A new review and meta-analysis of alcohol consumption has challenged the idea that light drinking lowers all-cause mortality versus abstaining from alcohol.

https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2802963

There is no statistically significant reduction in mortality from light drinking. And heavy drinking is associated with an increase in all-cause mortality. For women, this increased risk begins at a lower drinking threshold compared to men.

What they found was that past studies showing a protective effect had confounding. The abstainer comparison group in many studies contained former drinkers and occasional drinkers, instead of true abstainers. In addition light drinkers often had other beneficial health habits and characteristics compared to abstainers that offset the negative health effects of alcohol consumption.

We see this in a lot of observational studies on smoking as well that produce odd results suggesting some protective effects. For example those who don't smoke may be former smokers that had to quit due to cancer or emphysema. If you don't account for that you may bias against the no smoking group.

It is true that there are cardiovascular benefits to light drinking, but they get offset by the negative effects of alcohol on liver cirrhosis, cancer risk, etc.

For women the threshold was 25 grams of alcohol (roughly more than ~2 drinks a day). For men it was 45 grams or more.

The right side is the adjusted model. So statistically significant increase in risk for women at 25 grams or more. And for men at 45 grams or more.