Alcohol

The largest analysis of drinking amount and death combined individual data on 599,912 current drinkers without cardiovascular disease from 83 prospective studies in 19 high-income countries, with 40,310 deaths over 5.4 million person-years of follow-up. All-cause mortality rose along a curve with the amount drunk, with the minimum risk at or below about 100 g a week. The study corrected its estimates for the way individual drinking changes over time, using 152,640 repeat measurements [2].

Compared with drinking more than 0 and up to 100 g a week, drinking 100 to 200 g, 200 to 350 g, or more than 350 g a week was associated with a life expectancy at age 40 that was shorter by about 6 months, 1 to 2 years, and 4 to 5 years respectively [2]. At about 14 g of alcohol per US standard drink, those bands correspond to roughly 7 to 14, 14 to 25, and more than 25 drinks a week.

At the population level, the Global Burden of Disease Study 2016 attributed 6.8% of age-standardised male deaths and 2.2% of female deaths worldwide to alcohol. Among people aged 15 to 49, the leading alcohol-attributable causes of death were tuberculosis, road injuries and self-harm; from age 50, cancers made up 18.9% of alcohol-attributable deaths in men and 27.1% in women. Using data from 694 sources on drinking and 592 studies of risk, the analysis estimated that the level of consumption minimising total health loss was zero (uncertainty range 0.0 to 0.8 standard drinks a week) [1].

For decades, many studies found a J-shaped curve, in which light and moderate drinkers had lower mortality than non-drinkers. A central problem was the comparison group: people classed as abstainers often included former drinkers, many of whom had stopped because of illness, which made non-drinkers look less healthy. A 2016 meta-analysis (a study that pools the results of earlier studies) of 87 studies covering 3,998,626 people reproduced the J-curve without adjustment, with low-volume drinkers showing 14% lower mortality (relative risk 0.86, where relative risk is the risk in one group divided by the risk in another). Former drinkers had 22% higher mortality than the abstainer group. After adjustment for these abstainer biases and for study quality, low-volume drinking showed no reduction in mortality (relative risk 0.97, 95% CI 0.88 to 1.07) [5].

A 2023 update of 107 studies with 4,838,825 participants and 425,564 deaths reached the same conclusion. Compared with lifetime non-drinkers, occasional drinkers (under 1.3 g a day, relative risk 0.96) and low-volume drinkers (1.3 to 24 g a day, 0.93) showed no statistically significant reduction in mortality. Mortality was higher at 45 to 64 g a day (1.19) and at 65 g or more (1.35), and increased risk began at lower levels of intake in women than in men [4].

The same comparison-group problem affects other outcomes. In a meta-analysis of 38 studies of 1,902,605 people and type 2 diabetes, moderate drinking was associated with up to 18% lower risk, but the apparent benefit was seen only in women and non-Asian populations and was absent in studies that compared drinkers with people who had never drunk [8]. Some lower risks are specific to particular conditions: alcohol was associated with slightly lower risk of myocardial infarction (heart attack) in the 599,912-drinker analysis [2], and light and moderate drinking with lower risk of ischaemic stroke in a meta-analysis of 27 studies [9]. These associations sit alongside higher risks of other cardiovascular diseases, cancers and injuries at the same levels of intake.

All of this evidence is observational: it comes from tracking drinkers rather than assigning people to drink, and it relies mainly on self-reported intake, which tends to understate consumption. No randomised trial has measured the effect of long-term drinking on survival.

Alcohol is classified by the International Agency for Research on Cancer as a Group 1 carcinogen, the category for agents with sufficient evidence of causing cancer in humans. A meta-analysis of 572 studies including 486,538 cancer cases identified clear dose-related increases in risk for cancers of the mouth and throat, oesophagus, bowel, larynx and female breast, and higher risk in heavy drinkers for cancers of the stomach, liver, gallbladder, pancreas and lung [10].

A 2021 analysis estimated that 741,300 new cancer cases worldwide in 2020, 4.1% of the total, were attributable to alcohol. Men accounted for 76.7% of these cases. Cancers of the oesophagus (189,700 cases), liver (154,700) and breast (98,300) contributed most. Heavy drinking (over 60 g a day) accounted for 46.7% of the alcohol-attributable cases and risky drinking (20 to 60 g a day) for 39.4%, while moderate drinking (under 20 g a day) still accounted for 103,100 cases, 13.9% of the total [3].

