How Med Harms

Most people are quite skeptical when I tell them the standard estimate is near zero for the marginal effect of medicine on health. While they grant that much of medicine may be useless, they point to particular cases where medicine was helpful, and can’t imagine much of it being harmful. But as I posted in November:

In at least 0.4% of hospital stays, a medical mistake “caused or contributed to a patient’s death.” (more)

Also, even standard diagnostic tests can be quite harmful:

CT scans of the heart cause one cancer for every 270 [=0.37%] 40-year-old women who undergo the test, researchers estimate. Yet in a study of CT scans investigating abdominal, hip or pelvic pain, only 9 percent of emergency-room doctors knew that the scans increased cancer risk. (more)

29,000 future cancers could be related to CT scans received in 2007, with the greatest number of cancers projected in the abdomen and pelvis. The cancer risk was greatest for young patients. (more)

A medical treatment really has to be quite clearly and strongly beneficial to overcome such harms. Just sort of maybe hoping that it might be useful, cause, heh, you haven’t heard anything specifically bad about it, just isn’t good enough.

Added 8p: Reasonable doubts have been raised about both the 1/270 and the 9% figures.

GD Star Rating
Tagged as:
Trackback URL:
  • TheQuickBrownFox

    This seems to be supported by the fact that we continue to require more treatment for disease as time goes on despite improved medical care. We survive and and are propped up by technology but continue to lose the underlying fundamental stuff of health.

    Health is not about healthcare. It’s about lifestyle (including diet). Unfortunately, mainstream dietary advice often worsens health.

    Colonoscopies may also cause more cancers than they prevent. Undergoing a colonoscopy means losing all of your gut flora. This can cause poor health for a long time afterwards if the bacteria are not replaced. Worsened digestion means losing out on micronutrients that are protective against a whole host of diseases. The importance of gut flora is recognised by researchers but seems to be largely ignored in medical practice.

    • Robert Koslover

      Regarding the colonoscopy issue, consider eating some yogurt afterward.
      It is my understanding that colon cancer is both deadly and far too common, yet the slow-growing polyps that typically precede it are often detectable early enough (via colonoscopy) to be safely and fully removed, thus preventing the cancer from occurring.

      • TheQuickBrownFox

        I’m aware of the suggestion to eat some foods with live bacteria cultures post-colonoscopy but I doubt it’s a comprehensive correction of what was there before.

        If children can increase lifelong risk of various infections and disorders from not getting proper gut bacteria exposure during childbirth and early life then it’s reasonable to hypothesise that a post-colonoscopy yoghurt binge does not undo all of the damage. Modern yoghurt has a few cultures artificially placed in it after the milk is pasteurised. It does not contain a full spectrum of flora and there are no doubt species we don’t even know about. And the gut flora issue is just one of the dangers of a colonoscopy.

        The problem is ultimately that the benefits of such procedures and researched with great vigour and evangelised, while the costs are not. Also, it feels more useful to do something as opposed to nothing.

      • Sarah Sofia Granborg

        Consider eating some yogurt BEFORE!!!
        Seriously… why do you think people get cancer? Usually because of how they live and eat!
        I agree, meds are usually more harmful than useful and even though prophylactic treatments, tests and scans do actually really pose different kinds of risks, so do other things, like flying. you usually get a whole lot more radiation that way.
        But the real problem here is that people rely on test, scans, meds etc. to get their illness detected and to get well. Unfortunately that’s not how it works. Even if there were no dangers connected with those it would still leave the patient in an effect position. To be causative, you just have to adapt a very healthy lifestyle and a very healthy diet. it’s actually a lot of fun, too and I’ve never seen anyone who lived that way getting colon cancer etc.!

    • AaronM

      The studies of colonoscopies DO include the costs, in that they have looked at total cancer rates for people who do and do not get the procedures. They have observed that people who recieve colonoscopies have a lower overall risk of dying from colon cancer. So even positing that the loss of GI flora has some contribution to increasing risk (of which I am very skeptical), the total cost/benefit ratio is still in favor of benefit – ie, lower cancer risk.

