Seth's Blog

… is the title of a New York Times Magazine article about the ketogenic diet, a treatment for childhood epilepsy, which I’ve blogged about several times (here, here, here, here, here). It’s a very-high-fat diet. It interests me for two reasons: (a) It connects a high-fat diet with proper brain function, as my self-experiments have done. A curious feature of the ketogenic diet is that it isn’t permanent. After several years the child can go off it. My self-experimentation suggests that Americans eat far too little of certain fats. Perhaps eating enough of these fats would prevent childhood epilepsy. (b) It shows how someone who cares enough — in this case, Jim Abrahams, whose son had epilepsy — can be more effective than professional researchers and doctors. Abrahams rediscovered the diet. He saw its value, the professionals didn’t. I’ve argued that this is part of why my self-experimentation found new solutions to common problems: because I had those problems. I cared more about finding a workable solution than researchers in those areas, who had several other concerns (publication, funding, acceptance, etc.).

The details of the article reminded me of something I learned in the BBC series The Story of Science. For hundreds of years, medical students were told, following Aristotle, that the liver has three lobes. It doesn’t. You might think that examination of thousands of actual livers would have dispelled the wrong idea, but it didn’t. The article contains many examples of doctors ignoring perfectly good evidence in favor of nonsense they read in a book or heard in a lecture. Epilepsy is easy to measure. If a child has 100 seizures per day, and has been having them at this rate for years, and this goes down to 5 shortly after he starts the ketogenic diet, and goes up again when the child goes off the diet, there is no doubt the diet works. As early as the 1930s, this had been observed hundreds of times. This was overwhelming evidence of effectiveness. Doctors ignored it, probably based on the modern equivalent of the three-lobed liver. They complained, according to the article, that there was “no evidence it worked” or that the evidence wasn’t “controlled” or “scientific” (whatever that means). A study published in 2008 “answered doubts about keto’s clinical effectiveness” — as if doctors needed the equivalent of a very-large-type book to be able to read what most of us can read with normal-sized type.

According to the article, “by 2000, more people were asking about keto, but most pediatric neurologists still would not prescribe it” — as if the parents needed the approval of their doctor to try it. You don’t need a prescription to buy food.

Thanks to Tim Beneke, Michael Bowerman, Alex Chernavsky, David Cramer, and Peter Couvares.

John Cassidy, a staff writer at The New Yorker, understands clearly the poor judgment of economics professors. In How Markets Fail he said the Nobel Prize in Economics has made things worse, because it has often been given for worthless work. Outside of economics, however, he can write this:

That such a smart well-informed non-party-liner can believe Lipitor is wonderful shows Orwell was right: with enough repetition, people can be convinced war = peace. Here is the truth about Lipitor:

Statin therapy is extremely efficient in lowering cholesterol numbers, but unfortunately not without adverse effects on the body. To prevent a first heart attack, for every life that is saved – 1% over 10 years of use – statins cause an equal number of adverse deaths due to accidents, infection, suicide and cancer — 1% over 10 years’ use and significantly greater levels of serious side effects and suffering. . . . In a study to see the effects of raising the Lipitor levels from 10 to 80 mg (more sales) on patients, those taking 80 mg had increased liver problems, that is the rate of raised liver enzymes was six times higher than those given 10 mg of Lipitor. Even though the total deaths due to CVD in the 80 mg group was fewer (126) than in the 10 mg group (155), the total deaths due to other causes was higher in the 80 mg (158) than the 10 mg (127) group. There was no difference in the overall mortality rate.

Lipitor, the miracle drug. Taken by millions at a cost of billions. This is what happens when you — such as those in charge of health care — have little understanding of a problem: You aren’t good at solving it.

Cardiologists believe that high cholesterol causes heart attacks. Their depth of understanding was illustrated by the cardiologist at my Quantified Self talk about butter who said that the Framingham study showed that diet caused heart attacks (no, it found new correlations between heart disease and “risk factors” such as cholesterol — see also this) and that the recent reduction in heart attacks is evidence of our improved understanding (e.g., the science behind Lipitor). That a thousand other things changed over the same time period he apparently hadn’t considered. He simply couldn’t defend — at least then — his core belief that butter was bad. A cardiologist! How many thousands of people has he told to eat less butter?

