There are lies, damned lies, and statistics. 24,000 may sound a lot, but not true given the wide genomic diversity. Add to it self-reporting variation, any claim of so-called "adjustment for such and such factors" is what we normally call "fudge factors". Given known biochemical mechanisms how alcohol affects the brain, take this with a grain of salt. Just like similar studies claiming alcohol enhances the risk of dementia. The existence of both contradictory claims simply shows that the effect of alcohol on dementia is not definitive. One might have protective genes and another might not.