IN THE UNITED STATES, CYTOLOGIcal

Transcription

1 ORIGINAL CONTRIBUTION Cost-effectiveness of Alternative Triage Strategies for Atypical Squamous Cells of Undetermined Significance Jane J. Kim, MS Thomas C. Wright, MD Sue J. Goldie, MD, MPH IN THE UNITED STATES, CYTOLOGIcal screening has reduced the incidence of invasive cervical cancer, but as a result of screening, many women are diagnosed as having equivocal cytological abnormalities (eg, atypical squamous cells, herein referred to as ASC). 1 The management of an ASC result is controversial, and clinical practice patterns range from performing immediate colposcopy to repeating cervical cytology at specified intervals. 2 Clinicians who prefer immediate colposcopy argue that some women with ASC have cervical intraepithelial neoplasia grade 2-3 (CIN 2-3) or invasive cancer, and in fact, a substantial proportion of all biopsyconfirmed CIN 2-3 is identified in women with an ASC result. 3 Clinicians who prefer to use repeat cytology argue that most women have either no lesion or CIN grade 1, which is likely to regress in the absence of any treatment. A third option is DNA testing for high-risk types of human papillomavirus (HPV) and performing colposcopy only in women with positive test results. This strategy is increasingly adopted in the United States because HPV DNA testing appears to be more sensitive than a single repeat cervical cytology for identifying women with ASC who have CIN See also pp 2372 and Context Every year approximately 2 million US women are diagnosed as having a cervical cytological result of atypical squamous cells of undetermined significance (ASC-US). Objective To determine the most efficient and cost-effective management strategy for women in the United States diagnosed as having ASC-US. Design and Setting Cost-effectiveness analysis of data from clinical trials, prospective studies, and other published literature. A computer-based model was used to compare 4 management strategies for a cytological result of ASC-US: immediate colposcopy; human papillomavirus (HPV) triage, which includes colposcopy if high-risk HPV types are detected; repeat cytology, which includes follow-up cytology at 6 and 12 months and referral for colposcopy if a repeat abnormal result occurs; and reclassifying ASC-US as normal in which a cytological result of ASC-US is ignored. Reflex HPV DNA testing uses either residual liquid-based cytological specimens or samples cocollected at the time of the initial screening for conventional cytology. Another method, referred to as the 2-visit HPV DNA triage, requires a woman with an ASC-US result to return within 1 month to provide another speciman sample. Main Outcome Measures Years of life saved (YLS), quality-adjusted life-years (QALYs), and incremental cost-effectiveness ratios. Results The least costly strategy for biennial screening was to reclassify ASC-US as normal, resulting in a reduction in total cancer incidence of 75% for conventional cytology and 84% for liquid-based cytology compared with no screening. The next least costly strategy was HPV DNA testing resulting in a reduction in total cancer incidence of 86% for conventional cytology and 90% for liquid-based cytology, followed by immediate colposcopy with a reduction of 87% and 91%, respectively. Compared with reflex HPV DNA testing, a strategy of repeat cervical cytology or delayed HPV testing costs more but is less effective. When all strategies were compared simultaneously, varying frequency and type of cytological test, biennial (vs every 3 years) liquid-based cytology with reflex HPV testing had a cost of $ per YLS. In a similar comparison, liquid-based cytology with reflex HPV testing conducted every 3 years (vs every 5 years) had a cost of $59600 per YLS and was more effective and less costly than a strategy of conventional cytology incorporating repeat cytology or immediate colposcopy conducted biennially. Conclusion Reflex HPV DNA testing provides the same or greater life expectancy benefits and is more cost-effective than other management strategies for women diagnosed as having ASC-US. JAMA. 2002;287: Determining the most effective and efficient management approach for ASC in a given setting requires a formal analysis of costs and benefits. To contribute information prior to the deliberations of Author Affiliations and Financial Disclosure are listed at the end of this article. Corresponding Author and Reprints: Sue J. Goldie, MD, MPH, Department of Health Policy and Management, Harvard Center for Risk Analysis, 718 Huntington Ave, Second Floor, Boston, MA ( sgoldie@hsph.harvard.edu) JAMA, May 8, 2002 Vol 287, No. 18 (Reprinted) 2002 American Medical Association. All rights reserved.

2 professional organizations developing national guidelines for the optimal management of ASC, we conducted a comprehensive cost-effectiveness analysis comparing different strategies. Figure 1. Natural History Model HPV Undetectable CIN1 CIN 2-3 METHODS Natural History Model We modified a previously developed computer-based mathematical model, 9-11 which simulates the natural history of cervical carcinogenesis using a sequence of monthly transitions among health states (FIGURE 1). HPV infection is defined as either detectable or undetectable to accurately represent the proportion of CIN 1 and CIN 2-3 that would be detectable in a clinical practice setting. This structure was chosen to permit calibration to cross-sectional data on the distribution of detectable HPV DNA within low-grade squamous intraepithelial lesions (LSILs) and high-grade squamous intraepithelial lesions (HSILs). Biopsy-confirmed cervical disease is defined as either CIN 1 or CIN 2-3, whereas cytological results are classified as ASC, LSIL, or HSIL. 12 Invasive cervical cancer is stratified according to the cancer staging system of the National Cancer Institute s Surveillance, Epidemiology, and End-Results Program (local, regional, and distant cancer). 