The Annals of Internal Medicine has just published a cluster of systematic reviews and meta-analyses, all in the same issue, on the topic of eating meat and processed meat. All of these research papers show, consistently and with statistical significance, harms of eating more rather than less meat and processed meat. The specific harms these papers tabulate (see Table 1) include all-cause mortality, cardiovascular disease, cancer, and diabetes. Eating more meat and processed meat contribute to all of these- so say these research papers just out in the Annals, and sure to dominate the health news for some time (and likely doing so already).
Table 1: A summary data table from the research papers in the Annals of Internal Medicine, compiled by Frank Hu, MD, PhD, Chair of the Department of Nutrition at the Harvard School of Public Health. The tabulated data show consistent, statistically significant, clinically important increases in all-cause mortality, heart disease, cancer, and diabetes with higher intakes of red and processed meat.
Why would such a dull, well-established, “me too” finding dominate the health news? You and everyone else has surely heard already that to some extent red meat, and to a greater extent processed meat, is bad for your health- to say nothing of its calamitous effects on the planet.
Because the Annals made the unusual if not unprecedented decision to publish multiple papers from the same group of authors in a single issue, all but turning control of the journal over to them, and producing a “theme issue” without actually saying so. This large group of researchers is led by people with careers devoted to certain ways of pooling and scoring data from prior studies- statisticians, effectively; with almost no one with expertise in nutrition and health outcomes in the mix. The aim of this group was not to reaffirm that eating more meat produces more disease, although indeed their data show exactly that, in alignment with all the prior data. Their aim was to say that the evidence is weak.
We Don’t Trust What We Found, So We Recommend the Opposite
Accordingly, they scored their own data using the methods they prefer (there are other methods, better suited)- and declared their findings very weak. They then proceeded to publish- along with their research papers- so-called “guidelines” recommending the opposite of what they found. In other words, the conclusion they claimed to draw from their own work was: everyone should just go ahead and keep eating meat and processed meat. That is why the health news will be abuzz with this tale now, and for some time.
The Annals of Internal Medicine brought these pending publications to the attention of health journalists and content experts in advance under terms of an embargo. I was among these recipients, as were many of my close colleagues. We all received a press release from the Annals with the following headline:
“New guidelines: No need to reduce red or processed meat consumption for good health. A rigorous series of reviews of the evidence found little to no health benefits for reducing red or processed meat consumption.”
This was clickbait, and it worked as intended: we all clicked. Colleagues and I who have scrutinized the relevant data on diet and health, read the pertinent studies, and contributed to them with our own publications over entire careers were taken aback by the claim of findings at odds with so monumental and consistent a body of research.
So, we all dove in and quickly discovered what I state above: all of the new papers did in fact show exactly the expected harms of eating meat and processed meat, even despite all the ways the statisticians apparently biased their efforts against that conclusion (more on that below). But the Annals was nonetheless allowing them to publish “guidelines” opposing their own findings because they were able to call their own findings “very low certainty.”
An Ounce of Prevention
None of the colleagues with whom I have been corresponding all week- a who’s who in public health and nutrition- had ever seen a travesty of science quite like this before, nor one with as much potential to do harm. My friend, Dr. Walter Willett, said for public airing: "These papers and conclusions represent the most egregious abuse of evidence that I have seen." Accordingly, groups of leading scientists in nutrition and public health- one that I pulled together, and others- took the unprecedented step of asking the Editor in Chief and editorial office at the Annals to withhold publication of these papers pending a full review of related concerns (see our letter here). That I am here with this tale now means that we failed to persuade.
Why did we take that extraordinary step? We would certainly NOT have done so if new papers actually DID show lack of harm from eating processed meat. That would have surprised us, and we would have needed to figure out what led to such anomalous findings. That’s how science works: we produce the results we produce, then work to understand them. There is no guarantee that our results will be entirely consistent, and the truth, and genuine understanding, must be evinced by working through such inconsistencies, not covering them over.
But that kind of inconsistency is not what is going on here. The statisticians’ findings are entirely consistent with what we already knew to be true, but they are issuing guidelines that recommend the opposite anyway, on the basis of their own contrived technicality: they deem their own findings highly uncertain.
Many of my colleagues, and certainly I, saw this for an assault on public health, and public understanding. The nuance of WHY there were new guidelines encouraging meat and processed meat intake gracing the rarefied pages of the Annals of Internal Medicine would surely be lost in the hurly-burly of press reporting and social media. The original clickbait headline issued by the Annals itself would surely be transmogrified into far worse in a media feeding frenzy. We knew this, and we understood the consequences.
Distrust, Disgust, and Lost Opportunity
It conveys the notion that there is scientific dissent on so settled a topic as the harms of processed meat. It suggests that “the experts” can’t agree, and don’t know what’s what. It is prone, inevitably, to foment distrust in nutrition, science, and expertise- and even spill over into overt disgust for us so-called “experts” and all of our inconsistencies.
