#197 - The science of obesity & how to improve nutritional epidemiology | David Allison, Ph.D.

Hey everyone. Welcome to the Drive Podcast. I'm your host Peter Attia. This podcast, my website and my weekly newsletter all focus on the goal of translating the science of longevity into something accessible for everyone. Our goal is to provide the best content in health and wellness, full stop, and we've assembled a great team of analysts to make this happen. If you enjoy this podcast, we've created a membership program that brings you far more in depth content. If you want to take your knowledge of this space to the next level, at the end of this episode, I'll explain what those benefits are. Or if you want to learn more now, head over to peterattiamd.com subscribe Now without further delay. Here's today's episode. My guest this week is David Allison. David received his PhD from Hofstra University in 1990 and then he went on to complete his postdoctoral fellowship at the Johns Hopkins University School of Medicine and a second at the New York Obesity research center at St. Luke's Roosevelt Hospital Center. He's currently the Dean and Provost professor at the Indiana University Bloomington School of Public Health and prior to that he was an Endowed professor and a Director of an NIH funded nutrition Organization Research center at the University of Alabama in Birmingham. He's authored over 500 scientific publications and received many awards including the following the 2002 Lilly Scientific Achievement Award from the Obesity Society, the 2002 Andre Mayer Award from the International association for the Study of Obesity and the National Science foundation administered 2006 Presidential Award for Excellence in Science, Mathematics and engineering mentoring. In 2012, he was elected to the National Academy of Medicine and the National Academies. He serves on or has served on many of the editorial boards and currently serves on as an associate editor or statistical editor for obesity, the International Journal of Obesity, Nutrition Today, Obesity Reviews, Public Library of Science, plos Genetics, Surgery for Obesity and Related Diseases and the American Journal of Clinical Nutrition. He is also the founding field Chief editor of Frontiers in Genetics. David's research interests include obesity and nutrition, quantitative genetics, clinical trials, statistical and research methodology, and research rigor and integrity. I've known David Allison for probably about seven or eight years now. I have always found him to be one of the most insightful and thoughtful people on virtually any topic. And that means fly fishing and genetics. And you pick it in this episode. Of course we don't talk about fly fishing. What we do talk about is obesity and we spend a lot of time getting into the what is known and what is not known about obesity and also the science of obesity. In particular the science of nutrition and nutritional epidemiology, and we talk about all these pitfalls, which for many of my listeners will be familiar. But what I really appreciate about David is the lens to which he brings a very unemotional and a very insightful and a very logical and thoughtful approach to how he thinks about the pitfalls within this space. There are parts of this discussion, actually, that are quite frustrating in the sense that David acknowledges, for example, that most of the measures that we have in place from a public health perspective are probably not founded in any scientific basis. We then talk a little bit about the reproducibility of science, and we close with a discussion that, for me personally, was the most interesting, which was a very clear elucidation of the challenges that maybe science seems to be facing or maybe is not facing today as this kind of existential threat where science and scientists are often confounded, science and advocacy is often confounded, and there seems to be a bit of a crisis around this. David offers his thoughts on that and whether or not there is a crisis or not. We close our discussion with, I think, one of the most lucid explanations of something that I've been trying to wrap my head around for some time without much success, and that is the seeming lack of credibility that science has today. David is quick to point out that it's probably not science that is that which is being doubted. Certainly the scientific method is a means by which knowledge can be reliably and reproducibly gained, but rather it might be the confounding of science and advocacy. I won't try to reproduce what David said here because, heck, he does a much better job than me. But I would encourage you to listen all the way through, even if at the outset you find the topic of obesity not particularly to your interest. So without further delay, please enjoy my conversation with David Allison. Hey, David. Great to see you today.

My pleasure, as always, Peter. Good to see you.

It's been a while since we've seen each other in person. I feel like it's probably been, I don't know, three or four years, right?

Easily, easily. We've spent many an hour on Zoom Together in the last couple of years, but it's been a while since we've been in the same state, in the same building, in the same room, at the same time. Same time.

Now, you're no stranger to this podcast in the sense that I feel like you probably listen to podcasts before anybody. In fact, I think you and Nir Barzilai must hold the record for being the first people to listen to a podcast because they come out on Mondays and it's barely the early part of the morning, and usually you and Nir are the first two to send me an email with your thoughts. So I know you're a fan, and it's great to have you on. I've been wanting to do this for some time.

Well, thank you. I really admire the podcast and I admire Nia Barzilai, so to know I'm in league with him is pretty cool.

So you're one of the most interesting people I know, and I always love just spending time with you and talking with you about subjects. And I think one of the things I've always admired about you is you're known for being in a field such as obesity, which tends to attract a lot of dogmatic thinking, and yet you tend to approach this field with. With a distant lack of emotion. You just come at it very intellectually. And I realize, as I'm saying that it speaks disparagingly of people in the field of obesity, which is not really what I'm trying to say. But I think for many people, the field of obesity is a loaded field, scientifically, even, let alone politically, and all the other things that go with it. So I've always assumed that that's because your background is slightly different than maybe some of the people who came to obesity through physiology. So tell us a little bit about your background. What did you study? What did you do your PhD in that sort of thing?

I sometimes will have people say, you know, why are you like this? Or, why did you do this? Why did you choose that? And the truth is, even as a scientist often say, following my friend Don Rubin, who I admire a lot, we may be able to figure out what the causal effect of X is, but we may not be able to assess what. Whether X caused Y. So I don't really know why it turned out the way I turned out, but even as a kid, I was always the kid asking questions. Sometimes that would interest people, sometimes that would annoy people, and that's true today. But I was always one saying, are you sure? How do you know? Where did that come from? What makes you think that's true? Of course, my kids fed it back to me as they were growing up and heard me say it. And when I would say to them, don't do that, you'll get hurt, they'd say, what's your evidence for that, Dad? I went to college and I started out knowing I wanted to be a psychologist, but what that meant to me at the time was Hitchcock films. I was going to interpret people's dreams, and I would figure out what the meaning of the clock in the dream was or something like that. And then the person would be cured and it would be a great movie. And then as I got to college, I started to think about, well, what's the evidence for these things? And that sort of opens up this whole can of worms about asking about evidence. Became more and more cognitive, behaviorally oriented. Still thought I wanted to be a clinician. Got to graduate school, and I asked a professor at one point, as we're being trained in Graduate School, PhD program to administer IQ tests, and we're learning these different theories of intelligence. Some people think there's one factor, some people think three. One person thought, 144. And I said, well, who's right? And the professor says, well, they bring their different evidence to bear and they argue. I said, well, what's the evidence? He said, these factor analyses, I don't really know what that is. What's a factor analysis? He says, well, you have to go study multivariate statistics if you want to understand that. So I don't like to take anybody's word for anything. So I went and studied multivariate statistics and became more and more involved with statistical analysis to the point where eventually, later in my career, people started thinking I was a statistician, and I stopped fighting it. I said, okay, the American Statistical association thinks so, and the university thinks so, and the NIH and NSF thinks so. I think so, too. And so that's great. So I sort of became, by evolution, a statistician, having been trained as a psychologist. But I think all of that comes down to evidence and all of it thinking that people are no different than anything else. Body fat is no different than any other variable, and that still has to obey the same laws of probability and physics and mathematics and so on that we apply to anything else. And I think we often forget that people talk. And so folks who study people think about the words that people say. Whereas if you're studying atoms and molecules, you don't think about the words they say. You just apply scientific methods. The same person who would never question a diabetologist on what the beta cell of a pancreas is doing, unless they themselves are diabetologists studying the beta cells of the pancreas will say to an obesity researcher, well, this is how obesity works. And it's this aspect of food or that aspect of culture or that aspect of child rearing, based only on the fact that they were children or had children and had some experience. And so to me, it's just taking a step back and just treating, as my old boss in biostatistics used to say, whether it's X's and Y's, they're just variables, and applying the same laws of thinking to all those systems.

When was it either post PhD or during PhD that you were drawn to obesity in particular as a clinical field?

I actually started as an undergraduate, I think was a sophomore, and I took a class at Vassar College. And I should say that all my remarks today just reflect my own opinions. I'm not speaking for Indiana University or Vassar College or anybody else, just myself. And I took a class at Vassar College on human emotion and motivation. And we studied the theories of Stanley Schacter, who at the time was a professor of psychology at Columbia University. He's since deceased. An amazing, brilliant man who wrote a book called Emotion, Obesity and Crime. And he talked about the connections among these and how the distinction between our cognitive state and our physiologic arousal state might lead us to certain kinds of behaviors, including, in the case of some people, eating more calories than they might otherwise eat. His experiments were so creative. I just loved it. And he would put, like, a clock on the wall, and he would have the clock move a little faster, but not perceptibly faster. And then he would bring some students in and give them some work to do under some ruse. And then the clock would say for some of the students at 11am that it was noon. And he'd say, by the way, I have a big tray of roast beef sandwiches out here, so whenever you're hungry for lunch, just let me know and you can have some sandwiches. And for some people, the clock wouldn't say it was noon until 1pm and for some people would say it was noon at noon. And he'd say, who would be following the clock and who would be following the actual time? And obese or overweight persons were more likely to follow the clock. And this became the external theory. And then it didn't always work out. And then you'd have another experiment, say, well, it only holds up under this circumstance. And it was incredibly creative how we'd keep looping through. And so I just got hooked on it. And then we had to write lots and lots of papers at Vassar College. I got a great education in writing and thinking there. Not so many tests, lots of papers. So when I would take physiologic psychology, I would write about the physiologic psychology of obesity. And when I took behavioral psychology, I wrote Behavioral and Obesity Developmental. I'd write about developmental aspects of obesity. And I love the fact that you not only could, but in my opinion, had to look at obesity from so many of those different angles that no one thing was gonna do it. Everybody's got their pet thing. This nutrient is toxic. It's this food marketing practice, it's this cultural practice. It's exercise, it's not exercise. I think there are many factors involved and I enjoyed studying it and still do from many angles.

Now this is happening in the late 80s, correct? When you're in school finishing your PhD.

Mid to late 80s, right at the.

Time, what was the both political and medical view of obesity? I don't think it was the same issue it was today. Was it? Obesity rates weren't particularly high in the 80s, or were they? I don't really recall.

They weren't as high as they are now. There had been steady increases. If you go back and look at data from at least the 1700s forward, there have been relatively steady increases in obesity rates in westernized, industrialized countries since then. But there seemed to be this wake up moment in the early 90s with the NHANES 3. So at that point, we didn't have the NHANES, which is the National Health and Nutrition Examination Survey, which annually does today, a survey of a representative sample of Americans with measured heights and weights. That's how we track obesity levels. At that point, it was only done every few years. And so there was this big gap in years between NHANES 2 and NHANES 3. So NHANES 2 had happened, I think, in the early 80s or mid-80s, and obesity rates were climbing, but not dramatically. And so obesity was an issue, but it wasn't on top of everybody's mind, especially for kids. Pharmaceuticals were still seen as the realm of charlatans at that point. If you were a doc and you proposed the use of obesity medications, you were called a diet doc. And that was like a very bad stigma. There weren't many good medications available. Surgery was still looked at as a very peculiar thing for very extreme cases, and even many medical professionals disdained it. Then something happened in the very early 1990s, which is NH3 data came out, and what you saw was this big jump in obesity levels. And that got everybody's attention. And words like epidemic started to be used, especially around childhood. And the public health policy and environmental perspective came in. And things started to shift in the mid-90s. And that was when you had people like Kelly Brownell, probably one of the most forceful voices at the time, who was a sort of old school University of Pennsylvania colleague of Tom Wadden, devotee, a student of Mickey Stunkard, of the cognitive, behavioral, individual clinical treatment approach to obesity, suddenly starting to say, maybe we've got it wrong. Maybe it's this environment. And I don't know if Kelly coined the term toxic environment, but he certainly either coined it or popularized it. And people started to think about prevention and children and the overall environment, the political, economic, social, food, marketing environment we lived in, it all started to take off and ideas from the cigarette world. The same public health people who had been battling cigarette companies and tobacco for decades, many of they and their tools came in and said, we know how to deal with things that are environmental, social, political problems. And it began to be seen as an environmental, social, political problem. And in some way, a lot was gained because we got a lot more funding for it, we got a lot more attention, a lot more efforts, a lot more research. But in some ways, it was lost in the early 90s when I kind of grew up in the field. I grew up in the first federally funded obesity research center, which is the New York Obesity Research center, for a long time the only one. And by gosh, if as a young postdoc, I were to say something that didn't quite jibe with physiology or. Or genetics or anatomy or clinical medicine, there was a physiologist and a geneticist and an anatomist and a medical doctor who would say. And they knew all. They knew each other, they knew the arguments, they knew the literature for the last 20 years. Then the public health people came in fresh without knowing so much. So on the one hand, it launched things up a lot, which was great. But on the other hand, back to what you were saying earlier about the dilution of the rigor of the field intellectually, it became a lot more of opinion, a lot more of advocacy. In the absence of rigorous evidence showing what worked, just saying this seems like it ought to work.

