SCIENCE BASED DIET
Mediterranean Diet Etc.
How Big Data Revealed the Best Diets
Chris's Conference Talk
My interest in nutrition began in high school when my side hustle was caddying at the Orinda Country Club. One fateful day, the caddie master assigned me to caddie for Jack LaLanne, who was a TV star. His show on fitness and nutrition was popular for 36 years. This photo was taken when he was almost 40. His mother was a Seventh-Day Adventist, and he was influenced by several Adventist nutrition professors, some of whom are still alive. He was also influenced by Ancel Keys, who was a towering figure at that time for his bestselling books on Mediterranean-style diets and his appearance on the cover of "Time." ...
Carlton Fredericks was a radio star six days a week for 30 years, and his message was the opposite of TV star Jack's. My father embraced these books and fed us meat, eggs, and cheese. Starch was a bad word for him. Welcome to the diet wars, young Chris. I brought up Jack at the dinner table, and Dad said Jack is a muscle man — the word of the day for bodybuilders. Dr. Fredericks is a doctor. That landed with me. In addition to the steaks Dad grilled, we ate the geese, pheasants, and ducks he hunted with a shotgun. I remember the frequent sound of clinking as we would spit out the lead shot that remained in the meat. Not sure how many I swallowed.
That was before Herb Needleman's epic battles with the lead industry in the '70s, armed with epidemiology he and several scientists produced. He is one of my greatest heroes. My desk in second grade looked like this — it had an inkwell. I filled it with mercury, which you could buy cheap at the drugstore, and it made me popular because the kids could dip their dimes and quarters in it and get a really cool luster finish. Sometimes we would splash mercury onto the tile floor, and beads of it would dance around the room. No one thought anything of it.
Ironically, in my earth science days, I ended up working for a company with something like 34 water-testing labs when I was there. And so, I ended up getting a decent background in environmental epidemiology. Water-testing labs typically get samples from mining areas where lead and arsenic get into the streams and near chemical plants, which dump chemicals in rivers. We got samples from the Chesapeake Bay drainage area because chicken farms had concentrated manure that got into the bay and killed fish and crabs. The fishermen in New England were upset and thought the chicken farmers were ruining their livelihood, so it became industry against industry. But the companies with the biggest balance sheet usually win, so you can guess how that dispute came out.
Unfortunately, environmental epidemiology is depressing. Every industry goes after scientists like Herb Needleman, accusing them of fudging their data, doing poor science, being alarmist, or a bully. This is a riveting book about what Herb and the scientists he worked with went through. Companies with toxic products market them as healthy and natural. They find shoddy alternative science to claim the science is conflicting. They write stories for the press. They lobby the government. With their enormous budgets, they run circles around the baffled scientists who had believed kids with lead poisoning from paint chips would be the primary motivation for these companies. The lead industry were masters at it. They used the playbook that the tobacco industry wrote, but 30 years before they wrote it. It bought them 50 years of delay and denial, and, as everyone knows, we're still dealing with the tragic consequences.
If you get a PhD in physics from a top university and go to work for NASA, as many of my friends have done, you have a tough choice to make. If you choose to work in space exploration, you're gonna be a rockstar at every party you attend. But if you choose climate science, industry is coming for you, and they will get half of Americans to believe you're incompetent and promoting a lie. Some partygoers are gonna bring the rage.
If you see this photo and think, "Oh, that's Joanna Hoffman!" you have a good eye. You probably saw the Steve Jobs movie, but it's actually Kate Winslet. She won a Golden Globe for her portrayal of Joanna in a Steve Jobs movie. That's because it's in Apple's financial interest to lift her up as a rockstar. I was less qualified in tech than I was in earth science, but I was on stage at some of our company events, sometimes signing autographs. That never happens to epidemiologists who save kids from lead poisoning. By the way, we were at Joanna's house three weeks ago for a 35-year reunion of General Magic. We were all teasing Joanna and calling her Kate.
Living Long. As life went on, I kept learning that I am a walking collection of risk factors. If you're wondering why I have a hole in my head, here's your answer. Doctors have pulled me back from the dead at least five times, and they are my heroes. When I was 42, a carotid ultrasound revealed I had moderate atherosclerosis, which raised my life insurance higher than it already was, given that I'd had rheumatic fever as a child, which had scarred my heart, and I have a type of birth defect in my heart.