Apart from myocardial infarction, alcohol was associated with higher risk of every major type of cardiovascular disease examined in the 599,912-drinker analysis, with no clear threshold below which lower intake stopped lowering risk. Each additional 100 g a week was associated with 14% higher risk of stroke, 9% higher risk of heart failure, 24% higher risk of fatal hypertensive disease and 15% higher risk of fatal aortic aneurysm [2]. In a meta-analysis of 27 studies, heavy drinking (more than four drinks a day) was associated with 67% higher risk of intracerebral haemorrhage and 82% higher risk of subarachnoid haemorrhage, two forms of bleeding stroke [9].

Alcohol raises blood pressure. In a dose-response meta-analysis of 23 studies, compared with 12 g a day, the risk of developing hypertension was 11% lower at zero intake and 11%, 22% and 33% higher at 24, 36 and 48 g a day [11].

Alcohol is a major risk factor for liver cirrhosis (scarring of the liver). In nine studies with 2,629,272 participants, risk rose steeply with intake. In women, 5 to 6 drinks a day was associated with 12.4 times the risk of long-term abstainers and 7 or more drinks with 24.6 times the risk; in men the corresponding estimates were 3.8 and 6.9, with wide uncertainty. One drink a day was associated with higher risk in women but not in men [12].

In brain scans of 36,678 middle-aged and older adults in the UK Biobank, alcohol intake was associated with smaller total brain and grey matter volumes and poorer white matter structure. These associations were already apparent at an average of one to two UK units a day and grew stronger with higher intake [13]. Among 31.6 million adults discharged from French hospitals between 2008 and 2013, alcohol use disorders were the strongest modifiable risk factor for dementia, with about 3.3 times the risk in both women and men; of 57,353 cases of dementia beginning before age 65, 38.9% were alcohol-related by definition and a further 17.6% had a recorded alcohol use disorder [14]. Other contributors to dementia are described under dementia risk factors.

Alcohol shortens the time taken to fall asleep and deepens the first half of the night, but increases sleep disruption in the second half and delays the first period of REM sleep (rapid eye movement sleep, the stage associated with most dreaming) at all doses studied [15]. A 2024 meta-analysis of 27 studies found that REM sleep was reduced even after a low dose (about two standard drinks), and that the reduction grew with dose [16]. The wider factors that affect sleep are described under sleep quality.

Suicide risk rises with average drinking. In a dose-response meta-analysis of eight studies, an average of 10 g a day was associated with 11% higher risk of death by suicide in men and 64% higher risk in women, compared with lifetime abstention, and risk increased further at higher intake [17].

The pattern of drinking affects risk in addition to the weekly total. In a meta-analysis of 14 studies with 4,718 coronary heart disease events, people who had irregular heavy drinking occasions (more than 60 g, or five or more drinks, at least monthly) had 45% higher risk than regular moderate drinkers with the same average intake (relative risk 1.45, 95% CI 1.24 to 1.70) [6].

Most alcohol-related injuries follow a single occasion of drinking. In a meta-analysis of studies relating the amount drunk on one occasion to injury, the odds of a motor vehicle crash rose by 24% for each 10 g of alcohol, reaching 52 times the odds of not drinking at 120 g. For other injuries, the odds rose by 30% per 10 g, reaching 24 times at 140 g [18]. Road injuries were the second leading alcohol-attributable cause of death among people aged 15 to 49 worldwide in 2016 [1]. Other causes of road deaths are described under road safety.

Some risk markers change within weeks of stopping. In a prospective observational study, 94 moderate to heavy drinkers who chose to abstain for one month were compared with 47 who continued drinking. The abstaining group showed falls in insulin resistance (25.9%), systolic blood pressure (6.6%), diastolic blood pressure (6.3%), body weight (1.5%) and two cancer-related growth factors, VEGF (41.8%) and EGF (73.9%), with no significant changes in the group that kept drinking [7]. Participants chose their own group, so the two groups may have differed in other ways.

In a meta-analysis of 36 trials with 2,865 participants, reducing alcohol lowered blood pressure in people who drank more than two drinks a day, but not in those who drank two or fewer. The largest effect was in people drinking six or more drinks a day who cut their intake by about half: systolic pressure fell by 5.5 mm Hg and diastolic pressure by 4.0 mm Hg [19].

National guidelines vary. The UK Chief Medical Officers advise that, to keep health risks low, both men and women should not regularly drink more than 14 units (112 g) a week, and that people who drink that much should spread it over three or more days, because one or two heavy drinking episodes a week increase the risk of death from long-term illness and from injuries. The limit was set at the level at which the lifetime risk of dying from an alcohol-related condition is about 1% [20]. The 599,912-drinker analysis concluded that its findings supported limits lower than those in most guidelines at the time [2]. Tobacco use is described under smoking.

Full deep dive: Alcohol and health.