Cervical Cancer Prevention: Cost-Effective Screening

Authors: D. Scott McMeekin, MD, UCI Medical Center; Kathryn F. McGonigle, MD, MPH, UCLA School of Medicine; Steven A. Vasilev, MD, MBA, Women's Cancer Center

Back to WCC Published Papers

Abstract
Introduction
Cost Effectiveness Defined
At What Age to Screen?
Management of the Abnormal Pap Smear
Addressing the Limitations of the Pap Smear
Advances in Specimen Collection and Interpretation
Conclusions
Women at Risk
References

Abstract

The decrease in the incidence of invasive cervical cancer has been credited to the widespread use of Papanicolaou (Pap) smear screening. Although relatively inexpensive to perform, Pap smears, if positive, often result in further diagnostic work-up (eg, colposcopy, biopsy, endocervical curettage) and associated patient anxiety. Unfortunately, false positives are frequent with Pap smears, and even screened populations of patients continue to have a significant incidence of cervical cancer. Presumably, expanding screening programs to unscreened populations or screening selected, at-risk populations more frequently could further reduce the incidence of invasive cervical cancer. Yet, few rigorous, prospective studies exist to allow for the formulation of cost-effective guidelines that optimize screening resources. To determine just how much screening is cost-effective, the medical community will have to answer several questions regarding the definition of cost-effectiveness itself, the optimal age to begin screening, whether abnormal Pap smears can be better stratified according to risk, the limitations of Pap smear screening, and whether advances in technology can help increase the positive predictive value of current screening strategies.

Introduction

Cytologic screening for cancer of the cervix can identify premalignant lesions and early subclinical disease, thus allowing the prevention or cure of this cancer. After the adoption of widespread Papanicolaou (Pap) smear screening in the US, invasive cervical cancer incidence decreased by 36% from 1973 to 1991, accompanied by a 42% reduction in the age-specific mortality rate.^[1] For patients with early-stage (Stage I) cervical cancer, 5-year survival rates approach 90%.^[2] Although these successes are credited to cytologic screening, no prospective study has demonstrated that the reduction in deaths is due to screening.

Despite widespread availability of screening, women continue to develop cervical cancer. In 1996, the American Cancer Society estimated there were 15,700 new cases, 4,900 deaths, and 65,000 cases of cervical carcinoma in situ (CIS).^[3] Since nearly 50% of cervical cancers in the US occur in women who have never been screened, and 60% of cases develop in women who have not been screened in at least 5 years,^[4] one might suppose that widespread, or even periodic, screening of all women would dramatically reduce the overall incidence of cervical cancer.

Barriers to access of Pap smear surveillance have been identified and have been the subject of numerous reviews.^[5-9] However, even when Pap smears are available, some data suggest that they do not alter the clinical management of many screened women. For example, Macgregor and colleagues^[10] reported that screening had been done in 63% of women younger than 45 who died of cervical cancer in Scotland from 1982 to 1991. Gay and others^[11] reported that 20% of women with CIS or invasive cervical cancer had a normal Pap smear within the preceding year.

In a meta-analysis, Fahey and colleagues^[12] found that Pap smear sensitivity may be closer to 20% to 30% than to the commonly held 80% to 90%. Therefore, understanding the limitations of Pap smear surveillance strategies is essential in counseling patients and in providing safe, cost-effective screening. Although Pap smears themselves are relatively inexpensive, they can lead to further diagnostic work-up (eg, colposcopy, biopsy) and patient anxiety. Thus, the financial and emotional costs of Pap smears go beyond the Pap smear itself. These costs by no means diminish the many positive aspects of Pap smear screening, but they include variables that can be difficult to factor into any cost-effectiveness equation.

To develop cost-effective screening strategies that maintain quality patient care, several important questions need to be addressed. First, how should cost-effectiveness be defined? Second, at what age should screening begin? Third, how should abnormal Pap smears be interpreted and can lesions be stratified according to risk, thus allowing the diversion of follow-up resources to patients most likely to benefit from them? Fourth, what are the limitations of Pap smear screening? Fifth, can advances in technology help increase the positive predictive value of current screening strategies, so that fewer false positives result in unnecessary diagnostic work-ups? Unfortunately, rigorous, prospective studies required to answer these questions definitively are largely nonexistent. Below, however, we explore the current data relating to these questions.

Cost Effectiveness Defined

Cost-effectiveness analysis is the evaluation of health outcomes and resource costs of health interventions. It takes the benefit as a given and looks for the least costly alternative(s). Thus, its primary role is to identify the value of alternative interventions for improving health, taking as many costs as possible into consideration. Analysis is complicated by the fact that determination of "total" or "full" costs is very elusive.

While physicians are held increasingly accountable to function within health-care budgets, cost-effectiveness analysis is rarely used to make decisions regarding health services. In particular, few studies have specifically addressed the cost-effectiveness of cervical cancer screening and management.^[13-16] Most studies have applied standard reimbursement rates to estimate cost savings for particular management models, without actually examining real applications. Costs, charges, and reimbursement are too often used interchangeably. Medicare allowable reimbursement rates for common cervical cancer screening interventions for 1997 are shown in Table I. A recent panel of experts has produced a consensus report describing how such analysis should be performed and interpreted in the health-care setting.^[17]

Costs attributable to Pap smear screening and the management of abnormal smears are those related to the contact of the patient with the health-care system (office visit type) and to the incremental costs of cytologic evaluation, interpretation of the specimen, and the treatment of affected individuals.

