select search filters
briefings
roundups & rapid reactions
before the headlines
Fiona fox's blog
Five reviews, published in the Annals of Internal Medicine, analyse previously published evidence on the relationship between red and processed meat consumption and human health.
Prof Nita Forouhi, MRC Epidemiology Unit, University of Cambridge, said:
“The authors confirmed a link between lower red or processed meat consumption and lower risk of many health endpoints, but their conclusion that a reduction in red/processed meat is not indicated for better health is surprising and incongruous with this finding.
“The reasons for the authors’ conclusion need scrutiny on at least three points in the interest of public health.
“First, they stated that the magnitude of the link is small. Is it? Take the link with type 2 diabetes, for instance. For the impact of reducing 3 servings of processed meat every week, they estimated 12 fewer cases (confidence range of 9 to 16 fewer cases) of diabetes per 1000 persons over 10 years. For a common condition such as type 2 diabetes, at a population and country level, that is not trivial. And by the time one sums this across the modest absolute risk reduction for several health conditions, the potential public health impact cannot be ignored.
“Second, they stated that the quality of the evidence is of low or very low grade and hence uncertain, in large part because nutritional research typically relies on observational studies. And this is related to their third point – lack of evidence for causality, since randomised controlled trials failed to find a link between red meat intake and health endpoints. The scientific community needs to acknowledge that doing clinical trials of specific food interventions (unlike pharmaceutical products) and following people up over long periods till disease or death occurs are simply not feasible. There are many reasons for this, including but not limited to the lack of masking or blinding to the diet type, failure to adhere to prescribed diet, loss to follow-up and prohibitive cost.
“In fact in their systematic review of randomised controlled trials (RCTs) they demonstrated this point themselves when they had to effectively base their RCT findings on a single study – the Women’s Health Initiative among postmenopausal American women, yet this was inappropriate and problematic. This RCT tested the effect of a low-fat dietary pattern, not of red/processed meat directly. The intervention group reduced the consumption of a range of fat-rich foods, not exclusively meat. So, the reduction of healthful fat-rich foods (some dairy products or plant oils or nuts) would make the size of health effect ascribed to reducing red/processed meat consumption seem smaller, and the consumption of red/processed meat was different only by 1.4 servings per week between the experimental and control groups.
“Since the best available evidence points in the direction of a modest benefit of reducing red and particularly processed meat intake, it makes good public health sense to promote reducing meat intake. Indeed the authors themselves stated in one of their reviews that “Our results from the evaluation of randomized trials do not support the recommendations in the United Kingdom, United States, or World Cancer Research Fund guidelines on red meat intake. One could argue, however, that neither do they seriously challenge those recommendations.” So with the precautionary principle why negate the existing recommendations? Further research, including results from the European EPIC study that have been published since the studies included in the current reviews, are already providing further evidence for the benefits of diets lower in red and processed meat. Further research should continue, but at present there seems no reason to negate the current guidelines.”
Dr Giota Mitrou, Director of Research at World Cancer Research Fund, said:
“The recommendation from the NutriRECS International Consortium for the public to continue their current level of consumption of red meat (which they state as currently estimated to be three to four portions a week) is not dissimilar to our own recommendation on red meat: that eating three or less portions of red meat a week is best for cancer prevention.
“The public could be put at risk if they interpret this new recommendation to mean we can continue eating as much red and processed meat as they like without increasing their risk of cancer. This is not the case. The message people need to hear is that we should be eating no more than three portions of red meat a week and eat little, if any, processed meat. We stand by our rigorous research of the last 30 years and urge the public to follow the current recommendations on red and processed meat.
“It is important to remember that consumption of red and processed meat is one component of our overall diet and exercise pattern and it’s unlikely that specific foods are important single factors in causing or protecting against cancer. Instead, different patterns of diet and physical activity throughout life combine to make you more or less susceptible to cancer.”
Prof Susan Jebb, Professor of Diet and Population Health, University of Oxford, said:
“These reviews broadly cover the same evidence previous groups have considered – there is nothing new which has specifically challenged previous evidence of a link between meat consumption and an increased risk of some types of cancer, especially colorectal cancer. It also shows some evidence of an increase in the risk of cardiovascular disease. But this group of researchers have concluded that the evidence does not warrant making dietary recommendations. I cannot agree with this interpretation.
“They note that the quality of evidence is weak – true – it’s very hard to conduct trials which give high quality evidence of causal links. These reviews found only two relevant trials and excluded data from one because of “implausibly large effect sizes”. So there is only one trial, the Women’s Health Initiative, which was not specifically focused on meat reduction and where the difference in intake between control and intervention groups was small. Most of the evidence is from cohort studies and overall there is a mostly consistent trend towards increased risk with higher consumption of meat (perhaps with the exception of stroke outcomes) but including an increased risk of premature mortality.