13 For our analyses, a hypothetical cohort of adolescent girls were entered into the model at age 13 years. They have never had sex, are free of disease, but at each month face an age-dependent risk of acquiring HPV. Females with detectable high-risk types of HPV (defined as HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 61 detected using a sensitive molecular assay) have a greater risk of CIN 2-3 and invasive cancer than those with undetectable HPV Women with high-risk HPV types, which are not detected using a single test, and women with low-risk HPV types are considered to be in an undetectable dimensional health state. Women infected with high-risk HPV types, as well as some women with falsepositive test results (eg, low-risk types of HPV), are considered to be in a detectable dimensional health state. Healthy HPV Detectable CIN1 Women with HPV infection or established cervical lesions can regress to normal, progress to HSILs or cervical cancer, or stay the same. Unique health states are defined to distinguish women with previously abnormal screening results, prior treatment for CIN, and detected cervical disease (through symptoms or screening). Women may die at any age due to a cervical cancer-related illness. CIN 2-3 Local Cancer Regional Cancer Distant Cancer Health states were defined using 4 categories of cervical health (normal, infection with human papillomavirus [HPV], grade of cervical intraepithelial neoplasia [CIN], and stage of invasive cancer). See Methods for details. Strategies We discuss 3 main strategies to manage ASC: (1) immediate colposcopy (all women receive colposcopy); (2) HPV DNA triage (colposcopy only if highrisk HPV types are detected); (3) repeat cytology (follow-up cytology at 6 and 12 months and referral for colposcopy for a repeat cytological result of ASC, LSIL, HSIL, or cancer). We also evaluate the implications of using a higher threshold on repeat cytology (eg, HSIL) for referral to colposcopy, and of a fourth strategy in which ASC is reclassified as normal (ie, a cytological result of ASC is ignored). We used the new 2001 Bethesda System, 17 which subcategorizes ASC into atypical squamous cells of undetermined significance (ASC-US) and atypical squamous cells that cannot exclude HSIL (ASC-H) for our base case. The ASC-H category constitutes approximately 5% of all ASC, but includes women at greatest risk for CIN ,19 We assumed that all women with ASC-H receive colposcopy, although alternative options are evaluated in separate analyses. Since the majority of published literature has not distinguished between ASC-H and ASC- US, we also conducted analyses in which ASC was not stratified. Testing of HPV DNA could be incorporated into the management of women with ASC-US in several ways. One is reflex HPV DNA testing using either the residual in a liquid-based cytological specimen or, in the case of conventional cytology, a sample co-collected at the time of the initial screening. Both approaches eliminate the need for a second examination to collect a sample. Another way to incorporate HPV DNA testing with conventional cytology is to have women with ASC-US return for a second visit within 1 month to collect a sample for HPV DNA testing (referred to as 2-visit HPV DNA triage). In all HPV DNA testing strategies, women who receive a positive test result for high-risk types of HPV are referred to colposcopy, while all others continue with routine cytological screening. For the base case we assume: (1) general screening begins at age 18 years and occurs every 2 years in the absence of a cytological abnormality; (2) colposcopy is performed for all cytological results of SIL but treatment is reserved for biopsy-confirmed CIN 2-3; (3) women who are treated for CIN 2-3, or have biopsy-confirmed CIN 1 return for a repeat cytological test every year; (4) women with ASC-US in whom no CIN is identified at colposcopy return to routine screening; (5) compliance with primary screening and follow-up is 100%; 2002 American Medical Association. All rights reserved. (Reprinted) JAMA, May 8, 2002 Vol 287, No

3 Table 1. Model Variables: Baseline Values and Ranges Used in Sensitivity and Specificity Analyses* Variable Base Case Plausible Range Natural History Probability of disease progression Normal to HPV DNA HPV DNA to CIN grade CIN grade 1 to grade CIN grade 2-3 to local invasive cancer Local invasive cancer to regional invasive cancer Regional invasive cancer to distant invasive cancer Probability of disease regression HPV DNA to normal CIN grade 1 to normal CIN grade 2-3 to normal Probability of cytology result 41,42 among women with abnormal cytology and CIN grade 2-3 Atypical squamous cells ASC-US ASC-H LSIL HSIL Year cancer survival rate 13 Local invasive cancer Regional invasive cancer Distant invasive cancer Annual probability of symptom detection 13 Local invasive cancer Regional invasive cancer Distant invasive cancer Test Characteristics 27,31-33,45-55 ThinPrep cervical cytology, % Sensitivity LSIL HSIL Specificity Conventional cervical cytology, % Sensitivity LSIL HSIL Specificity Repeat cervical cytology HPV DNA test, % Sensitivity LSIL HSIL Specificity 75/ (continued) and (6) disease status is confirmed by colposcopy and biopsy. The implications of alternative assumptions were evaluated in sensitivity analyses. We compared the different options within each possible screening frequency for both liquid-based and conventional cytology separately. Next, to identify the most efficient screening options not conditional on any particular screening interval or cytological technique, we simultaneously compared ASC-US management options, screening frequencies, and type of cytology. We adopted a societal perspective and followed the recommendations of the Panel on Cost-effectiveness in Health and Medicine. 20 Costs are expressed in 2000 US dollars and clinical benefits are expressed as years of life saved and quality-adjusted life-years (QALYs) gained. Future costs and life-years are discounted at an annual rate of 3%. The performance of alternative screening strategies is measured using the incremental cost-effectiveness ratio, defined as the additional cost of a specific screening strategy, divided by its additional clinical benefit, compared with the next most expensive strategy. Clinical Data Selected data used for the base case are summarized in TABLE ,21-61 To check the face validity of our model, we projected the age-specific prevalence of HPV, CIN, and invasive cervical cancer and compared these outputs with population-based data not used to create the model Model corroboration was assessed by comparing the projected lifetime risk of invasive cervical cancer with estimates from other published analytic models. 11,27,28,73 Natural History Our model requires parameters for the natural history of cervical disease conditional on a woman s HPV DNA status. Since some of the required model parameters were not available directly from the literature, average probabilities of progression and regression of CIN were derived using data that were not stratified by HPV status. 32,33 Using an independent mathematical model, we applied the relative risk of CIN with detectable HPV and the prevalence of detectable HPV to split the overall average probabilities of progression and regression of cervical lesions for women with detectable high-risk HPV DNA and women without detectable high-risk HPV DNA. To achieve consistency with the agespecific cross-sectional data reported in the literature, the model was first used to project a series of intermediate and long-term outcomes for which there were suitable data in the absence of screening (eg, lifetime risk and age-specific cancer incidence, stage distribution of inva JAMA, May 8, 2002 Vol 287, No. 18 (Reprinted) 2002 American Medical Association. All rights reserved.