      Whether or not these benefits are worth the FINANCIAL cost of colonoscopies on the medical system is another questions altogether…

      Also, people have been getting cancer for much longer than Americans have been fat and lazy. Most are caused by genetic predisposition, viruses, alcohol, or smoking, or are idiopathic (ie, bad luck). Models that I have read that attempt to break down the various contributing factors for cancer predisposition have tended to find only small contributions (<10%) of diet and lifestyle (besides smoking and alcohol).

      • Treenat

        “The studies of colonoscopies DO include the costs, in that they have looked at total cancer rates for people who do and do not get the procedures. They have observed that people who recieve colonoscopies have a lower overall risk of dying from colon cancer.”

        The only results from comparing screened versus unscreened patients show MARGINAL (or even ZERO) decreased risk of colorectal cancer. What’s more, the SCREENED groups had a higher ALL-CAUSE mortality rate. Read that last part again.

        Colonoscopies are a sham. Period.

  • Bill

    They are just confusing marginal and infra-marginal units. The average marginal out-of-pocket cost of the marginal dollar spent is close to zero. The only way the benefit of the average marginal dollar spent can be higher than zero is if something weird is going on.

    Also, they are worried that your next rhetorical move is to deny the benefits of technological advance in medicine, and they want to get out ahead of that.

  • michael vassar

    I’d like the details behind that claim about CT scans. My impression is that risk estimates for radiation are very poor. Do we question that tobacco causes cancer but give this a free ride?

  • IVV

    I greatly prefer MRI for this reason.

  • Ray Gardner

    We survive and and are propped up by technology but continue to lose the underlying fundamental stuff of health.

    Well said.

    On colonoscopies; gut flora is relatively easy to produce so if one was to undergo a colonoscopy and not eat accordingly afterwards, they are probably on the road to poor health anyway. Likewise, one is already eating correctly along these lines, a colonoscopy would – I believe – be harmless.

    More to the post itself, I was fortunate to have a naturally skeptically minded RN for a mother, and have always been wary of taking medicine for the sake of taking medicine.

  • Dániel Varga

    “CT scans of the heart cause one cancer for every 270 [=0.37%] 40-year-old women who undergo the test, researchers estimate.”

    I found this very hard to believe, so I googled it. This was estimated from a ridiculously small 1000-person sample. But another study based on 10 million Medicare records estimated 0.02 to 0.04 percent: cancer-risk-ct-scans-overestimated-study They talk about the whole population, not just 40-year-old women with CT scans of the heart, so the findings are not in direct contradiction with each other. But I think the 0.04 number shows a different picture overall, and it has a much higher chance of being correct.

  • Epidemiologist

    As a physician, I can tell you with certainty that one hundred percent of doctors know that CT scans increase cancer risk.

    If you can prove that only 9% of of us are aware of this fact, then the government should revoke the license of 91% of physicians.

    I also very much doubt that the effect is as large as indicated in this post, but I don’t have time to look up the article. If this estimate is true, then it would only be applicable to high-radiation-dose scans

    • Eliezer Yudkowsky

      If you don’t know how much something increases cancer, you don’t know whether it increases cancer. Seriously, you might as well not know anything as have a piece of nonquantitative knowledge like that; it is the pure illusion of expertise.

      • Tyrrell McAllister

        Seriously, you might as well not know anything as have a piece of nonquantitative knowledge like that;

        Epidemiologist claimed quantitative knowledge—in particular, that the effect was very likely not as large as indicated in the post.

      • Epidemiologist

        I agree that I misused the word “know”.

        Doctors already need to have a very high amount of information packed into their brains. It is not plausible to require that we should have exact quantitative knowledge about the size of an effect. For medical practice, it is sufficient for to know that “Current clinical guidelines are to avoid CT scans unless necessary, because experts suspect it causes cancer”. This is what I should have suggested that 100% of doctors “know”. I don`t think this is an illusion of expertise, but a good heuristic for everyday practice.

      • ThePenileFamily

        The statement below is false.

        If you don’t know how much something increases, then you don’t know whether it increases.