Direct costs of a particular screening and management strategy must be assessed in the context of opportunity costs. Opportunity costs can be defined as the amount of alternative services that must be sacrificed to screen more patients or offer more effective (with incrementally increasing direct costs) screening services. A relationship is presumed to exist between Pap smear screening program expenditures and their impact on cervical cancer mortality. As additional resources are allocated, the expected effect is an incrementally decreasing death rate. At some point, however, additional input into a screening program will produce negligible effects in terms of further reducing cervical cancer. Economists define the change in total costs that occurs with a one unit change in output as the marginal cost. Health-care planners must consider how much it costs to prevent each additional case of cervical cancer when determining how to allocate scarce resources.

Identifying subsets of patients who require less frequent screening may lead to a better utilization of resources. Screening has the potential for generating excessive spending when patients who will never develop cancer are repeatedly screened or treated. Tests supplemental to the Pap smear in primary screening that triage patients into risk groups may ultimately allow for more cost-effective care.

However, a dilemma arises in that new technologies have the actual potential to increase total cost by increasing the identification and treatment of clinically insignificant lesions. If a patient group at high risk for cervical cancer is not screened, savings accrued by withholding services may be offset by the relatively high costs of managing advanced cancers.^[14] Likewise, performing multiple expensive tests on patients within the health-care system will likely be at the cost of recruiting new patients into screening systems and providing even the most basic of tests.

How best to divide relatively fixed health-care resources is unclear. Overutilization of some services requires a proportional decrease in the availability for others. This is especially important when considering at what frequency to screen patients, at which age to initiate screening, and whether or not to supplement the Pap smear with additional tests.

Cervical intraepithelial neoplasia (CIN) meets many of the criteria required to fit into an idealized screening model. It is a common problem, with nearly 2.5 million cases of low-grade dysplasia diagnosed from 50 million Pap smears performed yearly in the US.^[18] The time of progression from CIN to CIS is estimated to be 10 to 15 years, creating a lengthy lead time during which the disease may be identified.^[19] The Pap smear is easy to perform, well tolerated, and relatively inexpensive. However, the cost-effectiveness of any screening program must be measured not in terms of its ease of implementation but in terms of clear endpoints and measures of response. The information provided by the test must translate into decreased cancer incidence, decreased death rate, or decreased overall spending. Although mathematical models suggest that annual Pap smear screening reduces the rate of invasive cancer by 93%--versus 91% at 3 years and 84% at 5 years^[20]--there may be subsets of patients who benefit from less frequent screening.

The identification of such subsets of patients at sufficiently low risk of cancer development may translate not only into cost savings but also into fewer unnecessary diagnostic work-ups and less associated anxiety. [For a discussion of high-risk factors, see the section "Women at Risk."] However, no prospective data exist that adequately clarify optimal screening intervals. (The recent study by Pearce and colleagues^[21] that seriously questioned the usefulness of routine screening was a retrospective study of a selected population of women who had previously undergone hysterectomy for benign disease.) One key question in clarifying optimal screening intervals is the age at which women should begin screening.

At What Age to Screen?

The peak age-specific rate for CIN occurs in the late 20s, for CIS at approximately age 35, and for invasive cancer at age 55 to 60 years.^[22] Although this suggests that screening may be more effective in older women than younger women, some young women clearly do develop cervical cancer and thus would benefit from early detection. However, whether it is cost effective to initiate screening at age 18 as opposed to beginning with age groups that are more likely to harbor high-grade lesions (30- to 35-year-old women) is unclear. Current recommendations err on the side of caution. The American College of Obstetricians and Gynecologists (ACOG) has recommended annual Pap smear surveillance for all women who are or who have been sexually active or who have reached the age of 18.^[23] After 3 or more normal Pap smears, less frequent screening may be offered to "low-risk" patients. ACOG also notes that certain patients may be at low risk for dysplasia and cancer and, therefore, may be screened less frequently.^[23]

Several lines of evidence support earlier rather than later screening. For instance, Eddy^[24] found that a 20-year-old woman with average risk reduced her lifetime risk of squamous cell cancer of the cervix from 2.5% to 0.7% by screening. In England in the 1960s, following a decision not to pay for Pap smears in women younger than age 35 unless they had 3 children, cervical cancer deaths doubled.^[25] In one mathematical model, the reduction in cumulative rate of cervical cancer was estimated to improve by 7% by initiating screening at age 20 rather than 35.^[20] Some also support initiation of screening at age 18 based on concerns related to a younger age peak incidence of adenocarcinoma.^[2]

Sexual norms in the US also lend support to earlier screening because a large number of young women are sexually active and therefore have a higher risk of human papillomavirus (HPV) transmission. HPV is recognized as a causative agent in the development of cervical dysplasia and carcinoma. The benefits associated with contraceptive counseling and sexually transmitted disease screening performed concurrently with Pap screening are difficult to calculate, but must measure into the analysis.

