Here is a paper (series of three) from authors who are experts in this field. I've always been a bit ambivalent as to whether guidelines actually work - in terms of changing behaviour. I'm still not convinced but it's good to see a bit more work being done to identify the target audiences for guidelines as well as considering conflicts of interest etc. Getting them right is a lot more complex than just collating what seems to be best practice, writing it up and then distributing them widely.
Oh, and the full text of the papers is free - that's nice. http://www.implementationscience.com/series/ClinPracGuidelines