“The absolute increase in risk for an individual might be considered small. But a poor diet is largely the result of multiple small increases in risks rather than a single dietary factor which is particularly harmful. Even more important is that the impact of small increases in individual risk adds up to a large population impact, especially for a condition like colorectal cancer which is relatively common. Dietary recommendations are issued for the population as a whole and it seems surprising to me that these researchers conclude we should not caution people about the risks of consuming meat, particularly for those people whose consumption is above the average.
“Unusually these reviews also include a paper on public preferences regarding meat consumption. This reports that people who eat meat are generally unwilling to reduce their consumption. This is perhaps unsurprising – we eat what we like eating and most of us would like to continue doing so, whether it’s meat or any other part of our diet. The authors use this as further justification for not recommending change. However, on this basis we wouldn’t make many – if any – dietary recommendations. How many people want to eat less chocolate or fewer biscuits, or drink less alcohol either? This information serves to emphasise that dietary change is difficult to achieve because it isn’t always popular, but it is not a good justification for inaction given that a poor diet is the greatest modifiable risk factor for ill-health.
“But the major concern here is that the report states “The panel chose to exclusively focus on health outcomes because environmental and animal welfare concerns are very different issues that are challenging to integrate with health concerns, are possibly more societal than personal issues, and vary greatly in the extent to which people find them a priority.” That is true, integrating the multiple goals for our food system is complex, but it is a poor defence for ignoring evidence when devising dietary recommendations for populations.
“Livestock production is a major source of greenhouse gases, in some areas it strains scarce water resources, and can exacerbate soil erosion. For this Committee to issue dietary guidelines for meat consumption for individual health without considering this wider context seems remarkably reckless. Climate change is an existential threat to human life.”
Dr Giota Mitrou, Director of Research at World Cancer Research Fund, said:
“The recommendation from the NutriRECS International Consortium for the public to continue their current level of consumption of red meat (which they state as currently estimated to be three to four portions a week) is not dissimilar to our own recommendation on red meat: that eating three or less portions of red meat a week is best for cancer prevention.
“The public could be put at risk if they interpret this new recommendation to mean we can continue eating as much red and processed meat as they like without increasing their risk of cancer. This is not the case. The message people need to hear is that we should be eating no more than three portions of red meat a week and eat little, if any, processed meat. We stand by our rigorous research of the last 30 years and urge the public to follow the current recommendations on red and processed meat.
“It is important to remember that consumption of red and processed meat is one component of our overall diet and exercise pattern and it’s unlikely that specific foods are important single factors in causing or protecting against cancer. Instead, different patterns of diet and physical activity throughout life combine to make you more or less susceptible to cancer.”
Prof Tim Key, Professor of Epidemiology & Deputy Director of the Cancer Epidemiology Unit, University of Oxford, said:
“There’s substantial evidence that processed meat can cause bowel cancer – so much so that the World Health Organization has classified it as carcinogenic since 2015. Today’s new publication reports results essentially identical to the existing evidence, but describes the impact very differently, contradicting the general consensus among cancer research experts.
“The authors here have found the same evidence of an effect but they think it is so modest that it isn’t worth recommending we do anything about it. For processed meat and colorectal cancer the new estimate here equates to a 19% greater risk of colorectal cancer for every 50 g of processed meat eaten per day. The figure cited by IARC/WHO in 2015 was 18% greater, so very similar to the 19% here. This is not a large risk, but still contributes to substantial numbers of cancers because the exposure is widespread and the disease fairly common.
“Recent estimates suggest over 5,000 people in the UK develop bowel cancer due to the consumption of processed meat each year, which is why the government recommends that people keep their total intake of red and processed meat to no more than about 70 g per day.”
Prof David Spiegelhalter, Chair, Winton Centre for Risk and Evidence Communication, University of Cambridge, said:
“This rigorous, even ruthless, review does not find good evidence of important health benefits from reducing meat consumption. In fact it does not find any good evidence at all – all the studies are ranked as providing ‘low’ or ‘very low’ certainty. We might expect even more controversy when this group turn their attention to other ‘risky’ things we consume. In contrast to their conclusions about health, I think we can be confident that reduction in meat consumption would benefit the planet.”
Dr David Nunan, Senior Researcher at the Centre for Evidence Based Medicine, University of Oxford, said:
“The reviews in this series were performed by an experienced team of researchers, applying a well-respected and thorough methodology (GRADE) that they indeed created and that has been adopted by other guideline groups internationally. So they likely did a good job and the findings and judgements are likely to be robust and trustworthy.