So, David, prior to NHANES 3 and prior to Kelly and others saying, hey, we should look for something else, what was believed to be the environmental or otherwise trigger for obesity, in other words, what was viewed as the cause?

I don't know that there was a single cause. I mean, clearly, what some people today call the energy balance model, which we can go back to and ask whether that really is even a model in some realistic sense, I think was then and still is today by most people in the field accepted as valid, but maybe not as a model. Maybe as a description of what occurs.

Yeah, I was going to say that's sort of a tautology and doesn't really tell us anything. It's implied and obvious and not explanatory.

Right. I think not explanatory is maybe the best way of describing it in some sense. The analogy I would use to that is I think the energy balance statement, I won't say model, say the energy balance statement is as long as you accept the laws of thermodynamics, unequivocally correct, in the same sense as if you accept Euclidean geometry. The statements about the relationship between the legs and the hypotenuse of a triangle are unequivocally correct. But if I say to you, I have a triangle of this size with legs of length A and B, and I proportionately imagine another triangle that has legs greater than length A and B, it doesn't cause the hypotenuse to be bigger. It means the hypotenuse is bigger. And if I say I imagine one with a bigger hypotenuse, it doesn't cause the legs to be bigger, it means the legs are bigger. And any question about whether the legs cause the hypotenuse or the hypotenuse cause the legs is nonsense. They don't cause each other. They're inherent in the definition of triangle, just as inherent in the definition of energy is the law of conservation. And it just says delta energy stores equals delta energy in minus delta energy out. It's just a statement.

So was the belief, though, that this was behavioral prior to NHANES 3, was the belief that, well, if a person is overweight or obese, they're just eating more than they're expending because of a choice?

It's almost hard to answer that because I think the thinking was, and still remains so sloppy that people don't even distinguish among things. Well, I was just invited to a panel literally today that I'm going to be on in a few months, where somebody said we're going to contrast for obesity the relative influence of biology versus behavior, as though we could behave without biology. I don't know about you, but I'm not a disembodied soul. If you're a disembodied soul, maybe. But for the rest of us who are in the material world, you behave with a body, and it's got to obey the same laws of physics and and so on, that any other body has to behave So I think there's a lot of sloppy thinking. But I think if you really drill down and you got to smart people who understood, they would say, of course there are genetic component. Any rancher could have told you 100 or more years ago that there's a genetic component. We can selectively breed animals for being thinner or fatter. That shows you there's a genetic component. There are many other things that show you that, but that's probably the strongest prima facie evidence. You can selectively breed animals to be fatter or leaner. It's prima facie evidence.

What, by the way, is the concordance of identical twins separated at birth? I've always found that to be one of the most interesting ways to assess everything from autism to schizophrenia. Like, it's really a beautiful natural experiment. But I don't know what the concordance is with respect to obesity and identical twins separated. Do you?

Yeah, there are a lot of different ways to quantify that, and we've written a couple of papers on it, but I think the easiest way is just the Pearson product moment correlation coefficient, the ordinary correlation coefficient of BMI twin to twin. And it turns out that for monozygotic or identical twins separated at or nearly after birth, it's nearly the same magnitude as it is for twins reared together, which is about 0.9 ish. Wow. It varies. Varies from study to study. So it's very similar. Now, of course, you can argue, well, that also takes into account the intrauterine environment, maybe some epigenetic things, but bottom line is, whatever those things are, they're not just child rearing of the home. It looks like for many traits, including obesity, the child rearing of the home does have an influence, but it's not a huge influence.

That's an incredible degree of concordance.

It sure is.

When was that realized? When were those studies first done?

Well, it's interesting because you can go back to quite some period of time to find the hints of those studies. And often when you look at science, it's often that at some level, we kind of knew something. The information was sort of there. But there was this moment where a key figure had to come in and both see it and say it with a degree of crispness that wasn't there before, as I said. So lots of people could have seen the ranching data and the mouse data and family data in 1923. Davenport, who looked at now probably as a racist from the past. But at the time, Davenport published under the Carnegie foundation, these studies of concordance of families and you could have seen the genetic component in his 1923 data. But it was Albert J. Mickey Stunkard, who was someone I had the great privilege of knowing and a wonderful, wonderful man. And Mickey did the first major high quality twin and adoption studies. And he did what he was great at. He got on a plane and he went around the world and he met bright, young, interesting people. And he said, you have resources and you have ideas and you have data and you're smart. Can I work with you? Can we work on this together? And he got Thorkild, Sorensen and other people to start working on twin studies with the great Nordic data, Sweden, Denmark and so on, and adoption studies. And it was the twin and adoption studies coming out of Sweden and Denmark that I think really nailed it. That finally in New England Journal of Medicine, with Mickey Stunkard behind it, with clear writing with high quality data, people said, we got it. There's a big genetic component.

And this was approximately what year, David?

This would have been 83 ish, plus or minus a couple.

So again, trying to bring this back to kind of the clinical side of things. So if by the early 80s should be patently clear that genetics are playing an important role, it would be at least a decade before people would say, well, pharmacologic therapy might make sense here. Just as, for example, if somebody is genetically more likely to have hypothyroidism, a very relatively straightforward endocrine system to understand, we wouldn't really think twice about replacing or abrogating thyroid hormone as necessary. So where was that disconnect? Where it's the early 80s and we realize there's a very strong genetic component to this. We probably have some medical tools that we could use. But it still seems like that wasn't being done. What was being done was it just a case of counseling people to eat less and exercise more.

It was mostly developing the tools that would help people reduce energy intake. And I think today that's still most of what we do. So even surgery and drugs mostly, but not only help people reduce energy intake, we know that reducing energy intake works. It's just really hard to do. And telling people to do it and saying, I'll try to do it is not all that effective. And so all these things were tools. And we used the tools of behavior therapy. Starting in, I would say really the 60s is probably where the formal behavioral stuff really kicked in with, I think it was Stuart and the behavioral thinking about obesity. And then it continued to get better and better all the way through, present and still continues to get better and better. But my sense is that we've been asymptoting for a long time. That is we're getting incrementally better, not meaningfully better with many of the behavioral cognitive things. What happened then in the early 90s is, and I remember because I was a postdoc at the New York Obesity Research center at the time, is a few people started to really kick in with pharmaceuticals and they started to come out of this idea that you're a diet doc and it's a bad thing. And fen phen, which now we know was dangerous and is gone. At least one of the fens in fen phen that came up and that caught people's attention.

Tell folks a little bit about fen phen. Many people listening probably won't be familiar with what it is, how it worked, and what the unfortunate consequence was in a small but non trivial subset of people.

Fen phen is a nickname for two drugs used in combination, phentermine and fenfluramine. Phentermine is a drug that was and still is approved by the FDA for the treatment of obesity, though it's never been approved for long term, so goes back a long way. It's a catecholaminergic agonist. It's a relatively safe drug. No drug is perfectly safe, and it's modestly effective. Fenfluramine is a selective serotonergic reuptake inhibitor, which I think originally was used for depression. And people realized sometimes Those drugs, those SSRIs, cause weight loss. Somebody realized you could put those two together and seem to get better results. So that became a big craze. And there were many unscrupulous medical providers who did provide that in ways they shouldn't have. But there were many scrupulous medical providers who used it carefully and thoughtfully. And it did seem to have much better benefit than other people had predicted before that. And so suddenly pharmaceutical treatment was no longer the realm of drive through pill mills and charlatans and diet docs and rainbow pills and so on. It was starting to be credible. You had people like George Bray, who's still around today and working on obesity, jumping on board and starting to really think through all the cornucopia of pharmaceuticals we knew about and which ones might be useful, others weighing in. Then a very smart young person, whose name I no longer recall, but was a pathologist, not an obesity researcher, as far as I know, at the Mayo Clinic, started noticing some people coming through as cadavers, and she was being asked to do autopsies and did autopsies and started noticing a peculiar valvulopathy that she didn't see very often. And she started to see a little run of cases and then noticed that all of these cases had been on fen phen, reported it, and then many, many other people started to investigate it. And very quickly people realized, in fact, this did produce a certain valvulopathy. And further investigation made it clear it was the fenfluramine component, not the phentermine component, that did that. So the fenfluramine was withdrawn. Phentermine is still used today. But I think what was terrible, of course, is that one of the many drugs in the history of obesity that hurt people, but what was good about it is that it really started getting people thinking more about obesity as a serious medical disorder that merited serious attention from credible physicians and credible scientists working at the molecular, pharmaceutical and physiologic level.

And how much of fen phen's mechanism was understood? I mean, obviously at the superficial level it was, but in terms of its direct action was the belief that one component was simply increasing energy expenditure while the other was reducing appetite.

I don't think it was quite that simple. And I don't think with any of these things we ever fully know the components. This is where you know far more than I. But I mean, even in the case of statins, which we think so highly of and use so well, whether they really have their beneficial effects by reducing LDL cholesterol, which is sort of the initial thought, or through other mechanisms, I think is being debated now. I think we suspected and still would believe that they reduced appetite and therefore help people control their food intake better, that the two things did it through slightly different mechanisms, one being more serotonergic, one being more adrenergic. They may have also had effects on energy expenditure, albeit probably modest ones. And that was probably more the phentermine than the fenfluramine in terms of the mechanism for the valvulopathy. I'll have to leave that to you. That's out of my realm.

When was fen phen told? Was that? The mid to late 90s?

I think it was the late 90s, but I don't remember the exact date. Probably 97ish. Somewhere around there.

What about gastric bypass and other surgical approaches? Now we have sleeves, roux, en y, all sorts of different approaches to it. But when did that sort of go from being presumably incredibly fringe to quite frankly, where we are today, where this is something that will be covered by an insurance company? If the bmi is high enough, I.

Don'T know that there was a single moment. So I think there's been a sort of a relatively steady march. But I do think there was one key step that was kind of a step function in the credibility of it, and that was the Swedish obese subject study as from Lars Jostrom in Sweden. Surgery had been around for a while. Interestingly, even among physicians and scientists, it was very controversial and still is. Some people think it's barbaric, some people, people think it's abhorrent. Again, I think you just have to take the data as it is. Say you might say, wouldn't it be better if we had a world where no one needed surgery, where our solution to obesity was not let children start to gain weight and then when they become massively obese adults, give them surgery and rearrange their internal anatomy. Yeah, probably would be better. But that's not the world we live in. We live in this world and in this world that's the most effective life saving, life changing treatment we have. And it's a good one. So what happened is many physicians would battle among each other. I remember John Krall, who's since passed on, a very good surgeon, another physician colleague, chastising him over dinner at a conference and saying, when you're done with a patient, Dr. Carl, they will never eat normally again. And he said, no, when I'm done, they will never eat abnormally again. So there were these real vitriolic battles. I think what happened is the surgeries of course got better and better and better as everything else. People learned how to do things better. Mortality rates went way down, efficacy went up. But what Lars Sjostrom did is he said, I want to look at whether this prolongs life. And I remember him telling me many decades ago that his senior mentor, Per Bjornthorpe, who was known for fat cell theories and apple versus pear idea, it was originally Jean Vague was apple versus pear from France in the 1950s. But then in the 80s, really 70s and 80s, BjornTorp was one of the big people who picked it up. Bjorntorp laughed at his mentoree. He says, Lars, everybody knows this, everybody knows that surgery will cause weight loss and help people live longer. There's no point in doing the study. And Lars said, you may be right that it does, but we have to do this study because everybody doesn't accept and believe that and it hasn't been shown. We know over and over again you've got to do the experiment. As John Hunter famously said to Edward Jenner When Jenner says, I think, and he's thinking about the first vaccine and Hunter comes back, says, I think do the experiment. Got to do the experiment and show it. Sjostrom didn't do a pure experiment. It's not a randomized trial, but it's a controlled trial. And the IRB at the time didn't think it was ethical to randomly assign people to either surgery or not. So if they're willing to get surgery, he would find a closely matched control and give them usual care. And then what he showed, well, more than a decade later in the New England Journal of Medicine, is that the surgery reduced mortality rate, it reduced obesity, very powerful effects. Clearly a life saving and a beneficially life changing treatment, I think that was probably the big jump. And then since then, many other trials, some randomized, have been done showing positive benefits on many things, including longevity.

