Chapter  5:  Evaluation of Cervical Cytology: Evidence Report/Technology Assessment Number 5

Objective

This report compares new technologies for cervical cytological screening with conventional Papanicolaou (Pap) test screening in terms of diagnostic accuracy, costs, effectiveness, and cost-effectiveness in adult women of average cervical cancer risk

Search Strategy

Published literature on the accuracy of cervical cytological screening, costs of screening and treatment, and cost-effectiveness were identified in MEDLINE, CINAHL, CancerLit, EconLit, HealthSTAR, and EMBASE databases.

Selection Criteria

Diagnostic test studies were included if they compared cervical cytology diagnosis with concurrent colposcopy or biopsy and provided estimates of sensitivity and specificity. For the new technologies, studies were also included that used a cytology reference standard and allowed estimation of either sensitivity or specificity. Articles on costs and health outcomes were selected if they assessed the effect of screening on life expectancy or quality of life, number of cases of cervical cancer, or total health care costs.

Data Collection and Analysis

For diagnostic test studies, paired reviewers independently abstracted sensitivity and specificity data from each study. Quality scores were assessed on blind interpretation of screening test results, histological reference standard, verification of test negative subjects, description of disease spectrum, avoidance of bias in sample selection, publication type, and source of support. Diverse articles on costs and health outcomes were summarized and quality-scored according to criteria published by an expert panel.

Supplemental analyses include a meta-analysis to generate summary estimates of Pap test discrimination; cost analysis using claims databases to generate costs of treatment and screening; and a Markov model to estimate the effectiveness and costs of different technologies and clinical strategies.

Main Results

Conventional Pap smear screening, based on the few studies that avoided severe biases, showed specificity of 98 percent (95 percent confidence interval (CI); 97-99 percent) and sensitivity of 51 percent (95 percent CI; 37-66 percent). The sample prevalence of disease is strongly related to between-study variability in Pap test sensitivity and specificity and may reflect bias. Other indicators of study quality were not significant when prevalence was controlled. The Pap test is more accurate when a high-grade squamous intraepithelial lesion threshold is used with the goal of detecting a high-grade lesion than when lower thresholds, such as a low-grade squamous intraepithelial lesion (LSIL) or atypical squamous cells of uncertain significance (ASCUS), are used with the goal of detecting low- or high-grade dysplasia. Few studies of the new technologies used histology or colposcopy as a reference standard or allowed estimates of both sensitivity and specificity. In studies using a cytology reference standard, each of the new technologies appears to significantly improve sensitivity relative to conventional Pap smear screening; however, little information is available on the effects on specificity.

Cost-effectiveness ratios from published models comparing Pap smear screening with no screening fall into an acceptable range, but these models used parameters that overstate Pap test accuracy.

Base case estimates of the incremental cost-effectiveness of conventional Pap screening every 3 years compared with no Pap screening is $4,097 per life-year saved. A technology applied to the initial step in Pap screening that reduces the false negative rate by a factor of 0.6 at an incremental cost per slide of $10 has an incremental cost of $22,010 per life-year saved when performed every 3 years. With more frequent screening intervals, the cost per life-year saved is greater than $50,000. Technologies that allow 100 percent rescreening of slides initially read as normal by conventional Pap screening, at a reduction in false negative rate of 0.85 or higher, are more effective than technologies that improve initial screening with a reduction of 0.6. At these reductions in false negative rate, with identical incremental costs of $10 per slide and a screening interval of every 3 years, the cost per life-year saved of rescreening technologies compared with improved initial screening is $45,375. Findings were relatively insensitive to assumptions about cervical cancer incidence, cost of technologies, diagnostic strategies for abnormal screening results, and age at onset of screening. Findings were sensitive to both the reduction in false negative rate (i.e., improvement in sensitivity) and the relative specificity of the technologies compared with conventional Pap.

Conclusions

Estimates of the sensitivity of the conventional Pap test are biased in most studies; based on the least biased studies, sensitivity is near 50 percent, much lower than generally believed. Newer technologies improve sensitivity compared with conventional Pap screening; however, there are no precise estimates for their effect on specificity. Under assumptions favorable to improved initial screening technologies and rescreening technologies, either approach can result in acceptable cost per life-year saved at 3-year Pap screening intervals. However, the imprecision in estimates of effectiveness and cost of the new technologies makes drawing firm conclusions about their relative cost-effectiveness problematic.

This document is in the public domain and may be used and reprinted without permission, except for those copyrighted materials noted for which further reproduction is prohibited without the specific permission of copyright holders.

Suggested Citation:

McCrory DC, Matchar DB, Bastian L, et al. Evaluation of Cervical Cytology. Evidence Report/Technology Assessment No. 5. (Prepared by Duke University under Contract No. 290-97-0014.) AHCPR Publication No. 99-E010. Rockville, MD: Agency for Health Care Policy and Research. February 1999.

New Technologies Assessed

Recent developments in specimen processing and interpretation may substantially improve the Pap smear as a diagnostic test for cervical cancer and cancer precursors. Three new devices recently approved by the Food and Drug Administration (FDA) are considered in this report: ThinPrep^®, Papnet^®, and AutoPap^®. The three devices employ three different types of technology: thin-layer cytology (ThinPrep^®) and computerized rescreening utilizing neural-network technology (Papnet^®) or algorithmic classification (AutoPap^®).

Each of these technologies was developed to reduce the false negative rate associated with cervical cytological screening. The two major components to this false negative rate are false negatives related to sampling error and false negatives related to detection error. About two-thirds of false negatives are a result of sampling error and the remaining one-third a result of detection error. Each of the new technologies is directed at one of these components of false negatives. Thin-layer cytology aims primarily to fix sampling error, whereas computerized rescreening targets detection error. This implies that neither technology will be able to reduce false negatives beyond a certain threshold.

One newly approved device, Papnet^®, uses neural-network computerized rescreening of Pap smears initially read as negative by a cytotechnologist.The system works by using automated computerized imaging of Pap smear slides and interpretation of images using a computerized algorithm to identify slides that are likely to contain abnormal cells. The Papnet^® system (Neuromedical Systems, Inc.) identifies cells or clusters of cells that require review and can display up to 128 images of the slide likely to contain abnormalities. These images can be reviewed by a cytotechnologist who can decide whether or not to review the slide using light microscopy.

AutoPap^® 300 QC system (Neopath, Inc.), an algorithm-based decisionmaking technology, identifies slides exceeding a certain threshold for the likelihood of abnormal cells. The laboratfory can select different thresholds corresponding to 10, 15, and 20 percent review rates. In contrast to random rescreening, the population of slides selected by the AutoPap^® 300 QC system is enriched with abnormalities and, at the 10-15 percent sort rate, this population of slides should contain 70-80 percent of the slides containing abnormalities missed by manual screening.

A variety of other technologies or clinical strategies have been proposed to improve Pap testing including various devices for collecting a cytological sample from the cervix. Still other technologies have been proposed to augment or replace cervical cytological screening, including colposcopic photographs for review by experts (cervicography) and DNA testing for specific human papillomavirus (HPV). These technologies are not considered in the present report.

Patient Population and Settings

The primary target population for this evidence report is women of average cervical cancer risk in the United States who are candidates for Pap smear screening. For the purposes of our analysis, candidates for Pap smear screening include women between the age of onset of sexual activity and the age of 85.

Although a large proportion of cervical cancer occurs in women with very limited or no screening, we did not examine programs or policies designed to improve screening compliance. Some previous studies have focused on special populations such as elderly women and elderly women who have not previously been screened.

The principal practice setting considered is the primary care practice in the United States (general internal medicine, family practice, adolescent medicine, and obstetrics/gynecology) and government and nongovernment family planning clinics (e.g., Planned Parenthood, public health clinics).

Literature Sources Used

MEDLINE, CancerLit, HealthSTAR, CINAHL, EMBASE, and EconLit computerized database searches, supplemented by manual journal searches and querying experts and device manufacturers, were the sources used to identify English language reports on the accuracy of cervical cytological screening, costs associated with screening and treatment, and cost-effectiveness.

Citations for the review of accuracy of cervical cytological testing were retrieved with a search strategy that combined various text word and index terms for cervical cytological tests with cervical cancer or dysplasia and sensitivity and specificity. The strategy to retrieve articles on the costs and health outcomes associated with cervical cancer screening combined cervical cytological test terms with terms describing cost analysis and mathematical modeling. Experienced librarians assisted with the design and translation of these search strategies for each database searched.

Screening of Articles

Separate sets of criteria for including articles in the evidence report were developed for the two topics that were the subject of literature reviews (diagnostic testing and cost and health outcomes). In each case, final screening criteria were developed through an iterative process. Each iteration of criteria was pilot-tested by each reviewer/abstractor on a subset of randomly chosen articles.

Articles on diagnostic testing were first screened based on information available through the online databases (primarily title, authors, and abstract when available). Citations were eliminated in Step 1 of the screening process if cervical cytology was not evaluated as a screening test or if the screening test results were not compared with a reference standard. In Step 2 of the screening process, full texts of articles were reviewed to select articles in which a reference standard of colposcopy or histology was used, the screening test and references standard were reasonably concurrent (i.e., within 3 months), and sufficient data to calculate both sensitivity and specificity were provided (i.e., all cells of a two-by-two table). Of the 939 bibliographic references reviewed, 561, or approximately 60 percent, were excluded during the first screening, and another 293, or 31 percent, during the second screening. Eighty-six articles were included according to these criteria: 84 studies of conventional Pap screening and one study each of ThinPrep^® and Papnet^®. Because so few studies of the new technologies met the original criteria, we modified the criteria to include studies of the new technologies that used a cytology reference standard and allowed estimation of either sensitivity or specificity. We considered a total of 59 studies (12 on AutoPap^®, 27 on Papnet^®, and 20 on ThinPrep^®) during this final stage of the screening process (Step 3). The net result was the inclusion of 6 studies of AutoPap^®, 11 of Papnet^®, and 8 of ThinPrep^®.

Articles on cost and health outcomes of cervical cytological screening were selected if they assessed the effect of screening on life expectancy or quality, number of cases of cervical cancer, or total health care costs for any of the following cytological screening technologies: conventional Pap smears, thin-layer cytology, or Pap smears with computerized rescreening. Of the 672 articles identified, 638, or 95 percent, were eliminated during the screening process. Thirty-four articles were included in the review.

Data Abstraction Process

Key information was abstracted onto specially designed forms and verified by either duplicate abstraction (two-by-two tables) or overreading by paired clinician-abstractors. Differences were resolved by consensus.

For the diagnostic testing articles, both members of each abstractor team also independently completed two-by-two tables for each study, extracting the key data to calculate sensitivity, specificity, and prevalence and other data to be used in the meta-analysis. The main outcome measures considered were the sensitivity and specificity of cytological abnormality by Pap test for detecting cases, where "cytological abnormality" was defined by one of three thresholds ranging from atypical squamous cells of uncertain significance (ASCUS) (threshold 1) to low-grade squamous intraepithelial lesion (LSIL) (threshold 2) to high-grade squamous intraepithelial lesion (HSIL) (threshold 3), and where a "case" was defined as a histological diagnosis of dysplasia or carcinoma. Equivalent categories in other classification schemes were also used. Two-by-two tables were constructed for four different combinations of cytological versus histological thresholds: ASCUS/ cervical intraepithelial neoplasia (CIN1), LSIL/CIN1, LSIL/CIN2-3, and HSIL/CIN2-3.

Criteria for Evaluating the Quality of Articles

Quality scores for articles on diagnostic testing were assigned according to predetermined methodological criteria based on blind interpretation of screening test results, use of a reference standard of histology, selection of test negative patients for verification, avoidance of bias in sample collection, description of the spectrum of disease in the sample, publication as a full report (as opposed to abstract), and source of support.

The quality of articles on costs and health outcomes was described according to recently published criteria by an expert panel on cost and effectiveness in medicine.

Supplemental Analyses

Cost Analysis

Several available datasets were analyzed to estimate direct medical costs of screening, diagnosing, and treating cervical cancer, calculating separate estimates for women 20-64 years of age and those 65 years and older (eligible for Medicare). For women 20-64, the unit cost of screening, diagnosis, and treatment of cervical cancer was estimated from MEDSTAT data from 1992, 1993, and 1994, inflated to reflect 1994 charges and converted to costs using 1994 cost-to-charge ratios published by the American Hospital Association.

For women over 65, Medicare's resource-based relative value scale (RBRVS) fee schedule for physician services, Medicare's clinical laboratory fee schedule for laboratory services, and national average Diagnosis-related group (DRG) payments for hospital admissions were used to identify the payments associated with services received for cervical cancer screening, diagnosis, and treatment. Charges and payment information obtained from all sources were then converted to reflect costs associated with the services provided and all costs were inflated to 1997 dollars.

Cost-effectiveness Model

We constructed a 20-state Markov model that follows a cohort of women from age 15 to 85 and assumes that there are no prevalent cases of HPV infection or squamous intraepithelial lesion (SIL) at age 15. Cycle lengths are 1 year long. No Pap smear screening is compared with the following screening strategies: conventional Pap smears at 1-, 2- and 3-year intervals, thin-layer cytology smears at 1-, 2- and 3-year intervals, and 100 percent computerized rescreening at 1-, 2- and 3-year intervals.

We used a U.S. health system perspective and evaluated the direct and health-care specific costs associated with screening, diagnosis, and treatment of cervical cancer and its precursors. We did not consider other societal costs such as work loss. The model considers the following outcomes: cost per year of life saved, cost per cervical cancer death prevented and per cervical cancer case prevented, and the number of morbid therapies avoided.

We discounted costs and years of life at 3 percent annually in the base case and varied the discount rate from 0 to 5 percent in a sensitivity analysis. Specific parameter estimates were derived from a preliminary literature assessment conducted for this report and prior published models of cervical cancer screening.

Findings

Important findings regarding the discrimination about the accuracy of cervical cytological screening include the following:

• Despite the demonstrated ability of cervical cytological screening in reducing cervical cancer mortality, the conventional Pap test is less sensitive than it is generally believed to be.

• Few studies of primary screening were unaffected by "workup" bias, but the few that were provided estimates of the specificity of Pap smear screening of 0.98 (95 percent confidence interval; 0.97-0.99) and sensitivity of 0.51 (95 percent confidence interval; 0.37-0.66).

• The Pap test is more accurate when a higher cytological threshold (HSIL) is used with the goal of detecting a high-grade lesion. Lower test thresholds or use of the Pap test for detecting low-grade dysplasia results in poorer discrimination.

The accuracy of the Pap test is strongly affected by disease prevalence. Higher disease prevalence is associated with higher estimates of sensitivity and lower estimates of specificity (with a greater effect on specificity). These findings are consistent with prevalence as a marker for workup bias and perhaps also reflect an imperfect reference standard that is more specific than sensitive.

• Quality of the studies reviewed, based on previously described criteria, varied widely; however, quality score did not explain a statistically significant amount of the between-study variation in discrimination when the variation in the prevalence of disease was controlled.

• Existing information fails to provide accurate estimates for specificity of thin-layer cytology or computerized rescreening technologies. Our initial requirement for verification of test negatives with colposcopy or histology led to the exclusion of all but one study each of ThinPrep^® and Papnet^® and all studies of AutoPap^®. The values reported for sensitivity and specificity in the few studies that use histological or colposcopic reference standards are well within the range of sensitivity and specificity reported for the conventional Pap test. However, including studies that directly compare these new technologies with conventional Pap smear testing (screening or rescreening) using a cytological reference standard results in significant improvements in sensitivity.

Important findings regarding the costs of cervical cytological screening and cervical cancer diagnosis and treatment include the following:

• Pap smear screening cost is somewhat higher in older women than younger women chiefly because physician and total time spent in obtaining Pap smears during office visits is longer for older women.

• Estimated costs of cervical cancer treatment calculated from episodes of care are substantially higher than estimates based on average procedure-specific costs because of both the provision of related services and the effect of complicated cases with unusually high costs. Estimates based on procedure-related costs alone will underestimate the true direct medical costs.

Important findings from a review of previously published models of the cost and effectiveness of cervical cytological screening include the following:

• Published models examining the cost and effectiveness of Pap smear screening have consistently found Pap screening to have a significant impact on the incidence and mortality of cervical cancer and to have an acceptable range of cost-effectiveness ratios when compared with no screening.

• Estimates of Pap test accuracy used in these models generally overestimated Pap test performance, as determined by recent unbiased studies and the findings of this report, and a previously published meta-analyses. Best estimates of Pap test performance fall outside the range used in sensitivity analyses of some models.

Important findings from a new model of cost and effectiveness of cervical cytological screening include the following:

• The cost-effectiveness of either a technology that improves primary screening sensitivity (e.g., thin-layer cytology), or one that improves rescreening sensitivity (e.g., computerized rescreening), is directly related to the frequency of screening -- longer intervals result in lower estimates of cost per life year saved.

• Our findings were relatively insensitive to assumptions about cervical cancer incidence, the cost of technologies, diagnostic strategies for abnormal screening results, age at onset of screening, or most other variables tested.

