We linked a a paper in Scientific American a while back, Health Care Myth Busters: Is There a High Degree of Scientific Certainty in Modern Medicine?
We know, for example, that when a patient goes to his primary-care physician with a very common problem like lower back pain, the physician will deliver the right treatment with real clinical benefit about half of the time. Patients with the same health problem who go to different physicians will get wildly different treatments. Those physicians can't all be right.
Overall, physicians are said to get it wrong around 50% of the time. I suppose that is possible. I get it wrong on a regular basis. Dr. DB says he trusts no-one in medicine, including himself.
More from the Scientific American article:
Give surgeons a written description of a surgical problem, and half of the group will recommend surgery, while the other half will not. Survey them again two years later and as many as 40 percent of the same surgeons will disagree with their previous opinions and change their recommendations. Research studies back up all of these findings, according to Eddy.