Do you recall what the risk reduction was in all cause mortality in that study?

I don't recall in that study per se, but when you look across studies, it varies quite a bit from, from study to study. There's a lot of heterogeneity. It's usually well on the order of a 50% reduction, sometimes a little bit more.

Maybe getting too in the weeds on this. But how does that change when you do and do not correct for the presence of type 2 diabetes at the outset? In other words, I would imagine that that's a combination of people who had and did not have type 2 diabetes. And do you have a sense of what the differences are amongst those two cohorts?

Your premise is correct, that it is among people who do and don't have type 2 diabetes. But no, I don't know. Off the top of my head, I share your intuition that it's probably stronger among people who do. And we need to also look at many different factors. So, for example, in the Swedish obese subject study, and I'm not sure this would hold up at all, one of the things that was very puzzling and surprising, at least to me, was that although diabetes really came down and stayed down very well, even if weight came back up a little, which it did on average, hypertension came down very well, but did not stay down very well, it would come back up. Why that is, is it damage to the endothelial elasticity, that's not really repairable and so you get a short term effect of negative energy balance that's not sustained, I don't know. But it goes back to that John Hunter thing. You've got to do the experiment. We can't make a priori assumptions about the effects of treatments. We've got to do the experimental look at the effects of treatments.

It's interesting that hypertension would return and yet mortality would still improve. Which suggests that perhaps the benefits of improved insulin sensitivity, which do persevere, play a greater role in mortality than body weight, which does not surprise me. But also hypertension, which is somewhat surprising given how causally related hypertension is.

Atherosclerosis, well, it may not be a greater role as much as an additional role. It could be that if you got the hypertension down, you'd get an even bigger reduction.

When did this idea of an obesity paradox, which I know you've written about, when did that start to become observed?

That phrase obesity paradox has never been really crisply defined. And whether it really is something that's a paradox is not clear. People use it to mean at least two different things. One thing is that there's this so called U shaped curve, more accurately concave upward as a bathtub shaped curve. If you imagine a Cartesian or XY plot in which the Y or ordinate is mortality rate or some function of it, and the X is BMI or relative body weight or something. So one is that it is U shaped, that it's not monotonic increasing. So people at the very thin end also die earlier than people in the middle, just as people at the very heavy end die earlier. The Duchess of Windsor is supposedly said you can never be too rich or too thin. And she may have been right on the rich part. I don't know, haven't gotten there yet, but apparently not on the too thin part. Now we can argue whether it's causal and there are lots of arguments about whether it's causal, but that's one thing. It's the idea that thinner people than sort of intermediate levels of BMI also die earlier. That's one part. The other part is that even though obesity is associated with increased mortality rate or decreased longevity, that when you look at people who are sick or injured, they often live the longest. So somebody comes in with kidney failure or, or someone comes in after a major injury or a major infection, they're in the hospital, it's often the heaviest people that live longer. And that started to be talked about 10, 15 years ago maybe. It's very difficult to disentangle cause and effect. We can observe lots of associations, but it's just hard to know what to make of all of these associations and what's causal, as the philosophers say. The hypotheses are under determined by the data. There are multiple hypotheses that are consistent with the data. That's why the randomized experiment allows us to do things that otherwise we can't do. Sort of eliminates more competing hypotheses. So I don't think we know what's causal yet. We've thrown out a model man named Doug Childers. And I, who's a postdoc, was a good mathematician working with me in which we said, what if obesity makes it more likely that you get a major illness or injury? So it's not good for your health in that sense. Age, of course, also makes it more likely that you get a major illness or injury. But once you get a major illness or injury, being heavier, more obese may reduce your risk of dying from it. And an analogy I use is this. Suppose you and I are sort of going for a hike on the edge of the Grand Canyon and we go to some outfitter who's setting us up and he says to you, who's. You're about 10 years younger than me, I think, and he says to you, you have a choice. I can give you this fat suit you can wear, has lots of padding on it. If you fall off the cliff, it makes it much more likely you'll survive, but it makes you a little clumsy, so it makes you more likely to fall off the cliff. You might say, I've got good balance and good eyesight and I'm young and strong. And no comes to me and I say, I'm not quite as young and strong, maybe, I don't know, comes to me, somebody 10 years older than me, that person says, I'm really struggling with my eyesight or balance or strength. I'll take the fat suit. And so whether the fat suit is good for you may depend upon your probability of falling to begin with. And under that mathematical model, we can show in fact that the point of bmi, the nadir of that bathtub shaped concave upward curve should keep moving to the right as you age, which is exactly what it does. We have a model that's consistent with the data, but the data don't prove our model. There are other models that would be consistent with it.

Let's give people an idea of what those numbers look like. So just for someone listening to this, maybe and not even watching, it can be hard to imagine what we're talking about, right? But a U shaped curve that at its nadir is giving you the optimal bmi, this would be the lowest mortality and, and what you're saying is, rather than just plot one of these curves for everybody, let's plot them by decade. So what does the U shaped curve look like for people in their 20s, in their 30s, 40s, 50s, 60s, up to their 90s? And what you're saying is, if your hypothesis is correct, one place you would expect to see that is that these curves would not only not look the same by decade, but they would move rightward, meaning the nadir is getting to a heavier and heavier bmi. Can you give me a sense of how much movement there is by decade?

Yeah. In fact, I'm going to smooth over things. So the story I'm about to tell you is not perfectly true, but it's sort of a roughly true, and it conveys this sort of nice element of this. The average American might gain about a pound a year. For an average height American, about six or so pounds might be a BMI unit. If you gained about six pounds, you might gain about a BMI unit. And so that might mean that over six years, you'd be about one BMI unit heavier. And that's not too far off from how the nadir moves. It's almost as though your weight is increasing to keep you at the nadir, which is an interesting speculation. And if you looked at people who are maybe 20 years old, that nadir might not be too far from 20. In some populations, an astute reader of the epidemiologic literature who's listening now might appropriately be saying, hey, wait a minute now, Allison needs to make distinctions by age, race and sex. He's doing age, he's not doing race and sex and other factors. And they'd be right. We can come back to that if you wish. But putting that aside for the moment, very loosely speaking, you might say that nadir might be far down near 20. When you're 20 and by the time you're 80, it's not to 80, but by the time you're 80, it might be well above 30, might be in the low 30s where that nadir is, which is, we generally say 30 is the beginning of obesity. And for those who are not used to BMIs, I used to lay it out like this. The supermodel Kate Moss, at least from a few decades ago when she was the top model, had a BMI, I think in 16 to 17 range. We say that about 18.5 is sort of the beginning of normal weight. We think less than that is underweight. My BMI right now is probably 21ish. Bill Clinton's the height of his presidency since lost weight was probably 27, 28, we say 30 is where obesity begins. And a secretory sumo wrestler, top class sumo wrestler, about 43, 44. So that gives you a sense of what those BMIs are. And so it's really 30, 32, where you see people 70, 80 and above having that lowest mortality rate.

Is there a risk that we are looking at confounders here? I don't know if this would be a confounder. It wouldn't seem like it would be a confounder in this year, meaning in this time and place we're in. But would it be a confounder for affluence, for example, as it may have been hundreds of years ago? Other things like that. And the other thing about BMI is when you look at other data that look at body composition, do we see the same thing hold up? So you can have a BMI of 30 even though you're obese by BMI, no one would look at you and say you're obese. You might have a body fat of 15%, which is pretty low, and be incredibly muscular, for example. So how do we reconcile one the potential confounders of that type of an analysis with this other next layer of granularity in the data that look at adiposity versus lean mass?

Let's start with the confounding question. Now let's get to the body composition question. The confounding question is a nightmare. This is the challenge of observational research in general. And there's no simple solutions to it other than, in my opinion, trying to triangulate on the answer with multiple studies. What we really want is the pure randomized experiment in humans in large samples with perfect compliance for many years. And we want to randomly assign people to be obese or not obese. Obviously we can't do that. So what do we do? Well, we look at observational epidemiology, we look at randomized trials where we have people lose weight, we look at non randomized trials like Sjostrom's with surgery, we look at mouse studies, we look at all these different things. We try to just put the puzzle together as best we can, but there's absolutely a lot of possibility for confounding. Cigarette smoking was one of the early ones, very famous paper by Joanne Manson and colleagues published in jama, pointing out that the quote right way, according to them to analyze BMI and mortality data was you must exclude ever smokers. You only look at never smokers. Otherwise you may have confounding by smoking. Smoking makes you thinner. Smoking kills you earlier. You've got to take care of that. You've got to throw out the subjects who die early because subjects who die early may have been sick. Sickness makes you thinner, sickness makes you die earlier. That is a confounder. We've since proved mathematically that that's not a good thing to do, but for a long time it was believed.

Is that even true, David? On the low end? Because that does strike me as one of the most obvious explanations for the uptick in mortality at very low BMIs is all of the liver disease, the kidney disease, the types of chronic diseases that actually do lead to muscle wasting and things like that. Are you also including them in your analysis?

Yes, we are. What we showed is not that that confounding doesn't occur, that confounding absolutely can occur. We agree that it can occur. We speculate, we think it's likely. What we've shown is that throwing out subjects who die early in the analysis doesn't help or doesn't reliably help. It was very interesting. Back to sort of the discussion of the field. When I went to analyze my first BMI mortality data set. I had no training in this. So I get a data set and I'm about to try it. I don't know what I'm doing. So I call up my friends and I say, I read this paper in JAMA and it says, you have to eliminate the early deaths. How do I do that? How do I pick how many years to do exactly? How does this work? And it was interesting because when I would call the epidemiologists and I would say, how does this work? And then I'd say, by the way, is there a reference that explains and proves that this works? They'd say, no, it's obvious that it works. I said, well, it's not obvious. And they said, well, maybe you're not smart enough to be an epidemiologist.

Those words were actually uttered jokingly.

