Skip to content

An Interview with Distinguished University Professor Gordon Guyatt, OC, FRSC on Red Meat and Processed Meat Intake, Smoking, Antioxidants, NMAs Combined with GRADE (Part Five)


Author(s): Scott Douglas Jacobsen

Publication (Outlet/Website): In-Sight: Independent Interview-Based Journal

Publication Date (yyyy/mm/dd): 2019/12/15


Dr. Gordon Guyatt, OC, FRSC is a Distinguished University Professor in the Department of Health Research Methods, Evidence, and Impact at McMaster University. He is a Fellow of the Canadian Academy of Health Sciences. The British Medical Journal or BMJ had a list of 117 nominees in 2010 for the Lifetime Achievement Award. Guyatt was short-listed and came in second place in the end. He earned the title of an Officer of the Order of Canada based on contributions from evidence-based medicine and its teaching. He was elected a Fellow of the Royal Society of Canada in 2012 and a Member of the Canadian Medical Hall of Fame in 2015. For those with an interest in standardized metrics or academic rankings, he is the 15th most cited academic in the world in terms of H-Index at 245 and has a total citation count of more than 261,883 (at the time of publication). That is, he has among the highest H-Indexes, or the highest H-Index likely, of any Canadian academic living or dead. He discusses: ‘controversies’ over ordinary red meat intake and processed meat intake; coffee drinkers, reactions of the media; the GRADE approach in general; the GRADE approach applied to NMAs; making the research more precise; intellectual humility; and research in 2020; limits of automation intervention; technology and new advancements in medicine; and more advice to prospective medical students.

Keywords: anesthesiologist, Canada, evidence-based medicine, Gordon Guyatt, GRADE, McMaster University, medicine, NMA, P.J. Devereaux, red meat.

An Interview with Distinguished University Professor Gordon Guyatt, OC, FRSC on Red Meat and Processed Meat Intake, Smoking, Antioxidants, NMAs Combined with GRADE: Distinguished Professor, Health Research Methods, Evidence, and Impact, McMaster University; Co-Founder, Evidence-Based Medicine (Part Five)[1],[2]

*Please see the footnotes, bibliography, and citation style listing after the interview.*

1. Scott Douglas Jacobsen: I want to start a little bit more on a deep conversation on some of the recent research that has come out, which doesn’t have to do about “Branded Diets” as we talked about before.

It has to do with moderate red meat intakes and the previous recommendations to reduce those more. However, when you did a more GRADE-based approach, the recommendations came out that people are pretty much okay with their red meat and processed meat intake.

Can you walk us through some of the research there? And why and the previous research was not as robust? And why the GRADE research is better??

Distinguished Professor Gordon Guyatt: Perhaps, a slight correction, what you said is “people are okay to eat their meat,” not quite right. Our results were not very different from other people’s results.

So, they come largely from observational studies. Observational studies look at people who eat varying amounts of red meat and compare them to people who eat less red meat. Those observational studies show a relative increase of 10-15% in bad things happening.

Bad things being cardiovascular events, cancer, and cancer deaths. However, two things, I will go into it a little more. Whether the red meat is actually causing the heart disease or the cancer is uncertain, we would call this “Low Quality Evidence.”

Moreover, if it is true, the absolute effects are very small. In other words, for instance, if 1 were to stop one’s red meat intake by 3 servings per week, and average folks in Western countries eat about 3 servings of red meat a week, so, more or less, eliminating red meat for most folks, and if you did this for the rest of your life, you would reduce your cancer deaths by 7 in a 1,000.

Jacobsen: [Laughing].

Guyatt: Which most people would probably think is a small effect. So, there’s 2 things. First of all, the causal relationship is uncertain. Second, the effect, if it exists at all, is small. When you say, “It is okay to eat your read meat,” that depends on your attitude on a small, and some consider it a very small, and uncertain effect.

If you were the person who would say, “Well, it may be uncertain and the effect may be small. But I want to protect my health in any way that I possibly can,” then the message isn’t, “It’s okay to eat your red meat.” It should be, “You better cut down or starve.”

