There are studies (Eddy 1982; Gigerenzer and Hoffrage 1995) that show only a small percent of physicians understand even basic statistics.
Just Google same and you'll get a zillion links to freshman college homework discussing the problem.
After all, all those statistics are kind of meaningless. They rarely are split by sex, age, smoking behaviour, and familial history ("Where was the study of nonsmoking 36-year-old neurosurgeons?") It's a doctor's job to, somehow, find the best approach using incomplete data.
More importantly, all guidelines likely are about approaches that have changed since the data was collected. That new anesthetic procedure may have shifted things by a percent, that slightly adjusted monitoring protocol another, the improved CT scanner another, etc.
Medicine just isn't that much of an exact science as many of us think it is. In many cases, we haven't managed to make it a science (yet). For example, in IVF, even after years of working together to copy each other's best practices, one hospital might have a 30% success rate, another a 40% one, even after normalizing for every difference in patient population or treatment we can think of.