There is actually pretty good epidemiological evidence to support the notion that coffee protects against many types of liver disease. What there is not (to my knowledge) is any appropriately powered randomized trial to prove the point.
And, one should beware treatments based strictly on epidemiological considerations. Two examples come to mind where in each case randomized trials showed a net negative health effect despite overwhelming and rather compelling epidemiological evidence that suggested the contrary.
- estrogen for postmenopausal women
- vitamin E to prevent cancer