It really depends on your attitudes. We call them values and preferences. I will go back. We did a number of systematic reviews. We did systematic reviews of red meat and cardiovascular risk, red meat and cancer, and dietary patterns and cancer and cardiovascular.

They were consistent in showing 10-15% relative increases in those events for those people who ate more red meat rather than less red meat. Our results were not really that different. We did it more rigorously. We got all the studies available.

We did the GRADE approach. Our results were not that different. Our results were different in their interpretation. The nutritional epidemiologist before said, “On the basis of these observational studies, we conclude red meat causes cancer and cardiovascular disease.”

But the problem from the GRADE perspective is the problem with all observational studies. Germane to the nutritional world. I will give an obvious example, which everyone gets, easily, in terms of the problems with observational studies.

Let’s say you ask a question, “Are hospitals dangerous places?” You compare what happens to people in hospitals to people out of hospitals. You find that many more people die in the hospital. You, therefore, conclude that hospitals are dangerous places.

But if you want to avoid a premature death, then you should avoid the hospital. Most people understand there is a logical problem with the reasoning. It is more difficult to get that there is the same logical problem with red meat and these same bad events.

In other words, just as it isn’t that the hospital kills people, it is that the people in the hospital are different from the people who aren’t. Similarly, it may well be that the red meat does not causes cancer and cardiovascular disease. It is that the people who eat the meat are different from the people who don’t eat the meat.

There are a number of ways people who are in hospital – they’re sicker, clearly – are different than people out of hospital.

Jacobsen: [Laughing].

Guyatt: Here there may be a number of ways people are different, there may be a number of things going along with eating meat. Because the criminals in terms of the problems may not be the red meat but the things that go along with them.

What we saw in the dietary pattern studies support the hypothesis that it is, maybe, something else, secondly, maybe, they exercise differently. Or, maybe, they are more likely to live in areas where there is more pollution.

Or, maybe, their smoking is different, and so on and so forth. There may be things other than the red meat that are, in fact, causing it, just as there are things other than being in the hospital that causes you to be more likely to die in the hospital.

There’s one set of observational studies that highlights the issue. That is, the intake of antioxidant vitamins. So, as it turns out, big, nicely done observational studies of antioxidant vitamins showed that people who take antioxidant vitamins have less cardiovascular disease and less cancer than people who don’t take antioxidant vitamins.

It’s true! People who take antioxidant vitamins have less cardiovascular disease and less cancer than people who don’t take them. It just has nothing to do with antioxidant vitamins. So, when people have done the randomized trials of antioxidant vitamins, all the people who believe in the observational studies are saying, “For sure, we are going to show a reduction in cardiovascular disease and cancer.”

No reduction, zero! Zero reduction in cardiovascular disease and cancer. So, just like the people in the hospital are different than the people out of the hospital, that explains their increased risk of dying. The people who take antioxidant vitamins are different from the people who don’t take antioxidant vitamins.

It is those differences in the people rather than the antioxidant vitamins, which are responsible for the decreased cardiovascular risk and cancer. So, we are, for that reason, using a technical term, “confounding,” which means that the exposure of interest is associated with other differences in people that may, in fact, be responsible for the finding.

In the GRADE framework, we are mistrustful of observational studies. So, observational studies start as low-quality evidence. They, generally, end off as low-quality evidence.

Jacobsen: [Laughing].

Guyatt: If they have other problems, they may even be very low-quality evidence in the GRADE framework, which is high, moderate, low, and very low. Now, sometimes, there may be some things about the observational studies that make us raise the quality of the evidence and make results more trustworthy.

A great example of that is smoking and lung cancer. What makes us sure or very convinced that smoking causes lung cancer is that the relative effect is gigantic, in other words, it’s 10 times the relative effect if you’re a heavy smoker.

If you’re a heavy smoker, you have 10 times the chance of getting lung cancer than if you don’t. Secondly, there is a dose-response gradient. You smoke a little bit. Your risk goes up. You smoke a moderate amount. Your risk goes up more. Your smoke a lot. Your risk goes up even more. You smoke a ton. You have a very high risk.

