select search filters
roundups & rapid reactions
factsheets & briefing notes
before the headlines
Fiona fox's blog

expert reaction to the association between lifetime drinking and mortality and cancer risk

A new cohort study, published in PLOS Medicine, examines the association of lifetime alcohol use with mortality and cancer risk in older adults.


Dr Tony Rao, Visiting Researcher, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, said:

“This study is large enough to provide a meaningful interpretation of its results and is unique in that it includes former drinkers and examines lifetime drinking patterns in an older population. However, the lack of information in important areas such as socio-economic status could well have influenced the findings and needs to be looked at more closely in future research.

“It is important to note that the study did not show a protective effect of light lifetime alcohol consumption on mortality from suicide, homicide or accidents and provided no data on mental disorders such as dementia. Given the finding of lowest mortality seen in people drinking fewer than 0.5 units of alcohol per day, the results may also provide further evidence of the need to examine lower risk drinking guidance in older people”.


Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:

“Many previous studies have found a J-shaped relationship between alcohol consumption and death rates, particularly for deaths from heart disease and strokes. In other words, they have found that the chances of dying in a given period of time tend to be a bit larger in people who do not drink any alcohol than in people who drink a small amount. As the amount of alcohol increases above a certain small level (which varies a bit from one study to another), death rates increase again so that they rise above the level for people who drink no alcohol. So why is it worth doing another study like this?

“One problem is that the studies that produced these results (including this new one) are observational. People aren’t allocated an amount of alcohol to drink. Instead, researchers ask how much they drink or used to drink, and follow them up for years to observe when they fall ill or die. But people who drink no alcohol tend, on average, to be different from people who do drink, even a little, in several ways – for example in terms of culture, or religion, or upbringing – and perhaps it is something else to do with these differences that leads to the increased risk rather than the fact they don’t drink. And the group of people who do not drink might include people who have given up drinking for health reasons, and it could be those health reasons that cause the increased death rates compared to those who drink a little.

“Other studies have dealt with the issue of people who gave up drinking for health reasons by asking about people’s previous drinking habits, and perhaps comparing drinkers with people who have never drunk alcohol in their lives rather than with anyone who happens not to be drinking when they enter a study. This new study goes further down that path, by estimating people’s past drinking over their whole lives from age 18 onwards. The same J-shaped pattern was seen for overall death rates, and for death rates from heart attacks and strokes, but not for deaths from cancer, or for the rates at which new cases of cancer arise. Those rates for cancer are lowest for people who never drank alcohol, and increase steadily with alcohol consumption, in line with previous findings.

“This all does clarify the picture a bit, but there are still some difficulties of interpreting what it all means. Using past drinking across most of people’s lives does, in principle, get over the problem of people giving up alcohol because of their health, but it doesn’t deal with other differences between people that affect their health and, coincidentally, happen to be related to their alcohol consumption. The researchers did allow statistically for some possible differences, but they mention that differences in socio-economic status couldn’t be allowed for (because they did not have the data), and there are other possibilities too. And then there must be questions about the accuracy of the measures of past drinking. Participants (all in the USA) entered this study at age 55-74, when they were asked to recall how much alcohol they had drunk a week when they were much younger, starting at age 18 to 24. I’m in the age group that was recruited to the study, and I don’t think I can recall particularly accurately how much I used to drink all that time ago, at least not in terms of giving a representative figure covering the whole time from age 18 to 24. And perhaps my recall of how much I used to drink would be affected by my current state of health in some way, making the picture yet more complicated. In some other studies of alcohol, people seem not always to be able to be particularly accurate on stating the amount they drink now, let alone 50 years ago. And the different alcohol consumption groups used in the study were then based on average consumption right from age 18 till the participants entered the study, so can’t take account of different patterns of consumption over the lifespan. These issues certainly don’t mean that the results are wrong, just that one shouldn’t perhaps claim too much for their detailed reliability.

“These new results do show overall death rates beginning to increase from a fairly small average consumption level – from around 5 standard American ‘alcoholic drinks’ a week, which is around 9 UK units of alcohol a week. (One US standard alcoholic drink is about 1.75 UK units of alcohol.) But those results are statistically uncertain, and the pattern really only becomes statistically clear for people drinking 2 or 3 US drinks a day (or about 3.5 to 5 UK units), and overall death rates don’t get higher than in people who report that they never drank alcohol until the level of about 4 US drinks a day (7 UK units), though the effects are more marked in women than in men. I’m certainly not advocating that people should drink more – these results are based on statistical models, and (because of the limitations I’ve mentioned, and others listed in the study report) I don’t think that, on their own, they form any kind of a clear guide for action by individuals.

“On cancer risks, this study does indicate, as have previous studies, steady increases in cancer risk with the average amount of alcohol consumed. But the increase isn’t very rapid, particularly for incidence rate (the rate at which new cancers arise). The researchers found that rate to be around 10% higher in people who drank, on average, 7 US alcoholic drinks (over 12 UK units) a day, and indeed there remained a little statistical uncertainty about whether there really was any increase – though the increase was much greater for specific cancers such as breast cancer, bowel cancer and liver cancer that are already known to be related to alcohol consumption.


* ‘The association of lifetime alcohol use with mortality and cancer risk in older adults: A cohort study’ by Kunzmann et al. published in PLOS Medicine on Tuesday 19 June. 


Declared interests

Dr Tony Rao: “No conflict of interest to declare other than receiving a fee for speaking on alcohol misuse in older people at a Pharma sponsored educational event, but which did not contain any reference to the drug manufactured by that company.”

Prof Kevin McConway: “Kevin McConway is a member of the Science Media Centre’s Advisory Committee.”

in this section

filter RoundUps by year

search by tag