(ChemotherapyAdvisor) – An online survey of U.S. primary-care physicians has found that better education about how to interpret studies about cancer-screening tests is needed to ensure patients are not harmed, according to an article and an editorial published in the March 6 issue of Annals of Internal Medicine.
“Most primary-care physicians mistakenly interpreted improved survival and increased detection with screening as evidence that screening saves lives,” the investigators wrote. “Few correctly recognized that only reduced mortality in a randomized trial constitutes evidence of the benefit of screening.”
The parallel-group, randomized study surveyed a national sample of primary-care physicians comprising a research panel maintained by Harris Interactive. The cooperation rate was 79%. Physicians practiced both inpatient and outpatient medicine in 2010 (n=297) and exclusively outpatient medicine in 2011 (n=115). Scenarios about the effect of two hypothetical screening tests were presented; in one, the effect was described as improved five-year survival and increased early detection and the other as decreased cancer mortality and increased incidence.
Investigators found primary-care physicians to be more enthusiastic about the screening test supported by irrelevant evidence (five-year survival increased from 68% to 99%): 69% recommended the test vs. 23% for the test supported by relevant evidence (cancer mortality reduced from 2 to 1.6 in 1,000 persons)(P<0.001).
Many physicians did not distinguish between irrelevant and relevant screening evidence; 76% vs. 81%, respectively, stated that each of these statistics proves screening saves lives, and 47% incorrectly said finding more cases of cancer in screened as opposed to unscreened populations “proves that screening saves lives.”
The authors noted that one limitation of the study is that physicians’ recommendations for screening were based on hypothetical scenarios, not actual practice and, to avoid bias and/or the potential for misinformation, no specific cancer type was mentioned.