00006250-201011000-00023

Surveillance After Treatment for CervicalIntraepithelial NeoplasiaOutcomes, Costs, and Cost-Effectiveness

Joy Melnikow, MD, MPH, Shalini Kulasingam, PhD, Christina Slee, MPH, L. Jay Helms, PhD,Miriam Kuppermann, PhD, MPH, Stephen Birch, DPhil, Colleen E. McGahan, MSc,Andrew Coldman, PhD, Benjamin K. S. Chan, MS, and George F. Sawaya, MD

OBJECTIVE: To estimate outcomes and costs of surveil-lance strategies after treatment for high-grade cervicalintraepithelial neoplasia (CIN).

METHODS: A hypothetical cohort of women was evalu-ated after treatment for CIN 2 or 3 using a Markov modelincorporating data from a large study of women treatedfor CIN, systematic reviews of test accuracy, and individ-ual preferences. Surveillance strategies included initialconventional or liquid-based cytology, human papillo-mavirus testing, or colposcopy 6 months after treatment,followed by annual or triennial cytology. Estimated out-comes included CIN, cervical cancer, cervical cancerdeaths, life expectancy, costs, cost per life-year, and costper quality-adjusted life-year.

RESULTS: Conventional cytology at 6 and 12 months,followed by triennial cytology, was least costly. Com-pared with triennial cytology, annual cytology follow-upreduced expected cervical cancer deaths by 73% to 77%and had an average incremental cost per life-year gainedof $69,000 to $81,000. For colposcopy followed by annualcytology, the incremental cost per life-year gainedranged from $70,000 to more than $1 million, dependingon risk. Between-strategy differences in mean additionallife expectancy per woman were less than 4 days; differ-ences in mean incremental costs per woman were as highas $822. In the cost-utility analysis, colposcopy at 6months followed by annual cytology had an incrementalcost per quality-adjusted life-year of less than $5,500.Human papillomavirus testing or liquid-based cytologyadded little to no improvement to life-expectancy withhigher costs.

CONCLUSION: Annual conventional cytology surveil-lance reduced cervical cancers and cancer deaths com-pared with triennial cytology. For high risk of recurrence,a strategy of colposcopy at 6 months increased lifeexpectancy and quality-adjusted life expectancy. Humanpapillomavirus testing and liquid-based cytology in-creased costs, but not effectiveness, compared withtraditional approaches.(Obstet Gynecol 2010;116:1158–70)

Each year in the United States an estimated 500,000women have cervical intraepithelial neoplasia

(CIN) grade 2 and 3 diagnosed.1 Detection andtreatment of these high-grade precancerous lesionshave been accompanied by large reductions in cervi-cal cancer incidence and mortality in many countrieswith widespread cytology-based screening programs.2

After treatment of CIN, follow-up surveillancestrategies must strike a balance between the detectionand treatment of persistent or incident lesions and theoveruse of costly and invasive diagnostic procedures

From the Center for Healthcare Policy and Research and the Department ofEconomics, University of California, Davis, California; the Division of Epide-miology and Community Health, School of Public Health, University ofMinnesota, Minneapolis, Minnesota; the Departments of Obstetrics, Gynecology& Reproductive Sciences and Epidemiology & Biostatistics, University ofCalifornia, San Francisco, San Francisco, California; the Department ofClinical Epidemiology and Biostatistics, McMaster University, Hamilton, On-tario, Canada; the Surveillance and Outcomes Unit and Population Oncology,British Columbia Cancer Agency, Vancouver, British Columbia, Canada; andthe Oregon Evidence-based Practice Center, Oregon Health and Science Univer-sity, Portland, Oregon.

Supported by National Cancer Institute grant R01 CA109142.

Presented at the 25th International Papillomavirus Conference, May 8–14,2009, Malmo, Sweden.

Corresponding author: Joy Melnikow, MD, MPH, Department of Family andCommunity Medicine, Center for Healthcare Policy and Research, University ofCalifornia, Davis, 4860 Y St, Suite 2300, Sacramento, CA 95817; e-mail:[email protected].

Financial DisclosureDr. Kulasingam has previously received research support from Merck andCSL-Australia. She has also served as a consultant for CSL-New Zealand andSanofi-Pasteur MSD and is currently a consultant for Medtronic. The otherauthors did not report any potential conflicts of interest.

© 2010 by The American College of Obstetricians and Gynecologists. Publishedby Lippincott Williams & Wilkins.ISSN: 0029-7844/10

1158 VOL. 116, NO. 5, NOVEMBER 2010 OBSTETRICS & GYNECOLOGY

and tests. Evidence on the patterns of recurrence ofCIN after treatment indicates an elevated risk forapproximately 6 years after treatment, with the levelof risk depending on initial CIN grade and treatmenttype.3 The risk of invasive cervical cancer, althoughstill low, remains elevated for many years after treat-ment for CIN.3,4 Current recommendations by theAmerican College of Obstetricians and Gynecologistsand by the American Society of Colposcopy andCervical Pathology suggest that after treatmentwomen may be followed-up with a combination ofcytology, colposcopy, and human papillomavirus(HPV) testing.5,6 Specifically, both guidelines suggestthat surveillance can consist of either a single test foroncogenic HPV at 12 months or cytology testing at 6and 12 months; American Society of Colposcopy andCervical Pathology additionally endorses a combina-tion of cytology and colposcopy at 6-month intervals.If all test results are normal, then routine screening forat least 20 years is advised. If any test result isabnormal, then colposcopy is recommended. Themeaning of “routine” screening had been unspecified;recently released American College of Obstetriciansand Gynecologists recommendations suggest fol-low-up with annual cytology.7

Although many options for posttreatment surveil-lance are proposed, the precise ordering of tests andperiodicity of testing that maximize benefits andmake the most productive use of resources has notbeen determined. Given the large number of womendiagnosed with CIN 2 or 3, recommendations forposttreatment follow-up will affect hundreds of thou-sands of women each year in the United States. Weconducted a cost-effectiveness analysis of strategiesfor follow-up of women after treatment for CIN, takinginto consideration recurrence risk based on initial CINgrade and treatment type as well as the effect of thesestrategies on health-related quality-of-life.