4 sive disease). The model was then calibrated to the approximate distribution of detectable HPV within LSIL and HSIL. The calibration required (1) approximately 20% of acquired HPV in each monthly Markov cycle to be undetectable; (2) the probability of developing LSIL among women residing in the detectable and undetectable HPV sectors of the model to be the same; and (3) the transition of LSIL to HSIL among women residing in the undetectable HPV sector to be reduced by 80% relative to those in the detectable sector. Our model projected that approximately 5% of invasive cervical cancer would appear to be HPV-negative (ie, arising from the undetectable HPV sector). Screening Tests The distribution of abnormal cytological results based on cervical disease status was estimated using data from several different studies (Table 1). Data from the Columbia University Family Planning Colposcopy Clinic ( ) provided probabilities of each cytological test result (ASC-US, ASC-H, LSIL, HSIL), which were conditional on the underlying disease status (normal, CIN 1, CIN 2-3, invasive cancer) (T. C. W., written communication, September 6, 2001). We then applied the relative distributions for ASC-US and ASC-H to published data that were not stratified. 41,42 Estimates for the sensitivity and specificity of cervical cytology (when used for population screening) were obtained from recent published comprehensive reviews, 27,44,45 while those for repeat cervical cytology and HPV DNA testing of women with ASC-US were obtained from a formal literature review conducted for the American Society of Colposcopy and Cervical Pathology national consensus conference on management of cytological abnormalities. 4,5,7,8,74 Health-Related Quality of Life In the sensitivity analysis, the quality weights for the base case (and their plausible ranges) were 0.68 ( ) for local cancer, 0.56 ( ) for regional cancer, and 0.48 ( ) for distant Table 1. Model Variables: Baseline Values and Ranges Used in Sensitivity and Specificity Analyses* (cont) Variable Base Case Plausible Range Direct Medical Costs, US $ in ,27,56-61 ThinPrep cervical cytology Normal Papanicolaou test result Abnormal Papanicolaou test result with physician review Office visit Time cost Conventional cervical cytology Normal Papanicolaou test result Abnormal Papanicolaou test result with physician review Office visit Time cost HPV DNA test (Hybrid Capture II) Co-collection fee with conventional cytology Aggregate Costs, US $ in 2000# Colposcopy and biopsy CIN grade CIN grade Local invasive cervical cancer Regional invasive cervical cancer Distant invasive cervical cancer *HPV indicates human papillomavirus; CIN, cervical intraepithelial neoplasia; ASC-US, atypical squamous cells of undetermined significance; ASC-H, atypical squamous cells that cannot exclude a high-grade lesion; and LSIL, lowgrade squamous intraepithelial lesion; and HSIL, high-grade squamous intraepithelial lesion. Estimates are reported as monthly probabilities unless otherwise noted. Values are age-specific and are available from the authors on request. This value represents the average progression rate from undetectable and detectable HPV DNA to CIN grade 1. See Methods for details. Estimated using data from Kaiser Permanente. 41 See Methods for details. Sensitivity analyses were conducted using data from Columbia University Family Planning Colposcopy Clinic ( ; unpublished data), in which results for ASC-US were 46 (32.6%); ASC-H, 20 (14.2%); LSIL, 42 (29.8%); and HSIL, 33 (23.4%) among women with abnormal cytology and CIN grade 2-3. The same analyses performed by Jones and Novis 42 had the following results: 537 (10%) for ASC (unstratified), 1549 (29%) for LSIL, and 3258 (61%) for HSIL. LSIL sensitivity represents the probability of HPV-positive test result given CIN grade 1. HSIL sensitivity represents probability of HPV-positive test result given CIN grade 2-3. Specificity represents probability of HPV-negative test result given no CIN. Aggregate costs reflect the sum of the procedure, office visit, and a woman s time. The value of 75 was for women younger than 30 years; 85, women 30 years or older. #Aggregate costs reflect the sum of the procedure, office visit, and a woman s time. cancer. 75 Age-specific quality weights from the Beaver Dam Health Outcomes study 76 were used for noncancer states because these data were only available for women aged 40 years or older. We assumed a quality weight of 1.00 for women aged 39 years or younger. Costs Selected costs are shown in Table ,27,56-61 Direct medical costs included: screening (eg, Papanicolaou test, HPV test, treatment visit); clinician s fee(s); workup following an abnormal cytological result (eg, colposcopy) and any necessary treatment; and inpatient and/or ambulatory medical facility. Direct medical costs for screening and treatment were derived from 3 sources. 27,56-58 Cytological costs were estimated by having the model directly calculate a weighted average of normal and abnormal cervical cytological results since abnormal results require physician review and their evaluation is reimbursed at a higher rate. 57 For example, the total cost for liquid-based cytology when used for general screening ranged from $71 to $107 depending on whether physician review was required. We assumed a unit HPV DNA testing cost of $48.50 based on current Centers for Medicare and Medicaid Services (formerly the Health Care Financing Administration) reimbursement rate, but this var American Medical Association. All rights reserved. (Reprinted) JAMA, May 8, 2002 Vol 287, No

5 Figure 2. Prevalence of Human Papillomavirus (HPV) Prevalence of HPV, % Published Data Model Age, y The model s projected age-specific prevalence of HPV closely approximates a curve constructed by the weighted means of the prevalence reported in several clinical studies ied from $25 to $ For reflex HPV testing using liquid-based cytology, this cost was added to the cost of the initial screening visit, which included the office visit, screening cytology, and a woman s time. 59 For reflex HPV testing using conventional cytology, we included a $2 cost for the collection kit and handling of the specimen, but varied this cost from $1 to $10. For 2-visit HPV DNA testing, the cost of a second office visit and the additional time costs to the patient were included. The time spent undergoing screening was based on a prospective study of time costs associated with cervical cancer screening. 60 Previously published data incorporating time and transportation costs were used in sensitivity analysis. 9,10 To account for inflation, all costs were made constant by using dollar values from 2000 and by using the medical care component of the consumer price index. 