I watched my dad, former captain of his hockey team at Queens, decline in his 60s with ailments like gout and high blood pressure. He died at 70 of a heart attack. In the meantime, Jack pulled 70 boats with 70 people inside a mile across Long Beach Harbor to celebrate his 70th birthday. He swam in handcuffs, and he was shackled at the ankles. That white dot on the lower right of your screen is Jack. In the meantime, none of the popular diet doctors that my dad admired made it out of their 70s. Blake Donaldson on the left was a popular carnivore doctor. I knew those were anecdotes, so I tried not to let them affect me, but Jack kept making the news for thriving until he was 96. It made me wonder if this was like environmental epidemiology and you could actually discover enough good signal in the noise, like Herb did with lead, to understand what is healthy.
Big Data. In my decades in tech, I watched companies like Google hire a lot of PhD data scientists, some of whom I know, and crush companies like Yahoo, which mainly hired engineers who didn't understand big data. Data scientists at Google didn't call it epidemiology because that word means population health, but the methods they use are very similar. I wondered if my background would help me sift through the noise of nutrition. It is a shock to come from environmental epidemiology into nutrition and hear over and over from the most popular influencers, "You can't infer causality from observational data." What? In environmental epidemiology, that has always been your job, and you can't do RCTs (Randomized Controlled Trials) on paint chips in children. There are dozens of great books on inferring causality from observational data, which we do in every field I can think of. UC Berkeley alone has over a dozen scientists who list causal inference as one of their specialties. Perhaps the most influential advice I ever received from my professors was to not be that guy who reads thousands of papers hoping to understand something. You be the guy who thoroughly understands the relatively small number of papers that made a real difference either for or against a hypothesis. Most of the rest is noise.
The Famous 22-Country Paper. It became super popular among consumers and influencers in the early 2000s to reference this 1957 paper after the bestselling book Good Calories, Bad Calories mentioned it. That paper is still popular today, so I thought it would be important to fully understand it. For example, the movie Fat Fiction, which got millions of views, made this paper a centerpiece of the movie. The only copy I could find was a poor-quality scan on CrossFit's site, and it was a little bit hard to connect data points to countries, and there were variations of this chart about fat calories with no annotations at all. So I converted the paper to a searchable PDF, annotated the data points, and made it publicly available on the net.
If you have a background in epidemiology, the first thing to jump off the page is the lead author, Jacob Yerushalmy, the famous professor of biostatistics at UC Berkeley. Everyone knew him as Yak in his day. He was best known for torturing Richard Doll and Bradford Hill, who led the British Doctors Study, which inferred that smoking was a causal factor in lung cancer. Yak denied smoking was causal for lung cancer until the day he died in 1973. This book by Bradford Hill, an epidemiologist on the British Doctors Study, was a cornerstone of medical statistics for 50 years through 12 editions. Yak was known for challenging well-known scientists, so no surprise he led this paper with this famous chart from Ancel Keys from his 1953 paper. It showed a strong correlation between fat in the diet and heart disease in six countries. Yak charted FAO data for 22 countries and noted that the correlation was weaker. The movie said this looked like a scatterplot and there was no correlation. - And when you put in all of those countries, it just looked like a scatterplot. There was no clear relationship. - No correlation between fat and heart disease.
Nina's book makes the same bold claim. That's one reason I made a searchable PDF and made it public so people could easily read it. Yak was a statistician, so he computed the correlation coefficients from the two different graphs he provided on fat that characterized the data slightly differently. One would be a perfect correlation and zero is no correlation. I would call these correlations quite significant.
Notice he listed the correlation with animal fat and protein separately, and there was a fairly strong positive correlation tying them both to heart disease. Notice that vegetable fat and protein had fairly strong correlations in the other direction, meaning they were associated with lower incidents of heart disease. That was also true for carbohydrates. ...