So, it is those two things. To illustrate the difference, let’s say, you do not eat any red meat. Your risk of cancer is 1%. If you eat, according to the results of the studies, three servings of red meat a week, your risk goes up 1.15%.

Jacobsen: [Laughing].

Guyatt: Whereas with smoking, if your risk is 1%, and if you smoke heavily, the risk goes up to 10%. So, in those instances, when you have a very large relative risk like that, confounding cannot explain it. So, we believe it.

The relatively minor risk with red meat is very easily explained by confounding. So, where we disagree with the others in the nutritional community by applying the GRADE approach, we are much more skeptical of the results of observational studies and only consider low-quality evidence, and are not ready to declare red meat causes cardiovascular disease and cancer.

It might! It might. But the evidence is only low-quality. Previous authors have ignored the issue of the absolute effect. They have only presented the relative effects. They ignored or haven’t event calculated, in most cases, the absolute effects.

So, the other thing is, even if it is a true causal relationship, as I have just told you, the absolute effect is very small, and I gave you an example. Those are the two ways that we did things differently. By the way, we also looked at the randomized trials, which, further, have their own problems and only provide low-quality evidence.

But they have no association with the red meat in the bad outcomes at all in the most trustworthy randomized trials. Bottom lines: skepticism about whether there is a causal effect. If it is there, it is very small.

We also did a systematic review of looking at people’s values and preferences. We looked at how people like their red meat. Perhaps, no surprise, people like their red meat and are reluctant to give up their red meat.

Jacobsen: [Laughing].

Guyatt: Most people would want a convincing effect of some magnitude before giving up their red meat. Some would give up their red meat with a convincing effect of small magnitude. Most people would want something more than that.

That then led to the recommendation, a weak recommendation because people’s values and preferences differ if you’re only considering health effects.

2. Jacobsen: I recall some commentary by you. It had not to do with antioxidant intake, but with coffee drinkers and then some of the rather large claims about the health effects, positive health effects, of it.

Is the similar notion or set findings there too?

Guyatt: Sadly, you are about to uncover the limitations of my memory. I haven’t looked at coffee studies in a while; and I don’t really remember them. It would be the same issue. People who drink coffee.

In fact, most of us can say this by looking around us. People who drink coffee are different than the people who are abstainers. It might be any of the differences that are responsible for the different health outcomes.

3. Jacobsen: After the research with the GRADE approach on average levels of red meat intake and processed meat intake, by North Americans, say, there were mixed reactions in the popular media in general with varying levels of commentary too.

Some more emotive. Some questioning the studies legitimacy and validity. What were some of those? How would you respond to some of those commentaries?

Guyatt: You say there were varied responses. Overwhelmingly, the responses were hostile, I would say. In some cases, intensely hostile, and in some cases, verging on the hysterical, what are the responses?

The responses are really much as what I have just told you. Okay, I will tell you one. The response, “Observational studies are untrustworthy for the reasons that were said. Even if there is a true effect, which there may not be, the effect is very small. And when you look at people’s value and preferences, people are attached to their red meat. The evidence suggests people would be reluctant to reduce their red meat. Unless, there was really compelling evidence to do so.”

That is fundamentally our response.

There is one other thing. Some of the critics claim, “Nutrition should have different rules. GRADE is designed for randomized trials. Nutrition with its observational studies should have a different set of rules.”

Our answer to that. I try to illustrate it. Picture two bodies of evidence, that are identical. They are observational studies. Same number of studies. Same sample size in the studies. Same safeguards against bias. As far as one can tell, in terms of their credibility, they are identical bodies of evidence.

One is looking at the nutritional intervention in which there’s never going to be adequate randomized trials because of he obstacles. The other is a drug for which there will be randomized trials. But in terms of their credibility, sample size, risk of bias protection, and so on.

Is the credibility that you would give to causal inferences from those two bodies of evidence the same?

Jacobsen: [Laughing].

Guyatt: Or is it because in one you can do randomized trials and another is one in which you cannot? This is dealing with an area of study called Epistemology, which is how we know things. To us, it is profoundly illogical to say, “Two identical bodies of evidence, the strength of inference differs on whether you can do randomized trials or not.”

