Performance and Cost-Effectiveness of Screening Tests

Few studies of primary Pap screening were conducted in low-prevalence populations and unaffected by "workup" bias, but these few studies provided estimates of the specificity of Pap smear screening of 98 percent and sensitivity 51 percent. The sensitivity estimate is much lower than that generally believed to be true. Future decision models, cost-effectiveness studies, and health policy decisions should consider this lower estimate.

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

Thin-layer cytology and computerized rescreening devices primarily aim to reduce different sources of error: sampling error and detection error, respectively. Combined use of thin-layer cytology and computerized rescreening has the potential to exceed the effectiveness of either technology alone. Although with current pricing, this strategy would not be economically feasible, further research could lead to further improvements in sensitivity.

Comparisons with cytological reference standards attest to the validity of the new technologies compared with optimal Pap screening. Comparison with a histological reference standard provides a more relevant outcome for clinical decisionmakers, since histological diagnosis forms the basis of most clinical management decisions. Further research is needed on validating negative cytological diagnoses made with the new technologies with colposcopy in both low-prevalence and high-prevalence populations. This could be accomplished by subjecting a random sample of cytology-negative women to colposcopy, which would permit statistical correction for "workup" bias and estimation of test specificity. Precise estimates of the relative performance of new versus conventional technologies require verification of all patients in whom at least one test is positive. Methods such as discrepant analysis, in which subjects with concurrent test results (negative or positive) are not verified, should be avoided in future studies.

Differences in the performance of Pap testing have been observed between different types of laboratories. Attempts at quality control aimed at assessing and reducing between-laboratory variation in the interpretation of cervical cytology have had mixed success. It has been suggested that the 10 percent manual rescreening mandated by federal law is costly and ineffective in improving sensitivity; however, it may have an important role to play in assessing and maintaining quality in cervical cytology interpretation. The computerized rescreening strategy reviewed in this report functions as a quality control mechanism and will likely reduce between-laboratory variation in interpretation. Further research in this area may be warranted.

Patient and Provider Screening Behaviors

Many of the existing models of cervical cancer screening have provided insights into the effect of various screening intervals and strategies on cancer incidence and mortality and on the cost-effectiveness of screening. Despite consistent demonstration of extremely high marginal cost-effectiveness ratios for annual screening compared with less frequent biennial or triennial screening, many agencies recommend annual screening. This recommendation may be based on other considerations, including improving patient and provider compliance, providing marginal economies of scale, or even attempting to improve compliance with other health screening practices, such as breast or colorectal cancer screening. However, it is possible that a recommendation for annual screening might actually prevent those women who are apprehensive about pelvic examinations or Pap smears from receiving any care. Although this topic was beyond the scope of our analysis, increasing the number of women who have regular cervical cytological screening would appear to have a greater impact on mortality than improving the sensitivity of the screening test. Further research is needed in ways to improve patient and provider compliance with performing screening at appropriate intervals and in assuring followup of abnormal results, especially when screening is performed at less frequent intervals.

Quality of Life

Deaths from cervical cancer have been substantially reduced since the introduction of the Pap test. By limiting one's assessment of the impact of new technologies for cervical cancer screening to reduction in cervical cancer mortality, one fails to consider a number of other health outcomes that may be quite important. Pap screening has resulted in three trends: an increased proportion of preinvasive lesions, an increased proportion of earlier stage invasive cancer and, as suggested by our model, an increasing proportion of cases among younger women. Although these trends improve the likelihood of curative treatment, they also lead to an increase in treatment and to a longer life living with the potential sequelae of treatment. The shift toward diagnosis of more premalignant lesions, and the inconvenience, potential discomfort, and psychological distress associated with screening and treatment of precursors (many of which will never progress to become cancer), might well outweigh the negative impact of cancer itself on quality of life at the population level.

Further research on quality of life issues associated with cervical cancer screening and treatment is needed to quantify screening behaviors, quality of life associated with various treatments for cervical cancer, and quality of life associated with low-grade lesions.

Role of HPV Testing

Our current understanding of the biology and natural history of cervical cancer places HPV as the main causative agent. HPV testing may have an important role in cervical cancer screening or in triage of women with abnormal cytology. Studies are currently under way to evaluate the utility of HPV testing in predicting progression of atypia and low-grade lesions. The results of these and future studies may substantially change the clinical practice of cervical cancer screening. Our model has been constructed to allow assessment of the potential role of HPV testing or vaccination on various strategies for cervical cancer prevention.

Proposed New Primary Screening Devices

A new computerized device not reviewed in this report, the AutoPap^® Primary Screening System, has received FDA approval to be used for primary screening rather than rescreening of Pap smears. Because of lack of data, we did not specifically assess the cost-effectiveness of using this type of computerized primary screening. Our model suggests that improving the sensitivity of the primary screening step is more cost-effective at a given screening interval than improving rescreening sensitivity, so it is possible that this strategy might compare favorably with the strategies we evaluated. However, more reliable estimates of the incremental cost, sensitivity, and specificity of the AutoPap^® Primary Screening System are needed before meaningful comparisons can be made. Results of cervical cytological screening with this new device should also be compared with colposcopy or a histology reference standard, with verification of a random sample of test negative women.