Following up on Dave’s observations regarding doubts about the health benefits of moderate drinking:
(1) The politics of epidemiology are most obvious when you see people treat relatively weak correlations from observational studies as definitive proof of their pet hypothesis, and then turn around and come up with 100 reasons why such correlations don’t really prove anything when the correlations go the “wrong” way.
You see this all the time in obesity research, where slight increases in mortality risk at high weight levels are trumpeted as terribly significant, while much higher increases in mortality risk at weight levels within the supposed normal or recommended range are dismissed as products of other uncontrolled-for confounders.
On a related note, the NY Times story points out there are no randomized clinical trials demonstrating that moderate drinking is beneficial. Most people, I’m sure, would be surprised to learn that there aren’t even any observational studies suggesting that significant long-term weight loss is beneficial, let alone any randomized clinical trials demonstrating this.
(2) In a sense, the claim that the beneficial health effects of moderate drinking haven’t been demonstrated rigorously is rather beside the point. It’s true that the available data only suggest (more or less strongly, depending on how they are interpreted) this. But that data is much more powerful when considered in the light of a far weaker hypothesis: that moderate drinking isn’t a significant health risk. After all, while the somewhat better health of moderate drinkers in comparison with abstainers might be accounted by variables that have nothing to do with drinking, what’s extremely unlikely is that moderate drinking could have a significantly detrimental health effect, and yet produce these same correlations.
The relevance of this is that surely almost no one drinks moderately (as opposed to abstaining) because of the supposed health benefits. Whether moderate drinking is good for your health, or largely neutral (as opposed to being positively bad for you) is socially significant only if people were to treat moderate drinking like medicine — as something they’ll consume if it benefits their health, but otherwise not. The whole “is moderate drinking actually good for you?” argument is based implicitly on the assumption that it might be good for people to drink who otherwise wouldn’t, on, as it were, doctor’s orders. Which is a really bad idea.