Discussing how physicians and patients can cure their misunderstandings of medical statistics
Oct. 26, FRANKFURT —Several dozen physicians, scientists and amazingly enough, even journalists, have gathered at a red building on a research campus here with a health-care revolution on their minds. No, the terms "single payer" and "public option" have not been mentioned. Rather, attendees from Europe, Canada, Australia and the United States brainstormed all day about how to improve statistical literacy among health-care researchers, providers and consumers.
This meeting, hosted by Germany’s Ernst Strüngmann Foundation, is the latest effort to encourage out-of-the-box thinking about critical scientific issues. The brainchild of business-tycoon twin brothers who took over a pharmaceutical company founded by their father, Ernst Strüngmann Forums ask participants to reformulate "seemingly dead-end problems that require eclectic perusal," says program director Julia Lupp.
I’m here because I was invited to write a background paper for the meeting on the extent and causes of statistical illiteracy among journalists. Background papers are not read at the meeting. Participants read them beforehand and use them as launching pads for intense discussions held in groups of roughly a dozen people.
By the end of the week, four groups will issue reports:
- One report will address how researchers can generate evidence that the clinicians and patients can understand and use.
- Another report will present recommendations on how to improve communication about risks and benefits between physicians and patients, and between the media and the public.
- A third effort will examine how health-care decisions by physicians and patients can go wrong and how to describe accurately the often uncertain implications of medical tests.
- A final report will offer a prediction of how health-care professionals and patients will collaborate in 2020, from the examination room to the Internet.
Statistical illiteracy generates serious health-care problems that get little attention, says conference co-chairperson Gerd Gigerenzer, director of the Center for Adaptive Behavior and Cognition in Berlin. Not only patients, but physicians too, routinely overestimate the benefits of various medical tests and vastly underestimate their risks. Earlier this year, Gigerenzer and his colleagues reported that 92 percent of women interviewed in nine European countries overestimated — by a lot — the extent to which mammograms reduce one’s chances of dying from breast cancer.
For every 1,000 women who receive a mammogram, roughly one woman fewer will die of breast cancer over the next 6 to 20 years, compared with those who don’t receive mammograms. That's good news for the one woman who survives thanks to screening. But many participants in Gigerenzer’s study thought that mammograms reduced subsequent mortality by 100 or even 200 women per 1,000 screened.
Physicians and journalists come out looking comparably bad when asked to evaluate data on the risks and benefits of various medical tests. A big challenge for the future is to present evidence in understandable ways, such as using absolute risks (1 in 1,000 women who survived due to screening) rather than relative risks (survival increases by 20 percent after screening).
The problem is much bigger than that, though. Physicians, health insurers and journalists have their own vested interests in inflating the benefits and downplaying the risks of medical tests. And patients often don’t appreciate learning that their physicians offer medical tests and treatments with highly uncertain effects.
Maybe by the end of the week, a few promising leads on inoculating people against statistical illiteracy will emerge from a few conference rooms in Frankfurt.