This week, the Annals of Internal Medicine published five new research papers and a commentary arguing that red and processed meat aren’t bad for you. The authors argued that the quality of evidence is low to very low and that the impacts, if they do exist, are small to very small. Other public health scientists, including at the Harvard School of Public Health (HSPH), which issues its own public health guidelines, have argued strongly against the conclusions of these new papers. So what do these papers actually say? (And why are there so many?)
Aaron Carrol, a public health hero of mine, wrote the introductory commentary. He points out that nearly all the advice about avoiding red and processed meats is based on observational studies. These studies are limited in what they can tell us about cause and effect in nutrition. First, people have a hard time remembering what they’ve eaten even in the last 24 hours. Long-term studies of nutrition don’t ask people to record what they eat every day for a decade. Instead, these studies just ask people to fill out one or two surveys about what they eat. Second, if people stop eating one food, they replace it with another, making it difficult to separate the effects of the new food from the old food. Third, many of these observational studies leave out controls for healthy behaviors that could be correlated with things like red meat consumption – say, how long people sleep. And by definition, observational studies can’t control for factors we don’t even know could affect health. The solution to all these problems is to run randomized trials, but getting thousands of people to comply with randomly assigned diets for years or decades is almost impossible.
The first three papers are meta-analyses of every large, long-term observational trial that collected data on meat consumption. The first paper compared people in the highest and lowest categories of red and processed meat consumption. They found that deaths from all causes, heart disease, and cancer all went down by a little over 10% for people who ate the least compared to the most red and processed meat.
The second and third papers measured the benefit of reducing consumption by 3 servings per week, from whatever point you start at. Where the second paper focused on total mortality and heart disease, the third paper focused on cancer. Both papers found reductions in mortality by a little under 10%. The HSPH prepared a summary table for the key results of the first three papers, which I’ll copy here.
In the fourth paper, they attempted to conduct a meta-analysis of large, long-term randomized controlled trials of red and processed meat consumption. But there weren’t any. (They knew that when they started. They frame this as a failed attempt just to make a point.) There have only been three large, long-term randomized trials of diet.
The PREDIMED Study was the best diet study we’ve ever had, focusing on the Mediterranean diet. The authors found that a Mediterranean diet is good for you, but in 2018 they had to retract it because they messed up the randomization process. It’s still a useful study, but it’s now best viewed as an observational trial and not a randomized experiment. The Lyon Diet Heart Study was another randomized trial of the Mediterranean diet. Those experimenters ended the study early because results were already so strongly in favor of a Mediterranean diet. The authors of this new review ignored the PREDIMED Study and excluded the Lyon Study because they believed its results were implausibly large.
That left the reviewers with only one study: the Women’s Health Initiative (WHI). The WHI randomized 50,000 older women to a low-fat diet with group coaching sessions. It wasn’t focused on red and processed meat consumption, but the women in the treatment group reduced their consumption of those foods. The reviewers rated the WHI as low-quality evidence for questions about red meat because of its broader low-fat design. The WHI found no benefit for all-cause mortality, heart disease, or cancer.
The reviewers used the GRADE criteria to assess the quality of this whole body of evidence. In sum, they consider it to be low or very low. For those keeping track, the fifth paper in the series tallied up how much people love eating red and processed meat. But that’s a matter of personal taste and irrelevent to whether or not those foods are unhealthy.
So should we eat all the bacon and sausage we want? No. Their data agree with all prior meta-analyses of red and processed meat consumption. The preponderance of the evidence is that they’re bad for you. The only thing that’s new in these papers isn’t the actual results; it’s just the authors’ unusual interpretation.
First and foremost, a 13% reduction in all-cause mortality is not “very small.” If you enjoy a diet high in sausage and bacon enough to accept that tradeoff, go right ahead. I don’t. I’m not a vegetarian, but I’ll continue to choose chicken over red meat most of the time.
The health benefits would be even larger than the numbers here for the roughly 1/3rd of US adults who eat 1 serving/day or more of red and processed meat. The HSPH says, “These risk reductions are substantial at both individual and population levels. We currently spend tens of billion dollars per year on screening and treating risk factors for cardiovascular disease and diabetes that have benefits of this magnitude.”
Second, the evidence is not low quality. The HSPH says the “GRADE criteria were mainly developed for evaluating evidence from drug trials.” These reviewers are correct that randomized drug trials give us much greater certainty than the studies we have about red meat. But that’s true of everything in nutrition and even exercise. HSPH says that criteria tailored to cohort studies would rank this evidence as “moderate.” They also argue that you should upgrade the quality of evidence if short-term randomized trials show an increase in intermediate risk factors like an increase in LDL cholesterol. That’s exactly the case for red meat. The evidence against red meat is as strong as evidence can ever be for nutrition.
To quote the HSPH at length:
The “very low” evidence grade given to high-quality cohort studies is inappropriate. If the same procedure were used to evaluate the evidence for other dietary (such as low consumption of fruits and vegetables, high consumption of sugary beverages), lifestyle (such as physical inactivity and inadequate sleep), and environmental (such as passive smoking and air pollution) factors, none of the current recommendations on these factors would be supported by high- or even moderate-quality evidence. Basically, the foregone conclusion would echo that of this new report: that people should ‘eat whatever they want and do whatever they want;’ no need to bother with the systematic reviews and meta-analyses.
These HSPH scientists are really riled up. They go past talking about methods to attack the ethics of the scientists and editors involved.
It is puzzling that the journal would publish dietary guidelines developed by a self-appointed panel that are tantamount to promoting meat consumption, despite their own findings that high consumption is harmful to health….It is irresponsible and unethical.
Coming from a group of public health professors, those are fighting words.
I have a lot of respect for the work of Aaron Carrol. John Ioannidis, a Stanford professor, is another scientist I respect who spoke approvingly of these results and argues against conducting more observational trials. I like their crusade to improve the quality of nutrition research, but in this case I think they’ve gone too far. We have sufficient evidence to say with confidence that red meat is bad for you. It’s not as bad as smoking, but it’s still bad.