March 22, 2021

expert reaction to analysis for meat consumption and dementia

A cohort study published in the American Journal of Clinical Nutrition looks at meat consumption and risk of incident dementia.

Prof Paul Matthews FMedSci, Director of the UK Dementia Research Institute at Imperial College London, said:

“Risk as defined in this study means association, but does not provide evidence that eating processed meat causes dementia.  Similarly, the association between eating unprocessed meat and a lower incidence of dementia in the UK Biobank population does not provide evidence that unprocessed meat protects people from dementia.  The authors themselves note that other factors, like the higher salt content in processed meat might explain their most notable observation. 

Nonetheless, the results highlight that lifestyle or environmental factors can influence the risk of developing dementia.  However, if we are to reduce risks by change aspects of lifestyle or the environment, it is important that we understand how specifically why they might affect disease.   Interventions for reducing disease risk must be targeted appropriately to avoid more harm than benefit.  It is premature to propose changing dietary recommendations on the basis of associations like these alone.”

Prof Susan Jebb, Professor of Diet and Population Health at the University of Oxford, said:

“This is another study using the UK Biobank prospective cohort study to analyse the relationship between some aspect of diet and later risks of particular conditions. It looks at associations between diet and dementia – not causes of dementia. It is well a conducted analysis which has tried hard to adjust for the potential confounding factors, though inevitably others will remain. But to interpret the implications of this analysis the findings need to be considered in context – beyond the headlines in the press release.

“First, there are relatively few cases of dementia – only 2896 people, representing 0.6% of the population in the study. Second, the data on meat intake and dementia risk are mixed.  Third, the mechanisms which might explain these mixed findings for different types of meat are unclear.

“Higher intake of processed meat was associated with an increased  risk. The ‘headline’ finding is that a 25g per day increase in processed meat is associated with a 44% increase in risk, but even if it is causal, a 44% increase in a very small number of cases is a small absolute risk; also, the average intake of processed meat in the UK from national survey data is about 25g per day and declining, so a 25g per day increase would represent a doubling of current intake.

“Conversely, eating unprocessed red meat appeared protective against dementia. People who consumed 50g of red meat a day were reported to have a 19% lower risk of dementia – but average intakes in the UK are only about 30g per day.   A wider examination of the health consequences of eating red meat suggests that increasing consumption would do more harm than good. A paper published earlier this month from the UK Biobank cohort reported on a wide range of adverse health outcomes associated with higher red meat intake¹.

“There was no association in either direction with total meat intake and dementia risk.

“This study confirms the strong genetic association with dementia. It showed that APOE e4 genotype alone is associated with a threefold increased chance of developing all-cause dementia, a much bigger effect that that attributed to processed meat.”

¹ https://pubmed.ncbi.nlm.nih.gov/33648505/

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

“It’s very important not to read too much into this study. The top line on the press release implies that eating processed meat could actually cause an increase in dementia risk. The study itself does come close to claiming that too. But that claim goes well beyond a reasonable interpretation of what the researchers found – as indeed they effectively make clear themselves by including a substantial list of limitations of their findings.

“Really the fundamental problem with this study, and with any study of this general kind, is that it can never establish what causes what – at least not on its own. The problem is that the study is observational. The researchers didn’t make people eat different amounts of processed meat – they simply recorded what they said they ate, followed them up for some time, recorded when any of them got a diagnosis of dementia, and looked whether the chance of getting a dementia diagnosis varied according to how much meat they said they ate. The trouble is that people who eat different amounts of processed meat also differ in many other ways apart from their meat consumption. As the researchers reported, those who ate more processed meat were more likely (for instance) to be smokers, to be less educated, overweight, and to eat less fruit and vegetables. Those things have been identified as risk factors for dementia in previous research. So, if people who eat more processed meat turn out to be more likely to be diagnosed with dementia, it could be because of these other differences in risk factors rather than because of their processed meat consumption. The researchers did realize that this is an issue, and carried out statistical adjustments to try to take account of these factors and others. However, the process of adjustment can never be perfect. In particular, you can’t adjust for a factor for which you do not have good data. The researchers report that people who ate more processed meat also consumed more calories, protein, and fat. Those things might possibly affect their dementia risk, but it seems that no adjustment was made for them in the main statistical analyses. I think that may be because they had data on these factors only for about a quarter of the people that they studied. And there may well be other factors that also varied with the amount of processed meat people ate, which the researchers couldn’t adjust for or even describe in their report, because they had no data on them.

