Alcohol consumption and mortality from all causes, coronary heart disease, and stroke: results from a prospective cohort study of Scottish men with 21 years of follow up

The paper by Hart et al is intriguing, as is all research that
challenges accepted wisdom. However, the authors seem to assert their
finding that alcohol consumption was associated with excess mortality more
strongly than is supported by the data. They state 'Men who drank over 15
units of alcohol a week had significantly higher risks of dying from all
causes than the non-drinkers' and that 'drinkers of more than 15 units a
week had significantly higher risks [of stroke] than non-drinkers.'
Neither of those statements is in fact true after adjustment for smoking.

Smoking is likely to be a major confounder in studies of this kind,
yet the authors only touch on it briefly. It would have been interesting
to know how it was included in the statistical model (eg were people
simply classified as smokers or non-smokers, did it take account of former
smoking habits, was the number of cigarettes smoked daily included?), and
what the relative risks of mortality associated with smoking were in their
dataset.

It is certainly interesting that this study did not find the J-shaped
relationship between alcohol consumption and mortality found in many other
studies, and the reasons for that discordance is certainly worthy of
further study. However, this study has provided no conclusive evidence
that drinking fewer than 35 units per week is harmful.