Race and medicine

Share on FacebookShare on Google+Email this to someoneTweet about this on Twitter

Apropos of a previous post on race, PLoS Medicine has just published two (opinion) articles on the use of racial categories in medicine. There’s only a cursory treatment of genetics (and the treatment that’s there is pretty bad), but it’s sometimes useful to see another take on the issue. The message I get is that, well, doctors aren’t trained in genetics, so any “race-based” medicine (which is necessarily based on probabilites) is likely to become a sort of “black = medicine X, white = medicine Y” dogma.

Labels: ,

17 Comments

  1. well…medical doctors are some of the scariest fuck-ups when you look at human inability to engage in bayesian conditional probabilities….

  2. Some doctors are at least rudimentally trained in genetics. For instance, it is increasingly common for pathology residents to rotate through molecular diagnostics and cytogenetics labs for at least a few months. There are also post-residency fellowships available in clinical genetics and molecular diagnostics.

  3. Sorry, I meant “rudimentarilly.”

  4. of course molecular genetics and population genetics have only one thing in common: they both study genes ;-)

  5. doctors aren’t trained in genetics 
     
    It’s not genetics they need, but statistics. And not just doctors, everybody – certainly more important than calculus!

  6.  
    It’s not genetics they need, but statistics. And not just doctors, everybody – certainly more important than calculus!
     
     
    your IP tells me you’re not steve sailer…are you on his pay roll???

  7. Concur w/ Razib & David Boxenhorn — the fact that until recently medical students weren’t even taught how to use Bayes’ rule speaks volumes about the profession and scares the shit out of me. Their prejudice against statistical prediction rules that have been *proven* to be more accurate than human specialists is insane: 
     
    http://www.skeptic.com/eskeptic/07-08-22.html#feature 
     
    One study famously showed that a successful predictive instrument for acute ischemic heart disease (which reduced the false positive rate from 71% to 0) was, after its use in randomized trials, all but discarded by doctors (only 2.8% of the sample continued to use it). It is no secret many doctors despise evidence-based medicine. It is impersonal “cookbook medicine.” It is “dehumanizing,” treating people like statistics. Patients do not like it either. They think less of doctors’ abilities who rely on such aids. 
     
    The problem is that it is usually in patients’ best interest to be treated like a “statistic.” Doctors cannot outperform mechanical diagnoses because their own diagnoses are inconsistent. An algorithm guarantees the same input results in the same output, and whether one likes this or not, this maximizes accuracy. If the exact same information results in variable and individual output, error will increase. However, the psychological baggage associated with the use of statistics in medicine (doctors’ pride and patients’ insistence on “certainty”) makes this a difficult issue to overcome. 
     
    The statistics vs. clinical intuition debate has ensued for decades in psychology. Where one sides in the debate is largely determined by what one makes of a single phrase: “Group statistics don’t apply to individuals.” This claim, widely believed, ignores many of the most basic concepts of probability and statistics, such as error. Yes, individuals possess unique qualities, but they also share many features that allow for predictive power. If 95% of a sample with quality X has quality Y, insisting that someone with quality X may not have Y because “statistics don’t apply to individuals” will only decrease accuracy. Insistence on certainty decreases accuracy.

  8. yeah, I love cytogenetics and all, but I think everyone is right–it’s not genetics, but statistics that’s missing.

  9. I also concur with Razib and Matt. In my limited experience (as a Ph.D. biostatistician / epidemiologist) doctors avoid statistics and go for what they think is a “sure thing”. They also have difficulty in dealing with odds of success, blowing it off with comments like “I’m not a bookie!” I believe part of the reasoning is the extreme focus on the current case, rather than thinking about trends and groups.

  10. btw, i didn’t mean that medical docs are particular bad at bayesian probability. just that their mistakes are pretty consequential.

  11. Excuse me, but where is the statistical proof of such statements as “(Doctors)have difficulty in dealing with odds of success, blowing it off with comments like “I’m not a bookie!”? It seems that the commentors are using anecdotes to make a point about how physicians don’t use statistics in their practices. 
     
    I have a bias, being a physician, but we do deal with individuals and the application of evidenced based medicine is helpful at best in the light of the problem of incomplete evidence that we have to deal with all the time. 
     
    Take the example of dosage of medications. The FDA has “approved doses” that only reflect a mandated and structured set of experiments designed to provide information on safety and efficacy. They do not test what the limits are or what the broad range of receptor variants will accept. Drug testing’s sole purpose is to get a drug on the market. As a result, 60% of drug usage (I think that is the figure) is “off label”, uses that have no evidence per se yet seem to work pretty well. Most of the evidence gathered is after the fact when single case reports keep saying the same thing and someone (who usually does not have the BigPharma grants) runs a smallish study. 
     