The point of Yak's paper is his data shows the correlation between animal protein and heart disease was stronger than it was for fat. So he thought Keys may be chasing the wrong hypothesis. It's a reasonable thing to conclude from his data. So he plotted this chart of animal protein versus heart disease deaths. The movie Fat Fiction didn't show this chart. Yak would've known that a foundational theorem of epidemiology since the 1850s is that chronic disease is multi-causal, but the point of the movie was something different. It was that Keys had access to data from 22 countries, yet he cherry-picked from his six-country graph or his Seven Countries Study. {Yak suspected that} Keys was determined to prove himself right. He followed up with his famous Seven Countries Study, where he traveled to the countries cherry-picked from his graph. For his Seven Countries Study, he repeated the same mistake. I mean, he cherry-picked his countries. That study was to begin a year after Yak published this paper.
But here's how an epidemiologist would look at this chart: Epidemiology works best when you have a large effect size in a well-characterized population. Notice the largest effect size is in Japan compared to Finland and the USA. So they added Finland to the six countries, even though it was scattered to the left. The next thing you'd be asking is what explains the scatter in this graph? That's scientifically fascinating, no? Maybe it's just poor data from the FAO, but you don't know unless you investigate it, which brings us to the Netherlands.
According to Yak's data, they ate a lot more fat than France but had about the same mortality from heart disease, so add them to the study. Epidemiologists would also be thinking, I need a really good team of scientists in each country, and the countries would have to be willing to fund a very expensive long-term study. So they dropped Australia, Canada, and England from the six-country graph, and now they're left with five countries. So how would you pick more countries?
An epidemiologist would say picking from 22 countries when there are 200 in the world would be cherry-picking, and it wouldn't add blinding to the study. Note that data for Greece and Yugoslavia were not available, and seven of the 16 cohorts they picked were from them. The only scientist I'm aware of to write an entire book to explain their study, still available via Harvard University Press, is the Seven Countries team. I have it here, but they were not interested in country-wide data because that would not be well characterized. They were interested in rural populations who had eaten the traditional food of their areas for all their lives and would for a couple decades more before McDonald's and junk food showed up. The scientists couldn't expect those populations to accurately answer questionnaires, so a team of scientists had to visit every six months to collect representative samples of what they ate, freeze them, and send them to the lab for chemical analysis. They also collected blood samples, EKGs, etc.
All Yak needed for his study was the crude data from FAO and a slide rule. I can't think of a more expensive study to conduct than Seven Countries. And what did they find after following the 16 cohorts for 50 years?
Saturated Fat. That the food they were eating when the study started varied in its saturated fat content by five times, but heart disease deaths varied by nine times. The film did not show this chart, and it claimed there was no data showing saturated fat is harmful.
"There is no evidence against saturated fat. There just isn't." "Real food fats are not dangerous. They are not the enemy." "It's too simplistic and reductionist to say saturated fat is dangerous and needs to be avoided, and the evidence does not support that."
It's hard to fake this data or to blame it on Ancel Keys, who retired after the first 20 years and passed before this was published. This data is quite consistent with other data of the era. The famous French epidemiologist, Serge Renaud, observed that in the '60s and '70s, French populations in the north of France, which relied on meat and dairy, had about four times the incidences of heart disease and lived five years less than French populations near the Mediterranean who were eating the Mediterranean diet. He noted that by the time he co-founded the famous Lyon Diet Heart Study, that effect had faded because all types of food had become available everywhere. But his findings were still consistent with the Seven Countries Study, noting how healthy the Mediterranean diet is for the general population.
By the way, I talk to Henry Blackburn often with more questions about the Seven Countries Study. He'll turn 100 in a month and he's still playing in his band. He was a core member of the study for the full 60 years of it. I asked statisticians in AI, what are the odds that three of 18 scientists born in America in the early 1900s with no genetic relation would make it to 100? The answer they came back with was roughly one in a billion or one in 10 billion. Also, two others on the team died at 97, and a third at 95. And I've been thinking of interviewing Alessandro Menotti, who oversaw the Italian cohorts and is 91.
Gary Taubes Article on Epidemiology. Moving on — in 1995, Science published a very popular article by Gary Taubes with a message that epidemiology was near its limit. It was scaring the public with false positives, and it was a blunt instrument incapable of discerning small effect sizes. Perhaps epidemiology was done advancing, but time and science march on. Twenty-eight years later, a group of epidemiologists followed up on that Science article to see which claims were indeed false positives and needless scares.
