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

Results

Overview

The results are presented by specific aim. Each section refers to a table that presents the expected number of false-positives (defined as women with abnormal screening results and normal histology), colposcopies, cases of CIN2-3, cases of cancer, and cancer deaths per 1,000 women. The outcomes in these tables are presented with two significant digits so that the outcome per 100,000 women can also be determined. The base-case results for Specific Aim 1 and Sub-Aim 1 are summarized in the accompanying tables and figures. Each table presents only the strategies that are identified as efficient, in the sense that they provide a reasonable trade-off between the burden and benefits of screening. Each figure presents the same strategies using an efficiency curve; the metric used for each curve is colposcopies per (undiscounted) life-year. Three different sets of results are presented for Specific Aim 2. Key sensitivity analyses are presented as outlined previously.

Summary of Results for Specific Aim 1

Tables 7 through 9 and Figures 1 and 2 summarize the results for the base-case analyses for Specific Aim 1, regarding the age at which to begin screening. Tables 7 and 8 present estimates of expected false-positive test results, colposcopies, CIN2-3 cases, cancer cases, and cancer deaths associated with screening in 1-year age increments, beginning at age 15 years and ending at age 25 years. The results are grouped according to screening interval (every 1 year [q1], 2 years [q2], 3 years [q3], and 5 years [q5]). Tables 7 and 8 show results over a short time horizon (cohort is followed until age 30 years) and lifetime horizon (cohort is followed until age 100 years). When comparing by age across the row, increasing age at first screening is generally associated with fewer false-positive test results, fewer colposcopies, and fewer CIN2-3 cases, but more cancer cases and cancer deaths. There are some fluctuations between successive ages due to differences based on screening interval and age at which to end screening. However, some of the fluctuation in the estimates is also due to ASCCP guidelines,33 which allow for repeated screening in women younger than age 21 years with abnormal cytology results. If all adolescent women are assumed to attend colposcopy, the outcomes are more consistent by age (data not shown). The inclusion of this aspect of the guidelines in the decision model may also explain the shape of the curve for false-positive test results presented in Figure 1, in which the largest number of false-positive test results occurs in adolescents younger than age 21 years (range per 1,000, 190 at age 20 years to 232 at age 15 years). This age group also has the lowest number of expected cancer cases (range per 1,000, <1 [or 16 to 22 cancer cases per 100,000]). In contrast, the number of false-positive test results is lower (range per 1,000, 101 at age 25 years to 161 at age 22 years) with each successive year that screening is delayed beyond age 21 years (compared to beginning at age 21 years). However, as shown in Figure 1, the number of expected cancer cases begins to rise (range per 100,000, 31 at age 22 years to 58 at age 25 years) with each successive year that screening is delayed beyond age 21 years.

Table 7. Sensitivity Analysis Showing Expected False-Positives, Colposcopies, CIN2-3 Cases, Cancer Cases, and Cancer Deaths Associated With Screening Beginning at Age 15 Years and Increased in 1-Year Increments to Age 25 Years, Among Women Followed to Age 30 Years*.

Table 7

Sensitivity Analysis Showing Expected False-Positives, Colposcopies, CIN2-3 Cases, Cancer Cases, and Cancer Deaths Associated With Screening Beginning at Age 15 Years and Increased in 1-Year Increments to Age 25 Years, Among Women Followed to Age 30 Years*. (more...)

Table 8. Sensitivity Analysis Showing Expected False-Positives, Colposcopies, CIN2-3 Cases, Cancer Cases, and Cancer Deaths Associated With Screening Beginning at Age 15 Years and Increased in 1-Year Increments to Age 25 Years, Among Women Followed for a Lifetime.

Table 8

Sensitivity Analysis Showing Expected False-Positives, Colposcopies, CIN2-3 Cases, Cancer Cases, and Cancer Deaths Associated With Screening Beginning at Age 15 Years and Increased in 1-Year Increments to Age 25 Years, Among Women Followed for a Lifetime. (more...)

Table 9. Base-Case Analysis.

Table 9

Base-Case Analysis.

