The biggest flaw in this report is that it doesn't say what classifies as a non-drinker!? It seems to imply that anyone who drinks 1 or more units of alcohol day is a drinker, so that was suggest anything less has no demonstratable negative effects.

These kinds of studies are usually based self-reporting via questionnaires, so the data is going to be wildly inaccurate. Any small correlation is in the data is probably meaningless. The other problem is that the risks of consuming alcohol are different for each person. Failing to take someone's diet into account also invalidates the results.