“The major issues are that there are very few clinical trials and for the ones that exist, many don’t directly test higher versus lower red and processed meat consumption, and they are all too short to allow for enough data on outcomes of importance and so they report surrogate measures. The observational data is likely to be highly confounded and therefore means the findings from these sorts of studies are more uncertain. We are then left to infer from imperfect studies that produce uncertain data. For example, the differences in the amounts of meat eaten between higher and lower eaters was on average the equivalent of the size of one deck of playing cards in terms of steak. Then there are the issues with the generally poor quality of the methodologies for all the included study designs, again meaning we can’t put much certainty in the numbers that the reviews reports.
“As far as I can see, for the majority of important outcomes (death, cancer rates, heart disease), people choosing to eat less meat and those who are told to eat less meat or given less meat to eat (trials) have a small benefit, but this benefit might not be meaningful enough to impact on the population levels of these outcomes (e.g. 1 to 12 fewer events per 1000 persons over 8 to 17 years). For less important outcomes (blood pressure, cholesterol etc.), people who choose to eat less meat look like they have slightly better values. This is the same for people who are asked to eat less meat or given less meat to eat; but for both groups the amount of benefit doesn’t appear to reach the minimum for what would be considered meaningful.
“If you put all the evidence together, where the authors found a difference between higher and lower meat consumption that was less uncertain, it tended to favour less meat. But remember by ‘favouring’ we are talking about very small benefits. Where the authors found uncertain differences, it still tended to favour less meat. Only for a few outcomes did the effect estimates favour higher meat consumption, and again by ‘favour’ we are talking very small benefits.
“Overall, the science that tries to answer the question ‘What is the effect of red and processed meat consumption on health outcomes?’, at present, is weak and as such is open to different interpretations. I think it’s important to stress that the authors of these studies point out that their recommendations, whilst not in line with other groups, are based on similar findings to other reviews and in their own words “One could argue, however, that neither do they [their findings] seriously challenge those recommendations”. It appears this group have taken a direct interpretation of the data in that it is not good enough to support a reduction of red or processed meat below current consumption rates for health benefits; though the authors do not define this rate which itself is an important omission. Importantly, they do not recommend an increase in meat consumption beyond current consumption rates.
“Other guideline groups will have taken the view of one of caution, in that those most at risk of these health outcomes are likely to eat higher amounts of meat than the current average consumption rate, so a recommendation to reduce consumption will at best move more people to the average and if that also means some move from average to below average this is unlikely, for most, to lead to harm in terms of health outcomes. But again, that is if we believe the findings, which the authors of the current studies put little belief in. All this says nothing about individual risk, as even if we believe that at best 12 (out of 1000) people who consume slightly less red or processed meat will be saved from a bad health outcome, no one can ever predict if you will be one of those 12.”
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“All in all this is an extremely comprehensive piece of work. It starts from reviews of all the evidence that the researchers could find about associations between eating red and processed meat and serious health outcomes, such as cancers, heart disease, strokes, and type 2 diabetes, and also overall death rates. Then, in the light of that evidence, it makes new recommendations for individuals who eat red meat, processed or unprocessed. And you don’t need me to tell you that these recommendations don’t look much like the current recommendations from government agencies and other organisations in many countries.
“I’ll start with the reviews of the evidence. Inevitably these are based on statistical reasoning and calculations, and in my view the statistics is done well.
“There are actually five different reviews, and it may help if I spell out what each of them looked at. One puts together evidence from randomised trials where some people, chosen randomly from the participants, were allocated to eat diets, or to receive dietary advice, that led to them eating less red and processed meat than the other participants. Generally, randomised trials like these are better than other research methods at sorting out what causes what – in this case, whether eating less meat caused people to have better health in terms of whatever the trials was measuring. But the researchers found only 12 such trials (involving about 49,000 people). These are relatively small numbers, given the potential importance of the subject, and that may well be because they are difficult and expensive to perform. Then there were three different reviews that put together evidence from cohort studies. These are observational studies, where people aren’t allocated to diets by the researchers, but just eat what they would eat anyway. Their diets are recorded, and they are followed up to see whether and when they are diagnosed with, or die from, various diseases. Inevitably in such research, there are other differences between the people who eat different amounts of red and processed meat, apart from their meat consumption, and these other differences might be the real reason for any differences in health, rather than the meat consumption itself. Researchers can and do adjust their results to allow for such other differences, but only where they have relevant data. Also in studies like these, it’s not always easy or accurate to record what people eat. For these and other reasons, it’s pretty well impossible to decide what the true causes of differences in health actually is from a cohort study, and the evidence is generally of lower quality than that from randomised trials. But that said, the cohort studies reviewed here include several million participants. One of the reviews looks only at cases of, and deaths from, cancer. Another looks at overall mortality, and at other diseases (heart diseases, strokes, and diabetes, mainly). Those two put together evidence from studies that specifically looked at red and processed meat eating. The third review of cohort studies looks instead at research that compared different dietary patterns, such as Mediterranean-style diets or vegetarian diets, that tended to include different amounts of red and processed meat. The final review of evidence wasn’t directly about people’s health, but instead put together information from studies that had looked at people’s preferences and views on meat consumption.