If women are screened at a younger age, more rather than less screening will be performed, thus increasing costs and work-up-related anxiety. Data by Lynge and Poll,^[26] and by Arneson and Kao,^[27] however, suggest that even if periodic screening is begun at a young age, it need not necessarily be performed yearly. In Lynge and Poll's^[26] study of Danish women, the 5-year risk of developing an invasive cervical cancer was 48% lower in women after 1 negative smear and 69% lower after 2 to 4 negative smears than in women who were not screened. If a screened woman developed cervical cancer, she was more likely to present with an earlier stage of disease.^[27] According to the investigators, "women with 1 previous negative smear have a zero risk of developing cervical cancer during the first year following the negative smear. The incidence among these women increases with length of time since the negative smear and reaches the level of unscreened women during the fifth year of follow-up. Women with 2 to 4 previous negative smears also have a negligible risk of developing cervical cancer during the first 2 years following the last negative smear. The incidence among these women increases less over time than the incidence for women with 1 previous negative smear. No cases of cervical cancer were observed among 7716 women with 5 or more previous negative smears."^[27]

Current issues regarding screening are not limited solely to young women, however. One report noted that 25% of cervical cancers and 41% of all cervical cancer deaths occurred in women older than 65 years of age.^[28] In another study, 65% of a group of cervical cancer patients 65 years of age and older had not had a Pap smear until diagnosis, but 88% had seen a physician in the preceding 3 years.^[29] Thus, opportunities do exist to screen women in this age group without necessarily increasing the cost of physician contact. Surprisingly, not all states allow uniform coverage of Pap smears as a screening tool for women eligible for Medicaid.^[30]

Continued screening results of patients older than 65 years of age may translate into decreased mortality, and as a result, some authorities recommend continued screening throughout one's lifetime.^[2] Data have suggested, however, that low-risk groups of elderly patients can be identified as candidates for either no screening or screening at 3-year intervals, based on previous Pap smear histories.^[31,32] If this contention is supported in well-designed, large trials, a greater cost savings may be realized.

Management of the Abnormal Pap Smear

Changes in nomenclature. In its early form, the Pap smear was used to categorize smears into groups based on the suspicion of malignancy. As the importance of premalignant conditions became understood, the classification was modified. In the 1960s, Richart^[33] introduced the concept of CIN and the continuum of premalignant lesions (CIN I and CIN II) that progress toward cancer (CIN III). Multiple adaptations of the Pap smear classification have been in widespread use without clear correlation between cytologic and histologic diagnoses. Largely to address these inconsistencies between cytology and histology classifications, the National Cancer Institute (NCI) in 1988 convened a workshop of experts, who developed and proposed The Bethesda System (TBS) for reporting cervical and vaginal diagnoses. TBS has since become the standard classification system used by most pathology laboratories.^[34,35] Under TBS, a standardized nomenclature of cytologic diagnoses, which should more clearly correlate with the corresponding histologic diagnoses, is used (Table II). Smears are classified based on the technical adequacy of a specimen for interpretation, and are placed into a general category. For any smear not within normal limits, a descriptive diagnosis is reported. When an epithelial cell abnormality is identified on Pap smear, the standard of practice has been to perform colposcopy, biopsy, and endocervical curettage (ECC). Squamous epithelial abnormalities have been regrouped as low- and high-grade squamous intraepithelial lesions (SILs). Low-grade SILs (LG SILs) correspond with the previous diagnosis of CIN I, and high-grade SILs (HG SILs) correspond with the previous diagnoses of CIN II and CIN III. Generally, the incidence of Pap smears indicating dysplasia should not exceed 4% to 5% of screened women.^[36]

It has been suggested that introduction of TBS terminology--in particular, the epithelial cell abnormality, atypical squamous cells of undetermined significance (ASCUS)--may be responsible for at least a doubling of the number of "abnormal" Pap smears.^[37] ASCUS represents abnormalities greater than those expected for benign or reparative changes, but not sufficient to meet criteria for a SIL. (ASCUS should not be confused with the diagnosis of atypical glandular cells of undetermined significance [AGCUS]. The latter cells are of endocervical or endometrial origin with fewer nuclear abnormalities than those meeting the criteria for adenocarcinoma. The presence of AGCUS requires a thorough evaluation; as many as 30% to 50% of patients will have a SIL, and 2% to 5% will have an endometrial or endocervical adenocarcinoma.)^[38]

ASCUS represents an area of cytologic uncertainty, where the pathologist is reporting cells with mild nuclear enlargement, hyperchromasia, and variations in size and shape, the significance of which is uncertain. The problem is compounded because, in selected populations, 20% to 30% of patients with smears indicating ASCUS ultimately will be found to have SIL on follow-up evaluation with colposcopically directed biopsies. Most of these lesions are LG SIL.

The frequency of diagnoses of ASCUS should not exceed 5% of Pap test findings, or 2 to 3 times the frequency of smears that indicate dysplasia/SIL.^[39] Medicolegal concerns related to the inherent false-negative rate of Pap smears are perceived to be largely responsible for an excessive use of this category.^[36] True increases in the incidence of HPV infection also may play a role. Subtle abnormalities are increasingly diagnosed as "possibly dysplastic," as pathologists concerned about claims of missed diagnoses err on the side of caution.^[36,40] It is not clear whether the increased recognition of these abnormalities translates into diminished cancer deaths or merely results in increased allocation of resources to diagnosis and treatment of clinically insignificant lesions.

