Despite what you and your "experts" in your two-year-old links say, it does work. Not sure why you refuse to read Dr. Risch's paper, but oh well. If you did, you would learn non-parallel RCTs were used to "prove" it doesn't work. You would learn that not one parallel RCT was ever conducted (due to the inaccurate and later-retracted story in The Lancet). And you would learn every well-executed parallel observational trial shows it works.