Number One Pick on medical risk disclosure

Posted on by

Scott Alexander writes,

Moderna had to disclose to the FDA that one of the coronavirus vaccine patients got struck by lightning; after a review, this was declared probably unrelated. For the more serious version of this, read Get Ready For False Side Effects. Why does the FDA keep doing this if they know it makes their label information useless?

21 thoughts on “Number One Pick on medical risk disclosure”

  1. The article states that the FDA came to the conclusion that the patient who was diagnosed with arrhythmia likely got it from the lightning strike and not the vaccination.

  2. The reason the lightning strike victim was included was that the lightning strike (rather than the vaccine) was judged to be the reason for the heart arrhythmia that followed.

    Heart arrhythmias are serious possible side effects from many medicines. It was entirely appropriate, and very relevant, to explain that the lightning strike was a much more likely cause of the arrhythmia than the vaccine.

    In my opinion this post does not meet the much higher standards usually maintained on this blog. This one feels like clickbait.

    • I think the problem was with the underlying post by Scott Alexander, who misleadingly suggested the FDA ruled it “probably unrelated” to make it sound more ridiculous than it actually was and thereby give an excuse to write that whole post.

      When you read the underlying article, it’s clear that the FDA was saying unequivocally that they were “assessed as unrelated.” Scott makes it sound like the FDA had to hem and haw about the relatedness for the reasons he lays out in the post, when that doesn’t look to be the case.

      • Yes, the problem did propagate from Scott’s choice. But Arnold is also usually much more reliable in his choices about the items he curates. He can usually be counted to show us the best, not the worst, of those he recommends.

        • This objection strikes me as overly subjective. Whether the lightning anecdote was off-base or not, parts the post, particularly the side-by-side with aspirin and warfarin and whether that’s symptomatic of a broader disease, were interesting to me and I can’t find Professor Kling at fault for sharing it.

    • It’s from page 54 of the FDA briefing document for the advisory committee meeting that reviewed the Moderna vaccine.

      https://www.fda.gov/media/144434/download

      Greg has it exactly right. From a clinical perspective, the adverse event wasn’t the lightning strike, it was the arrhythmia. The lightning strike was invoked in assessing whether the arrhythmia was plausibly attributable to the vaccine.

      This isn’t going to get anywhere near the product labeling. It was irrelevant to Scott’s discussion about side effect warnings, and it was sloppy and misleading for him to mention it in that context.

  3. So with all due respect to Scott Alexander, the more serious version of this is how does a modern society use larger data streams to make effective medical decisions?

    More data is better, but that eventually means the amount of data is unmanageable for individual communication.

    We shouldn’t be asking why the FDA keep doing this if they know it makes their label information useless, we should recognize that adding warnings to the “label” concept is a construct from another age, and writing out of stuff most people cannot comprehend in one point type onto a sticker has been stupid for some time. Weird commercials on cable news listing cringe inducing side effects highlights how absurd this has gotten.

    A doctor talking to a patient face to face is even more limited. Asking for a patient to riff their medical history with a pen on to a piece of paper every time a doctor sees them is, again, stupid. The medical industry only cares about the data they own and is not seriously cooperating with efforts to give patients the means to maintain their own datasets and use them effectively.

    We have technical knowledge to connect datasets and analyze them better but we aren’t spending much time working at it.

    • +1
      Medical data connection and sharable should be a requirement for full gov’t re-imbursement. Could easily start taking off 1% per year, for 2 years, than 1% per quarter — until data is sharable in practice.

      Words in tiny type makes them worthless, as are too many side-effects. There should be some max % allowed, and only the top 2 or 3 listed with %:
      X (0.4%), Y (0.3%), Z(0.1%) – others on the website.

      Each drug maker should have a more standardized web page for each medicine approved by the FDA.

  4. The root cause of a lot of our problems today – much more than most people would expect – is simple legal uncertainty. There’s is an old and vast legal and economic literature on this problem. E.g., on the implications of the non-existence of a market to insure against a particular kind of legal risk. And my impression is that it’s been thoroughly mined out. We know a lot about the kind of jam we’re in.

    When people don’t know what they’ll get in trouble for, they have no choice but to turn off judgment, play it safe, and do stupid things with bad consequences, even though they know perfectly well that those things are stupid and have bad consequences. They are just doing their jobs and trying to stay out of trouble, and when inevitably things go wrong, no one will be able to blame them, because “doing what I was told” is the de facto ‘safe haven’ except for war crimes.

    In certain medical contexts this is euphemistically called “defensive medicine”, but the problems with this kind of risk aversion and complacency about it involve everything, everywhere.

    There are only two answers to this. (1) Keep treating people like risk-avoiding robots, but improve their programming so that they follow better instructions, or (2) to guarantee people legal safe haven for using reasonable judgment.

    The problem in both cases is a combination of “unpredictable judiciary”, “severe penalties for error”, and “the process is the punishment”.