MATERIALS AND METHODSThis study was reviewed and approved by the Uni-versity of California Davis Institutional ReviewBoard. A state-transition Markov model was devel-oped to predict outcomes after treatment of CIN. Weevaluated 12 different surveillance strategies to detectsubsequent disease in women who had previouslybeen treated for CIN 2 or 3 (Box 1). Four strategiesinvolving liquid-based cytology were subsequentlyexcluded from the analysis, leaving eight remainingstrategies. Model outcomes included recurrent (resid-ual or incident) CIN 1, 2, and 3, invasive cervicalcancer, incidental hysterectomy, death from cervicalcancer, and death from all other causes. Rates of

false-positive test results were estimated for eachstrategy, based on test specificity. Costs were assessedfrom the payer perspective, and estimates of out-comes and program costs were projected for a cohortof 500,000 women. Projected life expectancy, quality-adjusted life expectancy, total lifetime health care costsfor surveillance and treatment of CIN, and incrementalcosts and effects associated with each strategy weretabulated. Incremental cost-effectiveness ratios that mea-sure the average cost per life-year or quality adjustedlife-year were calculated, in which the lowest cost pro-gram was used as the baseline, and each alternativestrategy was compared with the next lower cost strategy.Strategies found to be more costly with no increase ineffectiveness or less costly but less cost-effective than analternative strategy were considered dominated by otherstrategies and excluded from further analysis. Healthoutcomes and costs occurring in the future were dis-counted at 3% annually for the base case.8

The model simulated a cohort of women at age30 just after treatment for CIN 2 or 3 and followedthem in 6-month cycles until age 85 years. Initialsurveillance testing and follow-up of recurrent CINoccurred in 6-month cycles. Women progressedthrough the different health states of the model (eg, nodisease, CIN 1, CIN 2 or 3, cancer [stages I–IV])

Box 1. Surveillance Strategies Evaluated in theMarkov Model

Conventional cytology at 6 and 12 months, then annuallyConventional cytology at 6 and 12 months, then trienniallyLiquid-based cytology at 6 and 12 months, then annually*Liquid-based cytology at 6 and 12 months, then triennially*HPV testing† at 6 months, cytology at 12 months, thenannual cytologyHPV testing† at 6 months, cytology at 12 months, thentriennial cytologyHPV testing† and cytology at 6 months, cytology at 12months, then annual cytologyHPV testing† and cytology at 6 months, cytology at 12months, then triennial cytologyHPV testing† with liquid-based cytology at 6 months,cytology at 12 months, then annual cytology*HPV testing† with liquid-based cytology at 6 months,cytology at 12 months, then triennial cytology*Colposcopy at 6 months, then annual cytologyColposcopy at 6 months, then triennial cytology

* Strategies involving liquid-based cytology subsequently wereeliminated from the base-case model.

† HPV testing performed with Hybrid Capture 2.

VOL. 116, NO. 5, NOVEMBER 2010 Melnikow et al Surveillance After Treatment for CIN 1159

based on estimates of residual disease after treatmentand incident CIN. Residual disease probabilities werestratified by type of treatment (loop excision or conebiopsy of the cervix, laser ablation or excision, orcryotherapy) and initial CIN grade (CIN 2 or 3). Anoverview of the model structure is shown in Figure 1.Women with residual or incident CIN could havetheir disease regress, progress, or persist. Women withcancer (stages I, II, III, IV) could have their cancerdetected during screening or if they presented to ahealth care provider because of symptoms. Womenwho did not have their disease detected couldprogress to the next stage, remain in the same stage,or die of cervical cancer. Each year, women also facedan age-specific risk of dying from other causes orundergoing hysterectomy for reasons other than cer-vical cancer. We assumed that women with a historyof treatment for CIN 2 or 3 would undergo total andnot supracervical hysterectomy.9

We assessed outcomes with and without qualityadjustment using utilities obtained from a diversegroup of women with differing levels of experience,with the outcomes included in the model.10 Selectedmodel parameters including probabilities, estimatesof test sensitivity and specificity, costs, and utilities arepresented in Table 1.

Movement between states over time was basedon 6-month transition probabilities derived from the

literature and a population-based retrospective cohortstudy of more than 37,000 women followed-up for upto 18 years in British Columbia (British Columbiacohort).3,11 The British Columbia cohort study was alarge, population-based, longitudinal follow-up studyof CIN recurrence in women after treatment for CIN.The mean age of women in the cohort who weretreated for CIN 2 or 3 was 30 years, so we applied thisas the age of women in the hypothetical modelcohort. Rates of residual and incident disease werebased on analysis of this cohort data. Residual diseaseestimates were based on disease detection within 3years of initial treatment among individuals who hadundergone at least one cytology test during each year.Incident disease estimates were based on diseasedetection outside this period in individuals who hadnegative cytology tests within the first 3 years aftertreatment. For base-case estimates, we used the high-est estimate of residual disease after treatment ofwomen for CIN 2 or 3 derived from this BritishColumbia cohort. We assumed that disease detectedwithin the first 3 years after treatment was residualand present at the onset of the model, and that after 3years CIN 2 or 3 was new (incident) disease. Womenwho had been successfully treated were assumed to beat risk for incident CIN. Residual and incident CINwere detected based on the sensitivity of the surveil-lance testing strategy. Undetected disease was as-

Cervical cancerStages I–IV

Death from cervical cancer

Death from other causes

CIN 2 or CIN 3

CIN 1

False positive

Normal

Annual cytology

Triennial cytology

Cytology

HPV testing

HPV testing and cytology

Colposcopy

Cryotherapy

Laser

Loop excision orcone biopsy

CIN 2

CIN 3

Pretreatment diagnosis

Treatment

Initial surveillance test at 6 months (Box 1)

Lifetime follow-up

Outcomes

Initial diagnosis and treatment Surveillance model

Fig. 1. Overview of the Markov model structure.Melnikow. Surveillance After Treatment for CIN. Obstet Gynecol 2010.