Not all women with SILs develop cancer. The incidence of low-grade changes is greater than that of high-grade changes and far greater than that of invasive cancer. Dysplasia has a relatively long lead time for progression, as the peak age is 15 years earlier than that for patients with invasive cancer.^[41]

In a study of more than 225,000 English women screened by Pap smears, 15,500 smears were found to have an abnormality, and roughly 6000 of these women underwent colposcopy. Using data collected from an era prior to the initiation of widespread Pap smear screening, 200 cases of cervical cancer would be anticipated from a similar population of 225,000 women. The investigators suggested that widespread screening results in the excessive identification of "at-risk" patients, leading to the unnecessary work-up of more than 95% of women (5800 of 6000), many of whom would also receive treatment.

Low-grade lesions. Approximately 2.5 million women in the US are cytologically diagnosed each year with low-grade lesions.^[18] Many investigators have attempted to define better management strategies to deal with this group of patients. The 3 studies described in Table III highlight the importance of spontaneous regression of low-grade lesions. Nash and colleagues^[42] addressed the group of patients with HPV infection, defined by strict criteria, as the only abnormality on colposcopically directed biopsies. The 45 patients were prospectively followed at 3- to 6-month intervals with Pap smear, colposcopy, biopsy, and ECC. While 40% of patients had spontaneous regression of their lesions at an average time of 13.7 months, another 33% of patients had progression to CIN. Whether this reflects true biologic behavior in terms of spontaneous regression of dysplastic lesions or other factors is unclear. This study as well as others has suggested that biopsy of lesions may either remove small lesions entirely or generate a local inflammatory effect that leads to the eradication of the adjacent dysplastic lesion.^[19]

Montz and coworkers^[43] reported on a cohort of 492 patients with either ASCUS or low-grade SIL on a referral Pap smear. They performed colposcopy at presentation and biopsied the cervix on patients only if colposcopy findings were suggestive of high-grade lesions. Moderate dysplasia or worse was encountered at initial colposcopy in 19% of patients. The study subgroup of 294 patients with initial Pap smear and colposcopic diagnosis of LG SIL was followed with Pap smears and colposcopic exams, but without biopsy, at 3-month intervals for 9 months. Nearly 71% of patients had spontaneous regression to normal, and only 3% progressed toward a more significant lesion.

In Sweden, patients are not routinely treated for dysplasia unless they have high-grade changes. Nasiell and coworkers^[44] reviewed 555 patients with a first abnormal Pap smear, showing mild dysplasia (low-grade lesion formerly classified as CIN I). All patients had an initial colposcopy and were followed with Pap smears at 3- to 12-month intervals. Progression was defined as cytology consistent with CIN III (HG SIL) or invasive cancer. A 62% regression rate with a mean follow-up of 39 months was noted. Persistent abnormalities were seen in 22% (10% mild, 12% moderate dysplasia), and progression occurred in 16%. Two cases of invasive cancer occurred in patients who were lost to follow-up for 2 to 6 years before cancer had been diagnosed. This underscores the importance of reliable, routine surveillance when managing patients expectantly.

These studies are representative of data reported in the literature. Despite estimated progression rates of CIN of 15% to 33%, the observed rates of cervical cancer are far lower than those predicted by mathematical models. While data support the use of follow-up Pap smears without treatment in select patients with low-grade changes, this approach is not without its difficulties. Most series have shown that as many as 20% of patients with low-grade abnormalities on Pap smear will actually have a higher-grade lesion when examined by colposcopy and biopsied.^[43] Colposcopy, therefore, is essential in the initial management. If no lesion is noted, or a well-trained colposcopist finds that the lesion is consistent with LG SIL, then serial Pap smears at 3- to 4-month intervals may be reasonable.^[45] For worsening cytology, repeat colposcopy and biopsy are indicated.

The duration of follow-up before treatment has not been defined. In Sweden, expectant management is the rule unless progression to CIN III (HG SIL) is seen. In the US, differences in access to care, compliance problems, and medicolegal constraints may force one to give earlier treatment. Patients should be counseled as to their options and be allowed to participate in management decisions. Those electing treatment over serial examinations require continued close surveillance at 3- to 6-month intervals after therapy, typically for a total of 3 consecutive normal smears.

Few studies have addressed conservative follow-up patterns in terms of cost. While it would seem reasonable to expect that less treatment would translate into less cost, in some practice settings, serial Pap smears with or without colposcopy may be more costly than initial colposcopy, biopsy, and treatment. The small but real incidence of treatment-related morbidity must be considered in these cases. Well-designed, prospective studies are needed to define the most cost-effective management. The National Institutes of Health (NIH) has instituted a prospective trial with an accrual goal of 7200 patients to determine whether the "wait and see" approach is appropriate in patients with low-grade abnormalities.^[46] Patients with ASCUS and LG SIL are being enrolled into a 3-arm, randomized trial. Patients are either managed by serial Pap smear follow-up at 6-month intervals, by immediate colposcopy and biopsy, or by HPV DNA testing to stratify patients by risk into expectant management or immediate colposcopy groups. Results from this trial will have a major impact on the management of low-grade lesions.

