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

expert reaction to study looking at alcohol consumption, including within UK guidelines, and structural images of organ systems

A study published in eLife uses UK Biobank data to look at alcohol consumption and structural changes in multiple organ systems.


Prof John Holmes, Sheffield Alcohol Research Group, ScHARR, University of Sheffield, said:

“The UK’s drinking guidelines provide advice on ‘low risk’ alcohol consumption, not ‘no risk’ alcohol consumption.  As such, they are set at a level meaning the average person will have only a very small risk of dying due to alcohol.  The guidelines take account of evidence that drinking small amounts of alcohol still carries some risk to people’s health, particularly in relation to cancer, but also for other health conditions.  While this new study adds further evidence that drinking small amounts of alcohol is not risk free, it does not provide any compelling reason to change the current guidelines.”


Prof Tim Spector, Professor of Genetic Epidemiology, King’s College London, said:

“This is an observational study of around 10,000 60 year olds without any control group that don’t drink.  While the risk of some MRI detected changes in the brain and heart increased with alcohol intakes, the risks these imply for an individual are unclear.  It is hard to put these risks in context against say the risk of driving a car which also has no lower risk limit or the risk of eating junk food that is likely much higher.  There are also no details on type of alcohol, as multiple studies have shown 1-2 glasses of red wine associated with cardiac protection.  While it is correct to point out the dangers of alcohol, these major limitations mean we should be wary of issuing any new guidelines that are unlikely to fit everyone.”


Dr Sadie Boniface, Head of Research, Institute of Alcohol Studies and Visiting Researcher, King’s College London, said:

“This study looks at the relationship between alcohol consumption and markers of brain, heart and liver health from medical imaging.

“The results do fit with a lot of existing evidence.  We know already that alcohol affects every organ of the body, and that heavy drinking is associated with Alzheimer’s, heart disease, and liver disease.  We also know there is ‘no safe level’ of alcohol consumption for certain other health conditions, such as the seven cancers that are caused by alcohol.

“The data come from the UK Biobank, which is an extremely large study.  There are however documented problems ( around how well the Biobank represents the population, for instance the response rate is very low and participants are healthier on average.  It’s not clear how this would have impacted the results of this particular study.

“A major review of the UK’s lower risk drinking guidelines was completed in 2016, and since then it has been recommended not to regularly exceed 14 units a week.  This was intended to balance the fact that many people enjoy drinking alcohol alongside what was seen as an acceptable risk of harm (  For example, drinking two units a day (i.e. the upper limit of the guideline level of 14 units a week) was calculated to carry a 16% relative increase in the risk of breast cancer among women, and an 11% relative increase in the risk of ischaemic stroke among men.  Because this was the approach taken by the recent guidelines review, I’m not sure if the drinking guidelines need to be amended as a result of this one study.

“Alcohol carries multiple health risks, and the only failsafe way to avoid these completely is not to drink at all.  While I don’t think this study means the lower risk drinking guidelines need to be amended, they do need to be better communicated.  The forthcoming consultation on alcohol labelling will provide a real opportunity to introduce independent health information on alcohol products so that so consumers can make fully informed decisions about their drinking.”


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

“I think the findings of this study need careful interpretation.  The researchers used data from a large group of late middle-aged people in the UK – the exact number of people is different for different measures in the study, but is around 10,000 for most of them.  They found associations between the amount of alcohol that people said they drank, and several measures of the structure of their brains, hearts, aorta (the body’s main artery, that runs down the abdomen), as well as with the amount of fat in their livers.  They found no evidence that these association were u-shaped or j-shaped, as has been found in some (but certainly not all) studies of association between alcohol consumption and various health measures.  In other words, it appeared that the differences on these measures between people who drank almost no alcohol, and people who drank very little, were always in the same direction as the differences between people who drank, say, a fairly small amount of alcohol and those who drank a rather larger amount than that.  But what might that all actually mean?  It’s very well known that drinking larger amounts of alcohol has many adverse effects on health.  Does this study show that drinking small amounts, possibly very small amounts, also has adverse health effects, on a considerably smaller scale?

“That’s certainly a possibility, but I think there are several other interpretations as well, and the way the study was conducted means that it can’t clearly distinguish between them.  It’s an observational study, and that makes it pretty well impossible to know what causes what, not from this study alone at any rate.  The issue is that there are many differences between the people in the study who drank different amounts of alcohol, apart from their alcohol consumption.  The differences that were found between people who drank different amounts could be caused, in whole or in part, by some combination of these other differences, and not just by the differences in alcohol consumption at all.  The researchers are aware of that, and clearly mention in in their report.  It’s possible to try to take into account some of these other differences, if the participants provided data on them, by using a statistical model that adjusts for those differences.  The researchers did do that, for several factors, and several associations between alcohol consumption and measures of the brain, heart and aortal structure and liver fat remained.  But we still can’t be sure that those associations are caused by the differences in alcohol consumption.