“Reviews like this have been done before, but these ones have good features that certainly don’t always appear. They combine evidence from more studies than most previous reviews have done, but I don’t think that’s the most important point. They include clear assessments of the quality of evidence of the studies that they review using a standard set of criteria (the GRADE criteria). That needs to be distinguished from the numerical evidence of the strength of associations, calculated from meta-analyses. Those numerical measures take the statistical results of the individual studies at face value, and essentially just take into account the statistical uncertainty that arises because people vary. But the GRADE criteria of study quality assess how likely it is that each study is subject to important biases or other issues. Overall, the results in reviews using cohort studies are assessed as providing low-certainty evidence at best. You might expect that the quality would be better for the review of randomised trials, but in fact it isn’t in these reviews. Partly because it is difficult to run such trials, and partly because they were mostly not set up to look specifically at meat consumption, the evidence in the review of randomised trials is also classed as being of low or very low certainty. Depressingly, all this tends to indicate that after all these years and all these millions of research participants, we still don’t know much.
“Other good points about these new reviews are that, for each outcome (disease or cause of death) that they consider, they give a plain-language summary of what they found, and they also give estimates of the absolute risk difference. As just one example of this, the review of cohort studies on cancer outcomes gives the following results for incidence (new cases) of colorectal (bowel) cancer, in relation to processed meat consumption. It says that they found 15 cohort studies of this, involving over 1.6 million participants. As well as giving a figure for the relative risk of eating 3 fewer servings a week (0.93, so a reduction in risk of 7%), they report that, in 1000 people, the number who will have a bowel cancer diagnosis (before the age of 75) is 20 (based on global cancer rates), and if all the 1000 people eat three fewer servings of processed meat a week, then there would be 1 fewer diagnosis (or perhaps 2, because of the statistical uncertainty). The GRADE certainty of the evidence for this conclusion is low, because it comes from observational studies. In lay terms, they say, this means that “Reduction of processed meat intake may result in a very small decrease in colorectal cancer incidence.”
“In a way, I think, it is these absolute risk differences and their interpretations that make the overall conclusions and guidelines from this research different from what we’ve usually seen up to now. I chose bowel cancer incidence, in relation to processed meat consumption, because the possible effects of eating processed meat such as bacon on bowel cancer rates seem rarely to have been out of the media for several years. Reports saying that processed meat causes cancer, and that we should eat less of it, have appeared repeatedly. Many of those report were based on a 2010 review from the World Cancer Research Fund (WCRF) saying that eating an extra 50g of processed meat, about two slices of bacon, increases bowel cancer risk by 18%. In a later (2015) review, they revised this downwards slightly to 16%. That’s a relative risk. The statistician Sir David Spiegelhalter pointed out at the time that, out of 100 people in the UK, eating all that extra bacon would put up the number out of 100 who would get bowel cancer in their life from 5 to 6, and that people might well find that acceptable – but that hasn’t stopped the WCRF, Cancer Research UK, and many other bodies recommending quite strongly that we should eat less processed meat. These relative risk figures are in fact pretty similar to what was found in the new research. The comparison isn’t obvious, partly because the WCRF figures are for 50g extra each day, and the figures from the new research are for three servings (of 50g) fewer each week. But translating the figures from the new research into the same terms as WCRF, they correspond to an relative risk of 18%, which is the same as WCRF’s 2010 figure and bigger than their 2015 revised figure. The difference is that the new research describes this as “Reduction of processed meat intake may result in a very small decrease in colorectal cancer incidence,” whereas WCRF, CRUK and others presumably think that this decrease is large enough to matter and to affect their recommendations. Perhaps the difference is that some would consider that any statistically significant evidence should affect guidelines, even if the absolute change in risk is small and the quality of the evidence is not high, whereas the authors of these new papers take a different view. My own sympathies are with the line taken by the new researchers, but not everyone will agree.
“Note that, in fact, there’s rather more evidence in relation to processed meat consumption and bowel cancer than there is for most of the diseases considered in the new research. In many other cases, I think the conclusions of the new research (that the researchers are uncertain about any effects, of that if there are any effects, they are very small) would be less controversial.