1160 Melnikow et al Surveillance After Treatment for CIN OBSTETRICS & GYNECOLOGY

sumed to progress or regress according to the modelparameters for the natural history of CIN. Noncervi-cal cancer deaths were estimated using data from theNational Center for Health Statistics.12 Incidentalhysterectomy rates were estimated using publishedage-specific estimates.13 Estimates of cancer symp-toms and progression rates between InternationalFederation of Gynecology and Obstetrics stages Ithrough IV were based on published estimates.14

Five-year stage-specific survival was based on datafrom Surveillance Epidemiology and End Results.15

Sensitivity and specificity of conventional cytol-ogy, Hybrid Capture 2 HPV DNA testing, and thecombination of the two were obtained from a meta-analysis of studies of Hybrid Capture 2 testing forhigh-risk HPV compared with cytology as follow-upafter treatment of CIN16 and an analysis of HybridCapture 2 and conventional cytology test perfor-mance of pooled data from primary screening stud-ies.17 Estimates for the performance of liquid-basedcytology were obtained from a large randomized trialthat compared the relative sensitivity and positivepredictive value of liquid-based compared with con-ventional cytology.18 Because the trial found thatliquid-based cytology had similar sensitivity andlower specificity than that of conventional cytology,with higher costs, liquid-based cytology strategieswere consistently more costly and less effective (dom-inated) and were subsequently excluded from thebase-case analysis. Strategies using liquid-based cytol-ogy were included in sensitivity analyses of testsensitivity and specificity, however. The sensitivity andspecificity of colposcopy were derived from a studycomparing visual colposcopy physician impression tofour-quadrant cervical biopsy in women undergoingcolposcopy and biopsy for diagnostic reasons.19

Costs were estimated from the payer’s perspec-tive. Costs for cervical cancer treatment (local, re-gional, and distant) were derived from estimates for agroup model health maintenance organization.20 Lo-cal cancer was assumed to be stage I, regional stagesII–III, and distant stage IV. Age-specific net costs forthe last year of life were based on Medicare reim-bursements.21 Costs of cervical cancer screening withconventional and liquid-based cytology (Pap tests)and Hybrid Capture 2 HPV testing, as well as colpos-copy, cryotherapy, laser, loop excision, cold-knifecone biopsy, and hysterectomy, were based on aver-age Medicare reimbursements for clinical pathwaysdeveloped by experienced physicians on the researchteam (J.M. and G.S.) using published algorithms as aguide.5,22 Each pathway was specified for a typicalpatient as a comprehensive series of diagnosis-related

group codes for hospital charges reimbursed under theprospective payment system and Current ProceduralTerminology codes for laboratory tests, anesthesiology,and all other professional services that were not coveredby the diagnosis-related group reimbursement.

The Medicare average reimbursement for hospi-tal care was calculated by dividing the total Medicarereimbursements for that diagnosis-related group bythe associated number of discharges using data fromthe Medicare Provider Analysis and Review.23 Costswere assigned to laboratory procedures on the basis ofthe midpoint for the Clinical Diagnostic LaboratoryFee Schedules used for Medicare claims payment.24

Costs for anesthesiology were based on the sum ofnational average time units and base units multipliedby the average anesthesia conversion factor.25 For allother professional services, costs were based on totaltransitioned facility relative value units multiplied bya conversion factor.26 All costs were adjusted toconstant year 2007 dollars using the Medical CareComponent of the Consumer Price Index. Onlydirect costs were included in the analyses.

Time trade-off utilities27 (health state preferences)for the various surveillance strategies and their asso-ciated potential outcomes were obtained from inter-views with a diverse group of 76 English-speaking orSpanish-speaking women who had at least one abnor-mal cytology test result within the past 2 years.10

Utilities for 11 different treatment or outcome com-binations were directly assessed. We assumed theutility for a negative HPV test result was equivalent tothe utility for a negative HPV test result with a normalcytology result. We interpolated two additional utilityscores not available from the data: an initial normalcolposcopy was assumed to be halfway between anormal cytology result and an abnormal cytology testfollowed by normal colposcopy, and a period of notesting after a normal surveillance test was assumed tobe halfway between a normal cytology result and aperiod of no testing after treatment for CIN 2 or 3 forthe first 15 years after treatment because of greateranxiety (Table 1). All interpolated values were ex-plored with wide ranges in sensitivity analyses.

Key assumptions of the model included:

Women who had excisional treatments had clearmargins (ie, no evidence of residual disease).

Women adhered to surveillance, follow-up, andtreatment protocols.

Human papillomavirus DNA testing was conductedusing the Hybrid Capture 2 test.


Table 1. Selected Parameters and Base Estimates Used in the Model

Parameter Base Estimate Range Source (References)

RESIDUAL CIN British Columbia DatabaseIndex disease CIN 2; treatment with cone

biopsy or loop excisionCIN 1 .039 .022–.039CIN 2 .021 .012–.021CIN 3 .018 .010–.018

Index disease CIN 2; treatment with laserCIN 1 .124 .071–.124CIN 2 .065 .037–.065CIN 3 .053 .030–.053

Index disease CIN 2; treatment withcryotherapy

CIN 1 .18 .103–.180CIN 2 .096 .055–.096CIN 3 .081 .046–.081

Index disease CIN 3; treatment with conebiopsy or loop excision

CIN 1 .047 .027–.047CIN 2 .023 .013–.023CIN 3 .056 .032–.056

Index disease CIN 3; treatment with laserCIN 1 .145 .083–.145CIN 2 .068 .039–.068CIN 3 .178 .099–.178

Index disease CIN 3; treatment withcryotherapy

CIN 1 .207 .118–.207CIN 2 .098 .056–.098CIN 3 .245 .140–.245

INCIDENT CIN British Columbia DatabaseIndex disease CIN 2

CIN 1 .0019 .0011–.0032CIN 2 .0015 .0008–.0028CIN 3 .0015 .0005–.0043

Index disease CIN 3CIN 1 .0019 .0011–.0032CIN 2 .0015 .0008–.0028CIN 3 .003 .0017–.0055

CIN PROGRESSION OVER 6 MO 14CIN 1 to CIN 2 (younger than 35 years) .015 .0075–.03CIN 1 to CIN 2 (35 years or older) .072 .036–.144CIN 1 to CIN 3 .015 .0075–.03CIN 2 to CIN 3 (younger than 35 years) .019 .019–.038

(35–44 years) .039 .019–0.78(45 years or older) .05 .025–.1

CIN 3 to stage I cancer .019 .0095–.038CIN REGRESSION OVER 6 MO 14

CIN 1 to normal (younger than 35 y) .051 .026–.102CIN 1 to normal (35 y or older) .031 .016–.062CIN 2 to CIN 1 .11 .055–.22CIN 2 to normal .1 .05–.2CIN 3 to CIN 2 .007 .0035–.014CIN 3 to CIN 1 or normal (younger than 45 y) .007 .0035–.014CIN 3 to CIN 1 or normal (45 y or older) .005 .0025–.01

(continued)


Table 1. Selected Parameters and Base Estimates Used in the Model (continued)

Parameter Base Estimate Range Source (References)

SURVEILLANCE TEST SENSITIVITY AND SPECIFICITY Base Alternate estimatesHPV 16,17,28

