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

expert reaction to observational study of moderate drinking and cognitive decline

An observational study published in PLOS Medicine uses UK Biobank data to look at moderate alcohol consumption, brain iron, and cognition.


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

“Although the summary reflects the way in which the study was conducted and its main findings, it leaves several questions un-answered, with the risk of interpretations that could far over-reach those that can be concluded from the study. The research is highly sophisticated, particularly in the way in combines the use of recognised tests of cognitive function with measures to assess iron deposition in the liver and a range of brain areas known to underly changes in memory, problem-solving and movement. Is also uses well-established genetic markers for alcohol use disorders and changes in the body’s iron control to explore associations between alcohol consumption and cognitive function. The study design is robust in that it considers several alternative mechanisms by which alcohol intake could affect cognitive function through the iron deposition in the brain. It is also the first to add to our knowledge of how even moderate alcohol consumption can affect cognitive function through abnormally high levels of iron in the brain. Previous studies have not been able to draw firm conclusions. The study’s bottom line is that it shows positive associations between self-reported alcohol consumption of more than an average of 11 units per week – equivalent to just under 5 pints of beer and about 3 ½ large glasses of wine per week. 

“The main problem with interpreting the findings from this study are that the brain areas identified as having abnormal iron deposition are more commonly associated with changes in movement and coordination rather than changes in cognitive function. The authors also state that the cognitive tests were limited in scope and concerns have been raised about the reliability of the tests used.  

“The authors have attempted to adjust for other possible factors that could account for the association between alcohol intake, brain iron deposition and changes in cognitive function. However, although the study makes a case for implicating abnormal levels of iron deposition in the brain in people with drink at moderate levels, the suggestion that moderate drinking has a direct association with cognitive impairment with potential for the prevention of Alzheimer’s Disease and Parkinson’s Disease appears a little far-fetched. The more sophisticated genetic measures used, failed to show an association between alcohol-related brain iron deposition and changes in cognitive function. The brain areas showing abnormal iron deposition are also more closely associated with movement but did not show an association with measures of motor control such as walking speed and grip strength. As such, a direct association between moderate drinking and cognitive function associated with abnormal levels of iron in the brain does not appear strong. We should not hold our breath in the light of these findings to suggest that the prevention of alcohol related dementia and Alzheimer’s Disease can be achieved through interventions to reduce brain iron deposition.”


Dr Shanti Shanker, Deputy Head (SciTech) Ageing & Dementia Research Centre, Bournemouth University, said:

  1. Does the press release accurately reflect the science?

“It very accurately reflects the science and the analysis is robust. The results reported are very accurate and they do not have seemed to overexaggerate the findings. Which is really important in this kind of large databank analysis.”

  1. Is this good quality research?  Are the conclusions backed up by solid data?

“Yes, it is really good quality research and conclusions are solid, based on what was available in the data bank.”

  1. How does this work fit with the existing evidence? 

“This is one of the largest sample study looking at the links between AUD and cognitive decline. Giving the UKB (UK BioBank) cognitive tasks, it is useful to be able to see some evidence suggesting decline in some aspects of executive function tasks (like Trail making). More importantly the evidence between subjective reports of alcohol and the genetic links are critical for this research. They help add more evidence to understand the link between AUD and cognitive decline.”

  1. Does this observational study show that the increased iron is because of alcohol?  Is that plausible / already established?

“This study shows that there is an indirect effect of alcohol through systemic iron, and the direct effect of alcohol is linked to specifically the increase of left putamen susceptibility which affected some performance in a set of executive task (measured by Trail Making). It is plausible and provides a more robust link to the effect of alcohol and importantly, this study accounts for the large sample size and also that the models were adjusted for a range of factors, such as sex, age, etc.”

  1. Does the link to ‘brain changes and cognitive decline’ automatically follow?  Why/why not?

“There definitely seems to be some link to a region in the brain that gets affected by increase of iron (due to alcohol consumption) and that affects one aspect of cognition. It is important to note that in this study it links to one task of executive functions. So yes, it does affect one are, but it is important to not claim that it affects all cognition!”

  1. Have the authors accounted for confounders?  Are there any important limitations to be aware of?

“Yes, with limitations – one needs to avoid overgeneralisation of cognition decline  – the authors have reported this well in the paper and it’s important to ensure that the message sent out is clear and well drafted.”

