Alcohol caused an estimated 741,000 cancers worldwide in 2020
Building on the IARC classification of alcohol as a Group 1 carcinogen, this global modelling study attributed 4.1% of all new cancers in 2020 to drinking, three-quarters of them in men and the largest numbers in oesophageal, liver and breast cancer.
IARC Monographs volumes 44, 96 and 100E established alcoholic beverages, and the acetaldehyde derived from them, as Group 1 carcinogens with sufficient evidence for cancers of the oral cavity, pharynx, larynx, oesophagus (squamous), liver, colorectum and female breast. Rumgay and colleagues combined per-capita alcohol consumption estimates (with a 10-year lag) and relative risks from meta-analyses with GLOBOCAN 2020 incidence to estimate the attributable burden by country, sex and site.
An estimated 741,300 new cancers (4.1% of all cases) were attributable to alcohol in 2020; 76.7% were in men. Oesophageal (189,700), liver (154,700) and breast (98,300) cancers had the largest numbers. Heavy drinking accounted for most of the burden, but moderate drinking (up to two drinks a day) contributed around 100,000 cases.
The paper drove policy interest in alcohol labelling and pricing, and the 2023 WHO and 2025 US Surgeon General statements on alcohol and cancer.
- 741,300 alcohol-attributable new cancers globally in 2020 (95% uncertainty interval 558,500-951,200), 4.1% of all cancers
- 76.7% of attributable cases were in men
- Largest contributions: oesophagus 189,700; liver 154,700; female breast 98,300
- Moderate drinking (less than 20 g/day) accounted for an estimated 103,100 cases (13.9% of the alcohol burden)
There is no safe level of alcohol for cancer risk, and the risk is highest for cancers of the mouth, throat, oesophagus, liver, bowel and breast. Public awareness is low; most people do not know alcohol causes breast cancer. Warning labels and minimum pricing are the policy levers being debated.
- Modelled estimates depend on consumption data of uneven quality and on relative risks from observational meta-analyses
- Unrecorded alcohol consumption is poorly captured in many countries
- Does not account for former drinkers or interactions with smoking
- Population-level estimates say nothing about individual risk trajectories