But, yeah, then I would call my friends who were statisticians, and I would say, there's this paper and it says you have to eliminate the early deaths. And they would laugh at me, like, what? That's the most ridiculous thing I've ever heard. We don't just throw out data and think it makes things better. You've got to have a model that you fit things to. We then got a small grant from the CDC and we looked at mathematical proofs, computer simulations and meta analyses, and the mathematical proofs showed it not only didn't have to help reduce the confounding, it could make it worse. Simulations showed the same thing in realistic data scenarios. And then the meta analyses showed that on average didn't make much difference at all. So throwing out the subjects who die early basically just reduces your power in most practical situations. Then they would say you've got to not control for the intermediary variables like the diabetes and the hypertension. That's probably true. I think I would agree with that. Later, in subsequent paper in 1995 in New England Journal of Medicine, that same group said you've got to throw out the subjects who have more weight variability. All the things they said in 1987 weren't enough to flatten out that left end of the U shaped curve. So they said throw out the subjects who have a lot of weight fluctuation. And that didn't completely get rid of it, but got rid of most of it. That's the bigger one. But what it means is unclear. Katherine Fleagle has written very well on this and talked about these different things and how much you're throwing away and pointing out, yes, these are the patterns we observe, but what they mean causally is unclear. So that's the whole confounding issue. The bottom line is it's likely that there's confounding by cigarette smoking and socioeconomic status and stigma. That's one of the big things, right? How much is does obesity kill you because it stigmatizes you and it creates some stress? Could be. And that may be, by the way, why the BMI associated with the lowest mortality has been increasing over calendar time. It's not just with age of individuals. But if you compare data collected in the 1970s to data collected in the 1990s, obesity doesn't, from an association point of view, look quite as bad in the 1990s as it did in the 1970s. And that's true in Denmark, and that's true in the us and it's true in meta analysis, although again, not for every age, race, sex, group. Why is that? And maybe it's stigma, maybe it's that if you think about, you watch like the Three Stooges, those old TV shows, and there was Curly and he was often mocked as the fat guy. Right. By today's standards, he's not that big. What stigmatized changes over time as different BMIs become normative. It may be that that's partially accounting for it. So there's lots of potential confounding going on there, but there's Lots of other possible explanations. So bottom line, we don't know. And the real important question then is what's the effect of intervention? This goes back to my friend and someone I really idolize in the field. One of the clearest thinkers, Don Rubin, statistician at Harvard, talks to me about the Rubin causal model. And he's always asking, what's the intervention? If you say, does this cause that, what do you mean? Compared to what? It's always got to be compared to what? Right. And so if you say weight loss is going to increase longevity, that's the question. Well, how are you going to achieve that weight? What are you going to do to get that? Well then if it's surgery, then what's the effect of surgery? It's a GLP1 agonist. What's the effect of a GLP1 agonist, et cetera. So that I think is the key question and I think what we're starting to see is some of those things do Prolong life. Surgery, SGLT2 inhibitors, GLP1 agonists, and so forth. Now to your other question about body composition. Many people like to point out that BMI is a measure of mass divided by stature. It was developed by Adolph Keit le, who was a Belgian astronomer, epidemiologist, statistician, mathematician back in the 19th century. Brilliant guy. We actually named a professorship after him and I held that title. I was very grateful for that opportunity. Back when I was at the University of Alabama at Birmingham before I came to iu, Kate Lay made it. You would think that BMI, or rather the mass of a three dimensional object ought to increase in proportion to the cube of a linear dimension. And if in fact we were spheres of uniform density, that would be true. But I don't know about you, I'm not a sphere of uniform density. So it turns out not to work that way. Exactly. Turns out that empirically it works closer to the square for adult humans. Kate Lay realized that and so he said, take mass or weight over stature squared, the square of stature. Then it was called Kate Lei's index. Now we call it BMI. It was sort of rediscovered in the 60s or 70s by Ancel Keys and term BMI, body mass index. And every few years some smart person likes to come along says it should be cubed. Yeah, yeah, yeah, we get it. We knew that a couple hundred years ago and we worked that out. Then they say, and it doesn't really take into account body composition, we say we know that. And they say the NBA center would have a BMI. Greater than 30. And yet look how strong and fit that NBA center is. We go, right, if the NBA center comes to your clinical practice, don't measure his BMI and diagnose him with obesity on that. And what physician in his or her right mind would do such a thing? So clearly we get that BMI is a useful tool for epidemiologic research and some simple physiologic research and some simple clinical trials. It's not a perfect clinical tool for the average person. It works. Okay.

Now, do we have to be worried about different ethnicities? Obviously, the two that come to mind would be Asian and East Asians, because I think most people are kind of familiar with this idea of skinny fat. So you have someone of East Asian descent whose BMI is 26, which by all intents sounds just wonderful, but they have metabolic syndrome. There's nothing about this person that's healthy. Every metric across the board, tons of visceral fat, very little muscle mass, et cetera. And that's a different phenotype. That's a slightly different phenotype than perhaps the model was built on. So how much is the metric that we use for BMI? 20 to 25 being perfect, 25 to 30 being overweight, 30 to whatever it is, 40 being obese, and then morbidly obese kicks in at some point, I'm sure. Is that validated on other ethnicities as well?

Yes and no. So with a few exceptions, the idea that there is a curve and that it's a generally concave up curve. Yes. Although again, there are exceptions, but the shape of that curve is not the same in every age, race, and sex group. It does seem that the right side, the part where you're getting too high in BMI and risk is going up, up, seems to occur a lot earlier among people of Middle Eastern and East Asian descent. So that's a very simple model, just sort of says, oh, it's all a curve for everybody. A slightly more complicated model says, well, we know some of the other factors that come in, like how much of your fat is subcutaneous versus how much is visceral. And it's just that that group has more visceral fat than that group for any given body mass index, and therefore adjust the curves. And there's probably some truth to that, but it's probably even more than that. So, for example, a lot of these statements we hear, and this goes back to way back when I said way at Vassar College. I loved the complexity of it that you had to look at it from obesity from so many different angles. And that's still true. The narratives we have about obesity, including about ethnicity and obesity, are grossly oversimplified. So we often hear obesity is selectively a disease of the poor and uneducated. That often in this country it's stated minority status leads to less income, less education, which in turn leads to poorer access to health care, poorer habits, poorer living environments, which leads to poorer health and reduced longevity. And there's probably some truth to that, but it's not one to one. It's not simple. So, for example, when we hear that there's this inverse relationship between socioeconomic status and obesity, which we hear over and over again, first shown by Mickey Stunkard, again in the 1950s in the Midtown Manhattan Project, that's true in white women. It's reliably true in adult white women. When you go outside the group of adult white women, not always true. If you go to African American women, virtually no association between socioeconomic status and obesity. When you go to African American men and you look at their obesity levels, they're very similar to the obesity levels of European American men, white American men. But if you look at African American women's obesity levels and you compare it to European or white American women, much higher. If you bring Hispanic Americans in, both men and women have higher levels of obesity than do European American men and women. So now you have an ethnicity difference, but not a gender by ethnicity difference. Whereas in African American you have a gender by ethnicity difference. When you look at mortality, the African American curves follow similarly but not identically to the European American curves. But in our research, we can't find an association between BMI and longevity in Hispanic Americans.

Wow, just say that again. That's pretty interesting. That's across all BMIs, we've been unable to find it.

Now, the data sets we have are not perfect. The follow ups may not be long enough, the sample sizes may not be big enough. You can always question things. I don't say it's causal. I'm just saying this is what we observed a few years ago when we got every data set that was publicly available that we could get our hands on that involved Hispanic Americans and BMI and longevity. And we analyzed them all together with common methods as carefully and thoroughly as we could. We could not find an association between bmi, elevated BMI and mortality.

As someone who sits on the opposite end of the spectrum, right where I don't at all concern myself with public health, I'm not trying to make grand observations about what's happening in society, across ethnicities or anything like that. I really just have the luxury of looking at one person at a time and trying to make a determination about the best course of action. My view of BMI is generally quite negative. It's maybe the least bad tool available to get massive data sets and make broad assessments of what's happening at the population level. But to your point earlier, at the individual level, it offers very little insight relative to other tools. When we have patients do DEXA scans, I tell them that we're going to get four important pieces of information out of this, but one of them's really the least important to me, which is your subcutaneous fat. That's the first thing people want to look at when they have a DEXA scan. They want to see what's their percentage of body fat. And they say there's three things that are much more important to us. One is your visceral fat, one is your bone mineral density, and one is your appendicular lean mass index. So how much muscle mass do you actually have? And we'll look at your subcutaneous fat. But the reality of it is that doesn't really seem to matter because a, that seems to be incredibly genetic body habitus in that regard. Very genetic and relatively uncoupled for metabolic health. So the vat, the visceral adipose tissue, seems to be much more tightly correlated to what we see when we look at more sophisticated biomarkers of insulin sensitivity and metabolic health. Any evidence of liver fat, all these other things tend to track much more closely with that. So my view on this is that I'm glad that I get to look at these other metrics. I have the luxury that a statistician or an epidemiologist doesn't have, which is when you do things at the individual level, you have much more data and you can be much more nuanced in your appreciation for things. But I can't help but wonder if at the population level, something better than BMI will come along one day for which there will be enough data to actually do what's out there. And again, this example with the Hispanic subset, that's mind boggling to me. And it really speaks to, I think, the futility of that measurement in other populations.

That is one plausible interpretation. And I don't think we're far away from these better tools. We published some work years ago on the idea of using 3D photography. Take a picture of somebody from a couple of angles. This came out of work experience way back at the New York Obesity Research center, where Steve Heimsfield, who was my mentor, would study professional basketball players. They would bring the teams in from New York to him. He was the king of body composition, and he would study them. In talking with some of the technicians there who did this every day, who measured people's body composition every day, they would say, I can look at a person and tell you how much fat they have, and I'll be very accurate. They were good at it. I said, well, if you can do it with your eye, why can't my camera do it? We got an NIH grant with Olivia Fuzo and I and Steve Heimsfield to look at this. Published a paper on it. Since then, many others have done it. And I think Amazon may be working on this or already have something out on this. So I think we'll get to the point where we have 3D photography. So that's Archimedes Principle. Eureka. I found it. The crown goes under the water. He knows how much gold is in it because he knows how much water it's displaced. So if you can weigh it and you can measure its. If you can measure something's weight and volume, you know its density. If you know a little about human anatomy, you can figure out from density, body fat. That's Archimedes principle. That's what we can do with a camera. So we have that. We have dexa, we have plethysmography, we have isotope dilution techniques. All of these can start to be used more en masse, on and on. And so I think these will be coming, and we'll do better with that. But I also think there's this idea of fit for purpose. If you were to say to me, is my vehicle that I drive, is this precision enough for the Indy 500? No. I mean, they need better vehicles. But I'm not in the Indy 500. I'm just driving five miles to and from work every day. It's fine. And so I think it's fit for purpose. If you say to me, what proportion of that country has obesity, BMI is probably perfectly reasonable. If you say, I just want to know on average, what's the approximate correlation between obesity level and whether you play golf or not, that's probably okay. But if you want to say, I want to help this individual patient, especially at the sort of the kind of artisan level that you go at, then you need better tools.

Well, the other thing that comes back to your friend Reuben is you're still limited in the what to do the now what question? So if you say, well, we're going to take our tool of BMI extrapolation and we're going to look at the state of Indiana, we can pretty accurately probably plot out the histogram of exactly what the BMI is by age and by demographic and all those things. It's the so what question? Well, what do we do with that information? What's the implication of this? Are people in Indiana more or less healthy than people in Kentucky? And are they more or less healthy than the average American? More importantly, are they more or less healthy than the average person on this planet? And what can we do to improve their health, even if we don't want to compare them to anybody other than an absolute standard and say, could everybody become healthier? Could we extend the life of the average person in the state of Indiana by three years? What would the intervention need to be? And that obviously becomes a much more complicated problem. But it's a more germane question, right?

And it's not a fanciful question. It's a question that we literally are asking. So this becomes tough because when I talk to people, even physicians, even highly educated scholars, there's this bifurcation, there's the data. And those who look at the data and if they really know the data and they're really honest, they say, we agree, surgery works, pharmaceuticals work, individualized or group based clinical treatment with behavioral cognitive techniques works somewhat for some period of time. Meal replacement form formulas work somewhat for some period of time. But all the public health stuff we've tried. Now if you really are honest and you really scrape away the obfuscated data, there's virtually no evidence thus far that any of those community, school, public health things for obesity work. Doesn't mean that we won't get them someday. But right now they don't. So when people come and say, well, what do we do about the state of Indiana? And I get this question a lot. What's your goal? Is your goal to expose people to ideas so that there's some really smart kid, the next Marie Curie, the next Einstein, the next George Washington Carver is there in that classroom and maybe the thing you do to try to reduce obesity has no effect on obesity, but that kid is thinking about about it for the next 15 years and then when they become an adult, they go, I got an idea. And suddenly someone smarter than we are or with new knowledge cracks it. Is that your goal? Then it's consciousness raising. Okay, fine. Is your goal to say to communities, we know you're suffering and we Know this is concerning and we want you to know we care and we're trying. We're not really going to necessarily reduce obesity levels. Want you to know we care. And you want farmer's markets in the school parking lot, you want vending machines changed, you want running tracks built in your neighborhood, we'll build running tracks and so on, and we'll feel better about we're caring for each other, but it's probably not going to affect obesity given what we know right now. Or do you say, I actually want to have less people suffering with and from obesity. I want obesity levels in some definable, countable number of people to go down and I want their health to improve. It's not going to win you any feel good awards. But surgery, pharmaceuticals, and to some extent individualized treatment, cognitive, behavioral, group based treatment, including things like meal replacements and so on, those are ways to go. You could take. If the state of Indiana handed me $10 million and said make a difference, I would not go build farmers markets in schoolyards. I would say it'll only be a small number of people. But let's give bariatric surgery to a subset of people and those subset of people will likely live longer on average.

What's the best explanation for why weight loss is not particularly difficult? But weight loss maintained is incredibly difficult?