• There is substantial uncertainty about the estimates of sensitivity and specificity of thin-layer cytology and computerized rescreening technologies compared with each other and with conventional Pap testing. The uncertainty is not reflected in the point estimates for effectiveness or cost-effectiveness. Although it is clear that both thin-layer cytology and computerized rescreening technologies provide an improvement in effectiveness at higher cost, the imprecision in estimates of effectiveness makes drawing conclusions about the relative cost-effectiveness of thin-layer cytology and computerized rescreening technologies problematic.

Future Research

Our research suggests several areas for possible future study.

• Future decision models, cost-effectiveness studies, and health policy decisions should consider the sensitivity of Pap smear screening close to 50 percent.

• Thin-layer cytology technology (ThinPrep^®), the computerized rescreening device (AutoPap^®), and the algorithm-based decisionmaking technology (Papnet^®) have received regulatory approval from the FDA based on their demonstration of improved sensitivity compared with conventional Pap smear techniques. However, the evidence currently available does not fully describe the impact of these technologies on the specificity of the screening process. It is possible that a new technology might simultaneously raise both sensitivity and specificity; however, this has not been conclusively demonstrated for the devices reviewed in this report. Future studies of these technologies should include verification of test negative subjects to allow estimation of specificity.

• Comparisons with cytological reference standards attest to the validity of the new technologies compared with optimal Pap screening, but comparison with a histological reference standard provides a more relevant outcome for clinical decisionmakers, since histological diagnosis forms the basis of most clinical management decisions. Further research is needed to validate negative cytological diagnoses made with the new technologies with colposcopy, in both low-prevalence and high-prevalence populations. This could be accomplished by subjecting a random sample of cytology-negative women to colposcopy, which would permit statistical correction for workup bias and estimation of test specificity.

• Further research is needed to quantify the effect of cervical cancer and premalignant cervical lesions and various treatments for cervical cancer or dysplasia on quality of life. These data will allow a more comprehensive assessment of the impact of technologies for cervical cytological screening.

Incidence and Prevalence

The majority of cases of invasive cervical cancer occur in women who have either never had a Pap smear or have not had a smear in the previous 5 years (Shingleton, Patrick, Johnston et al., 1995). Lack of screening may explain many of the differences in cervical cancer rates among ethnic groups (Miller, Kolonel, Bernstein et al., 1998); for example, overall mortality from cervical cancer in black women is higher than in white women; stage-specific survival is identical, but black women present with more advanced disease (Ries, Kosary, Hankey et al., 1997).

Although all sexually active women are at risk for cervical cancer, the disease is more common in women of low socioeconomic status and in those who smoke, have a history of multiple sex partners, or have early onset of sexual intercourse.

Certain types of HPV have a strong epidemiological association with cervical cancer. Types 16 and 18 are two of several oncogenic strains of the more than 70 types of HPV. Both incidence and prevalence of HPV are greatest in women under the age of 25 (Koutsky, 1997). Ho, Bierman, Beardsley et al. (1998) have recently reported a cumulative 3-year incidence of 43 percent, or an average annual incidence of 14 percent, in a cohort of college women. Prevalence of HPV DNA in women with normal cervical cytology decreases with age in a variety of populations (Bauer, Hildesheim, Schiffman et al., 1993; Coker, Jenkins, Busnardo et al., 1993; ; Figueroa, Ward, Luthi et al., 1995; Hildesheim, Gravitt, Schiffman et al., 1993; Wheeler, Parmenter, Hunt et al., 1993). A second peak is seen after age 40 in some studies (Figueroa, Ward, Luthi et al., 1995); whether this is due to reinfection, or a cohort effect secondary to differences in age at onset of sexual activity, is unclear. Data on postmenopausal women are rare. Reported prevalence in peri- and postmenopausal women ranges from 1 percent (Ferenczy, Gelfand, Franco et al., 1997) to 38.1 percent (Smith, Johnson, Figuerres et al., 1997), a difference that may be attributable to differences in populations, cohort effects, or viral assay techniques. An international study found prevalence in women between 50 and 59 ranging from 4.1 percent in Spain to 13.6 percent in Brazil (Munoz, Kato, Bosch et al., 1996). Prevalence in a German cohort of women over 55 was 3.2 percent (De Villiers, Wagner, Schneider et al., 1992).

The natural history of HPV infection is complex, with clearance and persistence of viral DNA, along with progression to squamous intraepithelial lesion (SIL), varying depending on the viral type, patient characteristics such as age, and study design and assay methods (Herrero, 1996; Kiviat, 1996; Koutsky, 1997; Mitchell, Tortolero-Luna, Wright et al., 1996 ; Schiffman, Bauer, Hoover et al., 1993).

Disease Biology

Although a small percentage of cervical cancers do not have detectable HPV DNA, even with sensitive assays, there is consensus that HPV infection is the causative agent for the vast majority of cervical cancers (Herrero, 1996; Kiviat, 1996; Koutsky, 1997; Schiffman, Bauer, Hoover et al., 1993). Certain HPV types are clearly more likely to progress to cancer than others, and identification of these types in cervical cells may have a role in determining optimal diagnostic and treatment strategies for patients with abnormal Pap smears (Cox, Lorincz, Schiffman et al., 1995).

Natural History

Invasive cervical cancer usually is preceded by a long, premalignant phase during which it is readily treatable (Cannistra and Niloff, 1996; Wright and Richart, 1992; Wright, Kurman, and Ferenczy,1994). Even if this premalignant phase progresses to invasive cancer, early-stage disease is highly curable: 5-year survival rates for early-stage disease are above 90 percent, but if the disease has spread outside of the pelvis (Stage IV), cure rates are only 10-15 percent.

Epidemiological data suggest that a substantial proportion of patients with low-grade squamous intraepithelial lesions (LSIL) will have regression if the lesion are not treated, a finding that supports a "watchful waiting" strategy involving retesting.

Burden of Illness

Despite the success of Pap smear screening, women continue to develop cervical cancer. Treatment of early-stage disease is effective, but both of the alternatives, radical hysterectomy or radiation therapy, can have significant short- and long-term morbidity (Cannistra and Niloff, 1996; Landoni, Maneo, Columbo et al., 1997). Mortality from late-stage disease remains high.

Target Populations

The primary target population for this evidence report is women of average cervical cancer risk in the United States who are candidates for Pap smear screening. For the purposes of our analysis, candidates for Pap smear screening include women between the age of onset of sexual activity and the age of 85.

Although a large proportion of cervical cancer occurs in women with very limited or no screening, we did not examine programs or policies designed to improve screening compliance. Some previous studies have focused on special populations such as elderly women (Fahs, Mandelblatt, Schechter et al., 1992; U.S. Congress Office of Technology Assessment, 1990) and elderly women who have not previously been screened (Mandelblatt and Fahs, 1988).

Practice Settings

The principal practice setting considered will be the primary care practice in the United States (general internal medicine, family practice, adolescent medicine, and obstetrics/gynecology) and government and nongovernment family planning clinics (e.g., Planned Parenthood, public health clinics).

Literature Sources Used

Primary sources of literature for the two key research questions that were the subject of literature reviews were six of the most widely used online bibliographic databases: MEDLINE, CancerLit, HealthSTAR, CINAHL, EMBASE, and EconLit. Although there is considerable overlap among these databases, each contains unique information.

Produced by the U.S. National Library of Medicine, the MEDLINE database is widely recognized as the premier source for bibliographic and abstract coverage of biomedical literature. MEDLINE coverage began in 1966. More than 8.7 million records from more than 3,600 journals are indexed, plus selected monographs of congresses and symposia. Abstracts are included for about 67 percent of the records.

Produced by the U.S. National Cancer Institute, CancerLit is an important source of bibliographic records (most with abstracts) beginning in 1983 and pertaining to all aspects of cancer therapy, including experimental and clinical cancer therapy; chemical, viral, and other cancer-causing agents; mechanisms of carcinogenesis; biochemistry, immunology, and physiology of cancer; and mutagen and growth factor studies. Approximately 200 core journals contribute a large percentage of the records. Other entries are drawn from proceedings of meetings, government reports, symposia reports, theses, and selected monographs. Indexed materials include articles from journals, abstracts of papers presented at professional meetings, government and technical reports, dissertations, and monographs.

HealthSTAR, produced cooperatively by the U.S. National Library of Medicine and the American Hospital Association, contains citations to the published literature on health services, technology, administration, and research from 1975 to the present. Topics included that were of particular interest to us are evaluation of patient outcomes; effectiveness of procedures, programs, products, services, and processes; health economics and financial management; and quality assurance. This database contains citations and abstracts (when available) to journal articles, monographs, technical reports, meeting abstracts and papers, book chapters, government documents, and newspaper articles.

The Cumulative Index to Nursing & Allied Health (CINAHL) provides authoritative coverage of the literature related to nursing and allied health from 1983 to the present. CINAHL was chosen for the literature searches on cervical cytology primarily because it indexes several medical laboratory journals.

EMBASE is a comprehensive pharmacological and biomedical database containing over 3 million records from 1980 to the present, indexed from 3,500 journals published in 70 countries. More than 65 percent of the records contain abstracts.

EconLit contains abstracts of journal articles, books, and working papers, book reviews, and citations to articles in collective volumes. It covers economic literature broadly, from 1969 to the present.

Computerized searches of these online databases were supplemented by secondary searches, including the manual scanning of newly published journal issues that had not yet been indexed in the online databases. This scanning was accomplished through online searches of journal websites as well as searches of printed journals obtained through subscriptions and the Duke University Medical Center Library. In addition, the World Wide Web of the manufacturers of the automated cytology devices reviewed in this report were checked several times a week, along with those of relevant professional societies.

Search Strategies for Computerized Databases

The two strategies used to search the above computerized databases were developed in consultation with a Duke University Medical Center librarian who specializes in evidence-based medicine research. There were several iterations of each search strategy before the final strategies were agreed upon. The final strategy for articles on diagnostic testing for cervical cancer was:

• 1

  Vaginal smears/

• 2

  ((pap or papan$) and (smear$ or test$)).tw

• 3

  (papnet or autopap or thinprep).tw.

• 4

  1 or 2 or 3

• 5

  exp Cervix neoplasms/

• 6

  Cervix dysplasia/

• 7

  Cervical Intraepithelial Neoplasia/

• 8

  dyskaryo$.tw.

• 9

  5 or 6 or 7 or 8

• 10

  exp "Sensitivity and Specificity"/

• 11

  (sensitivity or specificity).tw.

• 12

  exp Diagnostic errors/

• 13

  4 and (10 or 11 or 12)

• 14

  13 and 9

• 15

  limit 14 to (human and english language)

• 16

  Papillomavirus, Human/

• 17

  15 not 16

The final search strategy for articles on the costs and health outcomes associated with cervical cancer screening was as follows:

• 1

  Vaginal smears/

• 2

  ((pap or papan$) and (smear$ or test$)).tw

• 3

  (papnet or autopap or thinprep).tw.

• 4

  1 or 2 or 3

• 5

  exp "costs and cost analysis"/

• 6

  (cost$ or expenditure$).tw.

• 7

  ec.fs.

• 8

  5 or 6 or 7

• 9

  4 and 8

• 10

  limit 9 to (human and english language)

• 11

  exp Decision Support Techniques/

• 12

  exp Models, Statistical/

• 13

  Technology assessment, biomedical/

• 14

  Monte carlo method/ or Survival Analysis/

• 15

  11 or 12 or 13 or 14

• 16

  4 and 15

• 17

  limit 16 to (human and english language)

• 18

  10 or 17

These strategies were initially developed using the National Library of Medicine MeSH key word nomenclature developed for MEDLINE. The Duke University Medical Center librarians assisted with the translation of these search strategies to the key word structures used by other databases. The basic search terms remained the same. Each online search began with the first year of the database (see individual database descriptions above).

After the initial runs in January 1998, the computerized searches were updated twice, in February and March. The final yield of the two searches was 1,538 unduplicated articles--939 articles were identified by the diagnostic testing searches and 671 by the cost and health outcomes searches, including articles identified through manual searches. Approximately 67 percent of all articles were identified through MEDLINE, 23 percent through EMBASE, and 5 percent through manual searches, with the remaining 5 percent divided among CancerLit, HealthSTAR, and CINAHL. EconLit did not identify any relevant or unique articles.

Development of a Bibliographic Database

As online literature searches were conducted, results were imported into Pro-Cite Version 2.0, a database software program for storing, managing, and retrieving bibliographic information. Each article was given a unique identifier within the Pro-Cite database. An ongoing search for duplicates was conducted. Initially, each article was coded to indicate its source and key question(s). As screening and data abstraction progressed, each article was coded for the stage at which it was excluded from the evidence report.

Screening of Articles

Separate sets of criteria for including articles in the evidence report were developed for the two topics that were the subject of literature reviews (diagnostic testing and cost and health outcomes). In each case, final screening criteria were developed through an iterative process. Each iteration of criteria was pilot-tested by each reviewer/abstractor on a subset of randomly chosen articles.

Study Design of Included Articles

At the time the literature searches were conducted, there were no completed randomized clinical trials comparing conventional Pap tests with new technologies to detect cervical cancer and its precursors. Study designs included in the diagnostic testing review were diagnostic test evaluations, meta-analyses, and review articles. Cost and health outcomes articles included were cost studies, typically cost-effectiveness and cost-benefit analyses.

Data Abstraction Process

The review team responsible for including a given article was also responsible for abstracting data and other key information from that article. The goal of the data abstraction process was to abstract essential information directly into the template of the evidence table in the format required by the Agency for Health Care and Research (AHCPR).

The evidence table entries were created by one member of each review team, over-read by the other team member, and then revised. For the cost and health outcomes articles, each evidence table entry was again over-read by a clinician.

Figure 2. Map of cervical cytology classification schemes (more...)

Figure 2. Map of cervical cytology classification schemes

Classification System	Within Normal Limits	Benign Cellular Changes	Epithelial Abnormalities
The Bethesda System (National Cancer Institute Workshop, 1993)	Normal	Infection reactive repair	ASCUS	Squamous intraepithelial lesion (SIL)					Invasive carcinoma
				Low grade (LSIL)		High grade (HSIL)
Richart (1973)				Condylom a	Cervical intraepithelial neoplasia (CIN)
					Grade 1	Grade 2	Grade 3
Reagan (1979) (WHO)		Atypia		Mild dysplasia		Moderate dysplasia	Severe dysplasia	In situ carcinoma
Papanicolaou (Nyirjesy, 1972)	I	II		III			IV		V

ASCUS = atypical squamous cell of uncertain significance.

Some studies that failed to permit calculation of sensitivity and specificity directly nonetheless allowed calculation of relative TPR and relative FNR (false negative rate) according to the method of Chock, Irwig, Berry et al. (1997). This method was applied to studies comparing two independently applied screening tests in which verification was obtained on any patient testing positive on either test (but not those testing negative on both tests).

Criteria for Evaluating the Quality of Articles

Given the two very distinct sets of articles reviewed for the evidence report, two separate sets of criteria to judge their quality were needed.

Cost Estimates for Women 20-64 Years

To obtain cost estimates for this age group, Health Economics Research, Inc., analyzed the MarketScan database created by the MEDSTAT Group and previously used in cost analyses of AHCPR practice guidelines. The MEDSTAT MarketScan database provides claims information for a large sample of privately insured individuals employed at large private companies across the country. The claims data are collected from over 100 different insurance companies, Blue Cross and Blue Shield plans, and third-party administrators (for self-insured employers). The data represent the medical experiences of insured active employees, early retirees, Consolidated Omnibus Budget Reconciliation Act (COBRA) participants, and their dependents. These data were used to estimate screening and treatment costs for the population under the age of 65. For our analyses, we used four types of claims files: inpatient claims, physician claims, outpatient department claims, and pharmaceutical claims. Information on service dates, procedures, diagnoses, and payments was abstracted. We analyzed data for a 3-year period, 1991-93, which represented the medical experience for roughly 2.5 million privately insured beneficiaries with more than 500,000 inpatient admissions and over 76 million outpatient claims.

Although the MarketScan database is a rich source of medical utilization information for the under-65 population, a number of issues arise when these data are used that may or may not affect our analyses. First, our ability to track treatment costs over time due to individuals' dropping in and out of the MarketScan database because of employment changes is limited. Thus, the data were used to estimate average costs of treatment and screening at an aggregated level. It is also important to note that the MarketScan data represent a nonrandom sample of the under-65 population. The MarketScan database includes a disproportionate number of persons employed in large firms, especially in California, Michigan, Ohio, and Texas. To the extent that large employers generally offer more generous insurance benefits than do smaller firms, utilization estimates derived from the MarketScan database are likely to overstate the general under-65 utilization experience. This bias would most directly affect cost estimates for cancer treatment, rather than cost estimates for specific diagnostic procedures or therapeutic services, as episode of care estimates are most affected by differential utilization patterns. Better insured women may use more services as well as more expensive services than less well-insured women. Last, the MarketScan database reflects the mix of fee-for-service and managed care arrangements present in the early 1990s. To the extent that the risk sharing and payment arrangements for the set of large firms in the MarketScan database differ systematically from those observed for the general under-65 population, there might be biases in the derived cost estimates. However, the direction of the bias is not readily known.