Except that in this case, there weren’t any inconsistencies in the findings in the first place. These statisticians, the very ones doing this damage to understanding, found the same as the rest of us: eating more meat and processed meat will harm you, to say nothing of the planet.
Since they published guidelines at odds with their own findings, why didn’t they just publish guidelines based on their native, a priori inclinations- and skip all the work of the systematic reviews, the results of which they effectively ignored? Presumably the Annals could not justify letting a self-established group publish alternative meat-eating “guidelines” for public consumption based on… nothing. So, the 5 systematic reviews and meta-analyses appear to be the window dressing required to justify the one point the authors were so committed to making they made it in opposition to their own findings: go ahead and keep eating meat and processed meat.
Don’t. Not if you would like to reduce your risk of dying prematurely from any cause. Not if you would like to avoid heart disease. Not if you would like to avoid cancer. Not if you would like to avoid diabetes. And not if you care at all about the fate of the planet and all life currently on it.
That’s the gist of this sad tale, and the take-away message for those wondering where it leaves us. If that satisfies your appetite for information on the topic, push back from the table here. Thanks for coming, and digest well – you’re done. For anyone still hungry - I can serve up a bit more.
The Fine Print
Although conducting allegedly “systematic” reviews, the researchers here left out some obviously qualified, prominent trials well known to me and my colleagues, many with large effect sizes (such as this, and this, and this - to name only a few). We can discern no valid reason for the many such omissions and they belie the claim that these reviews were legitimately “systematic.” We might hope that these statisticians whatever the contortions in their conclusion might at least be faithful to their methods, but as best we can tell- they were not. And to be clear, I – and many of my colleagues reaching the same conclusion- know these methods well; we have published systematic reviews and meta-analyses of our own.
In the one case where these authors included a trial with a large effect size (The Lyon Diet Heart Study), they excluded it from their data analysis not because of any methodologic shortcoming but because of, by their own direct admission, its large effect size. Declaration in advance of what constitutes a “plausible” effect size in a research effort aimed at determining effect sizes is...highly irregular, at best.
The researchers then divided their retrieved evidence, perhaps also in the service of proactive conclusions, into observational and interventional silos, analyzed separately. This is not necessary even with GRADE, the method they used to score evidence, to say nothing of alternative strength of evidence (SoE) tools far better suited to this particular domain. This approach produced multiple publications where there might have been one, and reduced cell sizes, attenuating power to find strong effects with tight confidence intervals.
The entire analysis was predicated on trials which for the most part achieved a very tiny difference in the exposure variable. In the review of randomized trials, for instance, groups differed by just 1.4 meat servings per week. One is very unlikely to produce large outcome differences based on tiny exposure differences. This problem was compounded by nearly complete inattention to what was replacing meat among those eating a bit less; it could have been other meat (e.g., fried chicken replacing hamburger), ultraprocessed food, or anything else. That there is more than one way to eat badly and that certain such substitutions produce the same bad outcomes is well established.
Despite all of that- this group nonetheless found entirely consistent, clinically meaningful, statistically significant adverse effects of eating more meat and processed meat on all-cause mortality, on cardiovascular disease, on cancer, and on diabetes. That they did so despite the obstacles they put on the path to this finding is nearly incredible, and directly bespeaks the magnitude of adverse effects of meat and processed meat intake on health.
Out of the Corners of Our Eyes
Sometimes we see most clearly by looking away, taking in the view at the corner of our eyes. Such is the power of analogy, so let’s apply it here.
Imagine studying the health effects of walking. Imagine comparing studies generally not even about walking, but focused on other outcomes- such as TV viewing time. Imagine relying on such studies in which any change in walking is a secondary effect. Imagine thus winding up with very small apparent between-group differences in walking- say, 5000 average daily steps versus 4200. Imagine further having no information on what people are doing “in the place” of those 800 missing steps; they might be sprawling on the couch, but they might also be on a bike, or in a pool, or dancing.
Then, imagine finding that despite all of this- there is clear, consistent benefit from even those 800 additional steps daily. But finally imagine applying a statistical approach to data “scoring” that allows you to call your own findings unreliable- and then issuing guidelines that “people need not bother walking for their good health.”
Or, in the opposing direction: imagine researchers select studies with extremely small between-group differences in the number of cigarettes smoked per week. They avoid any studies that actually compare smoking to non-smoking. They find that despite the small differences in exposure, there is still a clear and consistent benefit of smoking less. They then apply methods of grading the evidence that strongly favor randomized trials over all other methods. Since there are few if any randomized trials of smoking, they conclude that they have very low confidence in the reliability of their own findings. On that basis, they publish guidelines recommending – that the public simply continue to smoke. After all, they reason, people who smoke like smoking.
If all of your data show benefit from even tiny reductions in smoking, but you have no confidence in your findings because of how you chose to score them- how can that possibly justify recommending the opposite for which you have no evidence at all, and in which you thus have zero confidence? If you are scratching your head asking these questions, stop before you make a bald spot. There are no good answers.
