Author: Seth Roberts

Nick Winter and Percentile Feedback

Seth Roberts1 Comment


How much did percentile feedback help Nick Winter? In 2011, he wanted to finish coding a mobile app for Skritter. This graph shows his daily work on that app. Each point is a different day. The lines are loess fits.

At first, he tried to work on the app but did not make it a priority. Then he made it a priority compared to other things he did. Then he set an explicit goal (a certain number of hours of work each day) and used Beeminder to help reach that goal.

Finally he tried percentile feedback. It helped a lot. His work per day increased about 60% (to 8.6 hours/day) compared to the previous phase (5.5 hours/day). Using percentile feedback, he finished the app. During the launch, he worked on the app much less.

The data has several interesting features. One is the sudden improvement when percentile feedback started. The same thing happened to me. Another is that it helped him even though he was already making steady progress. A third is that the sudden improvement happened after he had been working on the task a long time (more than six months). Things become easier to do the more we do them but surely this change was complete by the time he started percentile feedback. Apparently it engaged a different source of motivation. Finally, he tried several things before he tried percentile feedback, implying that its value wasn’t obvious. It wasn’t obvious to me, either. I originally tried it to see what would happen. I didn’t have a strong belief it would work.

Above all, this graph shows it’s possible to learn from long-term self-measurement and what you learn may not be obvious. People who know little about research sometimes say that randomized double-blind trials are the only convincing way to learn something.

The Rise and Fall of Heart Disease

Seth Roberts11 Comments

Heart disease was once the number one killer in rich countries. Maybe it still is. Huge amounts of time and money have gone into trying to reduce it — statins, risk factor measurement (e.g., cholesterol measurement), telling people to “eat healthy” and exercise more, and so on. Unfortunately for the poor souls who follow the advice (e.g., take statins), the advice givers, such as doctors, never make clear how little they know about what causes heart disease. Maybe they don’t realize how little they know.

I encountered an ignorant-without-knowing-it expert after a talk I gave about the effect of butter on brain function. I found that butter improved my brain function (measured by arithmetic speed). I had been eating lots of butter for more than a year. A cardiologist in the audience said I was killing myself. He thought butter caused heart disease. I said that I had experimental data that butter was good for me. Easy to interpret. The notion that butter is bad has come from epidemiological (non-experimental) data, which is hard to interpret. The cardiologist said that the epidemiology has not been misleading. One sign of our correct understanding, he said, is that heart disease has declined. I said there were many possible reasons for the decline.

A 2012 paper called “An epidemic of coronary heart disease” by David Grimes, a British doctor, could hardly make clearer how little we know about the cause of heart disease. Grimes points out that before 1920 heart disease was almost non-existent, that it rose sharply from 1930 to 1970 and since 1970 has declined sharply, at roughly the same rate that it rose. Both the rise and the fall are mysteries, says Grimes, in agreement with what I told the cardiologist. The rise and fall contradict all popular explanations. Heart disease cannot be due to obesity or wealth — both increased substantially at the same time heart disease fell sharply. Nor was the decline due to government intervention:

The decline of CHD deaths in the UK was further described in a UK Government report of 2004, Winning the War on Heart Disease. In this report, the government predictably but undeservedly assumed responsibility for the decline. Clearly, the NHS [National Health Service] in the UK could not have had an international effect [the decline is international].

“There [has been] no obvious effect of statin therapy or other medical intervention,” Grimes continues. Yet statins continue to be prescribed in very high amounts and very great expense. The NNT (number of people you need to treat to save one life) is often in the thousands, he noted.

Those who complain about the high cost of health care fatally fail to grasp this enormous ignorance — about many things, not just heart disease — and its consequences. Reducing the cost of health care (reducing the cost of statins, for example) would improve health if cost were the only thing deeply wrong with our health care system. It isn’t.

Assorted Links

Seth Roberts3 Comments

Thanks to Phil Alexander and Casey Manion.

