assessment of techniques for cervical ... copies of assessment of techniques for cervical cancer...

Canadian Coordinating Office for Health Technology Assessment

ASSESSMENT OF TECHNIQUESFOR

CERVICAL CANCER SCREENING

CCOHTA Report 1997: 2E

Additional copies of Assessment of Techniques for Cervical Cancer Screening are available fromCCOHTA.

Vous pouvez aussi vous procurer la version française Évaluation des techniques de dépistage du cancerdu col utérin à l’OCCETS.

Reproduction of this document for non-commercial purposes is permitted provided appropriate credit is given to CCOHTA.

Legal Deposit - 1997 National Library of CanadaISBN 1-895561-49-3


ASSESSMENT OF TECHNIQUESFOR CERVICAL CANCER SCREENING

PROJECT DIRECTOR:Hussein Z. Noorani

PROJECT TEAM:Cheryl Arratoon

Annie Hall

MAY 1997

110-955 Green Valley CrescentOttawa, Ontario, Canada K2C 3V4

Telephone (613) 226-2553Facsimile (613) 226-5392

Internet http://www.ccohta.caEmail [email protected]


This report was reviewed by externalreviewers and by members of a subcommitteeof CCOHTA’s Scientific Advisory Panel.These individuals kindly provided commentson drafts of this report. This final documentincorporates most of the Reviewers comments,however, the authors take sole responsibilityfor its form and content.

REVIEWERS

External Reviewers

Dr. William GeddieDirector, Cytopathology LaboratoryCredit Valley HospitalMississauga, Ontario

Dr. Jean ParboosinghSenior Medical ConsultantAdult Health DivisionHealth Promotion and Programs BranchHealth Canada

Dr. Gavin StuartDirector, Tom Baker Cancer CentreCalgary, Alberta

Subcommittee Members

Dr. David HaileyDirector, Health Technology AssessmentAlberta Heritage Foundation for Medical ResearchEdmonton, Alberta

Dr. John HamertonDistinguished Professor EmeritusUniversity of ManitobaWinnipeg, Manitoba

Dr. Murray KrahnStaff PhysicianDivision of General Internal Medicine & Clinical EpidemiologyThe Toronto HospitalToronto, Ontario


EXECUTIVE SUMMARY

Cervical cancer is one of the few readily preventable forms of cancer. Although the Papanicolaou (Pap)smear has had an enormous impact on the incidence and mortality rates of cervical cancer over the past fourdecades, concerns over false-negative rates and their adverse effects on patients have driven the search forstrategies to improve the Pap test, and development of emerging technologies.

The use of particular cell collection devices has minimal effect on smear quality. Instead, training for smeartakers and feedback on their performance from laboratories would significantly improve smear quality. Asignificant impact would result from dialogue between smear takers and cytotechnologists to reach consensusas to the components of an adequate smear. Automated methods of monolayer preparation that producecleaner, more evenly dispersed smears appear to improve detection of abnormal cases. However, furtheranalysis is required to balance this gain with the high cost of equipment and interpretive training of staff.

Automated devices for the rescreening of slides previously examined by humans are more effective, but alsomore costly than manual rescreening. Both of the automated systems compared in the economic evaluationappear to be of similar efficacy, but dissimilar cost. Sensitivity analyses over the range of values for each keyvariable revealed that the cost-effectiveness ratios for these systems were sensitive to the per rescreening costby automated device.

The potential role of human papillomavirus (HPV) testing in screening for cervical cancer depends uponclarification of the link between HPV infection and invasive disease, and the ability to overcome cost andtechnical barriers.

Canadian Coordinating Office for Health Technology AssessmentCanadian Coordinating Office for Health Technology Assessment

CONTENTS

OBJECTIVES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

METHODOLOGY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

EFFECTIVENESS OF THE PAP TEST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

STRATEGIES TO IMPROVE THE EFFECTIVENESS OF THE PAP TEST(i) Recruitment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7(ii) Sample Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8(iii) Preparation Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

(iv) Screening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9(v) Quality Assurance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

COST-EFFECTIVENESS OF AUTOMATED RESCREENING STRATEGIES (i) Type of Analysis, Viewpoint, Cost Measurement, & Outcome Measurement . . . . . . . . 11 (ii) Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

(iii) Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 (iv) Uncertainty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

(v) Related Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13(vi) Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

EMERGING TECHNOLOGIES(i) HPV Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14(ii) Flow Cytometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

TABLESTable 1: Databases Searched and Description of Searches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17Table 2: Automated Pap-smear Screening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18Table 3: Baseline Values Used in the Decision Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19Table 4: Baseline Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20Table 5: One-way Sensitivity Analysis on Key Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

FIGURE 1: Decision Model for Rescreening Strategies of Cervical Cancer . . . . . . . . . . . . . . . . . . . 22

APPENDIX 1: Cervical Cancer Screening - Questions re: Automated Screeners of Pap Smear . . . . . 23

REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

Canadian Coordinating Office for Health Technology AssessmentCanadian Coordinating Office for Health Technology Assessment 1

OBJECTIVES

The following four objectives were identified for study:

(1) To examine the effectiveness of the Pap test.

(2) To consider different strategies for improving the effectiveness of the Pap test.

(3) To compare the cost-effectiveness of automated rescreening strategies.

(4) To consider emerging technologies.

Canadian Coordinating Office for Health Technology AssessmentCanadian Coordinating Office for Health Technology Assessment2

I INTRODUCTION

Cervical cancer is uncommon in Canada, with an estimated incidence of 1,350 cases and 390 deaths in 1996.1

This relatively low incidence and mortality is largely attributed to the effectiveness of cytologic screening bythe Papanicolaou test (Pap test).2 Based on self-reported data from provincial and national surveys, it isestimated that over 4 million smears are taken annually in Canada, of which approximately eight percent (orover 320,000 smears) will have an abnormality requiring follow-up.3 Despite its low incidence and mortalityin Canada, given that it is a potentially preventable disease, cervical cancer is still the eleventh most commonmalignancy amongst females.1

Although cervical cancer can occur in a wide range of ages, it usually occurs in the fifth or sixth decade oflife. A long latency period normally exists between the detection of precursor lesions in cervical epitheliumand the development of invasive disease. Precursor lesions occur more frequently in younger women (oftenunder 35 years of age).4 Several factors that increase the risk for cervical cancer have been identified,including sexual intercourse at an early age, multiple male sexual partners, immunodeficiency, and smoking.4

Cervical cancer screening can be performed within organized programs (programmatic screening) oropportunistically.5 The latter comprises screening which is carried out by suggestion from a physician (whenwomen present for consultations for other health matters) or upon “demand” (upon a woman’s requestoutside a planned system of invitation). The contribution of the Pap test to reduction in cervical cancer inCanada has been through opportunistic screening. Programmatic screening, on the other hand, implies amechanism to identify the target population and invite all members of the target population to participate.To date, in Canada, only British Columbia (BC) and Nova Scotia (NS) have most of the components of anorganized screening program: a recruitment/retention strategy, a comprehensive quality management programthat spans all steps in the screening pathway, and information systems to facilitate the first two componentsand to provide a mechanism for monitoring and program evaluation. Complete coverage of the targetpopulation is still lacking in Canada.

Screening for cervical cancer began in 1949 in BC, and has gradually spread across the country.6 In 1976the Canadian Task Force on Cervical Cancer Screening Programs (“the Walton report”) recommended thedevelopment of organized screening programs designed to detect precursors of the disease.7 The NationalWorkshop on Screening for Cancer of the Cervix in 1989 estimated that an organized screening programcould save $20 million annually in laboratory costs alone across Canada, in addition to a reduction in the costof invasive cases and deaths.8 In the 20 years since the first recommendation of “the Walton report”, anddespite its reinforcement by subsequent reports, there has been little progress toward implementation oforganized screening programs in most provinces. As of this writing, in addition to BC and NS, onlyNewfoundland and Prince Edward Island (PEI) have established provincial registries,2,9 one component ofan organized screening program.

In response to renewed concern about the prevalence of cervical cancer among Canadian women, a workshop(Interchange ‘95: A Canadian Forum to Collaborate on Cervical Cancer Screening Program ImplementationStrategies)10 was held in Ottawa on February 27th- March 1st 1995, in order to identify the barriers to theimplementation of the 1989 National Workshop recommendations. Provincial and territorial representativesinvolved in cervical cancer screening on a clinical or programmatic level participated in the workshop alongwith policymakers from the provincial and federal governments and other relevant national organizations.


The Interchange 95' workshop concluded with suggestions that the federal government continue to encourageand facilitate information exchange between the provinces and to provide some direction and leadership inthe areas of standards and quality of care.2,10 In this regard, all provinces and territories were invited toparticipate in the Cervical Cancer Prevention Network, an informal association of provincial and federalrepresentatives with the relevant professional bodies.2,10 The challenge for the network was to overcome thebarriers to the implementation of the 1989 recommendations and for the provinces to emerge withcomprehensive, organized programs that use the technologic innovation and advances in knowledge that haveoccurred since 1989.2,10

In March, 1995, following Interchange ‘95, CCOHTA was requested to undertake a technology assessmenton cervical cancer screening by a provincial health ministry. Other provincial ministries were consulted withto determine the value/interest of undertaking this project. Based on comments from the representatives ofthe various health ministries around the country, and a clinical advisory committee convened for the project,the four objectives stated in the previous section were identified for study.

Section II summarizes the methodological aspects of the report, and the four study objectives are addressedsequentially in Sections III, IV, V, and VI. The conclusions of the assessment are in Section VII.