There's probably no single explanation. And I think that question of why is a tough one. Do we mean evolutionarily why that is what happened in evolution that got us to be what we are today that leads to that, or is it biochemically and otherwise why that is? What are the mechanisms from either point of view? I don't know all the answers from the evolutionary point of view for a long time the meme was it's the thrifty gene hypothesis from James Neal. Right. So the idea is that animals in general and including, and especially humans throughout evolutionary history were on the brink of starvation. Anything you could do to preserve energy, you did. And anything you did that when given the opportunity to get more energy, eat as much as you can while you can, and then you get into this modern environment where there's for practical purposes, for most of us, unlimited consumable energy you over consume. I think that's simplistic for many, many reasons. First of all, as the lawyers say, objection. It assumes facts not in evidence. That is, it's not at all clear that humans have been on the brink of starvation throughout history. In fact, Robert Fogel, who won the Nobel Prize for looking at these old data going back to by at least the 1700s of British naval recruits. And other places you see BMIs sort of on average they're going up over the centuries. But there's a little fluctuation as things get better and worse in places. Second thing is, how does this account for pregnancy? Maybe now the latest data from John Speakman and colleagues in science, with all the doubly labeled water about three months ago, maybe suggest that during pregnancy, women's energy intakes don't need to go up that much, but they still go up. And so if humans have been reproducing for millennia, where did all that extra energy come from if we were all on the brink of starvation all the time? And then the last thing I'll say is anybody who ever goes fishing knows that the idea that every animal is hungry all the time and is going to grab every bite of food you throw in front of it. Never been fishing. You can see that beautiful bass sitting in the clear water in front of you and you can dangle your worm or killie or whatever it is you've got and sometimes the fish just looks at you. So it's not at all clear that this is the case. Some animals do get obese when given unlimited food, some don't. Both within and across species, there's lots of differences. So that's that. John Speakman came along and he said, I'm not sure I'm buying this whole idea. He said, I think it's freedom from predation. Back in history, we were prey. Then. There was a certain point where we learned to use tools and hunt together and we stopped being prey and we started being predators. And when we switched from being prey to predators, then we didn't need to hide in our burrows and eat the least we could, because every time you came out of your burrow, you were prey, potentially exposed to a predator. We could sort of walk around and eat kind of ad lib. And in that case, the genes that were being selected for gave us satiety mechanisms that kept our weight down no longer were being selected for. It wasn't that nature was selecting for genes that made us fatter. It just wasn't selecting for genes that kept us thin. And then what happens is what's called drift. Mutations happen and things just drift.

It's less about the thinness. The thinness was really a consequence of what the genes were probably selecting for, which presumably was lower appetite or something like that.

Probably. There's not one factor. We see this, for example, in the evolution of sexual reproduction, which is called the queen of questions in biology. Nobody can really figure out why do we have sexual reproduction when asexual reproduction seems so much more efficient from a genetic fitness point of view and people have proposed different hypotheses for it and no one seems to work mathematically. And what it may be is that it's only by putting them all together that it mathematically works and it's just an inelegant solution. You see similar things in physics where it may be, you know, these beautiful simple mathematical things may not hold up. You may just have to have ugly composite theories.

Say a bit more about that. That's interesting. I've never heard the argument. I don't actually know much about asexual reproduction. I don't spend much time thinking about plants or other life forms that do it. But what's the argument for why we would be better off with asexual reproduction?

Well, think about something like Daff, which is a species that can produce both sexually and asexually. There are many species like this all the way up through some vertebrates. And if an organism reproduces itself asexually, just sort of makes a copy of itself, think about what it's done is it's reproduced all of its genes. And so from the Richard Dawkins point of view, the selfish gene, those genes all got their way, those genes all won, they got copied. And genes that are good at getting copied will get copied again in the future. Right? So you have more of those. That's how evolution works. Whereas if you reproduce sexually, you get another partner. And of course this assumes other. This invites other questions like, well, why are there only two sexes? Why in fact have sexes at all? You could exchange DNA without having sexes. Bacteria do it through conjugation. But we'll put that aside for the moment. Say there are two sexes, male and female. They come together, they mix up their genetic material. The offspring has roughly 50% of the genetic material of one parent and 50% of the other. And you only copied yourself one half. And so you didn't win as much as if you copied yourself entirely. So why would you ever switch to sexual reproduction? It's very inefficient from a genetic fitness point of view. Now you can hypothesize things. The most compelling hypothesis I've heard to me is the so called Red Queen hypothesis, which is from Alice in the Looking Glass, where the Red Queen is running with Alice. And Alice at one point says, we don't seem to be getting anywhere. And the Red Queen says, oh, in this world you have to run as fast as you can just to stay in place. And Alice says, oh, my world. We run and we actually get somewhere. And so the Red Queen hypothesis is the idea that you keep running as fast as you can just to stay steady. What does that mean is, well, as you're living a long time as a human, there are these microbes in you, and they're evolving much more rapidly because they have a much more shorter generation time. And as they evolve, they start to get good at getting past your defenses and your locks. They start developing keys to your locks, and you want to reset the locks. The way you reset the locks is by getting a partner and mixing up your DNA with them. So the idea that it's a way of keeping up with the Joneses, where the Joneses are the bacteria, the microbes in your body, that's called the Red Queen hypothesis.

That's super elegant, right? Because if you think about it, if it was all asexual, we'd have a population of identical people. I mean, we would, for all intents and purposes. It'd be interesting to do the math on what it would look like, but you might only have a few hundred thousand gene pools.

You'd have much less diversity, much less diversity. From a species point of view. If you believed in group selection, then you say, oh, right, it's good for the species. But the smart evolutionary biologists come along and say, yeah, but that's group selection. It doesn't make sense. Selection occurs at the individual or gene level. You've got to explain how it makes sense for that individual, how it enhances their fitness or their genes fitness. So you say, okay, well, if only half the genes get reproduced, then it's got to be double the fitness level to break even. That doesn't seem like it really holds up. So what people have said is, you know what? If you take a handful of things, which I can't explain them all right now, there's Mueller's ratchet, and there's this and that, and you take all of these things and you put them together, then maybe the math works. But it's very inelegant. It's not one nice little theory. And it's probably the same thing with people and evolution. So there was Neil with the thrifty gene, that is, you need to be selected to get food when you can. There's probably some truth to that. Maybe not everybody who's dying of starvation, but if you're not getting enough food, you may not be big enough to win the battle. For mates in a polygynous physical combat mating system that may select for wanting to eat more, you may not be. Think of the Frisch hypothesis. You get too thin as a woman, you stop menstruating. So it may be that it's not that you die of starvation, but that your reproductive fitness goes down. On the other hand, the predation, the freedom for predation idea that John Speakman puts out is also legitimate, and there are yet others. Gary Beauchamp from Monell Chemical Senses center has talked about the idea of the safety of food. Is it possible that if you're back in time and you're not eating out of a refrigerator in a modern, safe food supply like we have, you know, we love to consult our food supply. It's probably the safest food supply in history. Every time you eat something, you're exposing yourself potentially to microbes and toxins, not just predators. If you eat less, you're less exposure. So there's again, there's an optimization problem. But as you now have a safer food supply, you can relax that constraint a little bit. You can even think about it socially. If I were in a species where I'm just out on my own, if I'm a species that just eats eucalyptus leaves or something and doesn't depend on anything else, then maybe I can eat the last eucalyptus leaf. But if I'm a species that very much depends upon cooperative living, if I am so hungry all the time that I'm willing to try to kill you for the last bite of chicken and you're willing to try to kill me for the last, last bite of chicken, that's a bad situation, especially for me, because you're a little bigger than me and did more martial arts. So that's not good for fitness. It might be good to have satiety mechanisms just so we can preserve some social order. So after you and I each eat a little chicken, we can actually work together on building tools and engines and steam engines and airplanes and so on.

This is super fascinating. I could go down this path forever. I think we'll come back to this next time we have dinner, because having this discussion over a meal would make it even better. Let's march on to the topic of nutritional epidemiology. We've talked a little bit about epidemiology. Obviously, epidemiology is married very closely with statistics. Without statistics, you can't really do epidemiology. But it's a field that I think even the casual listener of this podcast will understand has its limitations. If by no Other means than they've heard me rail on it many times. So I don't think we need to define it. I think people understand the nature of it. But let's talk a little bit about your views of it, because I think. I wouldn't say you're in either camp. There's a camp of people that would say there is absolutely nothing wrong with epidemiology, nutritional epidemiology. It is a masterful tool that provides exceptional insights without which we would be lost. The other end of the spectrum, I'll acknowledge I'm a little closer to the other end of that spectrum. There are people who say this is a tool that has probably reached its peak of utility. The epidemiologist should probably focus on other problems outside of nutrition. Now, you're probably more in the middle of those, but I'd like to hear you talk a little bit about the bookends and how you settle out and why.

More importantly, I appreciate the opportunity to comment on this. It's a topic that. That I feel really does need more courageous addressing. I think that you have characterized it well. There are at one end the people I would call the abolitionists. I agree you're close. John Ioannidis, another mutual friend, who said, nutrition epidemiology is a dead science and it's time to bury the corpse. Gary Taubes has been very critical. Nina Tiekolz, you, many others have pointed out that perhaps this is just a worthless waste of time and misleading. At the other end, there are a number of people, they tend to be concentrated in Boston.

It seems there's a little school out east there. I forget the name of it. It's in Cambridge. Is it?

Yeah, I think I've heard it. I think I've heard of it.

Yeah.

There are other places that believe this, but it's probably strongest in Boston. That sort of seem to be the defenders of the status quo that say nutrition epidemiology is imperfect, as all tools are, but it is still a very valuable tool. There's nothing seriously wrong with it. Those who criticize it are naive. I look at this, naive and ignorant. And I look at a quotation like that and I think, really, you're telling me that Gary Taubes and John Ioannidis are naive and ignorant. If you want to tell me you don't agree with them, that's fine. Naive and ignorant. Come now. Not really. We know that there's something wrong there. When we look at the evidence, it's very clear that many findings from nutrition epidemiology have not held up once we've done randomized controlled trials. Now some people will say, no, you're wrong. And I've had arguments with people from Boston about this and they said you're wrong. If you look at these meta analyses, it looks like non randomized studies on average give very similar findings to randomized studies. But often they're citing non randomized intervention studies, non randomized observational studies, and they're often not citing nutrition studies, they're citing pharmaceutical or other medical treatment. So those are very different situations. When you look at the nutrition epi, it's not clear to me that these things hold up very well when they are studied. In fact, seems to be more the norm that they don't. If you look at other things going on as John Ioannidis probably better than anybody else has done, but many of us have done and you start to peel the hood open on these and you really look, you see a lot of things that look like obfuscation, exaggeration, sweeping under the rug of measurement error and so on. So we've got huge issues with confounding, we've got huge measurement problems. And so I think many people, including me, say no, the status quo is not okay. It's not just minor fine tuning. We need reformation, but we also need to use these tools. Well, this field is not going to go away whether you want it to or not. So my feeling is reformation is essential. The status quo is completely unacceptable. But abolition is neither realistic nor, nor desirable. And I want to cite one paper we did which I think makes an interesting point. This paper, you know, I often end reading the Observational Epidemiology Nutrition study in which the author correctly and honestly points out that it's an observational study, has association, not necessarily showing causation, and then says, but how could I be wrong? Well, it could be that this measurement error creates this problem. They say, but I used a validated food frequency questionnaire, so it's really okay. Could be confounding due to socioeconomic status, but all of my subjects had the same profession, so it's really okay. As though the nurse who works in a poor school in rural Indiana has the same socioeconomic status as the nurse who is married to a billionaire. They say we control for we only looked at never smokers and so on and so on. So really it was okay. They dismiss the idea that it's not really causation. I got this idea once and I said, you know, what they're saying is if I measured everything well, and I controlled for food intake and I controlled for diet composition and housing and socioeconomic status and genetic background, then it would be okay. So I said, you could never do this study in humans. But I realized I've done this in mice. And we had a mouse study where we randomly assigned mice to eat different amounts of calories or food energy, but all on the same diet. So composition was fixed. The mice had no choice. There were no restaurants, there was no grubhub, there was no doordash. We gave the mice the same food, they all ate the same thing. No self report. They're all genetically identical in probability because they're inbred. Isogenic strain, all C57 black six chain mice. They're all in the same housing conditions, there's no smoking. And then we take these mice and we randomly assign them to low calorie, medium calorie, high calorie, effectively, or ad lib, ad libitum. And what we find is the more calories we assign them to be allowed to eat, the shorter they live. No big surprise there. This has been shown a thousand times in the literature by Ray Walford and Rick Weindrucht and others over the the decades. But then we take the ad libitum group, and within that, some choose to eat more than others. And within that group now we have an observational epidemiologic study, and we correlate amount chosen to be eaten with longevity. And those mice that choose to eat more live longer. So the association in the observational component is exactly opposite to the causal effect in the experimental component. And what people say to me is, well, David, it's confounding. What you're seeing probably, almost certainly is that the mice that are the strongest and healthiest have the biggest appetites. They eat more. And what you're seeing is confounding by general health. To which I say, right, that's the point. The point is, even in a study, an observational study that is more pristine than anybody will ever be able to do, a human study, which there's no smoking, no restaurants, everybody eats the same thing, there's no measurement error. They're all genetically identical in probability. We can't get the observational study to reproduce causal effect. And that suggests to me why reform is so needed. And we talk about different kinds of observational studies. What I would love to never hear again is the puerile analogy of there's no randomized controlled trials with parachute jumping, you know what I'm talking about? That comes up every now and then. And there's a wonderful book called Randomistas, like, you know, Fashionistas Randomistas by Andrew Ley. And he says, actually there are randomized controlled trials of parachute jumping. So first of all, it's just not true. But beyond the fact that there really.

