Paul Matiuzzi's blog post nails the controversary about "empirically-based treatment" for therapists: Read it here: http://everydaypsychology.com/2011/09/yale-psychologist-calls-for-end-of.html
Empirically-based research is a helpful guide to what works for some people some of the time. I absolutely follow the research in my field. But if something doesn't work for the client in front of you, you must try something else.
In about an hour, I'll be explaining to my editor why I can't, in good conscience, write the protocol-driven trauma treatment plan book she's asking for. Trauma clients are individuals with ideosynchratic biologies, temperaments, histories, and cultures. One treatment plan can't fit all. My last book was all about knowing as many therapies as possible, so that you can find what works for a client in front of you.
For a scholarly (hilarious) article about the issue, check out this study on "Parachute use to prevent injury", that fails the empirical test for lack of a "no parachute use" control: http://www.bmj.com/content/327/7429/1459.full