The Bleeding Edge of Innovation

What type of hospital should you use if sick?  Well none if possible; you suffer about a 1 in 400 chance of dying from a medical error during each hospital stay.  But what if your problem is bad enough to be worth this risk?

Studies of how often patients die during or after hospital stays, correcting for patient sickness, give many conflicting results.   There may be benefits from hospitals with more nurses per patient, and from private non-profit teaching hospitals.  But it seems there is no benefit from hospitals where more doctors are board-certified specialists.   The best advice I can offer comes from these conflicting observations:

  • Big hospitals have about the same overall death rate as small ones.
  • Big hospitals tend to do more and newer procedures, relative to small ones.
  • For any given procedure, hospitals that do it more have lower death rates.

My resolution:  Big hospitals are better at any given procedure/treatment, but they do more new ones, which tend to be bad.  So avoid the bleeding edge of medical innovation: go to a big hospital, but only accept procedures small hospitals also do.

Added:  Most new ideas are discarded; you want the ones that last. 

GD Star Rating
loading...
Tagged as:
Trackback URL:
  • eddie

    Arnold Kling often talks about the weak relationship between the amount of health care services provided and outcomes; he seems convinced that much of what doctors and hospitals do is simply wasted effort. Robin’s argument here appears to take that idea to an absurd conclusion. “Avoid the bleeding edge of medical innovation [..] only accept procedures that small hospitals do.” If the alternative to the offered procedure is to continue to suffer from a debilitating or life-threatening condition, why should the chance of dying from a medical error deter you from the procedure?

    If the new procedure is merely an incremental improvement over an older procedure, there might be a reasonable case that the increased risk isn’t worth the benefits. But consider the early adopters of hip replacement or laparoscopic surgery. Those procedures, even when new, were much much better than the alternatives.

    Also, I note that Robin’s conclusion (innovative procedures are higher risk) is drawn indirectly from three other observations. Is there no data that shows the increased risk from innovative procedures directly? Surely some new procedures will be riskier than other equally new ones; the decision to have or not have the procedure should be made on a case-by-case basis, not by a broad rule such as Robin proposes here.

  • dks

    Could you provide some documentation for the claim that “you suffer about a 1 in 200 chance of dying from a medical error during each hospital stay”? The report that is linked offers no such evidence–it suggests that you have maybe a 1 in 50 chance of being subject to medication error (“adverse drug event”), but few of these are fatal. I also looked at the full text of the report it summarizes, and there is no effort to quantify mortality. Is the claim based on a source different from the one linked?

  • http://profile.typekey.com/robinhanson/ Robin Hanson

    Dks, I had the wrong link about; that is fixed now.

    Eddie, most innovations are failures and discarded. With hindsight you can say it would have been good to get the successes early, but we don’t have the foresight to know which new ideas will succeed.

  • dks

    Thanks for the link! Looks like 1 in 400, using the highest point estimate (1 in 800 with the lowest).

  • Nick Tarleton

    “Well none if possible; you suffer about a 1 in 200 chance of dying from a medical error during each hospital stay. But what if your problem is bad enough to be worth this risk?”

    But is this error evenly distributed over all patients? I would think that the worse their problem, the more and more invasive procedures they would perform, and the greater the chance of a fatal error. I doubt 1 in 200 (or even 1 in 800) people who go into the hospital with a broken finger come out in a body bag.

  • http://profile.typekey.com/robinhanson/ Robin Hanson

    DKS, OK, I changed my post to say one in 400.

  • Telnar

    I’m very suspicious of this claim unless the relevant studies have controlled for patient characteristics.

    One possible explanation for the higher death rate from new procedures is that at least some of the new procedures are used for situations which were previously considered incurable. That could lead to large hospitals getting a sicker population on average than small hospitals (since most with the time and resources to be able to choose will prefer to go to a hospital which says that it is willing to try to cure him).

  • http://profile.typekey.com/robinhanson/ Robin Hanson

    Telnar, would you be equally attentive to method detail if the studies came to the opposite conclusions?

  • Telnar

    Robin, probably not since I will admit to a preconception that medicine as a whole makes progress over the medium and long runs (even though there are specific examples where it’s been proven that newer treatments are no more effective than older ones).

    Still, while that bias has the potential to distort the way I handle evidence, not all of this preconception is bias. Medical technology clearly does improve over long periods (life expectancies at birth have been rising by on the order of 1 year for every 5 that pass, of late). The relevant question is how that manifests in the shorter run since (without data that I don’t have) I can’t rule out the possibility that treatments are sorted by quality well after they are first launched. Still, if that scenario were common, I’d expect to have heard more about inferior treatments which were weeded out — if only because of the volume of lawsuits they would generate.