“Because no single observational study of this kind can show whether the purported risk factor actually causes the risk to increase, it’s important to compare the findings with other research. The researchers on this study rightly made several comparisons of this sort, but they report that the findings from these studies are inconsistent. Some found no association between meat consumption and dementia,  and at least one found an apparent association that went the other way, that is, people who ate more meat had a lower risk of dementia. The researchers on the new study point out that a lot of different biological explanations have been put forward for how meat consumption might affect brain function, and that some of these potential explanations work in opposite directions. That could certainly explain the inconsistent results in previous studies, but it does raise the question of why the pattern of risk in the new study should somehow be more representative or reliable than the patterns in other studies. So it’s certainly not the case that this new study is adding to a lot of existing evidence that eating more processed meat could cause an increased risk of dementia. Neither this new study on its own, nor the whole pattern of previous findings, provides solid evidence that eating processed meat actually causes dementia risk to increase.

“These important limitations don’t tell us that eating more processed meat definitely doesn’t cause an increase in dementia risk, just that the study can’t tell us whether it does or not. Let’s suppose for the moment that the association is one of cause and effect, and see how big the effect might be. It sounds pretty dramatic to say that consuming 25g of processed meat a day, about one rasher of bacon a day, increases the risk of developing dementia by 44%. But what was the risk, in the context of this study? Yes, almost 500,000 people were followed up, but under 3,000 of them actually developed dementia during the study period. In other words, only about 6 in every thousand got a dementia diagnosis. Imagine a group of 1,000 people like those in the study. If we follow them up for 8 years, the average follow-up time in this study, 6 of them would get a dementia diagnosis. Imagine for now that it really is the processed meat eating that increases dementia risk by the amount found in this study, and now imagine that all the 1,000 people ate an extra 25g of bacon every day for the 8 years. (That’s a lot of bacon. For each individual it’s an extra 175 grams a week, and in this study the people in the analysis group with the highest processed meat consumption ate only 38g per week on average – and I’m asking you to imagine this going on every week for 8 years) The number who would get dementia would go up from 6 to 8. Actually there’s a bit of statistical uncertainty about that figure – it could be between 7 and 10. So all that extra bacon eating would increase the number of dementia cases in our group of 1,000 people by between 1 and 4 cases. That’s an increase, but a pretty small one in my view.

“The study can’t show that processed meat actually causes an increase in dementia risk, any more than it can show that eating non-processed red meat can cause dementia risk to decrease. The findings are interesting, but they don’t establish anything for certain.”

Prof Clive Ballard, Pro-Vice Chancellor and Executive Dean at the University of Exeter, said:

“Even though there are nearly half a million participants in biobank, less than 3000 people have developed dementia.  The power is therefore very limited to look at risk factors, and most of the results did not show significant links when adjusting for multiple testing.  Although this study will be important when we are in a position to combine the results from multiple studies, as the study is small for this type of work, we should definitely not over-interpret the results. 

“We should not assume from this research that one rasher of bacon a day increases your risk of dementia by 44% – it is simply impossible to demonstrate that in a study like this.”

Prof Robert Howard, Professor of Old Age Psychiatry at UCL, said:

“The study showed an association between consumption of processed meat (bacon, ham, sausages, meat pies, kebabs, burgers, chicken nuggets) and receiving a diagnosis of dementia over 8 years follow-up. It’s always important to remember when looking at studies like this that they can’t attribute causation and important confounding factors might account for apparent associations. For example, the authors found that male sex, less education and smoking were associated with both processed meat consumption and risk of developing dementia. Although the authors attempted to control for confounds, they can never do this completely because of limitations in the data and the likely presence of other factors that they have no information about.

“As a doctor who works clinically with people with dementia and conducts research into potential dementia treatments, the data wouldn’t persuade me to give up my breakfast bacon.”

‘Meat consumption and risk of incident dementia: cohort study of 493,888 UK Biobank participants’ by Huifeng Zhang et al. was published in the American Journal of Clinical Nutrition at 00.00 UK Time on Monday 22 March 2021.

doi: https://doi.org/10.1093/ajcn/nqab028

Declared interests

Prof Matthews: “I am on the Steering Committee of UK Biobank, Chair of the UK MRC Neurosciences and Mental Health Board and a consultant to Biogen and Novartis, both of whom are developing drugs intended to treat dementias.”

Prof Jebb: “I am co-Director of the Wellcome Trust funded LEAP (Livestock, Environment and People) programme.”

Prof 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 Howard:  “I have no relevant conflicts of interest.”

None others received.