Against IRBs

Once upon a time some researchers gave people diseases and prevented their treatment without those folks’ consent or knowledge. Other researchers let volunteers think that they were torturing folks. This so horrified many that they created a system of regulation where any academic “experiments” must have prior approval by an Institutional Review Board (IRB). And that system has expanded to the point of requiring prior approval for any interaction between researchers and non-researchers intended to be the basis of an academic publication.

That is, researchers seeking publication can’t talk to people (e.g., survey), or buy or sell something with them, or even pay them to do trivial tasks like correcting spelling mistakes, without first writing out a detailed plan months in advance and getting that approved by a committee of other academics.

One common rule is “informed consent” – people must be informed in great detail of the consequences of their interacting with the researchers – researchers must tell much more than ordinary people must tell when they deal with others. A second common rule is that people must benefit in some other way than money – they must gain some sort of intellectual insight or learning. A third common rule is that no record can be kept of people’s identities unless a really strong reason is offered to the contrary.

IRBs seem a good example of concern signaling leading to over-reaction and over-regulation. It might make sense to have extra regulations on certain kinds of interactions, such as giving people diseases on purpose or having them torture others. But it makes little sense to have extra regulation on researchers just because they are researchers. That mainly gets in the way of innovation, of which we already have too little.

Notice that researchers continue to be allowed to publish their results, and give lectures and media interviews, without such prior approval. Yet couldn’t ordinary people be harmed by reading articles that induce them to have unethical or unpleasant beliefs?  Of course they could – it is only an accident of history that regulation does not also require prior ethical review of publications.

Added noon:

In the United States, IRBs are governed by … provisions of the National Research Act of 1974, for example defining IRBs and requiring them for all research that receives support, directly or indirectly, from … the Department of Health and Human Services. (more)

GD Star Rating
a WordPress rating system
Tagged as: , ,
Trackback URL:
  • http://entitledtoanopinion.wordpress.com TGGP

    When did they infect subjects? The Tuskegee experiments was pretty notorious, but in that case they just weren’t treating existing infections (and possibly deceiving patients to prevent them from being treated).

    • Constant

      When did they infect subjects?

      Without their consent or knowledge. That’s important. Even these days subjects are infected with their knowledge.

      • q23

        IRB’s are not about human protection anymore- they have gone way overboard- and make it hard to even use already existing data in analysis that can save lives. The VA Medical Center system is especially choked up with institutional paranoia such that much good work is not done and resources are wasted,

    • Tyrrell McAllister
    • radical_negative_one

      Glancing at the wikipedia article, it was indeed the Tuskegee experiment which led to IRBs. Also correct that the Tuskegee study did not actually infect people, though they did prevent treatment after antibiotics were found to be effective.

      There is mention of a study in Guatemala which “infected soldiers, prisoners and mental patients with syphilis and other sexually transmitted diseases, without the informed consent of the subjects, and then treated them with antibiotics.” However, the Guatemala experiments were apparently only formally acknowledged last year, so it seems like they don’t exactly fit here.

      I had also assumed Robin Hanson’s first sentence was a reference to Tuskegee, and wouldn’t be surprised if most people assumed that the Tuskegee researchers actively infected their subjects.

      On a related note, in a Psychology 101 class i was told about the Milgram torture experiment, but i don’t recall it being famous primarily for being highly unethical.

      • Tyrrell McAllister

        On a related note, in a Psychology 101 class i was told about the Milgram torture experiment, but i don’t recall it being famous primarily for being highly unethical.

        Indeed, I think that it’s mostly famous as “one of those awesome experiments that the IRB wouldn’t let us get away with anymore”.

      • PrometheeFeu

        Think more the Stanford prison experiment.

    • Claire
    • Phil Goetz

      Edward Jenner, developing smallpox vaccine, 1796. But – it was a good thing for him to do so! Even though he had a very high chance of killing his first subject. He saved many, many lives by doing so.

  • Pingback: Club Troppo » The idiocies of regulation edition #473

  • Amelia de Seattle

    Regulation in this case isn’t bad, but it is poorly arranged. Mere polling is exceedingly unlikely to cause harm (at least that’s my understanding, I’d be both very surprised and interested by the opposite being the case), but dealing with medications is much more likely to cause unforseeable harm (isn’t that what the study looks for?)

    Wouldn’t a better solution to the problem be to class the potential research into two (or more) groups, one of which would be “almost certainly safe” for things like polls and passive data collection (ie buying something), and one for drug trials and the like.

    The drug trials would have proper scrutiny to keep the unethical and crazy out (infecting people without their knowledge or infecting people without enough of your own knowledge).

