Note: Part one of this three-part series is also accessible online – Nutrition science serves a big fat lie. On Wednesday, we learned how we got to today in the fat debate, learning from Nina Teicholz’ book, “The Big Fat Surprise: Why Butter, Meat and Cheese Belong in a Healthy Diet.” Today, we’ll talk about the research, or lack thereof.

Long-story short, the American Heart Association bloomed to be a leading health advocate at the same time heart attacks were the nation’s leading killer. They went with a hypothesis linking a low-fat diet to less heart risk, and the rest is history, or so we thought.

Studying a bad example

One of the first things I learned in my experimental design and statistics coursework at the University of Minnesota’s graduate school was how important a representative sample is. Without a representative sample, you have nothing. Or worse; you think you have something when you actually have nothing.

But over 60 years ago, Dr. Ancel Keys, eventually forgot that part of his studies, or possibly didn’t realize his error and was too vested in the results.

Regardless, when I reference Dr. Keys, I’m not talking about a half-brained person. By many counts, the man was a genius. An obituary tells that he was one of 1528 “gifted” students discovered through a test. He graduated from the University of California at Berkeley, with bachelor degrees in economics and political science and a master’s in zoology. He received a Ph.D in oceanography and biology and bounced around to many research positions and then taught at Harvard, earning another Ph.D. there in physiology.

Then, in 1936 he came to Minnesota with the Mayo Foundation. Not liking the Mayo work environment (not enough research), he headed to the University of Minnesota in 1937. There, he developed the infamous K-ration (K for Keys), a meal that fit in a box in a pocket, which millions of soldiers ate during World War II.  

He then studied the effects of testosterone, vitamin supplementation, and starvation.

But in the 1950s, seeing the contrast in heart disease between well-fed American businessmen and those in underfed Europe, he set his sights on cardio-vascular disease research. When he saw that southern Italy contained the highest number of centenarians, he looked further into the Mediterranean diet.

As Teicholz explains in the book, Keys’ hypothesis about the Mediterranean diet came at a time when the country was looking for answers on the heart disease problem. It also came as lipid (fat) research exploded, as gas-liquid chromatography allowed scientists to test different fats’ effects on human biology. Early experiments showed that replacing animal fats with vegetable fats dramatically reduced cholesterol levels – a suspect as a leading cause of heart attacks.

Keys gave different diets to schizophrenic patients at Minnesota hospital ranging from 9 to 24 percent fat content, and saw that lower fat diets performed a little better in lowering cholesterol. But, the study was performed on only 60 patients, and he included none of the details in his final paper. Using these results, he professed that a perfect curve existed between plaque accumulating on artery walls and fat in the diet. He called it the “diet-heart hypothesis.” Further, he thought that fat must make people fat.

The logic makes sense, but is that only because we’ve heard it for over 60 years? Fat does have 9 calories per gram, versus 4 calories per gram for protein and carbohydrates.

To prove his point – although there was already some data on the subject from Eskimos, the Masai of Africa, and the Navajo Indians – he and his wife traveled to many countries, measuring cholesterol levels of adult males in South Africa, Sardinia, Sweden, Spain, Italy, Finland and Japan. They would visit remote villages, testing subjects including fisherman, farmers, and loggers.

While Keys did write down what he saw some of his early subjects eat, he had no great way to track diets. Further, he did travel to many locations, but his sample sizes were quite small.

Back in Minnesota, he saw that cholesterol would go up after the schizophrenic men ate saturated fat (found in animal fats) and down when they consumed vegetable oils. With this knowledge, he published the “Keys equation,” where he advocated dropping eggs, dairy products, meats and all visible fats to lower the heart disease rate in a population. He told everyone to switch to vegetable fats.

Keys published a paper comparing 6 of the countries he visited, and linked mortality to percent fat calories in the diet. A biostatistics professor from UC-Berkeley thought the results looked fishy, and then compiled data for the 22 countries in which national data was available – the correlation disappeared.

But Keys refuted the other studies, calling the data flawed. Then, he used selected portions the same data set to defend his own study.

The study to end all studies

Keys would standardize data collection in 7 countries that fit his hypothesis well; Italy, Greece, Yugoslavia, Finland, the Netherlands, Japan, and the United States. Greece, Italy, and Yugoslavia were still recovering from WWII. It was called the Seven Countries study, and a 2004 analysis shows that it has been referenced over a million times since it was published.

Keys and his collaborators in each country measured body weight, blood pressure and cholesterol levels of adult men in each country. Sampling was done in the 1960s and revealed in the 1970s, in an American Heart Association journal.

But when Teicholz looked further into the Seven Countries data, she found glaring errors. Greek islands Crete and Corfu were sampled multiple times, and once during Lent. 60 percent of those on Crete were fasting one of the three times they were sampled. Based on this obviously errant sampling, we’re supposed to follow a Mediterranean diet.

Keys reported the study’s details in a little-known Dutch journal that he had previously been disappointed in, in terms of publicity. But the AHA journal still held the fireworks, allowing Keys to focus on his key takeaways. The paper also makes no mention of other problems with the study, like using multiple methods of evaluating fat in food samples.