II METHODOLOGY

Published literature was obtained using a number of bibliographic databases. These databases and the searchstrategies are listed in Table 1.

Searches were limited to English language articles and to human studies. Secondary references were alsoidentified and retrieved. Information on automated Pap-smear screening systems and related cost estimateswere obtained from manufacturers through a mail questionnaire (Appendix 1), personal communication, andfrom the literature. The clinical advisory committee provided advice on the project, and helped to identifyemerging technologies. The committee included a gynecologic oncologist, a medical consultant, and apathologist.

The methodology for the economic evaluation of the rescreening strategies is detailed in Section V. Whereappropriate, the reporting structure for the economic evaluation is in accordance with the “Guidelines foreconomic evaluation of pharmaceuticals: Canada”.11


III EFFECTIVENESS OF THE PAP TEST

Several cytologic classifications of Pap-smear findings have been proposed, with the Bethesda classificationcurrently the most widely used.12 The most common significant lesions detected by the Pap smear aresquamous cell abnormalities.4 The Bethesda classification recognizes two categories of squamousintraepithelial lesions (SIL).12 The first category, termed “low-grade” SIL (LSIL), encompasses cellularchanges associated with human papillomavirus (HPV) as well as cervical intraepithelial neoplasia 1 (CIN 1),a lesion that usually undergoes spontaneous resolution.12,13 The treatment and follow-up of patients withLSIL is controversial (there is currently a landmark study under way in the United States through theNational Cancer Institute that may serve to elucidate this issue).14 Although the finding of LSIL usuallyindicates the presence of a low-risk lesion, colposcopy and biopsy are indicated for persistent LSIL.13

The second category of abnormal findings on the Pap smear that suggests the presence of an intraepitheliallesion is termed “high-grade” SIL (HSIL).12 A Pap-smear finding of HSIL suggests the presence of CIN 2or 3 on biopsy.12 These are serious lesions that call for colposcopic evaluation.13

In addition to LSIL and HSIL, the Bethesda classification includes a group of lesions termed “atypicalsquamous cells of undetermined significance” (ASCUS).12 The ASCUS category refers to cells that are moreabnormal than cells seen in reactive or inflammatory lesions, but do not fulfill all the criteria for LSIL orHSIL.12 The management of patients in this category remains controversial since a Pap-smear finding ofASCUS may turn out to be SIL on follow-up.13,14

The purpose of a Pap smear is to screen for intraepithelial lesions before they progress to invasive disease.Patients have come to expect the Pap test to have 100% sensitivity: a false-negative result must meansomeone made a mistake. Many women (and unfortunately many clinicians) fail to realize that a Pap test isa screening test and therefore subject to limitations: it makes the promise of reducing the incidence of cervical(squamous-cell) cancer, but it makes no promise to detect all abnormalities in all women.9

Despite its demonstrated efficacy as a screening test, false-negative results do occur for a number of reasons.False-negative Pap smears may result from sampling errors (failure to obtain a smear representative of thebiologic abnormality), screening errors within the laboratory (failure to locate the biologic abnormality), orinterpretative errors (failure to interpret the biologic abnormality correctly when found). Consensus suggeststhat screening and interpretative errors are responsible for about a third of false-negative Pap smears.4 Theremaining two-thirds of false-negative Pap smears are attributed to sampling errors, primarily related toinadequate collection of cells from the transformation zone (see section on sample collection below).4 Thereported false-negative results in the literature vary widely, from 1% to about 90%,15 with the mostcomprehensive studies showing results between 20% and 30%.16,17 This corresponds to sensitivity rates of70% to 80%. Variability in interpretation (depending on the severity of the lesion) of abnormal Pap smearsmay contribute to the wide variation in false-negative results.15 In this regard, the Pap test works best whenlesions have progressed further than desirable.


The incidence of cervical cancer decreased by 85%, and mortality by 80%, in BC between 1955 and 1988,with the implementation of a provincial screening program.6 Also, comparisons among the Scandinaviancountries have demonstrated contrasting trends both in incidence and in mortality from the disease. On theone hand, sharp reductions were observed in Finland, Sweden, and Iceland, which implemented nationwide(70-80% coverage) screening programs in the mid-1960s, whereas during the same ensuing 20 years, theincidence and mortality in Norway, where screening was performed in only one county, did not change.18

Although not the results of randomized trials, these comparisons point to the efficacy of the Pap test inreducing the incidence of cervical cancer, and associated mortality.


IV STRATEGIES TO IMPROVE THE EFFECTIVENESS OFTHE PAP TEST

The use of the Pap test is part of a broader process that involves several discrete steps. These includerecruitment of the patient, collection of cervical cells and preparation of a Pap smear, processing, screeningand interpreting the smear, reporting the findings, possible treatment, and follow-up. Strategies to increaserecruitment of unscreened and underscreened women, improvements in the methods of obtaining the Papsmear, preparation techniques, the screening of smears, and quality assurance in the laboratory can eachimpact, in varying degree, on the results of the Pap test and the screening process.

(i) Recruitment

All women who have ever been sexually active are at risk for cervical cancer and are encouraged to haveregular Pap smears. In the absence of an organized program, as stated above, the usual method of recruitmentfor Pap smears is by suggestion from a physician or upon “demand”. This method is adequate for women whoregularly visit their physicians or are educated about their own health care. However, many women whodevelop invasive disease have never been screened. Canadian studies of invasive cervical carcinoma casesrevealed that in PEI and BC, 65% and 49% of these women, respectively, either never had a smear or failedto have a smear in recent years.19,20 These women could be targeted for recruitment to further reducefatalities from cervical cancer.

Women do not have regular contact with the health care system or do not have regular Pap smears forvarious reasons. Barriers that prevent these women from participating in screening include fear ofgynecological examination, fear of cancer, concern about the gender of the smear taker, and loss ofconfidence in the health care system in general.18,21 Language and communication barriers may exist and lackof transportation or child care can prevent women from initial screening or follow up of a “positive” test.10

Canadian studies found that in particular, women who had a low level of education, whose mother tonguewas not English or French, Native women, recent immigrants and women over 60 yrs, were found to be verylow users of the Pap smear.20,22 The absence of screening in women over 60 yrs is especially serious sinceabnormalities are more advanced by the time they are detected, resulting in more invasive cases and highermortality rates than in younger women.19,23

In order to be successful, methods to encourage cervical cancer screening need to be sensitive to culturalfactors. In addition, there is a need for awareness that women have other health, social, and economic needsthat may interfere with recruitment.8 Approaches such as health visitors in the home, that reach beyond thephysician’s office, could recruit women missed by traditional methods.24 Other women may gain access tothe health care system through venereal disease clinics or penal institutions.25

A registry linked to a population-based information system would have the potential to increase recruitmentthrough the consistent recall and retesting of women. It would facilitate follow-up of cases diagnosed as‘abnormal’ and recall of women for regular Pap smears.10

At present, the best methods of health promotion and recruitment are still being investigated in Canada,8,26

but both systematic follow-up and the targeted screening of women never tested have been consistently raisedas main strategies to reduce the number of invasive cases.


(ii) Sample Collection

The quality of the smear has a large impact on the sensitivity of the Pap test and can significantly influencethe false-negative rate. This can be influenced by the quality of the sample collected, the instrument used totake the sample, and the skill of the sampler.

Controversy exists between smear takers and cytotechnologists over the composition of an adequate sample.It is acknowledged that the transformation zone, the junction between the squamous and columnar epithelia,is the origin of most squamous epithelial abnormal changes and is therefore the main target area for cellcollection. Typically, the presence of endocervical cells has been used to indicate that the transformation zonehas been sampled. However, the location of the transformation zone varies with individual anatomy, movestowards the ectocervix after childbirth, and is indrawn into the endocervix with menopause.27 In addition,women who are post-menopausal, pregnant, or using oral contraceptives may show very few endocervicalcells in a smear.18 Also it is possible for a sample to have abnormal cells without the presence of endocervicalcells.28 It has been suggested that the presence of squamous metaplastic cells may be a more accurateindicator that the transformation zone has been sampled, although these cells may be more difficult toidentify.27,29

The quality of a smear may also be reduced by the presence of inflammatory cells, necrotic debris, or bloodwhich may obscure the viewing of cancer cells.16 Further cooperation is required between smear takers andthe laboratories to determine what is indicative of an adequate sample.

The appropriateness of the sampling instrument is best determined by the characteristics of the individualcervix and the location of the transformation zone.27,29 A device is desired that will sample both theendocervical canal and the ectocervix to ensure sampling of the transformation zone. A variety ofinstruments are available that can be used alone or in combination, however, results of comparative studiesare inconsistent as to whether these methods differ in their ability to detect neoplasia (CIN).28,30-36 Within thecontext of a screening program with regards to cost and ease of sampling, a simple technique would bepreferred.

