[Link] An argument for Low-hanging fruit in Medicine
Those of us who have found the arguments for stagnation in our near future by Peter Thiel and Tyler Cowen pretty convincing, usually look only to the information and computer industries as something that is and perhaps even can keep us afloat. On the excellent West Hunters blog (which he shares with Henry Harpending) Gregory Cochran speculates that there might be room for progress in a seemingly unlikely field.
Low-hanging fruit
In The Great Stagnation, Tyler Cowen discusses a real problem – a slowdown in technical innovation, with slow economic growth as a consequence.. I think his perspective is limited, since he doesn’t know much about the inward nature of innovation. He is kind enough to make absolutely clear how little he knows by mentioning Tang and Teflon as spinoffs of the space program, which is of course wrong. It is unfair to emphasize this too strongly, since hardly anybody in public life knows jack shit about technology and invention. Try to think of a pundit with a patent.
Anyhow, it strikes me that a certain amount of knowledge may lead to useful insights. In particular, it may help us find low-hanging-fruit, technical innovations that are tasty and relatively easy – the sort of thing that seems obvious after someone thinks of it.
If we look at cases where an innovation or discovery was possible – even easy – for a long time before it was actually developed, we might be able to find patterns that would help us detect the low-hanging fruit dangling right in front of us today.
For now, one example. We know that gastric and duodenal ulcer, and most cases of stomach cancer, are caused by an infectious organism, helicobacter pylori. It apparently causes amnesia as well. This organism was first seen in 1875 – nobody paid any attention.
Letulle showed that it induced gastritis in guinea pigs, 1888. Walery Jaworski rediscovered it in 1889, and suspected that it might cause gastric disease. Nobody paid any attention. Krienitz associated it with gastric cancer in 1906. Who cares?
Around 1940, some American researchers rediscovered it, found it more common in ulcerated stomachs, and published their results. Some of them thought that this might be the cause of ulcers – but Palmer, a famous pathologist, couldn’t find it when he looked in the early 50s, so it officially disappeared again. He had used the wrong stain. John Lykoudis, a Greek country doctor noticed that a heavy dose of antibiotics coincided with his ulcer’s disappearance, and started treating patients with antibiotics – successfully. He tried to interest pharmaceutical companies – wrote to Geigy, Hoechst, Bayer, etc. No joy. JAMA rejected his article. The local medical society referred him for disciplinary action and fined him
The Chinese noticed that antibiotics could cure ulcers in the early 70s, but they were Commies, so it didn’t count.
Think about it: peptic and duodenal ulcer were fairly common, and so were effective antibiotics, starting in the mid-40s. . Every internist in the world – every surgeon – every GP was accidentally curing ulcers – not just one or twice, but again and again. For decades. Almost none of them noticed it, even though it was happening over and over, right in front of their eyes. Those who did notice were ignored until the mid-80s, when Robin Warren and Barry Marshall finally made the discovery stick. Even then, it took something like 10 years for antibiotic treatment of ulcers to become common, even though it was cheap and effective. Or perhaps because it was cheap and effective.
This illustrates an important point: doctors are lousy scientists, lousy researchers. They’re memorizers, not puzzle solvers. Considering that Western medicine was an ineffective pseudoscience – actually, closer to a malignant pseudoscience – for its first two thousand years, we shouldn’t be surprised. Since we’re looking for low-hanging fruit, this is good news. It means that the great discoveries in medicine are probably not mined out. From our point of view, past incompetence predicts future progress. The worse, the better!
I think Greg is underestimating the slight problems of massive over-regulation and guild-like rent seeking that limits medical research and providing medical advice quite severely. He does however make a compelling case for there to still be low hanging fruit there which with a more scientific and rational approach could easily be plucked. I also can’t help but wonder if investigating older, supposedly disproved, treatments and theories together with novel research might bring up a few interesting things.
Many on LessWrong share Greg’s estimation of the incompetence of the medical establishment, but how many share his optimism that our lack of recent progress isn’t just the result of dealing with a really difficult problem set? It may be hard to tell if he is right.
This is unfair. Modern attempts to eradicate h pylori use “triple therapy” of two different antibiotics plus a PPI, taken simultaneously, all for a unusually long period of time (one to two weeks). If you just give a patient a random antibiotic for some other disease, that’s not going to produce astounding results: you need the PPI to alter the transport of the antibiotic and prevent it from getting immediately broken down in the acidic environment of the stomach, and you need both antibiotics in case the bacteria develop resistance to one. Although some doctors probably got lucky and eliminated h pylori by chance, it’s not like every time anyone took an antibiotic it was curing their stomach ulcers and no one noticed.