We used the MEDSTAT data to estimate the unit cost of screening, diagnosis, and treatment of cervical cancer. Since 3 years of data (1992, 1993, and 1994) were analyzed, all 1992 and 1993 charges were inflated to reflect 1994 charges. The national medical care services (includes hospital and physician services) inflation rate of 12.02 percent for 1992 charges and 5.17 percent for 1993 charges were utilized (CPI-U, U.S. Bureau of Labor Statistics, 1982-84).

The costs associated with hospital and physician services were derived from the MEDSTAT data estimates by utilizing appropriate adjustments. In the case of services provided in a hospital setting, the total covered charge for each service submitted by the hospital was identified from the MEDSTAT data. The inpatient hospital charges include room and board and ancillary charges. To convert these charges to costs, 1994 cost-to-charge ratios published by the American Hospital Association (AHA) (obtained by AHA through its Annual Survey of Hospitals) were utilized. The 1994 cost-to-charge ratio was 0.58. That is, on average across the Nation, the cost associated with the inpatient services provided was 58 percent of the charge.

Since no similar cost-to-charge ratio was available for physician charges, it was assumed that the payments made by the health care plans closely reflected the costs associated with providing these services. Thus, payments were used as a proxy for costs. This approach provides a "total" cost estimate for a given procedure or set of procedures. It not only includes the amount reimbursed by the insurance plan but also any copayments made by the patient. These payments differ from charges because they include only the amount eligible for payment under the benefit plan after pricing guidelines such as discounts and fee schedules are applied.

Cost Estimates for Women 65 Years and Older

Medicare's RBRVS fee schedule for physician services, Medicare's Clinical Laboratory fee schedule for laboratory services, national average diagnosis-related group (DRG) payments for hospital admissions, and ambulatory surgery center payment rates for outpatient procedures were used to identify the payments associated with services received for cervical cancer screening, diagnosis, and treatment. The 1994 Medicare payments were used to allow for comparability in the estimates between the over-65 and the under-65 age groups (MEDSTAT data estimates). The Medicare payments were assumed to closely approximate the national average cost of providing the service, and therefore no conversion factors were employed (Eisenberg, 1989).

Cost estimates for both the under-65 and the over-65 age groups were updated to reflect 1997 dollars. Separate inflators were employed for services provided at institutional and noninstitutional settings. Services received at institutions were updated to 1997 dollars using the DRI-Prospective Payment System (PPS) Hospital Market Index and Health Care Financing Administration (HCFA) Capital Index. A rate of 9.85 percent was utilized. The rate was calculated using DRI-PPS Hospital Consumer Price Survey (CPS) Market Index (DRI/McGraw Hill) for operating costs, and HCFA Capital Index (Federal Register) for capital cost. A rate combining the operating and capital inflator was obtained by assuming that 90 percent of costs were related to operations and 10 percent to capital. The Medicare Economic Index (MEI) was used to inflate 1994 costs of all noninstitutional services to 1997 dollars. The MEI of 8.40 percent was utilized.

Introduction

This section describes our response to the question, "What are the effects on total health care costs, morbidity, and mortality of regular cervical cytological screening using thin-layer cytology or computer rescreening compared with conventional Pap smear in women participating in a screening program?"

Our basic approach to this question was to create a decision analytic model to simulate the health and economic impact of various cervical cancer screening strategies based on the results of our literature review and meta-analysis. The model, borrowing from previous similar efforts, was created to generate a wide variety of health and cost outcome projections, including survival, cumulative cost, and incremental cost-effectiveness. However, because limited data were available regarding the natural history of cervical cancer and the costs and effectiveness of cervical cancer screening and treatment, we concluded that the sort of precise projections that would be of use to decisionmakers would not be possible. In particular, we determined that the aggregate effect of uncertainty in the model inputs (see below) led to substantial uncertainty in the outputs, such that a point estimate of incremental cost-effectiveness could not be taken as a reliable representation of the value of many of the screening strategies under investigation. As a consequence, we focused the model on the general question, "What characteristics would a cervical cancer screening strategy need to have in order to be cost-effective?"The results of this investigation are intended to guide future research regarding cervical cancer screening strategies.

Two model inputs that are crucial to the precise estimation of the cost-effectiveness of new screening technologies are particularly uncertain, namely, test operating characteristics (sensitivity and specificity) and test costs. Evaluation of the sensitivity and specificity of new screening technologies is particularly problematic given the discrepancy in reference standards. As previously stated, because the clinical management of the patient depends on cytology, colposcopy, and histology taken together (and certainly not cytology alone), histologic diagnosis, or at least colposcopic diagnosis, is the most appropriate reference standard. However, given the practical and methodological difficulties in applying this reference standard, and the paucity of studies available that used such a standard, we attempted to identify other studies that met alternative criteria, as outlined above. Unfortunately, there were few studies that met even these alternative criteria, and the resulting estimates of relative true positive rates and false positive rates had very wide confidence intervals.

Our estimates for the costs associated with implementation of ThinPrep ^®, AutoPap^®, and Papnet^® are based on a variety of sources, including the manufacturers. There is a wide range of estimates, and issues such as training costs, amortization of equipment, and labor costs are contestable. Our original estimate for one technology, based primarily on information from the manufacturer, was thought to be unacceptably low by every reviewer of the draft report except that manufacturer.

Because of the difficulty in estimating operating characteristics and costs that have face validity for all readers, we decided that presenting cost-effectiveness estimates for generic technologies across the range reported would ultimately prove more useful to readers than selecting base case estimates that were uncertain and lacked face validity. Our alternative approach is to provide cost-effectiveness estimates across a range of test characteristics and incremental costs and to identify ranges of these parameters (thresholds) that lead to cost-effective cervical cancer screening. This allows readers to use their own judgment about which costs and effectiveness measures are most appropriate. The thresholds also provide a "target" for future studies -- convincing evidence of either cost or effectiveness that meets these threshold values will be evidence for acceptable cost-effectiveness. To perform this analysis, we considered two generic strategies for improving Pap sensitivity -- 3/4improving the sensitivity of the initial screening test, or allowing 100 percent rescreening with improved sensitivity. By varying sensitivity, specificity, incremental cost, and screening frequency, we sought to identify threshold values where the incremental cost per life-year saved was within generally acceptable limits (in this case, we used a threshold value of $50,000/life-year).

To address the general question of what would constitute a substantially valuable improvement in Pap technology, we reframed the specific question to be addressed by the cost-effectiveness modeling to: "What are the ranges of incremental cost, sensitivity, specificity, and screening frequency that meet conventional levels of cost per life-year saved for technologies that improve Pap test performance by (1) improving the sensitivity of the initial screening step, or (2) allowing 100 percent rescreening at improved sensitivity?"

Also, since some decisionmakers prefer presentation of specific outcomes rather than an aggregate outcome, the incremental cost-effectiveness ratio, we pursued a secondary question: "What are the expected number of cervical cancer cases, deaths, and treatments prevented at different levels of test sensitivity and screening frequency?"

Structure of the Model

For the cost-effectiveness analysis, we constructed a model with two major components. The first component is a 20-state Markov model (Sonnenberg and Beck, 1993) intended to simulate the natural history of cervical cancer in the absence of screening. The second component is an intervention model that represents possible screening strategies. The model was developed in DATA 3.0 software (Boston, MA: TreeAge Software, Inc).

Natural History

Table 1. Markov Model Description

Table 1. Markov Model Description

State	Definition	Allowable Transitions
Well	Never infected with HPV, no history of SIL	Well, benign hysterectomy, death from other cause, undetected HPV
Benign hysterectomy	Hysterectomy for cause other than cervical cancer or SIL	Benign hysterectomy, death from other cause
Death from other cause	Death from cause other than cervical cancer	Absorbing state
Death from cervical cancer	Death from cervical cancer	Absorbing state
Undetected HPV	Undiagnosed cervical HPV infection	Well, death from other cause, benign hysterectomy, undetected HPV, LSIL
Detected HPV	Diagnosed cervical HPV infection or posttreatment for SIL	Well, detected HPV, death from other cause, benign hysterectomy LSIL, HSIL
LSIL	Low-grade squamous intraepithelial lesion	LSIL, death from other cause, benign hysterectomy, undetected HPV, detected HPV, well, HSIL
HSIL	High-grade squamous intraepithelial lesion	HSIL, death from other cause, benign hysterectomy, detected HPV, undetected HPV, LSIL, well
Unknown Stage I	Undiagnosed Stage I cervical cancer	Unknown Stage I, detected Stage I, death from other cause, unknown Stage II
Detected Stage I	Stage I cancer diagnosed by Pap or symptoms	Detected Stage I, death from cervical cancer (over 5 years), death from other cause
Stage I survivor	5 years after initial diagnosis of Stage I	Stage I survivor, death from other cause
Unknown Stage II	Undiagnosed Stage II cervical cancer	Unknown Stage II, detected Stage II, death from other cause, unknown Stage III
Detected Stage II	Stage II cancer diagnosed by Pap or symptoms	Detected Stage II, death from cervical cancer (over 5 years), death from other cause
Stage II survivor	5 years after initial diagnosis of Stage II	Stage II Survivor, death from other cause
Unknown Stage III	Undiagnosed Stage III cervical cancer	Unknown Stage III, detected Stage III, death from other cause, Unknown Stage IV
Detected Stage III	Stage III cancer diagnosed by Pap or symptoms	Detected Stage III, death from cervical cancer (over 5 years)
Stage III survivor	5 years after initial diagnosis of Stage III	Stage III Survivor, death from other cause
Unknown Stage IV	Undiagnosed Stage IV cervical cancer	Unknown Stage IV, detected Stage IV, death from other cause
Detected Stage IV	Stage IV cancer diagnosed by Pap or symptoms	Detected Stage IV, death from cervical cancer (over 5 years)
Stage IV survivor	5 years after initial diagnosis of Stage IV	Stage IV survivor, death from other cause

The model follows a cohort of women from age 15 to 85 and assumes that, at the beginning of the simulation, the prevalence of HPV infection is 10 percent and the prevalence of LSIL is 1 percent. Cycle lengths are 1 year long. Women infected with HPV can undergo regression, no change, or progression. Although most progress from HPV to LSIL, a proportion progresses directly to HSIL. Women with LSIL can undergo regression (to either "Well" or the HPV-infected state), no change, or progression to HSIL. Women with HSIL can regress, stay in the same state, or progress to Stage I cancer. Women with cancer either become symptomatic or progress through Stages II-IV. Once a cancer diagnosis is made, the probability of survival is stage-specific. Women without cancer are at risk for hysterectomy for other causes, and all women are at risk for death from other causes. The states and transitions of the natural history model are shown in Table 1.

Table 2. FIGO Staging of Cervical Cancer

Table 2. FIGO Staging of Cervical Cancer

Stage	Definition
	0 Carcinoma in situ (corresponds to CIN III or HSIL) Ia Confined to the cervix, detectable by microscopy only, depth of invasion not greater than 3 mm in depth or 7 mm in width Ib Confined to the cervix, larger than Ia IIa Involvement of upper 2/3 of vagina, no parametrial involvement IIb Involvement of parametria but not to pelvic sidewall IIIaInvolvement of lower 1/3 of vagina but not to pelvic sidewall if parametria involved
IIIb	Extension to pelvic sidewall and/or hydronephrosis or nonfunctioning kidney
IVa	Involvement of mucosa of bladder or rectum
IVb	Distant metastasis or disease outside of pelvis

Cervical cancer is staged clinically. The two most commonly used staging systems are the Federation Internationale de Gynecologie et Obstetriques (FIGO) staging system and the Tumor, Nodes and Metastases (TNM) staging system of the American Joint Committee on Cancer (AJCC). The two systems are compatible as described in the AJCC Manual (American Joint Committee on Cancer, 1997). The FIGO staging system is described in Table 2.

Because cervical cancer is clinically staged, findings at exploratory surgery or during subsequent treatment do not change the stage. Progression of disease also does not change the stage. In addition, imaging modalities such as computerized axial tomography (CAT) scanning or magnetic resonance imaging (MRI), while useful in determining the extent of the disease, are not permitted to be used in assigning stage.

The FIGO staging system is not used by the Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute, the source of the most reliable estimates of cervical cancer incidence and mortality. SEER reports cancer stages as "Local," "Regional," and "Distant." This classification is also used in the MEDSTAT data we used to generate cost estimates associated with cervical cancer. Although cervical cancer staging does not create as many difficulties in synthesizing data for the purposes of constructing a model as does the variety of classification systems for preinvasive cervical changes, there are still some issues that must be addressed.

First, the definition of Stage 0 has changed with time. Initially, this referred only to carcinoma in situ (CIS). Subsequently, when carcinoma in situ was included as CIN3, lesions that previously would not have been included as cervical cancer cases were classified as such at some SEER sites. Adaptation of The Bethesda System further increases the potential number of cases, since lesions previously classified as CIN2 are included along with CIN3 and CIS in the HSIL classification. Because of this, cases that would not meet FIGO criteria for invasive cancer may be included in SEER data (Noller, 1996).

Second, it is unclear whether the Local/Regional/Distant classification is based on clinical, radiological, or surgical findings. Cervical cancer spreads most commonly by local extension and/or by lymphatic spread. The likelihood of pelvic lymph node metastases increases with advancing FIGO stage (DiSaia and Creasman, 1997), but if the diagnosis of nodal metastases is made on the basis of surgery or imaging modalities such as CAT or MRI, the initial FIGO stage should not change.

Interventions

Screening strategies

Figure 3. Distribution of Pap results based on primary (more...)

   Figure 3. Distribution of Pap results based on primary and rescreening steps. The model assumes that the sensitivity of cytological screening using an ASCUS threshold is identical for all histological states

Figure 9. Distribution of cytological results for women (more...)

   Figure 9. Distribution of cytological results for women with the cancer. For a sensitivity of 51 percent, 49 percent of results after the primary screening step will be normal, 6.2 percent ASCUS, 3.7 percent LSIL, 10.6 percent HSIL, and 30.6 percent cancer

We compared no Pap smear screening to the following screening strategies: (a) conventional Pap smears at 1-, 2-, and 3-year intervals; (b) a technology that improves the initial screening step, increasing overall test sensitivity by 40, 60, or 90 percent at 1-, 2-, and 3-year intervals; and (c) a technology that allows 100 percent rescreening, again increasing overall test sensitivity by 40, 60, or 90 percent at the same screening intervals. Strategies for the evaluation of abnormal cytological results are discussed below. The intervention strategies are illustrated in Figures 3 - 9 . We compared screening with the different technologies at intervals of 1, 2, and 3 years, since this reflects current recommended maximum screening intervals in the United State (U.S. Preventive Services Task Force, 1996). Although several European countries recommend screening at 5-year intervals (Koopmanschap, van Oortmarssen, van Agt et al., 1990), no American organization recommends a screening interval of longer than 3 years.

We did not model combinations of screening intervals--for example, every 3 years after three consecutive negative annual Pap tests as recommended by the American Cancer Society (U.S. Preventive Services Task Force, 1996). Modeling such strategies significantly increases the complexity of the model. A model derived from age-specific incidence rates in unscreened populations suggests that the age at which screening occurs, rather than the interval between screens or the number of preceding negative smears, is the most important determinant of screening efficiency (Gustafsson and Adami, 1992). This is consistent with the natural history of the disease, where the incidence and prevalence of low-grade lesions that are unlikely to progress are high in younger ages, and high-grade lesions have peak incidence and prevalence between ages 30 and 45. Thus, three negative annual smears at age 18 or 20, just when HPV incidence is peaking, might have a low predictive value for the development of lesions 15-20 years later and would not significantly improve screening efficiency.

Measures of Effectiveness

We estimated the cost-effectiveness of cervical cytology using several outcomes. First, we calculated the cost per year of life saved. This allows comparison with other health interventions and with other cost-effectiveness analyses of cervical cancer screening. However, use of life expectancy, or life-years saved, may underestimate some of the benefits of screening. Early-stage cervical cancer is slow-growing and has a very high cure rate (approximate 85-90 percent 5-year survival for Stage I tumors). Thus, failure to detect preinvasive lesions that are then detected as Stage I tumors will not result in much decrement in life expectancy. However, the therapeutic options for all but the most minimally invasive Stage I tumors are radical hysterectomy or radiation therapy, both of which have substantial short- and long-term morbidity that may have significant impact on quality of life.

We were unable to identify any data that would allow calculation of quality-adjusted life years. We therefore calculated other outcomes that can be used to draw some inferences about morbidity. We produced estimates of cost per cervical cancer death prevented and per cervical cancer case prevented. In addition, by estimating the number of cases presenting in each stage and using data on the probability of specific treatments by stage, we also estimated the number of morbid therapies avoided using different screening strategies.

