But there seem to be some studies to the contrary (or maybe those with a deeper understanding of the area can reconcile these results)
This is quite common in the early stages of hunting for causation. See CETP inhibitors for instance - where the epidemiological data is very conflicted.