14 drinks a week carries a 1 in 25 lifetime risk of death from alcohol

For years, official US guidance allowed men up to 14 drinks per week, and the idea that light drinking might even protect the heart has shaped how people think about alcohol. Whether any level of drinking is genuinely safe is one of the most consequential questions in public health, because small risks spread across a whole population add up to a great many deaths. This study set out to put actual lifetime numbers on the risk of dying or falling ill because of alcohol at each level of average weekly consumption, and to test whether a protective effect exists at low intakes at all.

The researchers built a cause-specific risk model for the United States rather than running a trial. They combined drinking data from national health surveys with published relative risks for alcohol-related conditions, population figures from the US Census Bureau, death records from the Centers for Disease Control and Prevention, and illness data from the Institute for Health Metrics and Evaluation. From these inputs they estimated the lifetime risk of alcohol-attributable death and disease according to a person's average weekly consumption. A separate narrative review examined how the amount drunk per occasion, rather than per week, affects health.

No protective net effect appeared at low levels of drinking; risk simply rose with intake, starting at relatively low levels. Lifetime risk of an alcohol-attributable death passed 1 in 1,000 for men averaging more than 6.5 drinks per week and women averaging more than 7.0. Above 8.5 drinks per week, the risk for both sexes exceeded 1 in 100. At 14 drinks per week, the upper limit of the former Dietary Guidelines for men, the risk of an alcohol-caused death reached 1 in 25, or 4%. Pattern mattered too: above 1 drink per occasion, higher consumption was associated with progressively greater risks of breast cancer, cardiovascular disease and injury.

The absence of any protective floor adds to a growing pushback against the long-held belief that moderate drinking benefits the heart. This is a model rather than an experiment, so its figures depend on the quality of the survey data and the relative risks it draws on, and different assumptions would shift the exact thresholds. The authors conclude the results support tightening US alcohol guidance, for both men and women, to no more than 1 drink per day. What a randomised test of low-level drinking would show remains open, since such trials are rarely feasible.

If you drink at a level that feels moderate, this analysis is a reason to question whether that word still means what it used to. The model places meaningful lifetime risk at intakes well below the old guideline ceilings, and it found no level at which alcohol did net good. None of this dictates what anyone should drink; it is one modelling study, and personal risk also depends on pattern, since harm climbed with the amount consumed per occasion as well as per week. What it does remove is the comfort of assuming a couple of daily drinks is neutral.