The thing is, big data is getting much better with bigger data sets, more compute power, and big advances in causal reasoning. Wouldn't that be true of epidemiology? Indeed, the Harvard studies alone have become much bigger data. They've added another 200,000 people and another 30 years of observation since that article was published. Environmental epidemiologists can only dream of data sets like this.
We've done three interviews with epidemiologists at Harvard and asked them to their faces how they answer their toughest critics. The Framingham Study just celebrated their 75th year. There are 150 scientists and technicians involved in that study. I toured their campus with my camera rolling and interviewed the director. They followed three generations, sequenced their DNA, and they measure almost everything you can measure when the participants come in every two years for five hours of bone scans, blood tests, glucose tolerance tests, etc. No other study has data like this.
When The Washington Post compiled the list of 10 medical milestones for the last century, they listed Framingham as number four — and that's epidemiology. So how well did the Science article age? It didn't list many indisputable positives of epidemiology like lead, but it did pick a lot of things it considered sketchy in 1995. Here's a very partial list: 30 years later, no one is saying any of these are not causal factors, such as tanning beds and melanoma. But it didn't stop industries in America from trying to wreck the careers of scientists who dedicated their lives to studying these things.
Diet Confusion in America. So this is where we were in the '60s — the TV star and the radio star giving opposite advice. And this is where we are in America 60 years later. For every doctor and influencer who believes their diet is the one, there is another doctor and influencer who believes the exact opposite. No consumer can make sense of it.
I think we Americans have reason to be humble about where we are in this chart. There is so much corporate influence here, and we're making supplement salesmen and social media stars unimaginably rich selling what consumers want — like magic pills and a new miracle diet every few years. But here's the thing: we researched dietary guidelines around the world and identified 114 countries with independent scientists who create them. It's amazing how similar they all are.
So we interviewed scientists on their committees from countries like the US, Norway, Canada, Japan, and Switzerland. As far as we can tell, they all have epidemiologists on their committees who understand the big data modern epidemiology is producing. I'm not aware of any popular American influencers who invite credible epidemiologists on their shows. Instead, they vehemently deny that epidemiology is valid science. I've seen that movie before.
When Google's founders were students, they tried to sell their search engine to existing companies like Excite and Yahoo, but they couldn't find a buyer because Excite and Yahoo thought the web would be far too noisy for search to work. Anyone with common sense could see the way forward was human curation of the internet, as they both were doing. Whoops. Underestimating Google's talent with big data cost them their companies.
I made an episode about Japan school lunches. They have a dietician in each school who designs their meals based on Japan's food guide and teaches it to the children. The largest category on their visual guide is grains, just like it is in the US, but their obesity rate is 4%, and they have an exceptional life expectancy.
"So, Chris," you say, "you open with the bold statement that big data will revolutionize nutrition. How can you believe that?" That's because it's already happening in countries that typically lead us by a decade or two on topics like this. Denmark banned trans fats 15 years before we did. Uganda is leading the effort for African nations.
"But Chris," you say, "you are vegan, and yet none of those 114 food guides are 100% vegan, including Denmark's and Uganda's. Not even the epidemiologists at Harvard or Framingham are vegan. Why then are you?"
I know several of you have posted on social media that being vegan is an eating disorder. The thing is, I'm an earth scientist. All the prominent earth scientists I know advocate for a vegan diet: James Hansen, Johan Rockstrom, Stefan Rahmstorf, Hannah Ritchie, etc. We don't like the land use, water use, pollution, pandemic risk, pathogens, antibiotic resistance, greenhouse gases, deforestation, etc. I'm not aware of any of those scientists having an eating disorder. They're all models of health, far as I can tell, even into their 90s.
The Pushback.