Figure 1 shows the expected false-positives and cancer cases per 1000 adolescent women who begin screening at ages varying from 15 to 25 years and who are followed to age 30 years. The number of cancer and false-positive cases (per 1000) are presented on the y and z axes, respectively. The x-axis shows the age in years in which screening begins,varying from 15 to 25 years. The number of cancer cases varies from 0.16 per 1000 at age 15 years to 0.58 per 1000 at age 25 years. The number of false-positives per 1000 varies from 232.42 for screening beginning at age 15 to 100.88 for screening beginning at age 25 years.

Figure 1

Expected False-Positives and Cancer Cases for Adolescent Women Who Begin Screening at Ages Varying From 15 to 25 Years and Are Followed to Age 30 Years*. *Results presented assume an annual screening interval and are calculated per 1,000 women.

Figure 2 presents an efficiency curve comparing strategies that differ by age at first screening. The average number of colposcopies per 1000 women varies from 0 to 2250 along the x-axis. The average number of life-years varies from 69,000 to 69,300 per 1000 women. Strategies are compared by ranking them in order of average number of colposcopies per 1000 women. Adjacent strategies are then compared using an incremental colposcopy per life-year ratio, which is calculated using the difference in colposcopies divided by the difference in life-years. The strategies on the curve are those that are not dominated. A strategy that is “dominated” is eliminated from consideration because it is associated with more colposcopies and less effectiveness or fewer colposcopies but a higher colposcopy per life-year ratio than an adjacent strategy.

Figure 2

Efficiency Curve Comparing Strategies Differing By Age at First Screening*. *Strategies presented are those identified as efficient using incremental colposcopies per life-year.

In terms of screening interval, for both time horizons the patterns are similar: the number of false-positive test results and colposcopies increase as screening frequency increases, whereas the number of CIN2-3 cases, cancer cases, and cancer deaths decrease. Of the two time horizons, the lifetime horizon shows fewer fluctuations within successive age intervals. For both horizons, a screening interval of every year is associated with the highest number of colposcopies, exceeding one per woman screened for the lifetime horizon. Compared to screening beginning at age 21 years and conducted every 3 years (which is part of the current USPSTF recommendations), screening every year beginning at age 21 years results in more (1,931 vs. 758 per 1,000) colposcopies but is also associated with a reduction (approximately 3 vs. 9 per 1,000) in cancer cases. Screening every 2 years is associated with approximately 1,084 colposcopies and 6 cancer cases per 1,000 women. Taken together, these patterns can be used to explain the base-case findings presented in Table 9 and summarized in Figure 2. If the strategies that fall on the steepest part of the efficiency curve are assumed to represent a reasonable trade-off between colposcopies and life expectancy gained, then strategies of screening every 3 to 5 years beginning in the early 20s are more attractive, compared to those strategies that are identified as efficient but are based on screening every year in the teens. A strategy of screening every 5 years beginning at age 22 years is also more effective, but is associated with more colposcopies than screening every 5 years beginning at age 20 years. This is because the ASCCP guidelines for triage to immediate colposcopy start at age 21 years.33 Even though women aged 20 years have one more opportunity for screening (14 vs. 13, when screened through age 85 years), this additional screening occurs before age 21 years. As a result, those with abnormal results undergo repeat cytology instead of immediate referral to colposcopy; all women with abnormal test results are not referred for immediate colposcopy until age 26 years. This aspect of the guidelines may also explain why the currently recommended strategy of the USPSTF is not identified as an efficient strategy—the number of colposcopies is high, but fewer cancer cases are prevented at age 21 years compared to earlier ages.

CIN2 as a Percentage of CIN2-3

Table 10 uses different outcomes than other tables included in this report. Table 10 presents the total number of cases of CIN2-3 (per 1,000 women) over the shorter time horizon (until age 30 years), and the percentage of CIN2-3 cases estimated to actually be CIN2.

Table 10. Percentage of CIN2-3 Cases Estimated to Be CIN2.

Table 10

Percentage of CIN2-3 Cases Estimated to Be CIN2.