Are randomized, how did the IRB approve.

That you got to be in the army? I think the interesting thing there is that this is used often as an excuse to say, okay, I can't do the pure, perfect, pristine randomized control trial. So therefore, you have to accept any old observational study. And I think the answer to that is no. No, we don't. We may have to accept that we're going to draw some inferences from something other than the pure, classic pristine randomized control. But in between that and any old observational epidemiologic study, there's a lot of space. And what we're saying is we need to get to this space here. So co twin controls. You hinted at that earlier. How about we take your identical twin, we randomly assign you to one and the twin to the other? Or if we can't randomly assign, say, oh, Peter, you exercise a lot. Your twin brother doesn't exercise, but you have the same genotype. Okay, that's a tight control. How about things in which we intervene even if it's not randomly intervened, as opposed to just observe? So we say, okay, in this town, we're going to build a restaurant, in that town, we're not going to build a restaurant. It's still an intervention study even if it's not a randomized study. That's the realm of Brian L. Bell, for example, looks at food deserts and things, Gee, built a grocery store where there was a food desert. And we're told that food deserts with a problem and doesn't look like things got better when we built the grocery store. That's a much stronger design than just asking people, how far away do you live from a grocery store?

I think that's very eloquent. And that mouse study is remarkable, by the way. I remember when that came out. That needs to be a Sunday email to my group. So let's remember that. I would highlight one thing that you've already alluded to and one that you haven't. But I would just add it to my side of the ledger on why I struggle so much with the legitimacy of epidemiology. Anything that relies on a food frequency questionnaire I simply can't take seriously for the fact that at least with every patient I've ever come in contact with to try to accurately assess what they eat based on a food frequency questionnaire would be a fool's errand. It's simply unrelated to what they eat, full stop. The second issue I have with nutritional epidemiology comes down to the hazard ratios that very commonly show up and lead to grand statements. When you see hazard ratios like 1.16, and we talk about this like we have demonstrated the causal relationship between bacon and cancer, I mean, what would the actual fathers of epidemiology be saying in their graves if they were looking at these strength of association. I'm not necessarily saying there should be no epidemiology, but boy, there needs to be a referendum on how the media is taught how to interact with such studies, on how we scrutinize some of the methodologies behind these things. Because again, checking a food frequency questionnaire once in an eight year study, it just doesn't mean anything. It means less than nothing, actually. And yet it's amazing that that could be the basis of an observation. So it's hard for me to say.

Something good about epidemiology on the measurement issue, which is often, I think, incorrectly perceived by some people as the key issue, and say, yeah, we admit the measurement of food intake is a big problem, but other than that, everything's fine. Even when the measurement is near perfect, as in my mouse study, it's still not fine. You don't have to go back to the fathers of epidemiology or wait for the fathers of epidemiology. You can go back to at least Confucius who says to know what you know and to know what you do not know, that is true knowledge. This idea that we have to be honest with ourselves and each other about what we know and don't know and how we know it and don't know it is clear. And I think that's part of that reform. We need a greater level of honesty. Samin Vizier just had a paper in one of the peer reviewed journals looking at or writing about epistemic humility and saying, you know, when you get to the discussion section of a paper and you consider that the hypothesis that you made, which now seems to be supported by your data, and you, and you say but I might be wrong, and then you just systematically go through with the greatest effort and art to show how you're not wrong, how you can dismiss all the competing explanations, that's not an honest epistemic. An honest epistemic humility would be to say I really might be wrong. And here's all the ways I might be wrong that I and others should test going forward. Michael Strevens in his book the Knowledge Machine, does a beautiful composition, decomposition, construction, deconstruction of this and talks about the idea of communities doing this. And so that we need that constructive battle, but it needs to be an honest battle. The battle shouldn't be ad hominem. The battle shouldn't be undercutting each other by who you are and who you work for and what your beliefs are. It needs to be a battle about the data. He calls that the iron rule of evidence. So if I say, Peter, your theory that X causes Y could be mistaken because your measurement of X is not valid. You need to come back and say, I hear you. What if I measure X this way? That's a legitimate thing. Let's do it and see if we can rule it out and then we can go back and forth and well, maybe the measurement of Y is not right. I think we need to be more honest about that. I don't think the nutrition epidemiology field has been quite honest about the limits of its measurement, but so have the attackers not. I've been careful about saying, I think measurements of energy intake and expenditure from self report methods are so bad that they shouldn't even be used. That is not enough to say, well, it's not perfect, it's so bad you can't even be guaranteed to get the directional effect, so don't even use it. Better to not do the study at all. But on the other hand, if you said to me, I want to know if people eat vegetarian or not, or eat kosher or not, or eat after midnight or not. I don't know. Maybe people do report those accurately, maybe they don't. I honestly don't know. I think we have to ask for what purpose.

You mentioned very briefly Katherine Flegel earlier. She also wrote a very famous paper about the obesity wars. Give folks a bit of an idea about that, I thought. Very excellent paper.

Yeah. Katherine Fleagle's been a colleague from whom I've had great respect for many, many years and she's very, very bright and very capable and very careful. I've learned a lot, lot from her actually. She published a paper in, I think it was 2005 that really got people's attention and it was interesting in the way it was a meta analysis and combined pooling analysis showed a few things. What was interesting to me is what people harped on, particularly the Media, was that BMIs in the overweight range were actually not strongly associated and consistently associated with increased mortality rate. In some cases were associated with lower mortality rate. The media went crazy with this, as though it were a new finding. Just as every few years they write the new finding that BMI is mass and not fat. The new finding that there's a genetic component to obesity that we knew from decades ago. The media went crazy with this. The people that are the defenders of you can never be too rich or too thick went crazy and attacked her. And the bizarreness of it was, wasn't new. I mean, it was a meta analysis. How could it have been new? We all knew these data. You can go back to Linus Pauling, the double Nobel laureate, has a paper in the 1950s on BMI and life expectancy. And he's got a bathtub shaped curve there. So this is not new at all. But somehow it was seen as new. It was in jama. The newspapers went crazy. The people who were defending it went crazy and they started attacking her very vociferously. One investigator called it a worthless pile of rubbish. To me, these are very inappropriate statements.

Who were the most vocal critics?

Some people in Boston seem to be among the most vocal critics. So that was kind of a very interesting thing. But as I said, it was all old news. Anybody been reading the literature for decades would have said, but I already knew this. What was actually new and to me far more intriguing, but got much less attention, was that the nadir was moving over time. That was, we talked about earlier, that in 1990, it wasn't the same as 1970. That is to me, very intriguing. We actually have an active NIH grant built on that, trying to figure out what's going on with it. But anyway, they attacked her and these statements, there's rubbish and that she didn't know what she was talking about, that there was nobody with a medical background, and these are all ad hominem things. First of all, it's not clear. There was no one with a medical background on the paper. And even if it was clear, who cares? The data are the data. Whether you have an MD or you don't have an md, it's the data that matter. When I frequently say in science three things matter. The data, the methods used to collect the data which give them their probative value, and the logic which connects the data and the methods to conclusions. Everything else is not science. Now, if you want to look at other things as ways of helping you make a pragmatic decision, that's fine. So if I say, oh, I trust Peter, and he's a smart guy and I know he studies a lot, and he tells me I should eat this. That's maybe a very pragmatic way of making a decision, but it's not science. The science is the data, the methods and the logic connecting the data to conclusions. Not whether I trust you or not. Not. So all these ad hominem things were said. They attacked her. It was very, very vociferous, very inappropriate. But there are many other people who have been attacked. I've been attacked. I know lots of other people have been attacked for their beliefs. When Nina Tiekolz had an editorial, I think in Lancet, talking about some elements of nutrition and fat and carbohydrate, many people wrote in, tried to get it retracted. This is a sort of regular occurrence. Now. When Brad Johnson published On Red Meat roughly two years ago, big meta analysis showing that the association between red meat consumption and negative health outcomes was not strong and compelling as his words, again, vigorously attacked people tried to get it retracted before it was published, which thankfully didn't work. Instead of just engaging with the data, the methods, in the logic, it's terrible. This happens. It doesn't only happen in obesity and nutrition, but it happens a lot in obesity and nutrition.

It does seem to happen disproportionately in this field. Why is that?

I think it has to do with the everyday experience. Back to my example from early in our discussion about the beta cell of the pancreas. If you are at a cocktail party in your neighborhood and you say to somebody, well, I study diabetes and here's what I think about the beta cell of the pancreas and how this might work. Unless that person also studies it, they're not going to challenge you. They probably don't know what the beta cell of the pancreas is. They don't have strong feelings about it. They don't see it every day. All of us eat every day. Almost all of us eat almost every day. Food is culture, it's family, it's love, it's economy, it's commerce, its political beliefs, it's philosophical beliefs, it's ethical beliefs about whether you're a vegetarian or not. It's the sustainability of the environment. It's so connected to so many emotional things. And we all have that everyday experience. And we have to make decisions every day. Each of us decides what to eat, what to feed our kids, whether wear a seatbelt or not. All these things. We make these decisions and then we may want to justify them. We may need to believe they're good. We may mistake our experience for expertise. And that's why I think when you get into any fields where people have everyday experience, whether it's human sexuality, whether it's relationships, whether it's child rearing, TV watching, book reading, music, eating, these are things people have very strong feelings about and often will opine about, often quite aggressively. In the absence of data, it is not only obesity. Statements about violence in media, statements about how to best teach mathematics to children. People have very strong feelings that are often quite in contrast to what the data show.

So is there a path forward here from a nutritional standpoint? I mean, I think you've already touched on a few of them, but we've already seen some pretty exciting pharmacologic things come along. I think semaglutide, the study that we've talked about before on this podcast, really kind of a remarkable drug. It's certainly the most impressive thing that I've seen clinically for obesity. We're making great progress there. We talk about surgical treatments. 25 years ago when I was in medical school, I remember the first gastric bypass I ever saw done as a medical student watching surgical rotation. And this was an open, this was done before they were doing them laparoscopically. It was done as an open roux en y gastric bypass on a 400 pound man who would die 40 days later in the hospital of sepsis. He never got out of the hospital. He had an anastomotic leak. That was a very dangerous operation 25 years ago. Today that operation is done almost without exception laparoscopically. It is an incredibly safe procedure and it has also remarkable efficacy. So on this surgical front, on this pharmacologic front, we have made amazing progress. We haven't made progress on the nutritional front. Are we going to not clear?

