NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Vesco KK, Whitlock EP, Eder M, et al. Screening for Cervical Cancer: A Systematic Evidence Review for the U.S. Preventive Services Task Force [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2011 May. (Evidence Syntheses, No. 86.)

Cover of Screening for Cervical Cancer

Screening for Cervical Cancer: A Systematic Evidence Review for the U.S. Preventive Services Task Force [Internet].

Show details


Key Questions and Analytic Framework

Using the USPSTF’s methods (detailed in Appendix B),95 we developed an analytic framework (Figure 5) and five key questions (KQs) to guide our literature search. These KQs include:

Figure 5 is an analytic framework for the key questions of this report. It depicts the events that females may experience while undergoing screening and treatment for cervical cancer. In general, the figure illustrates how screening for cervical cancer may detect abnormal results and lead to the early detection of CIN1-3 and invasive cancer. Subsequent treatment may improve the length and/or quality of life. The key questions address the time to begin screening for cervical cancer and the benefits and harms of cervical cancer screening.

Figure 5

Analytic Framework and Key Questions. Key Questions KQ1: When should cervical cancer screening begin, and does this vary by screening technology or by age, sexual history, or other patient characteristics?


When should cervical cancer screening begin, and does this vary by screening technology or by age, sexual history, or other patient characteristics?


To what extent does liquid-based cytology improve sensitivity, specificity, and diagnostic yield and reduce indeterminate results and inadequate samples compared to conventional cervical cytology?


What are the benefits of using HPV testing as a screening test, either alone or in combination with cytology, compared with not testing for HPV?


What are the harms of liquid-based cytology?


What are the harms of using HPV testing as a screening test, either alone or in combination with cytology?

The scope of the review was set in early 2007 and was conducted in the intervening years without a change in scope. This report’s scope differs from the 2002 USPSTF evidence report in several ways. KQ1, which was not included in the 2002 evidence report, addresses when cervical cancer screening should begin. Both LBC and automated screening technologies were evaluated in the prior review, and the evidence was determined to be insufficient to recommend for or against the use of these technologies in cervical cancer screening programs. For this review, we updated the evidence regarding LBC (KQ2) and focused on studies that evaluated either ThinPrep or SurePath, which are both FDA approved. However, we did not update the direct evidence for automated screening technologies because they are less relevant to primary care clinicians. These technologies are implemented by laboratories and not performed by the clinician at the time of cervical cytology collection. The previous review evaluated the sensitivity and specificity of the HPV test for detection of histologically proven HSIL and LSIL. The authors also evaluated the use of the HPV test as a tool to facilitate triage of women with abnormal cytology. The current review expanded the scope of KQ3 to evaluate the evidence regarding the use of HPV testing in the following scenarios:

  1. Primary screening with HPV test alone.
  2. HPV testing with cytology triage of positive HPV (reflex cytology).
  3. Combination HPV and cytology testing (co-testing).
  4. Cytology testing with HPV triage of positive cytology (reflex HPV).

We restricted the scope of KQ3 to include only HC2 or PCR methods for HPV testing. We did not evaluate the use of HPV testing for followup after treatment for CIN. KQs 4 and 5, neither of which was framed as a separate KQ in the prior review, address the harms of LBC and HPV testing.

We addressed one contextual question that evaluated the efficacy of screening in women older than age 65 years according to the USPSTF’s specified nonsystematic approach.96 The previous review addressed this question systematically, and the USPSTF recommended against routinely screening women older than age 65 years, based on limited evidence regarding the benefits of continued screening in these women. We did not update the direct evidence for screening in women after a hysterectomy because the prior USPSTF recommendation to discontinue screening after hysterectomy for benign disease is clearly supported. Because the HPV vaccine is so new, data to determine the long-term efficacy of the vaccine or how the HPV vaccine will affect screening is limited. Therefore, the USPSTF did not include a KQ addressing the impact of the HPV vaccine on cervical cancer screening. This will be an important topic for future evidence reviews, when more data regarding this issue become available. The USPSTF judged that a thorough review of cost effectiveness analyses was beyond the scope of our review. We did not review evidence on appropriate screening intervals, as this issue is rarely studied directly and the concurrent modeling study addresses this topic. We did not systematically review the harms of treatment procedures such as LEEP, cryotherapy, and laser cone biopsy.

Literature Search Strategy

For all KQs, we searched for systematic reviews, meta-analyses, and evidence-based guidelines on cervical cancer screening in the Database of Abstracts of Reviews of Effects, the Cochrane Database of Systematic Reviews, PubMed, and the Health Technology Assessment database from 2000 through January 2007. We also conducted a series of searches for each KQ and reviewed the search results for applicability to all KQs. For KQs 1, 3, 4, and 5, we conducted searches to identify studies published since the previous USPSTF review (2000 through September 2010). We searched in MEDLINE and the Cochrane Collaboration Registry of Clinical Trials (CCRCT) without restrictions on study designs. For KQ5, we also searched PsycINFO to capture adverse psychological effects of HPV testing. The search period for LBC (KQ2) began in 2003 because two systematic reviews provided a complete and thorough search of the relevant literature through July 2003.97,98 We used these reviews as source documents to locate relevant studies from before 2003 and bridged their searches for LBC using MEDLINE and CCRCT, without restrictions on study designs, from the beginning of 2003 through September 2010. We evaluated the studies included in the previous review by Hartmann and colleagues99 against the inclusion and exclusion criteria for the current review, and found only one study was eligible for inclusion.100 We also obtained articles from outside experts and bibliographies of other relevant articles and systematic reviews. In addition to these searches for published trials, we searched federal agency trial registries for unpublished trials of cervical cancer screening. All searches were limited to articles published in the English language.