  1. What are the implications in the real world?  Are people who consume >7 units of alcohol per week at increased risk of Alzheimer’s and Parkinson’s disease? 

“The implications from this publications do showcase that in one of the largest study  they explored the links between self-reported alcohol consumption and brain iron levels. They assessed the association between alcohol intake (with blood and liver iron) and cognitive measures with higher brain iron. Indicating that there is one area (left Putamen) which seems to be susceptible to the iron level and is associated with performance on one of the executive tasks (Trail making). There have been papers (in the past, where higher brain iron have been linked to Alzheimer’s and Parkinson’s disease – this paper clearly indicated that no participants had dementia diagnosis at the time of imaging and there is no report of any Parkinson’s disease. So we cannot say people who consume >7 units of alcohol per week are at the increased risk of AD or Parkinson’s as THIS study DOES NOT Make that direct link. And that there are other factors that contribute to an individual being diagnosed with AD and/or Parkinson’s.”

  1. How concerned should we be? 

“We should be concerned to the point that increased alcohol consumption can effect cognition. But not get worried about AD or Parkinson’s with this paper alone!”

  1. Should the unit guidelines be revised in light of this research?   

“Worth thinking based on it affects one executive function tasks, and supports the various published evidence that alcohol consumption negatively impact the brain – so yes, reducing the unit guidelines will allow lower alcohol consumption.”


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

“This study shows that people with higher alcohol consumption tend to have higher iron levels generally, as measured in the liver, but also in some regions of the brain. However it’s less clear that these iron levels actually have a harmful effect on cognitive function. A number of different analyses are carried out so that it is difficult to decide which, if any, of the associations between brain iron and cognitive function represent real, causative effects.”


Professor Paul Matthews, Centre Director at the UK Dementia Research Institute at Imperial College London, Edmond and Lily Safra Chair, NIHR Senior Investigator, and Head of the Department of Brain Sciences, said: 

“The importance of UK Biobank as an open resource cannot be overstated.  By making its data available to different researchers across the world, observations can be repeated to confirm and extend important findings.  

“An earlier report1 from the UK Dementia Research Institute showed that even moderate alcohol consumption by UK Biobank participants was associated with disease-related changes in the brain, heart and liver.

“Topiwala and her colleagues have thoughtfully extended this analysis with a specific focus on potential roles for abnormal iron deposition in causing the associated tissue injury.  

“The new report reinforces the uncomfortable message that there may not be any fully “safe” level of regular alcohol consumption.”

1 10.7554/eLife.65325


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

“Our brains are incredibly complex, responsible for our memories, as well as what we think, feel and do. By keeping our brains healthy as we age, we can help stave off diseases like Alzheimer’s which physically attack it, striking at the very essence of who we are.

“While there’s no definitively safe level of alcohol consumption, research into how smaller amounts of alcohol can contribute to poorer brain health is an active research area. In this large study researchers asked volunteers to self-report their alcohol intake, while looking at the levels of iron in volunteers’ bodies, including their brains. This study points to an increased amount of iron as one potential mechanism for poorer brain health related to alcohol consumption, however the study can’t say for sure that this relationship is causal.

“It’s not clear how a build-up of iron affects dementia risk, and Alzheimer’s Research UK is funding research at the University of Oxford to further our understanding of how changes in iron levels track with other brain changes that occur in diseases like Alzheimer’s.

“This study didn’t look at dementia, or alcohol as a risk factor for the condition, but there are many good health reasons to keep an eye on how much alcohol you’re drinking. As well as only drinking in moderation, staying physically, mentally, and socially active, eating a healthy balanced diet, not smoking, and keeping weight, cholesterol and blood pressure in check are all good ways to keep the brain healthy as we age.”


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

“This is a complicated study. It uses several different kinds of statistical analysis, some of them not particularly familiar. Though in my view the statistical methods are appropriate, and carried out well, the results from most of them need careful interpretation. But perhaps the main issue in making sense of all of this is that the researchers have investigated a complicated set of possible effects and changes, and there remain gaps between the different findings.