61 RESULTS Face Validity of the Model The model predicts age-specific prevalence of HPV infection within a plausible range of results from studies using sensitive assays to detect HPV DNA (FIGURE 2) Prevalence for all CIN was also found to be within the range of other published estimates, with a peak prevalence of approximately 7.1% early in the third decade In the absence of screening, the lifetime risk of invasive cervical cancer was 3.3% and the peak annual incidence of cervical cancer was 67 per at age 57 years, similar to values reported in the literature. 11,27,28,73,80,81 Colposcopy referral rates were 52.5% with HPV testing, 49.7% with a single repeat cytological result of ASC-US or greater (LSIL, HSIL, or cancer), and 6.2% with a single repeat cytology result of HSIL. These results closely approximated those reported in published findings of the ASCUS/LSIL Triage Study. 8 Base-Case Analysis We first assessed the discounted costs and benefits associated with different approaches to managing ASC-US given a particular clinical screening practice (TABLE 2). In the context of biennial Table 2. Discounted Costs, Average Life Expectancy, and Incremental Cost-effectiveness of Alternative Approaches to Manage a Cytology Result of Atypical Squamous Cells Associated With Biennial Screening* Screening Strategy Total Average Lifetime Costs, $ Incremental Costs, $ Total Average Life Expectancy, y No screening Absolute Reduction in Cancer Incidence, % Cost-effectiveness Ratio $ Per Quality-Adjusted Life-Year $ Per Year of Life Saved Biennial screening using liquid-based cytology Ignore ASC-US Reflex HPV DNA testing Repeat cytology Dominated Dominated Immediate colposcopy Biennial screening using conventional cytology Ignore ASC-US HPV DNA testing Visit HPV DNA testing Dominated Dominated Repeat cytology Dominated Dominated Immediate colposcopy Biennial screening using conventional cytology# Ignore ASC HPV DNA testing Visit HPV DNA testing Dominated Dominated Repeat cytology Dominated Dominated Immediate colposcopy *HPV indicates human papillomavirus; ASC-US, atypical squamous cells of undetermined significance; and ASC, atypical squamous cells. Base case analyses were performed using data from Kaiser Permanente. 41 Costs expressed in dollars are based on US values. Incremental costs represent the difference between the strategy and the next best nondominated strategy. The difference in cost divided by the difference in quality-adjusted life expectancy or life expectancy for each strategy compared with the next best strategy. All strategies are assumed to be equally available. Incremental tests for dominated strategies. A strategy of repeat cervical cytology or delayed HPV testing cost more but was less effective than HPV DNA testing and was therefore dominated. This strategy assumes co-collection of a second specimen at the time of the initial Papanicolaou test, which is stored until the patient s cervical cytology result is available and used only if the initial cytology result is ASC. #Results reflect the traditional unstratified cytological category of ASC. See Methods for details JAMA, May 8, 2002 Vol 287, No. 18 (Reprinted) 2002 American Medical Association. All rights reserved.

6 screening using liquid-based cytology, the least costly strategy was to reclassify ASC-US as normal (ie, ignore a result of ASC-US), followed sequentially by HPV DNA testing, repeat cytology, and then immediate colposcopy. Compared with no screening, a strategy in which ASC-US was ignored increased discounted life expectancy by 32 days with a cost of $13700 per year of life gained. Compared with ignoring ASC- US, reflex HPV DNA testing was associated with an incremental life expectancy benefit of 2 days with a cost of $44400 per year of life gained. In comparison, repeat cytology was more expensive and less effective than HPV testing. Compared with HPV testing, immediate colposcopy provided an additional 2 hours in life expectancy and had an incremental cost-effectiveness ratio of $ per year of life gained. Results in which life expectancy was quality-adjusted were similar (Table 2). We repeated these analyses in the context of conventional cytology and evaluated these testing strategies: (1) reflex HPV testing using samples cocollected at the time of conventional cytology and (2) 2-visit HPV DNA testing in which a woman with an ASC-US result has to return for an HPV test. Provided that compliance with the repeat visit was 100%, the 2-visit HPV DNA testing strategy was just as effective as reflex HPV testing using co-collected samples. As estimates of compliance with the second visit decreased, the reflex HPV testing strategy using cocollected samples became more effective. As expected, under the base-case assumptions, the 2-visit HPV DNA testing was more costly than reflex HPV testing using co-collected samples. To quantify the trade-off between the cost of co-collection vs the cost of a second visit, we determined the monetary threshold at which the cost-effectiveness ratios for the 2 HPV testing strategies converge. We found that unless the cost of a repeat visit was less than $32 (including the cost of medical staff, office visit, transportation, and patient time), the co-collection strategy would be preferred. Figure 3. Discounted Costs and Life Expectancy of Alternative Management Strategies for Screening With Liquid-Based or Conventional Cytology Discounted Life Expectancy, y Co-collect HPV $12 100/YLS Ignore ASC $7100/YLS $20 300/YLS No Screening $59 600/YLS Results were similar for analyses in which the categories of ASC-US and ASC-H were not segregated but grouped under the single traditional cytological diagnosis of ASC (lower third of Table 2). HPV DNA testing with biennial screening using conventional cytology had a cost of $24700 per year of life gained and immediate colposcopy exceeded $ per year of life gained, making it more expensive than other strategies. The life-expectancy gains associated with all strategies except immediate colposcopy were somewhat lower than in the base case; however, these differences were almost immeasurable. Stratification of ASC had a significant impact only in the context of ignoring an ASC result. Biennial screening with conventional cytology provided a 79% reduction in cancer incidence using the new stratified terminology (ASC-US and ASC-H) compared with a 75% reduction using a single cytological category of ASC. $ /YLS $ /YLS Frequency of Screening, y eg, Liquid-Based, Ignore ASC Repeat Cytology $1.8 million/yls Type of Cytology Screening Strategy Conventional Liquid-Based Ignore ASC Colposcopy Repeat Cytology 2-Visit HPV Testing Discounted Lifetime Costs, $ YLS indicates years of life saved; ASC, atypical squamous cells; and HPV, human papillomavirus. FIGURE 3 shows the relationship between the discounted costs and life expectancies for all 37 potential screening strategies. This analysis assumes all strategies are available and are feasible options. The cost-effectiveness of moving from one screening strategy to a more costly alternative is represented by the difference in cost divided by the difference in life expectancy associated with the 2 strategies (this is expressed as a costeffectiveness ratio and is shown for each nondominated strategy). Strategies on the efficiency curve dominate those to the right of the curve because they are more effective, and either cost less or have a more attractive cost-effectiveness ratio than the next best strategy. Every 5-year screening using conventional cervical cytology and ignoring an ASC-US result had a costeffectiveness ratio of $7100 per year of life gained compared with no screening. Every 5-year screening with reflex HPV DNA testing yielded an incremental cost-effectiveness ratio of 2002 American Medical Association. All rights reserved. (Reprinted) JAMA, May 8, 2002 Vol 287, No

