The Structure of Comparative Effectiveness Revolutions

Liberal blogs have been bashing conservatives over the "comparative effectiveness issue".  And they're basically right.  The idea that the government shouldn't test the relative effectiveness of various treatments because this might, someday, lead some moronic bureaucrat to try to ban treatments, is not a good argument.  Comparative effectiveness research is one of those things that even a conservative should be willing to at least think about having the government do, because the government doesn't have a vested interest in the outcome.

I do want to raise some potential issues here, however, not because I am against the idea--I am for it--but because there is a danger that the government seal of approval may become far too powerful.  Governments do not have the obvious conflict of interest that plagues some pharmaceutical industry research.  But they have different problems, some more prevalent in government because there is no countervailing market discipline to weed them out.

  • Perhaps the most obvious problem is that we won't entirely eliminate the financial motive--government workers sometimes leave their agencies, and the obvious place for them to go is to the companies they regulate.  If you wall off this lucrative avenue of escape (say by dictating that they can't work in a regulated company for 1-5 years after they leave their agency), you may have trouble recruiting good people in the first place, because working for the government will become something like a prison sentence.
  • Science isn't always cut and dried, but government reports are supposed to produce answers.  There's a danger the bureaucrats will be more definite than the science calls for.  This is a risk in the private sector, too, but private sector errors of this sort are rarely as powerful as government errors of the same kind.  Once the government establishes a standard of care, private companies will probably follow, even if they are wrong, because it's
    • Easier than doing their own analysis
    • A lot easier than getting sued
    • Possibly cheaper than the more effective treatment
  • Government agencies are much more vulnerable to interest group pressure than private companies.  Researchers will come under tremendous pressure to say that things work when they don't--not just from big, bad Pharma companies, but from patients who do not want their insurance company to cut off access to the treatment.  And see above:  a government report saying snake oil might work has more impact than a dozen private company reports saying the same.
  • Government power can perpetuate a bad paradigm.  I'm currently reading a book called Cure Unknown by a science journalist who believes she and her family are suffering from chronic Lyme disease.  I don't know if Chronic Lyme Disease exists, or is a figment of the imaginations of people with some unspecified systemic or psychological problem.  But some of the things she's angry about ring true to me because they sound a lot like other episodes from the history of science. 

    The spirochete that causes Lyme is hard to detect, so treatment guidelines focus on the "bullseye rash", not because there's any particular reason to think it must follow infection by the borrelia bacterium, but because it's easy to diagnose, and . . . it's part of the diagnostic criteria.  Everyone who has "real" Lyme disease has the rash, because the definition of "real" Lyme disease is having a rash.  This, of course, makes it hard to test the theory that the spirochete might cause symptoms other than a rash.

    Weintraub makes a compelling case that these sorts of hard-and-fast diagnostic rules have, at the very least, left some indisputable cases of Lyme undiagnosed, including that of Weintraub's son.  The CDC has turned this into a major problem, since of course most physicians do not pour through the journals themselves; they glance at the CDC criteria, which are quite restrictive.  It's pretty clear that scientists who have a lot vested in the current model of Lyme (their careers, possible malpractice accusations), have at least for now won the debate.  It's not quite so clear that they should have.  And the government imprimatur has done a lot to seal the fate of the dissidents.  This is all standard stuff to anyone who's read The Structure of Scientific Revolutions.  But those revolutions happen because there are multiple possible centers of power.  The government has the ability to potentially shut the revolutionary centers down.

As I say, I am in favor of doing the research.  But the dangers of this sort of government sanction are not quite so far off and imaginary as Matthew Yglesias and Hilzoy seem to think. I don't think conservatives have done a very good job of articulating those dangers (and don't get me started on the pharmaceutical industry!)  But I still think they're worth keeping in mind.