Why else didn’t people catch onto h pylori’s role in stomach ulcers quicker? Well, at least 80% of people with h pylori don’t have ulcers or any symptoms whatsoever, and 20% of people with ulcers don’t have h pylori, so that’s going to confuse people. Second, h pylori is hard to stain and very hard to culture, so all you have are these Bigfoot-esque rumors of “I saw this bacterium in the stomach...I think...no, I don’t have any to show you.” A bunch of studies looked for bacteria in the stomach and find none, because of previously mentioned staining and culturing problems. And whenever people tried treating ulcers with antibiotics—and they did do the studies—they get equivocal results because they weren’t using the exact right drug combo to hit h pylori and keep it gone. The first person to successfully culture h pylori was the guy who won the Nobel Prize for discovering it.
Now, just from reading this article, you might believe that doctors are lax about looking at possible infective causes for chronic disease. Let’s look at what PubMed has to say about possible infective causes of schizophrenia. There are 195 studies on “schizophrenia + influenza”, 100 on “schizophrenia + herpesvirus”, 84 on “schizophrenia + toxoplasma”, 24 on “schizophrenia + cytomegalovirus”, 17 on “schizophrenia + varicella”, 10 on “schizophrenia + Lyme disease”, 4 on “schizophrenia + neurocysticercosis”, and one on “schizophrenia + trichinosis” (note that some of these studies are counted multiple times). This is not too atypical: after (among other things) the whole h pylori debacle people realized this was low hanging fruit and have been trying to pick it for the past thirty years.
I think that’s generally a useful principle: if you know something is low-hanging fruit, then unless you’re special so does everyone else, which means it’s not low-hanging fruit anymore. And that’s probably why h pylori seems so clear to Cochrane with thirty years of hindsight, while the people who figured it out at the time won the Nobel Prize, which generally isn’t given for pointing out the obvious.
Oh, be nice now.
Doctors seem more engineers than scientists to me. For some reason they seem to get offended by such assessment.
I’d be offended if I were an engineer.
I don’t have a bad opinion of engineers. Doctors seem to for some reason.
(I have a bad opinion of doctors, relative to degree of education.)
Was this discovery made as soon as was theoretically possible, or are we rationalizing reasons to justify why it was in fact fairly slow? Or to put it another way, it seems like it was fairly slow, could we do better if the process of Science was different?
Like for instance, balancing the cost of putting researchers on the case, trying different combinations of antibiotics (which as experiments go shouldn’t be too much of a health risk, neither too much of a material cost) on people with peptic ulcer, against the bayesian evidence for and against the bacterial hypothesis?
Even if the evidence was low, the cost to implement such a research project probably is lower than what’s invested in bogus medicine like homeopathics or the like, which has more evidence against itself (at least a bacteria is more likely a causal phenomenon for something than magical water memory).
The efficient research hypothesis?
...
Actually being an accomplished researcher in the field of medicine, having done more science in the field than vast majority doctors in the world, he has good reason to say this.
It is Cochran, without an e. I agree with those who enjoy West Hunter. It is a new blog, and among the best—highly informative.
An obvious low-hanging fruit, hiding in plain sight is medical errors. Due to poor recognition and under-reporting, what we notice and read about is only the tail of the bell curve of severity.
Interesting timing; I spent this morning in a risk management simulation: a mock hospital, with actors playing patients, surrounded by a bunch of doctors criticizing me when anything I did looked like it left potential open for a medical error. This was part of about fifty hours or so of training specifically in avoiding medical errors I’ve gone through in the past four years. Also, my record for “number of different doctors who have mentioned Atul Gawande to me in a single day” currently stands at three, and I’m expecting it to be broken before finishing school in May. It’s safe to say any trend that’s reached medical schools in southwestern Ireland is well established in the States, Britain, and the rest of the world.
There are dozens of journals, textbooks, and three-letter-acronym organizations dedicated solely to healthcare risk management, and most good hospitals will have their own on-site risk management team. Newer paper hospital charts have been specifically designed around risk management (for example, on the charts I used today, there are two spaces to sign for giving any drug: the signature of the person who gave it, and that of the person they checked it with to make sure it was safe), and those hospitals that have electronic charts have their built-in measures to prevent errors (for example, they’ll automatically pop up an alert if you prescribe an unusual dose of any drug, or two drugs that interact with each other, or a drug a patient is known to be allergic to; sometimes only certain authoritative staff at the hospital have the password to override these alerts.)