First of all, before I even gave my talk, Ken Berry came up to me, wished me the best, and was 100% gracious. He said he watches some of my episodes. But I noticed after the conference, he posted on Twitter that the carnivore diet is about to go mainstream and “poor vegans.” He's probably right because it does go mainstream every two generations. It went mainstream under Blake Donaldson two generations ago, Vilhjalmur Stefansson two generations before that, the Cutters in the early 1900s, James Salisbury in the last quarter of the 1800s, and I could go back centuries like that with my collection of old books. The trouble is populations who adopted these diets have significantly lower life expectancies on average compared to populations who eat Mediterranean, Asian, and vegetarian diets. So it gets forgotten for two generations before someone revives it again.
Other people who responded to me did so after both my talk and the 90-minute debate, so I should say something about that. One point I made is that there are no ruminants to emit methane in the Amazon except for a few small deer whose numbers are low because of jaguars and competition with other herbivores. We were in the Peruvian Amazon a few months ago watching it burn to the ground to make way for huge cattle ranches and slaughterhouses. They export beef and leather via meat-packing giant JBS, who are turning the Amazon into a savanna that converts oxygen from the atmosphere into the potent greenhouse gas methane, instead of trees turning carbon dioxide from the atmosphere into oxygen. That’s what the remaining trees in the Amazon do.
My opponent in the debate said, "Introducing cows to the Amazon was simply because poor people want better nutrition." Nick Norwitz said something to me in the hall, which he said several times before, that hit jobs on beef are fuel for the carnivore movement. He said his three most popular YouTube episodes were about the carnivore diet, and I noticed after the conference that he and Dave were both posting on Twitter promoting beef, but not mentioning the environmental disaster that it is. That is a gut punch to earth scientists like me.
So the question is, did I do some good by speaking at the conference or did I do harm? This is the question Herb Needleman had to wrestle with because the response to his research on lead was for Dutch Boy Paints to ramp up their ads, promoting the health benefits of lead for children. So I think I should learn a lesson from Nick that my talk at this conference just added fuel to the burning fires of the Amazon. No, I wouldn't do {such a conference} again. I wanted to leave it there, but when I sent out drafts for fact-checking, the feedback I got was I should explain what I would do instead.
What I'd Do Instead. The situation in America is that about 10 YouTube stars are responsible for more than 50% of nutrition views on YouTube, far as I can tell, and I'm trying to get more accurate data. None of them have any background in conducting credible scientific studies, and most of them sensationalize anti-science.
"Is there any such thing as good sugar?" "No, so even all the fruit that we find today in supermarkets is not natural." Some people believe that if the sugar is coming from a fruit, for example in a fruit smoothie, that's good sugar, but that's a total lie.
My perspective is that big brands have spent a fortune in America over the past century discrediting science via disinformation, and they have succeeded among American consumers. For example, France banned lead in all paint by 1908. The League of Nations advised countries to ban it in 1922. But in America, Dutch Boy Paints was distributing coloring books to children promoting the health benefits of lead until Herb Needleman prevailed in 1978. But we just lowered the permitted amount to 600 parts per million in residential paints. And the regulations for commercial paint are so complicated that, in the immortal words of Nate Bargatze: "Nobody knows!"
And now we're exporting American social media misinformation to the rest of the world, just like we exported McDonald's. In 1961, Time put scientists on the cover as Men of the Year. You don't see that anymore, but once upon a time it was common. For example, here’s A.J. Carlson, who was a giant in the field of physiology, on the cover in 1941, despite the big topic of the day being the war.
Also in 1961, they placed another giant in the field of physiology on the cover, Ancel Keys. It wasn't until he was almost 100 and could no longer defend himself, that influencers were able to wreck his legacy in the eyes of consumers by making up things about him. Actual scientists, however, were not fooled and continue to regard him as a giant in the field of nutrition. The reason they do is we now have 75 years of very big data to verify his results.
The thing about scientists who actually conduct credible studies is they either understand big data or they team up with epidemiologists who do. I have never seen that among popular social media influencers. The mission of our channel is to find credible scientists and help them tell their stories to earnest audiences who are trying to cut through the noise of dueling influencers. The most credible scientists are not YouTube stars. They're too busy actually conducting science.
For example, three weeks ago, we published an interview with Iman Lee, the co-director of the Harvard Women's Study. She's an epidemiologist, and there wasn’t a loose fact to be found. That has 200,000 views already and great comments about the things people learned. Doesn't it feel like I should stay focused on that mission?