It is essential that health care workers be provided with training on the proper technique of obtaining acervical smear. Several studies have demonstrated that theoretical and practical training can significantlyimprove the quality of cervical smears taken.37-40

(iii) Preparation Techniques

Despite adequate collection of cervical cells for a Pap smear, poor transfer of the cells to the slide may resultin a sample too thick or uneven for accurate viewing. Additionally, the cervical cells may be obscured byblood, mucus or inflammatory cells. In an attempt to improve the quality of the slide sample, new automatedmethods of monolayer preparation have been introduced. (ThinPrep, CYTYC Corp., Marlborough, MA,USA, and CytoRich, Roche Image Analysis Systems, Inc., Elon College, NC, USA).41,42 The collection ofcervical cells remains the same, but the cells are rinsed from the collection device into a vial of preservativesolution. Automated processors gently disperse the sample, then separate cervical epithelial cells from debrisby filtration (ThinPrep) or density gradient centrifugation (CytoRich). The monolayer is preparedautomatically from the diagnostic cells onto a glass slide and stained for viewing. An evaluation of these twodevices has been recently reported by McGoogan & Reith (1996) from the University of Edinburgh(Scotland) Department of Pathology.43


Investigators have compared the automated monolayer preparation systems with the conventional cervicalsmear mainly using patients that were high risk or with known abnormalities. In these cases agreement indiagnosis between the two methods is high, being in the range of 88.3% - 99%.44,45 The enhanced cellularpreservation and even distribution of cells unobscured by other material makes slides prepared by monolayereasier to view and the review time is decreased substantially,45,46 however, fatigue sets in more quickly.43

There is inconsistency in the amount of endocervical cells reported in monolayer preparations, but thereduced total number of cells can increase the number of unacceptable slides.44,46-50 Overall, many studiesreport that monolayer preparation slightly improves detection of low and high grade disease, perhaps due tothe superior cell preservation and distribution.42,44,47 However, to attain accurate diagnosis, substantial trainingfor cytotechnologists and pathologists is required to interpret the unfamiliar patterns and enhanced nucleardetail in monolayer samples.43 These monolayer preparation systems are not routinely used in Canada duein part to their high cost.51

(iv) Screening

The process of screening Pap smears is a time-consuming task, even for the most experiencedcytotechnologist. Considering that each smear contains as many as 300,000 epithelial cells and carefulscreening requires a minimum of 5 minutes per slide, even under optimal circumstances, no cytotechnologistcan adequately screen more than 90 smears in a average working day.4 (Many factors, however, influencehow many slides are screened by a cytotechnologist. These factors include, for example, the patientpopulation, specimen quality, and individual abilities of a cytotechnologist). Faulty interpretation may beattributable to professional fatigue and habituation (the preconceived notion that a smear will be normalbecause only a small number of smears is actually abnormal) during screening, among other factors.52

In 1988, the Clinical Laboratory Improvement Amendments (CLIA ‘88) mandated rescreening of at least10% of all gynecologic cases initially interpreted as “negative” by the cytotechnologist, as a means of qualitycontrol in laboratories across the United States.53 This review was to include “negative” cases selected atrandom as well as cases from patients classified as “high-risk” by clinical history.53 Despite criticism of itsappropriateness as a quality control tool, ten percent random rescreening is a common practice in the UnitedStates.54 In the panoply of rescreening strategies, targeted rescreening of preselected cases (e.g., previoushistory of SIL) and rapid (30-second) rescreening of “negative” smears, for example, have been proposedas alternative manual candidates for more efficient quality control in cytopathology laboratories.54,55

Research into automation of the visualization and diagnosis of Pap smears has increased in recent years inan attempt to reduce false-negative interpretations resulting from human error.52,56 Improvements in the price-to-performance ratio of electronics, shortages of cytotechnologists, and in the United States, the increasedscrutiny by the mass media as false-negative cases have been made public and the associated legal liability,have combined to increase interest in the development of automated techniques for cervical cancerscreening.56 Automated screening devices can be used for prescreening (primary screening) or rescreeningof Pap smears (Table 2).52,56 These two modes have different implications for cytopathology laboratories,although the automated device may be doing essentially the same thing in both situations.52,56

The Food and Drug Administration (FDA) of the United States has approved two automated devices forquality assurance rescreening of cervical cytologic smear slides. The first system to be approved “as anautomated cervical cytology rescreening device intended for use in the quality control and rescreening ofpreviously screened Papanicolaou smear slides”, the AutoPap 300 QC Automatic Pap Screener (NeoPath


Inc., Redmond, Washington, U.S.A.), is designed to process all “negative” slides from primarycytotechnologist screening, and to select 10 percent or more of the most “abnormal” fraction for manualreview.57 The second system which has received FDA approval in an “adjunctive” capacity in the qualitycontrol process, the PAPNET Testing System (Neuromedical Systems Inc., Suffern, New York, U.S.A.),is also designed to process all “negative” slides from primary cytotechnologist screening; however, 27 percentof the most “abnormal” fraction would be selected (human selection based on examination of “tiles”processed by the system) for manual review.58 It is important to note that even with these automated devices,cytotechnologists are still required for interpretation of smears. Two other systems, the AUTOcyteInteractive Screening System (Roche Image Analysis Systems, Inc., Elon College, North Carolina, U.S.A.),42

and the Cyto-Savant (Xillix Technologies Corp., Richmond, British Columbia, Canada),59 are currentlyundergoing clinical trials as part of FDA clearance. Another automated system, the CYMET A40(Morphometrix Technologies Inc., Toronto, Ontario, Canada), will be in clinical trials through summer, 1997(personal communication, Gordon Rosenblatt, February 11, 1997) (Table 2).

Regarding Health Protection Branch (Health Canada) notification status, AutoPap is “in clinical trials” as of3/6/94; PAPNET, AUTOcyte, and Cyto-Savant, on the other hand, have received notification (“normalstatus”) (Table 2). To date, none of the automated screening devices are being utilized in practice for primaryPap smear screening. An advisory committee to the FDA recently reviewed the AutoPap system for thispurpose and recommended that additional studies are needed on its effectiveness in the laboratory settingprior to recommendation for clearance to market the system as a primary screener for Pap smears.60 TheAutoPap system is to date the only primary automated screener to be reviewed by the FDA panel, althoughother systems mentioned above are under investigation.

(v) Quality Assurance

Quality assurance is essential for all aspects of a cervical cancer screening program and can be integrated intoeach component from sampling and screening to treatment.

As stated above, a major source of error in missed invasive cases is due to poor samples, but training of thesampler can significantly improve the quality of smears. Laboratories could work with clinicians bymonitoring smears from specific clinicians and offering feedback on their quality.61

The importance of laboratory quality control was recognized by the 1989 Canadian National Workshop onCervical Cancer Screening which advocated the 1989 revised guidelines for quality assurance programs incytopathology produced by the Canadian Society of Cytology.8 In addition, some provinces have developedtheir own quality assurance measures such as those recommended by the Laboratory Quality Assurance Sub-Committee of the College of Physicians and Surgeons of Saskatchewan62 and the Laboratory ProficiencyTesting Program in Ontario.51

However, there is not yet consistent, systematic guidelines for cervical cytopathology laboratories in Canadaas in other countries, such as in Scotland63 and in Australia.64 These guidelines address issues such asworkload and training of cytotechnologists, consistency of diagnostic terminology, rescreening methods, thecorrelation of cytology and histology, and standard record keeping, as areas for the development of internaland external quality assurance. Guidelines are also required for consistent management, treatment and follow-up by clinicians and specialists.65


V COST-EFFECTIVENESS OF AUTOMATEDRESCREENING STRATEGIES

(i) Type of Analysis, Viewpoint, Cost Measurement, & Outcome Measurement

A ‘cost-effectiveness’ analysis from the perspective of a third party payer, comparing AutoPap 300 QC(AutoPap) and PAPNET, with a strategy of 10 percent manual random rescreening was conducted. Asacknowledged earlier, despite being generally discredited as a QC practice, ten percent random rescreeninghas been used as the standard against which to measure the cost-effectiveness of the automated strategiesin this analysis to facilitate comparison between the strategies. AutoPap and PAPNET have been selectedas comparators for this analysis due to availability of published clinical trials. Both devices have beenavailable for use by the cytopathology community in the United States following clearance by the FDA forrescreening of slides previously examined by humans (In Canada, PAPNET is offered for QC purposesthrough private clinics). At present there is a lack of sufficient published clinical data for the other automateddevices identified in the previous section, for purposes of this economic evaluation.

Decision analysis66,67 was used to calculate the expected total direct laboratory costs of screening under eachof the three strategies, and to derive the cost per additional abnormal case discovered. The cost estimateswere derived from values quoted in the relevant literature, and in addition, for the automated strategies, froma mail questionnaire (Appendix 1) to the respective manufacturers. Clinic and follow-up (colposcopy andtreatment) costs were excluded from this analysis. The costs to patients (e.g., productivity loss, anxiety) havenot been included, due to the perspective taken for the analysis.