    So, how should one approach thinking about bias when it seems to point in the same direction as the available evidence, but that evidence is weak?

  • Stuart Armstrong

    Telnar, would you be equally attentive to method detail if the studies came to the opposite conclusions?

    I know questionning method detail selectively is a bias, and that Telnar has admitted bias, but I do feel it’s the natural question to ask of the study.

    The post starts What type of hospital should you use if sick? and then cites a study that says that 1 in 400 people will die in the hospital. Immediately I thought “what are the odds of me dying personally?” – a variant of Telnar’s query, and a relevant one to the rest of the post.

    Another question worth mentioning: if there is some initial medical error (or just an uncesseful procedure), then if it doesn’t kill you, it will leave you even more sick, staying in hospital for longer, and at the risk of dying (by a further medical error or just by not being cured). But that further death will be classified under another procedure. So can we say:
    If a procdure is kill-or-cure, without any middle-ground, we should consider it relatively safer than those with a middle-ground.

  • Telnar

    Robin,

    Also, preconceptions are often derived from indirectly related information which speaks inconclusively to the situation at hand. In a weak sense they may represent biases if we have a tendency to overrate the strength of indirect relationships. Ignoring that information isn’t right either, though, so the balancing act can be quite tricky.

  • http://profile.typekey.com/robinhanson/ Robin Hanson

    Telnar, you vastly overestimate the contribution of medicine to long term health improvements, and the scope of the grounds for malpractice suits.

  • http://profile.typekey.com/halfinney/ Hal Finney

    I am very happy to see Robin blogging about medical issues. This seems to be one of the strongest and most resistant areas of bias. Say bad things about advertisers or politicians and everyone chimes in, but say something bad about medicine and suddenly people are up in arms.

    We all need to recognize that we have a strong emotional investment in the idea that doctors have nearly miraculous healing powers. It is extremely comforting to believe that if we get sick, modern medicine can get us well. The role of medicine in society and our social and political views on the topic are subject to bias as well.

    Keep in mind that people have had these kinds of attitudes towards their shamen and healers throughout history, even in situations where we would recognize that the actual tools available to healers were extremely limited. These attitudes towards medicine evolved for a purpose but we need to look past our prejudices in order to clearly see the bias involved. I recommend approaching these topics with an uncluttered mind and trying to forget the things that “everyone knows” about doctors and medicine. Many of these commonly accepted beliefs are false or at least unjustified.

  • Stuart Armstrong

    Failure teaches more lessons that success: if you were altruistic, you might go in for a (slightly) higher risk procedure, in the knowledge that if it did fail, the world would benefit a bit from your death.

    (But no one ever seems to follow that sort of reasoning – people seemed wired to be altruistic towards the sick, but never altruistic when sick).

  • Telnar

    Robin, I’d be interested in understanding where you got the perception that most of the recent improvements in life expectancy in wealthy countries are unrelated to medicine.

    There were huge advances in public health in the 19th and early 20th centuries in the West (e.g. better handling of sewage, and reductions in food contamination). In addition, improvements in wealth have eliminated types of risks from everyday life. Again, though, the most visible differences were in the past (indoor plumbing and air conditioning are no longer considered luxury goods in the US). Do you believe that those kind of advances are still significant? If not, then what other factors aside from medicine do you believe are responsible for the fact that the long term trend of increasing life expectancies is continuing unabated as semi-anecdotal information suggests that medicine is improving (e.g. prognoses for patients with heart disease are much improved over the last 20-30 years).

    These days, the available incremental public health improvements (e.g. food irradiation) are often rejected by the public which has preferred alternatives which are less likely to make a difference like organic food. Perhaps I’m missing changes, but it’s much harder to see major positive developments in public health in the last 20 years. Increased regulation of the environment, yes, but many of those regulations don’t have a clear and significant connection to health.

    For good measure, consumer behavior is probably getting worse (with greater obesity more than offsetting lesser smoking).

  • http://profile.typekey.com/robinhanson/ Robin Hanson

    Telnar, medical skepticism will be a common theme here, which I will post on often. In the spirit of modularity, I will not try to say it all in one post.

  • michael vassar

    Telnar: Smoking has declined a lot. Driving safety has improved. Murder fallen.
    Still, good points.