My guess based on the article would be it is not just looking at a spot change but at an overall curve, and considers the shape of the curve too (e.g. the comme

My guess based on the article would be it is not just looking at a spot change but at an overall curve, and considers the shape of the curve too (e.g. the comment about it getting flatter depending on the amounts of fat and protein). So, I'd guess the shape for water intake might be so radically different it can be cancelled out. Just a guess, doing the (perhaps) scamsters' work for them, though...