I know "correlation/causation" feels like a dead horse, but I would want far better evidence that there's an actual link and not just confounding factors (ex. diet).