NCHR Comments on the USPSTF’s Draft Recommendation Statement, Evidence Review, and Modeling Report on Cervical Cancer Screening

National Center for Health Research: October 9, 2017

Thank you for the opportunity to express our views on the U.S. Preventive Services Task Force (USPSTF) draft recommendation statement, evidence review, and modeling report for cervical cancer screening. The National Center for Health Research is a nonprofit think tank that conducts, analyzes, and scrutinizes research, policies, and programs on a range of issues related to health and safety. We do not accept funding from companies that make products that are the subject of our work.

We agree that this is the right time for the U.S. Preventive Services Task Force to re-evaluate its cervical cancer screening recommendations, due to new research studies and decision modeling analysis. In 2012, the USPSTF recommended starting screening at age 21 and stopping at age 65. Women could get a Pap every 3 years, or starting at age 30, women could get both a Pap and HPV test (co-test) every 5 years. Based on the USPSTF’s review, we agree with maintaining most of the 2012 recommendations, but have significant concerns about changes USPSTF has proposed.

We are very concerned about the proposed grade “A” recommendation to exclusively screen women with the HPV test every 5 years starting at age 30. As the Task Force recognized, clinical trial evidence to support primary HPV screening as a preferred screening strategy is inadequate. There is also a lack of well-designed and implemented clinical trials comparing co-testing and primary HPV screening. The pivotal ATHENA trial, the U.S. prospective study supporting the efficacy of primary HPV screening, had significant shortcomings and does not provide clear and substantial evidence that primary HPV testing reduces cancer incidence or mortality. The Task Force appropriately classified this study as “poor quality” and excluded it from the draft evidence review. However, some medical societies are still quoting it.

Due to lack of available evidence, the Task Force relied heavily on mathematical models to arrive at the grade “A” recommendation. While mathematical models that incorporate real world data help to support available evidence, they are hypothetical, and conclusions must still be weighed against evidence from clinical trials and observational studies. This definitely is well below the standard of an “A” recommendation; in fact, we believe screening based exclusively on HPV tests should not be recommended.

Based on the USPSTF’s decision modeling analysis, primary HPV testing potentially confers a slightly higher mortality benefit (approximately 10 life-years per 1000 women) compared to cytology alone, but it requires significantly more colposcopies to be done for each possible cancer averted (766 vs. 39). Moreover, the models demonstrate that primary HPV testing and co-testing prevent approximately the same amount of deaths, but co-testing would require more total tests (19,806 vs. 12,151). However, models are necessarily based on assumptions that are not necessarily accurate.  The conclusions from these models are not based on research conducted on actual patients.

In addition, while the USPSTF depended on the draft modeling report to conclude that primary HPV screening may provide more benefits than harms, other recent modeling analyses by Felix et al., concluded that screening by co-testing may be superior to screening by HPV test alone. Therefore, the totality of the evidence does not support eliminating co-testing as an acceptable strategy.

We therefore conclude that the Task Force’s grade “A” recommendation assigned to primary HPV screening is inappropriate, because it is primarily based on modeling analysis and not clinical trials or real-world data. This grade “A” would certainly mislead physicians and women to believe that there is incontrovertible evidence to support an HPV testing-only screening strategy rather than co-testing or Pap smear only testing. Since the recommendation is that women can select a Pap smear every 3 years or the HPV test every 5 years, many women will likely choose the less frequent screening option, not realizing that they will be at increased risk of needing additional, more invasive testing with the HPV test only screening.

The seven cited randomized clinical trials are difficult to interpret. First, the studies cannot be assumed to be representative of women in the U.S. since they were conducted primarily in European countries with national screening programs and more standardized triage strategies. Second, the studies were extremely heterogeneous in terms of test technology, screening intervals, and follow-up protocols. Third, the study results were mixed. For instance, the NTCC phase II trial found that hrHPV test alone had a higher CIN3 detection rate only in the first screening round, and actually had a lower rate of detection in the second round. The Task Force’s review contends that the “cumulative rate was still double.” However, it is possible  that detection rates of CIN3 declined in the second screening round because the lesions regressed due to HPV clearance.  Thus, perhaps the colposcopies triggered by HPV positivity in the first screening round were unnecessary. Bottom line: these data are inconclusive. Last, the studies included surrogate endpoints for cancer incidence and mortality, such as CIN2 or CIN3 detection. These endpoints are likely not the most meaningful outcomes for women, since CIN2/3 lesions do not always progress to cervical cancer and sometimes regress instead.  There are several other determinants of disease trajectory that may differ among women in different countries.

While we understand why the USPSTF advises women to “discuss with their provider which testing strategy is best for them” it is obvious that this advice will not necessarily provide excellent guidance for the average patient.  Many providers will not realize that population-based studies and well-conducted clinical trials do not substantially support primary HPV as a preferred testing strategy. Instead, providers are likely to depend on the USPSTF recommendation or those of medical societies, rather than careful scrutiny of the data.  Furthermore, the triage strategy for a positive HPV test is not standardized; it varies from cytology test, HPV genotype test, or direct colposcopy. Such a decision would depend on organized follow-up and would necessitate an informed discussion, which may not be possible in a single office visit. As suggested in the ARTISTIC subsample, a positive HPV test is likely to cause anxiety, and many patients will undergo colposcopy they would not have been needed had they been screened with a Pap smear instead.

In conclusion, we agree with maintaining much of the 2012 recommendations, but we strongly disagree with the proposed “A” rating for HPV testing alone, and urge that the USPSTF reconsider the shortcomings of the evidence and revise their recommendation to give preference to co-testing and Pap smears alone, rather than stand-alone HPV testing.

Thank you for the opportunity to comment on this most important issue which will have an impact on the lives of many adult women on a national level.

For questions or more information, please contact Danielle Shapiro, MD, MPH at


Felix, JC., et. al. The Clinical and Economic Benefits of Co-Testing Versus Primary HPV Testing for Cervical Cancer Screening: A Modeling Analysis. J Women’s Health. 2016; 25(6):606-16

Draft Recommendation Statement: Cervical Cancer: Screening. U.S. Preventive Services Task Force. October 2017.