New York Times reporter called me a few years ago and this was in response. There was an article about, I think, was it President Taft who was very obese and somebody had found some letters between him and his physician talking about diet. And it was almost you could have picked those letters up and said this was between a president and his or her physician today and they would make equal sense. And the reporter said to me, this is a very bright science. Nutrition journalist said to me, david, why have we made so little progress? Why have we not been able to find the diet that reliably causes sustained weight loss? And I said, your question is premised on the idea that there is a diet that reliably causes sustained weight loss. Why should we believe that that's true. So that's an important thing to think about. I think one of the things that is perhaps a misperception and maybe a very problematic one in our field around nutrition and weight loss or food intake and weight loss is that there is a good diet with respect to weight loss, particularly such that for most, if not all people, if you just ate the right way, you wouldn't have to count your calories, you wouldn't have to be uncomfortable and hungry, you wouldn't have to feel deprived, and yet you would maintain a good healthy weight. I know of no reason to believe that's true. I know lots of people who argue, is it Diet A or Diet B? So this one thinks it's low carb, and this one thinks it's high carb or low fat. And this one thinks it's don't eat at night. And this one thinks it's. It's whatever it is, eat paleo, et cetera. Maybe the null hypothesis is doesn't matter that much. There isn't such a diet for many people. Now, for some people, they do maintain a normal, healthy, desirable weight without trying to restrict their energy. But maybe for others, it's just not the case. I think the paths forward are manifold, and I think in some cases we are on the good path, and in some cases we are wandering in the drunkard's walk. We're on the good path, I think on surgery and pharmaceuticals, clearly a long way to go, but they've gotten much better. I'd love to see more funding for those for good research. And I think we are on a good path, but I think we need to get on a much better path about, as a society, making those available to people. If you have cancer, we're willing to treat you. If you have obesity, maybe not. So if you're rich, you can pay for that. If you're not rich, what do you do? So I'd love to see more access to care. And I think we're on a better path, but we need to be on a better path. Still, I think we're on a good path on stigma, a long way to go, but I think as a society we've woken up to say stigmatizing obese people is not okay.

Do you think the pendulum's gone too far? I read something. This might have been a joke, but I literally read that Adele was actually shamed for losing weight. I don't know if that's true, but if it was, it would certainly suggest that the pendulum has gone a little too Far the other direction.

I would certainly say that that's ridiculous. I don't know if she was or wasn't, but if she was, that's ridiculous to me. The take home message is shaming people about their body habitus is not, not good in either direction. It's not a directional thing. It's just shaming people about their body habit. It's not okay. Sometimes the counter argument is, but it's good for them because it'll help them want to lose weight. And sometimes the argument made against the shaming is empirical. The evidence shows that people who experience a lot of weight shaming gain more weight and then the causal thing is thrown in. So therefore you shouldn't do it. You mean if shaming didn't cause weight gain, it would be okay to be immoral and cruel to others? No, I mean if you replace things like sex and race in that sentence, you're like, no, it's not okay to shame and denigrate people because of their sex, race, age, body habitus, regardless of whether it causes weight loss or weight gain, period. It's a moral issue, it's not an empirical issue. We have a long way to go, but I think we're making progress and people are going, yeah, I guess we can generalize this idea that you shouldn't denigrate people because of age, race, sex. Maybe you shouldn't denigrate them for other characteristics that are not moral failings. I think that on the basic science we're making progress. We could make better progress if we tightened up the rigor of our science. We could make better progress if we had more funding. Many groups, including the National Academy of Sciences and I'm on a strategic council to increase the rigor of science. All of science, not just obesity, nutrition. But we're making progress. We know so much more about genes and physiology and metabolism and cells with respect to obesity and nutrition than we knew 20, 30, 40 years ago. The one area where, as far as I can tell, we are not making progress and again, I don't think we are yet on a good path. I don't think we're on a path at all is this public health, community, school based, community based, policy based approach. I think we are continuing to look for our keys under the lampposts because that's where the light is as opposed to where the keys might be. I think we are continuing to ignore the data and keep saying the same old hackneyed suggestions that people have been trying for decades and that when you really look at the data, have Been at best, not been shown to work, and at worst, been shown to not work. And I think there are some people who are patently obfuscating those data. I think the cluster randomized trials we see in the childhood obesity literature bring to mind the phrase that rhymes with cluster muck. And it just. This is cluster muck. This is distorted evidence. This is science gone wrong in the worst sense. I think we've got to clean that up. We've got to clean up the quality of the science we do and start treating this like science just as much as the science of quarks or tires on automobiles, or beta cells of pancreases, and treat it like real science and take it just as seriously.

Can you tell folks briefly what the cluster randomization problem is?

Sure. So a cluster randomized trial is a trial in which, instead of randomizing, the individual unit of observation might be, let's say, a child in a school who you either assign to the treatment group, maybe it's exercise, or the control group, no special exercise. You assign the entire intact unit, such as a classroom or a school or a neighborhood or a family. There's nothing wrong with that as long as you have theoretically at least two clusters assigned to each condition. So you have some ability to estimate a variance, although frankly, with only two, you have so little power and robustness. For practical purposes, that would be invalid, but theoretically it's valid. But then you must analyze the data. Data to take into account both what's called the clustering and the nesting. The clustering is that you have people grouped, and that grouping leads to more similar individuals. So imagine that we did a trial, and the trial was you and your brother randomly assigned to eat low carb, and me and my brother randomly assigned to eat high carb. And at the end we see a difference by an ordinary T test. And you say, well, hold on a second. Was that really the effect of diet, or was that the Attia brothers are different than the Allison brothers? Well, you have to take that into account. So that's. You need more than one cluster. So if you get the Attia brothers and the Jones brothers and the Allison brothers and the Smith brothers, now, in theory you can do it, but now you can't treat us like we're eight different people. What you really have is four different clips clusters, four different sibships, four different sets of brothers. You've got to take that into account. You have less degrees of freedom. People don't do that reliably, and they don't do it correctly often. And that Leads to many papers being wrong. We've written countless letters to editors about this. I think we've probably had at least three or four cluster randomized trials retracted as a result of letters we written where people have had to come back and just say the results don't hold up. But until this has changed, we have people out there who understandably say, but I read these papers suggesting that this works. Gardening in schools makes kids thinner. That's not what the results showed, but that's what the paper says. And until we become more rigorous and more honest, we're not on a path.

So if I were to ask you now to speculate outside of known things, if you had to guess, and if you presuppose that there is an intervention or a set of interventions that could improve public health outcomes, what would your guesses be? What would you guess to test?

General education. Not nutrition education, general education. There is provocative data. I don't want to say definitive data. There are provocative data strongly suggesting that general education, especially for girls and women, leads to lower BMIs, lesser rates of obesity and lesser diabetes than less education. So there are some studies in Europe of policies where someone puts in a policy and it effectively gives a cohort of people more education. And then you see in that cohort less obesity, especially among women. There's a famous study by the Rameys, who were husband, wife, investigative team who actually worked at UAB when I first got there. And they started this study decades ago at unc. And it was sort of Head Start on steroids. They called it the ABC Adarian or Abecadarian study. And they gave these kids the super Head Start program. And it was mostly just general education. There may have been a little nutrition education, but mostly just general education. Wasn't a weight loss study. It wasn't intended to be. Thirty years later they followed them up. There's a paper in Science on this. Guess what? The women have less obesity. The Moving to Opportunity study, funded by the Department of Housing and Urban Development, took families who lived in so called poor neighborhoods. And they gave them either randomly assigned them either to control, but it basically got nothing, or to housing vouchers, but the housing vouchers required that they move to less poor neighborhoods. And what they then found years later in follow up, again published in Science and New England Journal of Medicine, is that there was less obesity and diabetes in those assigned to move to the less poor neighborhoods. And given the financial wherewithal to do so. So I could go on, but these are things that suggest to me that Education, general education may help. And I think that may speak to this whole socioeconomic thing. We started out way earlier. What is it about higher socioeconomic status that at least in some groups, groups at all, but at least in white women, seems to be associated with less obesity. And I don't know what the causal mechanisms are, but that might be my best. So if somebody said to me, you're going to be the king for a year and you've got the federal budget and you can take this big chunk of money and you can make an impact in obesity and diabetes, I would say I'm going to divide it into four pots. One pot is going to be surgery, and it's going to be both providing it and continue to study it. The next one's going to be pharmaceuticals, both providing it, continuing to study it. Third part is going to be some general education, maybe general well being, safety, security, starting in early childhood to see whether that alone is enough. And it may be really reducing disparities. Back to Confucius, Confucius said, we are not so concerned with an absence of wealth, we are concerned with a disparity of wealth. And so it may be that reducing disparities is really important. And then the fourth part would be basic research. And I'd like to say let's look at senolytics and let's look at microchimerism and all the things that you talk about so often your podcast that a. But against both metabolism and obesity and nutrition, but also the fundamental senescence. So I'd love to say, can we use microchimerism to restore people to younger metabolic states? Can we use senolytics to do that? That would be my fourth bucket. Those basic science questions.

Now, a moment ago, you touched briefly on the National Academy of Science, Engineering and Medicine. You were on a panel that looked at the question of reproducibility in science. Correct?

Correct.

This was two years ago, I think. The consensus, I believe, was that there was not a crisis, but we shouldn't let our guards down. Am I paraphrasing that correctly?

Close the phrase that Harvey Feinberg, who was the chair of it, I don't know if it's his phrase, but he started using and we followed is no crisis, but no cause for complacency. And the idea of no cause for complacency or no complaints complacency is we must make things better. The no crisis is a tricky one because it depends on what you mean by crisis. And some people say if you look at dictionary definitions of crisis, the idea is that it's a system about to collapse, so you need to go to the icu. If you don't get intervention, you're going to die, you're in crisis. There's no evidence that science is in crisis. There's no evidence that science is about to die or it doesn't work anymore. Not at all. Another way to say things are getting more much, much worse rapidly. There's also no clear evidence for that. There might be some things that are getting worse, but on average, there's a lot of evidence that things are getting better. My own belief is science is better and more rigorous than it's ever been in history. And so in that sense, no, I still don't think there's a crisis. But there's a third way, which is, I think, the spark for many social movements. If you think about what often happens as a social movement, is you have a group of people that, for example, are often, when it's around depression and they're oppressed, and they're sufficiently oppressed that people are unwilling to make too much noise. And the status quo is only very slowly and grudgingly moved. And then at some point, people feel a little bit more confident and a little less oppressed and they start to speak and people say, this state of affairs is not acceptable anymore. And if you were to say in that later point in time, but it's much better than it was in the earlier time, someone might say, it may be much better, but we ain't taking it anymore. Things may be much better than they used to be. Things in 2021 are probably much better for many situations than they were in 1921 and much better then than they were in 1821. But there comes a point where you say this lead in gasoline, or this lead in our paint, or this way in which we use heating, or these kinds of catalytic converters on cars, or this way in which we burn up fossil fuel, or this way in which we treat different age, race, sex groups. It's just not okay anymore. Yes, it may be better than it was years ago. The world has spoken. We won't take it anymore. I think that's where we are. Science saying, yes, science is probably better than it's ever been, but we also see all the flaws and it must get better. And that's why Marcia McNutt, the president of the National Academy of Sciences, instantiated this new Strategic Council on Trust and Integrity and Rigor in Science and very generously appointed me as one of the three co chairs. Marcia France Cordova, who's the director emeritus of the National Science foundation, and I are the three co chairs and are working on trying to see what we as the academies can do to try to help a little bit. And many other organizations are also trying to help on this.

How much of the situation in science is intrinsic, is within science itself, and how much of it is a result of the public, inclusive of the media's interface with science?

