In some trials, subjects have responded just as well to sugar pills as they have to real treatment. So how can you trust any trial results?