High-grade lesions. The management of high-grade lesions is less controversial. While their overall incidence is low, their risk of progression to cancer is substantial. Studies detailing the natural history of high-grade lesions show that the risk of progression to invasive cancer is approximately 6% by 3 years, and as much as 71% by 12 years.^[19] A variety of low-cost procedures with acceptable morbidity for treatment exist. For selected patients, some have advocated a "see-and-treat" approach in which patients with high-grade lesions undergo colposcopy and are treated at the same visit.^[47-49] However, colposcopy has a relatively high false-positive rate in patients with low-grade cytologic abnormalities, and "see-and-treat" in this group of patients has resulted in excessive treatment of patients without dysplastic lesions.^[19,47]

Addressing the Limitations of the Pap Smear

One must be cautious when criticizing the utility of the Pap smear, however. No one test in medicine has been as apparently successful in reducing disease and death rate at a nominal cost. The Pap smear can show the presence or absence of abnormal cells consistent with the histologic diagnosis of dysplasia or cancer. It does not, by itself, distinguish which patients with dysplasia will have a course marked by spontaneous regression from those who will ultimately develop a cancer if left untreated. For cervical cancer screening to be effective, repetitive screenings at regular intervals are required. A single, isolated, apparently normal Pap smear has less meaning than repeatedly normal smears, which are associated with an extremely low incidence of dysplasia or cancer.

One factor proposed to explain why women with a history of previously normal Pap smears sometimes have a newly diagnosed invasive cancer is related to the biology of the disease process. It is possible that, in some women, aggressive lesions may rapidly progress through the preinvasive stages prior to detection. De novo appearance of high-grade lesions has also been suggested but not proven.

A false-negative rate of 15% to 20% has been demonstrated in many series of re-reviewed Pap smears.^[50,53] This estimation of error comes from studies in which Pap smears are re-evaluated, from patients with high-grade dysplasia and cervical cancer who have had prior normal smears. In a series of 339 patients with biopsy-documented CIS and invasive cancer, 20% of smears obtained within 1 year of diagnosis were falsely negative on re-review.^[54] In another series of 735 patients with a new cytologic diagnosis of HG SIL, 80 (11%) had had at least 1 normal Pap smear within the preceding 5 years.^[55] These 80 cases represented only 0.2% of all Pap smears performed during the study period.

We have always assumed, as have other investigators,^[12] that if an early cervical cancer or precancer were present, the Pap smear would eventually detect it. Unfortunately, this assumption has led to studies in which the Pap smear's sensitivity was determined either by: 1) re-evaluating negative Pap smears and calculating sensitivity values based on the view that misread slides represent the only potentially missed cases of cancer or precancer or 2) retrospectively reviewing the Pap smear history of women with confirmed diagnoses of cervical cancer. In this second example, it should be apparent that the only way these women came to the attention of a clinician in the first place was due to an abnormal Pap smear, which resulted in colposcopy, biopsy, and diagnosis. Thus, accurate sensitivity and specificity estimates for the Pap test remain incompletely defined.

For a Pap smear to be correctly analyzed, the specimen must be adequately collected, the slide must be closely screened for abnormalities, and the abnormalities must be appropriately interpreted. Deficiencies in any one of these steps may explain a negative smear in the setting of a dysplastic lesion or cervical cancer. One study reported that sampling errors by the clinician accounted for 62% of false-negative smears, while interpretive errors by the pathologist accounted for 22%, and screening errors by the cytotechnician were responsible for 16%.^[54]

Inadequate sampling of the transformation zone, poor collection and fixation of the specimen, and inclusion of excessive blood, inflammatory material, or necrotic material can obscure or preclude the correct cytopathologic diagnosis. Clinicians may improve the quality of specimens submitted by using appropriate collection devices and following established techniques.^[45] Studies have shown that in 12% to 25% of patients diagnosed with CIS or cancer who had previous "negative" Pap smear results, the smears were actually unsatisfactory for interpretation on re-review.^[50,56] The lesions themselves may lead to errors in detection as some precancerous lesions may fail to shed cells in sufficient quantities for detection by routine cytology.

Steps to reduce laboratory screening errors were mandated under the Clinical Laboratory Improvement Act of 1988 (CLIA). Ten percent of Pap smears are required to be rescreened for quality control. In a case-control study by Lynge and colleagues^[57] evaluating smear misclassification, 106 Pap smears from 53 cases of invasive cervical cancer were matched against 530 controls without a diagnosis of CIN or cancer and were re-reviewed. The investigators found that with improved cytopathology that eliminated misclassification of positive smears, the proportion of prevented cancers could increase from between 62% and 72% to between 83% and 86%. This could be achieved with a 2% increase in the workload. However, to identify all cases from all smears correctly, including those read as unsatisfactory, a 31% increase in cost would be required.

Other limitations to rescreening all Pap smears include the relatively limited number of cytotechnologists available for reading, the physical and legal limitations on the number of smears that can be read in a day, and the fact that correction of misreading addresses only a portion of the reason behind false-negative smears. While some have reported the lack of usefulness in rescreening Pap smears, it is felt to be the gold standard of quality control by which other methods are compared.

A particular problem related to screening is the incidence of adenocarcinoma. The increasing proportion of adenocarcinomas of the cervix is attributed to the relative decrease in squamous cell tumors, as well as to possible true increases in incidence. Pap smears are somewhat limited in the ability to sample the endocervical canal completely.^[58] No supplemental test offers unique advantages in terms of reducing the risk of adenocarcinoma.

