They really are a joke, and about the only reason they get published in journals is as a way for the institutions that fund them (AARP, NCI, etc.) to save face for wasting loads of money on a virtually useless study.
All this study demonstrates is that there is a positive relationship between the amount of coffee one drinks and how long they live. Big deal. It doesn't do anything to explore causal linkages (which they admit); it could be that coffee increases longevity, but it could just be that people who live longer tend to drink more coffee. There are probably millions of factors positively correlated with a long life, but that doesn't mean they are contributors to one.