Sensitivity (CIN 2�) .91 0.944, 0.961Sensitivity (CIN 1) .76 0.792, 0.806Specificity (less than CIN) .75 0.750, 0.916

Conventional cytologySensitivity (CIN 2�) .77 0.642, 0.530Sensitivity (CIN 1) .68 0.573, 0.473Specificity (less than CIN) .90 0.963, 0.969

Liquid-based cytology 17,29Sensitivity (CIN 2�) .77 0.904Sensitivity (CIN 1) .68 0.807Specificity (less than CIN) .90 0.645

HPV and cytology 16Sensitivity (CIN 2�) .92 0.992Sensitivity (CIN 1) .68 0.807Specificity (less than CIN) .90 0.645

Colposcopy visual impression 19,30Sensitivity (CIN 1�) .88 .879–1Specificity (less than CIN) .57 .566–1

COSTS (United States 2007 $)Tests and procedures �25% Clinical pathway

Loop excision $484Cone biopsy $6,034Hysterectomy: vaginal $7,810Hysterectomy: abdominal $13,102Office visit $72

HPV test* $50 $0–$50Conventional cytology $15Liquid-based cytology $30

Cryotherapy $202Colposcopy with biopsy $271Colposcopy without biopsy $143

Cervical cancer treatment 20,31Local $30,252Regional $53,526Distant $142,371

Cost in last year of life $46,483 21TIME TRADE–OFF UTILITIES (mean scores) Interviews10

No surveillance† .912�15 y then .989 .989–1Normal cytology (Pap) test result .989 .989–1Normal HPV test result (alone or in

combination with cytology).953 .953–1

Abnormal cytology (Pap) test result, normalcolposcopy

.927 .927–1

Normal cytology (Pap) test result, abnormal(positive) HPV test

.909

Normal colposcopy surveillance result† .960 .927–1CIN 1 .897 .897–.998CIN 1 surveillance .897 .897–.998CIN 2–3 .806 .806–.996CIN 2–3 post-treatment surveillance .835 .897–.996Early (microinvasive) cervical cancer .724 .724–.950Early cervical cancer: follow-up, survivor .816 .816–.972Invasive cervical cancer .666 .666–.940Invasive cervical cancer: follow-up, survivor .754 .754–.943

CIN, cervical intraepithelial neoplasia; HPV, human papillomavirus.* HPV testing performed with Hybrid Capture 2.† Value is interpolated.


Women with normal initial surveillance test resultswere screened again with cytology in 6 months(1 year after treatment). If the cytology resultwas normal, women were followed-up with cy-tology either annually or triennially.

Abnormal cytology results of atypical squamouscells of undetermined significance or worse or apositive HPV test result led to diagnostic colpos-copy with biopsy of any abnormal areas.Women with a normal colposcopic visual im-pression returned to the first step of the surveil-lance strategy.

Women who had residual or recurrent CIN 1 werefollowed-up with repeat colposcopy. If repeatcolposcopy result was negative, then they re-turned to surveillance and follow-up. PersistentCIN 1 was assumed to be treated with loopexcision.

Residual CIN 2 or 3 detected at the first visit aftertreatment was assumed to be treated a secondtime with cone biopsy (75%), loop excision(20%), or hysterectomy (5%); those not treatedwith hysterectomy returned to the first step ofthe surveillance strategy.

Women with recurrent CIN 2 or 3 on biopsy foundafter the first 12 months were treated a secondtime with cone biopsy (75%), loop excision(20%), or hysterectomy (5%), and were not atrisk for additional recurrence.

Women with cancer received stage-specific treatment.The utility for HPV testing only with receipt nega-

tive results would be similar to the utility forundergoing both HPV and Pap testing andreceiving negative or normal results from thesetwo tests.

We examined the effect on projected model outcomesfrom varying test sensitivity, specificity, probability ofresidual CIN, costs, and utilities using one-way sensi-tivity analyses. We selected alternate estimates fromrecently published meta-analyses of primary screen-ing for the sensitivity and specificity of cytology,17

HPV testing,17,28 and liquid-based cytology.29 A re-cently published cohort study comparing colposcopywith cytology provided alternate estimates of sensitiv-ity and specificity for colposcopy and cytology.30

Alternate estimates of residual CIN were calculatedby applying a different approach to the data from theBritish Columbia Cytology Database. We used analternate estimate of cervical cancer costs from theliterature adjusted to 2007 dollars.31 Costs of cytology,HPV testing, and colposcopy were varied individu-

ally by 25%. The discount rate for costs and outcomeswas varied from 1% to 5%. We assessed the effect ofsubstituting median utility scores in place of means,and in separate analyses we varied the two interpo-lated utility scores to the extremes of their ranges.Multi-way sensitivity analysis of test characteristicssimultaneously varied sensitivity and specificity ofcytology and HPV testing. Multi-way scenario analy-ses examined the effect of simultaneous changes toestimates of sensitivity, specificity, costs, and utilitiesin analyses that favored liquid-based cytology or HPVtesting.

RESULTSConsistent with findings in the British Columbiacohort study, women with an initial diagnosis of CIN2 treated by loop excision or cone biopsy had thelowest estimated risk of subsequent disease, whereaswomen with an initial diagnosis of CIN 3 treated bycryotherapy had the highest risk.3 Women treatedwith laser consistently had estimated outcomes inter-mediate between women treated with excision andwomen treated with cryotherapy (data not shown).We calibrated the model by comparing the observedcervical cancer rate in the British Columbia cohortstudy to the expected cervical cancer rates predictedby the model for conventional cytology screeningstrategies, assuming a distribution of CIN grade andtreatment type comparable to the British Columbiacohort. The expected cervical cancer rates for annual(21.3/100,000 woman-years) and triennial screening(51.2/100,000 woman-years) bracketed the observedrate (46.6/100,000 woman-years).