Assumptions of the Model

Parameter Estimates from Available Data: Natural History

The sources and rationale for specific transition probability estimates and ranges for sensitivity analysis are discussed in detail below. Some transition probabilities are based on large datasets using standard reporting methods (e.g., age-specific death or hysterectomy rates, or stage-specific mortality for cervical cancer); in such cases, estimation was relatively straightforward. However, some transition probabilities had to be estimated from smaller datasets that used nonstandard reporting; in these cases, it was difficult to estimate transition probabilities related to HPV and SIL from reports of incidence, regression, persistence, and progression rates probabilities (Miller and Homan, 1994). Regression and progression are usually reported as percentages of either prevalent or incident infections, with followup time varying widely among studies. In addition, very few data are available that allow calculation of transition probabilities for untreated cervical cancer, such as the annual stage-specific probability of symptom development or the annual probability of progression between stages. Estimates for these probabilities were derived from the literature and prior published models of cervical cancer screening, especially those of Eddy (1990) and Muller, Mandelblatt, Schechter et al. (1990), which have served as the basis for several subsequent models (Brown and Garber, 1998; Radensky and Mango, 1998; Schechter, 1996).

Human Papillomavirus (HPV) Infection

Although not essential for the current analysis, HPV is included in the model. In this section, we document this component of our cervical cancer screening model is documented.

Incidence of HPV infection

The model assumes that all cases of cervical cancer begin with infection with human papillomavirus. We have incorporated HPV status into the model for two reasons. First, although a small percentage of cervical cancers do not have detectable HPV DNA, even with sensitive assays, there is consensus that HPV infection is the causative agent for the vast majority of cervical cancers (Herrero, 1996; Kiviat, 1996; Koutsky, 1997; Schiffman, Bauer, Hoover et al., 1993). Second, certain HPV types are clearly more likely to progress to cancer than others, and identification of these types in cervical cells may have a role in determining optimal diagnostic and treatment strategies for patients with abnormal Pap smears (Cox, Lorincz, Schiffman et al., 1995). Although this role is still unclear, the current model can be adapted in the future to allow consideration of technologies such as HPV testing in screening for cervical cancer and its precursors. In addition, incorporating HPV data also allows our model to be used for assessing the cost-effectiveness of vaccines against HPV.

The natural history of HPV infection is complex, with clearance and persistence of viral DNA, along with progression to SIL, varying depending on the viral type, patient characteristics such as age and immune status, and study design and assay methods (Herrero, 1996; Kiviat, 1996; Koutsky, 1997; Mitchell, Tortolero-Luna, Wright et al., 1996; Schiffman, Bauer, Hoover et al., 1993). We do not distinguish between different types of HPV. Our incidence, progression, and regression estimates are averages for all viral types. The risk of developing cervical cancer after infection is clearly related to HPV type. With further data on the test characteristics of screening tests incorporating HPV typing, this model can be adapted for estimating the utility and cost-effectiveness of such tests.

Table 3. Age-specific Annual Incidence of HPV Infection (more...)

Table 3. Age-specific Annual Incidence of HPV Infection

Age	Incidence (percent per year)
15	0.1
16	0.1
17	0.12
18	0.15
19	0.17
20	0.15
21	0.12
22	0.1
23	0.1
24-29	0.05
30-49	0.01
50+	0.005

We used the following baseline age-specific estimates for HPV incidence in the model and varied them over ranges that, combined with our estimates of regression and progression, resulted in age-specific prevalence of HPV infection and age-specific incidence of cervical cancer within the ranges reported in the literature (Table 3). We varied the incidence rates by factors of 0.5-2 to examine the effects of changing HPV incidence on cervical cancer incidence in unscreened populations. We also varied the age of peak incidence from 20 to 30 in order to examine the effects of a later onset of sexual activity on age-specific cervical cancer incidence.

Regression, persistence, and progression of HPV infection

Estimates of regression and progression rates of HPV infection are subject to variability in study design, patient population, and viral assay techniques and are complicated by the mathematical difficulty of converting rates to probabilities. Reported regression rates for prevalent cases include 70 percent after 2 years in a cohort of adolescents and college-age women (Moscicki, Shiboski, Broering et al., 1998), 68 percent over 14 months for women under 25, and 35 percent for women over 30 (Hildesheim, Schiffman, Gravitt et al., 1994). Ho, Bierman, Beardsley et al. (1998) reported a 1-year regression rate of 70 percent. All of these results were in women with normal cytology. The overall regression rate in a large Finnish cohort was 42.8 percent over 50 months (Kataja, Syrjanen, Mantyjarvi et al., 1989; Kataja, Syrjanen, Syrjanen et al., 1990). However, determination of disease status in this group was made on the basis of cytological evidence of HPV infection. Under the Bethesda System, these would be classified as LSIL.

As with incidence, regression rates and possibly progression rates appear to vary with age. Whether this is truly related to age or is a function of duration of infection (the longer HPV persists, the less likely it is to regress) is unclear. Again, this question can be further explored with the current model. For the purposes of a cohort simulation, there is not much difference between an age-based or duration-based progression rate as long as the age-specific incidence is constant.

Table 4. Regression and Progression Rates of HPV Infection (more...)

Table 4. Regression and Progression Rates of HPV Infection

Rate	Estimate	Range
Regression:
Age 15-24	0.7/18 months	0.60-0.9
Age 25-29	0.50/18 months	0.45-0.6
Age 30-39	0.25/18 months	0.10-0.2
40-49	0.15/18 months	0.10-0.2
Age 50 +	0.05/18 months	0.05-0.2
Progression:	0.2/36 months	0.15-0.3
Proportion progressing to HSIL directly	0.1	0.05-0.5

Progression rates are equally difficult to determine. Moscicki, Shiboski, Broering et al.(1998) reported progression to SIL in 15.7 percent over 40 months, and 7.6 percent of these progressed directly to HSIL. Ho, Bierman, Beardsley et al. (1998) reported a cumulative 36-month incidence of SIL of approximately 25 percent, and 6.5 percent of these were high-grade lesions. Koutsky, Holmes, Critchlow et al. (1992) found a 2-year cumulative incidence of HSIL of 28 percent in women with HPV DNA, and only 36 percent of these had had a prior LSIL smear. Consistent with our other estimates, our base case estimates (Table 4) are derived from the higher end of the reported ranges. Cases that progress directly to HSIL are similar to the "rapidly progressive" cases of the Eddy (1990) model. Again, we varied age-specific regression rates in order to achieve an age-specific cancer incidence curve similar to that seen in unscreened populations.

Because there is some uncertainty in the literature about whether treatment of SIL results in total eradication of HPV, we created a Diagnosed HPV state to represent women who have been treated for SIL. We assumed that the probability of progression was 5 percent of the age-specific progression rates. This is equivalent to a 95 percent cure rate for SIL treatment used in other models (Eddy, 1990; Fahs, Mandelblatt, Schechter et al., 1992; Schechter, 1996).

Low-grade and High-grade Squamous Intraepithelial Neoplasia

Table 5. Progression and Regression Estimates for LSIL (more...)

Table 5. Progression and Regression Estimates for LSIL and HSIL

	Estimate	Range
LSIL
Regression
Age 15-34	0.65/72 months	0.6-0.8
Age 35 +	0.4/72 months	0.3-0.6
Progression
Age 15-34	0.1/72 months	0.1-0.3
Age 35 +	0.35/72 months	0.3-0.5
HSIL
Regression	0.35/72 months	0.3-0.5
Progression	0.4/120 months	0.3-0.5

Determining transition probabilities from the literature that accurately reflect natural history is as problematic for SIL as it is for HPV infection. The difficulties in converting rates collected over varying, often unspecified times and in heterogeneous populations are further magnified by differences in terminology. For example, many studies report transitions from HPV-associated cytologic changes to CIN 1 to CIN 2 to CIN 3 to CIS, which may be difficult to translate into the Bethesda System terminology of LSIL and HSIL. For our base case, we use the estimates of Syrjanen, Kataja, Yliskoski et al. (1992), the largest cohort that reports results using the Bethesda System. We assume an age dependence in regression and progression rates (Hildesheim, Schiffman, Gravitt et al., 1994; Munoz, Kato, Bosch et al., 1996; van Oortmarssen and Habbema, 1991). The length of time for HSIL progression is more difficult to estimate from these data than the other parameters; in this cohort, all patients who progressed to carcinoma in situ under an older classification system were treated. Prior models have estimated the duration from severe dysplasia/CIS to invasive cancer of 10-15 years. We use 12 years for the base case (Table 5), since this interval resulted in an age-specific incidence of cervical cancer most consistent with observed data.

Conditional Probabilities of Cytologic Results for Given Histologic States

Because the model is based on transitions from true histological states of LSIL, HSIL, and invasive cancer and Pap smear results are categorical, the probability of a particular cytological diagnosis for a given histological state requires estimates beyond "sensitivity" and "specificity" for cytology.

In the model, any Pap smear result of ASCUS or above is considered abnormal. For patients without histological changes (Well, Unknown HPV Infection, and Detected HPV Infection), the probability of a normal smear will be the specificity of the Pap smear or other test for a cytological threshold of ASCUS or above, since specificity is defined as the probability of a normal test result given no pathology, whereas the probability of an abnormal result, or false positive, is 1.0 minus the specificity.

For patients with histological changes (LSIL and higher), our base case estimate of the overall probability of an abnormal result is equal to our sensitivity estimate, using an ASCUS threshold, from the three studies by Baldauf, Dreyfus, Lehmann et al. (1995); Davison and Marty (1994); and Hockstad (1992). This estimate was 51 percent. We varied the sensitivity up to 78 percent, the mean estimate of the meta-analysis.

For the base case for conventional Pap smears, we used an estimate of specificity based on the three studies of the use of conventional Pap smear in primary screening that were unaffected by verification bias (Baldauf, Dreyfus, Lehmann et al., 1995; Davison and Marty, 1994; Hockstad, 1992). This estimate is 97 percent, which we varied down to 0.88 percent, the mean specificity estimate for all of the studies included in the meta-analysis. For the base case, we assumed that new technologies would have no impact on specificity. We varied the relative specificity of the new technologies from a 50 percent increase in specificity to a 50 percent decrease in specificity.

Figure 4. Effect of a new technology that improves the (more...)

   Figure 4. Effect of a new technology that improves the sensitivity of the primary screening step on the distribution of normal and abnormal Pap results. The model assumes that 10 percent of slides intially read as normal are rescreened at the sensitivity and specificity of the new technology.

Figure 5. Effect of a new technology that improves (more...)

   Figure 5. Effect of a new technology that improves the sensitivity of the rescreening step on the distribution of Pap results. 100 percent of slides that are read as normal by conventional Pap are re-read at the sensitivity and specificity of the new technology

Figure 6. Distribution of Pap results when underlying (more...)

   Figure 6. Distribution of Pap results when underlying histology is normal(including patients with HPV infection but no cytological or histological change) (Jones and Novis, 1996). If specificity of conventional Pap is 97 percent, then 97 percent of cytology results will be normal, 3 percent X 52.5 percent = 1.6 percent will be ASCUS, 1.2 percent LSIL, 0.3 percent HSIL, and 0.01 percent cancer after the primary screening step

Figure 7. Distribution of specific abnormal Pap results (more...)

   Figure 7. Distribution of specific abnormal Pap results when underlying pathology is LSIL. If sensitivity is 51 percent, then 49 percent of patients with LSIL will have initially Paps, 51% X 23.5% = 12.0 percent will have ASCUS, 34.8 percent LSIL, 4.1 percent HSIL, and 0.08 percent cancer as the cytological diagnosis.

Figure 8. Distribution of cytological resutls for underlying (more...)

   Figure 8. Distribution of cytological resutls for underlying histological diagnosis of HSIL. At sensitivity of 51 percent, 49 percent of Paps will be normal, 5.3 percent will be ASCUS, 15.3 percent LSIL, 29.9 percent HSIL, and 0.56 percent cancer.

Abnormal results on Pap testing include ASCUS, LSIL, HSIL, and cancer. For each histological state, we calculated the proportion of all abnormal results represented by each possible cytological diagnosis using data from a large data set of biopsy-cytology pairs (Jones and Novis, 1996). For our model, we collapsed the categories as follows: ASCUS and AGUS = ASCUS (this is not meant to imply comparability in the relative clinical significance of these two diagnoses, but rather only to simplify the calculations, as discussed below), indeterminate SIL and LSIL = LSIL, any malignancy or glandular carcinoma-in-situ = cancer. Because data correlating ThinPrep^®, AutoPap^®, or Papnet^® results with histological results were limited, we assumed that these proportions are similar for new technologies. We present these proportions graphically in Figures 3 - 9

We assumed that the sensitivity of the screening test is independent of the underlying histology. In other words, the sensitivity using an ASCUS threshold is identical for LSIL, HSIL, or cancer. This assumption may not be true. This is a complex issue that would be difficult to model in a useful way. In keeping with other models, we have assumed constant sensitivity and specificity across histological states. Misclassification of cases, whether to a higher or lower stage, could significantly alter procedure or cost parameters.

Natural History of Invasive Cancer

Table 6. Progression Rates for Cervical Cancer Stages (more...)

Table 6. Progression Rates for Cervical Cancer Stages I through IV

Cancer Stage	Progression Rate	Annual Probability of Symptoms	Proportion of Cases in Unscreened Population Predicted by Model	Proportion Reported in Literature
I	0.9/4 years	0.15	0.463	0.34-0.55
II	0.9/3 years	0.225	0.270	0.13-0.28
III	0.9/1.25 years	0.6	0.181	0.12-0.19
IV	N/A	0.9	0.085	0.06-0.10

As stated above, there are almost no data for estimating the rates of progression from Stage I through Stage IV cervical carcinoma. There is also no way to determine the likelihood of developing symptoms. Since the distribution of cases by stage in an unscreened population should be a function of the progression rate and the likelihood of symptoms (since cases would only be detected upon presentation with symptoms), we have adopted the approach taken by others (Eddy, 1990; Fahs, Mandelblatt, Schechter et al., 1992) and adjusted these estimates so that the proportion of cases represented by each stage is similar to that seen in cervical cancer patients who have never been screened (Bearman, MacMillan, and Creasman, 1987; Janerich, Hadjimichael, Schwartz et al., 1995; Pretorius, Semrad, Watring et al., 1991; Schwartz, Hadjimichael, Lowell et al., 1996). Our base case estimates are described in Table 6.

Stage-Specific Treatment Probabilities

Table 7. Stage-specific Treatment Probabilities

Table 7. Stage-specific Treatment Probabilities

Treatment	Stage I	Stage II	Stage III	Stage IV
Simple hysterectomy	0.146	0.006	0.004	0.003
Radical hysterectomy	0.355	0.043	0.013	0.009
Radiation therapy	0.204	0.647	0.537	0.398
Radiation therapy and surgery	0.146	0.072	0.046	0.017
Radiation therapy and chemotherapy	0.022	0.161	0.320	0.307
Chemotherapy only	0.002	0.004	0.004	0.103
Other therapy	0.038	0.028	0.027	0.020
No Therapy	0.087	0.039	0.049	0.143

The proportion of women in each stage of cervical cancer who received a specific treatment was estimated using 1990 data from the American College of Surgeons' Patient Care Evaluation (PCE) study for cervical cancer (A. Fremgen, personal communication to E. Myers, 1998; Jones, Shingleton, Russell et al., 1995; Russell, Shingleton, Jones et al., 1996). Table 7 shows the stage-specific data. Of note, 21.8 percent of the patients in this dataset had an unknown stage. Of those, almost one-half were treated with either a simple or radical hysterectomy, suggesting that most were earlier stage disease.

These proportions were used to calculate the number of specified procedures resulting from each diagnosed case of a specific stage of cervical cancer predicted by the model. These proportions were also used to generate estimates of stage-specific treatment costs (see below).

Parameter Estimates from Available Data: Test Characteristics

We derived our estimates for conventional Pap sensitivity and specificity from the systematic literature review and meta-analysis. For sensitivity and specificity of the new technologies, we used articles reviewed under the revised criteria, the review of Brown and Garber (1998), and estimates provided by the manufacturers. The derivation of these estimates is described in greater detail below.

Sensitivity Estimates for Initial Screening Step

Table 8. Calculations of Reduction in False Negative (more...)

Table 8. Calculations of Reduction in False Negative Rate for Initial Screening Technologies

Source	Thin Prep® Sensitivity	Conventional Pap Sensitivity	Calculated Reduction in False Negative Rate
Roberts, Gurley, Thurloe et al (1997)	Relative TPR 1.13	NA	0.91 (assumes same sensitivity and specificity as Bolick and Hellman, 1998)
Bolick and Hellman, (1998)	0.94	0.85	0.6
Brown and Garber (1998)	0.91	0.8	0.55

One way to improve Pap test sensitivity is to improve the sensitivity of the initial screening step, either through reducing potential sampling errors and interpretive errors, such as monolayer preparations, or through interpretive errors only, such as computerized technologies. We modeled an improvement in sensitivity by using the reduction in false negative rate (see Figures 3 - 5). As mentioned previously, we assumed that 10 percent of normal smears would be rescreened at the same sensitivity and specificity as primary screening. As a starting point, we used data from studies of ThinPrep^® included in our meta-analysis as well as estimates of reduction in FNR used by Brown and Garber (1998) (Table 8).