Key Sensitivity Analyses

Tables 11 through 16 present the results of the sensitivity analyses. The range of sensitivity analyses confirm the base-case findings, namely that screening frequently in the teens is associated with a large number of colposcopies, but relatively small gains in life expectancy. Results were similar to the base case when the sensitivity and specificity of colposcopy and biopsy were varied or different conditional probabilities of cytology results given underlying histology were used (data not shown). As shown in Table 10, approximately 70 percent of high-grade disease detected in this younger age group may be CIN2 rather than CIN3. This suggests that disease detected in these early years may be very likely to regress and that overdiagnosis and treatment of these lesions are potential concerns. The sensitivity analysis that uses screening cytology tests per life-year also shows that screening in the teens is associated with a high number of cytology tests per life-year gained. Finally, results from the revised natural history model, in which higher disease regression and lower progression is modeled, suggests that, if correct, screening could potentially be delayed past the early 20s.

Summary of Results for Specific Aim 1–Sub-Aim 1

Tables 17 through 19 and Figures 3 and 4 summarize the results for the base-case analyses for Sub-Aim 1, regarding the age at which to end screening. Outcomes for women, grouped by screening interval and whether they have never been screened or have been screened annually prior to age 65 years, are presented in Table 17. Among women who have never been screened, varying the age at which to end screening has a relatively small impact on cancer cases and cancer deaths, but a large impact on the number of colposcopies and false-positive test results. For instance, cancer deaths range from approximately 9 per 1,000 women (860 per 100,000) if screening is conducted every 5 years and ends at age 70 years, to approximately 8 per 1,000 women (789 per 100,000) if screening ends at age 90 years. For the same comparison, colposcopies range from 36.95 to 135.78 per 1,000 women. A similar pattern is seen with increasing the frequency of screening—small reductions in cancer cases and deaths, but large increases in false-positives and colposcopies. Although a similar pattern is also seen among women who have been screened every 3 years prior to age 65 years (i.e., many more colposcopies and false-positive test results associated with small decreases in cancer cases and deaths), compared to those who have never been screened, there is a large increase in colposcopies. For example, among women who have never been screened prior to age 65 years, screening every 5 years and ending at age 70 years is associated with approximately 37 colposcopies per 1,000 women, compared to 621 if screening is conducted every 3 years. Although deaths are increased, as expected, if age-specific survival ratios are used instead of pooled estimates of survival, the patterns observed above are similar. As a result, among those who have never been screened, strategies associated with infrequent screening (every 2 through 5 years) and ending at age 70 years fall on the steep part of the efficiency curve (Table 18 and Figure 3).

Table 17. Sensitivity Analysis Showing Expected False-Positives, Colposcopies, CIN2-3 Cases, Cancer Cases, and Cancer Deaths Associated With Different Ages at Which to End Screening, Varying in 5-Year Increments From Age 65 to 90 Years.

Table 17

Sensitivity Analysis Showing Expected False-Positives, Colposcopies, CIN2-3 Cases, Cancer Cases, and Cancer Deaths Associated With Different Ages at Which to End Screening, Varying in 5-Year Increments From Age 65 to 90 Years.

Table 18. Base-Case Analysis for Strategies Identified as Efficient Among Women Who Have Never Been Screened Prior to Age 65 Years.

Table 18

Base-Case Analysis for Strategies Identified as Efficient Among Women Who Have Never Been Screened Prior to Age 65 Years.

Table 19. Base-Case Analysis for Strategies Identified as Efficient Among Women Who Have Been Screened Every 3 Years Prior to Age 65 Years.

Table 19

Base-Case Analysis for Strategies Identified as Efficient Among Women Who Have Been Screened Every 3 Years Prior to Age 65 Years.