Cassidy’s article about the harm done by the financial industry, from which that quote was taken, is excellent.

After I moved to China in September, I was surprised that my arithmetic speed went down. (That is, I got faster.) I had lowered it from about 630 msec/problem to 600 msec/problem by eating lots of butter. I had no idea how to lower it further. I didn’t deliberately change my diet in China but it was quite different. I kept some things the same: the amount and brand of butter/day, the amount and brand of flaxseed oil/day.

I failed to figure out why I had gotten faster. I reduced the amount of flaxseed oil from 3 T (tablespoons) per day to 2 T per day. It made no difference. (In the beginning of my interest in flaxseed oil, change from 2 T/day to 3 T/day had made a difference.) Perhaps because of the butter.

Surprised that the change from 3 T/day to 2 T/day hadn’t made a difference, I went down to 1 T/day for two weeks, then back to 2 T/day. Both changes made a difference:

Each point is a separate test. Each test had 32 arithmetic problems (e.g., 3+4, 11-3). In the beginning of the data shown in the figure I tested myself once per day. After 12 days I started doing two tests/day, one right after the other. I was curious about the repeatability of the numbers; it wasn’t hard; it was a way to get better measurements. Averaging over the tests for each day to get one value per day, combining the 19 2-T/day (before) days and the 11 2-T/day (after) days, and comparing the combination to the final 7 1-T/day days, t(38) = 6.5. If you’re not familiar with t values, t = 2 is a barely reliable difference, t = 4 is a very clear difference.

This is more evidence that flaxseed oil improves brain function. It interests me because it implies the optimum dose is close to 2 T/day. It cost about $20 and took 1 person-month. In contrast, the DHA-Alzheimer’s study I mentioned two days ago cost about $1 million and took about 7000 person-months. And used (a) a cruder something-versus-nothing comparison, b) a less-sensitive between-subjects comparison, and (c) a more ethically-problematic placebo-controlled design.

A 1997 epidemiological study, which I just learned about, found that increases in saturated fat intake were associated with a lower risk of stroke. Sampling from among the papers that cited it, this study found a non-significant change in the same direction. This study found a significant change in the same direction in Japanese, who eat a low amount of saturated fat. This is especially interesting because many people assumed that the high rate of stroke among Japanese was due to high salt intake. This finding suggests it is due to low saturated fat intake.

As I said in a recent post, this sort of consistency across studies on a question of enormous interest argues against the severest critics of epidemiology, such as John Ioannidis.

Last week a journalist asked me why the 5% improvement in arithmetic speed produced by butter was important. In an earlier post I said I’d given a poor answer.A few days later I figured out what I should have said. The article was delayed, it turned out, so there was time to use my new strategy. I answered the question like this:

I was excited by this discovery because it was so big and unexpected. Someone once found a correlation between IQ and reaction time. The higher your IQ, the faster your reaction time. I don’t know what the exact function was but a decrease of 30 milliseconds might correspond to 10 more IQ points. I felt a little bit smarter. It was so unexpected because hardly anyone was going around saying butter is good for you — and thousands of people were saying it is bad for you. The only ones saying butter is good for you were the followers of Weston Price, and they had almost no evidence for what they were saying. Compared to their evidence, my evidence was crystal clear. Among mainstream nutritionists, butter is universally scorned. Yet my data suggested exactly the opposite — that it had a large amount of an important nutrient I wasn’t getting enough of. If mainstream nutrition advice could be so wrong, it would have big implications for what we eat. Maybe other things we are constantly told about what to eat are also wrong.

I discovered this big effect of butter by substituting butter for pork fat. So the reason butter was so helpful wasn’t anything as simple as animal fat is food for us. I ate plenty of animal fat before I started eating lots of butter. The reason was something more specific.

