His notion that in the field of medicine you should disregard contrarian anecdotes because there's statistical evidence is horrible advice. If you actually looked at said statistical evidence, you'd realize it is very rarely strong, and often only relevant to 70% or so of the population (which is great for 2 out of 3, but useless for the third one).
"Best practices" often aren't, and common sense is not at all common in medicine. A significant number of published results are plain wrong (see http://saveyourself.ca/articles/ioannidis.php and the paper it references). A lot of medical advice is wrong, harmful or useless; The archives of seth roberts' blog are an enlightening read.
The problem is that this is impossible for most people.