Figure 3 presents an efficiency curve comparing strategies that differ by the age at which to end screening. Women are assumed to not have been screened prior to age 65. The age at which to end screening varies in 5-year increments from age 70 to age 90. The number of colposcopies per 1000 women varies from 0 to 500 along the x-axis. The number of life-years varies from 69,015 to 69,040 per 1000 women. Strategies are compared by ranking them in order of average number of colposcopies per 1000 women. Adjacent strategies are then compared using an incremental colposcopy per life-year ratio, which is calculated using the difference in colposcopies divided by the difference in life-years. The strategies on the curve are those that are not dominated. A strategy that is “dominated” is eliminated from consideration because it is associated with more colposcopies and less effectiveness or fewer colposcopies but a higher colposcopy per life-year ratio than an adjacent strategy.

Figure 3

Efficiency Curve Comparing Strategies Differing By Age at Which to End Screening, Among Women Who Have Not Been Screened Prior to Age 65 Years*. *Strategies presented are those identified as efficient using incremental colposcopies per life-year.

Figure 4 presents an efficiency curve comparing strategies that differ by the age at which to end screening. Women are assumed to have been screened every 3 years prior to age 65. The age at which to end screening varies in 5-year increments from age 70 to age 90. The number of colposcopies per 1000 women varies from 590 to 790 along the x-axis. The number of life-years varies from 69,204 to 69,214 per 1000 women. Strategies are compared by ranking them in order of average number of colposcopies per 1000 women. Adjacent strategies are then compared using an incremental colposcopy per life-year ratio, which is calculated using the difference in colposcopies divided by the difference in life-years. The strategies on the curve are those that are not dominated. A strategy that is “dominated” is eliminated from consideration because it is associated with more colposcopies and less effectiveness or fewer colposcopies but a higher colposcopy per life-year ratio than an adjacent strategy.

Figure 4

Efficiency Curve Comparing Strategies Differing By Age at Which to End Screening, Among Women Who Have Been Screened Every 3 Years Prior to Age 65 Years*. *Screening is assumed to begin at age 21 years. After age 65 years, screening is then varied by (more...)

In contrast, there are much smaller differences between the strategies identified as efficient for women who have been screened frequently prior to age 65 years (Table 19 and Figure 4). The strategies cluster very closely together based on life expectancy, with an approximate 1-year gain in life expectancy per 1,000 women at most, which represents less than 1 day’s gain in life expectancy per woman. These results are robust across a range of sensitivity analyses (Tables 20-29), including the analyses which assume less than perfect compliance with screening and low estimates for sensitivity. Results are also robust when the sensitivity and specificity of colposcopy and biopsy are varied or different conditional probabilities of cytology results given underlying histology are used (data not shown).

Summary of Results for Specific Aim 2

Tables 30ac and 31ac and Figures 5 through 7 summarize the results for Specific Aim 2, which compares strategies that include HPV DNA testing to cytology only using three different sets of test accuracy estimates. As shown in Tables 30ac, the results are similar regardless of estimates used; there are fewer colposcopies but more cancer cases and cancer deaths associated with screening using cytology tests conducted at intervals of every 2, 3, and 5 years, compared to screening with cytology only prior to age 30 years, and then with cytology and HPV beginning at age 30 years. At an annual screening interval, the cytology-only strategy is associated with more colposcopies but fewer cases of cancer and cancer deaths, compared to the cytology and HPV strategy. The reason for this switch is that at less frequent intervals, a sufficient amount of disease is detected by adding the HPV test, which offsets any loss in disease detection due to women with dually negative test results being screened once every 3 years (beginning at age 30 years) for the cytology and HPV strategy. However, at the most frequent screening interval, the impact of women with dually negative results being screened every 3 years becomes evident, with fewer colposcopies and false-positives, but more cases of disease for the HPV and cytology strategy compared to the cytology-only strategy.

Table 30a. Vesco et al: Sensitivity Analysis Showing Expected False-Positives, Colposcopies, CIN2-3 Cases, Cancer Cases, and Cancer Deaths Associated With Cytology and HPV Test-Based Strategies Either Alone or in Combination.

Table 30a

Vesco et al: Sensitivity Analysis Showing Expected False-Positives, Colposcopies, CIN2-3 Cases, Cancer Cases, and Cancer Deaths Associated With Cytology and HPV Test-Based Strategies Either Alone or in Combination.

