Survey Suggests Physicians May Have Difficulty Interpreting Cancer Statistics

A recent survey (abstract) of more than 400 primary care physicians suggests that many physicians fail to grasp the finer points of interpreting cancer screening statistics, a fact that an accompanying editorial contends could be tied to society's overenthusiasm for cancer screening. The results of the survey were published in the March 6 issue of Annals of Internal Medicine.

The survey introduced two hypothetical screening tests that were created using data on prostate cancer. To avoid bias, however, researchers described the two tests merely in terms of being noninvasive, free and capable of detecting cases of cancer for which treatment is available. In one scenario, the effects of screening were given in terms of five-year survival rate and increased early detection. In the second scenario, the effects were given in terms of decreased mortality and incidence.

The researchers reported that 76 percent of primary care physicians surveyed indicated that survival data prove that screening saves lives. About one-half of the physicians also said that finding more cancer cases in screened versus unscreened populations proved that screening saves lives.

According to the National Cancer Institute, however, because survival and early detection rates are susceptible to lead-time and overdiagnosis biases, while mortality rates are not, reduced mortality is the only statistic that reliably proves the benefit of cancer screening.

The results come as no surprise to Doug Campos-Outcalt, M.D., chairman of the Department of Family, Community and Preventive Medicine at the University of Arizona College of Medicine, Phoenix, and the AAFP liaison to the U.S. Preventive Services Task Force. It's a problem, he said, that most likely extends to other types of physicians and throughout society.

"It's not just doctors," Campos-Outcalt told AAFP News Now. "It's newspaper reporters. It's organizations. I've argued with congressmen over this sort of thing."

The researchers said that in the context of screening, the term "survival" takes on a different meaning from that in common language because the calculation of survival has different starting points for screened and unscreened people. Moreover, because population mortality relative to survival is low in this setting, even an effective screening test can result in only a small absolute reduction in deaths from cancer (typically 1 percent or lower).

"This is an inability to really grasp numbers and analyze the true numbers," Campos-Outcalt said. "You have a lot of physicians practicing who were never trained in this." All the more reason, he added, for physicians to "look to organizations that know how to analyze the evidence and make solid, evidence-based recommendations, such as the U.S. Preventive Services Task Force."

For his part, Campos-Outcalt said he takes pains to emphasize these concepts in the classroom, and he thinks that most medical schools now teach students these concepts and point out these kinds of errors.

"I think it's going to be less of a problem as time goes on and this becomes more ingrained in medical school curricula," he said.

When it comes to patients' views on screening for cancer, however, the idea that it's not always the best course may be a hard sell.

In an editorial accompanying the research (excerpt), Virginia Moyer, M.D., M.P.H., of the Baylor College of Medicine in Houston, said that, in general, patients believe cancer is much more common than it is and assume that early detection is synonymous with cure. This is reflected in a 2004 study about enthusiasm for cancer screening in which 87 percent of adults surveyed said they thought routine cancer screening is almost always a good idea and 74 percent believed that finding cancer early saves lives.

"In the face of all this enthusiasm, many people will not appreciate that for some screening tests, there is evidence of no benefit and even potential harm to persons who receive the service," she said.