Something outside of the evidence should not determine the credibility of the evidence. So, we would argue rather strongly that one is making an epistemological error by saying, ‘We have different standards of knowledge for one body of evidence over another because what is possible in terms of randomized trials.

4. Jacobsen: When it comes to the GRADE approach in general, are the same critiques repeated when similar large-scale studies are done?

Guyatt: In general, and I should say I am sympathetic to this, the folks who do public health and toxicology, and, in this case, nutrition, have reservations about the GRADE approach. Their reservations are based on the fact that their evidence will seldom be better than “low.”

That makes them unhappy. But if I were in their position, I’d be unhappy too.

Jacobsen: [Laughing].

Guyatt: Because you want to sell your public health intervention, e.g., putting fluoride in the water or getting the public to stop eating red meat. Then someone says, “What is the quality of evidence supporting the advocacy for this public health position?”

They say a little embarrassed, “Oh, it is low-quality evidence But we still think that you should do it.” Not a particularly happy position to be in. But unfortunately, that is the way it is. That doesn’t mean that we shouldn’t act.

Perhaps, we should act on the basis of low-quality evidence. But it is low-quality evidence. These communities with low-quality evidence without randomized trials tend to not be enthusiastic about the GRADE approach.

5. Jacobsen: How are Network Meta-Analyses (NMA) linking up to the GRADE approach?

Guyatt: Historically, meta-analyses, systematic meta-analyses, compared Treatment A to Treatment B. It was a standard comparison. Starting 15 years ago, it started more and more with people presenting the same problem.

If you have 10, or in the case of antidepressants 25, different treatments, then they will seldom be compared A versus B, B versus C, and so on. A lot of the time there will not be a lot of parity comparisons.

A lot of people start to think, “Wouldn’t there be some nice way to summarize the evidence, so we can take all 25 treatments and say which ones are the better ones and the best one?” The statisticians went to work. They made a statistical methodology that compares A versus B and through C.

A versus C shows a big effect. B versus C shows no effect. A is probably better than C. These statistical methods have been around a decade or more. It is early in the game in terms of a new statistical approach.

So, there is lots of work going on now. A few years ago, 2014, maybe, it became very evident that the GRADE approach was needed with NMAs. When we first came up with the initial GRADE guidelines in 2004, it was based on dozens, perhaps hundreds, of examples that we applied GRADE.

It was pretty solid right from the beginning. With respect to this NMA, GRADE guidance was needed, but we hadn’t applied this in nearly so many vases. But we did offer it. Since then, as a result, we knew it was going to happen.

As we applied it more and more, we have refined guidance. There are, at least, 3 other articles out that provide updates and refinement to the GRADE applied to NMA. Bottom line, we have this new statistical approach.

It raises challenges for deciding on the quality and certainty of the evidence, to which GRADE has responded.

Jacobsen: When we’re talking about antioxidants and coffee, and the users thereof, those who come out healthier when using them. Rather than general statements, has or could NMA with a GRADE approach tell us in more detail? They exercise. They eat better, etc.

Guyatt: Probably not, or we’d be no further ahead, then you’d say, “It is the exercise.” But maybe, it isn’t the exercise.

Jacobsen: [Laughing].

Guyatt: The people who exercise are different than the people who don’t exercise is a whole host of ways in a similar way it is harder to do it.

6. Jacobsen: What are some of the next steps in making the research more precise?

Guyatt: The next step is to realize that sometimes: you never know. This is one of those times that we will never know. We may not like that. But we argue, “Better to recognize the best evidence you have is low-quality than to pretend you know when you don’t know.”

7. Jacobsen: [Laughing] would that be a good principle moving forward with intellectual humility, the old one?

Guyatt: Yes, I think so.

8. Jacobsen: [Laughing] so, we are in the end of the year. We did a review of some of the work being done for you. What are we looking forward to in 2020 in terms of some of the next steps in terms of the research?

Guyatt: In research in general, there are thousands of things ongoing. Immediately coming to mind is the work my colleague P.J. Devereaux is doing with perioperative medicine. It is really exciting and might make a big difference.