“It’s worth mentioning, I think, that the new guidelines, which (in very brief summary) suggest that adults aged 18 or over should continue their current consumption of processed or unprocessed red meat, are said to be ‘weak recommendations’ based on ‘low-certainty evidence’. And they were voted for by 11 of the 14 panellists involved, while the other 3 voted for weak recommendations to reduce consumption. The fact that the majority group and the dissenters all described their preferred recommendations as ‘weak’ would again tend to indicate that, despite all the research, we really don’t know much about this. I’d add that research on these topics seems to have so many difficulties that it’s unclear to me personally whether we’ll ever know a lot more, at least not in the near future. Also, the recommendations are based only on the health consequences of eating red and processed meat. Eating animals might not have much effect on your health, but it sure doesn’t do much for the health of the animals being eaten. The guidelines document makes it explicit that no account was taken of animal welfare or environmental issues.”
Dr Marco Springmann, Senior Researcher on Environmental Sustainability and Public Health, Oxford Martin Programme on the Future of Food, University of Oxford, said:
“The recommendation that adults continue current red and processed meat consumption is based on a skewed reading and presentation of the scientific evidence.
“By presenting the evidence for a change in consumption that is less than half of what is customary (for a change of less than half a serving a day compared to a change of one serving per day as is customarily used), it was perhaps inevitable that the authors would report only small potential health benefits of reductions in red and processed meat consumption.
“Even with this skewed way of presenting the evidence, the reviews clearly indicate the benefits of reducing red and processed meat consumption.
“The presented evidence from cohort studies shows that even small reductions in red and processed meat were associated with statistically significant reductions in mortality from all major diet-related diseases, including overall cancer mortality, cardiovascular-disease mortality, and, for processed meat, even all-cause mortality.
“The authors interpreted the statistically significant health benefits from reduced red and processed meat consumption as small, because they focused on small changes in consumption.
“The review of randomised trials was dominated by a trial that was not designed for analysing the effects of changes in red meat consumption and whose design and findings have long been a topic of debate, not on meat consumption, but on total fat intake.
“The recommendations also took into account a review of what the authors called people’s preferences related to meat consumption. However, the review of preferences was based on a small number of unrepresentative studies from high-income countries. This raises further problems. Recommendations that reflect the preferences expressed by a small number of people from high-income countries can hardly be judged as scientific.
“Finally, the fact the authors didn’t take into account the environmental impacts of red and processed meat consumption is extremely short-sighted in light of its disproportionate impact on climate change. The production of red meat is responsible for about two thirds of all food-related greenhouse gas emissions and is therefore a major driver of climate change. The meat sector in general is also the biggest driver of deforestation and a major polluter of oceans through its run-off. Even with rapid decarbonisation and technological improvements, projections indicate that by the end of the century, the emissions caused by meat and dairy production could take up the complete emissions budget we have for limiting global warming to below 2 degrees Celsius. Without significant reductions in red and processed meat consumption, there is little chance to avoid dangerous levels of climate change.
“In sum, the recommendations are dangerously misguided. The recommendations downplay the scientific evidence, and are based, in part, on the preferences expressed by a small number of meat-eating individuals from high-income countries, whilst ignoring a large body of evidence that highlights the global importance of reducing red and processed meat consumption to avoid dangerous levels of climate change, resource depletion and environmental degradation.”
Dr Ian Johnson, Nutrition researcher and Emeritus Fellow, Quadram Institute Bioscience, said:
“This is a comprehensive and meticulous analysis of virtually all the available evidence on the relationship between meat consumption and human health. The authors have applied very rigorous criteria to assess the quality of the evidence from human trials. Because of the difficulty of conducting long-term experiments on complex human diets, and of measuring peoples’ dietary habits over long periods of time, the evidence from human nutrition research is often thought to be less reliable than that obtained in other branches of biomedical science.
“Overall, the results of this major review are reasonably consistent with previous studies, but the dietary recommendations derived from the review are not. This is because the authors conclude that both the rigorous criteria they have applied when assessing the quality of the evidence, and the generally rather small adverse health effects of meat consumption that have been detected, tend to reduce the confidence with which any recommendations for the overall health benefits of reducing meat consumption can be made. It is worth noting that the final conclusion of the review panel, that people need not reduce their current meat consumption, was not entirely unanimous. Three of the fourteen panel members did favour a “weak” recommendation to reduce red meat intake.
“There are strong environmental and ethical arguments for reducing meat consumption in the modern world. In my opinion, people who choose to do so can still reasonably believe that they will experience modest reductions in their risks of cardiovascular diseases and cancers over a lifetime. On the other hand this study will, I hope, help to eliminate the incorrect impression sometimes given that some meat products are as carcinogenic as cigarette smoke, and to discourage dramatic media headlines claiming that ‘bacon is killing us’.”