Conclusions

The cost effectiveness of tests beyond the Pap smear has not been clearly demonstrated. For an individual, higher costs may appear to be worth the perceived additional security that some tests offer. Manufacturers of tests such as PAPNET are counting on this and are directly marketing services to the consumer. Medicolegal concerns no doubt raise the costs of medicine by favoring overdiagnosis and overtreatment. Legal reforms would remedy some of the problems and possibly reduce costs. For health-care planning of large segments of the population, however, cost effectiveness must be considered. The natural history of dysplasia and invasive cancer make the likelihood of eventually detecting a dysplastic lesion in a patient who is regularly screened high, thus reducing the clinical impact of most false-negative smears. Nothing supports this contention more than the marked reduction in cervical cancer incidence since institution of widespread screening. Nevertheless, limitations of Pap smear screening must be recognized, and patients need to be adequately counseled. Each adjuvant test adds cost, and the cost effectiveness of new technology needs to be addressed in well-designed trials. Finally, there is the question of whether cervical cancer incidence can be decreased more by improving the tests for those already screened or by improving access to the unscreened population.

In the absence of prospective studies supporting the routine use of adjunctive tests, annual Pap smear screening, by itself, is advocated for most women. The limitations of Pap smears make the development of adjunctive tests attractive not only by their potential to reduce the false- negative rate of Pap smears, but more importantly to distinguish a low-risk group of women who require diminished surveillance from those at sufficient risk who require more costly evaluation and follow-up.

About the Authors

Dr. Vasilev is Associate Gynecologic Oncologist, Women's Cancer Center, Pasadena, Calif. Dr. McGonigle is Assistant Professor, Division of Gynecologic Oncology, UCLA School of Medicine, Los Angeles, Calif. Dr. McMeekin is Clinical Instructor, Division of Gynecologic Oncology, UCI Medical Center, Orange, Calif.