Table 30b. Mayrand et al: Sensitivity Analysis Showing Expected False-Positives, Colposcopies, CIN2-3 Cases, Cancer Cases, and Cancer Deaths Associated With Cytology and HPV Test-Based Strategies Either Alone or in Combination.

Table 30b

Mayrand et al: Sensitivity Analysis Showing Expected False-Positives, Colposcopies, CIN2-3 Cases, Cancer Cases, and Cancer Deaths Associated With Cytology and HPV Test-Based Strategies Either Alone or in Combination.

Table 30c. Koliopoulos et al: Sensitivity Analysis Showing Expected False-Positives, Colposcopies, CIN2-3 Cases, Cancer Cases, and Cancer Deaths Associated With Cytology and HPV Test-Based Strategies Either Alone or in Combination.

Table 30c

Koliopoulos et al: Sensitivity Analysis Showing Expected False-Positives, Colposcopies, CIN2-3 Cases, Cancer Cases, and Cancer Deaths Associated With Cytology and HPV Test-Based Strategies Either Alone or in Combination.

Table 31a. Vesco et al: Sensitivity Analysis Showing Expected Colposcopies, Incremental Colposcopies, Life-Years, Incremental Life-Years, and Incremental Colposcopies per Life-Year for Strategies Identified as Efficient.

Table 31a

Vesco et al: Sensitivity Analysis Showing Expected Colposcopies, Incremental Colposcopies, Life-Years, Incremental Life-Years, and Incremental Colposcopies per Life-Year for Strategies Identified as Efficient.

Table 31b. Mayrand et al: Sensitivity Analysis Showing Expected Colposcopies, Incremental Colposcopies, Life-Years, Incremental Life-Years, and Incremental Colposcopies per Life-Year for Strategies Identified as Efficient.

Table 31b

Mayrand et al: Sensitivity Analysis Showing Expected Colposcopies, Incremental Colposcopies, Life-Years, Incremental Life-Years, and Incremental Colposcopies per Life-Year for Strategies Identified as Efficient.

Table 31c. Koliopoulos et al: Sensitivity Analysis Showing Expected Colposcopies, Incremental Colposcopies, Life-Years, Incremental Life-Years, and Incremental Colposcopies per Life-Year for Strategies Identified as Efficient.

Table 31c

Koliopoulos et al: Sensitivity Analysis Showing Expected Colposcopies, Incremental Colposcopies, Life-Years, Incremental Life-Years, and Incremental Colposcopies per Life-Year for Strategies Identified as Efficient.

Figure 5 presents an efficiency curve comparing strategies based on cytology alone or in combination with HPV testing. The number of colposcopies per 1000 women varies from 0 to 2500 along the x-axis. The number of life-years varies from 69,000 to 69,300 per 1000 women. Strategies are compared by ranking them in order of average number of colposcopies per 1000 women. Adjacent strategies are then compared using an incremental colposcopy per life-year ratio, which is calculated using the difference in colposcopies divided by the difference in life-years. The strategies on the curve are those that are not dominated. A strategy that is “dominated” is eliminated from consideration because it is associated with more colposcopies and less effectiveness or fewer colposcopies but a higher colposcopy per life year ratio than an adjacent strategy.

Figure 5

Efficiency Curve Comparing Strategies Based on Cytology Either Alone or in Combination With HPV*. *For cytology and HPV combined strategies, women are assumed to be screened with cytology only (with a repeat cytology test for ASC-US results) before age (more...)

Figure 6 presents an efficiency curve comparing strategies based on cytology alone or in combination with HPV testing. The number of colposcopies per 1000 women varies from 0 to 1200 along the x-axis. The number of life-years varies from 69,000 to 69,300 per 1000 women. Strategies are compared by ranking them in order of average number of colposcopies per 1000 women. Adjacent strategies are then compared using an incremental colposcopy per life-year ratio, which is calculated using the difference in colposcopies divided by the difference in life-years. The strategies on the curve are those that are not dominated. A strategy that is “dominated” is eliminated from consideration because it is associated with more colposcopies and less effectiveness or fewer colposcopies but a higher colposcopy per life-year ratio than an adjacent strategy.

