This study uses an inscrutable statistical model to arrive at this conclusion. On the other hand I have yet to see an epidemiological study that shows abstaining results in a lower hazard ratio than moderate drinking.
I'm sure that similar issues exist with "Alcohol use and burden for 195 countries and territories" but the statistical model is too complicated to really see what's going on. I agree that at a population level alcohol is bad for longevity but at an individual level there seems to exist no evidence to suggest that moderate drinking is.
Consider the following study: https://www.thelancet.com/pdfs/journals/lancet/PIIS0140-6736... This study only calculates HRs between current drinkers, finding that the lowest mortality rates occur below 100g/week of alcohol. The issues with the conclusions of this study (issues which are apparent in many other alcohol studies) are addressed in the follow-up here: https://www.thelancet.com/pdfs/journals/lancet/PIIS0140-6736...
I'm sure that similar issues exist with "Alcohol use and burden for 195 countries and territories" but the statistical model is too complicated to really see what's going on. I agree that at a population level alcohol is bad for longevity but at an individual level there seems to exist no evidence to suggest that moderate drinking is.