  • Jess Riedel

    Most people are quite skeptical when I tell them the standard estimate is near zero for the marginal effect of medicine on health

    Are you sure that it’s not just the fact that people are confused about what “marginal” means? It’s perfectly consistent that there are many cases were medicine is helpful, very few cases where it is harmful, and for the marginal effect to be zero even though the total effect is massively positive. This is exactly what you’d expect if people value health at infinity when deciding on treatment for existing conditions, which, in fact, is exactly what they often claim to do.

    Yes, this means people are inconsistent over time since there are much cheaper ways to improve health preventatively (not driving, eating healthy, etc.). But that just tells us something about people’s irrational preference for treatment over prevention, not anything profound about modern medicine.

    In other words: when someone goes to the hospital, their family tells the doctor “do everything you can”, i.e. keeping spending money until it doesn’t help anymore. And that’s exactly what they get.

    This isn’t new. The only thing that’s changed is that in past there simply weren’t many options for spending money on healthcare. The marginal improvement was still zero (implying, again, that people irrationally over-valued treatment of healthcare), but nobody cared because blood-letting was pretty cheap. Now that we can spend $100,000 on a cancer treatment, it’s starting to become a problem.

  • Yvain

    There’s no citation on that “only 9% of emergency room doctors knew CT scans increased cancer risk” statement, and I give 99% odds that this is a gross misinterpretation of something or other. I can’t find any studies on doctors, probably because doctors knowing about CT radiation raising cancer risk is so obvious that no one would bother studying it, but two studies on the general population say 30% and 45ish % understand it, and a study on medical students show 70% of them understand it (I suspect the medical school studied is incompetent; I would estimate the knowledge level among my classmates at >90%).

    So unless you’re willing to say doctors are four times as ignorant as patients and eight times as ignorant as med students, something’s wrong with that study.

    It’s not your mistake, because that’s what it says in that Post article, but I think if you knew anything about medicine the mistake would be pretty obvious. And when you write about medicine without understanding it, this happens.

    Moving through the article: CT coronary angiographies have much higher cancer risk than an ordinary CT procedure. Doctors know this and are especially careful when performing them. That having been said, if your heart’s going to stop soon, that is something you will want your doctor to know, or else your death rate’s going to be a whole lot higher than 1/250.

    CT scans of the head have about a 1/10000 cancer rate. If your doctor has a clinical suspicion of (say) 10% that you have (let’s say) head cancer, his choice is to operate on you, leave you alone, or get a CT. Leaving you alone means there’s a 10% chance he just ignored your cancer and it will spread and kill you. Operating on you means there’s a 90% chance he’s going to cut open your head for no reason – and I imagine the death rate from major head surgery is more than 1/10000. Getting a CT is by far the best of these non-perfect options. I have just finished a rotation under one of Ireland’s top head-and-neck surgeons, and he could recite off the top of his head every single statistic necessary to make this decision (ie “He’s got a mass in his neck, which has X% chance of being a cancer, which would have an X% chance of killing him within 5 years, so I suppose we better operate”).

    Overuse of CT scans is a problem, but this is mostly because when given a choice between giving the CT – and risking cancer 20 years down the line in a way that can never be linked to the test – or not giving the CT, and risking that the patient dies tomorrow because the doctor missed something he could have detected on CT – many doctors will choose the CT for emotional and legal reasons whether or not the accurate multiplication of risks would recommend it. There are also random scumbags who peddle “screening” CT tests outside the normal medical system, but this is a known problem and everyone hates them.

    When you say “Just sort of maybe hoping that it might be useful, cause, heh, you haven’t heard anything specifically bad about it, just isn’t good enough” I assume you have read the 29,385 studies on CT scans on PubMed, including the thousands of specific ones plus the broader ones showing that CT scans in general raise life expectancy in areas and diseases where they are employed, and decided none of them meet your exacting standards for evidence?

    Radiologists have read a bunch of them, the doctors prescribing CT scans have read others, and mostly everyone involved in the system has read more than an article in the Washington Post that doesn’t cite sources and screws up its statistics. Please stop doing this. Please.

