It is not limitations. It is just that the findings should be presented truthfully and the authors should not try to suggest more than can actually be inferred

It is not limitations. It is just that the findings should be presented truthfully and the authors should not try to suggest more than can actually be inferred from the data. Correlation between facts is interesting in itself, but to general public it should be clearly presented that "although we can prove these things are correlated, we don't know why and how and whether one causes another or vice versa or there is something else causing dementia as well as making people more likely to supplement with vitamin D". Without putting this warning label I treat it as kind of "scientific clickbait"