“Overall, though the study has thrown a certain amount of light on a complicated set of potentially interacting measurements, it couldn’t be in a position to sort out what actually causes what, or even sort out all aspects of the association between the quantities that were measured. And it certainly doesn’t show that drinking moderate amounts of alcohol is going to lead to major cognitive decline or dementia in later life. There might well be a connection between alcohol consumption and cognitive decline, and this study throws a certain amount of light on one way in which such a connection might or might not work (via iron in the brain). But despite generally being a good piece of research, inevitably it leaves a lot of questions unanswered.

“The researchers set out to investigate the possibility that drinking alcohol may affect the amount of iron present in some parts of the brain, and that in turn, greater amounts of iron in the brain may affect the decline of cognitive processes. So it’s important to understand that the researchers did not, in this study, look directly at associations between alcohol consumption and cognitive decline. They looked only at separate parts of the possible process – associations between alcohol consumption and measures of iron in the brain and elsewhere in the body, and associations between iron in the brain and findings on some tests of cognition.

“In terms of the cognitive measurements, it’s important to understand that the study did not anyway directly measure cognitive decline, even though the possible effects that they aimed to investigate involved cognitive decline. Many of the participants took the cognitive tests only once, and though some of them would have had cognitive data from more than one point in time, it seems that the researchers did not directly analyse changes in cognitive performance in individuals.

“Instead they looked at how quickly performance on these tests changed with the age of the participants. They found – and this is no surprise at all – that over a range of ages, older participants tended to perform rather worse than younger ones. That’s certainly not saying that the older participants had dementia – it’s just that cognitive performance does, on average get a bit worse with age, just as body strength, athletic performance, and so many other things do tend to decline as we age. So, if the way that the cognitive results are related to age is different in groups of people that drink different amounts of alcohol, that might possibly indicate that alcohol consumption affects cognitive performance. But there are other possible explanations. I’ll say more about that below, but one possibility is perhaps that, for some reason, those who were born in different decades would have had different cognitive performance anyway even if they had all done the cognitive tests at the same age. That would be an example of what’s known as a cohort effect, and these researchers do point out that some of the apparent changes with age could represent a cohort effect.

“But this is just a minor part of a broader difficulty with the study. The researchers had a hypothesis about changes over time – over time, drinking alcohol may lead to increases in the amount of iron in an individual’s brain and the rest of their body, and over time, this extra iron might lead that individual’s cognition to get worse than it otherwise would. But they investigated this using cross-sectional data, rather than directly measuring changes over time in individuals, and that makes the interpretation more awkward.

“Another general difficulty of interpretation is that the study is observational. There was no intervention from the researchers to change what the participants did – instead the participants just got on with their lives and the researchers asked them questions and carried out various measurements on them. The trouble with any observational study, and this is no exception, is that it’s very difficult to sort out what is causing what. So, for instance, when the researchers found that participants who drank more alcohol tended to have higher measurements of iron in some parts of their brain, that could be because the alcohol drinking had caused the higher iron levels. But there were many other differences between people who drank different amounts of alcohol, in all kinds of other factors, and those other differences could, in whole or in part, be the reasons for the differences in brain iron measurements.

“The researchers were, naturally, very aware of this issue. It’s possible to make statistical adjustments to try to allow for differences in those other factors, and the researchers did make many such adjustments. But that doesn’t remove the difficulty in deciding what causes what – one can never be sure that adjustments have been made for all the relevant factors, and in any case you can’t make adjustments for things that weren’t measured.

“As well as making the adjustments in their observational data analyses, the researchers, admirably, carried out an extra set of analyses in relation to some associations. They did what are known as mendelian randomisations. The idea here is to use genetic information to try to pick out the part of an association which can be said to be causal. Since people’s genetic make-up is fixed at conception, it won’t be affected by other factors that arise during their lives. An issue is that mendelian randomisation makes sense only if certain assumptions can be made about how all the various factors fit together, and what could affect what, and these assumptions can’t all be checked.