Prof Gunter Kuhnle, Professor of Nutrition and Food Science, University of Reading, said:
“This is a very interesting set of publications of very good quality that investigate the effect of meat intake on health. The key limitation of these studies is that they use very narrow terms of reference which downgrade a large number of studies. Due to my expertise, I will limit my comments to the associations between meat and cancer. The studies are systematic reviews and meta-analyses, combining all the evidence currently available.
Is this good quality research? Are the conclusions backed up by solid data?
“This is high quality research and the conclusions are backed up by the data. The key limitation of these kind of studies is the availability of data and the terms of reference used. The authors point out that a lot of studies that provide evidence are of low quality, and this affects their recommendations. In the context of cancer, in particular colorectal cancer, this is an important limitation: colorectal cancer develops over a long time, and is therefore difficult to investigate in dietary intervention studies. This means that evidence has to be based mainly on observational studies, which were rated as being of low quality due to their methodology (difficulty in assessing intake, confounding etc). This study therefore highlights very well the challenges of investigating the link between diet and long-term diseases. It is therefore not surprising that the outcome of this study disagrees with recommendations of other organisations, such as Cancer Research UK, the WCRF or the NHS, as they use a wider reference base.
How does this work fit with the existing evidence?
“The study summarises the current evidence available, but uses different methods to assess its quality. It has been known for some time that red and processed meat intake increases the risk of cancer, but also that this increase is rather modest (about 1 additional case per 100 people for colorectal cancer). In this study, the authors estimate that reducing meat intake would result in about 10 fewer people dying early due to cancer out of every 1,000 that die of cancer. Cancer Research UK’s general cancer stats say that there are 164,000 cancer deaths in the UK in total, so if the authors’ estimate is correct this would mean there would be around 1,640 fewer deaths if red and processed meat consumption is reduced. The authors conclude that the quality of the evidence available is insufficient to recommend a reduction in meat intake. Reviews by other organisations, such as for example the World Cancer Research Fund (WCRF) disagree, and based on their review they recommend a reduction in red and processed meat intake (about 350-500 g /week of red meat and very little processed meat). The main reason for the difference here is that here the authors downgrade the importance of observational data, because of its limitations. However, observational data, in combination with mechanistic data (which are not part of the review), clearly establish a link between meat intake and cancer.
“The results of this study highlight in a nice way the difference between hazard and risk. While IARC/WHO has categorised processed and red meat as carcinogens or probable carcinogens, i.e. a hazard to health, this study shows that the actual exposure to red and processed meat is for many people sufficiently small not to be of concern. This does not however mean that there is no risk associated with increasing intakes.
Have the authors accounted for confounders? Are there important limitations to be aware of?
“The key limitations of the study are:
What are the implications in the real world? Is there any overspeculation?
“The authors recommend to continue eating red and processed meat and do not recommend any change, although there was a minority opinion that suggested a weak recommendation to reduce red and processed meat intake. There is quite convincing mechanistic and other data, assessed by the WHO/IARC, showing that red and processed meat can increase the risk of cancer – these data here show (and confirm) that the impact on a population level is rather small, and also that it is generally difficult for people to change their diet. The data clearly show that the while the association between meat and cancer does not have to be addressed urgently, it should not be ignored. Small dietary changes can mitigate the effect of red and processed meat on cancer risk, for example a high fibre diet. Alternative approaches that are yet in their infancy are to include fibre or plant extracts in meat products in order to reduce the risk of cancer. The data however also shows that simply removing nitrite, a compound that has been seen as driving the association between meat and cancer, is unlikely to have a strong effect.”
Comments from the New Zealand SMC:
Prof Rod Jackson, Professor of Epidemiology, University of Auckland, comments:
“I’ve reviewed much of the evidence covered in these papers over the years and have come to the conclusion that it is impossible to undertake a useful medium- to long-term (more than about 6 months) randomised controlled trial or cohort study to assess the effects of common foods like meat, vegetables or dairy products, or common nutrients like fats, proteins, or carbohydrates, on ‘hard’ outcomes like coronary disease, cancer or death.
“The implications of this conclusion are three-fold:
1. We have spent hundreds of millions of dollar on studies incapable of giving us useful information because of inherent biases in medium- to long-term nutrition studies that are almost impossible to deal with. This is not a criticism of the researchers’ ability to design studies, although they should have realised by now that the studies can’t be done well.
2. The misinformation given to the public, based on the result of these two types of seriously flawed study designs (for examining the medium- to long-term effects of common foods and nutrients), have led to huge confusion and likely harm. This includes the conclusions of these latest articles.