Going back 25 years in 1999, the lead Italian researcher re-analyzed data for his country and found a higher correlation with sweets (0.821 – and not including chocolate, ice cream, and soft drinks) than that of animal foods (0.798 – including margarine, in this case). Sugar was only lightly touched upon in Keys’ paper. Yet, this Seven Countries study is considered the final evidence of the theory that saturated fat in animal foods causes heart disease.

But again, this was only a correlation, or an association. It was not a cause-effect study because it wasn’t a clinical trial, instead it was an epidemiological survey.

Meanwhile, reputable studies in North Dakota, Ireland, India, Pennsylvania, Massachusetts, of competing tribes in Africa (one vegetarian and one meat-eater), Eskimos, and even an egg-eating doctor countered some of Keys’ research. But, today, most of those dissenters have died, and the “logical” cholesterol leading to heart disease theory stands firm, according to Teicholz.

Keys and a colleague jumped on an American Heart Association committee in the late 1950s, despite neither being trained in nutrition. That committee advocated dumping saturated fats found in meat, cheese, whole milk, and other dairy products in 1961, and Ancel Keys wound up on the front of Time magazine.

The genius, Keys, was a persuader. The animal fat theory won out over the carbohydrate and sugar theories. Keys would refute, rather than just comment, on opposing papers.

The polyunsaturated puzzle

With the American Heart Association, National Institutes of Health, and USDA all favoring a low-fat diet, it can be hard to imagine the new “food pyramid” or “MyPlate” tools coming out with a positive view of fat. But until 1910 the only fats found in the kitchen came from animals; lard, butter, eggs, meat and dairy products. When vegetable oils leapt on the scene, they advertised the polyunsaturated oils as a huge health benefit for lowering cholesterol. “Take this ad to your doctor,” read one Mazola corn oil advertisement. But none of the other effects of these oils had not been studied, like the later-found existence of trans-fats.

The National Heart, Lung and Blood Institute (NHLBI) tried to create a study using the right kind of science – a clinical trial. But Keys was put in charge, along with a colleague, and they attempted to create new foods from oils that would resemble animal meats. This included milk “filled” with soybean oil.

The NHLBI held a pre-study trial that, if fruitful, would be sized up for a much bigger study to put the stamp on the diet-heart hypothesis. But the trial didn’t last long, as 25% of men dropped out because they found it too hard to keep the diet between always eating at home and convincing their wives to cook without animal fats.

So, NIH peeled back to fund two smaller $250 million studies, still the largest nutritional trial at the time. The results were a disaster for the diet-heart hypothesis. Keys’ colleague who ran the study called it “troublesome” and “unexpected.”

But, there have since been three studies that put the diet-heart hypothesis to rest. However, with an overwhelming crowd shouting them down, they’ve never been well-publicized.

In the 1980s, Dr. Dean Ornish controlled the airwaves and promoted the near vegetarian diet. He competed with Dr. Atkins, studying thing this exact issue at the same time but advocating in the other direction for a high-fat diet. Both sold millions of their books. But Ornish’s single trial turned into several scientific publications. Meanwhile Atkins’ handful of smaller trials didn’t receive as much respect.

What should we eat?

In summary, the science simply isn’t there to advocate a low-fat diet for all Americans. Worse yet, while the studies were done on adult males, we’ve pushed the terrible suggestion onto women and children over age 2, as well.

If you read the book, you’ll see that Teicholz serves up even more evidence against a low-fat diet. She notes that before 1910, when hydrogenated oils burst on the scene, we were eating far more saturated fat than we are today. Also rising with heart attacks between 1910 and the studies in the 1950s was vegetable oil usage, sugar consumption, smoking, and car exhaust fumes, in nearly every country.

We’ve learned much about cholesterol since then, including the difference between LDL (bad) and HDL (good) cholesterol.

Fat is the soul of flavor, as Teicholz puts it – we’ve learned this as we removed it from the diet. We traded saturated fats for Trans fats, and put the American Heart Association logo on cookies, crackers and even Rice Crispie bars. The oils, except for olive oil, create toxic aldehydes when heated, yet we cook with these almost everywhere.

If you trust Teicholz’ advice, it is saturated fat we need back in our kitchens. Fat does not cause obesity. We’ve reduced our consumption of saturated fat by 14 percent since the AHA guidelines first arrived. Obesity is at record levels, near 40 percent for those aged 45 to 64.

While I’m not able to provide the full picture in this already lengthy article, pick up Teicholz’ book to learn more. Stick to whole fat dairy, eggs, and meat, even the fatty stuff. Olive oil can also stay on your “good list,” as

not part of the polyunsaturated family of oils that become unstable when heated. The high fat foods don’t have real links to diabetes or heart disease, either.

When asked to follow advice of a book author versus the scientific establishment, it seems counter-intuitive to follow one person’s opinion. But, for my health, I’ll be happily moving towards a higher-saturated fat diet.

The trend may be changing already. Butter is back, whole milk sales are up, and lard dinners are a new fad in New York City. We’ve been part of one of the worst-tasting experiments of all time. Chew the fat, drink the [whole] milk, fry some bacon, and toast with a stick of real butter for a healthier planet.


Part 1 – Nutrition science serves a big fat lie (Friday, June 25)

Part 3 – The sad state of science (Monday, June 30)