J-curve probably not real
Many studies have replicated the J-shaped curve obtained by plotting all-cause mortality against alcohol consumption. The implication is that moderate drinking is associated with lower likelihood of death compared to teetotallers and heavy drinkers.
A great deal of work has gone into challenging this apparent correlation, and various research groups have attempted to identify and correct for confounders - like, for example, the issue that some people may be forced teetotallers because they are not healthy enough to drink. An example would be people with alcohol-induced liver damage. It didn't stop George Best.
Anyway, there are studies that attempt to correct for this, and other confounders. An example is
"Do “Moderate” Drinkers Have Reduced Mortality Risk? A Systematic Review and Meta-Analysis of Alcohol Consumption and All-Cause Mortality
Journal of Studies on Alcohol and Drugs, 77(2), 185–198 (2016). "
http://www.jsad.com/doi/10.15288/jsad.2016.77.185
Note it is from 2016, and has been able to learn from criticism of other meta-analyses.
The conclusion is that after adjustment for confounders, the correlation goes away
"...adjusting for these factors find that low-volume alcohol consumption has no net mortality benefit compared with lifetime abstention or occasional drinking."
There's a commentary on the paper here:
http://ebm.bmj.com/content/ebmed/early/2016/08/23/ebmed-2016-110490.full.pdf
And a press release/article on the study, which is quite accessible here:
https://www.eurekalert.org/pub_releases/2016-03/joso-imd031616.php
I do like a drink, but I also like evidence-based thinking and policy. I may no longer be able to claim that (flawed/incomplete) epidemiological studies show that moderate drinking is healthy.
Which the drinker in me thinks is a shame. Sigh.
(By the way, if I'm wrong, please don't just downvote me, but post a brief explanation or link explaining why I'm wrong, so I can learn - thanks).