    The “almost certainly safe” stuff could be rushed through quickly by someone reading maybe a 5-10 page outline of the experiment and making sure it’s not worthy of more scrutiny.

    • Jeffrey Soreff

      Agreed. The “almost certainly safe” research category would be much like the GRAS category for food additives.

      • http://silasx.blogspot.com Silas Barta

        Am I the only one who finds it amusing that “gras” is French for “fat”?

  • Julia

    I believe the problem you’re referring to only applies to institutional (university) IRB’s. My experience is limited, but when doing research outside of a university, I’ve only needed IRB approval for clinical trials. I’ve done simple marketing and linguistics studies as a university researcher, which required IRB approval, but there is no government body that also must approve the trial. In clinical trials, the FDA must also approve it. The quirk in the system is likely not a regulation, but a university code of ethics. Your gripe should be the university system.

    • Lex Spoon

      Right, IRBs are for universities, and they are an additional requirement on top of other regulatory review.

  • Tyrrell McAllister

    IRBs seem a good example of concern signaling leading to over-reaction and over-regulation. . . .

    Notice that researchers continue to be allowed to publish their results, and give lectures and media interviews, without such prior approval. Yet couldn’t ordinary people be harmed by reading articles that induce them to have unethical or unpleasant beliefs? Of course they could – it is only an accident of history that regulation does not also require prior ethical review of publications.

    I normally buy your “concern signaling” arguments. But surely a major contributor to IRBs is our fear of being objectified as the subjects of experiments by “cold-hearted” scientists.

    Being a subject of an experiment feels more like a loss of autonomy than does reading an article. It may be irrational to feel this way, but the origin of this irrationality doesn’t seem to be concern-signaling.

  • Michael Vassar

    This was the very second instance I ever noticed, growing up, right after the drug war, of policy that was simply not explicable by any attempt at a credible justification.

  • http://hanson.gmu.edu Robin Hanson

    TGGP, the Nazi experiments were important, and they infected folks.
    Amelia, the existing regulations for non-researchers seem sufficient. They already limit drug trials.
    Julia, see my added to the post.
    Tyrrell, I can also see IRBs as trying to bring low the scientists who see themselves too far above us.

    • Tyrrell McAllister

      Tyrrell, I can also see IRBs as trying to bring low the scientists who see themselves too far above us.

      Yes, but it’s not just that non-scientists resenting scientists’ status. That wouldn’t explain the lack of restrictions on publications that you point out. IRBs don’t exist primarily to counteract the scientist’s dominant position in the scientist/non-scientist relationship. That dominance might be resented, but it appears minor and innocuous compared to the scientist’s dominant position in the scientist/subject relationship. It is our fear of being placed under the latter kind of dominance that motivates public support for IRBs. That, at any rate, is how it seems to me.

  • Dave

    The Tuskegee experiments were certainly unethical by today’s standards. They were not as bad at the time as they have been viewed in retrospect. They were designed to determine the long term outlook for untreated syphilis. When started no satisfactory drugs were available other than highly toxic arsenic compounds. Effective drugs like penicillin were not widely available until after World War Two,about 1946. By this time all the patients would be in the latent phase. By this time about two thirds of the patients would have cured themselves. The other one third would have eventually developed neurosyphilis or syphilitic aortitis. This may have happened before the penicillin was available or later. No information about the efficacy ,dosage or any thing else was available at the time penicillin came out. It was quickly found to cure primary and secondary( early) syphilis.
    Once penicillin became available is it alleged that the doctors would have let actively sick patients die without treating them? If so they were indeed criminal but this issue is not well addressed. Since the patients were getting health check ups detection of any complications would be more likely than the people who received no follow up. An allegation I have heard is that the untreated patients spread syphilis to their wives and children but this needs further explanation,since late syphilis is not infectious. It only spreads in the early stages,and this would be long passed when penicillin came along. No nuanced explanation of the Tuskegee experiment is commonly presented to the public since the simple story plays so well.

    • Claire

      What about the ethical responsibility of a doctor to tell a patient they have a disease if they know it to be true? I do not think telling the patients they have “bad blood” was sufficient.

  • Julia

    Dr. Hansen,
    Wow. That is a lot of regulation. My exemptions makes sense.

    Outside of having to remove subjects names though, which is also a good policy, I can’t imagine this normally affects many people outside of university researchers. Who’s doing privately funded social science research? Think tanks? (There are costs involved with getting IRB approval. Private companies usually pay private IRBs because they are faster, so there’s that.)