References

1. Ries L, Miller B, Hankey B, et al (eds): SEER Cancer Statistic Review 1973-1991, National Cancer Institute, NIH publication no. 94-2789, 1994.
2. DiSaia P, Creasman W: Invasive cervical cancer, in Clinical Gynecologic Oncology. St. Louis, Mo., Mosby-Year Book, 1993, pp 58-125.
3. Parker S, Tong T, Bolden S, et al: Cancer statistics 1996. CA Cancer J Clin 46(1):5-27, 1996.
4. Singleton H, Orr J: Screening in Cancer of the Cervix. Philadelphia, Pa., Lippincott-Raven, 1995, pp 17-35.
5. Reelick N, de Haes W, Schuurman J: Psychological side effects of the mass screening of cervical cancer. Soc Sci Med 18:1089-1083, 1984.
6. National Cancer Institute: Healthy People 2000. Bethesda, Md., US Dept. Health and Human Services, 1990. DHS Pub. No. (PHS) 91-50212.
7. Black B, Scwitzer R, Dezelsky T: Report on the American Cancer Society workshop on community cancer detection, education, and prevention demonstration projects for underserved populations. CA Cancer J Clin 43:226-233, 1993.
8. The National Cancer Institute Cancer Screening Consortium for Underserved Women: Breast and cervical cancer screening among underserved women: Baseline survey of results from six states. Arch Fam Med 4:617, 1995.
9. Orbell S, Crombie I, Robertson A, et al: Assessing the effectiveness of a screening campaign: Who is missed by 80% cervical screening coverage? J R Soc Med 88:389-394, 1995.
10. Macgregor J, Campbell M, Mann E, et al: Screening for CIN in northeast Scotland shows fall in incidence and mortality from invasive cancer with concomitant rise in preinvasive disease. BMJ 308:1407-1411, 1994.
11. Gay J, Donaldson L, Goellner J: False-negative results in cervical cytologic studies. Acta Cytol 29:1043-1046, 1985.
12. Fahey MT, Irwig L, Macaskill P: Meta-analysis of Pap test accuracy. Am J Epidemiol 141(7):680-689, 1995.
13. Massad L, Lonky N, Mutch D, et al: Use of speculoscopy in the evaluation of women with atypical Papanicolaou smears: Improved cost-effectiveness by selective colposcopy. J Reprod Med 38(3):163-169, 1993.
14. van Ballegooijeen M, Koopmanschap M, Tjokrowardojo AJ, et al: Care and costs for advanced cervical cancer. Eur J Cancer 28A(10):1703-1708, 1992.
15. Miller A: The cost effectiveness of cervical cancer screening. Ann Intern Med 117(6):529-530, 1992.
16. Roland P, Naumann R, Alvarez R, et al: A decision analysis of practice patterns used in evaluating and treating abnormal Pap smears. Gynecol Oncol 59:75-80, 1995.
17. Siegel J, Weinstein M, Russell L, et al: Recommendations for reporting cost-effectiveness analyses. JAMA 276:1339-1341, 1996.
18. National Cancer Institute: National Health Interview Survey Supplement on Cancer Control. Bethesda, National Center for Health Statistics, 1989.
19. Follen-Mitchell M, Schottenfeld D: The natural history of cervical intraepithelial neoplasia and the management of the abnormal papanicolaou smear in cervical cancer and preinvasive neoplasia. Philadelphia, Pa., Lippincott-Raven, 1996, pp 103-113.
20. Miller A: Screening for cervical cancer, in Rubin SC, Hoskins, WJ (eds): Cervical Cancer and Preinvasive Neoplasia. Philadelphia, Pa., Lippincott-Raven, 1996, pp 13-25.
21. Pearce K, Heafner H, Sawar S, et al: Cytopathological findings on vaginal Papanicolaou smears after hysterectomy for benign disease. N Engl J Med 335(21):1563-1567, 1996.
22. Bibbo M, Keebler C, Wied G: Prevalence and incidence rates of cervical atypia: A computerized file analysis on 148,735 patients. J Reprod Med 6:79-83, 1971.
23. Committee on Gynecologic Practice: Recommendations on the Frequency of Pap Test Screening. ACOG Committee Opinion Number 152, March 1995.
24. Eddy D: Screening for cervical cancer. Ann Intern Med 113: 214-226, 1990.
25. Knox E: Ages and frequencies for cervical cancer screening. Br J Cancer 34:444, 1976.
26. Lynge E, Poll P: Incidence of cervical cancer following negative smear: A cohort study from Marido County, Denmark. Am J Epidemiol 124:345-352, 1986.
27. Arneson A, Kao M: Long-term observations of cervical cancer. Am J Obstet Gynecol 156:614-625, 1987.
28. Mandelblatt J, Gopaul F, Wistreich M: Gynecologic care of the elderly. JAMA 256:367, 1986.
29. Norman S, Talbott E, Kuller L: The relationship of Papanicolaou testing and contacts with the medical care system to stage at diagnosis of cervical cancer. Arch Intern Med 151:58-64, 1991.
30. Boss L, Guckes F: Medicaid coverage of screening tests for breast and cervical cancers. Am J Pub Health 82(2):252-253, 1992.
31. Miller A, Anderson G, Brisson J: Report of a national workshop on screening for cancer of the cervix. Can Med Assoc J 145:1301-1325, 1991.
32. Baker B: Regular Pap testing held unnecessary past age 64. OB-GYN News: Feb 15, 1997, p 18.
33. Richart R: Cervical intraepithelial neoplasia. Pathology Annual 8:301-328, 1973.
34. National Cancer Institute Workshop: The 1988 Bethesda System for reporting cervical/vaginal cytological diagnoses. JAMA 262:931-934, 1989.
35. Broder S: National Institutes of Health: Rapid Communication: The Bethesda System for reporting cervical/vaginal cytological diagnoses. Report of the 1991 workshop. JAMA 267(14):1892, 1992.
36. Richart R, Wright T: Controversies in the management of low-grade cervical intraepithelial neoplasia. Cancer 71:1413-1421, 1993.
37. Julian T: The minimally abnormal Pap smear and cervical cancer screening. Colposcopist XXV(3), 1993.
38. Nasu I, Meurer W, Fu Y: Endocervical glandular atypia and adenocarcinoma: A correlation of cytology and histology. Int J Gynecol Pathol 12:208-218, 1993.
39. Kurman R, Henson D, Herbst A, et al: National Cancer Institute workshop: Interim guidelines for management of abnormal cervical cytology. JAMA 271(23):1866-1869, 1994.
40. Raffle A, Alden B, Mackenzie E: Detection rates for abnormal cervical smears: What are we screening for? Lancet 345:1469-1473, 1995.
41. Cramer D, Cutler S: Incidence and histopathology of malignancies of the female genital tract in the US. Am J Obstet Gynecol 118:443-460, 1974.
42. Nash J, Burke T, Hoskins W: Biologic course of cervical human papillomavirus infection. Obstet Gynecol 69:160-162, 1987.
43. Montz F, Monk B, Fowler J, et al: Natural history of the minimally abnormal Papanicolaou smear. Obstet Gynecol 80:385-388, 1992.
44. Nasiell K, Roger V, Nasiell M: Behavior of mild cervical dysplasia during long-term follow-up. Obstet Gynecol 67: 665-669, 1986.
45. ACOG Technical Bulletin: Cervical Cytology: Evaluation and Management of Abnormalities. No. 183, Aug. 1993.
46. National Cancer Institute: Getting a handle on ASCUS: A new clinical trial. JNCI 87(11):787, 1995.
47. Higgins R, Hall J, McGee J, et al: Appraisal of the modalities to evaluate an initial abnormal Pap smear. Obstet Gynecol 84:174-178, 1994.
48. Burger R, Monk B, Van Nostrand K, et al: Single-visit program for cervical cancer prevention in a high-risk population. Obstet Gynecol 86:491-498, 1995.
49. Megevand E, Van Wyk W, Knight B, et al: Can cervical cancer be prevented by a see, screen, and treat program? A pilot study. Am J Obstet Gynecol 174:923-928, 1996.
50. Koss L: Cervical (Pap) smear: New directions. Cancer 71:1406-1412, 1993.
51. Bogdanich W: The Wall Street Journal, Nov. 2, 1987, p 1.
52. Bogdanich W: The Wall Street Journal, Dec. 29, 1987, p 17.
53. Koss L: The Papanicolaou test for cervical cancer detection: A triumph and a tragedy. JAMA 261:737-743, 1989.
54. Gay J, Donaldson L, Goellner J: False-negative results in cervical cytologic studies. Acta Cytol 29:1043-1046, 1985.
55. Allen K, Zaleski S, Cohen M: Review of negative Papanicolaou tests: Is the retrospective 5-year review necessary? Am J Clin Pathol 101:19-21, 1994.
56. van der Graaf Y, Voois G, Gaillard H: False-negative results in cervical cytology screening. Acta Cytol 31:434-438, 1987.
57. Lynge E, Arffmann E, Poll P, et al: Smear misclassification in a cervical cancer screening programme. Br J Cancer 68:368-373, 1993.
58. Kristensen G, Skyggebjerg K, Holund B, et al: Analysis of cervical smears obtained within three years of the diagnosis of invasive cervical cancer. Acta Cytol 35:47-50, 1991.
59. Tazuka F, Oikawa H, Shuki H, et al: Diagnostic efficacy and validity of the ThinPrep method in cervical cytology. Acta Cytol 40:513-518, 1996.
60. Neuromedical Systems Inc: Industrial developments in automated cytology as submitted by their developers. Acta Cytol 15:358-369, 1993.
61. Sherman M, Mango L, Kelly D, et al: Papnet analysis of reportedly negative smears preceding the diagnosis of a high-grade SIL or carcinoma. Mod Pathol 7:578-581, 1994.
62. Keyhani-Rofagha S, Palma T, O'Toole R: Automated screening for quality control using Papnet: A study of 638 negative Pap smears. Diagn Cytolpathol 14:316-320, 1996.
63. Ryan M, Stastny J, Remmers R, et al: Papnet directed rescreening of cervicovaginal smears: A study of 101 cases of ASCUS. Am J Clin Pathol 105:711-718, 1996.
64. Birdsong G: Automated rescreening of Pap smears: What are the implications? Diagn Cytopathol 13:283-286, 1995.
65. ACOG Technical Bulletin: Genital Human Papillomavirus Infections. No. 193, June 1994.
66. Johnson K: Screening for HPV infection in asymptomatic women. Can Med Assoc J 152:483-493, 1995.
67. Szarewski J, Ho G, Maddox P, et al: Human papillomavirus testing in primary cervical screening. Lancet 345:1533-1536,1995.
68. Koutsky L, Holmes K, Critchlow C, et al: A cohort study of the risk of CIN grade 2 or 3 in relation to papillomavirus infection. N Engl J Med 327:1272-1278, 1992.
69. Lonky N, Edwards G: Comparison of chemiluminescent light versus incandescent light in the visualization of acetowhite epithelium. Am J Gynecol Health VI:11-15, 1992.
70. Mann W, Lonky N, Massad S, et al: Papanicolaou smear screening augmented by magnified chemiluminescent exam. Int J Gynecol Obstet 43:289-296, 1993.
71. Lonky N, Mann W, Massad S, et al: Ability of visual tests to predict underlying cervical neoplasia: Colposcopy and speculoscopy. J Reprod Med 40:530-536, 1995.
72. Spitzer M, Krumholz B, Chernys A, et al: Comparative utility of repeat Papanicolaou smears, cervicography, and colposcopy in the evaluation of atypical Papanicolaou smears. Obstet Gynecol 69:731-735, 1987.
73. Schneider A, Zahm D, Kirchmayr R, et al: Screening for cervical intraepithelial neoplasia grade 2/3: Validity of cytologic study, cervicography, and human papillomavirus detection. Am J Obstet Gynecol 174:1534-1541, 1996.
74. Liao S, Brewer C, Zavada J, et al: Identification of the MN antigen as a diagnostic biomarker of cervical intraepithelial squamous and glandular neoplasia and cervical carcinoma. Am J Pathol 145:598-609, 1994.
75. Liao S, Stanbridge E: Expression of the MN antigen is an early predictor (biomarker) of cervical dysplasia. Cancer Epidemiol Biomarkers Prev 5:549-557, 1996.