7 $12100 per year of life gained compared with ignoring ASC-US. Every 5-year screening with liquid-based cytology and reflex HPV DNA testing was substantially more effective and cost $20300 per year of life gained. The costs for the same strategy conducted per year of life gained are $59600 for every 3 years and $ for every 2 years. All strategies that relied on annual screening, regardless of the ASC-US management option, provided minimal incremental life-expectancy gains and had cost-effectiveness ratios exceeding $ per year of life gained. The inconvenience of co-collection may preclude its widespread adoption in settings where liquid-based cytology is considered too expensive for routine use. However, a longer screening interval of 3 years using the liquid-based cytology strategy of reflex HPV testing is more effective and less costly than a screening interval of every 2 years using the conventional cytology strategies of repeat cytology or immediate colposcopy. Sensitivity Analyses These general results were not sensitive to analyses of screening test performance, alternative management options for CIN 1, cost of the workup for an abnormal screening test result, and the costs of CIN 2-3 and cervical cancer treatment. Repeat cytology was no longer dominated when the cost of HPV DNA testing exceeded $190. The rank ordering of strategies did not change despite varying the costs of colposcopy of between $200 and $600. When noncompliance was assumed to be random (ie, all women are equally likely to be noncompliant), lower rates of compliance resulted in lower costs and more attractive cost-effectiveness ratios. For example, when compliance was decreased to 60% for each return visit in the strategies of repeat cytology, reflex HPV testing no longer dominated repeat cytology because the cost savings associated with losing patients to follow-up was relatively greater than the loss in clinical benefits. When noncompliance was assumed to be systematic (ie, women at greater risk for cervical cancer are more likely to be noncompliant), this counterintuitive result was somewhat blunted lower rates of compliance were still associated with lower costs, but the loss in clinical benefits resulting from women at high risk being lost to follow-up were greater than when women were at average risk. In both situations, however, HPV DNA testing remained more effective than repeat cytology and had an attractive costeffectiveness ratio of between $19000 and $29000 per year of life gained compared with repeat cytology strategy when using biennial screening with liquid cytology. This pattern was similar for noncompliance with screening, follow-up visits, and colposcopy. COMMENT Managing the approximately 2 million US women with equivocal cervical cytological results each year is a significant clinical and public health problem. 82 This year a national consensus conference was held to develop evidence-based guidelines for the management of women with cervical cytological abnormalities. 74 To provide information for the guideline process, we conducted a comprehensive cost-effectiveness analysis of alternative strategies for women with ASC. We first assessed the best option for ASC-US management given a particular clinical screening practice, since all possible strategies evaluated in this article may not be universally accessible and feasible. Our results indicate that a policy of ignoring a cytological result of ASC-US reduces the effectiveness of cervical cancer screening compared with HPV testing, repeat cytology, or immediate colposcopy, especially when longer screening intervals or less sensitive screening methods are used. There are extremely small differences between the 3 ASC-US management strategies in terms of cancer incidence reduction, although under all conditions repeat cervical cytology is somewhat less effective than either HPV DNA testing or immediate colposcopy. In contrast, the costs associated with ASC-US management strategies differed substantially. Immediate colposcopy was always more costly than either repeat cervical cytology or HPV DNA testing. Regardless of the clinical scenario evaluated, referral of all women with ASC-US for colposcopy had incremental cost-effectiveness ratios exceeding $ per year of life gained compared with HPV DNA testing. Reflex HPV DNA testing was always less costly than repeat cytology, in part because it eliminates the need for a repeat clinical examination to obtain a cervical specimen in the case of ASC-US and reduces the number of colposcopic examinations by 40% to 60% compared with immediate colposcopy. Our results were not affected by reasonable changes in screening costs, performance of HPV testingandcytology,andtimecosts.ourbase case incorporated the new 2001 Bethesda System that subcategorizes ASC into ASC-US and ASC-H. Recent clinical practice guidelines recommend that since women with ASC-H are at increased risk for having CIN 2-3, they should have immediate colposcopy. However, our analyses indicate that subcategorizing ASC makes little difference in terms of clinical benefits, provided some follow-up of ASC-US occurs. We then assessed the most efficient screening options by considering alternative strategies to manage ASC-US while simultaneously varying all possible screening frequencies and type of cytological tests. This was done from a broad health policy perspective and assuming all strategies are equally available and feasible. Similar to others, we found that all strategies that rely on annual screening, regardless of the ASC-US management option, provide minimal incremental life-expectancy gains compared with less frequent screening strategies and have extremely high and unattractive cost-effectiveness ratios ( $ per year of life gained). 27,28,73,83-86 A biennial screening program using liquid-based cytology combined with reflex HPV DNA testing for women with ASC-US is more effective and less costly than annual conventional cytology coupled with either repeat cervical cytology or colposcopy for women with ASC-US. However, the cost JAMA, May 8, 2002 Vol 287, No. 18 (Reprinted) 2002 American Medical Association. All rights reserved.