“The researchers used mendelian randomisation to investigate the association between alcohol consumption and brain iron, and between alcohol consumption and the general level of iron in the body. For various unavoidable technical reasons, the mendelian randomisations investigating alcohol consumption and brain iron provided only what the researchers describe as “weak evidence” that the association is one of cause and effect. It’s still very far from certain, therefore, that higher levels of alcohol consumption cause there to be more iron in the brain (though that’s one possibility). The mendelian randomisations investigating alcohol consumption and general levels of iron in the body seem to have been more successful, but even if that association is causal, it isn’t directly saying anything about alcohol and brain iron. Another analysis that was done (causal mediation analysis) showed that the association between alcohol consumption and brain iron has aspects that don’t seem to involve iron in other parts of the body. So, overall, the position in terms of cause and effect of alcohol and iron in the brain remains unclear, I’d say.

“The researchers did not carry out any mendelian randomisation to examine the association between brain iron levels and cognitive performance. Therefore the evidence, that they did find, of some associations between brain iron and cognitive performance comes entirely from observational data, and that makes it very difficult to conclude that differences in iron levels in the brain actually cause differences in cognitive performance.

“As the researchers explain, there are some issues about the way they measured iron in the brain, of two kinds. First, they measured the iron using MRI scanning, specifically using measures involving what’s known as magnetic susceptibility – and the researchers say that these measures can possibly also pick up other substances, not just iron. And incidentally, though the measurements are called ‘susceptibility’, this is a term from physics describing how easily something can be magnetised, and it has nothing to do with susceptibility to brain damage or to ill health. Second, the associations that were found between the brain iron measurements and performance on cognitive tests could be picking up an association between something else in the brain, that happens to be associated with iron levels, and cognitive performance. Even if there is cause and effect here, therefore, it doesn’t necessarily mean that the iron levels themselves are causing the cognitive levels.

“I’d have liked to finish this comment by giving some idea of the absolute size of the association between alcohol consumption and cognitive performance – but I can’t. First, in this study, the researchers mostly did not directly measure association between alcohol consumption and cognition. (They did, in what they describe as a ‘post hoc analysis’, find some association between alcohol and one aspect of cognitive performance, but this showed up only in a large sample of about 100,000 participants that did not have the MRI scans, so had no iron measurements, so the study provided essentially no evidence on what the causes of this association might be.)

“But, second, I can’t even give you a very useful measure of how big the association was between the brain iron measurements and the cognitive tests. That’s because the cognitive test results aren’t expressed in terms of everyday quantities, and, more importantly, because (for good statistical reasons) the researchers didn’t use the immediate results of the tests in their analyses anyway. It looks as if having a raised iron measurement in certain parts of the brain might correspond to having an average performance on one of the tests (of executive function) that matches the performance of an average person, who doesn’t have high brain iron but is just a few years older. But we don’t know for sure that any such difference is caused by the brain iron, and even if it is, we can’t be confident that the differences in brain iron are caused by alcohol consumption.”


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

“Alcohol can cause brain damage among people who drink at very heavy and dependent levels, including alcohol-related dementia and Wernicke-Korsakoff’s Syndrome. This study could be an important addition to our understanding of the relationship with alcohol and dementia at lower levels of alcohol consumption. Earlier studies have had mixed findings, and it is difficult to disentangle this relationship from other aspects of health and life circumstances.

“We already know of other health risks at ‘moderate’ levels of alcohol consumption, for example certain cancers, and this was taken into account in the 2016 alcohol guidelines review. The review considered what an acceptable level of risk was, and developed low risk drinking guidelines, not safe drinking guidelines. However research shows most people are not aware of the health risks of alcohol, something which could be improved by better labelling of products with health warnings.”



‘Associations between moderate alcohol consumption, brain iron, and cognition in UK Biobank participants: Observational and mendelian randomization analyses’ by Anya Topiwala et al. was published in PLOS Medicine at 7pm UK time on Thursday 14th July.




Declared interests

Dr Tony Rao: “Previous member of the Alcohol Health Alliance, current member of the Society for the Study of Addiction and have lobbied for a comprehensive government alcohol strategy. I am also Editor-in-Chief of the journal Advances in Dual Diagnosis and have published an editorial on alcohol and the brain with the lead author.”

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

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

Prof Paul Matthews is Chair of the UKRI MRC Neurosciences and Mental Health Board, a member of the UK Biobank Strategic Oversight Committee and Chair of the UK Biobank Imaging Working Group.  He is Director of the UK DRI Centre at Imperial College London.

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

For all other experts, no reply to our request for DOIs was received.



in this section

filter RoundUps by year

search by tag