We therefore use 0.6 as our base case reduction in FNR for the initial primary step. For sensitivity analysis, we varied the estimate of reduction in FNR from 0.4 to 0.9. This range is consistent with the "50 percent increase in detection of disease" cited by the manufacturer (R.Silverman, personal communication to D.C. McCrory, 1998 ).

Because these values are calculated relative to the sensitivity of the conventional Pap testing using adjudicated review, they should, in theory, be independent of the actual sensitivity of the conventional Pap testing. In fact, this argument is the justification for not requiring histological confirmation studies for FDA approval of these technologies. Therefore, in our model, the base case sensitivity of the initial screening technology will be

0.51 + (0.6 * 0.49) = 0.804.

We assume that 10 percent of slides will be rescreened at the same improved sensitivity, for an overall sensitivity of

0.51 + (0.6 * 0.49) + (0.1 * 0.6 * 0.49) = 0.833.

It is important to note that these estimates do not distinguish test sensitivity for screening and rescreening, nor do they differentiate sampling error and interpretive error. Test performance is likely to be different in a rescreening application, since the population subjected to rescreening is different from the primary screening population. For practical purposes, if sensitivity is improved sufficiently to make a new screening technology cost-effective, then it should not matter if that differential is due to technological superiority or difference in the type of error that is reduced. However, to examine the impact of this difference, we model differential sensitivities between initial and rescreening testing. Regarding different types of test errors, sampling errors may result from either poor technique on the part of the person obtaining the smear, or placement of cells on the slide which may not be representative of the potential areas of abnormality. Determining the relative effect of errors in technique (inadequate visualization of the cervix, inadequate sampling of the transformation zone, failure to clear off blood and mucous prior to obtaining the smear) versus errors in cell sampling from the collection device itself would require more histological confirmation than is currently available. Again, to permit examination of this issue, we model the effect of changes in differential sensitivities between initial and rescreening technologies.

Sensitivity Estimates for Rescreening Step

We also examined the potential impact of technologies that improve the sensitivity of the rescreening step. We assumed that the sensitivity of the primary screening step was that of the conventional Pap and that 100 percent of smears read as normal would be rescreened at the improved sensitivity and specificity (see Figure 5). We used a similar approach to the one described above.

For rescreening technologies, the overall improvement in sensitivity can be calculated as either

Sensitivity Conventional Pap + {1.0 * [Sensitivity Conventional Pap +x (1-Sensitivity
Conventional Pap)] * (1-Sensitivity Conventional Pap)}

or as

Sensitivity Conventional Pap + [1.0 * Sensitivity Rescreening Device * (1-Sensitivity
Conventional Pap)].

Table 9. Calculation of Reduction in False Negative (more...)

Table 9. Calculation of Reduction in False Negative Rate for Rescreening Technologies

Source	Rescreening Sensitivity	Calculated Reduction in False Negative Rate
AutoPap® (Rescreening Only)
Stevens, Milne, James et al. (1997)	0.43	0.43
Patten, Lee, Wilbur et al. (1997a)	0.51	0.51
Colgan, Patten, and Lee (1995)	0.663	0.663
Brown and Garber (1998)	0.77	0.77
Papnet®
Kaufman, Schreiber, and Carter (1998)	0.38-0.41	0.38-0.41
Jenny, Isenegger, Boon et al. (1997)	0.89	0.89
Brown and Garber (1998)	0.88	0.88

Table 9 shows the sources and calculated reduction in false negative rate used to derive our estimated range for rescreening technologies.

Again, we chose 0.6 as the base case for the rescreening technology and varied it between 0.4 and 0.9. The base case estimate was chosen because it was in the mid-range of estimates. By beginning at identical values, identification of thresholds between the two types of technologies was simplified.

Parameter Estimates from Available Data: Cost

Cost estimates for cervical cancer screening and for the diagnosis and treatment of LSIL, HSIL, and invasive cervical cancer were derived primarily from the review of MEDSTAT data conducted by Health Economics Research, Inc.

Costs of Screening

Table 10. Costs of Screening Tests

Table 10. Costs of Screening Tests

Test	Base Case	Range	Medicare Cost
Conventional Pap, 20-64	$38.68	$35.32-43.7
Conventional Pap, 65+	$47.73	$44.24-56.98	$35.01 (range: $34.55-36.92)
New technology which improves primary screening sensitivity	Incremental cost: $10.00	$5.00-15.00	Not available
New technology which improves rescreening sensitivity	Incremental cost: $10.00	$5.00-15.00	Not available

Estimates for the incremental cost per slide for conventional Pap smears, Paps smears obtained using ThinPrep^® media, and Paps with 100 percent rescreening with AutoPap^® or Papnet^® were assigned on the basis of data from Health Economics Research, Inc., and estimates made by Brown and Garber (1998), Schechter (1996), Radensky and Mango (1998), peer reviewers of the draft evidence report, and the manufacturers. These costs are summarized in Table 10.

Costs of Diagnostic Evaluation for LSIL, HSIL, and Invasive Cancer

Diagnostic costs were estimated in two ways. First, the mean cost for diagnosis and treatment of LSIL, HSIL, and Stages I, II, III, and IV cervical cancer from MEDSTAT data was used to assign a global diagnosis and treatment cost to each level of pathology. Sensitivity analyses were run using the 25^th and 75^th percentiles as the minimum and maximum values. The MEDSTAT dataset did not contain information specific to FIGO stage, only World Health Organization (WHO) stage (Local, Regional, Distant). We assumed that "local" disease equaled Stage I, "regional" equaled Stage II or III, and "distant" equaled Stage IV.

Table 11. Procedure-specific Costs for Diagnostic and (more...)

Table 11. Procedure-specific Costs for Diagnostic and Treatment Procedures

Diagnostic Procedure	Base Case	Range	Medicare Cost
Colposcopy + biopsy + endocervical curettage	$375	$244-477	$244
Colposcopy + biopsy only	$276.57	$195-351	$172.63
LEEP	$564	$352-710	$205
Cone biopsy	$919	$623-1114	$409

The model was also evaluated using procedure-specific costs for diagnostic and treatment procedures. Means and 25^th and 75^th percentiles from the MEDSTAT data were used to establish base estimates and ranges for colposcopy and biopsy, LEEP, and cone biopsy (Table 11).

Table 12. Proportion of Patients in Each Stage Undergoing (more...)

Table 12. Proportion of Patients in Each Stage Undergoing Specified Diagnostic Procedures

Procedure	Average Cost	Stage I	Stage II	Stage III	Stage IV
Examination under anesthesia	$201	0.414	0.458	0.462	0.462
Cystoscopy	$982	0.201	0.388	0.454	0.454
Proctoscopy	$774	0.152	0.281	0.316	0.316
Chest X-ray	$43	0.727	0.781	0.779	0.779
Intravenous pyelogram	$339	0.313	0.327	0.311	0.311
Barium enema	$108	0.136	0.199	0.213	0.213
Bone scan	$236	0.045	0.112	0.136	0.136
CT scan	$312	0.408	0.696	0.768	0.768
MRI scan	$563	0.046	0.067	0.071	0.071
Lymphangiogram	NA	0.013	0.032	0.039	0.039
Fine needle biopsy	NA	0.017	0.035	0.042	0.042

NA = not available.

Source: Russell, Shingleton, Jones et al., 1996

The diagnostic costs associated with each stage of cervical cancer were estimated by calculating a weighted average cost using the proportion of patients within each stage who underwent a given diagnostic procedure in 1990 (Russell, Shingleton, Jones et al., 1996) and procedure-specific MEDSTAT costs (Table 12).

Table 13. Stage-specific Cost Estimates of a Diagnostic (more...)

Table 13. Stage-specific Cost Estimates of a Diagnostic Workup

Cervical Cancer Stage	Base Case	Range	Medicare Costs
I	$714	$274-1042	$348
II	$1138	$429-1642	$520
III	$1257	$470-1837	$563
IV	$1257	$470-1837	$563

The base case estimates (Table 13) were derived using the average MEDSTAT costs, with lower and upper estimates calculated using the 25^th and 75^th percentiles. Similar estimates were made based on Medicare costs.

Because MEDSTAT data did not include professional charges for anesthesia for pelvic examination under anesthesia, or charges for lymphangiography or fine needle biopsy, the diagnostic costs are likely to underestimate the true average costs of diagnosis of cervical cancer to some extent.

Costs of Treatment

Table 14. Global Costs of Diagnosis and Treatment

Table 14. Global Costs of Diagnosis and Treatment

Pathological Findings	Base Case	Range
LSIL	$1728	$675-2274
HSIL	$3049	$1384-4203
Stage I	$17,645	$9,439-21,946
Stage II-III	$27,069	$11,734-29,089
Stage IV	$40,280	$12,670-46,509

Like diagnostic costs, treatment costs were estimated in two ways. Global diagnostic and treatment costs estimated from MEDSTAT data (Table 14) were one source.

Table 15. Procedure-specific Costs

Table 15. Procedure-specific Costs

Therapeutic Procedure	Base Case	Range	Medicare Costs
LEEP	$564	$352-710	$205
Cryotherapy	$156	$108-184	$111
Laser ablation	$730	$390-976	$205

Procedure-specific costs were also calculated and used to generate a second set of estimates (Table 15). Costs for procedures used in the treatment of SIL were obtained directly from MEDSTAT and Medicare data.

As noted elsewhere in the report, mean costs for cervical cancer treatment were often substantially higher than median costs, especially for more advanced procedures. This probably represents the effects of varying severity of disease within stages or of complications of treatment, which substantially increase costs. We have chosen to use mean costs for our base case estimate in order to include the economic impact of varying disease severity, comorbidities, and complications. Because of the extremely wide standard deviations resulting from the effect of outliers, we use the 25^th and 75^th percentiles as endpoints in sensitivity analysis.

Table 16. Number of Inpatient and Outpatient Treatment (more...)

Table 16. Number of Inpatient and Outpatient Treatment Procedures

Treatment	Number of Inpatient Procedures	Number of Outpatient Procedures
Simple hysterectomy only	1	0
Radical hysterectomy only	1	0
Surgery + radiation	1 Radical hysterectomy or 1 Simple hysterectomy + 2 radiation treatments	20 outpatient
Radiation + chemotherapy	2 radiation treatments and 0 inpatient chemotherapy, or 6 inpatient chemotherapy	6 outpatient (if no inpatient)
Chemotherapy	6	6 (if no inpatient)
Radiation therapy	2	20

Estimates of the proportion of patients undergoing a specified treatment in each stage were calculated from the American College of Surgeons data (see section on Stage-specific Treatment Probabilities above). Because the data were not specific about the exact radiation therapy or chemotherapy regimens used, we assumed that patients received standard regimens of each, which are described in Table 16 (A. Berchuk and D.L. Clarke-Pearson, personal communication to E. Myers, 1998; DiSaia and Creasman, 1997).

The cost estimates resulting from these assumptions about treatment regimens for each stage were then multiplied by the proportion of patients in each stage receiving each treatment in the American College of Surgeons PCE dataset to arrive at the following weighted averages for treatment of cervical cancer (Tables 17 and 18):

Table 18. Procedure-specific Costs for Diagnosis and (more...)

Table 18. Procedure-specific Costs for Diagnosis and Treatment Compared with MEDSTAT Global Estimates

Stage	Procedure-Specific Estimates			MEDSTAT Global Estimate
	Diagnosis	Treatment	Total
I	$714	$12,243	$12,957	$17,645
II	$1138	$15,244	$16,382	$27,069
III	$1257	$15,244	$16,501	$27,069
IV	$1257	$17,431	$18,688	$40,280

Table 17. Stage-specific Costs for Treatment (based (more...)

Table 17. Stage-specific Costs for Treatment (based on MEDSTAT procedure-specific cost data, standard treatment regimens, and distribution of treatments in American College of Surgeons dataset)

Stage	Base Case	Range	Medicare
I	$12,243	$7320-15185	$7724
II	$15,244	$7196-21,660	$10,105
III	$15,244	$7526-25,144	$10,159
IV	$17,431	$9296-34,365	$8,001

The mean MEDSTAT global estimates (see Table 18) are substantially higher than the estimates derived by applying procedure-specific MEDSTAT costs to the distribution of treatments reported in the American College of Cancer data set (see Table 17). This may reflect differences in patient populations, professional charges, changes in practice patterns between 1990 and 1994, or additional costs, such as antiemetic treatment for chemotherapy patients, or followup outpatient clinic visits, not captured using procedure-specific costs. Again, to bias the model in favor of more sensitive screening strategies, we used the global costs for the base case estimate. In the base case, we also did not use the lower Medicare estimates for patients over 65, again to deliberately bias the results in favor of more sensitive screening strategies.

Calibration Against Epidemiological Data

Epidemiological data on the incidence and prevalence of cervical cancer and its precursors (LSIL, HSIL, and HPV infection) were used to calibrate the cost-effectiveness model.

Converting reported incidence, regression, persistence, and progression rates for HPV and SIL to reliable transition probabilities for the model was difficult because of the way most of these values are reported in the literature. Regression and progression rates are usually reported as percentages of either prevalent or incident infections, but followup time varied widely among studies. Data are rarely provided in a way that easily allows conversion of these rates to yearly transition probabilities (Miller and Homan, 1994). In addition, very few data are available that allow calculation of transition probabilities for untreated cervical cancer, such as the annual stage-specific probability of symptom development or the annual probability of progression between stages. Estimates for these probabilities were derived from the epidemiological literature and prior published models of cervical cancer screening, especially those of Eddy (1990) and Muller, Mandelblatt, Schechter et al. (1990).

Figure 10. Predicted age-specific incidence of cervical cancer (more...)

   Figure 10. Predicted age-specific incidence of cervical cancer

Figure 10

Gustafsson, Ponten, Bergstrom et al. (1997) described two separate curves, one with a peak between ages 40 and 50 with a more rapid decline, and one similar to the U.S. data with a peak between ages 50 and 65 and a more gradual decline. Their modeling suggests that some of the difference results from increased incidence in successive birth cohorts. The early peak curves primarily were observed in western European countries with data collected predominantly between 1950 and 1975. Conversely, the U.S. data comes from Connecticut between 1935 and 1949. It is likely that early sexual activity, and therefore exposure to HPV, was more frequent in western Europe and Scandinavia during the 1950s and 1960s than in 1940s Connecticut. In addition, there are likely to be differences in access to care, or willingness to seek care for symptoms, across time and cultures.

Figure 11. Change in peak HPV prevalence

   Figure 11. Change in peak HPV prevalence

Figure 12. Effect of peak age of HPV on cervical cancer incidence (more...)

   Figure 12. Effect of peak age of HPV on cervical cancer incidence

Figure 13. Effect of age of peak HPV incidence and (more...)

   Figure 13. Effect of age of peak HPV incidence and delay in seeking medical attention on cancer incidence

Figure 11

Figure 12

Figure 13

We elected to use values resulting in an earlier onset for the base case scenario. Prevention of fatal disease occurring at younger ages results in greater increases in life expectancy compared with prevention of disease at older ages; thus, our base case model is biased in favor of screening strategies that have increased sensitivity.

The peak incidence of cervical cancer used in the model (81/100,000) is somewhat higher than that observed in an unscreened U.S. population in the 1930s (approximately 60/100,000) but lower than that seen in developing nations (Gustafsson, Ponten, Bergstrom et al., 1997). Because there have been significant increases in sexual activity in adolescents over time (Centers for Disease Control and Prevention, 1991), it seems likely that the actual expected rate in an unscreened population would be higher than the rate observed 60 years ago, especially since mortality from other causes, particularly in younger women, has declined. This higher incidence estimate is consistent with our choice of other base case estimates for the model, in that we have tried to bias the model in favor of screening strategies with increased sensitivity compared with conventional Pap smears at longer screening intervals.

In an unscreened population, age-specific incidence of cervical cancer will be dependent both on the rate of progression of the disease and the stage-specific probability of developing symptoms that lead to diagnosis. Very few data are available to validate parameter estimates for rates of progression between stages of cancer or the probability of developing symptoms in any given stage. Therefore, these parameters were adjusted to result in a distribution of cases by stage that approximates that seen in patients without Pap smear screening (Bearman, MacMillan, and Creasman, 1987; Eddy, 1990). These proportions are also similar to those observed in larger series of cervical cancer patients (Jones, Shingleton, Russell et al., 1995; Pretorius, Semrad, Watring et al., 1991), presumably because at least 30-40 percent of cervical cancer patients have never been screened or have not been screened within the past 5 years.