    The only truly reliable studies of human disease involve thousands of patients and decades of research. I read these studies and honor them even when my intuition says they are false (usually they are not, by the way) but there are not enough of these studies around to help the complex and often poorly defined illnesses that our patients bring to us. 
     
    It is not a matter of ignoring the statistics as much as there not being that many reliable statistics around for all the problems presented (Not to mention difficulties with diagnosis, multiple illnesses, ignorance and lies that we meet all the time in the practice of medicine.

  12. Let me make one other set of comments: 
     
    I don’t disagree with the statements about doctors and statistics nor do I disagree that a lot of doctors ignore evidenced based medicine (or even care about it.) But the problem with the suggestion that the body of medical genetics should be used in making treatment decisions (as good as it may be) is that no one will pay for it. In fact, the great concern about medical genetics becomes political. The problem of what will be done with the information comes down to who has the greatest economic interest in its use and who is the most organized to direct that interest. It won’t be the patients unless there is a government imposed solution of not being able to use the genetic information to set insurance rates (or some other way is set up to pay for treatment and then you just switch overseerers.) 
     
    I would love to have a set of inexpensive genetic tests that would help implement Dillenger’s Law (“You rob banks because that is where the money is.”) Any physician would. We are not Luddites after all, in fact are probably more accepting of technology than many other professions.  
     
    Genetics takes a special knowledge and an interest to really understand it. It is simple if that is all you focus on and in a geeky way very sexy. But who has time to figure all the permutations or keep them all straight? I have a hard enough time with all the liver enzymes that effect the medicines I use. Computer based tools are very helpful, but not practical when you don’t have enough information. Decision Matrices are quite good at making diagnoses but it is still a matter of GIGO and I alluded to the fact that we don’t always get the best information to start with, especially if the patient is a drug abuser or an alcoholic who wants to hide that fact. 
     
    Structured interviews work especially well in diagnosis but they take a lot of time to do, and time is often not available to make critical decisions, especially if the patient is not able to communicate well. 
     
    In my mind, everyone who has commented has very good points. If these suggestions were more practical they would be a tremendous boon to the profession and to patients. Structuring and implementing these ideas is possible, but will require more thinking and planning – maybe not only should physicians get the Ph.D in genetics, but an MBA to boot. ;’]

  13. mike, i don’t make shit up. go here, search “bayesian physician” using the search inside feature. select page 60 out of the results. the money shot: “in other words the probability of breast cancer is an order of magnitude lower….”

  14. But who has time to figure all the permutations or keep them all straight? 
     
    well, that’s a good point, but two points 
     
    1) perhaps licensing for medicine should be loosened or deregulated so that more people can get involved?  
     
    2) the example above was a “classic” bayesian fallacy. this is a normal human cognitive default. “fixing” it really isn’t that hard. doctors ignore or overrule human intuition all the time, because medicine is based on science. so they should be trained in the most basic of bayesian operations, it would only take one class.

  15. razib, 
     
    I don’t dispute anything that you say and I was asking the question about data on physicians because I was not aware of the research. 
     
    I would love to take a class in bayesian operations, I’ll have to google it up. 
     
    I did do an informal ( and I am sure non-Bayesian) survey of my younger colleagues, and most of them were familiar with the concepts of genetics that you refer to (unlike us old farts) but the universal comment was: “I bet the insurance companies would love to get their hands on that information.” The reason for this is the universal distrust of medical insurance companies who have open access to all medical records (if you don’t agree, no insurance) and will no doubt use this information to their economic advantage unless there is a fundamental change in privacy laws. 
     
    Another aspect is that many patients don’t want to know if they are genetically vulnerable since they equate this with a death sentence. Good education of patients will help this, but I have several patients whose parent have a devastating genetically transmitted disease and to a person they don’t want to know if they carry the gene. While it is clear to me that they should know this information (transmission, future planning, etc.), they all saw how their parent progressed, realize that the disease is not curable or even treatable, and freak out.  
     
    In that informal survey each physician also said that they would love to have the genetic information to help develop health plans for patients but that the political and economic sequelae make them wary. I don’t see the lack of use of this information being due to ignorance of the field as much as it is a concern about the non-medical uses of the information.

  16. OK, 
     
    I went here (http://yudkowsky.net/bayes/bayes.html) for the explanation and it wasn’t all that hard. In fact I got the first problem more or less correct (I didn’t do the math, but 10% was a close estimate based on the facts given in the first presentation of the question.) Pretty good stuff and what I have advocated to medical students without the knowlege of being Bayesian. 
     
    Thanks for the tip.

  17. mike, i assume that genetic information is going to make universal health care of some sort pretty inevitable. everyone is scared that they’ll lose their coverage, even if the reality is that it would only improve median health care. so just wait.

a