(ii) Analysis

A decision-analysis model was constructed to compare the direct laboratory costs of the alternativerescreening strategies for cervical cancer (Figure 1). The three options (10% Manual, upper branch;AutoPap, middle branch; PAPNET, lower branch) are located at the square decision node. For either of thethree strategies, circular nodes represent events that may occur by chance, and rectangles represent total costsfor each pathway. All three options enter the subtree depicted to the right of the bracket. For all options,the subtree has a similar structure. The first chance node represents the true rate of abnormal cases (pRate)in the screening population (Table 3). Subsequent nodes represent chance events in terms of routine practicein a cytopathology laboratory where the three rescreening strategies are operational.57,58,74,75 In the process,every smear is examined by a cytotechnologist and is interpreted to be either negative or positive, with allpositive interpretations being reviewed by a pathologist. If both a cytotechnologist and a pathologist interpretthe smear to be positive, it is given a final interpretation of positive. Alternatively, the pathologist may revisea positive interpretation from the cytotechnologist to a final result of negative. Under the ten percent manualrescreening strategy, a random 10% of all negative determinations from initial cytotechnologist screening willundergo review by the cytotechnologist. Under each of the automated strategies, on the other hand, allnegative determinations from the cytotechnologist are submitted to the automated device in question. If theautomated device determines the smear to be suspect (“review”), then the smear is routed back to thecytotechnologist for review and then, if determined to be positive, to the pathologist. Any smears that theautomated device accepts as negative, or redirects as “review” but the cytotechnologist still classifies asnegative, are given the final determination of negative. The decision analysis model was constructed andevaluated using the software program DATA™ 2.6.76


Seven major assumptions were made in the decision model: 1) the number of Pap smears to be screened is4,000,000. This number is derived from the estimated number of Pap tests performed per year in Canada,based on data from national and provincial surveys;3 2) the proportion of true abnormal Pap smears is 10%.This percentage is derived from the proportion of women having an abnormality on the first Pap test (8%),3and an estimate of 25% for the false-negative rate of current screening programs;16,17 3) the initial screeningand review sensitivity of the cytotechnologist is 75%.16,17 In practice, in comparison to initial sensitivity, itmay be reasonable to assume a higher review sensitivity of the cytotechnologist, for example due to increasedvigilance;74 4) the pathologist review sensitivity is 100%. This ideal sensitivity rate is assumed for thepathologist in order to accurately interpret the cost/additional abnormal case within the laboratory; 5) thecytotechnologist initial screening and review specificity is 95%.71 Similar to the case of sensitivity, inpractice, a heightened vigilance may increase the review specificity of the cytotechnologist (and/orpathologist). However, it may also be reasonable to assume, for example due to a tendency to “believe themachine”, a lower specificity for redirected smears;74 6) the direct cost estimates per slide forcytotechnologist screening is Can$8, and Can$5 for either cytotechnologist or pathologist review. Theseestimates were derived (adjusted for overhead) from the Ontario Ministry of Health Schedule of Benefits,under services listed under Laboratory Medicine;73 7) the direct cost estimates per slide for rescreening byAutoPap is Can$7, and Can$14 for rescreening by PAPNET (Appendix 1). The baseline values used in thedecision model are summarized in Table 3.

(iii) Results

For each additional abnormal case found by 10% random rescreening, the cost would be $250 (Table 4). Foreach additional case found by the least expensive automated strategy (AutoPap) the cost rises to over $400(Table 4). PAPNET is marginally more effective (rescreening sensitivity 83% vs. 80%), but the cost to findan abnormal case, in comparison to AutoPap is very high: over $10,000 (Table 4). This result highlights thefact that both automated rescreening strategies are virtually equal in efficacy, but quite different in cost, andit is this cost difference which accounts for the difference in economic attractiveness of the two strategies.

There is no widely accepted threshold for cost-effectiveness analysis at which programs become economicallyattractive. It may well be the case that society is willing to pay $10,000 to find an additional abnormal result.However, given the precision of our estimates of screening efficacy (low), we suggest that the data beinterpreted as follows: Automated rescreening methods are more effective, but also more costly than manualrescreening. The cost per case found for AutoPap appears to be reasonable. The automated programsappear to be of similar efficacy, but dissimilar cost. Given these estimates, the marginal cost difference forPAPNET does not appear justified.

(iv) Uncertainty

Values used in the decision model were varied widely from the baseline case to examine the impact ofuncertainty of these values (Table 5). 10% random rescreening was found to have a better cost-effectivenessratio than either AutoPap or PAPNET over a wide range of values for key variables under one-way sensitivityanalysis (Table 5). The baseline result in terms of a favorable cost-effectiveness ratio of the 10% manualroute in comparison to the automated devices, however, is sensitive to the per rescreening cost by automateddevice, and to the per review cost by a cytotechnologist or pathologist. Assuming each of the remainingvariables constant, 10% random rescreening would, for example, be less cost-effective than AutoPap if theper rescreening cost under AutoPap falls below $4, or the per review cost exceeds $7.


The cost-effectiveness ratios for each of the automated devices are sensitive to the per rescreening cost byautomated device. Assuming each of the remaining baseline variables constant, AutoPap will be less cost-effective than PAPNET if the per rescreening cost under each automated device begins to exceed $5 (forexample, at a per rescreening cost by automated device of $6, the cost-effectiveness ratio for AutoPap is$359 and for PAPNET $358; at a per rescreening cost of $10, the cost-effectiveness ratio for AutoPapincreases to $594 and for PAPNET $584).

(v) Related Studies

In a recent analysis of these automated strategies with alternative methods of manual rescreening,78

Hutchinson (1996), using a conventional mathematical model considering overall sensitivity in a three-stagescreening algorithm (cytotechnologist screening, rescreening by the automated device, and cytotechnologistreview) reported that the most effective route for recovering false negatives in the “negative” pool wouldbe 100% manual rescreening. In comparison to a no rescreen strategy, Hutchinson estimated that theincremental cost per additional abnormal case (LSIL+: prevalence, 2.50%) discovered by AutoPap (US$2,197) was twice as high than that discovered by 100% rescreening (US$ 1,049), and the incremental costper additional abnormal case discovered by PAPNET (US$ 4,486) was four times that of 100% rescreening.78

Despite being the least favorable of all rescreening strategies (manual and automated) in terms ofeffectiveness, ten percent random rescreening was reported to have the same cost-effectiveness ratio as thatof 100% manual rescreening.78

One hundred percent manual rescreening of all “negative” slides, however, is an unrealistic goal in the currentclimate of fiscal constraints and shortage of human resources in cytopathology laboratories.56 Of theremaining manual rescreening strategies, Hutchinson reports that rapid rescreening shows the greatest benefitper unit of cost (US $348 per additional abnormal case discovered).78 Rapid rescreening has recently beenproposed as an alternative strategy for quality control of “negative” slides by Baker et al (1995):55 Under thisproposal, all “negative” slides are rescreened by an ‘experienced’ cytotechnologist for a 30-second period.Based on over 100,000 “negative” slides over a period of four years, Baker et al. reported a fivefold increasein the pickup of false-negatives in comparison to 10% random rescreening.55 Concerns, however, over bothcytotechnologist fatigue and the high false-positive rates (“alarms”) of the screening cytotechnologists, atleast during the initial stages of utilization, need to be addressed before adoption of rapid rescreening incytopathology laboratories.51

(vi) Limitations

Two broad limitations of the current analysis need to be highlighted. The first limitation relates to thedifficulty in deriving estimates of the operating characteristics of the automated devices. These devices havegenerally been evaluated under controlled conditions (small sample, selected group of lesions) bymanufacturer-sponsored researchers as opposed to independent analysis in routine practice. The comparator(and study design) against which each of the devices have been measured, moreover, varies between studiesmaking it difficult to compare studies and assess the performance (screening efficacy) of the devices. Relatedto this latter point is to date the lack of a common definition of the gold standard for Pap-smearrescreening.79 Second, the downstream costs and effects of the competing rescreening strategies were notconsidered in this analysis. This may dramatically change the economic attractiveness of all of thesecompeting strategies. For example, while automated devices are designed to decrease false-negative Papsmears, it is too early to tell if they will significantly increase the number of false positives, thereby increasingpatient anxiety and suffering and adding still more unnecessary costs to screening programs.80


VI EMERGING TECHNOLOGIES

In addition to recruitment, sampling, preparation and screening techniques, two other strategies which couldimpact on cervical cancer screening in the future may include screening for HPV in asymptomatic womenand flow cytometry.

(i) HPV Testing

There is a strong body of evidence to associate epithelial cell transformation by human papillomaviruses withcervical cancer; although there is evidence that the presence of other social, sexual or nutritional factors canfurther increase the risk of cervical cancer.81-83 HPV infection may be observed visually by colposcopy as thepresence of raised plaques on the cervix (condylomas), or microscopically as the presence of epithelial cellswith enlarged atypical nuclei and perinuclear clearing (koilocytosis). Recent methods more precisely identifyHPV through the detection of its DNA, although there may be no other visible signs of HPV infection.

HPV is sexually transmitted and over 70 types have been identified. Specific types of HPV (e.g., types 16and 18), designated 'high risk' types, are found more frequently in invasive neoplasia and carcinoma.84-86 ‘Lowrisk’ HPV types (e.g., types 6 and 11) are most often associated with condyloma or mild dysplasia which donot usually progress to invasive disease.87 It is well established that the frequency of HPV in cervical cellsis directly related to the severity of CIN. The occurrence of HPV is about 9-27% in normal smears, 18-31%in atypical smears, and 48.6-92% in CIN diagnosis. The prevalence is highest in biopsy (46-92%) andcarcinoma (88%) samples.84,88-91 This connection was substantiated in an international study by Bosch et al.(1996) who found that 93% of invasive-disease samples from patients in 22 countries were HPV positive.85

In Canada, HPV testing methods are still confined to the research arena due to their complexity and cost.These methods include either gene amplification (polymerase chain reaction, PCR), hybridization againstspecific DNA probes (dot blot, Southern blot, or in situ hybridization) or the use of RNA probes (hybridcapture test).92 Direct comparison studies show that PCR is the most sensitive test compared to dot blot,Southern blot, and in situ hybridization tests (34%, 68.4% and 72.7% detection of HPV as compared to PCR,respectively),93-95 but this high sensitivity requires careful skill to avoid false-positives from external DNAcontamination. The Southern blot technique offers both high sensitivity and specificity, but is labour intensiveand not suited for routine clinical use.96,97 In situ hybridization is used most commonly, and like hybridcapture techniques, it is more economical and can be used on fixed specimens. Studies are constantlyemerging that improve methods to increase flexibility, sensitivity, specificity and ease of performance. Forexample, PCR can be adapted to fixed tissue samples,98 combined with enzyme immunoassay99 and there isa move to automate in situ methods.100

Although these tests can separate HPV types associated with lower or higher risk of invasive neoplasia, theclinical predictive applications are uncertain. Many researchers have suggested a predictive role for HPVtesting in conjunction with cytology or colposcopy to identify abnormal cases that are more likely to progressto cervical cancer. 90,101-103 Long term studies show that type HPV 16 infections are more persistent and morelikely to progress to an invasive stage, but a high percentage will regress spontaneously in both normalpatients (83.7%)101 and patients with CIN 1 or 2 (55.7-84%).104,105 Also, patients positive for HPV can shownormal cytological diagnosis.106 Since a positive HPV test does not confidently predict the development ofcervical cancer, identification of women who may be at higher risk for cervical cancer by present HPV testing


methods may lead to unnecessary aggressive treatment. In 1995, the Canadian Task Force on the PeriodicHealth Examination recommended that HPV screening not be performed on asymptomatic women whichreinforced a similar recommendation of the 1989 National Workshop.8

Presently, a large trial at National Cancer Institute (U.S.A.), is evaluating whether HPV testing candistinguish which women diagnosed as ASCUS (mildly atypical) should undergo colposcopy or closeobservation.14 Since an ASCUS diagnosis is often unclear, this would be a potentially important role for HPVtesting. However, through a mathematical model Jenkins et al (1996)107 project that HPV testing of womenwith mildly abnormal diagnoses would only minimally reduce the number of cancers detected. They reasonthat the greater number of undetected invasive cancers occur in women who have never been screened orwho are inadequately screened.