3. We need to re-educate many of the researchers who write these papers and the ‘experts’ who write so-called evidence based guidelines that they have to radically rethink what is considered acceptable evidence. These study designs are usually considered the gold standard study designs and many researchers and most guideline writers have yet to appreciate that they are next to useless.
“The key bias in randomised controlled trials is cross-over between intervention and control groups. Not surprisingly, it has proven impossible to keep different groups on the diets they are randomised to for more than a month or two. The reason this is not surprising is that in randomised trials of a once a day drug versus an identical placebo for a couple of years we are very lucky if we get a 60-70% adherence rate. So how anyone can assume one could achieve anything like even a 50% adherence rate is beyond me.
“Also, you can’t measure adherence because most people cannot accurately remember what they have been eating and tend to report what the researchers want to hear so we overestimate adherence.
“As a result most randomised trials of diet on hard outcomes show minimal or no effects on outcomes, because the comparison groups gradually converge to have quite similar diets. That’s why the current studies report small or negligible effects.
“The cohort studies are even worse. Firstly, questionnaires used to divide people into different baseline groups based on their diets are notoriously inaccurate so the studies start off with a major bias – the groups’ diets are probably not as different as the researchers think they are. Secondly the different baseline groups always differ in other ways over and above their diets and these other factors commonly cloud or exaggerate any real effects of diet. This is called confounding. Thirdly, long term cohort studies also suffer from the same cross-over problem that randomised trials suffer from. As they are generally longer than randomised trials, cross-over can be even worse.
“I hope this explains why I don’t think these new studies reported in this journal are meaningful.
“What we need to do instead is to bring together ALL the other evidence, including: from short-term randomised studies that are short enough to limit cross-over but can also only measure proxy outcomes like blood lipid levels or blood pressure levels; from ecological studies of whole populations; from biochemistry; from pathology and from long-term drug trials (e.g. with statins) etc. This is messy and requires people with serious expertise and experience, but evidence is messy. It doesn’t matter how many meta-analyses of randomised trials and cohort studies are done and how many millions of people are included. They are still seriously flawed.”
Prof Jim Mann, Professor of Medicine and Human Nutrition and co-director, Edgar Diabetes and Obesity Research Centre, University of Otago; Director, Healthier Lives National Science Challenge, comments:
“In my opinion the ‘weak recommendations’ based on ‘low certainty’ evidence that adults ‘continue current consumption of unprocessed red meat and processed meat’ are potentially unhelpful and could be misleading.
“The recommendations were based on a majority decision of a 14-member panel appointed by a core leadership team. While I do not question the integrity, expertise and good intentions of the leadership team and panel, I have some reservations about the ‘guidelines’ which they have issued.
“[The panel’s] recommendations for red meat intake by individuals is not appreciably different from that made by other organisations (e.g. World Cancer Research Fund (WCRF)) given that current consumption in many countries is around 3 to 4 portions per week. It should be noted that recommendations to limit intake of red meat (in terms of effects on human health) are principally based on the relationship between red meat and colorectal cancer.
“The panel opted to consider personal preferences along with cancer and cardiovascular outcomes but not to take into account environmental and animal welfare issues when making their recommendations. In my opinion it is irresponsible not to consider sustainability and planetary health (a key, if not the major, determinant of the health of future generations) when developing nutrient and food-based dietary guidelines.
“The panel appears to have relied exclusively on the GRADE approach for assessing quality of evidence and guideline development. While this approach is appropriate and widely used (including by the WHO) there are some situations where additional methods are needed. GRADE quality assessment is based on a hierarchy of study types which are used to examine the relationship between an exposure – in this situation a particular food – and disease outcome or mortality. For evidence to be classified as being of high quality or high certainty, it must generally be derived from high quality randomised controlled trials (RCTs). In some situations it is not possible to undertake RCTs, and other approaches – such as experimental studies – which are not sanctioned by GRADE, are essential to complement the epidemiological studies included in the GRADE hierarchy. The conclusion of the International Agency for Research on Cancer that processed meat is carcinogenic to humans is based on experimental (i.e. not considered by the Panel) as well as other types of evidence.
“I have not been able to review all the papers which underpin the guidelines but the article ‘Effect of lower versus higher intake of red meat intake on cardiometabolic and cancer outcomes’ (Zeraatkar and colleagues) provides an illustration of my concern. The paper is a systematic review of randomised trials which have examined the effects of lower versus higher intakes of red meat on relevant cardiometabolic and cancer endpoints. It concludes: ‘Low- to very-low-certainty evidence suggests that diets restricted in red meat may have little or no effect on major cardiometabolic outcomes and cancer mortality and incidence.’ The 12 trials included in the review were required to involve a comparison of groups which differed by a gradient of at least one serving of red meat per week for 6 months or more. The potential difference between the two groups – which could be as little as one serving of meat per week – may be too small and the duration of the trial potentially so short that there would be little or no chance of detecting any adverse effect of red meat, should this exist. Any risk (especially cancer) associated with a dietary attribute is likely to develop over a period of years, if not a life time, and a short term, partial reduction in exposure will almost certainly also not show benefit. While the authors acknowledge the limitations of their approach, the low quality of the evidence and their conclusions are tentative, I am not persuaded that it is even appropriate to suggest on the basis of these data that red meat may have little or no effect on disease incidence and mortality.