8 effectiveness ratio of this strategy ($ per year of life gained) is higher than many clinical preventive interventions that have been adopted in the United States. 87 There is no consensus that defines the cost per QALY and represents an acceptable value for money, however, cost-effectiveness ratios are often placed in context by comparisons with interventions that are widely mandated, such as hemodialysis for endstage renal disease of between $60000 and $ per year of life gained. 88 In our model, screening performed every 3 years, using liquid-based cytology coupled with reflex HPV DNA testing, has an incremental cost-effectiveness ratio of approximately $60000 per year of life gained over screening performed every 5 years and would be considered a cost-effective intervention. It is important to emphasize, however, that in the context of current cervical cancer screening practices in the United States, shifting women currently being screened annually with conventional cytology coupled with repeated cytology for an ASC-US result to biennial liquid-based cytology and reflex HPV DNA testing would save more than $15 billion over the lifetime of a typical cohort of 18-yearold to 24-year-old women. 89 Our analysis has a number of limitations. There is considerable uncertainty with respect to the longitudinal nature of HPV infection and our results may be further refined as better data become available. There is also considerable variability with respect to the distribution of cytological results in women with CIN 2-3, although our results did not change significantly despite using parameters from 3 different clinical studies Unknown factors that contribute to population heterogeneity and possible cofactors that contribute to an individual s risk of disease progression could not be modeled, 90 although we did try to explore the implications of certain heterogeneous behaviors, such as those reflecting noncompliance. 91 This analysis provides important insights for both policy makers developing management guidelines for women with ASC results, as well as for clinicians caring for these women. In interpreting these findings, it is important to recognize that not only are women s relative preferences for a potential outcome important when deciding the best option in a given clinical setting, but other factors must also be considered. 92 For example, there are settings in which liquidbased cytology is already the standard of care, and others in which conventional cytology is the only option. Some settings will find co-collection suited to their daily operational systems and others will not. However, many of our key findings were robust in a variety of settings. First, in the context of current cervical cancer screening practice in the United States, follow-up for ASC-US provides clinical benefits at a reasonable cost. Second, under the conditions of our base case, reflex HPV DNA testing appears to provide clinical benefits similar to those of immediate colposcopy for women with ASC-US, but is less costly. Finally, a triennial or biennial cervical cancer screening program incorporating liquidbased cytology and reflex HPV DNA testing is more effective and less costly than annual screening with conventional cytology and repeat cytology for women with ASC. Thus, the aggregate health economic costs saved by adopting such a policy would be enormous. Author Affiliations: Department of Health Policy and Management, Harvard Center for Risk Analysis, Harvard School of Public Health, Boston, Mass (Ms Kim and Dr Goldie) and the Department of Pathology, College of Physicians and Surgeons of Columbia University, New York, NY (Dr Wright). Author Contributions: Study concept and design: Kim, Wright, Goldie. Acquisition of data: Wright. Analysis and interpretation of data: Kim, Wright, Goldie. Drafting of the manuscript: Kim, Wright, Goldie. Critical revision of the manuscript for important intellectual content: Kim, Wright, Goldie. Statistical expertise: Kim, Goldie. Administrative, technical, or material support: Goldie. Study supervision: Goldie. Financial Disclosure: Dr Wright was the principal investigator for clinical trials investigating HPV DNA testing and liquid-based cytology funded by Digene Corp and Cytyc Corp via formal grants to Columbia University. He currently serves on the speaker s bureaus of Cytyc Corp and Tripath Inc, which make liquidbased cytology test kits. REFERENCES 1. Jones BA, Davey DD. Quality management in gynecologic cytology using interlaboratory comparison. Arch Pathol Lab Med. 2000;124: Kurman RJ, Henson DE, Herbst AL, et al. Interim guidelines for management of abnormal cervical cytology. JAMA. 1994;271: Kinney WK, Manos MM, Hurley LB, Ransley JE. Where s the high-grade cervical neoplasia? Obstet Gynecol. 1998;91: Wright TC, Sun XW, Koulos J. Comparison of management algorithms for the evaluation of women with low-grade cytologic abnormalities. Obstet Gynecol. 1995;85: Ferris DG, Wright TC Jr, Litaker MS, et al. Triage of women with ASCUS and LSIL on Pap smear reports. J Fam Pract. 1998;46: Manos MM, Kinney WK, Hurley LB, et al. Identifying women with cervical neoplasia. JAMA. 1999; 281: Wright TC Jr, Lorincz A, Ferris DG, et al. Reflex HPV DNA testing in women with abnormal Papanicolaou smears. Am J Obstet Gynecol. 1998;178: Solomon D, Schiffman M, Tarrone R. Comparison of three management strategies for patients with atypical squamous cells of undetermined significance. J Natl Cancer Inst. 2001;93: Goldie SJ, Weinstein MC, Kuntz KM, Freedberg KA. The costs, clinical benefits, and cost-effectiveness of screening for cervical cancer in HIV-infected women. Ann Intern Med. 1999;130: Goldie SJ, Freedberg KA, Weinstein MC, et al. Cost effectiveness of human papillomavirus testing to augment cervical cancer screening in women infected with the human immunodeficiency virus. Am J Med. 2001; 111: Goldie SJ, Kuhn L, Denny L, et al. Policy analysis of cervical cancer screening strategies in lowresource settings. JAMA. 2001;285: Wright TC, Ferenczy AF, Kurman RJ. Precancerous lesions of the cervix. In: Kurman RJ, ed. Blaustein s Pathology of the Female Genital Tract. 