Population studies don't have control groups and are largely worthless for recommending supplementation. There are just too many confounding factors.