“It should be noted that the group does not represent any national or international organisation or government. Guidelines are generally issued by authoritative bodies rather than self-selected groups. While the bona fides and expertise of the leadership team and panel are not questioned, I would dispute the criticisms which appear to be levelled at all other groups which have suggested guidelines in the past.”
* Five reviews:
‘Effect of Lower Versus Higher Red Meat Intake on Cardiometabolic and Cancer Outcomes: A Systematic Review of Randomized Trials’ by Dena Zeraatkar et al. was published in the Annals of Internal Medicine at 22:00 BST on Monday 30 September 2019.
‘Health-Related Values and Preferences Regarding Meat Consumption: A Mixed-Methods Systematic Review’ by Claudia Valli et al. was published in the Annals of Internal Medicine at 22:00 BST on Monday 30 September 2019.
‘Patterns of Red and Processed Meat Consumption and Risk for Cardiometabolic and Cancer Outcomes: A Systematic Review and Meta-analysis of Cohort Studies’ by Robin W.M. Vernooij et al. was published in the Annals of Internal Medicine at 22:00 BST on Monday 30 September 2019.
‘Reduction of Red and Processed Meat Intake and Cancer Mortality and Incidence: A Systematic Review and Meta-analysis of Cohort Studies’ by Mi Ah Han et al. was published in the Annals of Internal Medicine at 22:00 BST on Monday 30 September 2019.
‘Red and Processed Meat Consumption and Risk for All-Cause Mortality and Cardiometabolic Outcomes: A Systematic Review and Meta-analysis of Cohort Studies’ by Dena Zeraatkar et al. was published in the Annals of Internal Medicine at 22:00 BST on Monday 30 September 2019.
Clinical guidance:
‘Unprocessed Red Meat and Processed Meat Consumption: Dietary Guideline Recommendations’ by Bradley C. Johnston et al. was published in the Annals of Internal Medicine at 22:00 BST on Monday 30 September 2019.
Editorial:
‘Meat Consumption and Health: Food for Thought’ by Aaron E. Carroll and Tiffany S. Doherty was published in the Annals of Internal Medicine at 22:00 BST on Monday 30 September 2019.
Declared interests
Prof Nita Forouhi: “I am a member of the Joint SACN/NHS-England/Diabetes-UK Working Group on ‘lower carbohydrate diets compared to current government advice for adults with type 2 diabetes’. Views expressed are my own, not the Group’s.”
Prof Susan Jebb: Susan Jebb is co-Director of the Livestock, Environment and People (LEAP) project; a research grant from the Wellcome Trust to the University of Oxford.
Prof Tim Key: “I have no interests to declare.”
Prof David Spiegelhalter: “No COIs.”
Dr David Nunan: Dr Nunan is a member of the Royal College of General Practitioners (RCGP) steering committee to support the new Physical Activity and Lifestyle clinical priority. He has received funding for research from the NHS National Institute for Health Research School for Primary Care Research (NIHR SPCR) and the RCGP for independent research projects related to physical activity and dietary interventions. The views expressed are those of the author and not necessarily those of the NHS, the NIHR, the RCGP or the Department of Health.
Prof Kevin McConway: Prof McConway is a member of the SMC Advisory Committee, but his quote above is in his capacity as a professional statistician.
Dr Marco Springmann: “I am funded by the Wellcome Trust for a project called ‘Livestock, Environment and People (LEAP)’. Before that, I contributed to the EAT-Lancet report on Healthy Diets from Sustainable Food Systems for which I received a small honorarium that technically also came from Wellcome who support of the report.”
Dr Ian Johnson: “No conflict of interest.”
Prof Gunter Kuhnle: “I don’t have any conflicts to declare here.”
Prof Rod Jackson: Prof Jackson has been a full time university employee for almost 30 years (and 10 years before that as a publicly funded research fellow). All his research funding has come from public good sources like Health Research Council, Heart Foundation, MBIE.
Prof Jim Mann: Prof Mann is a member of WHO Nutrition Guidance Advisory Group and a former member of the World Cancer Research Fund panel which developed the Second Report on Diet, Physical Activity and Cancer.
None others received.