4th ed. New York, NY: Springer-Verlag; 1994: Surveillance, Epidemiology, End Results (SEER) Cancer Statistics Review, Bethesda, Md: National Cancer Institute; Available at: www-seer.ims.nci.nih.gov. Accessed August 1, Schiffman M. New epidemiology of human papillomavirus infection and cervical neoplasia. J Natl Cancer Inst. 1995;87: Koutsky LA, Kiviat NB. Genital human papillomavirus. In: Holmes K, Mardh P, Sparling P, et al, eds. Sexually Transmitted Disease. 3rd ed. New York, NY: McGraw-Hill; 1999: Schlecht NF, Kulaga S, Robitaille J, et al. Persistant human papillomavirus infection as a predictor of cervical intraepithelial neoplasia. JAMA. 2001;286: Solomon D, Davey DD, Moriarty A, et al. Terminology for Reporting the Results of Cervical Cytology. Bethesda, Md: National Cancer Institute; Available at: Accessed February 1, Sherman ME, Tabbara SO, Scott DR, et al. ASCUS, rule out HSIL. Mod Pathol. 1999;12: Malik SN, Wilkinson EJ, Drew PA, et al. Do qualifiers of ASCUS distinguish between low- and highrisk patients? Acta Cytol. 1999;43: Gold MR, Siegel JE, Russel LB, Weinstein MC, eds. Cost-effectiveness in Health and Medicine. New York, NY: Oxford University Press; Hildesheim A, Schiffman MH, Gravitt PE, et al. Persistence of type-specific human papillomavirus infection among cytologically normal women. J Infect Dis. 1994;169: Ho GYF, Bierman R, Beardsley L, et al. Natural history of cervicovaginal papillomavirus infection in young women. N Engl J Med. 1998;338: Liu T, Seng-jaw S, Alvarez R, Butterworth C Jr. A longitudinal analysis of human papillomavirus 16 infection, nutritional status, and cervical dysplasia progression. Cancer Epidemiol Biomarkers Prev. 1995; 4: Cuzick J. Human papillomavirus testing for cer American Medical Association. All rights reserved. (Reprinted) JAMA, May 8, 2002 Vol 287, No

9 ical screening. Health Technol Assess. 1999;3: Lawson HW, Lee NC, Thames SF, et al. Cervical cancer screening among low-income women. Obstet Gynecol. 1998;92: Sun XW, Kuhn L, Ellerbrock TV, Chiasson MA, Bush TJ, Wright TC Jr. Human papillomavirus infection in women infected with the human immunodeficiency virus. N Engl J Med. 1998;337: McCrory D, Mather D, Bastain L, et al. Evaluation of Cervical Cytology: Evidence Report/ Technology Assessment No.5. Rockville, Md: Agency for Health Care Policy and Research; AHCPR publication 99-E Myers E, McCrory D, Nanda K, Matchar D. Mathematical model for the natural history of human papillomavirus infection and cervical carcinogenesis. Am J Epidemiol. 2000;151: Londesborough P, Ho L, Terry G, Cuzick J, Wheeler C, Singer A. Human papillomavirus genotype as a predictor of persistence and development of high-grade lesions in women with minor cervical abnormalities. Int J Cancer. 1996;69: Syrjanen K, Kataja V, Yliskoski M, et al. Natural history of cervical human papillomavirus lesions does not substantiate the biologic relevance of the Bethesda System. Obstet Gynecol. 1992;79: Ho GY, Burk RD, Klein S, et al. Persistent genital human papillomavirus infection as a risk factor for persistent cervical dysplasia. J Natl Cancer Inst. 1995;87: Ostor AG. Natural history of cervical intraepithelial neoplasia. Int J Gynecol Pathol. 1993;12: Melnikow J, Nuovo J, Willan AR, et al. Natural history of cervical squamous intraepithelial lesions. Obstet Gynecol. 1998;92(4 pt 2): Ho GY, Kadish AS, Burk RD, et al. HPV 16 and cigarette smoking as risk factors for high-grade cervical intra-epithelial neoplasia. Int J Cancer. 1998;78: Hopman EH, Rozendaal L, Voorhorst FJ, et al. High risk human papillomavirus in women with normal cervical cytology prior to the development of abnormal cytology and colposcopy. BJOG. 2000;107: Rozendaal L, Walboomers JM, van der Linden JC, et al. PCR-based high-risk HPV test in cervical cancer screening gives objective risk assessment of women with cytomorphologically normal cervical smears. Int J Cancer. 1996;68: Ellerbrock TV, Chiasson MA, Bush TJ, et al. Incidence of cervical squamous intraepithelial lesions in HIV-infected women. JAMA. 2000;283: Koutsky LA, Holmes KK, Critchlow CW, et al. A cohort study of the risk of cervical intraepithelial neoplasia grade 2 or 3 in relation to papillomavirus infection. N Engl J Med. 1992;327: Nobbenhuis MA, Walboomers JM, Helmerhorst TJ, et al. Relation of human papillomavirus status to cervical lesions and consequences for cervical-cancer screening: a prospective study. Lancet. 1999;354: Remmink AJ, Walboomers JM, Helmerhorst TJ, et al. The presence of persistent high-risk HPV genotypes in dysplastic cervical lesions is associated with progressive disease. Int J Cancer. 1995;61: Lonky NM, Sadeghi M, Tsadik GW, Petitti D. The clinical significance of the poor correlation of cervical dysplasia and cervical malignancy with referral cytologic results. Am J Obstet Gynecol. 1999;181: Jones BA, Novis DA. Follow-up of abnormal gynecologic cytology. Arch Pathol Lab Med. 2000;124: Statistical Abstract of the United States: Consumer Price Index: All Urban Consumers. Washington, DC: National Center for Health Statistics; Available at: /gm291_1.pdf. Accessed September 1, Fahey MT, Irwig L, Macaskill P. Meta-analysis of Pap test accuracy. Am J Epidemiol. 1995;141: Nanda K, McCrory DC, Myers ER, et al. Accuracy of the Papanicolaou test in screening for and follow-up of cervical cytologic abnormalities: a systematic review. Ann Intern Med. 2000;132: Bergeron C, Jeannel D, Poveda J, Cassonnet P, Orth G. Human papillomavirus testing in women with mild cytologic atypia. Obstet Gynecol. 2000;95: Ferris DG, Wright TC Jr, Litaker MS, et al. Comparison of two tests for detecting carcinogenic HPV in women with Papanicolaou smear reports of ASCUS and LSIL. J Fam Pract. 1998;46: Lin CT, Tseng CJ, Lai CH, et al. High-risk HPV DNA detection by Hybrid Capture II. J Reprod Med. 2000; 45: Shlay JC, Dunn T, Byers T, Baron AE, Douglas JM Jr. Prediction of cervical intraepithelial neoplasia grade 2-3 using risk assessment and human papillomavirus testing in women with atypia on Papanicolaou smears. Obstet Gynecol. 