An analysis of over 400,000 people's self-reported drinking habits finds that statistically drinking alcohol 4 or more times per week increases the risk of premature death by 20 percent. The increased risk was consistent across age groups.

Because this is epidemiology, a lot of caution must be placed near the top. Recently, The Lancer published a paper they knew was bogus but had a media friendly claim; that the safe level of alcohol was none. But that was just a review of studies and included binge drinkers and alcoholics. And every paper of this kind will suffer from a glaring weakness; it relies on people accurately reporting how much they drink over time, which is then used to sift through data and see how many people got a disease. As you can infer, if you have enough diseases you are almost certain to find any chemical that will cause or cure it. 

This analysis of survey results looked at 340,668 participants, ages 18-85, in the National Health Interview Survey, and another 93,653 individuals, ages 40-60 who were treated as outpatients at Veterans Administration clinics. 

Twenty percent means different things to different people, which leads to the other problem in epidemiology - the difference between absolute and relative risk. If your risk is 1 in 10,000,000 and it goes up 20 percent, the actual concern level is minimal. If your risk is 1 in 10 and it goes up 20 percent, that is worrisome. But because this is just survey results, it doesn't lend itself to personalizing the results, it will just be used by lawyers who want to sue the alcohol industry to convince a jury to give them a large judgment. 

Citation: Hartz SM, Oehlert M, Horton AC, Grucza R, Fisher SL, Culverhouse RC, Nelson KG, Sumerall SC, Neal C, Pegnier P, Chen G, Williams A, Bhattarai J, Evanoff B, Bierut LJ. Daily drinking is associated with increased mortality. Alcoholism: Clinical&Experimental Research, Oct. 3, 2018.