Level of alcohol for the lowest mortality

What is the level of alcohol consumption at which mortality is minimised?

One of Bandolier's healthy living tips is to drink alcohol regularly ( Bandolier 78 ). Moderate alcohol consumers have lower mortality than either non-drinkers or heavy drinkers. This meta-analysis aims to quantify the level of alcohol which provides the most protection against death and disease.


Limited results could be extrapolated from this meta-analysis due to differences between studies. However, the level of alcohol consumption to minimise mortality was found to be much lower for women than for men. The results suggest that the estimated levels of alcohol to minimise mortality in the UK are approximately half the current recommendations (which are 28 units a week for men and 21 units for women).


I White. The level of alcohol consumption at which all-cause mortality is least. Journal of Clinical Epidemiology 1999 52: 967-975.


MEDLINE was searched up until the end of 1995. Studies were excluded if mortality was not reported for men and women separately or if fewer than three groups of alcohol consumption were reported. This resulted in twenty cohort studies with a total of 451,930 men (60,224 deaths) and 758,615 women (74,824 deaths).

Studies measured alcohol consumption differently, although all methods were self-reporting. For this meta-analysis, amounts of alcohol were converted into units, where one unit of alcohol = 9g.


The estimated levels of alcohol consumption for the lowest mortality were:

for UK men, 12.9 units per week (95% confidence interval 10.8 to 15.1)

for US men, 7.7 units per week (95% confidence interval 6.4 to 9.1)

for US women, 2.9 units per week (95% confidence interval 2.0 to 4.0).

Table 1 gives the number of studies, participants and deaths contributing to these results. All results were adjusted for age, all but two US male studies were adjusted for smoking. Participants were followed for 8 to 22 years.

Table 1. Estimated levels of alcohol consumption at which mortality was lowest.


Studies (Participants)


Estimated Alcohol Level (per week)

95% Confidence Interval

UK men

3 (17,973)



10.8 - 15.1

US men

10 (361,280)



6.4 - 9.1

US women

7 (750,321)



2.0 - 4.0



Studies were conducted in different countries and results varied substantially between them. Consequently, results could only be pooled by country. Only the US and UK had more than one set of results to pool (although all contributing studies had large numbers of participants). The UK studies only investigated men.

The UK guidelines for alcohol consumption are no more than 28 units a week for men and 21 units for women (where one unit = 8g). Results from this meta-analysis suggest that approximately half these amounts are the optimum levels to minimise mortality. However, self-reporting alcohol consumption (used in the studies) is likely to underestimate true consumption, which may result in lower estimates.

The high rates of coronary heart disease in the UK may explain the UK's higher estimated level of alcohol (given alcohol's protective effect against coronary heart disease). As mortality from coronary heart disease declines, the estimated level of alcohol would be expected to fall.