Technologic advances for evaluation of cervical cytology: is newer better?

Among those women who have cervical cancer and have been screened, 14% to 33% of the cases represent failure to detect abnormalities that existed at the time of screening. New technologies intended to improve detection of cytologic abnormalities include liquid-based, thin-layer cytology (ThinPrep, AutoCyte), computerized rescreening (PAPNET), and algorithm-based computer rescreening (AutoPap). This report combines evidence reviews conducted for the U.S. Preventive Services Task Force and the Agency for Healthcare Research and Quality, in which we systematically identified articles on cervical neoplasia, cervical dysplasia, and screening published between January 1966 and March 2001. We note the challenges for improving screening methods, providing an overview of methods for collecting and evaluating cytologic samples, and examining the evidence about the diagnostic performance of new technologies for detecting cervical lesions. Using standard criteria for evaluation of the diagnostic tests, we determined that knowledge about the sensitivity, specificity, and predictive values of new technologies is meager. Only one study of liquid-based cytology used a reference standard of colposcopy, with histology as indicated, to assess participants with normal screening results. Lack of an adequate reference standard is the overwhelming reason that test characteristics cannot be properly assessed or compared. Most publications compare results of screening using the new technology with expert panel review of the cytologic specimen. In that case, the tests are not independent measures and do nothing to relate the screening test findings to the true status of the cervix, making determination of false-negatives, and thus sensitivity, specificity, and negative predictive value, impossible. We did not identify any literature about health outcomes or cost effectiveness of using these tools in a system of screening. For the purposes of guiding decision making about choice of screening tools, the current evidence is inadequate to gauge whether new technologies are "better" than conventional cytology..