The reliability of HPV testing as a screening method would require further research on the natural historyof HPV infections and the clarification of the link between HPV presence, HPV type, and the subsequentdevelopment of cervical cancer. In addition, until the cost and complexity of HPV testing is reduced, thepractical benefits of HPV testing in cervical cancer screening are questionable.

(ii) Flow Cytometry

Flow cytometry (FCM) is a well established method in pathology for detection and quantification of differenttypes of cells.108 FCM is capable of rapidly evaluating large numbers of cells and is accorded with a measureof objectivity because its results are reported as graphic displays.109 FCM can rapidly quantitate cell size,nuclear size, nuclear:cytoplasmic ratio, and the expression of cellular antigens.110 FCM is widely used formeasurement of cellular DNA content, which 1) provides an estimate of a tumor’s proliferative activity, and2) identifies populations of cells with abnormal amounts of DNA.

DNA analysis by FCM requires a fluid suspension of single cells or nuclei.109 The suspended cells are stainedwith a DNA-specific dye (e.g., propidium iodide), and then examined by a flow cytometer as they flow insingle file past a narrowly focused beam of light, the source of which is usually a laser or a mercury arc lamp.The stained cells will emit a fluorescent pulse, the height of which will be proportional to the amount of DNAin the cell. The pulse heights of thousands of cells can be measured in minutes, and the data displayed as aDNA histogram. The nature of the peaks (diploid/aneuploid) will determine the (ab)normality of the cells.109

Attempts have been made to screen cervical samples using FCM DNA analysis in a research context. Therehave been claims in earlier published studies of 94-97% sensitivity110,111 and 82-88% specificity110,111 of FCMin detection of cervical abnormalities. Sample preparation times are longer than for the Pap test and therecan be difficulties in disaggregating cells.112 FCM, at present, is still seen as demanding (problems ofstandardization, complex S-phase analysis) for routine pathology services, and expensive (>$50 per test) inthe context of screening.108


VII CONCLUSIONS

The above techniques for cervical cancer screening may increase the effectiveness of the Pap test for thosewomen presently screened, but will not affect detection in women who are rarely or never screened.Barriers that prevent these women from participating in screening may put them at a higher risk for cervicalcancer. Opportunistic screening leads to over screening of younger, affluent, lower-risk women and under-screening of older, less affluent and minority group women.113 The targeted recruitment of women incoordination with an information system that would allow regular follow-up and recall of women couldsignificantly reduce the number of cervical cancer cases.

It is essential that promotion of technological improvements does not divert resources and effort from theimplementation of the main recommendations of the 1989 National Workshop, namely recruitment,information systems, and training and quality-control requirements for laboratories and programs.8

Canadian Coordinating Office for Health Technology Assessment 17

TABLES

Table 1: Databases Searched and Description of Searches

DATABASE DESCRIPTION OF SEARCH

MEDLINECancerlitHlth.Plan&AdminEMBASEPascal

1985-1997

(Vaginal Smears or Cervix Neoplasms) + Terms below invarious strategies. Instrumentation; (Economics or Cost-BenefitAnalysis); (Mass Screening or Screening); Barrier*;(Histocytological Preparation Techniques or Cytodiagnosis orCytological Techniques); (Equipment Failure or EquipmentDesign); Education, Medical; (Physicians, Family or Nurses orNurse Practitioners); Automat*; DNA Probes, HPV and FlowCytometry); Staining, Quality, Sensitivity or Specificity; Review

MEDLINE1985-1997

1. ((Cervix Neoplasms or Vaginal Smears) or IntraepithelialNeoplasia or Vaginal Neoplasms; Cervix))2. (Flow Cytometry or DNA Probe? or Pap smear? orCervicograph?) and (Screen? or Test? or Diagnosis or Lab?)3. 1 or 2 and Review Literature or Meta-Analysis

DIALOG PTS Newsletter Search for recent news on cervical cancer screening.

FDA Online Database The Gray Sheet for new approvals of tests, instruments, etc.Slide Preparation techniques

ECRINet 1. Cervix and Screening2. Vaginal smears

CURRENT CONTENTSClinical Medicine

Searches conducted as a current awareness service.(Vaginal Smears or Cervix neoplasms) and Screening

CCOHTA Library Database All relevant terms

* or ? = truncated term

Canadian Coordinating Office for Health Technology Assessment18

Table 2: Automated Pap-smear Screening

SLIDE COST/ FDA/HPB*MANUFACTURER: PRODUCT: MODE: PREPARATION: SLIDE: (APPROVAL/NOTIFICATION):

NeoPath, Inc. AutoPap 300 QC QC† conventional US $5 FDA- 9/29/95Redmond, Wash., USA HPB- in clinical trials as of 3/6/94

AutoPap Primary prescreen‡ conventional - -Screener

Neuromedical Systems, PAPNET QC† conventional US $10 FDA- 11/8/95Inc., Suffern, NY, USA HPB- “normal status” as of 10/27/94

Roche Image Analysis AUTOcyte prescreen‡ monolayer - FDA- in clinical trials through 9/96Systems, Inc., Elon College, HPB- “normal status” as of 3/20/95NY, USA

Morphometrix Technologies CYMET A40 prescreen‡ monolayer - in clinical trials-summer/97Inc., Toronto, ON, Canada

Xillix Technologies Corp., Cyto-Savant prescreen‡/ conventional/ - FDA- in clinical trialsRichmond, BC, Canada QC† monolayer HPB- “normal status” as of 6/23/93

* FDA = Food and Drug Administration (U.S.A.); HPB = Health Protection Branch (Health Canada)† In the QC mode, slides would first be screened by cytotechnologists, then “negative” cases would be rescreened by the automated

device (refer to text)‡ In the prescreening mode, slides are examined first by the automated device and subsequently by the cytotechnologist


Table 3: Baseline Values used in the Decision Model

BASELINE STUDYVARIABLE: INTERPRETATION: VALUE: (Ref no.):

pRate True rate of abnormal Pap smears 0.10 (3,16,17)pSens1 Cytotechnologist screening sensitivity 0.75 (16,17)pSens2 Pathologist review sensitivity 1.00 (*)

pReview Review rate 10% random 0.10 (53) AutoPap 0.10 (68) PAPNET 0.27 (69)

pSens3 Rescreening sensitivity 10% random 1.00 AutoPap 0.80 (68,70) PAPNET 0.83 (71)

pSens4 Cytotechnologist review sensitivity 0.75 (*)pSens5 Pathologist review sensitivity 1.00 (*)pSpec1 Cytotechnologist screening specificity 0.95 (*,71)†

pSpec2 Rescreening specificity 10% random 0.00 AutoPap 0.80 (70) PAPNET 0.80 (*,71,72)†

pSpec3 Cytotechnologist review specificity 0.95 (*)pSpec4 Pathologist review specificity 1.00 (*)N Number of Pap smears 4,000,000 (3)cScreen Per screening cost by a cytotechnologist 8 (73)

cRescreen Per rescreening cost by automated device AutoPap 7 (‡) PAPNET 14 (‡)

cReview Per review cost by a cytotechnologist or pathologist 5 (73)

* authors’ estimate† authors’ estimate based on cited studies‡ mail questionnaire (Appendix 1)


Table 4: Baseline Analysis

TOTAL ADDITIONAL COST/ADDITIONALCOSTS: ABNORMAL ABNORMAL ABNORMAL CASE ($)

STRATEGY: (1,000$): CASES: CASES: A:* B:†

No rescreen 34,400 300,000 0 Base case Base case

10% rescreen 36,283 307,500 7,500 251 251

AutoPap 59,469 360,000 60,000 418 442

PAPNET 84,846 362,250 62,250 810 11,278

* incremental cost/additional abnormal case in comparison to a no rescreen strategy.† incremental cost/additional abnormal case in comparison to the next most expensive strategy.