There is no good information in these new publications in the Annals either. There are no new or surprising findings; there are no inconsistencies. There is the presumption of guidance to the public by statisticians who contrived a conclusion at odds not only with robustly established understanding- but with their own data besides.
Perhaps you do like to eat meat. If so, these statisticians are wrong to think that’s an insurmountable barrier between you and what the health data really show. You are likely aware of the many ways, some quite new, to avoid your meat and eat it, too. The statisticians factored preference into their data scoring, but even here they appear to have been blinkered. The massive uptake by markets and consumers of meat alternatives shows that the widespread preference is not for a license from statisticians to keep eating meat despite the established harms to human health, abuse of animals, and calamitous damage to the planet and biodiversity. The preference is to find a way to love food that loves our health and the planet back. The statisticians not only took liberties with their own data, they seem to have overlooked massive trends in the world all around them.
Everything about these new publications and “guidelines” is wrong, down to the timing of them. Whatever sound and fury they may propagate in news cycles and social media, they truly do signify nothing to challenge the established consensus of science: for the health of people and planet alike, limit your intake of beef/red meat, and minimize or avoid intake of processed meat.
Still here? Left a little room for intellectual dessert? OK, here it is:
Of Diet & Exercise, a Juxtapositional Post-Script:
One final thing for those as frustrated as I am over the seemingly perpetual state of dietary pseudo-confusion. How can we know so much, yet so redundantly seem to know so little?
The answer resides, I think, in the unique disrespect our culture directs at nutrition- along with the reality that it is a massively complex, infinitely nuanced variable.
We can sort out what we do and don’t know, and how we manage to conflate the two so often and badly, by juxtaposing diet- where pseudoconfusion reigns- and exercise, where everyone knows that some is better than none, and more is better than some. Why the divide?
Imagine for a moment if we studied diet like exercise, or exercise like diet.
If we studied diet like exercise- a consummation devoutly to be wished- we would consistently highlight the common ground and global consensus. We would highlight the fact that there is a perfectly clear, enormously important, impressively consistent, evidence-based theme of healthful eating that could be elaborated in a great many words or as few as seven. All of our debate about this dietary approach versus that, these foods versus those, would take place within the realm of our established agreement about diets of high objective quality in which minimally processed whole foods, mostly plants, predominate. If we studied diet like exercise, research would be directed at adding to and refining what we already know, not replacing it from scratch with every study and news cycle.
Conversely, what if we studied exercise like diet? Well, then, we would toss out the agreement about the benefits of being active in general, and replace that entirely with a fixation on the one best way to exercise, and the relentless pursuit of the “active ingredient” in any given activity pattern.
So, for example, we would study people walking 10,000 steps daily and seek to identify which cluster of 50 steps was responsible for the fitness benefit. We would look at these 50 steps and fail to find decisive benefit; we would look at those 50 steps, and fail again. By the time we were done, we would have examined every set of 50 steps and found no decisive benefit attached to any of them. We would thus conclude that there was no benefit from any of the 10,000 steps- and advise the public to give up walking.
Further, rather than propound our agreement about the benefits of being active over couch surfing, exercise researchers would tribalize. Proponents of biking would design studies that helped show biking is better than, say, walking. Walking advocates would do the converse. Swimming devotees would design studies inevitably showing that swimming is better than either.
News cycles would gyrate accordingly: “walking beats biking;” “biking beats walking;” “swimming beats biking;” “walking beats swimming.” The public, hearing almost never about the benefits of activity in general, and hearing incessantly instead the discord among factions- would be left to conclude that exercise experts can’t agree on anything, and know nothing reliably.
They would all retire to their couches and limit exercise to rolling their eyes at all concerned.
This column was reviewed and approved for content accuracy by one of the authors of the newly published papers in the Annals of Internal Medicine prior to posting.
Dr. David L. Katz is author of The Truth about Food; founder & President of the True Health Initiative; founder/CEO of Diet ID; and founding director of Yale University’s Yale-Griffin Prevention Research Center. He is a 2019 James Beard Foundation Award finalist in health journalism.
David L. Katz, MD, MPH, FACPM, FACP, FACLM, is the Founding Director (1998) of Yale University’s Yale-Griffin Prevention Research Center, and current President of the American College of Lifestyle Medicine. He has published roughly 200 scientific articles and textbook chapters, and 15 books to date, including multiple editions of leading textbooks in both preventive medicine, and nutrition. He has made important contributions in the areas of lifestyle interventions for health promotion; nutrient profiling; behavior modification; holistic care; and evidence-based medicine. David earned his BA degree from Dartmouth College (1984); his MD from the Albert Einstein College of Medicine (1988); and his MPH from the Yale University School of Public Health (1993). He completed sequential residency training in Internal Medicine, and Preventive Medicine/Public Health. He is a two-time diplomate of the American Board of Internal Medicine, and a board-certified specialist in Preventive Medicine/Public Health. He has received two Honorary Doctorates.