    You can write better instructions, but you can’t predict if the judges will say those are valid.

    Or, you can try to give people safe haven, but indemnification is not a good answer, there is no such thing as “professional liability / malpractice insurance” in this context (see ‘unpredictable’ above), you can’t give officials unlimited authority to people to do just anything, and you can’t predict how the judges will interpret the limits.

    So your only option is to fix the judges.

    In the short term, you can’t do anything about them, you go to war with the judges you have.

    So again, you have no choice but to take the matter completely out of their hands, make the neutralization of safe-haven ultra vires, exercise sovereign immunity, eliminate all the relevant causes of action, strip jurisdiction, and also use the interstate commerce power to impose the same regime on all state and local jurisdictions.

    Assuming that by some miracle SCOTUS will let you get away with this, one can easily imagine alternatives to “list absolutely everything” for FDA requirements for side-effect disclosure warning labels. As any alternative to “everything” has to involve kicking some things out, you need a test that says the things you are going to weed out don’t meet some threshold. As possible examples, you could have a “top three” rule of most common side effects, or a “above 10%” rule for the number of users reporting that side-effect. There are many other possibilities, and at least one of them should be considered superior by most reasonable experts to the “everything and the kitchen sink” approach.

    But here’s the catch, actually, multiple catches.

    First, you essentially had to drop the legal equivalent of a small tactical nuke to get this one little thing done. Maybe SCOTUS will let you get away with it, but most judges are going to hate it and vow revenge. And most of the people who wanted to be able to sue and get a judge to order their desired outcome will be outraged, and they aren’t just going to passively acquiesce to the new state of affairs.

    Second, you had to get both houses of Congress and the President to agree to go nuclear for the sake of this one little thing. Good luck.

    But third and most important of all, literally everything that everyone does in government is like FDA warning labels and suffers chronically from the same fundamental problem. So you would have to go Full MAD into global thermonuclear war, really against the law itself and current constitutional structure (such as it is) in order to have any chance whatsoever on moving the needle in the direction of sanity, reason, and wisdom.

    And no one is willing to do any of that. Yet.

  5. I am sincerely incapable of comprehending how any sane person who has spent any amount of time investigating this vaccine could possibly reach the conclusion that the thing we should be most worried about is the FDA being too cautious.

    I have nothing but disdain for the FDA and want it abolished yesterday, but these sorts of milquetoast, Libertarianism-101 style criticisms of the institution that are being forwarded by people like Scott nowadays just seem so totally out of touch with reality as far as I can see it.

    It truly baffles me that people still can’t see how the problems with the FDA are so much more severe and insidious than just: “Oh no! This inept, overly-cautious government bureaucracy is interfering with the Free Market and hampering our noble, private pharmaceutical corporations with unnecessary regulations and safety standards that are preventing them from pumping us full of the chemicals we so desperately need to keep us safe!”.

    • Thanks for that link, very interesting. I found the data in Table one breaking down pain for Shingrix v Saline placebo for different age cohorts fascinating.

      The older people got, the *less* pain or severe pain they reported, and that *includes* reporting for the placebo!

      Shingrix (Pain/Severe Pain %)
      50s: 88.4 / 10.3
      60s: 82.8 / 6.9
      70s: 69.2 / 4.0

      Saline Placebo (Pain/Severe Pain %)
      50s: 14.4 / 0.5
      60s: 11.1 / 0.5
      70s: 8.8 / 0.2

      Not sure what the right interpretation is, but the trend is obvious and the differences substantial.

  9. The a few things I can think of that drive polarization:

    1) More things in government purview. I wouldn’t have imagined a year ago that many of the impositions I deal with daily would be within the governments purview nor that it would have such a huge partisan angle (red states open, blue states closed).

    2) More things in the cultural purview. I would say that employers and service providers are dramatically more partisan/political than they used to be. “Republicans buy Nike’s too” replaced by Woke Capital:
    https://twitter.com/WokeCapital/status/1362038612418846725

    3) I think there is a genuine fear of demographic driven Californification on the right and a genuine inevitable triumphalism of Californification on the left. If you think the ascent of the left is a fundamental demographic inevitability, then the left doesn’t need to adapt or compromise at all. Just ram that bill in 51-50, who cares, those losers are on the wrong side of history anyway. If we lose, just wait for a little more demographic change. We get XX million more voters in eight years, right.

    I could buy that if we had the demographics of 50 years ago, that any shift to far to one partisan side would trigger a backlash to the center, and so there wasn’t any need to get too worried over anything. Now I fear continuous leftward ratchet with no countervailing force in the long run.

  10. I don’t have any good policy ideas here. I have a “wish” that there was a way to make these frivolous lawsuits less risky and expensive to inventors, while still capturing the true negligence. Maybe making it a bit easier to counter sue for fees?

Comments are closed.