Table 5: One-way Sensitivity Analysis on Key Variables

BASELINE RANGE OF COST/ADDITIONAL ABNORMAL CASE* THRESHOLDVARIABLE: INTERPRETATION: VALUE: VALUES: 10% random: AutoPap: PAPNET: VALUE:†

pRate True rate of abnormal 0.10 0.01-0.20 2,645-118 4,468-193 8,658-374 Not found‡Pap smears

pSens1 Cytotechnologist 0.75 0.50-0.99 131-5,997 215-10,137 418-19,644 Not found‡screening sensitivity

pSens3 Rescreening sensitivity 0.50-0.99 (251) 668-338 1,343-680 Not found‡ AutoPap 0.80 PAPNET 0.83

pSens4 Cytotechnologist review 0.75 0.50-0.99 374-191 626-317 1,215-614 Not found‡sensitivity

pSpec1 Cytotechnologist 0.95 0.50-0.99 138-261 226-435 439-843 Not found‡screening specificity

pSpec2 Rescreening specificity 0.50-0.99 (251) 427-412 834-796 Not found‡ AutoPap 0.80 PAPNET 0.80

pSpec3 Cytotechnologist review 0.95 0.50-0.99 354-242 420-418 817-810 Not found‡specificity

cRescreen Per rescreening cost by automated device 2-20 (251) 124-1,180 132-1,150 5 AutoPap 7 PAPNET 14

cReview Per review cost by a 5 2-20 100-1,004 413-439 799-866 Not found‡cytotechnologist or pathologist

incremental cost/additional abnormal case in comparison to a no rescreen strategy.† The threshold (77) value is a value where the two automated devices have equal expected cost-effectiveness ratios. If a given variable has a value less than a threshold, than

one strategy is more cost-effective; if the variable has a value greater than a threshold, than the alternative strategy is more cost-effective.‡ AutoPap is more cost-effective than PAPNET over all the given range of values (no threshold).


10% Manual

AutoPap

PAPNET

ABNORMAL

NORMAL

Positive

Negative

“Review”

Positive

pSens1

Positive

Negative

Negative

Positive

Negative

Negative

Rescreen

“Review”

Positive

Negative

Positive

Positive

Negative

pSens2

pSens3

pSens4

pSens5

pSpec1

pSpec2

pSpec3

pRate

#

#

#

#

#

#

#

#

#

#

pSpec4

#

#

NegativepSpec4

RescreenpReview

pReview

Negative

Negative

Negative

Negative

= decision = chance event = costs for each pathway

Positive#

# = complementary probability

Figure 1. Decision Model for Rescreening Strategies of Cervical CancerN(cScreen+cReview)

N(cScreen+cReview)

N(cScreen+cRescreen+2(cReview))


N(cScreen+cRescreen+cReview)

N(cScreen+cRescreen)


N(cScreen+cRescreen+cReview)





N(cScreen+cReview)

N(cScreen+cReview)

FIGURES


APPENDIX

Appendix 1: Cervical Cancer Screening - Questions re: Automatic Screeners of Pap Smear

1. Name of screener (e.g. AutoPap 300)

2. Is it to be used for prescreening of slides or for quality assurance ?

3. What is it’s sensitivity? specificity?and rates of false positives? false negatives?

4. What are the hardware and software requirements?

5. Do the slides to be read require special preparation? Yes No

If yes, please elaborate.

6. Would it be possible for you to send any product information sheets that you might have? Yes No

7. What is the cost of the system?

8. Do you know what the cost/slide will be approximately?

9. What is the current status of this system? (e.g. still being studied, routine clinical use, etc.)

10. May I contact you again if I require further information? Yes No


REFERENCES

1. National Cancer Institute of Canada. (1996). Canadian Cancer Statistics 1996. Toronto: The Institute.

2. Parboosingh, E. J., Anderson, G., Clarke, A., Inhaber, S., Kaegi, S., Mills, C., et al. Cervicalcancer screening: Are the 1989 recommendations still valid? Canadian Medical AssociationJournal, 154 (12) : 1847-1853.

3. Goel, V., Agha, M. Developing a business case for cervical screening information systems.Report prepared for the Preventive Health Services Division, Health Canada. (Discussion Draft,November 6, 1995).

4. Shingleton, H. M., Orr, J. W., Jr. (1995). Cancer of the cervix. Philadelphia: J.P. Lippincott.

5. Australian Health Ministers’ Advisory Council. Cervical Cancer Screening Evaluation Committee.Cervical cancer screening in Australia: Options for change. Australian Institute of Health:Preventive Program Evaluation Series No 2. AGPS, Canberra, 1991.

6. Benedet, J. L., Anderson, G. H., Matisic, J. P. (1992). A comprehensive program for cervicalcancer detection and management. American Journal of Obstetrics and Gynecology, 166 (4) :1254-1259.

7. Task Force on Cervical Cancer Screening Programs. (1976). Cervical cancer screening programs.Canadian Medical Association Journal, 114 : 1003-1033.

88. Miller, A.B., Anderson, G., Brisson, J., Laidlaw, J., Le Pitre, N., Malcolmson, P., et al. (1991).Report of a national workshop on screening for cancer of the cervix. Canadian MedicalAssociation Journal, 145 : 1301-1325.

9. Cohen, M. M. (1996). Why is there no progress against cervical cancer? Canadian MedicalAssociation Journal, 154 : 1867-1869.

10. Health Canada. (1995). Proceedings: Interchange ‘95: A Canadian forum to collaborate oncervical cancer screening program implementation strategies. (February 27, 28 and March 1,1995, Ottawa, Ontario, Canada). Draft.

11. Canadian Coordinating Office for Health Technology Assessment. (1994). Guidelines for economicevaluation of pharmaceuticals: Canada. (1st ed.). Ottawa: CCOHTA.


12. Kurman, R. J., Solomon, D. (1993). The Bethesda System for reporting cervical/vaginalcytologic diagnoses: definitions, criteria, and explanatory notes for terminology and specimenadequacy. New York: Springer-Verlag.

13. Lieu, D. (1996). The Papanicolaou smear: its value and limitations. Journal of Family Practice,42 (4) : 391-399.

14. Titus, K. (1996). Abnormal Pap smears, ASCUS still OB/GYN puzzle. JAMA, 276 : 1014,1016.

15. Fahey, M. T., Irwig, L., Macaskill, P. (1995). Meta-analysis of Pap test accuracy. AmericanJournal of Epidemiology, 141 (7) : 680-689.

16. Gay, J. D., Donaldson, L. D., Goellner, J. P. (1985). False-negative results in cervical cytologicstudies. Acta Cytologica, 29 (6) : 1043-1046.

17. van der Graaf, Y., Vooijs, G. P., Gaillard, H. J., Daisy, M. D. (1987). Screening errors in cervicalcytology. Acta Cytologica, 31 (4) : 434-438.

18. Coleman, D., Day, N., Douglas, G., Farmery, E., Lynge, E., Philip, J., et al. (1993). Europeanguidelines for quality assurance in cervical cancer screening: Europe Against Cancer Programme.European Journal of Cancer, 29A (Suppl 4) : S1-S38.

19. Sweet, L., Tesch, M., Dryer, D., McAleer, C. (1991). A review of the cervical cytology screeninghistory of PEI women diagnosed with carcinoma of the cervix, 1981-1986. Chronic Diseases inCanada, 12 (1) : 1-3.

20. Goel, V. Factors associated with cervical cancer screening: Results from the Ontario Health Survey.(1994). Canadian Journal of Public Health, 85 : 125-127.

21. Carey, P., Gjerdingen, D. K. (1993). Follow-up of abnormal Papanicolaou smears among womenof different races. Journal of Family Practice, 37 (6) : 583-587.

22. Anderson, G. H., Benedet, J. L., Le Riche, J. C., Matisic, J. P., Thompson, J. E., (1992). Invasivecancer of the cervix in British Columbia: A review of the demography and screening histories of 437cases seen from 1985-1988. Obstetrics and Gynecology, 80 (1) : 1-4.

23. Remington, P., Lantz, P., Phillips, J. L. (1990). Cervical cancer deaths among older women:implications for prevention. Wisconsin Medical Journal, 89 (1) : 30,32-34.

24. Megevand, E., Van Wyk, W., Knight, B., Bloch, B. (1996). Can cervical cancer be prevented bya see, screen, and treat program? A pilot study. American Journal of Obstetrics and Gynecology,174 (3) : 923-928.


25. Canadian Task Force on Cervical Screening Programs. (1982). Cervical cancer screening programs:Summary of the 1982 Canadian task force report. Canadian Medical Association Journal, 127(7) : 581-589.

26. Clarke, E. A. Progress in screening for cervical cancer [Letter]. (1996). Canadian MedicalAssociation Journal, 155 (8) : 1037.

27. Thompson, D. W. (1993). Adequate “pap” smears: a guide for sampling techniques inscreening for abnormalities of the uterine cervix. Toronto: Laboratory Proficiency TestingProgram.

28. Sidawy, M. K., Tabbara, S. O., Silverberg, S. G. (1992). Should we report cervical smears lackingendocervical component as unsatisfactory? Diagnostic Cytopathology, 8 (6) : 567-570.

29. Sedlis, A., Saigo, P. E. (1992). Cervicovaginal cytology: New terminology. Current Problems inObstetrics, Gynecology and Fertility, 15 (5) : 207-241.

30. Bauman, B. J. (1993). Use of a cervical brush for Papanicolaou smear collection: A meta analysis.Journal of Nurse Midwifery, 38 (5) : 267-275.

31. Buntinx, F., Knottnerus, J. A., André J., Crebolder, H. F., Essed, G. G. (1994). The effect ofdifferent sampling devices on the presence of endocervical cells in cervical smears: A systematicliterature review. European Journal of Cancer Prevention, 3 (1) : 23-30.

