Background Recent guidelines from the UK Chief Medical Officer recommend both men and women should drink no more than 14 units of alcohol a week. This change in guideline has been influenced in part by the increased risk alcohol exerts on several cancers; however, there is a dearth of high quality cohort data from the UK on this topic. This study aims to establish whether various measures of alcohol consumption are associated with breast, lung, and prostate or colorectal, which combined account for half of UK cancer deaths.

Methods A cohort of 8,670 participants representative of the UK population were interviewed in 1984/5 about alcohol consumption and had follow-up data from cancer registries available for 25 years. The primary exposure measures were number of units consumed, and self-reported alcohol consumption (‘non/occasional’ to ‘heavy’). Secondary exposures included persistent consumption over recommended guidelines and different alcohol types. Cox proportional hazards models were used to estimate hazard ratios (HR) for each alcohol measure on the risk of each cancer separately.

Results The number of cancers ranged from 110 for prostate to 182 for breast cancer (incidence rate of 1.88 per 1,000 person-years). Eight statistical comparisons out of 36 were significant. The number of units consumed fitted as a liner term was not significantly associated with any of the four cancers. However, a HR of 1.12 per 10 units/week (95% CIs: 1.03–1.23) was found for number of units with lung cancer when restricted to men. Also, in the sensitivity analyses that were restricted to those whose alcohol diaries were ‘typical’, HRs of 1.33 (95% CIs: 1.04–1.68) and 1.14 (95% CIs: 1.04–1.25) were observed per 10 units/week with breast and lung cancer respectively. For spirits alone, there were HRs of 1.63 (95% CIs: 1.01–2.64) and 1.33 (95% CIs: 1.01–1.76) per 10 units/week with breast and colorectal cancer respectively. Finally, self-reported heavy consumption was associated with an increased risk of breast cancer (HR: 4.13, 95% CIs: 1.25–13.63).

Conclusion Alcohol consumption is associated with breast cancer, particularly at high levels, and lung cancer, however the latter may be due to residual confounding by smoking. These results do not support the recent changes in alcohol guidelines, as they do not find an increased risk at low consumption levels, although this may be due to the study being underpowered to detect small effect sizes. The results also demonstrate alcohol type is important when assessing risk and this should be researched further.