What is the landscape of uncertainty outside the clinical trial’s methods?

I live in the province of British Columbia in the country of Canada (right, this post is not by Andrew, it is by Lizzie). Recently one of our top provincial health officials, Dr. Bonnie Henry, has received extra scrutiny based on her decision to delay second doses of the vaccine. The general argument against this is the one I have heard from Dr. Fauci of the US, who has been various levels of adamant that you do what the trial did (I would say very adamant, very adamant, very adamant, then slightly less adamant after the kerfuffle with the UK). You don’t deviate from the methods of the trial.

This has got me wondering what the landscape of uncertainty looks like as you move away from the methods of a clinical trial. And what progress we’ve made — if any — on this in the last couple of decades, when I first realized how stark the divide between inside and outside the trial methods is for many.

Over 15 years ago I was helping take care of a 50-year-old family member who had cancer and was struggling to get through a 6 week regime of radiation + chemotherapy at a major cancer institute in Boston. She had gotten through the first couple of weeks okay, even driving herself the 6-8+ round-trip hours from her home to the institute five days a week for her daily radiation appointments. But things got progressively worse in the third and following weeks (when I was her trusty chauffeur and companion). By her fifth week she was in and out of the ER with various major issues and was receiving various infusions to attempt to prop up her system so she could survive the next dose of radiation. Every day before radiation she needed a series of tests followed by a visit to her oncologist to get approval for that day’s dose of radiation, and this did not seem out of the ordinary for the later weeks of high-dose radiation therapy.

At one visit, when things were going particularly poorly, the radiation oncologist was brought in to consult on whether to continue treatment. He was advising for continuing, though it would be hard. It took a lot of her energy to speak, so she was often quiet, but on this day she asked him: ‘why do I have to do this? I have done this for most of the 30 visits, why do these last few matter so much?’ And he told her the truth — “because we only have data on the people who get the full dose. We don’t know what happens if you don’t take the full dose, or you take a few days off before continuing.” It was very helpful. I remember she said something along the lines of ‘okay,’ and we drove home in semi-shock, but at least we knew why they were pushing for this now. It was always her choice, but until then neither of us realized how gaping the uncertainty was between, say, going for 27 of your total 30 radiation visits, and going for all 30.

Clearly, the ethics matter, and that’s especially clear with a highly infectious and deadly disease like Covid. I assume that many of these deviant public health officials who have delayed second doses have done the simple SIR model math and figured out that: (n higher number of people vaccinated at X% efficacy given Y weeks of delaying the second dose)*(black box uncertainty as you deviate away from the trial methods)=likely more lives saved. Henry has cited studies showing >90% efficacy for the three weeks after the first dose of the Moderna and Pfizer vaccines, so I suspect she’s feeling good that her X in extending the second dose to four months is still fairly high and thus has some internal estimate on the landscape of uncertainty beyond the methods of the trial, and there’s growing data on this.

But if you listen to various interviews with Fauci and other public health officials, I start drifting into memories of discussions that start with, ‘what is the variance of a fixed effect? It’s either 0 or infinity.’ Now, I don’t mean exactly that — but I do mean there seems to be a large gap in perspectives here. In one — the clinical trial methods must be followed to a T, until a new or properly vetted trial of any deviation is approved, conducted and reviewed. And in another — some adjustments happen given the potential for lives saved despite the uncertainty and that ‘population health data’ is then used to make further adjustments on the fly (in conjunction with other ways of viewing the clinical trial data you have).

These debates have made me wonder what progress have we made addressing this uncertainty from both a bioethics, and data collection and design standpoint? I am not (at all) a bioethicist but the rigid adherence to the trial methods doesn’t feel terribly ethical to me, and I think Covid has highlighted that. So I wonder how much has changed in last 10, 20 or 30 years of how those who deviate or ‘drop out’ of clinical trials are handled as datapoints. Are they required to be tracked? Or is it better to save money by focusing only on those who follow the trial perfectly? Is there an incentive for research or new methods or databases that compile these deviants to start fleshing out that landscape of uncertainty beyond the clinical trial methods? Or is everything beyond basically zero, or maybe infinity? Or maybe somewhere in between.