Internal Quality Control Mechanisms

Quality control procedures were integrated into each step of the literature review. The search strategies used were checked and finalized by two professional medical librarians. For additional checks on completeness, the articles cited in reference lists of reviewed articles, particularly meta-analyses, were compared with those in the Pro-Cite literature database developed for this evidence report, and when appropriate, articles were copied, reviewed, and added to the Pro-Cite database. Further, the manufacturers of AutoPap^®, Papnet^®, and ThinPrep^® were asked to submit articles and other documents that met the specified inclusion criteria for the evidence report.

With regard to the content of the evidence tables and data analyses, quality control procedures included: screening of all articles by at least two clinicians, abstracting of information into the evidence report by two clinicians, and independent construction of each 2-by-2 table by two clinicians and over-reading by a third. No data or other information was inserted into the evidence table or data analyses without absolute agreement by two, and often three, clinicians. Further, once the content was agreed upon by the clinicians, the information for the evidence table and data for the meta-analysis and cost-effectiveness analysis were double-entered and compared, and any discrepancies reconciled. Where appropriate, kappa statistics were calculated to show the strength of agreement between clinician-reviewers, particularly in the article screening stage and the construction of the 2X2 tables. When the strength of agreement was low, all reviewers met to discuss the nature of the discrepancies and to come to common agreement on the interpretation of the criteria and their application to the articles as well as agreeing on responses to articles that were in a "gray" area. In the latter case, the decision was to err on the side of including the article, recognizing that the data abstraction process and the development of the 2X2 tables would further illuminate the inclusion/exclusion status of the article.

Quality control in the cost-effectiveness analysis was provided by extensive sensitivity analyses, by validation with epidemiological data, and by comparisons with previously published cost-effectiveness models.

External Quality Control Mechanisms

Two external oversight and review panels were established. The Advisory Panel of Technical Experts was initiated early in the 11-month time frame for the study. The Advisory Panel of Technical Experts reviewed progress on the evidence report at four key stages of its development: the identification of the key literature search questions, the first drafts of the evidence tables, the plans for the cost-effectiveness analysis and meta-analysis, and the draft of the evidence report. The Technical Experts were sent draft documents that were discussed as a group in two conference calls and also in individual phone calls and written communications. The eight-member Panel consisted of clinical and methodological experts in relevant specialty areas, including clinical and laboratory pathology, obstetrics/gynecology, family practice, oncology, women's health, meta-analysis, cost-effectiveness analysis, and epidemiology. Consumers were represented on the Advisory Board of Technical Experts by Family Health International, a not-for-profit organization committed to improving the health of women and children; helping women and men have access to safe, effective, acceptable, and affordable family planning methods; and preventing the spread of AIDS and other sexually transmitted diseases (STDs).

The primary function of the second group, the Peer Review Panel, was to review and comment on the complete draft of the evidence report. The 23-member panel, which included AHCPR staff and the Advisory Panel of Technical Experts, consisted of clinical and methodological experts in relevant specialty areas, including clinical and laboratory pathology, obstetrics/gynecology, family practice, oncology, women's health, meta-analysis, cost-effectiveness analysis, and epidemiology, as well as the manufacturers of the Pap screening devices reviewed. The report was modified on the basis of their comments, with close attention to relevant studies not included in the draft report, misinterpretation of findings, and other issues deserving revision within the constraints of the methodology, time frame, and budget. The format for the report was designed by AHCPR.

New Technologies

Because very few studies of the new technologies met the original Step 1 and Step 2 screening criteria (no studies of AutoPap^®, one study of Papnet^®, and one study of ThinPrep^®), we modified the criteria to include studies pertaining to any of the new technologies that used a cytological reference standard if applied by an independent panel and if at least half of high-grade cytological results were verified histologically, as suggested in guidelines produced by the Intersociety Working Group for Cytology Technologies (1998). We further modified the screening criteria to include studies that failed to verify diagnosis in patients negative on two independent tests being compared. Though such studies do not estimate sensitivity and specificity, they can still provide estimates of relative TPR and relative FPR, as described by Chock, Irwig, Berry et al. (1997).

We considered a total of 59 studies (12 on AutoPap^®, 27 on Papnet^® and 20 on ThinPrep^®) during this final stage of the screening process (Step 3). Forty-three of these articles had initially been excluded in Step 2, and 16 were brought to our attention at this time by reviewers or manufacturers (7 unpublished manuscripts and 2 published only in abstract form). Ultimately, we decided to describe in the evidence table and the text of this report all the studies considered for Step 3 that also met the criteria requiring the application of a concurrent reference standard (histology, colposcopy, or cytology). The net result was the inclusion of 6 studies of AutoPap^®, 11 of Papnet^®, and 8 of ThinPrep^®.

Six studies of either AutoPap 300 QC^® or the AutoPap Primary Screening System^® permitted estimates of sensitivity. Little information was available on which to base an estimate of the specificity of rescreening. In three studies of the performance of AutoPap 300 QC^® in slides manually screened as negative, the estimated sensitivity for detecting ASCUS or worse at the 20 percent review rate was 0.43 (Stevens, Milne, James et al., 1997), 0.51 (Patten, Lee, Wilbur et al., 1997a), and 0.663 (Colgan, Patten, and Lee, 1995). The estimated sensitivity for detecting LSIL or worse at the 20 percent review rate was 0.66 and 0.77 in the latter two studies. Studies suggested that sensitivity at the 20 percent review rate for ASCUS or worse was higher when these technologies were used for screening than when they were used for rescreening: 0.86 (Wilbur, Bonfiglio, Rutkowski et al., 1996) for the AutoPap 300 QC^®, and 0.86 (Wilbur, Prey, Miller et al., 1998) and 0.92 (Lee, Kuan, Oh et al., 1998) for the AutoPap Primary Screening System^®. The estimated sensitivity for detecting LSIL or worse at the 20 percent review rate was 0.92 (Wilbur, Prey, Miller et al., 1998). An estimate of specificity of 0.98 was obtained for screening use of the AutoPap Primary Screening System^® (Lee, Kuan, Oh et al., 1998); however, this estimate was based on initial screening rather than rescreening. Because the spectrum of disease is different in patients whose smears are initially manually screened as negative compared with patients undergoing initial screening, this specificity estimate may be inaccurate. Furthermore, the Primary Screening System uses a somewhat different algorithm for classification than does the AutoPap 300 QC^® and therefore could be expected to have a different specificity.

Conventional Pap Test

Table 20. Quality of Selected Studies

Table 20. Quality of Selected Studies

Quality Criterion	Number of Studies (N = 84)	Percent
Reference standard
Histology	70	83
Histology or negative colposcopy	14	17
Independence of assessments Blinded Not blinded	22 62	26 74
Verification All test positives and test negatives verified Test positives and random fraction of test negatives verified Test positives and selected sample or none of test negatives verified	23 3 58	27 4 69
Study sample Consecutive or random Other	73 11	87 13
Spectrum of disease/nondisease Defined Not defined	9 75	11 89
Publication type Paper Abstract	84 0	100
Industry relationship Neither done nor supported by a manufacturer Supported by a manufacturer Done by a manufacturer	82 1 1	98 1 1

Data were available according to different combinations of test thresholds and reference standard thresholds: ASCUS/CIN1 (31 studies), LSIL/CIN1 (69 studies), LSIL/CIN2-3 (43 studies), and HSIL/CIN2-3 (45 studies). In two studies, the test and reference standard thresholds were not explicitly stated and could not be precisely inferred. Most studies permitted us to construct 2-by-2 tables using more than one combination of test and reference standard threshold.

Sensitivity estimates from these studies covered nearly the entire spectrum, ranging from 0.06 to 0.99. Similarly, specificity estimates ranged from 0.06 to 0.99. Because sensitivity and specificity are interdependent, simple statistics such as means do not provide an accurate description of the diagnostic performance of the test. We used several analytical strategies to determine the cause of the large amount of variability in sensitivity and specificity estimates and to ascertain a reasonable estimate of the performance of Pap tests in a low prevalence screened population. These analyses are described in the section on "Meta-analysis of Pap Smear Accuracy."

The setting in which most of these studies were conducted has important implications for interpreting their results and explaining the differences among them. Most of the studies were conducted in colposcopy clinics in patients who had been referred for colposcopy because of a cytological abnormality on an initial Pap smear screening. In such studies, the repeat Pap smear taken at the time of colposcopy usually served as the "test" result that was compared to the result of either colposcopy (available for all subjects) or histology (available only for women who underwent biopsy; that is, women with a negative colposcopy would be systematically excluded).

The assembly of the study sample from a population referred because of cytological abnormality on initial Pap screening can be expected to bias the spectrum of disease. Women who have a negative Pap smear at the time of colposcopy (following an initial positive smear) are likely to be different from women with a negative result on initial smear. Some have suggested that the removal of cervical epithelium with the initial Pap smear might remove enough abnormal cells that the subsequent Pap may be normal. Our analysis assumes that women with subsequent normal Pap smears had an initial false positive smear. If untrue, this assumption would lead to underestimation of Pap test accuracy.

Returning to the above example, when disease status verification is preferentially obtained in women with a colposcopically visible lesion who undergo biopsy, the study sample is further biased. This selection of subjects for verification results not only in a high prevalence of histological abnormalities in the study sample, but also in "workup" bias (Ransohoff and Feinstein, 1978). Although sensitivity and specificity are relatively invariant to prevalence, they are subject to workup bias. This bias can be expected to lead to elevated estimates of sensitivity and lowered estimates of specificity (Feinstein, 1985).

Many studies comparing conventional Pap screening with new technologies applied the reference standard (adjudicated cytology or histology) only to discrepant cases. Concordant positive and concordant negative test results are assumed to be true positives and true negatives, respectively, but may actually be concordant false results. This study design has a consistent underlying bias that can be expected to overestimate sensitivity and specificity of the new test (Miller, 1998). When the conventional and new test are conditionally dependent, as when tests may have similar problems with sample collection or interpretation of mild disease, this bias can be substantial.

Our goal was to obtain estimates of Pap test performance that are applicable to a low prevalence screened population. However, only three studies identified patients undergoing initial Pap smear screening and verified all (or a random fraction) of test negative subjects. The study of Baldauf, Dreyfus, Lehmann et al. (1995) was the largest of these, comprising 1,539 women. In this study, a 10 percent random sample of Pap negative women underwent colposcopy and biopsy to verify their disease status; this allowed for correction of any verification bias. Davison and Marty (1994) studied 200 women, and Hockstad (1992) tested 73 women; all test negative women in these two studies had their disease status verified with the reference standard test. These studies estimated sensitivity at 56 percent, 53 percent, and 29 percent, respectively, and specificity at 98 percent, 100 percent, and 97 percent, respectively.

Cost of Screening

Estimating the costs of using the Pap smear to screen for cervical cancer is a complex process because costs need to be estimated for both obtaining and processing the smear. Estimates of these screening costs, and the methods used to obtain the estimates, are described below. Estimates of the total cost of Pap smear screening and of the cost of new screening techniques are also provided.

Cost of Obtaining Pap Smears

Several steps were taken in this study to estimate the cost of collecting cytology samples. In the past, studies have used either the cost of an entire office visit or the amount of reimbursement made for the sample collection as a proxy. This overestimates the cost, as Pap smears are usually performed as part of an annual gynecological exam, which involves other procedures as well. In this study, attempts were made to arrive at more accurate estimates. Only the cost of the proportion of the office visit that was directly attributable to obtaining a Pap smear was included. To do so, first the time spent by the physician and other health care staff in obtaining smears was estimated. Second, the cost per minute of office visit time was calculated. Finally, the time devoted to collecting the samples during the office visit was multiplied by the cost per minute to obtain the cost attributable to the collection process.

The physician's time spent to obtain the smears was estimated from the National Ambulatory Medical Care Survey (NAMCS). This survey provides data on ambulatory medical care provided in physicians' offices and is based on a sample of nationally representative patient visits. The survey contains information on age, race, and sex of the patient, organized by selected physician characteristics such as type of specialty and geographic location. Information provided about the clinical substance of the visit includes the patient's problem, the physician's diagnosis (ICD-9-CM), the diagnostic/screening services provided, the surgical procedures performed, and the duration of the visit. Health Economics Research, Inc., analyzed the 1992 NAMCS data, which contained 34,606 patient records from 1,558 doctors who participated in the survey.

The NAMCS data provide information on the amount of time (minutes) spent in direct face-to-face contact with the physician. Separate estimates of physician time were obtained for patients who were 20-64 years old and for those who were 65 years or older. A regression technique was employed to derive the proportion of time spent obtaining a Pap smear. The model used is shown below:

Log (MINUTES) = B₁ + B₂ PAP_TEST

where

PAP_TEST = 1 if performed; = 0 otherwise

The B₁ value is an intercept, and the B₂ value is the proportion of time spent obtaining the smears. A log distribution was used to model the skewed distribution of office visit minutes.

Table 24. Estimates of Physician and Total Time Spent (more...)

Table 24. Estimates of Physician and Total Time Spent Obtaining Pap Smears During Office Visits

Age (yr)	Percent. of Time Spent Obtaining Pap Smear 1 (%)	X	Average Duration of Pap Smear Visit (minutes)	=	Physician Time Spent Obtaining Pap Smear (minutes)	+	Other Medical Staff Time Spent 2 (minutes)	=	Total Time Spent Obtaining Pap Smears (minutes)
Age 20-64	7.67		18.78		1.44		5.86		7.30
Age 65 and older	27.0		20.07		5.42		5.86		11.28

1

Percentage increase in time was obtained by econometric estimation: log(MINUTES) = B1 + B2 PAP_TEST, where PAP_TEST = 1 if performed; 0 otherwise.

2

Other medical staff time was calculated as total time minus physician time (7.30 - 1.44 = 5.86 minutes). For total time spent obtaining Pap smear, see Waugh, Smith, Robertson et al. 1996a

SOURCE: Analysis by Health Economics Research, Inc., of data from the National Ambulatory Medical Care Survey.

Table 24 shows the calculations used to derive estimates of time spent by physicians to obtain Pap smears during office visits. The physician time spent in obtaining the smears was 1.44 minutes for women under the age of 65 and 5.42 minutes for those 65 and older. The difference in the amounts of physician time reflects the difficulties that may be experienced in obtaining cytology samples from women as they age.

The NAMCS data only contain information on time the doctor spent with the patient, not on time the patient spent receiving care from someone else, for instance, a nurse or a technician. Therefore, time spent by other medical staff who assisted in obtaining the smears was estimated by subtracting physician time from total time, calculated by Waugh, Smith, Robertson et al. (1996a), for collecting smears from the under-65 population. When the time spent by other staff was included, the total time spent obtaining the smears was estimated as 7.30 minutes for the younger age group and 11.28 minutes for the older age group.

Table 25. Calculating of Office Visits for Pap Smears (more...)

Table 25. Calculating of Office Visits for Pap Smears

Age 20-64							Age 65 & Older
CPT Code	No. of Pap Visits	% Smears	Cost per Visit (MEDSTAT)	Time (minutes)	No. of Pap Visits	% Smears	Cost per Visit (MEDSTAT)	Cost per Visit (Medicare)	Time (minutes)
99211	2,539	1.9	$32.09	5	26	2.2	$32.09	$13.89	5
99212	10,651	8.0	38.59	10	41	3.5	38.59	24.85	10
99213	41,188	30.9	50.62	15	278	23.9	50.62	35.45	15
99214	44,059	33.1	67.21	25	366	31.5	67.21	54.83	25
99215	16,292	12.2	89.65	40	346	29.8	89.65	86.62	40
99201	690	0.5	54.63	10	8	0.7	54.63	28.88	10
99202	2,031	1.5	60.27	20	10	0.9	60.27	46.42	20
99203	6,486	4.9	75.01	30	23	2.0	75.01	63.60	30
99204	6,429	4.8	94.96	45	33	2.8	94.96	95.03	45
99205	2,725	2.0	113.06	60	30	2.6	113.06	119.16	60
Average	--	--	$64.35	23.93	--	--	$70.11	$60.42	27.52
Range	--	--	$50.62-$89.65	15-40	--	--	$50.62-$89.65	$35.45-$86.62	15-40
Cost per minute	--	--	$2.70	--	--	--	$2.55	$2.20	--
Cost per minute- range	--	--	$2.24-$3.37	--	--	--	$2.24-$3.37	$2.6-$2.37	--

NOTE: All cost estimates are reported in 1997 dollars. CPT = Current Procedures Terminology. SOURCE: Analysis by Health Economics Research, Inc., of MEDSTAT data.

In Table 25, the cost per minute of office visits for Pap smears is shown. For the 20- to 64-year-old age group, office visit cost was obtained from MEDSTAT data; for patients 65 years old and older, data from both MEDSTAT and Medicare payments were used. Weighted average costs per visit and time spent were obtained using the distribution of Pap smears by type of office visit. The cost per minute derived from these estimates was $2.70 for those 20-64 years old (from MEDSTAT data), and $2.55 and $2.20 for the 65 years and older group (from MEDSTAT data and from the Medicare Fee Schedule, respectively). Ranges of the cost estimates were also provided to allow for sensitivity analyses.