I think it's both. I think science is hard. Michael Streven's book, which I mentioned earlier, the Knowledge Machine, does a wonderful treatise on that. And I think we have to recognize that, but I think we have to make the distinction between normative errors and non normative errors. And let's take an example of each. So a few hundred years ago, Galileo was under house arrest. The Pope has said, bad writing about Copernicus, stay locked up in your house, but we won't kill you. And from his house, Galileo directs an experiment, or not really an experiment, a study. And he has two colleagues go out to two tops of mounds or hills or mountains far apart, each holding a lantern with shutters and a synchronized watch or timepiece. And he says to them, at a predetermined moment, you guys open your shutters and you record when you see the other guy's light. And we'll figure out whether light travels instantaneously or not. Is there some delay until the light gets to you and they conclude that it's instantaneous? They can't discern any time. Now, we know today, of course, that that's wrong. Ole Romer probably is the first person who convincingly shows it's wrong when he shows that a moon comes around at a time different than his mentor Dominic Cassini predicted. But at the time, that's the answer they get with their instrumentation. They couldn't have done better. I would call that a normative error. I would not mock Galileo. I don't think Galileo did anything wrong. I think we look at him as brilliant. My God. Great question, great way of working on it. But things move along. Similarly, when Linus Pauling says DNA is a triple helix, before Watson and Crick come out, it's a double helix. He didn't have good X ray crystallography data. He's working at the edge. Normative error. Then there are other things very famous. Non normative error is people working up the size of the thymus gland of children. They worked it out from cadavers in the early 20th century. Guess what? People don't become cadavers at random. Poorer people tend to become cadavers. Poor children tend to be undernourished. Undernourished children tend to have smaller thymus glands. So when physicians start seeing richer children coming in, dying of sudden infant death syndrome, and they examine them and go, hmm, this is a big thymus gland. Well, it was actually a normal thymus gland because their norms were from undernourished kids. So let's irradiate these kids with big thymus glands and prevent sudden infant death syndrome and probably cause lots of cases of thyroid cancer because the thyroid and the thymus are very close to each other. So that's an example of a non normative error, because even 100 years ago, any epidemiologist or statistician could have told you that is bad sampling and bad inference from bad sampling. We need to make that distinction between normative and non normative errors. And a lot of the errors that we have in nutrition epidemiology today, I don't think can be called normative errors anymore. I think we have to say that was non normative. The misanalysis of the cluster randomized trials. These are not normative. Any statistician knows how to do it. People are either obfuscating or they're just woefully ignorant and not using professional statisticians when they need to. People using food frequency questionnaires to draw causal inferences about some of these things we've discussed. These are not normative errors. People should know better. Then there's the stuff about the sort of more general public about believing things and how we promote our ideas. And here's where I think. I think at root is us and the scientific community that have to take responsibility for it, but it branches out beyond us. I think we need to be prepared to lose some battles in order to win the intellectual war. And what I mean by that is we need to be prepared to not use all the rhetorical tools at our disposal at any one point in time to convince somebody that X is true. Even when we're really worked up and think it's important we believe X is true, we think it's important that others believe X is true because we want them to eat what we think is good. We want them to eat more broccoli and less ice cream. We want them to take their vaccine, wear their mask, wear their seatbelt, stop smoking. We may be right about all those things. Maybe people should eat more broccoli and less ice cream and wear their seatbelt getting vaccinated and wear their mask and sipo. But if we Use rhetorical devices such as if you and I are debating and I attack you on ad hominem grounds, or I exaggerate the strength of my evidence, or I don't honestly say that I've shown an association and not causation, then in fact, I may win that battle that day on convincing people to eat broccoli instead of ice cream. But I've lost the battle in helping people think through what good evidence is and elevating our level of dialogue. And I think if we can get to the point where, just as today we changed our dialogue, we think about the things people would say, think about what a late night comedian would have said 30 years ago in making jokes about wives and husbands and race and sex, that would never be considered acceptable. Today, we are able to change societal norms about dialogue. Can we elevate our societal norms of dialogue on epistemologic and empirical issues so that we can get people to. You don't have to be a genius to say, oh, you're telling me you have a treatment for X? Was there a study? Was the study in humans? Was it a randomized study? Was it a study of the actual outcome you're making a claim about? Was it a study that was long enough for this to be a meaningful outcome? Was there a statistically significant result? Was the result big enough to matter? Was the dose a dose I might realistically take? Those are not all that difficult questions to train ourselves and each other to just reliably ask. And if we just reliably ask those and reliably and honestly answer them, it would go a long way.

Yeah, it's interesting. I mean, it's impossible to have this discussion without thinking about COVID because Covid has been such a polarizing scientific phenomenon in a way that I've become quite frustrated in watching it. And I'm trying to maintain a distance from it and ask myself the question is the reason that the head of the CDC makes assertions about masks or vaccines or mandates because she doesn't think that the people to whom she's speaking are intelligent enough to appreciate nuance? Or is it because she doesn't actually appreciate the nuance herself? And this gets to exactly the point you made a moment ago, which is the high ground is being lost. I think the public's faith in science, in the institution of science, and is eroding dramatically. And I think Covid has accelerated that in a way that I wouldn't have predicted. I remember having a discussion with a friend in March of 2020, and I very Very naively said to him. I think that what we're about to see with respect to the speed with which medications and vaccines are going to be developed is going to elevate the consciousness of science in the public's eye. I think the public is going to look back in a year and say, wow, science is great. Just as in the 1960s, the best and the brightest kids went into engineering because they were inspired by the space race. I stupidly thought this might be the end of every smart kid becoming an investment banker, and instead we might just see more kids enrolling in science. How silly of me to not envision what was coming, which was a world in which well meaning public health officials simply failed to communicate the nuance of science and lost so much credibility.

Covid has been unique. One of the things I was saying to my wife recently is I hope I live at least another 20 years for many reasons. But one of the reasons is that I really want to see what the historians say about the period of roughly 2016 to a year or two from now when hopefully we are somewhat out of this pandemic. I can see it in the present. I want to see it in the 2020 hindsight. Let me say two things about your points. One is about motivations, and the other is about trust. The motivations issue, I think, is an interesting one. You know, I was having a conversation with a colleague many years ago, and I was ranting about university accountants and how they drive me crazy with little things. And this friend of mine said to me, he says, think about you and your science. He says, you're so fastidious about it, you're so passionate about it, you're so committed to it. One iota of bending the truth is not okay for you. He says, the accountant feels that way about the way the accounting is done, and that's their domain. Different people have their different domains. And if I were to say to you, Peter, what are you professionally? Or you might say, or what are you just more generally? You might say, you might say, I'm a father. You might say, I'm a physician, I'm a healer. You might say, I'm a scientist. You might say a few other things. When I came to iu, I was at a big retreat and a friend of mine was there, and we started going around and introducing ourselves, and I was sort of the new kid on the block. And I said, you know, my name is David Allison, and I'm a scientist and I study, blah, blah, blah. And he said, you know, the first thing you said, like, you're the new dean here. And the first thing you said is, I'm a scientist. You didn't say, I'm the dean of the School of Public Health. He says that shows how you think of yourself. If you were to say to me in your professional role, what's the one thing you must never compromise? Truth. What's my one sacred duty? Pursue and communicate truth to the best of my ability. But if you said to some other people in my field who. Who are equally good people, maybe better people, who knows? And you said to them, what's your one uncompromisable duty? Some of them would say, help people make their health better, enact justice. Can't say they're wrong. That's their value. And I think that's what comes out a lot in public health. There are true believers who think they know the right answer. They think eating this is better than eating, that in many cases, they're probably right. They think not smoking is better than smoking. I agree. I think they're right. And given those premises, they say, whatever it takes to convince people to eat A and not B, to wear their seatbelt, to not smoke. I'm willing to say it. That's my sacred duty, is making better. I think that's part of the problem of what's our identity, who's speaking. And I think we need to be clear. When we are speaking as scientists, then we must not compromise truth. And I think when we're speaking as advocates, that's fine. Just say you're an advocate. Say, I'm not being a scientist right now. I'm just telling you what I want you to do and say whatever I need to say to convince you. I think that's important. The other thing I want to pick up is this trust idea. I hear a lot, especially from people within science and academia, especially from people who are somewhat on the left, but not only on the left side of the political spectrum who say trust in science is really weighed down in the last few years. I'm not sure that's true. In fact, I don't think it is true. It depends what you mean by science. If what you mean is science as a process of developing and finding knowledge, I know of no evidence that it's down. The results of the Pew Charitable Trust and their surveys, for example, suggest that it's not. If what you mean by science is, is trust in individual elements of the scientific community, then I'm not sure it's down either. But it's spread around. So some people think Fauci is trustworthy, and some people think Gwyneth Paltrow is trustworthy. Some people think David Allison is, and some people think Peter Ortiz is. And some people think we're corrupt and ignorant and confused and terrible. And I think the challenge is that there are people who cannot distinguish between the statements of a Tony Fauci and the extent to which they are or are not backed by evidence and the statements of a Johnny Yannidis and the extent to which they are not backed versus the statements of a Gwyneth Paltrow or somebody else. And I think that's the challenge. It's not that people don't trust science. They don't know which voice to trust as a communicator of the science, and therefore they don't trust individual elements of the canon of science. And we see that very strongly in nutrition. There's a nice summary on the Pew Charitable Trust website now indicating that trust in science is high. Trust in dietitians is high. Trust in medical doctors who talk about nutrition in treating their patients is high. But trust in nutrition scientists compared to dietitians and medical doctors who talk about nutrition and compared to other scientists is low. So in nutrition we have met the enemy and it is us. We have shot ourselves in the credibility foot with our obfuscation and our exaggeration and our hype. And I think that pocket of trust is gone, even though trust in science overall all is not down.

David, this was a fantastic discussion. You are always so lucid in the way that you can talk about an idea, including this very one. I think this point that we're ending on is really, really spot on. And I think that the difference between science and advocacy can't be overstated. And I think it would be amazing if people, myself included, all of us, had the self awareness to speak and know which hat we were wearing. Think about the problems that could be solved if one, you allowed people to wear both hats, but they always had to have a hat on. You couldn't blind the listener from which hat you're wearing. And if you were wearing your scientist hat, you would be delivering the nuanced hard truth, as messy as it might be, as unclear as it might be, with no regard for how a person feels when they hear it and what action they may or may not take as a result of it. And of course, if you're really just in the business of saying I want to change your behavior because I think it's in your best interest, I'm going to put the different hat on. I like this idea. I think there's something here. We can come up with a two hat system and everybody gets to own two hats.

Every one of us in our lives have to deal with this. We're both fathers. As fathers, you face this all the time. Sometimes your kid says to you, can I do this? And you say, no. And they could say, why? Well, because I think you might get hurt or it's a bad idea or something bad might happen or you won't get that good thing. And they say, do you know? Are you sure? And they have a counter argument. And if you're honest, I don't know and I'm not sure. I have no randomized controlled trial that climbing that tree is too dangerous. I'm just telling you I don't think it's a good idea. I have no need to pretend that I did a scientific experiment. I'm your dad and that tree looks dangerous to me and I'm telling you to get down from it right now. And then there are other times I'm acting as a scientist and I say this is what would constitute adequate evidence. And this is what we know. Just like Confucius said, know what you know and know what you don't know.

Brilliant. Thank you David. It's been a pleasure sitting here with you today.

Thank you my friend. Great to see you Peter.

Thank you for listening to this week's episode of the Drive. If you're interested in diving deep deeper into any topics we discuss, we've created a membership program that allows us to bring you more in depth exclusive content without relying on paid ads. It's our goal to ensure members get back much more than the price of the subscription. Now to that end, membership benefits include a bunch of things. One Totally kick ass comprehensive podcast show notes that detail every topic paper, person, thing we discuss on each episode. The word on the street is nobody's show notes rival these monthly AMA episodes or Ask Me Anything episodes. Hearing these episodes completely access to our private podcast feed that allows you to hear everything without having to listen to spiels like this. The Qualys, which are a super short podcast that we release every Tuesday through Friday highlighting the best questions, topics and tactics discussed on previous episodes of the Drive. This is a great way to catch up on previous episodes without having to go back and necessarily listen to everyone. Steep discounts on products that I believe in but for which I'm not getting paid to endorse and a whole bunch of other benefits that we continue to trickle in as time goes on. If you want to learn more and access these member only benefits, you can head over to Peterattiamd.com forward/subscribe. You can find me on Twitter, Instagram and Facebook, all with the ID Peter Atia. You can also leave us a review on Apple Podcasts or whatever podcast player you listen on. This podcast is for general informational purposes only and does not constitute the practice of medicine, nursing or other professional healthcare services, including the giving of medical advice. No doctor patient relationship is formed. The use of this information and the materials linked to this podcast is at the user's own risk. The content on this podcast is is not intended to be a substitute for professional medical advice, diagnosis or treatment. Users should not disregard or delay in obtaining medical advice from any medical condition they have and they should seek the assistance of their healthcare professionals for any such conditions. Finally, I take conflicts of interest very seriously. For all of my disclosures and the companies I invest in or Advise, please visit PeterAttiamD.com about where I keep an up to date and active list of such companies. Sa.