Walking and Learning: GRE Words

Seth Roberts3 Comments

Most of my earlier examples of the benefits of walking while studying involved treadmills and learning a foreign language. A Stanford student named Govind writes:

I found I was able to memorize GRE words very effectively while walking [compared to sitting]. It not only made the process much more enjoyable, but since I walked outside (around Oxford [England]), I also was able to associate words with physical cues. The difference between propitiate and propitious is now inextricably linked to Cowley. (I am now a memory palace convert.)

At Berkeley, I once assigned my intro psych students to do self-experiments. One of them measured how many French words she could study before falling asleep. She tried three body positions: sitting at a desk, lying on her side on her bed, lying on her stomach on her bed. She also tried three audio environments: silence, classical music, heavy metal. Best combination: lying on stomach, heavy metal. Worst combination: sitting at desk, silence. This amused me, but I now see that the real lesson of her experiment is that she didn’t try walking. It shows how little-known the walking-helps-memorize idea is, even though the effect is easy to notice, as Govind’s story shows.

Aquatic Ape Theory Revised

Seth Roberts17 Comments

I became interested in the aquatic ape theory of evolution because it pointed me in a fruitful direction — testing omega-3 fatty acids (e.g., flaxseed oil), which turned out to have easy-to-detect benefits (better brain function, better gums). That is more than I can say for alternatives to that theory, such as the savanna theory. Marc Verhaegen, a Belgian doctor, has recently proposed a new version of the aquatic ape theory. Some of his main points:

  • An extensive overview of the literature by Stephen Munro showed that virtually all known archaic Homo [= pre-Homo sapien] sites (including those in ‘savanna’) were associated with permanent water and edible shellfish.
  • Only regular diving can explain archaic Homo’s pachy-osteo-sclerosis (POS), the extreme thickness and density of cranial and postcranial bones of most erectus-like fossils. . . . POS is only seen in slow littoral divers, e.g. dugong and manatee, walrus, Kolponomos, pakicetids, Odobenocetops, and Thalassocnus spp. Marine biologists agree POS has a hydrostatic function (ballast).
  • The abundant brain-specific nutrients in aquatic foods (e.g. DHA, iodine) facilitated fast brain growth (sapiens’ poorer post-aquatic diet required a longer youth to grow the same brain size).
  • Man is the opposite of a savanna inhabitant. Humans lack sun-reflecting fur, but have thermo-insulative subcutaneous fat layers, which are never seen in savanna mammals. We have a water- and sodium-wasting cooling system of abundant sweat glands, totally unfit for a dry environment. Our maximal urine concentration is much too low for a savanna-dwelling mammal. We need much more water than other primates, and have to drink more often than savanna inhabitants, yet we cannot drink large quantities at a time.
  • Maps of human population densities show that, although we have become fully terrestrial today, we are still a waterside species, and almost half of human dietary calories still come from the water (e.g. rice, aquaculture, fish, shell- and crayfish).

I find the water-drinking point especially persuasive. We need to drink throughout the day, or at least feel bad if we cannot. Almost all workplaces, including cafes, have a source of water. This is inconsistent with savanna living and consistent with waterside living. The term aquatic ape is somewhat misleading. A better name would be aquatic-food ape.

Resistance to Fecal Transplants as Treatment for C. diff. Infection

Seth Roberts7 Comments

One of the worst infections you can get in a hospital is C. difficile. It is notoriously unpleasant and hard to get rid of. It has recently been discovered that fecal transplants are highly effective against this infection. Here’s what happened next:

The Food & Drug Administration (FDA) [decided] to require an Investigational New Drug (IND) application for stool transplants—formally known as “fecal microbiota transplants (FMT)”—for the treatment of C. difficile colitis. “C. diff,” as it is known, is a severe inflammation of the bowel . . .

Over the last 10 years of my practice, I saw a change in the patients I treated for C. diff. More patients were affected, they were generally more severely ill, and the infection became increasingly difficult to treat. . . . often being refractory to therapy. . . . I also began to see patients floridly septic from C. diff, occasionally needing emergency surgery to remove their colon (colectomy). [I began] to wonder whether we shouldn’t be treating severe cases of acute C. diff with stool transplants. I reasoned that it was a better alternative to an emergency colectomy. . . .