32. Rivlin, M. E., Woodliff, J. M., Bowlin, R. B, Moore, J. L., Jr., Martin, R. W., Grossman, J. H.(1993). Comparison of cytobrush and cotton swab for Papanicolaou smears in pregnancy. Journalof Reproductive Medicine, 38 (2) : 147-150.

33. Szarewski, A., Curran, G., Edwards, R., Cuzik, J., Kocjan, G., Bounds, W., et al. (1993). Comparison of four cytologic sampling techniques in a large family planning center. ActaCytologica, 37 (4) : 457-460.

34. Germain, M., Heaton, R., Erickson, D., Henry, M., Nash, J., O’Connor, D. (1994). A comparisonof the three most common Papanicolaou smear collection techniques. Obstetrics and Gynecology,84 (2) : 168-173.

35. Pretorius, R. G., Sadeghi, M., Fotheringham, N. (1991). A randomized trial of three methods ofobtaining Papanicolaou smears. Obstetrics and Gynecology, 78 (5) : 831-836.

36. Schettino, F., Sideri, M., Cangini, L., Candiani, M., Zannoni, E., Maggi, R. (1993). Endocervicaldetection of CIN: Cytobrush versus cotton. European Journal of Gynaecological Oncology, 14(3) : 234-236.


37. Kant, A. C., Palm, B. T., Dona, D. J., Makkus, A. F., Vooijs, G. P., Van Weel, C. (1995). Cellularcomposition of cervical smears taken by general practitioners. International Journal of Qualityin Health Care, 7 (1) : 11-16.

38. Burkman, R.T., Ward, R., Balchandani, K., Kini, S. (1994). A continous quality improvementproject to improve the quality of cervical Papanicolaou smears. Obstetrics and Gynecology, 84(3) : 470-475.

39. Mitchell, H. (1993). Pap smears collected by nurse practitioners: A comparison with smearscollected by medical practitioners. Oncology Nursing Forum, 20 (5) : 807-810.

40. Thomassen, H., Lenci, P., Brake, I., Anderson, G. (1996). Cervical cancer screening performed bya nurse. Evaluation in family practice. Canadian Family Physician, 42 : 2179-2183.

41. Zahniser, D.J., Sullivan, P.J. (1996). CYTYC Corporation. Acta Cytologica, 40 (1) : 37-44.

42. Knesel, E.A., Jr. Roche Image Analysis Systems, Inc. (1996). Acta Cytologica, 40 (1) : 60-66.

43. McGoogan, E., Reith, A. (1996). Would monolayers provide more representative samples andimproved preparations for cervical screening?: Overview and evaluation of systems available. ActaCytologica, 40 (1) : 107-119.

44. Wilbur, D.C., Cibas, E.S., Merritt S, James, L. P., Berger, B. M., Bongiglio, T. A. (1994).ThinPrep™ Processor 1: Clinical trials demonstrate an increased detection rate of abnormal cervicalcytologic specimens. American Journal of Clinical Pathology, 101 : 209-214.

45. Geyer, J.W., Hancock, F., Carrico, C., Kirkpatrick, M. (1993). Preliminary evaluation of Cyto-Rich:An improved automated cytology preparation. Diagnostic Cytopathology, 9 (4) : 417-422.

46. Bur, M., Knowles, K., Pekow, P., Corral, O., Donovan, J. (1995). Comparison of ThinPreppreparations with conventional cervicovaginal smears: Practical considerations. Acta Cytologica,39 (4) : 631-642.

47. Awen, C., Hathway, S., Eddy, W., Voskuil, R., Janes, C. (1994). Efficacy of ThinPrep®preparation of cervical smears: A 1,000-case, investigator-sponsored study. DiagnosticCytopathology, 11 (1) : 33-37.

48. Laverty, C. R., Thurloe, J.K., Redman, N.L., Farnsworth, A. (1995). An Australian trial ofThinPrep: A new cytopreparatory technique. Cytopathology, 6 (3) : 140-148.

49. Hutchinson, M. L., Agarwal, P., Denault, T, Berger, B., Cibas, E. S. (1992). A new look at cervicalcytology: ThinPrep multicenter trial results. Acta Cytologica, 36 (4) : 499-504.


50. Sprenger, E., Schwarzmann, P., Kirkpatrick, M., Fox, W., Heinzerling, R. H., Geyer, J. W., et al.(1996). The false negative rate in cervical cytology: Comparison of monolayers to conventionalsmears. Acta Cytologica, 40 (1) : 81-89.

51. Geddie, W. R. (1996). Quality assurance tools in cervical cytology: Which ones, when and how?LPTP Newsletter, 176 (April 16) : 1-6.

52. Birdsong, G. G. (1996). Automated screening of cervical cytology specimens. Human Pathology,27 (5) : 468-481.

53. Clinical Laboratory Improvement Amendments of 1988, P.L.100-578. Congressional Record 1988;134 : 3828-3863.

54. Tabbara, S. O., Sidawy, M. K. (1996). Evaluation of the 10% rescreen of negative gynecologicsmears as a quality assurance measure. Diagnostic Cytopathology, 14 (1) : 84-6.

55. Baker, A., Melcher, D., Smith, R. (1995). Rapid rescreening of cervical smears: An effective methodof quality control. Acta Cytologica, 39 (2) : 273.

56. Grohs, H. K. (1994). Reflection on a changing enviroment: Gynecological screening in the age ofautomation. In: H. K. Grohs, O. A. Husain, (Eds.), Automated cervical cancer screening (pp. xi-xv). New York : IGAKU-SHOIN Medical Publishers.

57. Patten, S. F., Lee, J. S., Nelson, A. C. (1996). Neopath, Inc.: Neopath AutoPap 300 Automatic PapScreener System. Acta Cytologica, 40 (1) : 45-52.

58. Mango, L. J. Neuromedical Systems, Inc. 1996; Acta Cytologica, 40 (1) : 53-59.

59. Palcic, B., Garner, D. M., MacAulay, C. E., Matisic, J., Anderson, G. H. (1996). OncometricsImaging Corporation and Xillix Technologies Corporation: Use of the Cyto-Savant in quantitativecytology. Acta Cytologica, 40 (1) : 67-72.

60. (1996). Automated system for Pap slide review needs further study, says FDA panel. HealthTechnology Trends, 8 (11) : 4-5.

61. Turner-Smith, L. Quality assurance indicators (personal communication, September 18, 1995).

62. Health Services Utilization and Research Commission. (1997). A comprehensive approach tocervical cancer screening. (Summary Report no. 8). Saskatoon, SK.: Health Services Utilizationand Research Commission.


63. Working Party on Internal Quality Control for Cervical Cytopathology Laboratories. (1995). Reportof Working Party on Internal Quality Control for Cervical Cytopathology Laboratories.Edinburgh: Scottish Office Home and Health.

64. Working Party on Quality Assurance in Cervical Cytology. (1994). Performance standards forAustralian laboratories reporting cervical cytology. (draft). Australia. Department of HumanServices and Health, Cervical Cancer Screening Section.

65. Stuart, G. C., Parboosingh, J. (1996). Implementation of comprehensive screening for cervicalcancer in Canada: impediments and facilitators. Journal SOGC, 18 (12) : 1241-1250.

66. Sox, H. C., Blatt, M. A., Higgins, M. C., Marton, K. I. (1988). Medical decision making.Stoneham, MA : Butterworths.

67. Kassirer, J. P., Moskowitz, A. J., Lau, J., Pauker, S.G. (1987). Decision analysis: A progress report.Annals of Internal Medicine, 106 (2) : 275-291.

68. Wilbur, D. C., Bonfiglio, T. A., Rutkowski, M. A., Atkison, R. M., Lee, J. S., Patten, S. F. (1996).Sensitivity of the AutoPap 300 QC system for cervical cytologic abnormalities: Biopsy dataconfirmation. Acta Cytologica, 40 : 127-132.

69. Ouwerkerk-Noordam, E., Boon, M., Beck, S. (1994). Computer-assisted primary screening ofcervical smears using the PAPNET method: Comparison with conventional screening and evaluationof the role of the cytologist. Cytopathology, 5 (4) : 211-218.

70. Anderson, T. L. (1994). Automatic screening of conventional papanicolaou smears: the AutoPap®300 and AutoPap® 300 QC Systems. In G. L. Wied, P. H. Bartels, D. L. Rosenthal, U. Schenck,(Eds.), Compendium on the computerized cytology and histology laboratory (pp.306-311).Chicago: Tutorials of Cytology.

71. Rosenthal, D. L., Acosta, D., Peters, R. K. (1996). Computer-assisted rescreening of clinicallyimportant false-negative cervical smears using the PAPNET testing system. Acta Cytologica, 40(1) : 120-126.

72. Mango, L. J. (1994). Computer-assisted cervical cancer screening using neural networks. CancerLetters, 77 (2-3) : 155-162.

73. Ontario Ministry of Health. (1992). Schedule of benefits: physician services under the HealthInsurance Act. Toronto: Ontario Ministry of Health.

74. Benneyan, J. C. Mathematical models for evaluating overall process sensitivity, specificity, andcost of automated rescreening in cervical cytology. (personal communication, December 3,1996).


75. Benneyan, J. C., Kaminsky, F. C. (1996). Statistical and economic models for analysis and optimaldesign of laboratory screening policies for cervical cancer. Annals of Operations Research, 67 :235-285.

76. Decision Analysis by TreeAge (1995). Data for Windows (Version 2.6) [Computer software].Boston, MA: TreeAge Software, Inc.

77. Pauker, S. G., Kassirer, J. P. (1980). The threshold approach to clinical decision making. NewEngland Journal of Medicine, 302 (20) : 1109-1117.

