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

Kulasingam SL, Havrilesky L, Ghebre R, et al. Screening for Cervical Cancer: A Decision Analysis for the U.S. Preventive Services Task Force [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2011 May. (Evidence Syntheses, No. 86s.)

Cover of Screening for Cervical Cancer: A Decision Analysis for the U.S. Preventive Services Task Force

Screening for Cervical Cancer: A Decision Analysis for the U.S. Preventive Services Task Force [Internet].

Show details

Methods

An overview of the model, including inputs, is provided in Appendix B. The following provides a summary of the main changes made to the model to address the specific aims and sub-aims.

Natural History

Recent evidence suggests that the natural history of HPV in young women (aged <30 years) may be such that establishment of a high-grade CIN lesion occurs early in the course (within 2 years) of a high-risk HPV infection.22-24 Studies also suggest that the burden of CIN may be higher than previously thought in young women, but that progression to cancer from high-grade CIN is low—approximately 1 percent per year.25-26 In the original model, only a small percentage (5 percent) of infections were assumed to directly result in CIN2-3. Approximately 4 percent of women were assumed to progress from CIN to cancer each year. To address this, a revised natural history model was developed (details presented in Appendix B). This model, which was used in sensitivity analyses, incorporates estimates of HPV and CIN incidence and regression that are higher than those used in the original model, but also includes lower rates of progression between CIN states and from CIN2-3 to cancer.

CIN2 or 3 Versus CIN2-3

There is evidence to suggest that CIN2 behaves similarly to CIN1 (i.e., a high proportion regress), especially in young women.27-28 It is also a much less reproducible histologic result than CIN3.29-30 The current model retains CIN2-3 as a single disease state instead of two separate states based on clinical guidelines that treat these outcomes in a similar manner.31 However, in order to address the possibility that CIN2 may be a false-positive result (that can lead to overdiagnosis and treatment), especially in young women (aged <30 years), a sensitivity analysis is conducted in which CIN2-3 is further stratified into CIN2 and CIN3 to estimate the percentage of CIN2-3 outcomes that are CIN2 for those strategies that are identified as “efficient” in the base-case analysis. For this sensitivity analysis, the percentage of CIN2-3 that is CIN2 is approximated by age based on data from a study of women undergoing screening in Kaiser Permanente Northwest by Insinga et al.32 These estimates are presented in Table 1. The number of CIN2-3 cases (per 1,000 women) is estimated using the model; the percentage of women younger than age 30 years whose disease is categorized as CIN2 is then calculated by multiplying the estimates of the number of CIN2-3 cases by the age-specific percentage in Table 1.

Table 1. Percentage of CIN2 Diagnoses By Age Based on Incidence of CIN2 and CIN3 per 1,000 Women Younger Than Age 30 Years.

Table 1

Percentage of CIN2 Diagnoses By Age Based on Incidence of CIN2 and CIN3 per 1,000 Women Younger Than Age 30 Years.

Screening

Four strategies (based on discussions and agreement with the USPSTF) are examined in this report. The first three strategies, recommended in recent guidelines, are as follows:

1.

Cytology, with a repeat cytology test for results of ASC-US. For this strategy, all women are screened with cytology. Women with a cytology result of atypical squamous cells–high grade (ASC-H), low-grade squamous intraepithelial lesion (LSIL), or high-grade squamous intraepithelial lesion (HSIL) are referred for followup and treatment based on American Society for Colposcopy and Cervical Pathology (ASCCP) guidelines.33 Women with an ASC-US cytology result are assumed to receive a repeat cytology test at 6 and 12 months, consistent with ASCCP guidelines.33 Women with normal cytology results are assumed to return to routine screening conducted every 1, 2, 3, or 5 years. This strategy is used to estimate outcomes for Specific Aim 1, Sub-Aim 1, and Specific Aim 2.

2.