“Even without this concern about what causes what, several aspects of the findings are difficult to interpret, I’d say.  For example, the researchers found that a measure of brain volume was smaller, on average, in people who said they drank more alcohol.  The title of the research paper, “Alcohol consumption in the general population is associated with structural changes in multiple organ systems”, seems to imply that the sizes of the brains of those who drank more alcohol were changed by their alcohol consumption.  I’ve explained that the study can’t tell us whether the association between alcohol and brain volume is one of cause and effect.  But there’s a further issue.  Despite the word “changes” in the title of the paper, this study didn’t look at any changes in individual’s brain volumes or in any of the other measures.  What it found was that, after allowing for some other factors, people who drank larger amounts of alcohol had slightly smaller brains than those who drank smaller amounts.  Maybe those people had always had slightly smaller brains.  Because the study couldn’t look at change and couldn’t establish cause, it can’t actually tell is whether alcohol drinking changes brain volume.  It might, or it might not.  And, even if the effect of differences in alcohol consumption is related to changes and is a cause and effect relationship, I think we need to look at the size of the associations.  The researchers estimate that, if we compare this brain size measure in two groups of people like those in this study, one of which drinks four times as much alcohol as the other, the brain size measure is about 0.34% lower in the group that drinks the larger amount of alcohol.  How serious might that be?  The researchers also report that this measure of brain size goes down, after adjustment for other factors (including alcohol consumption), by 0.3% per year of age.  So, if the association between alcohol consumption and brain size is one of cause and effect (which we can’t be sure of), then in terms of this size measure, the group that drinks four times as much alcohol will have brains that correspond to those in the lower drinking group that are about one year older than them.  Adding a year to your brain age doesn’t sound wonderful – but to get a whole year difference involves drinking four times as much alcohol, and that’s a pretty big difference in alcohol consumption.  The issue is that some of these differences in the structure of organs just don’t seem to get very large unless the difference in alcohol consumption is really quite large.

“Also, this study didn’t measure the health of the participants, and it didn’t follow up the state of their brains, heart, and so on over time.  So does the extra year of brain age actually have a large or a small association with dementia or other problems in later life?  This study can’t tell us directly – it can indicate, alongside other studies, that there might possibly be issues, but it can’t be precise on what they are, even if the association is one of cause and effect.  That would need considerably more research, and the combination of findings from many different studies, I think.

“The position on heart structure is more complicated still, I think.  The researchers looked at seven different measurements of the structure of the participants’ hearts.  All of them were different in people of different ages, which isn’t surprising.  There was evidence that five of the measures differed between people that drank different amount of alcohol – for the other two measures, the size of the differences with alcohol consumption were small enough that they might (or might not) have been due to chance alone.  But of those five measures that went beyond chance variability, only one went in the same direction as aging, in that it was higher in people that drank more alcohol and also higher in people who were older.  The other four went in the opposite direction – that is, in comparing two groups, one of which drank more, the association went in the other direction from comparing two groups, one of which was older.  That’s not like the position with brain volume.  Instead, on some of these heart measures, if alcohol consumption does have a causal effect, it goes in the opposite direction to aging.  Because the study couldn’t look at change and couldn’t establish cause and effect, we certainly couldn’t say that drinking a little more seems to make your heart a little younger, but that’s the direction of some of these associations.  This complicated pattern makes it important to try to establish the association between direct health measures and alcohol consumption, and this study can’t do that.  The researchers do argue that the association that goes in the same direction as aging, a measure of the volume of part of the heart (the right atrium), is particularly important for health, but this study can’t show how that fits in with the other associations in affecting health.

“I think, overall, that it is certainly plausible that differences in alcohol consumption may correspond to differences in the structure of brains, hearts and aortas in a reasonably consistent way, and it’s possible that those associations are causal, even if this study can’t show that.  But the study doesn’t directly show what the effect of these structural changes on people’s health is, and re-examining the recommended alcohol limits surely ought to be based on reconsidering the sizes of health effects.  There are many ways in which I could change aspects of the structure of my body, by exercising differently or eating differently for example, and the important aspect for my health is not necessarily the details of how my structure changed, but what those changes might do to my health.  And, as with everything else in life, there needs to be a balance between the enjoyment that someone might get from an activity, such as drinking small or moderate amounts of alcohol, and the harms to health that might be caused.  People might, or might not, wish to cut down on alcohol consumption on the basis of health harms, but to decide on that, they have to know how big the health harms might be, and this study can’t tell them that.”


Prof David Curtis, Honorary Professor, UCL Genetics Institute, said:

“This study provides no evidence at all that moderate use of alcohol is harmful.  For one thing, it does not even include any non-drinkers so there is no information at all about whether people drinking moderate amounts are more or less healthy than those drinking nothing at all.