Projected program costs, rates of CIN 1, CIN 2,and CIN 3, invasive cervical cancers, cervical cancerdeaths, and false-positive tests with follow-up throughage 85 for the two posttreatment states with thehighest and lowest rates of subsequent disease areshown for a hypothetical cohort of 500,000 women inTable 2. Nondominated strategies are shown in bold.Whether the initial diagnosis was CIN 2 or CIN 3, thelowest cost strategy was surveillance with conven-tional cytology at 6 and 12 months, followed bytriennial screening if initial results were normal. Thisstrategy, however, also resulted in the largest numberof subsequent cancers and cancer deaths. Reductionsin cancer deaths noted are compared with this base-line. For women with an initial diagnosis of CIN 2treated with loop excision (lowest risk of residualdisease), follow-up with annual cytology reduced thenumber of cervical cancers and reduced cancerdeaths by approximately 74% compared with trien-nial cytology, but with greater program costs (an


additional $411 million). For women with an initialdiagnosis of CIN 3 treated with cryotherapy (highestrisk of residual disease), colposcopy followed bytriennial cytology surveillance reduced cancers andreduced cancer deaths by approximately 10% com-pared with triennial cytology with a relatively smallprogram cost increase ($10 million). Greater reduc-tions were found for follow-up with annual cytology(cancer deaths reduced by 73%), but at substantiallygreater program costs (an additional $330 million).

Table 3 shows projected total costs and effective-ness, incremental costs and effectiveness, and incre-mental cost-effectiveness ratios for nondominatedstrategies for initial diagnoses of CIN 2 or CIN 3treated with cryotherapy or excisional procedures,with life expectancy as the outcome. More intensivestrategies had higher average costs per additionallife-year for women at greater risk for subsequentdisease, but differences in average effectiveness weregenerally less than .01 year (less than 4 additional

days of life expectancy per woman). Annual follow-upwith cytology had greater effectiveness and highercosts than triennial cytology follow-up for all initialdiagnosis or treatment combinations. The averagecost per additional life-year for annual cytology fol-low-up was less than $100,000 for all combinations ofinitial diagnosis and treatment. For initial diagnosis ortreatment combinations with higher risks of residualdisease (CIN 3 treated with cryotherapy), initial col-poscopy followed by annual cytology was more effec-tive but more costly than cytology as an initial testfollowed by annual cytology. Strategies involvingHPV testing were dominated except for women withCIN 2 treated by excision for which the average costper additional life-year was more than $1 million.

Including women’s utilities in the analysis re-sulted in much greater projected differences by fol-low-up strategy for quality-adjusted life-years than thedifferences projected for unadjusted life-years (Table4). The strategy of initial cytology followed by annual

Table 2. Projected Outcomes and Program Costs for a Hypothetical Cohort of 500,000 WomenEvaluated From Age 30 to 85 Years Showing Expected False-Positive Results, CervicalIntraepithelial Neoplasia 1 Cases, Cervical Intraepithelial Neoplasia 2 or 3 Cases, Cancers (AllStages), and Cancer Deaths

StrategyFalse

Positives CIN 1CIN 2or 3 Cancer

CancerDeaths

ProgramCosts

(Millions)

INITIAL DIAGNOSIS OF CIN 2 TREATED WITHEITHER CONE BIOPSY OR LOOP EXCISION

Conventional cytology (q3) 901,958 62,330 121,478 6,582 673 $2,212HPV testing (q3) 976,705 62,528 122,524 6,549 670 $2,256HPV and conventional cytology (q3) 999,663 62,562 122,638 6,546 669 $2,272Colposcopy (q3) 1,067,333 63,241 123,041 6,546 669 $2,279

Conventional cytology (q1) 1,357,754 62,846 117,754 1,958 153 $2,623HPV testing (q1) 1,436,328 62,915 118,646 1,940 151 $2,670

HPV and conventional cytology (q1) 1,459,838 62,924 118,742 1,938 151 $2,687Colposcopy (q1) 1,533,583 63,114 119,137 1,937 151 $2,699

INITIAL DIAGNOSIS OF CIN 3 TREATED WITHCRYOTHERAPY

Conventional cytology (q3) 787,136 136,740 252,072 7,858 796 $3,043Colposcopy (q3) 926,524 144,000 266,838 7,032 717 $3,053

HPV testing (q3) 864,998 138,730 260,429 7,285 741 $3,081HPV and conventional cytology (q3) 881,701 139,067 261,455 7,220 734 $3,094

Conventional cytology (q1) 1,163,267 140,860 250,527 2,680 211 $3,373HPV testing (q1) 1,268,554 142,237 257,628 2,367 186 $3,439

Colposcopy (q1) 1,365,853 144,915 263,807 2,211 173 $3,445HPV and conventional cytology (q1) 1,289,051 142,445 258,517 2,331 183 $3,456

CIN, cervical intraepithelial neoplasia; HPV, human papillomavirus.q1 indicates screening with conventional cytology every year after initial surveillance test and subsequent cytology test 6 months apart

with normal results.q3 indicates screening with conventional cytology every 3 years after initial surveillance test and subsequent cytology test 6 months

apart with normal results.Strategies are ordered by ascending program cost. Strategies found to be more costly with no increase in effectiveness or less costly but

less cost-effective than an alternative strategy were considered dominated by other strategies. Nondominated strategies are shown inbold. Dominance is calculated based on average additional discounted costs and average additional discounted life-years (notshown in Table). Small differences in cancer deaths are obscured by rounding.


cytology surveillance was dominated. Initial colpos-copy followed by annual cytology surveillance re-sulted in substantially higher quality-adjusted effec-tiveness than in the analysis that used life-expectancyas an outcome. It added little average cost per quality-adjusted life year gained and showed little variationrelated to risk of disease recurrence. Strategies involv-ing HPV testing were dominated for all combinationsof initial diagnosis and initial treatment in the cost-utility analysis.

The pattern of findings in the range of sensitivityanalyses was remarkably robust to variations in testsensitivity, specificity, costs, and utilities. In the lifeexpectancy analysis, both HPV testing and liquid-based cytology projected average costs per life-yeargained of more than $350,000 when their respectivesensitivities or specificities were changed to the valuesreported in the systematic reviews of primary screen-ing by Arbyn29 or Cuzick.17 After adjusting the sensi-tivity and specificity of conventional cytology byapplying primary screening values from the Cuzicksystematic review16 or values from the Soutter study,30

colposcopy remained the only nondominated strategyin the life expectancy analysis for initial surveillance

of women who had CIN 3 previously treated withcryotherapy. When sensitivity and specificity of bothHPV testing with Hybrid Capture 2 and cytologywere changed in a multi-way analysis to primaryscreening values, HPV testing for women with CIN 2treated with excision was associated with an averagecost per life-year gained of $40,595 compared withmore than $1 million in the base case. Humanpapillomavirus testing continued to be dominated forwomen treated with cryotherapy for CIN 3. In thequality-adjusted analysis, no adjustment of surveil-lance test sensitivity and specificity altered the patternof findings of the base-case analysis.

