Adams has a good point relating to alcohol, though. They haven't been unable to untangle correlation and causation. I used to be a "moderate drinker" until I developed (unrelated) health problems that mean taking medication every day. When I take my meds I can't drink. These health problems are statistically likely to end me sooner than most people.

So I'm likely going to be in the "doesn't drink dies earlier than average" column, but it has nothing to do with alcohol.