Women at Risk

A clear understanding of factors that place women at high risk for cervical cancer will help health-care professionals and their patients make more informed choices. According to the American College of Obstetrics and Gynecology, high-risk factors are:

• Multiple sexual partners
• Male partners with multiple sexual partners
• Early age of first intercourse
• Male sex partner who has had a partner with cervical cancer
• Current or prior human papillomavirus infection or condyloma
• History of herpes simplex infection
• HIV infection
• History of sexually transmitted diseases
• Immunosuppression (eg, renal transplant)
• Smokers, or abusers of other substances including alcohol
• History of lower genital tract dysplasia or cancer
• Lower socioeconomic status

Tables

Table I - 1997 Medicare Allowable Rates for Select Gynecologic Procedures

Table II - The Bethesda System for Pap Smear Reporting (TBS)

• Adequacy of the specimen
  • Satisfactory for evaluation
  • Satisfactory but limited (by reason specified)
  • Unsatisfactory for evaluation
• General categorization
  • Within normal limits
  • Benign cellular changes
  • Epithelial cell abnormality
• Descriptive diagnoses
  • Benign cellular changes
    • Infection (specify infectious agent)
    • Other
  • Reactive changes
    • Inflammation
    • Atrophy
    • Radiation
    • Intrauterine contraceptive device (IUD)
    • Other
  • Epithelial cell abnormalities
  • Squamous cell
    • Atypical squamous cells of undetermined significance (ASCUS)
    • Low-grade squamous intraepithelial lesion (LG SIL)
    • High-grade squamous intraepithelial lesion (HG SIL)
    • Squamous cell carcinoma
  • Glandular cell
    • Endometrial cells
    • Atypical glandular cells of undetermined significance (AGCUS)
  • Adenocarcinoma
    • Endocervical adenocarcinoma
    • Endometrial adenocarcinoma
    • Extrauterine adenocarcinoma
  • Other malignant neoplasm
  • Hormonal evaluation expressed as a P/I/S ratio of parabasal/intermediate/superficial cells

Table III - Spontaneous Regression and Progression of Low-Grade Lesions in 3 Studies