Cytology, with HPV DNA testing for cytology results of ASC-US. For this strategy, all women are screened with cytology. Women with a cytology result of ASC-H, LSIL, or HSIL are referred for followup and treatment based on ASCCP guidelines.33 Women with an ASC-US cytology result are assumed to undergo HPV testing, with triage to colposcopy if HPV positive or repeat testing at 1 year if HPV negative. Women with normal cytology results are assumed to return to routine screening conducted every 1, 2, 3, or 5 years. This strategy is used to determine outcomes for Specific Aim 2.

3.

Cytology and HPV. This strategy is recommended for women aged 30 years and older.34 For this strategy, women are screened with both HC2 and cytology. Women with a cytology result of ASC-H, LSIL, or higher are referred to colposcopy. Women with a cytology result of ASC-US undergo HPV DNA testing using HC2, with triage to colposcopy if HPV positive or repeat testing at 1 year if HPV negative. Women with a normal cytology result who have a positive HPV test result are assumed to undergo repeat testing with both tests 1 year later, with triage to colposcopy if the cytology result shows ASC-US or higher or if the HPV test is positive. Women with a normal cytology result who have a negative HPV test result are assumed to be screened with both HPV and cytology every 3 years, as per ASCCP guidelines.33 Prior to age 30 years, women are assumed to be screened with cytology only, with a repeat cytology test for ASC-US results. Women with normal cytology results are assumed to return to routine screening conducted every 1, 2, 3, or 5 years.

The fourth strategy, which is not part of current recommendations, is as follows:

4.

HPV followed by cytology if HPV positive. This strategy is included in a sensitivity analysis for Specific Aim 2. For this strategy, women aged 30 years or older are assumed to be screened with an HPV DNA test. Women who have a positive HPV test result undergo cytology testing. Women with ASC-US or a more severe cytology result are assumed to be referred to colposcopy. Women with a normal cytology result who are HPV positive are assumed to return in 1 year for repeat testing with cytology and HPV. Women who have an HPV negative test result are assumed to return to routine screening conducted every 1, 2, 3, or 5 years. Prior to age 30 years, women are assumed to be screened with cytology only, with a repeat test for ASC-US results. Women with normal cytology results are assumed to return to routine screening conducted every 1, 2, 3, or 5 years.

For Specific Aim 1 and Sub-Aim 1, the age at which to begin screening is varied. For Specific Aim 2, the age at which to begin screening is fixed at age 21 years. Except for Sub-Aim 1 (in which age to end screening is varied), all women are assumed to be screened until age 85 years. Women are assumed to enter the model at age 12 years and to be followed until age 100 years or death.

Screening Test Characteristics

Estimates of sensitivity and specificity for the different tests (HC2 and cytology) are based on estimates provided by the Oregon EPC in a separate evidence report.35 Details of the studies used to derive the estimates are presented in Appendix C. For the purpose of this report, we do not distinguish between liquid-based cytology (LBC) and conventional cytology (CC). We use the term “cytology” to refer to both LBC and CC; this decision is based on recent data showing no significant difference between the two tests in terms of sensitivity and specificity.35 For Specific Aim 2, three different sets of estimates for HC2 and cytology test accuracy are used, including estimates from a large randomized controlled trial of HPV testing and cytology by Mayrand et al,8 as well as summary estimates for sensitivity and specificity for cytology and HC2 from a meta-analysis by Koliopoulos et al.36 The third set of estimates is based on the predicted relative increase in sensitivity and decrease in specificity based on differences in cytology and HPV test performance reported by Mayrand et al8 and Bigras et al.37 The differences (delta) in sensitivity and specificity (for CIN2+ compared to cytology) are also presented for the HPV test accuracy estimates.

Table 2. Sensitivity and Specificity of Cytology and HPV Testing for Primary Screening and Triage of Abnormal Cytology Results.

Table 2

Sensitivity and Specificity of Cytology and HPV Testing for Primary Screening and Triage of Abnormal Cytology Results.

Colposcopy and Biopsy Sensitivity and Specificity

Colposcopy and biopsy are assumed to be perfectly sensitive and specific in the base case, to allow for comparison with previous cost-effectiveness analyses. The impact of less than perfect sensitivity and specificity is explored in sensitivity analyses using estimates of the performance of colposcopy (with biopsy as the gold standard) from Mitchell et al.48

Table 3. Sensitivity and Specificity of Colposcopy and Biopsy.

Table 3

Sensitivity and Specificity of Colposcopy and Biopsy.

Cytology-Histology Conditional Probabilities

Conditional probabilities of cytology results among women with an abnormal histology result are presented in Table 4. These results are used to determine the percentage of women with an abnormal cytology result who are categorized as ASC, LSIL, HSIL, or cancer. A study comparing cytology-histology correlations for LBC and CC found no significant differences between the two.49 However, to account for findings reported by Ronco et al,50 which show a higher percentage of women with ASC-US among those screened with LBC, compared to those screened with CC, a sensitivity analysis is conducted using estimates of conditional probabilities of LBC given histology from a screening study in Seattle, Washington (Akhila Balasubramanian, personal communication).51 In the absence of data for determining the conditional probabilities of LBC for detecting cancer, a distribution similar to that of CC is assumed, and the ratio of HSIL to cancer cytology given a certain histology result is assumed to be the same as that reported for CC.

Table 4. Conditional Probabilities of Cytology Results for a Given Histology Result.

Table 4

Conditional Probabilities of Cytology Results for a Given Histology Result.

Conditional probabilities for ASC-H. Among women with ASC cytology results, 14 percent of those with underlying CIN1 and 72 percent of those with underlying CIN2-3 or cancer are categorized as ASC-H.52 In the absence of data on the percentage of ASC classified as ASC-H with underlying normal histology, we assume a similar percentage to those with underlying CIN1 (14 percent).52

Followup for Abnormal Screening Test Results

Followup for abnormal screening test results and abnormal histology is based on recently published ASCCP guidelines.44 Women younger than age 21 years are treated according to guidelines for adolescent women who have abnormal screening test or histology results.

Adherence to Screening, Followup, and Treatment

Adherence to screening, followup, and treatment is assumed to be 100 percent for the base case. Age-specific estimates of screening used in sensitivity analyses are based on a study of rates of screening in a population of women at Kaiser Permanente Northwest by Schabert et al.53 These estimates should be treated with caution, however, since they may overrepresent screening estimates by age, due to the inclusion of more than one (multiple) screening test from some women. Although this study is from a health maintenance organization population, it was used to provide information on screening adherence because it provides data based on chart review rather than self-report. A concern with self-reported screening is that studies have shown that only 65 to 70 percent of self-reports of cervical cancer screening within 3 years can be validated through subsequent chart review.54-57 An additional sensitivity analysis for Specific Aim 1, in which age to begin screening is varied, is conducted using self-reported screening from the National Survey of Family Growth, since this survey provides information on self-reported screening in 1 year intervals and includes data on young women (aged <21 years).58

Table 5. Screening Adherence Estimates.

Table 5

Screening Adherence Estimates.

Table 6. Screening Adherence Estimates From the National Survey of Family Growth.

Table 6

Screening Adherence Estimates From the National Survey of Family Growth.

Analytic Approach

Base-Case Analyses

Base-case analyses (using a single set of test accuracy estimates) are conducted for Specific Aim 1. The analyses for Specific Aim 2 are conducted as preliminary analyses, with three different sets of test accuracy estimates to reflect a range in the published literature as well as a lack of meta-analytic results from the accompanying EPC report.35 For each question, outcomes presented include (per 1,000 women): expected false-positive test results (defined as a positive screening test but normal colposcopy-biopsy result), colposcopies performed, CIN2-3 lesions detected, cervical cancer cases, and cervical cancer deaths. If a result is less than 1 per 1,000, the outcome is also presented using a denominator of 100,000. The main outcome is colposcopies per (undiscounted) life-year gained. This outcome was requested by the USPSTF, which bases its recommendations on the trade-off between clinical benefits and harms. A previous decision analysis conducted for the Task Force on screening for colorectal cancer used colonoscopies per life-year gained as the primary outcome.59 Colposcopy, the current standard for definitive diagnosis after an abnormal cervical cancer screening result, was chosen as the closest analogue to colonoscopy.