In sensitivity analyses of the costs, we varied thecosts of HPV testing, cytology, colposcopy, and can-cer individually. For HPV testing, we lowered the costto less than that of conventional cytology. Even at ahypothetical zero cost per HPV test, the conventionalcytology strategies remained less costly because ofhigher specificity. Raising the cost of cytology by $10eliminated all but the colposcopy strategy with trien-nial follow-up for surveillance of women at high risk(those with CIN 3 treated with cryotherapy). Raising orlowering the cost of colposcopy by 25% had minimal

Table 3. Life Expectancy Analysis

StrategyAverage

CostIncremental

CostAverage Life

Expectancy (y)Incremental

Effectiveness (y)Average Cost perLife-Year Gained

INITIAL DIAGNOSIS OF CIN 2Cone or loop excision

Conventional cytology (q3) $4,423 25.2271Conventional cytology (q1) $5,245 $822 25.2372 .0102 $80,882HPV testing (q1) $5,340 $94 25.2373 .0001 $1,068,182Colposcopy (q1) $5,398 $58 25.2374 .0001 $1,160,000

CryotherapyConventional cytology (q3) $5,175 25.2254Colposcopy (q3) $5,253 $78 25.2268 .0014 $55,437Conventional cytology (q1) $5,915 $662 25.2365 .0097 $68,318Colposcopy (q1) $6,066 $151 25.2371 .0006 $255,499

INITIAL DIAGNOSIS OF CIN 3Cone or loop excision

Conventional cytology (q3) $4,642 25.2257Conventional cytology (q1) $5,445 $803 25.2367 .0108 $74,214Colposcopy (q1) $5,596 $151 25.2372 .0005 $319,239

CryotherapyConventional cytology (q3) $6,086 25.2202Colposcopy (q3) $6,105 $19 25.2249 .0047 $4,083Conventional cytology (q1) $6,747 $641 25.2342 .0093 $69,036Colposcopy (q1) $6,890 $143 25.2362 .002 $69,961

CIN, cervical intraepithelial neoplasia.q1 indicates screening with conventional cytology every year after initial surveillance test and subsequent cytology test 6 months apart


apart with normal results.Average lifetime health care costs, incremental costs, average life expectancy, incremental effectiveness, and average costs per

additional life-year saved for surveillance strategies not dominated in the analysis, by initial diagnosis, and treatment type.


effects on the results of the analysis. Substitution ofalternate cervical cancer costs from the literature31 hadessentially no effect on model outcomes.

Sensitivity analysis of the utility scores had littleeffect on the pattern of findings in the cost-utilityanalysis. When median utilities were substituted formeans throughout the model, the average cost perquality-adjusted life-year gained for colposcopy fol-lowed by annual cytology remained less than $50,000for all combinations of initial diagnosis or treatment,but was higher than values in the base-case analysis.Varying the interpolated utilities had even less effecton the model results, except when the value assignedto intervals of no testing was raised to that of havingjust had a normal cytology test. In that case, the onlynondominated strategies were initial cytology or col-poscopy with triennial cytology follow-up.

In multi-way scenario analyses favoring HPVtesting, we simultaneously lowered the cost of HPVtesting by 20%, increased the cost of colposcopy by20%, and substituted the values for sensitivity andspecificity for HPV testing and for cytology from theCuzick systematic review of primary screening.17 Us-ing this set of highly favorable assumptions, HPVtesting with triennial cytology (after CIN 2 treatedwith loop excision) or annual cytology (after CIN 3treated with cryotherapy) produced an average cost

per life-year gained that ranged from $28,422 to$38,066 in the life expectancy analysis but remaineddominated in the cost-utility analysis. The multi-wayscenario analysis of liquid-based cytology was ap-proached similarly and included lowering the cost ofliquid-based cytology by 20% and raising the cost ofcolposcopy by 20%. Test sensitivity and specificityvalues for liquid-based cytology were applied fromArbyn.29 Despite these favorable assumptions, theaverage cost per life-year gained for liquid-basedcytology exceeded $1 million, and liquid-based cytol-ogy was dominated in the cost-utility analysis.

DISCUSSIONPublished recommendations for surveillance aftertreatment for CIN were based on expert review ofcurrent medical literature at the time that the recom-mendations were formulated.5,6,7,22 These recommen-dations do not distinguish between women at higherand lower risk for recurrence. No randomized, con-trolled trials have compared different follow-up strat-egies after treatment for CIN. Our model was basedon recently published data, including a large obser-vational study3 that indicated that initial treatmenttype and CIN grade are associated with markeddifferences in recurrence risks and a rigorous system-atic review that defined the test sensitivity and speci-

Table 4. Cost Utility Analysis

StrategyAverage

CostIncremental

CostAverageQALY

IncrementalQALY

Average Cost perQALY Gained

INITIAL DIAGNOSIS CIN 2Cone biopsy or loop excision

Conventional cytology (q3) $4,423 23.907Colposcopy (q3) $4,558 $134 24.270 .363 $369Colposcopy (q1) $5,398 $840 24.431 .161 $5,217

CryotherapyConventional cytology (q3) $5,175 23.825Colposcopy (q3) $5,253 $78 24.178 .353 $221Colposcopy (q1) $6,066 $813 24.333 .155 $5,246

INITIAL DIAGNOSIS CIN 3Cone biopsy or loop excision

Conventional cytology (q3) $4,642 23.891Colposcopy (q3) $4,763 $120 24.254 .362 $331Colposcopy (q1) $5,596 $833 24.414 .160 $5,193

CryotherapyConventional cytology (q3) $6,086 23.762Colposcopy (q3) $6,105 $19 24.113 .351 $54Colposcopy (q1) $6,890 $784 24.266 .153 $5,133

CIN, cervical intraepithelial neoplasia; QALY, quality-adjusted life-year.q1 indicates screening with conventional cytology every year after initial surveillance test and subsequent cytology test 6 months apart


apart with normal results.Average lifetime quality-adjusted costs, incremental costs, average QALY, incremental QALY, and average cost per additional QALY for

all strategies not dominated in the analysis by initial diagnosis and treatment type.


ficity of HPV testing with Hybrid Capture 2 andcytology for follow-up after treatment.16 These datawere not available when the most recent guidelineswere formulated.

Our study is subject to the limitations of model-ing, which can never fully represent the realities ofclinical care and the health care system. We modeledclinical practice strategies and applied costs for sur-veillance of women being cared for in the UnitedStates. Although some model parameters were basedon limited data and therefore their precise valueswere uncertain, the stability of the model projectionsover a range of sensitivity analyses suggests that ourestimates of costs and outcomes are reasonable forwomen in surveillance after treatment of CIN.

