years. The USPSTF/ACS recommend a second screening strategy of co-testing with cytology and HPV, which allows extension of the screening interval from 3 to 5 years for women negative on both tests. Colposcopy is indicated in women with cytology results of ASC-US/HPV positive, or cytology worse than ASC-US, regardless of HPV result. Women with normal cytology but who are HPV positive return for follow-up co-testing in 12 months. Although 5-year screening intervals are recommended for women negative on both tests, in practice, a 3-year interval is frequently used. Both intervals were modeled (see Figure 1B).

Co-testing with genotyping every 3 or 5 years. Another option for women with co-testing results of normal cytology but who are HPV positive is to genotype for HPV 16/18. Women testing positive for 16/18 are sent to colposcopy, whereas women positive for HPV but negative for 16/18 repeat co-testing in 12 months. All other co-testing results are managed the same way as for co-testing without genotyping (see Figure 1B).

HPV primary screening every 3 or 5 years. This strategy utilizes HPV with genotyping as the primary screening modality. Women who are HPV negative return for routine screening in 3 or 5 years. Women who are HPV 16/18 positive are referred for immediate colposcopy. HPV positive women who are HPV 16/18 negative have cytology performed on the residual sample. A cytology result of ASC-US or worse leads to immediate colposcopy, whereas normal results from cytology return women for follow-up testing in 12 months (see Figure 1C).

Model Structure

Consistent with published US rates, the model assumes a 75% probability of compliance with follow-up testing and routine screening intervals.7,8 Similarly, patients lost to follow-up at the time of re-test are assumed to have a 75% probability of returning to routine screening at the next interval. In the interim, patients with HPV infection/CIN may persist, progress, or regress from one stage to another.

The progression and regression of HPV and CIN were modeled using a Markov state transition model with a 1-month cycle, which captures the probability of a screened population of individuals 30 years or older, transitioning to a more or less advanced stage of CIN or HPV infection. Women enter the Markov model following results of the initial screen in 1 of the following 8 health states: well and HPV negative, non-16/18 HPV positive, 16/18 HPV positive, CIN1, CIN2, CIN3, invasive cervical cancer (ICC), or death.

Figure 2 shows the graphical representation of the Markov model. We assume only CIN3 may directly progress to ICC. Patients face a probability of death from ICC; however, death from other causes is not considered.

The model was used to assess the impact of the screening strategies over 2 screening cycles (2x interval length). The results of the model were then annualized to arrive at a 1-year time horizon that reports the expected annual incidence of cervical cancer and average annual cost of screening women 30 years or older (see eAppendix [eAppendices available at] for calculation). Annual outcomes were reported in order to normalize results across screening strategies with different interval lengths and to present the data on a basis that is easier for payers to compare. The model uses probabilities instead of a cohort approach to allow each payer to assess the impact on their population by multiplying the annual per-screened-woman outcomes by their relevant member population. The costs are reported annually and are assumed to be applicable in the short term (6-10 years) as the basis of the calculation is 2 screening cycles. The results assume that as long as the national population of screened women 30 and older are representative of a health plan’s population, the entry/exit of individual members should not impact the overall results, allowing the results to be representative of individual health plans.


Epidemiological and test performance inputs were taken from the Addressing THE Need for Advanced HPV Diagnostics (ATHENA) trial and are based on women 30 years or older (mean age = 44.7 ± 10.1 years). The ATHENA trial has been described elsewhere.9-11 Briefly, as a prospective cohort study which enrolled 47,000 women undergoing cervical cancer screening in the United States, it is the largest cervical cancer screening registrational trial to evaluate HPV testing.

Data used for the natural history of cervical cancer were taken from US and international studies. Clinical inputs are shown in Table 1.9-32 Where multiple sources existed, inputs were based on a weighted average, with results from studies with larger populations weighted more heavily than studies with smaller populations.

Costs include all screening costs in addition to costs for the diagnosis and treatment of CIN and ICC. Costs for screening, diagnosis, and treatment of CIN are taken from the US Medicare fee schedule.33 Cost for HPV testing was based on the cobas HPV Test, which includes simultaneous testing for strains 16/18, and therefore, no additional cost was assumed for genotyping. Direct costs for treating ICC were taken from published US studies and assume the average cost of treatment and follow-up across all stages of cervical cancer.34,35 (Cost inputs are available in eAppendix 2 [Table]). All costs were adjusted to 2014 US dollars.

A 1-way sensitivity analysis and probabilistic sensitivity analysis (PSA) were undertaken to assess the impact of parameter uncertainty on modeled results. Clinical inputs were varied across the ranges reported in the literature and assumed a beta distribution, while costs were varied by ±50% and assumed a gamma distribution. The correlation between sensitivity and specificity was controlled using the diagnostic odds ratio.36 The ranges used are shown in Table 1.9-32 The PSA followed a standard Monte Carlo approach based on 5000 randomly generated simulations of parameter values.

When assessing the costs and effectiveness of each strategy relative to alternatives, screening with cytology alone results in an increase in the incidence of cancer and higher mortality due to missed cancers than any other strategy, and at a cost higher than that of strategies incorporating a 5-year interval. We can thus consider cytology to be inferior to alternatives with 5-year screening intervals since it is both less effective and more expensive.

Of the remaining strategies, co-testing every 3 or 5 years without genotyping has similar costs as co-testing every 3 or 5 years with genotyping, but results in more cancer. Consequently, we can consider the co-testing with genotyping strategies to be superior to co-testing without genotyping. Thus, the strategies that utilize genotyping represent a desired combination of improving screening effectiveness while reducing cost. For instance, HPV primary screening at 5 years, when compared with the current guideline-recommended strategies of: 1) primary cytology every 3 years; and 2) co-testing without genotyping every 5 years, leads to reduced cervical cancer incidence and 27% and 19% reductions in cost, respectively.

Of all strategies modeled, the one that incorporates co-testing with genotyping and HPV primary screening at 3-year intervals results in the lowest annual incidence of cervical cancer (5.5 and 6.2 per 100,000 women, respectively). However, such strategies may require an increase in overall financial investment.

Copyright AJMC 2006-2019 Clinical Care Targeted Communications Group, LLC. All Rights Reserved.
Welcome the the new and improved, the premier managed market network. Tell us about yourself so that we can serve you better.
Sign Up