2000;96: Hutchinson ML, Zahniser DJ, Sherman ME, et al. Utility of liquid-based cytology for cervical carcinoma screening. Cancer. 1999;87: Diaz-Rosario LA, Kabawat SE. Performance of a fluid-based, thin-layer Papanicolaou smear method in the clinical setting of an independent laboratory and an outpatient screening population in New England. Arch Pathol Lab Med. 1999;123: Ashfaq R, Gibbons D, Vela C, et al. ThinPrep Pap test. Acta Cytol. 1999;43: Bai H, Sung CJ, Steinhoff MM. ThinPrep Pap test promotes detection of glandular lesions of the endocervix. Diagn Cytopathol. 2000;23: Quddus MR, Sung CJ, Steinhoff MM, et al. Atypical squamous metaplastic cells. Cancer. 2001;93: Franco EL, Ferenczy A. Assessing gains in diagnostic utility when human papillomavirus testing is used as an adjunct to Papanicolaou smear in the triage of women with cervical cytologic abnormalities. Am J Obstet Gynecol. 1999;181: Helms LJ, Melnikow J. Determining costs of health care services for cost-effectiveness analyses. Med Care. 1999;37: National physician fee schedule relative value file: Medicare fee schedule. Available at: Accessed October 1, Muller C, Mandelblatt J, Schechter C. The Costs and Effectiveness of Screening for Cervical Cancer in Elderly Women. Washington, DC: Office of Technology Assessment; clinical diagnostic laboratory fee schedule: public use file. Available at: /stats/pufiles.htm#labfee. Accessed October 1, Shireman TI, Tsevat J, Goldie SJ. Time costs associated with cervical cancer screening. Int J Technol Assess Health Care. 2001;17: US Bureau of Labor Statistics. Statistical Abstract of the United States. Washington, DC: US Government Printing Office; Publication Moscicki AB. Human papillomavirus infection in adolescents. Pediatr Clin North Am. 1999;46: Burk RD, Kelly P, Feldman J, et al. Declining prevalence of cervicovaginal human papillomavirus infection with age is independent of other risk factors. Sex Transm Dis. 1996;23: Bauer HM, Hildesheim A, Schiffman MH, et al. Determinants of genital human papillomavirus infection in low-risk women in Portland, Oregon. Sex Transm Dis. 1993;20: Melkert PJ, Hopman E, van den Brule J, et al. Prevalence of HPV in cytomorphlogically normal cervical smears, as determined by the polymerase chain reaction, is age-dependent. Int J Cancer. 1993;53: Burk RD, Ho GYF, Beardsley L, Lempa M, Peters M, Bierman R. Sexual behavior and partner characteristics are the predominant risk factors for genital human papillomavirus infection in young women. JInfect Dis. 1996;174: Kotloff KL, Wasserman SS, Russ K, et al. Detection of genital human papillomavirus and associated cytological abnormalities among college women. Sex Transm Dis. 1998;25: Sellors JW. Cervical cancer prevention for all Canadians. Can Fam Physician. 1999;45: , Moscicki AB, Shiboski S, Broering J, et al. The natural history of human papillomavirus infection as measured by repeated DNA testing in adolescent and young women. J Pediatr. 1998;132: Franco E, Villa L, Sobrinho J, et al. Epidemiology of acquisition and clearance of cervical human papillomavirus infection in women from a high-risk area for cervical cancer. J Infect Dis. 1999;180: Gustafsson L, Adami HO. Optimization of cervical cancer screening. Cancer Causes Control. 1992; 3: Ponten J, Adami HO, Bergstrom R, et al. Strategies for global control of cervical cancer. Int J Cancer. 1995;60: Sherlaw-Johnson C, Gallivan S, Jenkins D. Evaluating cervical cancer screening programmes for developing countries. Int J Cancer. 1997;72: Wright TC Jr, Cox JT, Massad LS, et al consensus guidelines for the management of women with cervical cytological abnormalities. JAMA. 2002;287: Gold M, Franks P, McCoy K, Fryback DG. Toward consistency in cost-utility analyses. Med Care. 1998;36: Fryback D, Dasbach E, Klein R, et al. Initial catalog of health-state quality factors. Med Decis Making. 1993;13: Sawaya GF, Kerlikowske K, Lee NC, Gildengorin G, Washington AE. Frequency of cervical smear abnormalities within 3 years of normal cytology. Obstet Gynecol. 2000;96: Results from the National Breast and Cervical Cancer Early Detection Program, October 31, September 30, MMWR Morb Mortal Wkly Rep. 1994;43: Mount SL, Papillo JL. A study of 10,296 pediatric and adolescent Papanicolaou smear diagnoses in northern New England. Pediatrics. 1999;103: Gustafsson L, Ponten J, Bergstrom R, Adami HO. International incidence rates of invasive cervical cancer before cytological screening. Int J Cancer. 1997; 71: Pettersson F, Bjorkholm E, Naslund I. Evaluation of screening for cervical cancer in Sweden. Int J Epidemiol. 1985;14: Cox JT, Wilkinson EJ, Lonky N, et al. Management guidelines for the follow-up of atypical squamous cells of undetermined significance (ASCUS). J Lower Genital Tract. 2000;4: Brown AD, Garber AM. Cost-effectiveness of 3 methods to enhance the sensitivity of Papanicolaou testing. JAMA. 1999;281: Eddy DM. Screening for cervical cancer. Ann Intern Med. 1990;113: Fahs MC, Plichta SB, Mandelblatt JS. Costeffective policies for cervical cancer screening. Pharmacoeconomics. 1996;9: Fahs MC, Mandelblatt J, Schechter C, Muller C. Cost-effectiveness of cervical cancer screening for the elderly. Ann Intern Med. 1992;117: Graham JD, Corso PS, Morris JM, et al. Evaluating the cost-effectiveness of clinical and public health measures. Annu Rev Public Health. 1998;19: Weinstein MC. High-priced technology can be good value for money. Ann Intern Med. 1999;130: Population Estimates Program. Washington, DC: US Census Bureau; Available at: Accessed August 24, Kuntz KM, Goldie SJ. Defining the effects of population heterogeneity in Markov models. Med Decis Making. 2000;20: Goldie SJ, Kuntz KM. Modeling compliance in costeffectiveness analyses of screening programs. Med Decis Making. 2001;20: Barry M. Health decision aids to facilitate shared decision making in office practice. Ann Intern Med. 2002;136: JAMA, May 8, 2002 Vol 287, No. 18 (Reprinted) 2002 American Medical Association. All rights reserved.