Now, none of this means that risk management is any good. You’re still making a bunch of chronically sleep-deprived and very busy people play a game of Chinese whispers with complicated biochemical data. And I sort of worry that some hospital manager has done the math and decided that the amount of money saved in malpractice suits is less than what it would cost to hire lots of extra doctors so each one is individually less busy and more awake.
But I do think that once there are thousands of people and tens of millions of dollars in a field, you don’t get to call it “hiding in plain sight” or “low hanging fruit” anymore.
This is at least a subset of the kind of software my dad works on. I remember him talking once about how someone was trying to get him to force doctors to dismiss every alert individually instead of just being able to select-all-dismiss, and my dad said that all that would do is waste time because the doctor would wind up mechanically dismissing 50 alerts without looking at them instead of doing it all at once, and trying to force the doctor to be conscientious with software was not the right way to go about making doctors be appropriately meticulous.
(Disclaimer: I probably misremember some substantial part of this.)
Yeah, the problem with introducing any safety procedure is that if you have to do it a thousand times a day, it eventually becomes mentally automated. It’s like clicking “I agree” to EULAs, or “let this program access to the Internet” on one of those versions of Windows that would incessantly ask you whether you wanted to let a program access the Internet (never had one, but heard awful things from people who did).
I’ve never used a program like that, so I don’t know whether the bar for alerts is set high enough that most alerts will be real errors, or whether it gets into so many nitpicky things (you’re using an antihypertensive with another antihypertensive! What if that causes hypotension?!) that you eventually develop a reflex of clicking through them. I’d hope the former.
If you want more detail about how much care should go into designing a safety system, try The Checklist Manifesto.
Do you know if every hospital has implemented the Surgical Safety Checklist and mandates and ensures its use?
This could be true, but I don’t think it can be taken for granted.
How about the Heimlich method? There were millions of dollars and thousands of doctors in the field when he invented what he invented. Not a very high hanging fruit.
Probably not a good example
Another relatively low hanging fruit. What is really the best response for chocking? Heimlich method OR “back slap” or something else?
Can you think of a way to test this that would get past an ethics committee, though?
I would go for statistics. How many have died/survived after this or another approach.
I agree with Greg’s estimate of medical practitioners- they are not scientists, by and large. It isn’t what they’re trained for.
And while it is certainly true that some medical problems are really difficult, I expect there is also a lot of low-hanging fruit. Part of the problem is likely that medical research is to a large extent governed not by “What is easy to cure?” but by “What is profitable to treat?” or “What would be prestigious to research?”
Do any countries’ national health services research and produce their own drugs and treatments? I would like to know how their goals and results differ from other researchers.
These questions are not unrelated.
A contrasting perspective: http://www.wired.com/magazine/2011/12/ff_causation/all/1
I’m thinking that determining whether medicine has low-hanging fruit is itself not low-hanging fruit.
The previous discussion of that article can be found here.
The most obvious cause of the mistake about ulcers was having a simple plausible theory that bacteria couldn’t survive in stomach acid. I wonder how the Chinese got it right. Did the theory about bacteria never happen to take hold there? Do stomach ulcers look like other infections?
More generally, I’m inclined to think that being a patient is a way of being low status. The basic need for sleep is made subordinate to hospital routine.
There may be cultural variation on that one—most of what I hear about medicine is about American medicine.
Every doctor should be required to study The Art of War before being allowed to practice. Practicing medicine is not like balancing an equation. Medicine is not a game of perfect information, nor are its rules unchanging. Indeed, each person’s health battles take place upon an idiosyncratic battlefield with various assortments of forces. The war of evolution, which has been raging for millennia, has developed in Homo Sapien what is likely one of the most highly adapted immune systems in the history of the planet. It is not surprising that doctors fail so often when they survey the battlefield and can think only to higher suspect mercenaries to fight proxy battles (drugs on secondary symptoms)...
We should think of our brain as a GrandMaster General, the spine her chain of command, every specialized organ a legion brimming with age-old veteran forces. Should we think the whole army, which have never yet succumb to defeat in millenia of evolution, could be defeated by average maladies? Should we assume that the General of millions of victories makes simple mistakes like misallocating resources (like making too much cholesterol)?
And when we’ve given the majestic human body its due respect, is it reasonable to expect people with 4 years of training, who are bombarded by biased solicitations and have to side-step personal biases, who primarily wield simple synthetic compounds and inelegant machinery… to have mastery over it?
The key thing to remember is: Correlation does not equal causation.