Study Selection

While differences in inclusion, exclusion, and quality criteria precluded us from incorporating any of the existing systematic reviews or meta-analyses that were identified, the high-quality reviews and meta-analyses were used to check the completeness of our searches for primary studies.

For KQ1, in the absence of RCTs addressing when to begin screening, we included cohort studies that evaluated the incidence and prevalence of cervical cancer in young screened populations, natural history studies of CIN and HPV infection in young women, and studies reporting outcomes of population-based screening programs targeting young women.

For KQs 2 and 3, evaluating LBC and HPV testing, we included studies that provided evidence regarding absolute and relative test performance. Our specific criteria were as follows:

  1. To determine absolute test performance, we required that the reference standard of colposcopy and/or biopsy was systematically applied to all those screening positive and at least a random sample of screen negatives, with valid adjustment for verification bias when necessary. The reference standard must have been independent of the screening test (i.e., the screening test results were not used to establish the final diagnosis).
  2. If a study did not test negatives appropriately with the gold standard, we could not use their absolute test performance estimates. However, if the study was an RCT, compared test performance within the randomization scheme, and was of appropriate quality, then we included relative test performance measures.
  3. To evaluate screening demands and potential harms, we abstracted the following data where available as absolute or relative measures: test positivity, colposcopy referrals, colposcopy compliance, positive predictive value (PPV), false-positive proportion, and appropriately calculated specificity. For trials with multiple screening rounds, we looked for round-specific and cumulative data. For all studies, we looked for age-specific data.
  4. Many studies reported theoretical test performance by estimating results for different screening and management programs, rather than by what was actually done in the trials. We determined these calculations could not be included if the assumptions required to estimate performance introduced potential threats to validity. We usually could not determine how to fairly assess whether these assumptions affected the validity of the calculated test performance, and if they did, what direction or degree of bias was introduced.

Studies of LBC and HPV primary screening must have been conducted in routine screening populations. Other inclusion and exclusion criteria specific to each question are detailed in Appendix B.

Data Extraction and Quality Assessment

Two investigators independently reviewed all abstracts for all KQs. Two investigators evaluated articles against a set of inclusion/exclusion criteria. Each investigator independently reviewed articles for quality using design-specific quality criteria based on the USPSTF methods, supplemented by the National Institute of Health and Clinical Excellence criteria for quality of systematic reviews and the QUADAS tool for quality assessment of diagnostic accuracy studies (Appendix B, Table 3).95,101,102 Two investigators critically appraised all studies and agreed when articles were excluded for quality reasons. One reviewer abstracted data from included studies into evidence tables, and a second reviewer checked the data.

Data Synthesis and Analysis

Except for cytology testing with HPV triage of positive cytology (KQ3), data synthesis for all questions was qualitative because heterogeneity in the samples, settings, study designs, and instruments did not allow for quantitative synthesis. In the results section, studies are summarized qualitatively within the KQs. For KQ3 addressing HPV testing, studies are categorized by four different uses of HPV testing in cervical cancer screening. For each question, we first describe RCTs comparing cytology with HPV-enhanced screening strategies within existing screening programs that report absolute and relative CIN2+/CIN3+ detection for one or two screening rounds, followed by cross-sectional studies reporting absolute test performance data. Studies from countries with less developed cervical cancer screening programs are discussed separately due to their lower applicability to the U.S. population.

For evidence on the benefits of using HPV testing to triage women with ASC-US cytology, we estimated the combined difference in sensitivity and specificity between HPV and repeat CC. A random effects model was used to incorporate variation among studies. For the difference in sensitivity and specificity between HC2 and cytology, we used risk difference as the effect measure. Statistical heterogeneity was assessed by Cochran’s Q test and the I2 statistic.103 All analyses were performed using Stata 10.0 (StataCorp, College Station, TX).

Many of the results reported in the evidence and summary tables are calculated from data provided in the articles using methods cited in Appendix B. Such calculations are indicated in the evidence tables by “(calc).” In the RCTs, results were generally reported using women screened (instead of women randomized, as in an “intention-to-screen” analysis, which also includes opportunistic screening results) within each arm and each round. To be consistent, we abstracted from the articles or calculated results using the number of women screened within each randomized arm as the denominator unless noted as otherwise in the tables. Consideration of program results among women screened only may be less appropriate to determine overall population impact, but acceptable when primarily evaluating the relative merits (including false positives and other adverse effects) of efficacious screening alternatives.

Evidence tables for all KQs are in Appendix C. Detailed methods can be found in Appendix B.

USPSTF Involvement

This research was funded by AHRQ under a contract to support the work of the USPSTF. The authors worked with eight USPSTF liaisons at key points throughout the review process to develop and refine the scope, analytic framework, and KQs; to resolve issues around the review process; and to finalize the evidence synthesis. AHRQ had no role in study selection, quality assessment, or synthesis, although AHRQ staff provided project oversight, reviewed the draft evidence synthesis, and distributed the initial evidence report for external review of content by outside experts, including representatives of professional societies and federal agencies. The final published systematic evidence review was revised based on comments from these external reviewers.

PubReader format: click here to try


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...