Eating a half-stick of butter (60 g) every day apparently improved how fast I can do simple arithmetic problems (e.g., 7-5, 3+1). I improved about 5% — from 630 to 600 msec per problem. My scores had been at 630 msec for months. They suddenly dropped.

A reporter said to me that a 5% improvement isn’t much. You couldn’t notice it. Why did this matter?

I did not reply “what good is a newborn baby?” I said it mattered for three reasons:

1. You cannot easily produce such an improvement. I was already doing very well. For example, I had already lowered my scores a lot via omega-3. Imagine the world record for the 100 meter dash suddenly dropping 5% due to eating something you can find in a supermarket.

2. A 5% improvement is just the beginning. There is room for optimization — better dosage, better timing of taking the butter, and so on.

3. The brain is a mirror of the rest of the body. Learning the best diet for the brain, at least in terms of fat, will help us learn the best diet for the rest of the body, just as learning what house current is best for one electrical appliance is a guide to what other electrical appliances are designed for. They’re all designed to work with the same house current.

Alas, this is not just a poor answer, it’s what I actually said. I give myself a C+. Reason 1 is almost gibberish. Reason 2 is technocratic. A good answer is more emotional. Reason 3 is okay, if not very clear.

At Berkeley I knew a student who had transferred from a junior college. He is/was black. He had probably gotten into Berkeley because Berkeley administrators wanted to admit more black students. He complained one day that he got C’s on his essays even though “all the words were spelled correctly.” It was frustrating, he said. I am in a similar situation here. My answer is poor but I cannot easily do better.

To my pleasant surprise, Mark Frauenfelder posted this call for volunteers. Will eating half a stick of butter per day or a similar amount of coconut fat improve your performance on arithmetic problems? Eri Gentry is organizing a simple trial to find out. The trial is inspired by my recent Quantified Self talk. Study details.

During the question period of my talk, I responded to a question about a trial with 100 volunteers by saying I would suggest starting with 2 volunteers. A reader has written to ask why.

What’s your reasoning behind suggesting only 2 volunteers to test the eating more butter results? You seem highly convinced earlier in the video, but if you were so convinced why not have a larger trial?

Because the trial will be harder than the people running it expect. If you’re going to make mistakes, make small ones.

This is my first rule of science: Do less. A grad student in English once told me that a little Derrida goes a long way and a lot of Derrida goes a little way. Same with data collection. A little goes a long way and a lot goes a little way. A tiny amount of data collection will teach you more than you expect. A large amount will teach you less.

My entire history of self-experimentation started with a small amount of data collection: An experiment about the effectiveness of an acne medicine. It was far more informative than I expected. My doctor was wrong, I was wrong — and it had been so easy to find out.

This may sound like I am criticizing Eri’s study. I’m not. What’s important is to do something, however flawed, that can tell you something you didn’t know. Maybe that should be the first rule, or the zeroth rule. It has the pleasant and unusual property of being easier than you might think.

Thanks to Carl Willat.

Two studies of the effect of dairy consumption on health have recently appeared. Both suggest it is healthy. One of them– a prospective study where about 1500 people were followed for 16 years — found no association of dairy intake with overall mortality but did find a protective effect of full-fat dairy against heart disease. The study considered lots of possibilities and the authors write ” it is important to take into account the large number of comparisons considered in this study and thus we cannot rule out the possibility that the protective association between full-fat dairy intake and cardiovascular mortality was due to chance.”

I mentioned this study earlier. It gains more credence because of the other study, which is a meta-analysis. The second study found protective effects of dairy products on several outcomes, including overall mortality:

Meta-analyses suggest a reduction in risk in the subjects with the highest dairy consumption relative to those with the lowest intake: 0.87 (0.77, 0.98) for all-cause deaths, 0.92 (0.80, 0.99) for ischaemic heart disease, 0.79 (0.68, 0.91) for stroke and 0.85 (0.75, 0.96) for incident diabetes.

This is good news for me since I eat yogurt and butter every day.

Thanks to Peter Spero.