There are barriers to doing so, however:

First, there is the “ick” factor. Thus far, resistance to transplants I have recommended has not come from patients or their families, who are desperate for relief. It has come from other health care workers, especially physicians, who seem to find the idea particularly distasteful. [emphasis added. This article supports the idea that doctors are a major source of resistance to this treatment.]

There is cost and time—while the “medicine” is inexpensive and readily available, current recommendations are that the stool donor be tested for a variety of infectious diseases at a cost of $1500-2000. There might be a week’s delay, while the donor is tested for hepatitis and other infections. . . . And now there is the new FDA requirement for an IND, which will be the coup de grace for this treatment. . . . INDs are incredibly burdensome, time-consuming, and expensive for an independent practitioner to obtain. They involve hours of paperwork (my office practice consisted of me and 1-1.5 secretaries; who has time?).

Given the awfulness and danger of this infection, I think it is fair to say that the home-treatment approach (via enema) is very easy. The author of this post, Dr. Judy Stone, complains about home treatment:

Then the sole data will come from some ambitious citizen science group [which is terrible because . . . ? — Seth], and acutely or seriously ill hospitalized patients, too ill to be treated at home, will be deprived of potentially life-saving treatment.

Dr. Stone is serious — deadly serious, you could say. According to this article, “more than 9% of C. diff-related hospitalizations end in death.” Fecal transplants are very effective. Stone predicts that patients will die because “hours of paperwork” are too much trouble, at least for her (“who has time?”). A more persuasive article would have explained why patients who need this treatment cannot be sent to doctors who decide that “hours of paperwork” are doable if that is what it takes to save lives.

Thanks to Paul Nash.

“The $2.7 Trillion Medical Bill”

Seth Roberts12 Comments

The New York Times has started a series called Paying Till It Hurts about high medical costs. The first installment is called “The $2.7 Trillion Medical Bill” and is about the high cost of common procedures, such as colonoscopies, in the United States compared to other countries. (Which I blogged about quite recently.) The most extreme example is that a certain (unspecified) amount of lipitor costs $124 in the United States and $6 in New Zealand. Other treatments that cost much more in the United States include hip replacements and MRI scans.

This series might be a turning point, leading to government regulation of what health care providers can charge, which is how other countries control health care costs. To read the huge number of comments (already > 1000) is to see the suffering caused by these prices. One comment: “An acne medication was over $550 for a small tube of ointment. The [prescribing] Dr. had no idea it was that expensive.”

The high prices are the tip of the iceberg of American health care dysfunction. Less obvious is the poor research that sustains them. Acne is an example. It surely has environmental causes (probably diet). If we knew what those are, you wouldn’t have to pay anything to cure acne.

Assorted Links

Seth RobertsLeave a comment

Thanks to Nick Gibb.

“Because It Costs More”: An Example of Medical Reasoning

Seth Roberts10 Comments

Melody McLaren, a friend of mine, lives in London. Her husband has Parkinson’s Disease and receives treatment through the National Health Service. His treatment has included “ deep brain stimulation” — implantation of an electrical device that stimulates subcortical brain areas. It is a standard treatment for Parkinson’s. It cost the National Health Service about £35,000.

She was surprised to discover that in the United States, the same procedure involved implantation of two batteries, one for each side of the brain. The device implanted in England has only one battery. It worked fine. My friend wondered why two batteries were used in the United States. She asked her husband’s neurologist, a French woman practicing in London. “Because it costs more,” she said. There was no other reason.

At the San Mateo Maker’s Faire a few weeks ago, I heard a talk by a doctor named Amy Baxter, who had developed a device for pediatricians that makes shots hurt less. (She had a child and noticed the problem.) She went to considerable trouble to develop a product that could be used by working doctors and presented the product several times to potential buyers. Again and again she was told It has to be disposable. Meaning one use per package. Nothing else will fit the supply chain. She did not say why she was told this, but the obvious reason was that disposable products are more profitable.

Steven Brill’s cover story in Time two months ago was about one way American health care takes advantage of sick people with little choice — hospitals, including nonprofits, charge patients for products and services far more than what they cost the hospital. This is another way.