Pap tests don’t work as well as many believe

But analysis supports 3-year screening intervals

Pap smears only detect about half of existing cervical abnormalities, a lower accuracy rate than is generally cited in cervical cancer screening recommendations, according to an analysis by the Duke University Evidence-based Practice Center in Durham, NC, a project of the Agency for Health Care Policy and Research.

However, since the overall prevalence of cervical cancer is low, screening every three years still is an effective goal, particularly if it is enhanced by new Pap test technologies, says Douglas McCrory, MD, MHS, assistant professor of medicine at Duke’s Center for Clinical Health Policy Research.

"For any one time a woman goes in and gets a Pap smear, for women who have an abnormality present, there’s only a 50% chance that that abnormality will be picked up," says McCrory, who is also a research associate with the Department of Veterans Affairs Medical Center in Durham. "That’s part of the rationale for doing regular screening. Any one, single Pap smear is not very accurate."

If the Pap screening occurs every three years instead of every one to two years, new technologies can cost-effectively increase the accuracy of the test, he says. Currently, the U.S. Preventive Health Task Force recommends screening at least once every three years after a baseline and two annual tests. With 13,700 new cases in 1998, cervical cancer is one of the most common cancers among women. Yet there has been a 70% decrease in cervical cancer mortality with the introduction of Pap testing some 50 years ago.

New studies shed light on accuracy

The models of Pap screening have been built around studies that focused on cervical cancer screening in a high-prevalence population, says McCrory. But recent studies focus on the accuracy of the test in a low-prevalence environment, which mirrors the general recommended screening population of women ages 21 to 64.

"There had been some very good new research that was included in our summaries that wasn’t included in previous analysis," says McCrory.

Still, some experts believe that there still isn’t sufficient evidence to show a false-negative rate that’s greater than the previous accepted rate of 20% to 30%. Even a false-negative rate as high as 50% isn’t cause for alarm, cautions Kenneth Noller, MD, professor and chair of the department of obstetrics and gynecology at the University of Massachusetts Medical Center in Worcester. "We’ve got the single best cancer screening test ever invented," he says. "There are just not that many cases that are missed."

The Duke team, working with Health Economics Research of Waltham, MA, also analyzed studies relating to three new technologies:

  • AutoPap and Papnet involve a computerized rescreening of slides to detect ones with suspicious cells that were initially read as normal.

  • ThinPrep is a thin-layer cytology technology that uses a different method of collecting cells and analyzing them.

    In May 1998, the American College of Obstetricians and Gynecologists issued an opinion from the Committee on Gynecologic Practice stating that "appropriate use of these new techniques requires further investigation." No research was available on whether the techniques lower the incidence rate of invasive cervical cancer or improve the survival rate, the committee said.

    McCrory’s review examined the available literature and analyzed the possible impact of greater detection of abnormalities. "[The new methods] seem to improve the detection rate of the cytologic screening by a very similar margin [of roughly 60%]," he says. "They do it in different ways, but the overall effect was modeled in a very similar way based on the studies that had been conducted.

    "To really understand the implications for cervical cancer screening, we built a cost-effectiveness model that takes into account our new estimates of accuracy of the Pap smear with our understanding of the natural history of the disease and annual screening," he says.

    In annual screening, using the new technologies could detect an additional four cancers per 100,000 women over a lifetime of screening. They would also detect many more mild abnormalities, and it’s not clear how the tests affect false positives, or screening that incorrectly identifies cells as abnormal.

    At three-year screening intervals, 18 cancers would be detected per 100,000 women over a lifetime of screening. "With longer screening intervals, that increase in accuracy of new technologies becomes important," says McCrory.

    The Duke analysis doesn’t take into account some issues, such as whether the average lab differs in quality from the labs used in the clinical trials. The cost analysis also doesn’t include potential legal liability from missed diagnoses.

    Regular screenings are most important

    Amid this analysis of the Pap test accuracy, McCrory stresses one important fact: Most cervical cancers occur among women who were not screened at all or who did not have regular screenings of any interval.

    Improvements in the overall screening of women can save lives, he says. "Getting a screening in the first place, regardless of which technology you use, is the most important issue. It’s a major public health problem, as opposed to the problem we addressed [in the analysis], which is a clinical issue."

    In fact, the finding that the Pap test is less accurate than believed simply underscores the need for regular screening. "Women need to get the Pap smear repeatedly because a Pap smear isn’t perfect," he says.

    [Editor’s note: A summary of "Evaluation of Cervical Cytology" (AHCPR 99-E009) is available on the Web site of the Agency for Health Care Policy and Research at http://www.ahcpr.gov. Free print copies are available from the AHCPR Publications Clearinghouse, P.O. Box 8547, Silver Spring, MD 20907. Telephone: (800) 358-9295. InstantFax: (301) 594-2800.]