Livre d'Or

I suspect we're looking at the difference between clinically significant and statistically significant. If it's true that for the individual, losing 5% of body weight is medically important, it's also true that for the next individual, gaining 3% is likely to be.

The individual patient who is being offered this intervention should be told both that it might help them lose weight, and that they might put in lots of effort and gain weight. In other words, they should treat it more like prescribing a drug: a competent doctor, if she offers Prozac, will warn the patient that there is a small but important chance that it will increase rather than relieving depression.