    It can’t be stifling that much innovation. If you are getting approval for something that’s not dangerous, its only a waste of time. Time is a cost and I don’t discount it, because it can make subject recruitment a real pain!, and then your grant money can run out… but its not specifically preventing anyone from doing the research they would like in the social sciences. What prevents research is almost always lack of funding. Also, if you’ve handled subject recruitment, you know how paranoid people are. These excessive regulations help with recruitment, because they make people feel comfortable about participating. You’d have to pay a lot more for subjects if you removed the regulations.

    Anyways, I think they are a pain, so they need to pulled back a bit, but I have little faith it would increase innovation.

  • Robert Koslover

    Notice that researchers continue to be allowed to publish their results, and give lectures and media interviews, without such prior approval.

    Shhh! Don’t give them any ideas!

  • http://entitledtoanopinion.wordpress.com TGGP

    OrgTheory presents an unusual IRB request.

  • Chris T

    I work in a laboratory at a university that does subject trials. The amount of bureaucracy that we have to go through to get anything done is pretty ridiculous. Last year, we got approval to do a set of completely noninvasive experiments (some basic vision tests), but weren’t allowed to advertise it until we got approval. Word of mouth was fine, but putting up fliers was out.

    We also couldn’t compensate without approval, and adding on an additional (noninvasive) test took months.

  • Lex Spoon

    The Institutional Review Blog is a good read for people interested in this topic.

    http://www.institutionalreviewblog.com/

    I fully concur that IRBs are out of hand in social sciences and in computer science. People stand up for them in principle, but in practice they’re really stifling university research.

    • Chris T

      I think the main problem with IRBs is that they have gone from ‘Protect Research Subjects’ to ‘Prevent the Institution from Ever Getting Sued’. Thus we get the ridiculous over-focus on minutia and consent forms that read like legal documents.

  • RJB

    I remember that IRB oversight really stepped up at my school when this happened at Columbia.

  • Bill

    And that system has expanded to the point of requiring prior approval for any interaction between researchers and non-researchers intended to be the basis of an academic publication.

    You are underestimating. University IRBs require you to interact with them before you do any human subjects research. This includes (yes, it does!) downloading the publicly available CPS and analyzing it. For that particular example, you are exempt from human subjects review. But, you, the researcher, are not allowed to determine for yourself that you are exempt. The IRB still does that determination (though it is usually fast). Now, I know that most researchers do not, in fact, seek IRB approval every time they analyze the CPS and that they are never (?) punished for doing this. But, there is a formal requirement to do so. In addition, universities also require researchers to go through training courses before they are permitted to apply to the IRB (sometimes enforced, sometimes not). As far as I can tell, they only really care about this stuff for federally funded grants, since the granting agencies require pieces of paper certifying that IRB review has been done, but, in principle, the review requirements are universal.

  • Pingback: Did You Know the Federal Government Regulates Almost All Research Involving Human Subjects? | John Goodman's Health Policy Blog | NCPA.org

  • mdb

    You leave out all the reporting requirements after approval takes. I am familiar with IRB in Biotech research and in this field reports have to be sent to the FDA and the IRB and both followed up. I don’t know about social science, but the reporting in Biotech is as bad as the approval. Both require huge data dumps, interviews, etc.

  • Phil Goetz

    This is also a good place to mention

    – researchers are required to treat mice *much* more kindly than ranchers are required to treat cows

    – in some research, informed consent statements have been ruled invalid, IIRC under the assumption that the average person is not smart enough to give informed consent, so not only do you need to get consent, you need to provide proof that the subjects understood what they were consenting to

  • Douglas Knight

    IRBs are made of professors policing other professors at the same school. It seems like they should have strong incentives to be lax, to allow their colleagues to increase their reputation, and thus the reputation of the school. It would be another matter if there were a single national IRB policing more distant strangers.

    • Robert Koslover

      Seems to me that since so much university research is sponsored by the Federal Government anyway, the sponsors already constitute a semi-IRB. And frankly, given that our Federal Government is already broke, I personally would prefer that we not add any more bureaucracy to it.

  • http://deevybee.blogspot.com/ deevybee

    A commentator on my blog has just pointed me to your post.
    As a researcher, I’ve been very exercised on this issue.
    Oh, and increasingly you *can’t* publish your results without ethics approval – journals want confirmation it was obtained.
    In my post I suggest a mechanism, based on evolution, as to why we have got to this sorry state of affairs,
    http://deevybee.blogspot.com/
    In the UK, the Academy of Medical Sciences has done a great report which attacks overregulation (see my blog for details) – government at last seems to be listening.