It has to do with monitoring after surgery. So, I think I told you at that last conversation that the complications of anesthesia have gone down 100-fold since the start. The reason: you have an anesthesiologist sitting by the bedside monitoring every aspect of the condition.

As soon as he or she notice something wrong on the monitor, they are able to react immediately. Then the patient finishes in the O.R. All these monitors are taken off. Then they go to a ward, where a nurse may look after them once every few hours.

We go from this intense monitoring reducing complications by 100-fold to in essence an unmonitored situation. So, we’ve eliminated – not eliminated – or next to eliminated bad things happening in the O.R.

Once people are in the O.R., bad things start to happen. What potentially allows us to do something is the changes in technology, which is relatively inexpensive, and allows people to wear these things for a long time, it may be that instead of walking around checking this patient, that patient, the next patient.

By the end of 8 hours, you have checked all the patients, but the first patient hasn’t been checked for an hour. The nurses can sit at the nurses’ station with the monitors in front of them. After 10 minutes, they can look at the monitors and then go back to the first monitor. In a much, much, much shorter period of time, you can pick up when something is wrong.

You can call the doctor. There are a number of actions that can be taken. I think that really could change the picture. Maybe, not quite in the same way with monitoring with the anesthesiologist with the bedside, but a lot; also, as it turns out, according to P.J. Devereaux’s research, 30% of the bad things that happen, like deaths, after surgery happen after people go home.

A surprising thing, I think most of us were surprised at that finding. Solution, they keep wearing the monitors when the bad events happen. So, I think P.J. says, “I want to cut post-operative mortality in half.”

He might just pull it off.

It, of course, would be a gigantic event. That’s, maybe, in the world of people who I work with, the most exciting potential.

9. Jacobsen: You mentioned something as one subtext to that. When you have an anesthesiologist by the bedside of a patient, followed by a nurse, followed by a nurse checking the readouts, say, there’s an automation of some healthcare there.

Where does that borderline hit where you will still need someone like an anesthesiologist or someone like a nurse to do consistent monitoring of a patient in those cases?

Guyatt: Always, until, we can teach patients to monitor themselves. There will always have to be someone who can understand the outputs.

10. Jacobsen: Any developments on the technology side that you know that are making things even more deep into that field?

Guyatt: The short answer is: what I know about all of this is what P.J. Devereaux has told me, so, the details are there. Certainly, the thing will go, “Beep! Beep! Beep!”, when something is not good. But [Laughing] someone will have to look at the thing if there is a problem.

11. Jacobsen: To any prospective medical students, they will look for various experts in different areas, or take advice. You have been doing this your whole professional life. Let’s take a note from a veteran.

What do prospective medical students need to know and have going into medical school?

Guyatt: I would like to think that they would, ideally, have a fair bit of intellectual curiosity, and they, ideally, would genuinely care about other people. One way to put it: if you cannot treat every patient as if it is your mother or father, of someone who you dearly care about, perhaps, medicine isn’t the right career for you.

The caring about people and being ready to make some degree of always putting the patient above, “It is late in the day. It is time to get for dinner. I do not feel like getting up early this morning,” or taking a short cut is tempting.

It is to care enough that you would put the patient first. I don’t know. That is the prime attribute that I would like to see.

12. Jacobsen: Thank you for the opportunity and your time, Professor Guyatt.

Guyatt: Alright, good!

Appendix I: Footnotes

[1] Distinguished Professor, Health Research Methods, Evidence, and Impact, McMaster University; Co-Founder, Evidence-Based Medicine.

[2] Individual Publication Date: December 15, 2019:; Full Issue Publication Date: January 1, 2020:


In-Sight Publishing by Scott Douglas Jacobsen is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. Based on a work at


© Scott Douglas Jacobsen and In-Sight Publishing 2012-Present. Unauthorized use and/or duplication of this material without express and written permission from this site’s author and/or owner is strictly prohibited. Excerpts and links may be used, provided that full and clear credit is given to Scott Douglas Jacobsen and In-Sight Publishing with appropriate and specific direction to the original content. All interviewees and authors co-copyright their material and may disseminate for their independent purposes.

Leave a Comment

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: