47 Living with Uncertainty

InForm Fitness Podcast


Adam Zickerman and Mike Rogers of InForm Fitness are joined by Dr. Peter Attia to discuss the studying of the studies that we are constantly barraged with through the news and in our social media feeds.  Dr. Attia explains the mechanics of scientific research and how to distinguish the relationship between showing cause and effect in an effort to become more equipped in understanding and possibly mistrusting the information we are given regarding exercise, nutrition, disease prevention, and more.

Dr. Peter Attia is the founder of Attia Medical, PC, a medical practice in New York City and San Diego that focuses on the applied science of longevity. Dr. Atti also happens to be a client of InForm Fitness.

Dr. Peter Attia’s Website: https://peterattiamd.com

Richard Feynman on Scientific Method (1964) https://www.youtube.com/watch?v=0KmimDq4cSU

Do We Really Know What Makes Us Healthy? By GARY TAUBES https://www.nytimes.com/2007/09/16/magazine/16epidemiology-t.html

Adam Zickerman – Power of 10: The Once-A-Week Slow Motion Fitness Revolution http://bit.ly/ThePowerofTen

For a FREE 20-Minute strength training full-body workout and to find an Inform Fitness location nearest you, please visit http://bit.ly/Podcast_FreeWorkout

Read more at http://theinformfitnesspodcast.libsyn.com/podcast#Bs0pQ2HIVD2YWOce.99

< Return to all episodes

Peter:   You know, cognitive dissonance and bias are so strong and once we walk in on an idea, it becomes very difficult to see evidence that contradicts that. And it also becomes very difficult to scrutinize evidence that supports it.

Tim:     Welcome InForm Nation, to episode 47 of the InForm Fitness Podcast. I’m Tim Edwards with the InBound Podcasting Network and in a minute, Adam Zickerman, the founder of InForm Fitness and general manager/trainer at InForm Fitness in New  York City, will be joined by the gentlemen’s whose voice opened the show, Dr. Peter Attia. Who is the founder of Attia Medical P.C., a medical practice with offices in San Diego and in New York City. Focusing on the applied science of longevity. He also happens to be a client of InForm Fitness like myself. Dr. Attia joins us today to discussing the studying of studies, regarding the mechanics of scientific research. And how to distinguish the relationship between showing cause and effect.

Now, you might be asking, what does any of this have to do with slow motion, high intensity strength training? Well, plenty. How many times in this podcast have we referenced a study, based upon exercise and nutrition alone. You see it every day in your social media feeds and on the news. But how many of these studies actually true or even accurate? How are these studies administered and can you trust the results? Well, to quote Adam Zickerman, hopefully, after listening to this episode, you’ll be a bit more equipped to understand the barrage of information you read and hear about. And not to fall, hook, line and sinker for every claim that is reported as a study.

Adam: Hello everybody, this is Adam. Very excited about this podcast, it’s kind of a little heady, gets into statistics, but hopefully we’ll break it down for the layperson. So that when you read headlines that make such crazy claims, well basically your bullshit detector goes off, and you have to figure out how to navigate that. So we have with us, to talk about this, Dr. Peter Attia. He’s a physician practicing in New York City and California, and his practice basically focuses on longevity, which he calls the overlap of lifespan and healthspan, which is kind of cool. His clinical interests are in nutritional biochemistry, exercise physiology, lipidology, endocrinology, and a few other cool things. He’s also somebody who works out — he just worked out here just now, didn’t you?

Peter:   I sure did.

Adam:   How are you feeling?

Peter:   That hurt.

Mike:   Sorry, Peter.

Adam: Mike trained him.

Tim:     Sorry not sorry.

Adam: He is also an incredible athlete, endurance athlete and he understands the efficacy of these high intensity workouts that we talk about. So doctor, thank you for coming on the show.

Peter:   Thank you for having me, but I will add one correction. I don’t think I can use the word athlete to describe anything I do anymore.

Adam: Former athlete.

Mike:   He looks like an athlete; you haven’t seen him, he looks pretty good.

Adam: A good looking guy. So Peter, I wanted to do this podcast because I read your series. You wrote a five piece series, called “Studying Studies” and you ventured into much detail about the mechanics, if you will, of scientific research. And how difficult it is to distinguish, without a shadow of a doubt, the relationship between of course, what all research tries to do, show cause and effect. So Dr. Attia, you start off your series quoting two iconic individuals. First, from Mark Twain, who I guess you mentioned as quoting Disraeli, Benjamin Disraeli. Who says, “there are lies, damn lies, and statistics.” And you also quote Nobel Prize winning physicist, Richard Feynman, one of my favorites actually. I read his posthumously published CalTech lecture called, “The Meaning of it All.” Is that where you got this quote from, by the way?

Peter:   To be honest with you, I don’t remember. I mean, he said it so famously so many times. But my favorite book, of course, of his is, Surely You’re Joking, Mr. Fineman, which is the sort of entertainment value of Fineman.

Adam: So Dr. Feynman says, “the first principle is that you must not fool yourself and you are the easiest person to fool.” So you have to be very careful about that and this is our responsibility as scientists, certainly to other scientists, and I think also to laymen. So doctor, tell me if I’m wrong, but your main point of this whole series of essays is something that you actually say in your first essay. Which is, and I’m quoting you, “statistics can be both persuasive and misleading, if we’re not careful. And it’s self-persuasion that we must vigilantly guard against, if we really want reliable knowledge about the world we live in.” So can you tell us why this is so important to understand?

Peter:   Sure. Before I do, I also want to acknowledge the work that goes into the stuff that I write. It couldn’t really be done without kind of a team of analysts I have. So inside my medical practice, I have a team of four analysts, whose full time job it is is to help me read, assimilate, digest and understand the absolue barrage of data that’s out there. I think in one of our posts, we even that last I checked, there are just under 100,000 pieces literature produced monthly, on PubMed in the English language. And I think our estimation is probably somewhere in the neighborhood of 100 of them would be relevant to what we do. Meaning they overlap with our field of interest, but also would rise to the standard of being relevant of our attention. And to funnel that down from 100,000 to 100 is something that I could use a team of 400 analysts to do. But the four guys I have are amazing, and in particular, Bob Caplan, who’s sort of in charge of the whole operation and my right hand. Without him, this whole thing that we do wouldn’t exist. So shout out to Bob and also, I think — I would say we wrote this series knowing that it was not going to be that “interesting.” This was not something that we wrote — there’s no clickbait involved here. This is kind of something that we hope will, over time, become the kind of thing that people go back to. I forget who said it, but there’s an interesting quote that said something to the effect of like, you only learn something when you’re actually ready to learn it. And of course it was stated much more eloquently but the point is, a lot of times, these things fall on deaf ears. So anyway, with that said, what we’ve created is a bit of a repository called studying studies. And it begins with, I think our first one is on relative versus absolute risk, which I’m sure we’ll get to. But to your point, which is the important one, we are all, myself included, so guilty of this. It is just — you know, cognitive dissonance and bias are so strong and once we walk in on an idea, it becomes very difficult to see evidence that contradicts that. And it also becomes very difficult to scrutinize evidence that supports it. And so I think what Fineman says quite eloquently, to be also stated by one of the guys when I was doing my postdoc, who I remember very well. He said to me, you have to learn to kill your babies. So you’re going to do — you’re going to have a hypothesis, you’re going to do a whole bunch of experiments. You’re going to think you’re getting an answer and you’re going to start to drive that answer. You’re going to look at ways to see your data, in a way that makes your answer seem more believable. But you have to be able to scrutinize to the point where you ultimately end up killing most of your babies. So I don’t pretend to say this is easy. I think this is actually very difficult, and in many ways, the easiest way to do it is doing it the way we try to do it. Which is, have more than one person involved, and have opposing sets of eyes. I’ve spoken about this, certainly in sort of other discussions around the notion of creating a blue team, red team. So this is something that is not uncommon in sort of experimental disciplines of science, like physics. But I think it’s frankly even valuable in things like a hedge fund. So for example, if you’re trying to decide if this stock is worth the price that it’s being asked right now, you typically will send of a team of analysts to do a bunch of research on that. But that exercise can be much more valuable if you send off a blue team and a red team. That don’t necessarily communicate with each other in any way, shape, or form, but go and look at data and force themselves to make the for and against case, and then you compare it.

Mike:   Create a debate, hoping it will be a debate.

Peter:   Absolutely. So we internally kind of talk about this idea of blue team, red team, as a way to keep ourselves honest because I don’t really trust myself that much. And I don’t know that any individual should be able to trust themselves.

Adam: No. I know I’m biased. Totally, when it comes to just the way we train people and how we view exercise. And that’s why I — like you said, you don’t learn something until you’re ready for it and that’s where I am in my career. Trying to push the envelope with exercise science and knowledge and this can’t be it. So there’s got to be more to all of this. I think it was Carl Sagan that said, if there’s an exception to the rule in science, the rule is wrong. And I’m finding exceptions in exercise all the time. So even I need to wake myself up and push the envelope and question everything I do,  so that’s why I have you on the show today.

Peter:   I think that speaks to something that makes biology much more difficult than physics or mathematics. So the first thing is, I think there are very few rules in biology, in the way that we think of rules in these other disciplines. So to your point, I don’t think I could, with a straight face, make the case that there is one exercise regimen, one nutritional regimen, one endocrinological regimen etc that is optimal for every person. I simply know that’s not the case. I could argue it on first principles, I could argue it empirically. So what’s a corollary of that? A corollary of that is that there are no proofs in biology and I’m going to be very careful with my use of that word. If I had a dollar for every time I saw a study that said, such and such proves X. Or this study proves Y. It’s incredible sloppiness, either on the part of the journalist or the scientists, because unlike mathematics where there are proofs, there are no proofs in biology. Everything is is stochastic, everything is about probability. Everything is about confidence, but there is no proof and therefore, as an extension, there are likely very few rules, outside of things like central dogma.

Adam: Exactly. So let’s get to this idea of self persuasion and use some examples. An example you used was a headline and the title of this journal article said, “Cholesterol Fighting Drugs Lower Risk of Alzheimer’s Disease.” And you use that as an example of this self persuasion.

Peter:   Well, I was thinking about that after we wrote it, and I was like, that wasn’t even a great example, because as — it was a good example because I think there is a far more egregious example actually, which is the use of hormone replacement therapy in postmenopausal women and the risk of breast cancer associated with it. I think I remember Bob and I talking about this before we put that piece out and deciding we weren’t going to use that example, since that warrants an entire post on its own, which we’ll get to. The gist of that is the following; I think many people, if they’re up in the peanut gallery and not digging down into the data, will still have some sort of vague notion that hormone replacement therapy increases the risk of breast cancer in women. So the question is, where does that inclination come from, and that inclination comes from something called the Women’s Health Initiative. Which was a study that was done in response to, interestingly, an epidemiological series of studies that suggested that women who went on hormone replacement therapy actually saw improvements in many aspects of morbidity. Obviously not useful to look at mortality, but for example, greater cognition, improved bone density, improved heart health, less breast cancer, etc. But as was the reasonable thing to do, the question was, well, let’s put this to the test with a prospective, randomized double blind and control trial. And that was done, I won’t get into the complete shortcomings of the actual study, because the study itself had at least four enormous structural shortcomings. That are not a fault of the people who studied it, in other words, at the time that the study was done, I think it had to be done incorrectly. As ridiculous as that sounds. But when we look back today, it’s very clear that you would do at least four things orthogonally different, not just slightly differently. Nevertheless, when the study came out, the headline, the only thing that most people is remember is, “breast cancer risk went up 25% in the women getting HRT.” So to this day, if I ever want to have a discussion about HRT with a patient, it begins with about a 30 minute lecture on the Women’s Health Initiative, which is one of the papers I actually have sitting in my office, on my desk, so I can just easily refer to it.

Adam: It kind of reminds me of the Autism and vaccination type of…

Peter:   That’s an even worse example because that actually was perpetuated by fraud. What effectively turned out to be fraud and sort of — at least scientific misconduct. This was an example where yes, it is true that when you compared the use of estrogen and  progesterone to a a placebo, the hazard ratio for breast cancer was 1.24. So what does that mean? A hazard ratio of 1.24 means a 24% increase in the relative risk of breast cancer. I’ll resist all urges to explain why I don’t even believe that is correct, but let’s assume it is correct. Let’s assume there is a 24% increase in response to the use of estrogen and/or progesterone. The important question that should be asked, but was often overlooked was, what was the absolute increase in risk? Because a relative increase in risk doesn’t tell you anything by itself. And when you look back at the data, which is very easy to do. You don’t actually have to be a statistician or a mathematician to do this. It’s actually only basic arithmetic that I think will walk readers through in that series. The  absolute increase in risk was approximately going from four cases, per thousand, to five cases per thousand. Now, going from four cases per thousand to five cases per thousand is indeed, a 25%, it was actually 24% increase in relative risk. But it’s only one in 1,000 or 0.1% absolute increase in risk. So the way I try to help women understand HRT is, let’s assume that’s correct and I’ll give you seven reasons why I don’t believe it is. But if it that is correct, if that becomes our ceiling of risk and let’s assume it is, you then have to ask the question. Is 0.1% increase in your risk of breast cancer worth the alleviation of some of the symptoms that you might actually have during menopause? And of course, I think the answer would be, it depends on how bad those symptoms are. If a woman is sailing through menopause and she has no symptoms, then one could maybe convincingly say, look, it’s not worth the hassle/the risk. But when a woman is debilitated by a number of symptoms, including all of the basic motor symptoms, which are  usually the first ones. You have to ask the question, how bad is 0.1% absolute increase in a disease that has a one in twelve chance of killing you?

Adam: Especially now, since you could probably figure out their family history of breast cancer and…

Peter:   Absolutely. We have much better ways to further refine risk for an individual. So this is, to me, one of the most important things that the layperson wants to have in their toolkit. Which is, if ever I read something, whether it be in the actual paper itself or, more typically, in the way it’s reported. I want to be sure I understand the difference between absolute and relative risk.

Adam: So Peter, is it too wieldy a question to ask you to help us figure out the difference between — our listeners to figure out what relative risk is versus absolute risk is?

Peter:   It might be, but I can give it a try. So the example I just gave is probably as good a one as you can use, although I think we talk about Alzheimer’s Disease elsewhere. The relative risk — at the risk of not using the word relative to define it — says, from wherever you start, and I don’t know where you’re starting, or I’m going to ignore where you’re starting. How much does the probability of this event go up? And that relative risk, in this example, is 25%. So you went from something that has 0.4% chance of happening, to something that has 0.5% chance of happening. And going from 0.4 to 0.5, mathematically, is an increase in 25%. So the relative risk is 25%. But, of course, if you went from 40% likelihood to happening to 50% likelihood of happening, that would also be a 25% relative risk. The difference is, in the latter example, your absolute increase in risk is 10%. Whereas in the former, your absolute level of increase in risk is 0.1%. And that’s why it’s very difficult to make decisions, clinically, without knowing both. But because we are, I think, inherently a bunch of lazy people, and I would include that in that just as much as anybody, we want headlines. We want — what’s going to sound more exciting; there was a 0.1% increase in the risk of something, or there was a 25% increase? Both are correct, but both are incomplete, in isolation.

Adam: Yeah, you gave another example about the relative risk for this new drug that reduces cancer incidents by 50%, that was the relative risk. But the absolute risk really goes from two in 1,000, going to one in 1,000. So is that really a — everyone thinks, wow, I have a 50% chance of reducing my risk of cancer, when really, that’s actually not true. Well, it’s true…

Peter:   It’s true in a relative chance, but you then have to evaluate the risk. Is dropping my absolute risk by 0.1% worth the tradeoff of taking whatever this drug may be.

Adam: And as you say, we are messy creatures and there are so many moving parts and pieces to what will influence the results of the study. For example, the exercise habits of individuals that make up a sample or that access their healthcare, their smoking history can confound the results of the study you’re trying to prove in association of. So one confounder I’d love you to talk about is this healthy user bias. That’s a common one and it’s kind of easy to understand, and how that plays into studies. And how quickly and — what I’m trying to do is to show how easy it is to have things screw up a study and the results really might not be what we think they are.

Peter:   So I mean, Richard Feynman does a great job and there’s a great clip on YouTube and I know we’ve linked to it somewhere in the blog about — he’s giving a lecture at, it’s either CalTech or Cornell, but maybe you can find it in the show notes and link to it. But it’s a beautiful, beautiful exercise of him on a blackboard, walking through the scientific method. And he explains it in a way that I won’t even try to reproduce, because it’s just so priceless and Fineman-esque. But it’s effectively, you make an observation, you make a guess.  You design an experiment and compute the consequences of what your guess would be, if the — in other words, what would be the experimental outcome if your guess or hypothesis or correct? You design the experiment and then you measure the outcome versus the outcome you’ve predicted. Or that would be true if the thing is correct. Now, he explains it much more eloquently. The gist of it is, outside of doing experiments, we actually can not establish cause and effect, outside of the most extreme circumstances, that tend to be, by far the exceptions and rarely the rule. So what you’re talking about, a healthy user bias, becomes an issue when we rely on things outside of experiments. Which we unfortunately have to do very often, or choose to do very often, in human biology because as you pointed out, humans are messy. We live for long periods of time, you can’t study us in captivity. So it’s very difficult to do an experiment. For example, if you wanted to demonstrate, do people who exercise four times a week or more have lower risks of, pick your favorite disease, versus people who don’t exercise? Well, to do that experiment is almost impossible because what are you going to do, get thousands of people and randomize them into two groups. Meaning, by randomization, that means you’ve mathematically created a large enough sample, that you know that your two groups are statistically identical, and now, you have one group exercise five times a week and the other group never exercise. First of all, you assume that they’re doing that. How you could make them do that without putting them in captivity for 20 years, I don’t know. But then, at the end of 20 years, you ask the question — or even ten years, whatever. Is there a difference in heart disease, cancer, Alzheimer’s Disease, etc. Now, we certainly think there would be, but we don’t do that. Instead, what we do is we say, let’s just go and actually survey the population. Let’s take a backwards look and ask the question, okay, let’s find people and do surveys and find out how much exercise everybody does, and then, we’ll stratify people and we’ll then do a mathematical analysis to try to simplify for other variables, and see if it gives us the answer. The problem with that is, if you’re comparing people who, on their own, are choosing to exercise five times a week, versus people who, on their own, are choosing not to exercise. The likelihood that you are able to also tease out every other difference. For example, their sleep habits, their eating habits; it’s very unlikely. Yes, you can using statistical analyses probably simplify some of the more obvious differences, such as smoking. As a general rule, I think you’d find a higher incidence of smoking in the non-exercising crowd than the smoking crowd and that could probably be extrapolated. But many of these things can not, and there’s actually an amazing essay on this — essay may be the wrong word, but Gary Taubes, who I know you must know. He wrote a great piece in 2007, in The New York Times magazine. I don’t remember the title to it but you should also link to this, because I think it’s a real gem of a piece. The gist of it is, what do we not know about studies, and I think it’s one of the better pieces on the limitations of epidemiology, and specifically, this healthy user bias that really makes it difficult to understand the impact of nutritional choices, exercise choices, other “lifestyle choices.” Frankly, even drug choices, on hard outcomes, because when you do these analyses, the hazard ratios, meaning the magnitude of difference between the groups is usually so small, that it falls well below the threshold of epidemiology. To rise to the level of, say — and it’s not significance. You can often find statistical significance, it’s just, can you be confident that there’s a causal relationship here. And I would say, unfortunately, the answer today is virtually never.

Adam: Another example of healthy user bias that comes to mind is, I hear often that people that floss their teeth have less cardiac disease. As if flossing your teeth itself has an actual cause — an effect on your heart.

Peter:   Now, it could. The point is, we can’t learn it from that study.

Adam: We can’t learn it because people that floss their teeth are also usually healthier in other ways, and they have healthier habits in other ways. And that might be the reason why they don’t have heart disease as often, as that’s the healthy user bias.

Peter:   That’s exactly right and that’s a great example of a question that would be quite vexing to me, because I think that, using cardiac disease as an example, it’s certainly a disease that’s both driven by lipoproteins, inflammation, endothelial function and all of these other things. So is there a plausible mechanism, by which having poor dentition could reduce your risk of heart disease? Absolutely; there’s a plausible mechanism there. But we’re not going to — my guess is the hazard ratio on those studies is sufficiently low, and therefore, at best, it’s generating a hypothesis to be tested.

Adam: Well, regardless, you should — whether it affects your heart or not, you should floss your teeth everybody.

Mike:   We think. Probably.

Adam: So you’ve thrown around a couple words like epidemiology and epidemiology is another word for an observational study. An observational study or an epidemiological study is different from a, what they call a random control study, and you’re kind of talking about hard that is with people in captivity. Because you have to control for all of these variables, otherwise, they all confound the results. So why — and you touched on this a little bit but maybe you could touch on it a little bit more. Why aren’t there more random control studies, which are considered the researchers’ gold standard of studies? Why are we relying on all of these epidemiological studies, which, as you just pointed out, have all of these confounded problems?

Peter:   So I think it comes down to a couple of things, but one of the biggest issues is the most obvious one, which is a logistics one. It is very difficult to do randomized control trials, and it is often much, much more expensive, but also takes much longer. So if you want to study the effect of — let’s go back to the previous example we had. Does the frequency of exercise impact — even just pick one metric — does it impact your risk of Alzheimer’s Disease? Which is a very important question and certainly, one that logically would make sense. More exercise, better blood flow. A number of neurotrophic factors, BDNF, all of these things. You can come up with a hundred mechanisms why, but if put to the test, are we going to — say we’re going to get two groups, several hundred people, and force one of them not exercise. Force another one of them to exercise, and again, we’re not going to put these people in captivity, that’s impossible. So they’re going to have to live in the free world and hope that the compliance is high enough, that you create enough discrimination between the two groups, and then follow for an outcome. Now, that’s not to say that nothing I described there cannot be done; all of that can be done. The question is, how difficult is it, how expensive is it, and oh by the way, who’s paying for that study? Because it’s very difficult to remunerate on an exercise study. So while we can — we have a much higher appetite for doing randomized control trials in pharmacology, and part of it is because the FDA says, you have to. We’re not approving this drug if you haven’t done — if you haven’t demonstrated safety and efficacy and effectiveness in prospective clinical trials. But there’s an incentive to spend a billion dollars, which is about how much it will cost today. A little over a billion today, to get a drug approved through that process. But it’s very difficult to imagine doing that, with something which there is no remuneration [for]. That’s the first fundamental problem. The second one comes down to the ethical problem. Sometimes the most interesting questions are ones in which we just really don’t think it would be ethical to randomize people to one of the two groups.

Adam: In other words, you can’t trust harmful effects.

Peter:   Yeah, we can’t —  not that this is debatable today, but certainly there was a great period of time after the landmark surgeon general’s report in the early 1960s demonizing smoking. Between that and the widespread acceptance of the role that smoking played in lung cancer. But the question is, could you, with all of that mounting evidence, and those were examples were the hazard ratios were more than ten. So this now gets into the territory of where epidemiology may actually be sufficient for determining causality. But would we feel — if we were scientists and physicians, designing that study, would be comfortable randomizing people to a forced smoking group? The answer is no, and similarly today, I don’t know that if I were involved in a clinical trial, that I would be terribly excited about randomizing people to a group that don’t exercise. Or let’s disrupt your sleep for the next ten years. Certainly on a short term basis, it’s probably reasonable to do sleep disruption studies over a period of weeks or even months, to test the validity of theories around the importance of sleep and for example, glucose homeostasis.

Mike:   In a silly way, what the guy did in Super Size Me. How he just ate all of that crappy food; he chose to do that to himself and then tested his health markers throughout the period.

Adam: Sample size of one.

Peter:   But basically, these are the fundamental issues which are, compliance, logistics, duration, cost, and ethics.

Adam: Okay, so hence the reason why we’re really relying on so many observational studies, or epidemiological studies. You even pointed out that 52 observational studies were looked at, and these particular 52 observational studies were actually tested by a random control study. And the random control study showed that every single conclusion of the 52, all 52 conclusion of the observational studies were wrong. They didn’t — it was 0/52, that’s ridiculous. So what gives?

Peter:   I think a lot of the low hanging fruit in epidemiology is gone; I think that’s the bottom line. I think when you think of the real gems, the real case studies of what made epidemiology great, it was in areas where — so epidemiology is such a blunt tool, that you need enormous amounts of discrimination between what you’re trying to detect. And so smokers getting lung cancer or not getting lung cancer. People exposed to massive amounts of asbestos getting mestholemia, yes or no. Chimney sweepers or non chimney sweepers getting scrotal cancer, yes or no. These questions, they had such enormous impacts that the epidemiology could give us much, much more confidence in an answer. Today, we’re dealing with things where if they’re — and to be clear, going back to those 52 cases, I’m not suggesting that every one of those epidemiology studies was incorrect. It’s also possible that some of those randomized control trials were so poorly done that they missed the mark. What I’m saying is, at that point, we’re now outside of the discriminatory capacity of the tool to measure. I don’t know if I — I think we’ve written about this as well and an even more upsetting feature is not that. What’s more upsetting is when you look at — and John Ioannidis, if anybody is interested in this space and they’re looking for one person to be reading, in terms of just like who is a very thoughtful academician who I think has some of the most insights on this topic, it’s this fellow by the name of John Ioannidis at Stanford University. And John has written some of the most cited papers on this topic, including a very famous paper, I believe in 2005 [Inaudible: 31:52] that gets at this notion of how most published research is incorrect.

Adam: Which kind of brings me to the grand finale question. Now that we’re thoroughly confused and we can’t trust observational studies or random control studies, because they’re so difficult to do and not done well. So what do we do as laypeople? We’re trying to lead healthy lives, we’re trying to improve ourselves, we’re trying to make decisions of what drug to take or not take. What do we do?

Mike:   If you’re not prepared to do the deep dive, like what’s step one, I think?

Adam: Well, even with a deep dive it seems like you can read a book on statistics, do your homework and still figure out, holy cow, we still don’t know for sure. So what do we do?

Peter:   I guess there’s a couple of ways to think about this. I think the first way to think about this is to get comfortable with uncertainty. I don’t think we’re wired to deal with uncertainty very well, we don’t really think..

Adam: That’s why there’s religion.

Peter:   Yeah, I think that’s why there’s a lot of actual problems which is, frankly, we — and again, when I say we, I’m being very deliberate to include myself, because even though I’m trained in mathematics and I think as probabilistically as anybody, I know that in my darkest moments, I tend to revert to binary thinking. So I think that…

Adam: What does that mean?

Peter:   Meaning it’s black or white; it’s this or it’s that. As opposed to thinking in probability distributions. So the real way we should always be thinking about life is probabilities. Now, in some cases, we love the examples where the probabilities are so clear. If I drop this pen, what is the probability that it will hit the floor? Well, that is actually — you can describe that with a probability function in physics and mathematics, and it will show you that…

Adam: There’s a probability, a small probability that it’ll actually not hit the floor.

Peter:   There’s, theoretically, a small probability that it won’t hit the floor. From a practical standpoint, in that example, the answer is zero. So we have a bunch of rules that govern our universe, that get us, I think, overly comfortable in the notion of yes or no, black or white. Zero or one, binary type answers. In sort of engineering, we describe that as being digital; on or off. The opposing concept is called analog, where you have a sliding scale, from zero to one; you turn up the lever. So that’s, to me, biology is much more analog than digital, and therefore, for every question, you just have to say, we may never actually know the answer to this. What I have to do, if I’m making a decision, or what my doctor has to do if she’s making the decision or at least advising me on the decision. I have to be able to understand the risk adjusted return on this investment. So the very simplest way I try to describe this to patients is using a 2×2 matrix. On the X axis is where I talk about the reward; on the Y axis, the vertical axis, is where I talk about the risk.

Adam: The hormone replacement therapy might be a good example of…

Peter:   Sure. Frankly, everything that we do is an example of this. In fact, Bob and I will often draw on a whiteboard, like this 2×2 matrix, and shade in the parts of it that we think represent any type of intervention. So where does high intensity interval training fit in in this? Where does long distance endurance training fit into this? Where does metformin fit into this? Where does intermittent fasting into this? Like anything that you can do should be able to be placed on that matrix and it’s not a point. It’s not a dot, by the way. It’s like a shaded curve, so it’s complicated. But you can simplify this into a 2×2 as opposed to just a continuum, and the 2×2 would be when dealing with the reward or the payoff, think about picking up a Bitcoin versus picking up a dollar bill. Now, this example might be irrelevant in a few years if Bitcoin — but let the record show that at the moment of this, Bitcoin is still worth something. So you’re either asking the question of, am I picking up a Bitcoin or am I picking up a dollar bill? Or make it a penny for that matter, something that we would prescribe very little value to. On the risk side, the question is, am I picking this thing up while it’s sitting in front of a moving tricycle or a moving train? So I try to look at everything I do through that lens and the first thing you want to realise is, you never, ever want to be picking up dollar bills in front of moving trains, that’s an obvious statement. But it’s worth thinking through things that you might do. There are lots of things that I think people propose to do, that in my mind, amount to that; picking up a dollar bill in front of a moving train. Yeah, you could get a dollar, but it also could be a catastrophic outcome. Conversely, there are very opportunities that we aren’t already aware of that are akin to picking up Bitcoins in front of tricycles. Most of those things have been realized. For example, not smoking is picking up a Bitcoin in front of a tricycle. It’s got a huge multiplier effect and it’s relatively safe to not smoke. Similarly, being insulin sensitive as an outcome is another one of those things where there’s just no disagreement. There’s complete convergence along the importance of insulin sensitivity, with respect to cardiovascular health, cerebral health, cancer, etc. Now, the how to becomes more problematic. So should one do this type of exercise versus that type of exercise? Should one take this drug versus that drug? Metformin I brought up before a moment ago because I think about ten years ago, it started to become pretty clear in what are called cohort studies. Where you took backwards looks at data that were collected for other reasons.

Adam: Retrospective.

Peter:   Yeah, these are retrospective cohort studies that are looking at patients with diabetes, who were taking metformin versus those who weren’t taking metformin and they corrected for all sorts of factors. And the suggestions looked pretty interesting actually. Both on an absolute and relative basis, the metformin takers were getting a lot less cancer. Their relative risk, reduction of cancer, was about 25%. And their relative risk in mortality was about 38-40%. Again, I don’t remember the absolute numbers but they weren’t trivial. It wasn’t like one of these 0.1% questions. So it begged the question, should we be taking metformin for cancer prevention? Now, at least three out of five patients will come in my office and they want to know if they should be taking metformin. And by full disclosure, I do. I’ve been taking metformin for ten years. But I also like to point out to patients, that I’m taking it in a really off-label way, because I don’t actually have data to talk about insulin sensitive people taking metformin to reduce the risk of cancer. So to me, I don’t think — so the benefit of taking metformin, if you have diabetes, might be high enough that that’s like your Bitcoin. And I think metformin is a relatively safe drug, that it’s probably closer to the tricycle than it is the train. But in someone like me, if I’m going to be brutally honest, it’s really picking up a dollar bill. I don’t believe I’m getting nearly the benefit of the patients with Type II diabetes. And I say that based on subsequent cohort studies that looked at obese non diabetics that were insulin resistant versus not etc. So to your macro question, how the hell does one actually make sense of this, I think the short answer is, you don’t. There is no making total sense of this; there is no knowing what to do at all times. There’s simply a process by which you think about things and notice, everything in that process involves your own lens for risk.

Adam: Well when you talked about — you were describing the tricycle versus the train and picking up a penny. That says it all, I think. That’s the answer because I relate — when you were saying that, I was thinking about a comment that one of my clients made to me. Saying, listen, Adam, I remember what you told me because this is a guy that I considered overtrained. He just worked out way too much, always getting hurt, always sore. I said to him, I’m more of your risk manager than I am your trainer. In other words, squats are a great exercise, deadlifts are a great exercise, but at what risk? Because doing squats with barbells and weights on your spine is just not worth the benefit because it can be catastrophic. You might as well do leg press, and maybe a leg press machine is not as effective as a squat with dumbbells on it, but it’s a lot less risky. So these are decisions that we’re making on how we train people, how we exercise. Yeah, maybe squats are better, who knows. I know they are a lot riskier, so let’s take the less risky route.

Peter:   I think that’s a great example. I’m a huge proponent of squats and deadlifts, but I’ll still go through periods of time — Mike, we were talking earlier with the workout you asked me before we started. Hey, you got anything kind of bugging you and it’s like, yeah, my right side joint has been bugging me for a little while. And I actually took six weeks off deadlifting, just focused on all single leg, isolateral. Lots of lunges, lots of lateral stuff and basically just had to give the thing a rest. Now, was that absolutely necessary? No, I’m sure I could have pushed through it, but yeah, the view was, look. I can probably get 80% of the benefit, without the exposure and the risk, and I think when you are doing very, very heavy compound joint movements, rule number one is, don’t get hurt. You talk to any good investor, they’ll say, rule number one is, don’t lose money. Charlie Munger probably gets credit for being the first to say that, but I think any investor will say that. Rule number one, don’t lose money. Rule number two, don’t forget rule number one. And rule number one of exercise is, don’t get hurt.

Adam: Thank you so much, that was great and sorry that we didn’t have any great conclusion for you people, but that’s reality. There’s a lot of uncertainty in this world. Now, I highly, highly recommend you read everything that Dr. Attia writes. He has a website peterattiamd.com. Read everything he writes, I highly recommend it. Thank you so much for coming on our show, it was very informative, very articulate.

Peter:   Thanks for having me. Thanks for the workout, Mike.

Mike:   It’s my pleasure, thanks for being here.

Tim:     Thanks guys. We’ll include links in the show notes not only to Dr. Attia’s website, but also, to an article that he referenced earlier in the episode, written by Gary Taubes. The article is titled, “Do We Really Know What Makes Us Healthy?” Dr. Attia also referenced a video that you can find on YouTube. It’s Richard Feynman’s, “Scientific Method.” We’ll have a link in the show notes to that as well. You’ll find additional links in the show notes, that will direct you to the InForm Fitness website. Where you’ll find a free, slow motion, high intensity workout waiting for you. Just click the “try us free” button, right there on the homepage. Fill out the form, pick your location, and then, you can experience a free, full body workout that you can complete in just 20-30 minutes. It’s informfitness.com.

And yet, another link in the show notes is for Adam’s book titled, Power of 10: The Once a Week, Slow Motion, Fitness Revolution. That link will take you to Amazon and for less than 15 bucks, you’ll find a ton of nutritional tips, including a handy list of foods that support the Power of 10 protocol, and some effective demonstrations of exercises that you can perform in the comfort of your own home. You know, we have close to 50 episodes for you to binge listen if you’re new to the podcast. So don’t forget to hit subscribe in whichever podcast app you might be listening. And if you don’t mind, we’d really appreciate it if you took a couple of moments to leave us a review. Until next time, for Adam Zickerman and Mike Rogers of InForm Fitness, I’m Tim Edwards with the InBound Podcasting Network.