Our findings do not apply to women undergoingprimary screening, nor do they apply to women withpositive margins after excisional treatment, becausewe had no data on outcomes of surveillance forwomen with positive margins. Our assumption of100% adherence to follow-up was not realistic, but toaffect the relative cost-effectiveness of the follow-upstrategies the rates or distribution of adherence wouldhave to differ between strategies. Identified factorsinfluencing adherence to follow-up with abnormalcytology include health beliefs, social support, patienteducation, counseling and reminder systems, andlesion severity,32 but type of recommended follow-uphas not been shown to influence adherence.33 Weinterpolated two utility values and assumed that theutility for normal cytology with a negative HPV testresult was equivalent to the utility for a negative HPVtest result alone. Although these values were notdirectly obtained from interviews, varying them insensitivity analysis had little effect on the modelresults.

Our findings suggest that surveillance strategiesmay be tailored to recurrence risk defined by initialCIN grade and treatment type. Traditional methodsof surveillance using conventional cytology or initialcolposcopy, depending on recurrence risk, providethe greatest effectiveness relative to resources in-vested in surveillance. Initial conventional cytology at6 and 12 months followed by surveillance with annualcytology, consistent with recent American College ofObstetricians and Gynecologists recommendations,7

appears to represent the most productive use of healthcare resources for this patient group to reduce subse-quent cancers and cancer deaths. Initial screeningwith colposcopy adds some additional life expectancyfor women at higher risk for recurrence and washighly valued, as indicated by findings from thecost-utility analysis.

Liquid-based cytology was consistently more ex-pensive and less effective than alternative strategies.Human papillomavirus testing using Hybrid Capture2, when it was not dominated in the analysis, addedlittle life expectancy at high cost compared with olderapproaches to surveillance based on conventionalcytology. Human papillomavirus testing became amore reasonable alternative when we altered thesensitivity and specificity to values derived fromstudies of primary screening. These values are likelynot applicable to the posttreatment surveillance set-ting, in which the prevalence of HPV is much higherthan in screening settings.16 New tests and technolo-gies often diffuse into practice before a full evaluationof their potential benefits, harms, and costs. Alex-ander and Stafford,34 arguing for the importance oftimely comparative effectiveness data, have pointedout that innovations in medical practice are oftenadopted with great enthusiasm and only later arefound to be wanting. Human papillomavirus testingusing Hybrid Capture 2 and liquid-based cytologyhave diffused widely in the United States and else-where, with limited evaluation of which circum-stances may justify the additional resources requiredby these tests.35,36 Although American Cancer Societyguidelines recommend that negative results fromhighly sensitive HPV testing could enable longerintervals for primary screening,37 practicing physi-cians in the United States have generally continued toperform annual screening,38 a practice that appears tobe preferred by many women.39,40

Our base-case assumptions related to the perfor-mance of these technologies were supported by re-cently published large studies of their use in primaryscreening. A recent randomized trial of liquid-basedcytology compared with conventional cytology inmore than 85,000 women found no difference inrelative sensitivity and positive predictive value.40 Arandomized trial of the addition of HPV testing tocytology among nearly 25,000 women followed-upover the course of two rounds of triennial screeningfound no difference in the detection of CIN 3 orhigher.41 This result indicates that our estimate of thesensitivity of HPV testing may have been overlyfavorable, based on studies comparing only a firstround of screening.

Our findings, however, relate only to posttreat-ment surveillance. They suggest that newer technolo-gies have increased the cost of surveillance aftertreatment without providing any meaningful increasein life expectancy or reductions in cancer, comparedwith traditional approaches to surveillance. Increas-ing use of conventional cytology and initial colpos-


copy for surveillance of women at high risk wouldreduce the costs of posttreatment surveillance andprovide health care resources that could be used moreproductively to broaden access to cervical cancerscreening, treatment, and surveillance.

REFERENCES1. Jones BA, Davey DD. Quality management in gynecologic

cytology using interlaboratory comparison. Arch Pathol LabMed 2000;124:672–81.

2. Woolf T. Screening for cervical cancer. Preventing disease:Beyond the rhetoric. New York (NY): Springer Verlag; 1990.p. 319–23.

3. Melnikow J, McGahan C, Sawaya GF, Ehlen T, Coldman A.Cervical intraepithelial neoplasia outcomes after treatment:long-term follow-up from the British Columbia Cohort Study.J Natl Cancer Inst 2009;101:721–8.

4. Strander B, Andersson-Ellstrom A, Milsom I, Sparen P. Longterm risk of invasive cancer after treatment for cervical intra-epithelial neoplasia grade 3: population based cohort study.BMJ 2007;335:1077.

5. Wright TC Jr, Massad LS, Dunton CJ, Spitzer M, WilkinsonEJ, Solomon D, et al. 2006 consensus guidelines for themanagement of women with cervical intraepithelial neoplasiaor adenocarcinoma in situ. Am J Obstet Gynecol 2007;197:340–5.

6. Management of abnormal cervical cytology and histology.ACOG Practice Bulletin No. 99. American College of Obste-tricians and Gynecologists. Obstet Gynecol 2008;112:1419–44.

7. Cervical cytology screening. ACOG Practice Bulletin No. 109.American College of Obstetricians and Gynecologists. ObstetGynecol 2009;114:1409–20.

8. Gold M SJ, Russell L, Weinstein M. Cost-effectiveness inhealth and medicine. New York (NY): Oxford UniversityPress; 1996.

9. American College of Obstetricians and Gynecologists. ACOGCommittee Opinion No. 388 November 2007: supracervicalhysterectomy. Obstet Gynecol 2007;110:1215–7.

10. Kuppermann M, Melnikow J, Slee C, Tancredi DJ, Kulas-ingam S, Birch S, et al. Preferences for surveillance strategiesfor women treated for high-grade precancerous cervicallesions. Gynecol Oncol 2010;118:108–15.

11. Canfell K, Barnabas R, Patnick J, Beral V. The predicted effectof changes in cervical screening practice in the UK: resultsfrom a modelling study. Br J Cancer 2004;91:530–6.

12. Minino AM, Heron MP, Murphy SL, Kochanek KD. Deaths:final data for 2004. Natl Vital Stat Rep. 2007;55:1–119.

13. Keshavarz H. Hysterectomy surveillance–United States,1994–1999. MMWR 2002; 51:1–8.

14. Myers ER, McCrory DC, Nanda K, Bastian L, Matchar DB.Mathematical model for the natural history of human papillo-mavirus infection and cervical carcinogenesis. Am J Epidemiol2000;151:1158–71.