Strategies are compared using incremental ratios based on the difference in expected number of colposcopies, divided by the difference in life expectancy. Strategies that are associated with more colposcopies and less effectiveness or fewer colposcopies but a higher colposcopy per life-year ratio than an adjacent strategy are considered to be dominated. The remaining strategies (after this elimination process) lie on an “efficiency” frontier. It should be noted, however, that the use of the term “efficient” is non-traditional, since this is not a cost-effectiveness analysis. Efficiency in this case refers to a strategy that represents a potentially reasonable trade-off between the burden and benefits of screening. Strategies that fall on the steepest part of the efficiency frontier are noted, since these are considered to be the most efficient. However, it is important to note that there is no formal definition for what constitutes a “high burden” using colposcopies per life-year gained. As a result, the incremental colposcopies per life-year associated with strategies identified as efficient vary by question.

For the purpose of interpreting the results, the current USPSTF recommendations regarding ages at which to begin and end screening are used. The USPSTF recommends that women begin screening within 3 years of onset of sexual activity or at age 21 years (whichever comes first), and that screening be conducted at least every 3 years.5 In terms of the age at which to end screening, the USPSTF recommendation states that screening should not be routinely conducted in women aged 65 years or older who have a history of normal cytology tests and are not otherwise at high risk for cervical cancer. As such, for Specific Aim 1, the strategies are also compared and contrasted with a strategy based on the current USPSTF recommendations. For Sub-Aim 1, in which it is assumed that women have been screened every 3 years since age 21 years, a baseline strategy that assumes that screening ends at age 65 years is included.

Sensitivity Analyses

For each of the specific aims, an analysis is conducted using a different model of the natural history of HPV and CIN (details provided in Appendix B). In addition, one-way sensitivity analyses are conducted in which screening adherence and the sensitivity and specificity of colposcopy and biopsy are varied. Additional analyses (for Specific Aim 1 and Sub-Aim 1) include using the lowest and highest estimates for sensitivity and specificity of the screening tests and using different conditional probabilities of CIN given an abnormal cytology result to account for potential differences due to the use of LBC. The following sensitivity analyses are also conducted.

Specific Aim 1: Age at which to begin screening

1.

A shorter time horizon, in which all women are followed to age 30 years, is included to determine whether there are patterns in outcomes affected by age at first screening that are obscured if women are followed for a longer timeframe.

2.

Estimates of the percentage of CIN2-3 that is CIN2 are presented for strategies identified as efficient using colposcopies per life-year.

3.

One concern with the use of colposcopies per life-year for the analysis of age at which to begin screening is that this measure may underestimate the burden of screening in this population, since ASCCP guidelines33 now allow for adolescent women younger than age 21 years to be rescreened if they have an abnormal cytology test result. To address this, a sensitivity analysis is conducted that estimates the number of screening cytology tests per life-year.

Specific Aim 1–Sub-Aim 1: Age at which to end screening

4.

Studies show that older women are at an increased risk of dying from cervical cancer even after adjusting for cancer stage.60 To address this, a sensitivity analysis is conducted using age- and stage-specific hazard ratios based on data from Ries et al for women aged 50 to 69 years and 70 years and older. These analyses should be interpreted with caution, however, since the estimates were not provided in detail to determine conditional 5-year survival probabilities. As such, the same ratio is applied to each year for 5 years, and may overestimate the impact of age on stage-specific survival.

Specific Aim 2: Role of HPV DNA testing

5.

Since the strategy of HPV testing followed by cytology (described previously) is not a currently recommended strategy, this strategy is examined in a sensitivity analysis.

6.

A sensitivity analysis is conducted in which screening and triage tests are used to quantify burden instead of colposcopies. This analysis is conducted to address the fact that the use of colposcopies may underestimate the burden of screening for HPV-based strategies, since women with discordant (HPV positive, normal cytology) test results undergo repeat testing instead of immediate referral to colposcopy.

7.

A sensitivity analysis is conducted in which women who are screened with HPV and cytology and are dually negative are assumed to be screened every 5 years instead of every 3 years.

Views

  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (1.1M)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...