And here's a second question to those that have suggested that...

And here's a second question to those that have suggested that the differences in the outcomes of the 2004 trials were an artifact of differing drop out rates among the dosage groups:

Assuming that drop-outs were self-selected from those who weren't doing well (an entirely reasonable hypothesis from my point of view, and the key hypothesis of those proposing this as an explanation for the trial results), why did the placebo group, with a dropout rate of 18% do worse than the 10 and 20 mg treatment groups with the dropout rates of 11% and 6% respectively, if the drug is ineffective?