    • Robin Hanson

      Seems you are suggesting I stop posting on med media reports unless I’ve read and judged the original research myself? I accept that you have reasonable doubts about the conclusions, and don’t like that they didn’t give a citation. I surely didn’t mean to claim that all radiologists know about radiation is only one media article; not sure where you get that from.

    • Douglas Knight

      I am much more skeptical of your ability to off-handedly assess your classmates than of the study you link. Sure, call the medical school in the study “incompetent” but that is quite compatible with it being of average competence. Similarly, invoking the knowledge of the top surgeon does not distinguish between 9% and 90% (though it suggests your medical school is not typical). Also, studies show that typical doctors cannot turn statistical knowledge into false positive rates.

  • Ray

    One indirect “harm” of medicine is that people would rather take a pill than take care of themselves.

    So blood pressure meds, cholesterol meds, etc take the place of actually taking care of one self. Regardless of how effective the meds really are since people will routinely try one drug after another (and be on several meds for several conditions simultaneously).

  • TGGP

    Those percentages sounded pretty low to me.

  • Silas Barta

    Most people are quite skeptical when I tell them the standard estimate is near zero for the marginal effect of medicine on health.

    Stupid question: isn’t that how it’s _supposed_ to work for *anything*? IOW, it’s optimal to spend resources on something *until* the marginal net benefit is zero, right? Where MC = MB?

    • Tyrrell McAllister

      I take the claim to be this: Most people think that it would be good if we had the resources to spend more on medicine while not spending less on anything else. Robin is saying, No, even ignoring opportunity costs, more resources spent on medicine would do us direct harm.

  • Pingback: Linkblogging For 09/02/11 « Sci-Ence! Justice Leak!

  • AaronM

    From someone who will have a medical degree in a few months, my best guess is that the “zero marginal benefit from increased medical care” is correct, or close to it.

    I would even go one step further: after removing antibiotics, vaccines, prenatal care, and care for traumatic injuries from the calculus, I would expect that the net benefit from all other care is close to zero. Or, at least, that it pales in comparison.

  • Jordan

    My uncle is a doctor; my father is a doctor; my mother was a doctor, now retired, who battled with major health issues; I have battled with major health issues myself.

    Based on a (relatively speaking) very large body of anecdotal evidence based on personal experiences and conversations with my family, my conclusion is that the basic level of health education among the general populace is low enough that people don’t actually know when they should seek professional medical care; this means that cultural traditions are the primary factor in overall medical spending. Medicine has no real incentive to turn away people who don’t really have a problem – or at any rate, a problem that medicine is good at – so if more treatment is kind of a wash, well… the treatment-provider now has more money. It would take EXTREMELY STRONG evidence – not just PRETTY GOOD evidence – that treatment was a complete waste of time in order to overcome this monetary incentive to provide treatment. And while it seems to be getting better, historically over the last few decades there has been a very strong incentive to run lots and lots of tests or risk a malpractice lawsuit (in short: “malpractice is not about medicine”). On the other hand, the medicine applied to myself and my mother (in most fields) has been remarkably effective; I attribute this to the fact that we are familiar enough with basic medical principles to judge when a problem is worth investigating with the help a medical professional.

    If you will entertain the above notion for a while, this explains why medicine can seem so darn marginally useless despite the massive amounts of money and research pumped into it: bad decisions are drowning out the good ones. This would indeed probably be improved by cutting medical spending nation-wide in half by forcing people to be more economical about medicine, but if you want an even bigger improvement in good-to-bad spending ratio – if you want to really maximize bad spending cuts while minimizing effective spending cuts – then a more productive approach may be to increase general education in medicine.

    • Jordan

      To clarify: I don’t mean to say the goal is maximizing bad cuts to spending and minimizing good cuts to spending; I mean maximizing cuts to bad spending and minimizing cuts to good spending. This should be obvious from context, but I apologize for the ambiguous grammar.

  • Pingback: How Med Harms « Daniel Smith

  • Pingback: Stares at the World » Recap: The Collapse is Coming