This interesting study looks ripe for a detailed examination of the causal claim and reasoning. Would be lovely if true. But can studies of this kind ever amount to convincing evidence? If so, how? If not, can claims of this kind ever be established? 

http://m.aje.oxfordjournals.org/content/182/12/1010.abstract?etoc
Various questions present themselves:

• Is there a clearly defined intervention, as the Potential Outcomes people would insist?
• Is there a clearly specified mechanism, as some philosophers of science would ask for?
• Is there a better explanation than causality, as Hill would ask?
• Would any of these help us be more sure that the association was not due to confounding (lusty, vigorous, or rich people drinking more coffee, for instance)?
• Could triangulation, crossword-type reasoning, and similarly hard-to-quantify approaches help?
• Does it even make sense to think of coffee as having a uniform effect on health, if you take caffeine out of the equation, given the variety of drinks going by that name?

I am going to have a cup of coffee and think about these questions.