Figure 6

Efficiency Curve Comparing Strategies Based on Cytology Either Alone or in Combination With HPV*. *For cytology and HPV combined strategies, women are assumed to be screened with cytology only (with a repeat cytology test for ASC-US results) before age (more...)

Figure 7 presents an efficiency curve comparing strategies based on cytology alone or in combination with HPV testing. The number of colposcopies per 1000 women varies from 0 to 3000 along the x-axis. The number of life-years varies from 69,000 to 69,300 per 1000 women. Strategies are compared by ranking them in order of average number of colposcopies per 1000 women. Adjacent strategies are then compared using an incremental colposcopy per life-year ratio, which is calculated using the difference in colposcopies divided by the difference in life-years. The strategies on the curve are those that are not dominated. A strategy that is “dominated” is eliminated from consideration because it is associated with more colposcopies and less effectiveness or fewer colposcopies but a higher colposcopy per life-year ratio than an adjacent strategy.

Figure 7

Efficiency Curve Comparing Strategies Based on Cytology Either Alone or in Combination With HPV*. *For cytology and HPV combined strategies, women are assumed to be screened with cytology only (with a repeat cytology test for ASC-US results) before age (more...)

As a result, HPV and cytology is identified as a strategy that may provide a reasonable trade-off between the burden and benefits of screening, especially when conducted every 3 or 5 years (Tables 31ac and Figures 57). The base-case findings are generally similar across a range of sensitivity analyses (Tables 32-37ac), including varying the sensitivity and specificity of colposcopy and biopsy as well as the conditional probability of cytology given underlying histology (data not shown). If women with normal cytology results who are HPV negative are assumed to be screened every 5 years instead of every 3 (Tables 33ac), and the difference in test accuracy between cytology and HPV is large (Tables 33a and 33b), cytology conducted every 3 years beginning at age 21 years is dominated (more colposcopies and fewer gains in life expectancy) by the cytology and HPV strategy. Exceptions to the findings for the base case (for the three sets of estimates) are when screening and triage tests are used to reflect the burden of screening instead of colposcopies (Tables 32ac) and when a strategy of HPV followed by cytology for HPV positive women is modeled (Tables 36ac and 37ac). When screening and triage tests are used instead of colposcopies to quantify burden, cytology-only strategies are primarily identified as efficient. While the currently recommended strategy of cytology and HPV conducted every 3 years for women with dually negative results also falls on the efficiency curve, there is a large number of additional tests per life-year gained (approximately 1,000 to 1,200). A sensitivity analysis of HPV followed by cytology for HPV positive women shows that this is a potentially efficient strategy whether tests or colposcopies are used to quantify the burden of screening. This strategy, although not currently recommended, is more efficient than either the cytology-only or cytology and HPV strategies. This is because only those women with positive results on both tests are referred to colposcopy (compared to cytology-only strategies), reducing the burden of colposcopies due to false-positive results. Those women with discordant results (HPV positive, normal cytology) are assumed to undergo repeat screening 1 year later, with referral to colposcopy only if repeat testing is abnormal; thus, this strategy detects more disease than cytology only. Although the cytology and HPV strategy is associated with greater gains in life expectancy compared to HPV followed by cytology, it is associated with more colposcopies at the less frequent screening intervals (every 3 and 5 years). At the more frequent screening intervals (every 1 and 2 years), HPV followed by cytology is associated with greater gains in life expectancy because only a small proportion of women undergo routine screening at these intervals; the majority (with negative HPV and normal cytology results) are screened every 3 years. Use of tests instead of colposcopies produces similar results except when estimates of test accuracy from Koliopoulos et al36 are used. In this instance, cytology-based screening strategies are identified as efficient, which suggests that the magnitude of the difference in test accuracy between HPV and cytology, as well as the metric used to quantify burden of screening, influences the degree to which this strategy is considered efficient.

Key Sensitivity Analyses

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...