15. National Cancer Institute. Surveillance, Epidemiology, EndResults (SEER) Cancer Statistics Review, 1973–1994. NIHPublication No. 97-2789. Available at: http://seer.cancer.gov/csr/1973_1994/. Retrieved October 19, 2009.

16. Chan BK, Melnikow J, Slee CA, Arellanes R, Sawaya GF.Posttreatment human papillomavirus testing for recurrent cer-vical intraepithelial neoplasia: a systematic review. Am JObstet Gynecol 2009;200:422.e421–9.

17. Cuzick J, Clavel C, Petry KU, Meijer CJ, Hoyer H, Ratnam S,et al. Overview of the European and North American studieson HPV testing in primary cervical cancer screening. Int JCancer 2006;119:1095–101.

18. Ronco G, Cuzick J, Pierotti P, Cariaggi MP, Dalla Palma P,Naldoni C, et al. Accuracy of liquid based versus conventionalcytology: overall results of new technologies for cervical cancerscreening: randomised controlled trial. BMJ 2007;335:28.

19. Cantor SB, Cardenas-Turanzas M, Cox DD, Atkinson EN,Nogueras-Gonzalez GM, Beck JR, et al. Accuracy of colpos-copy in the diagnostic setting compared with the screeningsetting. Obstet Gynecol 2008;111:7–14.

20. Helms LJ, Melnikow J. Determining costs of health careservices for cost-effectiveness analyses: the case of cervicalcancer prevention and treatment. Med Care 1999;37:652–61.

21. Hoover DR, Crystal S, Kumar R, Sambamoorthi U, Cantor JC.Medical expenditures during the last year of life: findings fromthe 1992–1996 Medicare current beneficiary survey. HealthServ Res 2002;37:1625–42.

22. Wright TC Jr, Massad LS, Dunton CJ, Spitzer M, WilkinsonEJ, Solomon D, et al. 2006 consensus guidelines for themanagement of women with abnormal cervical screening tests.J Low Genit Tract Dis 2007;11:201–22.

23. Centers for Medicare and Medicaid Services. 100% MedicareProvider Analysis and Review (MEDPAR). Available at: http://www.cms.hhs.gov/MedicareFeeforSvcPartsAB/03_MEDPAR.asp. Retrieved October 19, 2009.

24. Centers for Medicare and Medicaid Services. Clinical diagnosticlaboratory fee schedule files. Available at: http://www.cms.hhs.gov/ClinicalLabFeeSched. Retrieved October 19, 2009.

25. Centers for Medicare and Medicaid Service. Anesthesiologistsbilling and payment. Available at: http://www.cms.hhs.gov/center/anesth.asp. Retrieved October 19, 2009.

26. Centers for Medicare and Medicaid Services. National physicianfee schedule relative value file. Available at: http://www.cms.hhs.gov/PhysicianFeeSched/PFSRVF. Retrieved October19, 2009.

27. Torrance GW, Thomas WH, Sackett DL. A utility maximiza-tion model for evaluation of health care programs. Health ServRes 1972;7:118–33.

28. Arbyn M, Sasieni P, Meijer CJ, Clavel C, Koliopoulos G,Dillner J. Chapter 9: clinical applications of HPV testing: asummary of meta-analyses. Vaccine 2006;24(Suppl 3):S3/78–89.

29. Arbyn M, Bergeron C, Klinkhamer P, Martin-Hirsch P,Siebers AG, Bulten J. Liquid compared with conventionalcervical cytology: a systematic review and meta-analysis.Obstet Gynecol 2008;111:167–77.

30. Soutter WP, Butler JS, Tipples M. The role of colposcopy inthe follow up of women treated for cervical intraepithelialneoplasia. BJOG 2006;113:511–4.

31. Insinga RP, Glass AG, Rush BB. The health care costs ofcervical human papillomavirus–related disease. Am J ObstetGynecol 2004;191:114–20.

32. Eggleston KS, Coker AL, Prabhu Das I, Cordray ST, LuchokKJ. Understanding barriers for adherence to follow-up care forabnormal Pap tests. J Women Health 2007;16:311–30.

33. Melnikow J, Chan BK, Stewart GK. Do follow-up recommen-dations for abnormal Papanicolaou smears influence patientadherence? Arch Fam Med 1999;8:510–4.

34. Alexander GC, Stafford RS. Does comparative effectivenesshave a comparative edge? JAMA 2009;301:2488–90.

35. Bidus MA, Maxwell GL, Kulasingam S, Rose GS, Elkas JC,Chernofsky M, et al. Cost-effectiveness analysis of liquid-based


cytology and human papillomavirus testing in cervical cancerscreening. Obstet Gynecol. 2006;107:997–1005.

36. Cochand-Priollet B, Cartier I, de Cremoux P, Le Gales C, ZiolM, Molinie V, et al. Cost-effectiveness of liquid-based cytologywith or without hybrid-capture II HPV test compared withconventional Pap smears: a study by the French Society ofClinical Cytology. Diagn Cytopathol 2005;33:338–43.

37. Saslow D, Runowicz CD, Solomon D, Moscicki AB, SmithRA, Eyre HJ, et al. American Cancer Society guideline for theearly detection of cervical neoplasia and cancer. CA CancerJ Clin 2002;52:342–62.

38. Sirovich BE, Welch HG. The frequency of Pap smear screen-ing in the United States. J Gen Int Med 2004;19:243–50.

39. Sawaya GF, Iwaoka-Scott AY, Kim S, Wong ST, Huang AJ,Washington AE, et al. Ending cervical cancer screening:attitudes and beliefs from ethnically diverse older women.Am J Obstet Gynecol 2009;200:40.e41–7.

40. Siebers AG, Klinkhamer PJ, Grefte JM, Massuger LF, VedderJE, Beijers-Broos A, et al. Comparison of liquid-based cytologywith conventional cytology for detection of cervical cancerprecursors: a randomized controlled trial. JAMA 2009;302:1757–64.

41. Kitchener HC, Almonte M, Thomson C, Wheeler P, SargentA, Stoykova B, et al. HPV testing in combination with liquid-based cytology in primary cervical screening (ARTISTIC): arandomised controlled trial. Lancet Oncol 2009;10:672–82.


00006250-201011000-00023

Documents