“There is no suggestion from that data that health problems increase with alcohol consumption below recommended limits.  The study authors have simply made the assumption that there is a log-linear relationship, meaning essentially that they have drawn a straight line through their data points.  Given that excess alcohol use causes problems, it is inevitable that it will appear as though the risk of problems increases gradually even at low levels but this is a simple consequence of trying to fit the data to a straight line.  No attempt was made to test alternative hypotheses, for example that there is a safe level of consumption not associated with harm.

“In fact, the linear-log model is used is obviously inappropriate.  For one thing, the log of zero is negative infinity, which would imply that somebody drinking zero or minimal amounts of alcohol would have, for example, negative amounts of fat in their liver.  This is clearly absurd.  Given that the linear-log model cannot be used for low levels of alcohol intake it makes no sense to use it try to make inferences about the effects of drinking small amounts of alcohol.

“The study provides no evidence at all for any association between alcohol intake and risk of Alzheimer’s disease.  I have no idea why this is even mentioned.

“It is well established that excess alcohol consumption is associated with important health risks.  It is important not to distract from this message.  There is no evidence that small quantities of alcohol are harmful in any way.  If they do have any impact, the effects are minimal.  Public health messaging should focus on important, well-established risk factors which, along with excess alcohol, would include smoking, lack of exercise, poor diet and obesity.”


Prof David Nutt, The Edmond J Safra Chair in Neuropsychopharmacology, Imperial College London, said:

“Very large empirical study with clear findings.  These are not surprising given almost all prior studies have revealed monotonic or exponential relationships between alcohol intake and organ damage.  These data further question the idea of a j-shaped relationship for cvs disease in men given the linear increase in cardiac damage.

“However I question their proposed solution of lowering safe drinking guidelines.  We know that reducing recommended levels of personal consumption have little impact on drinking behaviour – what is needed are new approaches.  These should include increased taxation on alcohol especially Minimum unit pricing, limiting availability e.g. reduce supermarket sales and encouraging alternative drinks that give the desired effects of alcohol with less harm – as suggested by the 2005 UK government Foresight review  See also e.g.”


Dr Sara Imarisio, Head of Research at Alzheimer’s Research UK, said:

“This study adds to existing research highlighting how drinking alcohol can have negative consequences for our health.  While some previous studies have suggested no health consequences or even possible benefits associated with moderate drinking, this study—which looked at the brain, the heart, and the liver—did not.

“The researchers found that any level of alcohol consumption was linked to potentially harmful changes in the body, so the less people drink the lower this impact will be.

“While this UK Biobank study involves many thousands of people, it relied on participants reporting their own alcohol consumption, which is not always reliable.  Heavy drinkers may under-report how much they drink, and this may have influenced the apparent effects of moderate drinking in this study.

“These types of study provide a snapshot in time of a person’s health and so it’s not possible to conclude how these changes in the body may go on to affect long-term health or risk of dementia or heart disease.  However, previous studies have linked conditions like stroke and heart disease to levels of alcohol lower than the current recommended guidelines.

“Further research is vital to fully understand the effects drinking alcohol can have on the brain and how it might affect the risk of diseases like Alzheimer’s.  As well as limiting alcohol consumption, existing evidence points to a number of other lifestyle changes that can people maintain a healthy brain for longer.  These include not smoking, eating a balanced diet, staying mentally and physically active and keeping blood pressure and cholesterol in check.  We must do all we can to help reduce the number of people who will go on to develop dementia in future and Alzheimer’s Research UK has launched the Think Brain Health campaign as an important first step,”



‘Alcohol consumption in the general population is associated with structural changes in multiple organ systems’ by Evangelos Evangelou et al. was published in eLife on Tuesday 1 June 2021.

DOI: 10.7554/eLife.65325



Declared interests

Prof John Holmes: “I provided advice and epidemiological modelling to the expert group that developed recommendations for the UK Chief Medical Officers’ low risk drinking guidelines.”

Prof Tim Spector: “No COI other than author of nutrition books and consultant to ZOE global ltd, a data science company.”

Dr Sadie Boniface: “I work at the Institute of Alcohol Studies which receives funding from the Alliance House Foundation.”

Prof Kevin McConway: “I am a Trustee of the SMC and a member of its Advisory Committee.  However, my quote above is in my capacity as an independent professional statistician.”

Prof David Curtis: “I have no conflict of interest.”

Prof David Nutt: “DJN declares an interest in that he is CSO in, a company that is developing less harmful alternatives to alcohol.  He is also author of the new book on alcohol – Drink?”

Dr Sara Imarisio: “Alzheimer’s Research UK is a founding funder of the UK Dementia Research Institute.”




in this section

filter RoundUps by year

search by tag