Cost of Processing Pap Smears

Table 26. Average Cost for Processing Cytology Samples (more...)

Table 26. Average Cost for Processing Cytology Samples from Pap Smears

CPT Code	Description	% of Smears	Age 20-64 MEDSTAT Costs	Age 65 & Older Medicare Payments
88150	Screening by technician under physician supervision	80.7	$18.75	$8.33
88151	Requiring interpretation by physician	6.1	$19.08	$36.16
88155	With definitive hormonal evaluation	13.2	$20.81	$9.65
--	Average cost per smear	--	$18.97	$10.19
--	STD	--	$6.94	--
--	Range +/- STD	--	$12.03-$25.91	--

All cost estimates are reported in 1997 dollars.

CPT = Current Procedures Terminology; RBRVS = resource based relative value scale; STD = standard deviation.

SOURCE: MEDSTAT data analysis; Medicare's RBRVS fee schedule and Clinical Laboratory fee schedule.

The average cost of processing cytology samples from Pap smears is shown in Table 26. The proportion of smears requiring physician interpretation and the proportion of those with definitive hormonal evaluation are presented, along with the charges associated with processing the cytology samples. The laboratory charges were derived from MEDSTAT data and from Medicare's Clinical Laboratory fee schedule and from the resource-based relative value scale fee schedule (RBRVS). The weighted average cost per smear read from the MEDSTAT data was $18.97, with a standard deviation of $6.94, for patients aged 20 to 64. The average Medicare reimbursement for processing Pap smears was estimated to be $10.19 for patients aged 65 years and older.

Total Cost of Pap Smear Screening

Table 27. Total Cost of Pap Smear Screening

Table 27. Total Cost of Pap Smear Screening

Age (yrs)	Total Time Spent Obtaining Pap Smear (minutes)	X	Cost per Minute of Office Visit Time	=	Office Visit Cost for Pap Smear	+	Pap Smear Lab Processing cost	=	Total Cost of Performing Pap Smears
Age 20-64	7.30		$2.70		$19.71		$18.97		$38.68
Range	--		$2.24-$3.37		$16.35-$24.60		--		$35.32-$43.57
Age 65 and Older
MEDSTAT	11.28		$2.55		$28.76		$18.97		$47.73
Range	--		$2.24-$3.37		$25.27-$38.01		--		$44.24-$56.98
Medicare	11.28		$2.20		$24.82		$10.19		$35.01
Range	--		$2.6-$2.37		$24.36-$26.73		--		$34.55-$36.92

All cost estimates are reported in 1997 dollars.

SOURCE: Analysis by Health Economics Research, Inc., of National Ambulatory Medical Care Survey and MEDSTAT data, and Clinical Laboratory fee schedule.

In Table 27, the total cost of performing Pap smears is presented. Total cost, which included the office visit cost and the processing cost, was calculated for the two groups separately. For women ages 20-64 years, the total cost in 1997 dollars was $38.68. For women 65 years or older, the MEDSTAT estimate was $47.73 and the Medicare estimate was $35.01. Range estimates were provided for all cost estimates to allow for sensitivity analyses.

Cost of New Screening Techniques

Table 28. Estimates of Screening Costs for Cervical (more...)

Table 28. Estimates of Screening Costs for Cervical Cancer

Screening Procedure	Office Visit Cost for Pap Test	Cost of Conventional Pap Smear	Cost of New Technology	Total Cost Estimate
Conventional Pap Age 20-64 Average Range	$19.71 $16.35-$24.80	$18.97 --	-- --	$38.68 $35.32-$43.77
Age 65 and older -- MEDSTAT Average Range	$28.76 $25.27-$38.01	$18.97 --	-- --	$47.73 $44.24-$56.98
Age 65 and older -- Medicare Average Range	$24.82 $24.36-$26.73	$10.19 --	--	$35.01 $34.55-$36.92
AutoPap® Age 20-64 Average Range	$19.71 $16.35-$24.80	$18.97 --	$7.58 $7.15-$8.00	$46.26 $42.47-$51.77
Age 65 and older -- MEDSTAT Average Range	$28.76 $25.27-$38.01	$18.97 --	$7.58 $7.15-$8.00	$55.31 $51.39-$64.98
Age 65 and older -- Medicare Average Range	$24.82 $24.36-$26.73	$10.19 --	$7.58 $7.15-$8.00	$42.59 $41.70-$44.92
Papnet® Age 20 to 64 Average Range	$19.71 $16.35-$24.80	$18.97 --	$15.00 $10.00-$18.00	$53.68 $45.32-$61.77
Age 65 and older -- MEDSTAT Average Range	$28.76 $25.27-$38.01	$18.97 --	$15.00 $10.00-$18.00	$62.73 $54.24-$74.98
Age 65 and older -- Medicare Average Range	$24.82 $24.36-$26.73	$10.19 --	$15.00 $10.00-$18.00	$50.01 $44.55-$54.92
ThinPrep® Age 20 to 64 Average Range	$19.71 $16.35-$24.80	-- --	$24.57 $15.00-$30.00	$44.28 $31.35-$54.80
Age 65 and older -- MEDSTAT Average Range	$28.76 $25.27-$38.01	-- --	$24.57 $15.00-$30.00	$53.33 $40.27-$68.01
Age 65 and older -- Medicare Average Range	$24.82 $24.36-$26.73	-- --	$24.57 $15.00-$30.00	$49.39 $39.36-$56.73

®

SOURCE: Analysis by Health Economics Research, Inc., of MEDSTAT data from the National Ambulatory Medical Care Survey, and information from ThinPrep, AutoPap, and Papnetdevelopers.

Numerous technological advances have been made in the collection of specimens and in the interpretation of Pap smears. Table 28 contains the cost or charge per smear, using various screening procedures available in the market today -- conventional Pap, ThinPrep^®, AutoPap^®, and Papnet^®. Cost estimates were provided for the conventional Pap smear, and charges were presented for the others (cost information was not available).

ThinPrep^®

The costs to perform a ThinPrep^® Pap test can be broken into the following categories for a clinical laboratory: supplies, preparation costs, evaluation costs, and indirect expenses. The laboratory supplies consist of the ThinPrep^® Pap test kit that is distributed to clinicians along with stains and fixatives. The current retail price to laboratories for the ThinPrep^® Pap test kit and related supplies is $9.75. Preparation costs include the ThinPrep^® 2000 Processor cost of $45,000 as well as costs associated with personnel for preparing the slide for processing and disposing of waste after processing. Amortizing the cost of the ThinPrep^® 2000 Processor over a 5-year period and assuming that 40,000 slides are processed annually yields an estimated $0.22 equipment cost.

The last two categories of costs, evaluation and indirect expenses, as well as the personnel costs for preparing the slide for processing and disposing of waste, are essentially equivalent between the ThinPrep^® technology and the conventional Pap smear technology. A recent study conducted by the College of American Pathologists (CAP) estimated these costs for conventional Pap smears to be approximately $14.60. Thus, a total cost estimate of $24.57 can be derived by summing the components.

This estimate, however, excludes the cost savings attributed to a reduction in the number of Pap smears that may be repeated because the specimen is determined to have a "Satisfactory but limited by" (SBLB) diagnosis. The manufacturer of ThinPrep^® argues that specimen quality is significantly improved with this new technology, and it can reduce the number of specimens labeled SBLB (currently estimated to be 40 percent) as well as the number of repeat Pap smears that result from this labeling. Cytyc Corporation estimates a 50 percent reduction in the number of specimens labeled as SBLB; however, estimates on the number of women who return for a repeat Pap smear as a result of an SBLB diagnosis are not well known. The cost estimate of $24.57 does not include any adjustments to reflect these potential cost savings. Ignoring these costs will tend to favor conventional Pap smears over the new ThinPrep^® technology, and this is a conservative assumption.

However, the cost analysis has been conducted from the perspective of the health care system, with the majority of costs estimated using payments or allowed charges as a proxy for costs. Thus, it seems reasonable to take a similar approach to the estimation of costs for the new technologies. Cytyc Corporation estimates from a large number of private insurers that the average payment to laboratories for primary screening using the ThinPrep^® technology ranges from $15.00 to $30.00 and includes all nonoffice costs associated with the processing and evaluation of the ThinPrep^® smears, including pathologist fees for reviewing abnormal slides. Because Medicare does not currently reimburse laboratories at a different level from conventional Pap smears when ThinPrep^® smears are performed, we do not have current payment estimates for the Medicare population. Thus, the ThinPrep^® payment rates from the private sector will be used as an estimate for the Medicare population as well.

Using the payment levels from the private sector for the ThinPrep^® technology and office visit payment estimates derived from MEDSTAT data and Medicare's RBRVS physician fee schedule (see Table 27), the cost estimates for ThinPrep^® smears were obtained (Table 28).

Papnet^®

The incremental costs associated with the Papnet^® technology may be broken down into the following categories: the costs of preparation and shipment of the "normal" slides to a central facility for processing, the costs associated with the Papnet^® neural network computer system's processing of the slide at a central facility, the costs of a viewing station to review the data shipped back to the laboratory on compact disks, and the costs of triaging and reviewing selected slides by the cytotechnologist. Neuromedical Systems Inc. (NSI) estimates that the average cost to laboratories for these costs ranges from $12.00 to $18.00 depending on the laboratory's volume. These costs are in addition to the cytotechnologist and pathologist costs associated with the manual review of normal and abnormal slides and the costs of supplies for obtaining and processing the sample, costs that are already occurring with conventional Pap smears.

The charge for Papnet^® rescreening has been reported in the literature and from NSI to range from $30.00 (Schechter, 1996) to $40.00 (literature provided by NSI). This charge is in addition to charges for the conventional Pap smear, which includes all nonoffice costs associated with the processing and evaluation of the conventional Pap smear, including pathologist fees for reviewing abnormal slides. However, we were unable to obtain information from NSI regarding current average payments for rescreening with Papnet^® technology. Because cost estimates for the competing technologies were derived using average payments, rather than charges, we believe that it would be inappropriate to use charges for Papnet^®. Thus, we model Papnet^®'s cost-effectiveness using a cost estimate that more closely approximates the projected costs, rather than reflecting current charges (see Table 28).

AutoPap^®

The incremental costs associated with the AutoPap^® technology may be broken down into the following categories: equipment, software, service, and software licensing. NeoPath, Inc., estimates the total average cost to laboratories for these four cost components to be $4.50, with a range from $3.75 to $4.50. These costs are in addition to the cytotechnologist and pathologist costs associated with manual review of normal and abnormal slides and the costs of supplies for obtaining and processing the sample, which are already occurring with conventional Pap smears. When AutoPap^® is used for quality control purposes, there are no additional incremental costs or savings associated with labor. In contrast, when it is used as a primary screening vehicle, NeoPath, Inc., estimates that there is an overall reduction in cytotechnologist labor of about 20 percent across all slides. The reduction is a function of fewer slides being processed for review by the laboratory's cytotechnologist.

This savings in cytotechnologist labor can be demonstrated through a simple example. Assuming a laboratory currently processes 100 patient slides using the conventional technique of manual review of all 100 slides by a cytotechnologist, the cytotechnologist must actually review 110 slides because of the CLIA quality control requirement of a 10 percent re-review. In contrast, the AutoPap^® scores all 100 slides and archives 25 percent of them as the "most normal" of all slides. There is no manual review of these slides by the cytotechnologist. There is manual review of the remaining 75 slides as well as a review of 15 of the slides to meet CLIA quality control standards (set at 15 percent for AutoPap^® versus 10 percent for conventional manual review). Thus, the total number of slides undergoing manual review by the cytotechnologist is actually 90 when the AutoPap^® technology is used versus 110 when conventional manual review is done. Ignoring this cost will tend to favor conventional Pap smears relative to AutoPap^®.

The cost analysis has been conducted from the perspective of the health care system, however, with the majority of costs estimated using payments or allowed charges as a proxy for costs. Thus, it seems reasonable to take a similar approach to the estimation of costs for the new technologies. NeoPath, Inc., estimates that average payment to laboratories for quality control screening using the AutoPap^® technology ranges from $7.15 to $8.00 from a large number of public and private insurers, respectively. This payment is in addition to payment for the conventional Pap smear, which includes all nonoffice costs associated with the processing and evaluation of the conventional Pap smears, including pathologist fees for reviewing abnormal slides. Because Medicare does not currently reimburse laboratories at a different level from conventional Pap smears when AutoPap^® is used to identify slides for review, we do not have current payment estimates for the Medicare population. Thus, the AutoPap^® payment rates from the non-Medicare public and private sector will be used as an estimate for the Medicare population as well.

Using the above-referenced payment levels for the AutoPap^® technology, conventional Pap smear payment rates to account for cytotechnologist and pathologist review costs and slide preparation costs, and office visit payment estimates derived from MEDSTAT data and Medicare's RBRVS physician fee schedule (see Table 27), the cost estimates for AutoPap^® smears were obtained (see Table 28).

Cost of Diagnostic Procedures and Treatment

A claims analysis was performed to obtain the cost of diagnostic procedures and treatments for women 20-64 years old. Because no primary data analysis was undertaken for patients 65 years old and older, Medicare's RBRVS fee schedule, Clinical Laboratory fee schedule, diagnosis-related group payment rates, and ambulatory surgery center payment rates were reported.

All outpatient and inpatient claims for women (20-64 years old) with ICD-9 diagnosis codes for cervical cancer, carcinoma in situ, and dysplasia were included in the sample analyzed. The specific diagnosis codes used are shown below:

Pregnant women were excluded from the analysis, as costs associated with diagnosing and treating this group differ from those of the general population. Cases with only an ICD-9 code of 180.8, "other specified sites of cervix," and no other relevant diagnostic codes (180.0, 180.1, or 180.9) were excluded from the analysis as they may have indicated cases requiring specialized treatment. Very few cases were excluded because of this selection criterion.

Costs for procedures performed on an inpatient and on an outpatient basis are provided. Certain procedures are provided only on an inpatient basis (for example, exenteration), though others (colposcopy, cone biopsy) may be performed in either an inpatient or an outpatient setting. For the procedures in the latter category, the appropriate setting or the setting where the procedure was most often performed for women with cervical cancer was utilized to derive the estimates. Most of these procedures were estimated using outpatient (physician and hospital outpatient department) costs. For procedures where both inpatient and outpatient settings were deemed reasonable, the cost of the procedure was estimated for both the outpatient and the inpatient setting.

The analysis was limited to those individuals with a single procedure on a given day or admission to minimize differences related to multiple procedures. Thus, individuals who had both cone biopsy and LEEP performed on the same day would not have been included in the estimation because accurate allocation of costs to each specific procedure was not always possible. For outpatient procedures, in addition to the costs specifically related to the procedure (identified by the procedure code), we also estimated the cost of related services inherent in the performance of the procedure. Therefore, we evaluated all other outpatient services that were performed on the same day to determine if they were appropriate to include in our cost definitions. For instance, costs associated with anesthesia were added to the cost estimates to capture the total cost of performing the procedure. Supplementing procedure-specific costs with other related services provided allowed for a more comprehensive estimate of the costs associated with the procedure. In addition, all separately billed pathology services (CPT codes 88305 and 88307) were included in the final procedure cost.

Table 29. Cost Estimates for Diagnostic Procedures (more...)

Table 29. Cost Estimates for Diagnostic Procedures for Cervical Cancer

Age 20-64							Age 65 and Older Medicare Payments
Procedure	No. of Obs.	MEDSTAT Average Cost	Standard Deviation	25% Q1	50% Med.	75% Q3
Colposcopy
Exploration	5,412	$142.63	73.32	$81	$130	$179	$68.60
With Biopsy	16,878	$276.57	101.30	$195	$267	$351	172.63
Cervical Biopsy	1,266	$189.81	121.64	$115	$164	$217	131.46

1

All costs presented are in 1997 dollars.

Med. = median; Obs. = observations; Q1 = quartile 1; Q3 = quartile 3; RBRVS = resource based relative value scale.

SOURCE: Analysis by Health Economics Research, Inc., of MEDSTAT data and Medicare's RBRVS fee schedule.

Table 30. Cost Estimates for Cervical Cancer Treatment (more...)

Table 30. Cost Estimates for Cervical Cancer Treatment¹

Age 20 to 64								Age 65 and Older Medicare Payments
Treatment	No. of Obs.	MEDSTAT Average Cost	Standard Deviation	Days	25% Q1	50% Med	75% Q3
1. Cryotherapy	5,659	$156	$63	--	$108	$142	$184	$111
2. Loop electrode excision procedure	1,972	564	268	--	352	542	710	205
3. Laser ablation	1,393	730	457	--	390	650	976	627
4. Cone biopsy
Oupatient	4,856	919	443	--	623	821	1,114	831
Inpatient	10	4,052	2,200	1.40	2,640	3,268	4,564	2,459
5. Simple hysterectomy	518	5,899	2,348	3.05	4,364	5,407	6,934	5,694
6. Radical hysterectomy	110	13,077	6,395	6.54	8,688	11,453	15,105	8,535
7. Exenteration	9	51,413	61,663	21.78	17,011	19,373	53,261	9,370
8. Radiation therapy
Inpatient (per admission)								3,653
STAGE I	72	3,766	1,631	2.38	2,616	3,397	4,537	--
STAGE II-III	17	3,883	2,335	2.82	2,439	3,861	4,363	--
STAGE IV	14	7,074	5,810	7.36	3,332	4,695	10,838	--
Outpatient (per service)	4,531	410	468	--	133	258	526	196 2
9. Chemotherapy
Inpatient (per admission)								2,523
All cases	51	5,226	4,970	3.63	2,253	3,492	6,228	--
Less than 5 days	39	3,543	1,676	2.64	2,212	3,246	3,932	--
Outpatient (per service)	2,756	191	241	--	56	105	222	56 3

1

All costs presented are in 1997 dollars.

