Evidence-based medicine is critical to improving the effectiveness (and cost-effectiveness) of our medical system. In order to do expand its reach, we need doctors who understand medical evidence. Unfortunately, as recent debates about PSA testing and mammograms for women under 50 have shown, many doctors have (at best) a fuzzy grasp of statistical reasoning. Without a good understanding of statistics, physicians are as vulnerable as any other "man on the street" to serious statistical mistakes--mistakes that affect their clinical judgment.
Some of that can be reformed in medical schools: the medical curriculum reform movement, which often focuses on improving teamwork and coordination of care, has also incorporated calls to teach med students about medical evidence and how to properly interpret it when making clinical decisions.
What can we do, though, about the doctors already out there practicing medicine? PharmedOut, a project of the Georgetown Department of Pharmacology, is offering a new "5-Minute Fast Stats" powerpoint that provides information about a few critical concepts in medical statistics: the difference between absolute and relative risk (and between absolute and relative risk reduction), the number needed to treat (NNT), and the number needed to harm (NNH). The powerpoint is quite good for understanding the critical difference between absolute and relative risk--and most importantly, it points out the ways that drug manufacturers use the different numbers to emphasize benefits and minimize risk. The section on NNT and NNH is slightly more confusing--we would have focused more on interpreting the numbers and what they mean for patients, rather than computing them--but the concepts remain important.
We certainly hope that doctors will take advantage of the presentation (and the other resources on PharmedOut's website), and take the time to share it with their colleagues (download here). The more our doctors understand the messages they're getting from medical research and pharmaceutical marketing, the better off we'll be.