78. Hutchinson, M. L. (1996). Assessing the costs and benefits of alternative rescreening strategies.Acta Cytologica, 40 (1) : 4-8.

79. Linder J. (1997). Automation of the Papanicolaou smear: A technology assessment perspective.Archives of Pathology and Laboratory Medicine, 121 : 282-6.

80. Austin, R. M., McLendon, W. W. (1996). The Papanicolaou smear: medicine’s most successfulcancer screening procedure is threatened. JAMA, 277 (9) : 754-755.

81. Bauer, H. M, Hildesheim, A., Schiffman, M. H., Glass, A. G., Rush, B. B., Scott, D. R., et al.(1993). Determinants of genital human papillomavirus infection in low-risk women in Portland,Oregon. Sexually Transmitted Diseases, 20 (5) : 274-278.

82. Sikstrom, B., Hellberg, D., Nilsson, S., Brihmer, C., Mardh, P. A. (1995). Contraceptive use andreproductive history in women with cervical human papillomavirus infection. Advances inContraception, 11 (4) : 273-284.

83. Potischman, N., Brinton, L. A. (1996). Nutrition and cervical neoplasia. Cancer Causes andControl, 7 (1) : 113-126.

84. Liaw, K. L., Hsing, A. W., Chen, C. J., Schiffman, M. H., Zhang, T. Y., Hsieh, C. Y., et al. (1995).Human papillomavirus and cervical neoplasia: a case-control study in Taiwan. InternationalJournal of Cancer, 62 (5) : 565-571.

85. Bosch, F. X., Manos, M. M., Munoz, N., Sherman, M., Jansen, A. M., Peto, J., et al. (1995).Prevalence of human papillomavirus in cervical cancer: A worldwide perspective. Internationalbiological study on cervical cancer (IBSCC) Study Group. (1995). Journal of the NationalCancer Institute, 87 (11) : 796-802.


86. Sherman, M. E., Schiffman, M. H., Lorincz, A. T., Manos, M. M., Scott, D. R., Kuman, R. J. etal. (1994). Toward objective quality assurance in cervical cytopathology. Correlation ofcytopathologic diagnoses with detection of high-risk human papillomavirus types. AmericanJournal of Clinical Pathology, 102 (2) : 182-187.

87. Cannistra, S. A., Niloff, J. M. (1996). Cancer of the uterine cervix. New England Journal ofMedicine, 334 (16) : 1030-1038.

88. Chapman, W. B., Lorincz, A. T., Willett, G. D., Wright, V. C., Kurman, R. J. (1993). Evaluationof two commercially available in situ hybridization kits for detection of human papillomavirus DNAin cervical biopsies: Comparison to Southern blot hybridization. Modern Pathology, 6 (1) : 73-79.

89. Kurz, J., Mitra, K., Adam, R., Miao, X., Mackay, J. S., Isa, N. N., et al. (1993). PCR detection andtyping of genital papillomavirus in a New Brunswick population. International Journal of Cancer,55 : 604-608.

90. Cox, J. T., Lorincz, A. T., Schiffman, M. H., Sherman, M. E., Cullen, A., Kurman, R. J. (1995).Human papillomavirus testing by hybrid capture appears to be useful in triaging women with acytologic diagnosis of atypical squamous cells of undetermined significance. American Journal ofObstetrics and Gynecology, 172 (3) : 946-954.

91. Chan, M. K., Lau, K. M., Tsui Y., Wong, F. W., Huang, D. P. (1996). Human papillomavirusinfection in Hong Kong Chinese women with normal and abnormal cervix: Detection by polymerasechain reaction method on cervical scrapes. Gynecologic Oncology, 60 (2) : 217-223.

92. Johnson, K. (1995). Periodic Health Examination, 1995 Update: 1. Screening for human papillovirusinfection in asymptomatic women. Canadian Medical Association Journal, 152 (4) : 483-493.

93. Duggan, M. A., Inoue, M., McGregor, S. E., Stuart, G. C., Morris, S., Chang-Poon, V., et al.(1994). A paired comparison of dot blot hybridization and PCR amplification for HPV testing ofcervical scrapes interpreted as CIN 1. European Journal of Gynaecological Oncology, 15 (3) :178-187.

94. Labropoulou, V., Diakomanolis, E., Dailianas, S., Kalpaktsoglou, K., Rodolakis, A., Beaudenon,S., et al. (1996). Genital papillomavirus in Greek women with high-grade cervical intraepithelialneoplasia and cervical carcinoma. Journal of Medical Virology, 48 (1) : 80-87.

95. Adams, V., Moll, C., Schmid, M., Rodrigues, C., Moos, R., Briner, J. (1996). Detection and typingof human papillomavirus in biopsy and cytological specimens by polymerase chain reaction andrestriction enzyme analysis: A method suitable for semiautomation. Journal of Medical Virology,48 (2) : 161-170.


96. Dalence, C. R. (1993). Human papillomavirus DNA testing in the management of cervical neoplasia.(Diagnostic and Therapeutic Technology Assessment). Chicago, ILL: American MedicalAssociation.

97. Ferenczy, A. (1995). Viral testing for genital human papillomavirus infections: Recent progress andclinical potentials. International Journal of Gynaecological Cancer, 5 : 321-328.

98. Baay, M., Quint, W. G., Koudstaal, J., Hollema, H., Duk, J. M., Bruger, M. P., et al. (1996).Comprehensive study of several general and type-specific primer pairs for detection of humanpapillomavirus DNA by PCR in paraffin-embedded cervical carcinomas. Journal of ClinicalMicrobiology, 34 (3) : 745-747.

99. Coutlee, F., Provencher, D., Voyer, H. (1995). Detection of human papillomavirus DNA in cervicallavage specimens by a nonisotopic consensus PCR assay. Journal of Clinical Microbiology, 33 (8): 1973-1978.

100. Siadat-Pajouh, M., Periasamy, A., Ayscue, A. H., Moscicki, A. B., Palefsky, J. M., Walton, L., etal. (1994). Detection of human papillomavirus type 16/18 DNA in cervicovaginal cells byfluorescence based in situ hybridization and automated image cytometry. Cytometry, 15 (3) :245-257.

101. Saito, J., Sumiyoshi, M., Nakatani, H., Hoshaiai, M. I., Noda, K. (1995). Dysplasia and HPVinfection initially detected by DNA analysis in cytomorphologically normal cervical smears.International Journal of Gynecology and Obstetrics, 51 (1) : 43-48.

102. Namkoong, S. E. (1995). Clinical application of HPV typing in cervical cancer. InternationalJournal of Gynecology and Obstetrics, 49 (Suppl) : S59-S67.

103. Reid, R., Greenberg, M. D., Lorincz, A., Jenson, A. B., Laverty, C. R., Husain, M., et al. (1991).Should cervical cytologic testing be augmented by cervicography or human papillomavirusdeoxyribonucleic detection? American Journal of Obstetrics and Gynecology, 164 (6 Pt. 1) :1461-1469, discussion 1469-1471.

104. Koffa, M., Simiakaki, H., Ergazaki, M., Papaefthimiou, M., Karakatsani, K., Diakomanolis, E., etal. (1995). HPV detection in stained cytological cervical specimens and correlation with cytologyand histology. Oncology Reports, 2 :1085-1088.

105. Syrjänen, K. J. (1996). Spontaneous evolution of intrepithelial lesions according to the grade andtype of the implicated human papillomavirus (HPV). European Journal of Obstetrics andGynecology and Reproductive Biology, 65 (1) : 45-53.


106. Burger, M. P., Hollema, H., Pieters, W. J., Schroder, F. P., Quint, W. G. (1996). Epidemiologicalevidence of cervical intraepithelial neoplasia without the presence of human papillomavirus. BritishJournal of Cancer, 73 (6) : 831-836.

107. Jenkins, D., Sherlaw-Johnson, C., Gallivan S. (1995). Can papilloma virus testing be used toimprove cervical cancer screening? International Journal of Cancer, 65 : 768-773.

108. Hailey, D. M., Lea R. (1995). Prospects for newer technologies in cervical cancer screeningprogrammes. Journal of Quality in Clinical Practice, 15 : 139-145.

109. Cibas, E. S. (1995). Applications of flow cytometric DNA analysis to diagnostic cytology. Diagnostic Cytopathology, 13 (2) : 166-171.

110.110. Fowlkes, B. J., Herman, C. J., Cassidy, M. (1976). Flow microfluorometric system for screeninggynecologic cytology specimens using propidium iodide-fluorescein isothiocyanate. Journal ofHistochemistry and Cytochemistry, 24 (1) : 322-331.

111. Wheeless, L. L., Patten, S. F., Berkan, T. K., Brooks, C. L., Gorman, K. M., Lesh, S. R., et al.(1984). Multidimensional slit-scan prescreening system: preliminary results of a single blind clinicalstudy. Cytometry, 5 : 1-8.

112. Braggett, D., Lea, A, Carter, R. C., Hailey, D., Ludowyk, P. (1993). Issues in cervical cancerscreening and treatment: new technologies and costs of alternative management strategies.Canberra: Australian Institute of Health and Welfare.

111. Fahs, M. C., Plichta, S. B., Mandelblatt, J. S. (1996). Cost-effective policies for cervical cancerscreening: An international review. PharmacoEconomics, 9 (3) : 211-230.

Canadian Coordinating Office for Health Technology Assessment 110-955 Green Valley Crescent Ottawa, Ontario, Canada K2C 3V4

assessment of techniques for cervical ... copies of assessment of techniques for cervical cancer...

Documents