2

Medicare payment for most common radiation treatment received by women with cervical cancer -- weekly therapy (CPT code 77430).

3

Medicare payment for most common chemotherapy treatment received by women with cervical cancer -- infusion technique (CPT code 96410). This estimate does not include the cost of the chemotherapeutic agent.

CPT = Current Procedures Terminololgy; DRG = diagnostic-related group; Med. = median; Obs. = observations; Q1 = quartile 1; Q3 = quartile 3; RBRVS = resource based relative value scale.

SOURCE: Analysis by Health Economics Research, Inc., of MEDSTAT data and Medicare's RBRVS fee schedule/DRG payment.

Table 31. Cost Estimates for Other Procedures Used (more...)

Table 31. Cost Estimates for Other Procedures Used to Diagnose and Treat Cervical Cancer ¹ (Continued)

Procedures	Age 20-64						Age 65 and Older Medicare Payments
	No. of Obs.	MEDSTAT Average Cost	Standard Deviation	25% Q1	50% Med.	75% Q3
Endocervical curettage	2,546	98	75	$49	$81	$126	$71
Dilation and curettage	925	252	73	$163	$280	$322	642
Dilation of cervical canal	97	119	88	$75	$82	$136	52
Endometrial sampling (biopsy)	2,833	199	6	$162	$199	$236	58
Cauterization of cervix	703	178	247	$55	$108	$201	95
Pelvic exam under anesthesia	199	201	170	$81	$184	$341	136
Chest X-ray	5,534	43	58	$24	$31	$48	35
CAT scan	2,104	312	248	$170	$193	$379	271
Barium enema	1,204	108	56	$56	$107	$132	101
Bone scan	23	236	251	$56	$143	$261	122 2
Magnetic resonance imaging	156	563	453	$314	$490	$916	497
Intravenous pyelogram	7	339	267	$105	$248	$477	149
Cystoscopy	28	982	851	$226	$674	$1,546	184 3
Proctoscopy	7	774	543	$335	$672	$1,179	196 3

1

All costs presented are in 1997 dollars.

2

This estimate does not include the cost of the scanning agent.

3

These estimates only reflect costs associated with the professional component.

Med. = median; Obs. = observations; Q1 = quartile 1; Q3 = quartile 3; RBRVS = resource based relative value scale; CAT = computerized axial tomography.

SOURCE: Analysis by Health Economics Research, Inc., of MEDSTAT data and Medicare's RBRVS fee schedule.

Cost by Stage of Disease

Table 32. Identifying Stages of Cervical Cancer

Table 32. Identifying Stages of Cervical Cancer

Stage	Primary ICD-9 Codes	Secondary ICD-9 Codes
LSIL	622.1	No diagnosis indicating cancer or CIS
HSIL	233.1	No diagnosis indicating cancer
Stage I	180.0, 180.1, 180.9	No diagnosis indicating other forms of cancer
Stage II	180.0, 180.1, 180.9	198.82, 198.6, 179, 181, 182.x, 183.x, 184.x
Stage III	180.0, 180.1, 180.9	198.82, 198.6, 179, 181, 182.x, 183.x, 184.x and either of these codes 593.5, 599.6, 593.89, 591
Stage IV	180.0, 180.1, 180.9	199.1, 199.0, 140.x - 176.x, 188.x - 195.x, 197.x , 198.0, 198.1, 198.2, 198.3, 198.4, 198.5, 198.7, 198.81, 198.89

HSIL = high-grade squamous intraepithelial lesions; ICD-9 = International Classification of Diseases, version 9; LSIL = low-grade squamous intraepithelial lesions; CIS = Carcinoma in situ

SOURCE: ICD-9-CM Code Book, St. Anthony Publishing.

Table 33. Cost Associated with Diagnosis, Treatment, (more...)

Table 33. Cost Associated with Diagnosis, Treatment, & Follow-up of Cervical Cancer by Stage of Disease

Lesions/Stage of Disease	No. of Obs.	Average Inpatient Cost	Average Outpatient Cost	Average Total Cost	Hospital Days
LSIL	9,329	--	$1,728	$1,728	--
HSIL
With oupatient services only	1,257	--	3,049	3,049	--
With oupatient & inpatient Services	337	$6,352	2,806	9,159	3.17
Stage I	248	10,948	6,697	17,645	6.31
Stage II/III
Only Stage II/III costs	39	20,298	6,771	27,069	9.59
Stages I and II/III	54	19,734	11,439	31,173	13.17
Stage IV
Only Stage IV costs	17	31,115	9,164	40,280	20.29
All stages (final diagnosis Stage IV)	70	37,258	19,552	56,810	28.46

All costs presented are in 1997 dollars; costs are associated with women ages 20-64 years old diagnosed with cervical cancer.

HSIL = high-grade squamous intraepithelial lesions; LSIL = low-grade squamous intraepithelial lesions; Obs. = observations.

SOURCE: Analysis by Health Economics Research, Inc., of MEDSTAT data.

Table 34. Mean and Standard Deviation of Total Cost (more...)

Table 34. Mean and Standard Deviation of Total Cost for Cervical Cancer by Stage of Disease¹

Lesions/Stage of Disease	Total Cost	Standard Deviation	25% Q1	50% Med.	75% Q3
LSIL	$1,728	$1,498	$675	$1,143	$2,274
HSIL
With oupatient services only	3,049	2,033	1,384	2,733	4,203
With oupatient & inpatient services	9,159	4,207	6,710	8,556	11,540
Stage I	17,645	17,572	9,439	14,005	21,946
Stage II/III
Only Stage II/III costs	27,069	55,953	11,734	15,392	29,089
Stages I and II/III	31,173	51,231	13,303	22,631	32,858
Stage IV
Only Stage IV costs	40,280	53,390	12,670	17,772	46,509
All stages (final diagnosis Stage IV)	56,810	43,930	27,595	50,705	75,350

1

All costs presented are in 1997 dollars; costs are associated with women 20-64 years old diagnosed with cervical cancer.

HSIL = high-grade squamous intraepithelial lesions; LSIL = low-grade squamous intraepithelial lesions; Med. = median; Q1 = quartile 1; Q3 = quartile 3. SOURCE: Analysis by Health Economics Research, Inc., of MEDSTAT data.

Cost-effectiveness of Conventional Pap Testing

A series of comprehensive models of the cost-effectiveness of cervical cancer screening was initiated with a 1981 OTA report (U.S. Congress Office of Technology Assessment, 1981). Subsequently, efforts by Eddy (1990), Mandelblatt and Fahs (1988), and OTA (1990) (and a related study by Fahs, Mandelblatt, Schechter et al., 1992) addressed similar questions in somewhat different populations and perspectives. Eddy's target population for initial screening was asymptomatic 20-year-old women of average risk, whereas Fahs, Mandelblatt, Schechter et al. (1992) assessed women over the age of 65. Both analyses were conducted from the payer's perspective. All of these models used Markov processes, with the modification that transition probabilities change as patients or population age. Of all the models, Eddy's is considered to be the most general.

There are significant discrepancies in the results of different models of cervical cancer screening. For example, both Eddy (1990) and Fahs, Mandelblatt, Schechter et al. (1992) reported the marginal cost-effectiveness of triennial Pap smear screening for women age 65 or over under two conditions: (1) assuming no prior screening and (2) assuming previous regular screening. Triennial Pap screening beginning at age 65 in women with no prior screening was estimated to cost $22,448 per year of life saved, according to Eddy (1990), but was actually cost-saving in the Fahs, Mandelblatt, Schechter et al. (1992) analysis. When considering screening in women with prior regular screening, Eddy and Fahs et al. estimated different marginal cost-effectiveness ratios: $52,241 per year of life saved versus $33,572 per year of life saved, respectively (Eddy, 1990; Fahs, Mandelblatt, Schechter et al., 1992). It is difficult to assess which differences in model assumptions account for the differing results. However, sensitivity of Pap smears is known to strongly influence cost-effectiveness ratios. Eddy used 85 percent sensitivity in his base case analysis, whereas Fahs et al. used 75 percent sensitivity in their model. Another principal difference between Eddy (1990) and Fahs, Mandelblatt, Schechter et al. (1992) was whether the models assessed implementing a change in screening practices that might detect a large, one-time benefit by detecting prevalent cases in an unscreened population versus assessing the ongoing benefit one might see in a continuing policy. By attempting to recreate the results of Eddy (1990) and Fahs, Mandelblatt, Schechter et al. (1992) with the model created for this report, we were able to identify other differences, which are discussed in greater detail below.

Cost-Effectiveness of Three New Technologies

Two analyses have addressed the cost-effectiveness of Papnet^®, which provides interactive neural-network rescreening, versus unassisted manual interpretation of Pap smears (Schechter, 1996; Radensky and Mango, 1998). Schechter (1996) revised a previously published model constructed for OTA that examined the cost-effectiveness of implementing Pap screening as a Medicare benefit (Fahs, Mandelblatt, Schechter et al., 1992; U.S. Congress Office of Technology Assessment, 1990). Parameters in the model were updated to reflect the general U.S. population of women undergoing Pap testing, and the addition of a Papnet^® strategy with information on the sensitivity of the new technology was supplied by the manufacturer. The primary result of this study, $48,474 per life-year saved for biennially screened women, is given in terms of the overall cost-effectiveness ratio (which compares Papnet^® screening to no screening) rather than in terms of the conventional marginal (or incremental) cost-effectiveness ratio (which would require comparing Papnet^® screening with conventional Pap screening). This makes it appear that Papnet^® is highly cost-effective when, in fact, most of the cost-effectiveness is from conventional Pap testing; the incremental cost-effectiveness ratio for Papnet^®-assisted compared with that for conventional Pap testing is much higher.

Radensky and Mango (1998) adapted Eddy's (1990) model to estimate the marginal cost-effectiveness ratio of using interactive neural network-assisted (INNA) rescreening of negative smears compared with unassisted manual examination of cervical smears at a frequency of every 3 years. The cost-effectiveness ratios ranged from $39,087 to $79,440, depending on the sensitivity of the Pap smear examination and on the specificity of INNA rescreening. More frequent screening intervals than every 3 years were not examined in this analysis.

Brown and Garber (1998) performed a cost-effectiveness analysis of Papnet^®, AutoPap^® and ThinPrep^® for cervical cancer screening for the Blue Cross and Blue Shield Association's Technology Evaluation Center. This analysis was based upon the model by Eddy (1990). Results from this model show that Papnet^®, AutoPap^®, and ThinPrep^® modestly improve on the diagnostic accuracy of conventional Pap screening. These screening technologies are likely to increase life expectancy by only 1 or 2 days for most women and only by a few hours for those women who already get Pap tests frequently. The marginal cost-effectiveness ratios of the new technologies were found to exceed $80,000 per additional year of life saved with biannual screening and far greater with annual screening. However, the cost-effectiveness ratios are sensitive to the costs of each technology, sensitivity of the initial test, prevalence of cervical cancer, and proportion of tumors that grow rapidly.

Terminology

The sensitivity of a test is defined as the conditional probability that, given the presence of disease, the test will be positive. In the setting of our model, where the presence of disease is defined by the various Markov states, the sensitivity is equivalent to the true positive rate. The FNR is equal to the quantity:
1 - sensitivity.

Because modeling probabilities is easiest with variables between 0 and 1, we expressed the potential improvement in Pap test performance as a reduction in the FNR:
X (1 -sensitivity).

A reduction in the FNR is equivalent to improving sensitivity within the context of the model. The two expressions are used interchangeably throughout this section of the report.

Base Case Parameter Estimates

Table 36. Probability Estimates for Natural History (more...)

Table 36. Probability Estimates for Natural History Model

Parameter	Base Case	Range	References
Prevalence of HPV infection, age 15	0.10	0-0.25	Assumption
Prevalence of LSIL, age 15	0.01	0-0.1	Assumption
Age-specific incidence of HPV infection			Kiviat, 1996; Koutsky, 1997;Ho, Bierman, Beardsley et al., 1998;Hildesheim, Schiffman, Gravitt et al., 1994
15	0.1	0.5-2 X base estimate
16	0.1
17	0.12
18	0.15
19	0.17
20	0.15
21	0.12
22	0.10
23	0.10
24-29	0.05
30-49	0.01
50 +	0.005
Age-specific regression rate, HPV infection (HPV to Well)			Ho, Bierman, Beardsley, et al, 1998;Moscicki, Shiboski, Broering et al., 1998; Koutsky, Holmes, Critchlow et al., 1992
15-24	0.7/18 months	0.6-0.9
25-29	0.5/18 months	0.45-0.6
30+	0.15/18 months	0.1-0.2
Progression rate (HPV to LSIL)	0.2/36 months	0.15-0.3
Proportion of infections progressing directly to HSIL	0.1	0.05-0.5
Probability of progression after treatment for SIL	0.05		Eddy 1990; Fahs, Mandelblatt, Schechter et al., 1992; Schechter 1996
Regression rate (LSIL to unknown HPV)			Syrjanen, Kataja, Yliskoski et al., 1992;van Oortmarssen and Habbema 1991; Hildesheim, Schiffman, Gravitt et al., 1994; Munoz, Kato, Bosch et al., 1996;Bearman, MacMillan, and Creasman, 1987; Pretorius, Semrad, Watring et al., 1991; Janerich, Hadjimichael, Schwartz et al., 1995; Schwartz, Hadjimichael, Lowell et al., 1996
Age 15-34	0.65/72 months	0.6-0.8
35+	0.4/72 months	0.3-0.6
Progression rate (LSIL to HSIL)
Age 15-34	0.1/72 months	0.1-0.3
35+	0.35/72 months	0.3-0.5
Regression rate (HSIL to LSIL)	0.35/72 months	0.3-0.5
Progression rate (HSIL to Stage I cancer)	0.4/120 months	0.3-0.5
Progression rates and probability of symptoms in unscreened patients with cancer
Stage I
Progression rate (Stage I to Stage II)	0.9/4 years
Annual probability of symptoms	0.15
Stage II
Progression rate (Stage II to Stage III)	0.9/3 years
Annual Probability of Symptoms	0.225
Stage III
Progression Rate (Stage III to Stage IV)	0.9/2 years
Annual probability of symptoms	0.6
Stage IV
Annual probability of symptoms	0.9
Annual probability of survival after diagnosis by stage			Fremgen, personal communication; Jones, Shingleton, Russell et al., 1995
Stage I
Year 1	0.97
Year 2	0.97
Year 3	0.97
Year 4	0.97
Year 5	0.97
Stage II
Year 1	0.90
Year 2	0.89
Year 3	0.91
Year 4	0.95
Year 5	0.96
Stage III
Year 1	0.70
Year 2	0.74
Year 3	0.86
Year 4	0.93
Year 5	0.83
Stage IV
Year 1	0.40
Year 2	0.50
Year 3	0.86
Year 4	0.85
Year 5	0.90

Table 37. Probability Estimates for Screening

Table 37. Probability Estimates for Screening

Parameter	Base Case	Range	References
Conventional Pap
Sensitivity, cytology ASCUS+ for histology LSIL +	0.51	0.51-0.788	See meta-analysis
Specificity, cytology ASCUS+ for histology LSIL +	0.97	0.888-0.97	See meta-analysis
Percentage of smears read as normal in initial screening step re-read at same sensitivity and specificity	0.10	NA	NA
Technology improving initial screening step
Relative decrease in false negative rate of initial screening step	0.6	0.4-0.9	Roberts, Gurley, Thurloe et al., 1997;Bolick and Hellman, 1998; Brown and Garber, 1998
Relative specificity	1	0.9-1.0	Assumption
Percentage of smears read as normal in initial screening step re-read at same sensitivity and specificity	10%	NA	NA
Technology improving rescreening step
Relative decrease in false negative rate of rescreening step	0.6	0.4-0.9	Stevens, Milne, James et al., 1997;Patten, Lee, Wilbur et al., 1997aColgan, Patten, and Lee, 1995;Brown and Garber, 1998; Kaufman, Schreiber and Carter, 1998; Jenny, Isenegger, Boon et al., 1997
Relative specificity of rescreening step	1	0.9-1.0	Assumption
Percentage of smears read as normal in initial screening step re-read at same sensitivity and specificity	100%	NA	NA

NA = not available.