194 thoughts on “What is the landscape of uncertainty outside the clinical trial’s methods?

  1. I’d think that there’s not only the trial but also some medical/biochemical knowledge behind the decision to run the trial like this. Surely when making decisions about deviation from what was investigated by the trial this knowledge must be brought in somehow, and may or may not suggest how bad or not so bad an idea it could be to deviate?

  2. Lizzie –

    >… but the rigid adherence to the trial methods doesn’t feel terribly ethical to me,

    You have an estimate that the vaccine has an efficacy within a particular range.

    You can project those probabilities one way or the other but if you alter the protocol you run the risk of 0% efficacy, or even worse, is there some likelihood of creating a phenomenon of resistence by allowing a virus to linger and regroup?

    That seems to me to be a huge burden for policy-makers. This is very difficult decision-making about potentially very high damage (if low probability) risk. At some level values come into play. But I don’t understand why you put this within a framework of ethics.

    • I definitely see this as an ethical question (like much of public health and healthcare). In the absence of complete data, decisions still need to be made (but based on the data that is available.) And when priors, data, and models don’t give a feasible decision (which would be to vaccinate EVERYONE with two doses), ethics is the name of the game.

    • There’s definitely a tail risk with deviating from the protocol but I think there’s also a tail risk with staying with the protocol. If you believe delaying the second dose is more effective then that means sticking with the current scheme means the pandemic will be unnecessarily longer which means more time for troublesome variant shenanigans to arise. As a layperson it’s hard to say which is more of a worrisome tail risk (and I don’t know if even experts would know this, there’s just so much uncertainty) but it seems like ending the pandemic as soon as possible should be the primary priority.

      Also, I think an ethical framework might be beneficial here because well thought out heuristics may outperform formal models when there’s a lot of uncertainty. Though to be honest I don’t know much about bioethics and what they have to contribute to this issue.

      • Michael –

        > There’s definitely a tail risk with deviating from the protocol but I think there’s also a tail risk with staying with the protocol.

        I agree. I wasn’t dismissing that possibility. In fact, part of my thinking is that tails wag in lots of different directions and I don’t see how choosing a direction here boils down to an ethical choice.

        Particularly if you know with a reasonable degree of certainty that the existing protocol brings benefit. It might make sense to risk that benefit for the potential of a greater benefit, and one shouldn’t just ignore that potential, but I just can’t see how one determines which choice is ethically superior unless the balance of uncertainties is relatively unambiguous. I don’t think that’s the case here although I could be convinced otherwise.

        • Ethics can come in a few different ways. I don’t know much about it so maybe I’m off the mark but this is how I’m thinking about it:

          There are ethical concerns within a utilitarian / cost-benefit approach where you model, say deaths, under the two regimes and calculate point estimates and uncertainty intervals and then compare the two. The ethics decision here might be how to weigh expectation maximization against variance minimization. I think this is how you may be thinking about ethics? And in this scenario ethics would not be relevant if we can’t create models to begin with (more on this point at the end).

          Alternatively, I think one can use ethics to reject the entire idea of minimizing deaths or whatever outcome. Like Michael Nelson mentions below about “do no harm”. Under that framework, the existing policy is the default and given a lot of weight. So even if the two regimes have the same point estimates and uncertainty intervals you would strongly favor the existing policy. At least that’s how I see it but I think proponents of that way of thinking may disagree (but I do acknowledge that it’s a bit more complicated than how I said it). In general, I don’t really like these alternative frameworks but maybe some are useful.

          Back to this specific situation: there’s a recent preprint (hat tip Marginal Revolution) that seems to argue that there’s more risk regarding mutations on the do not delay second dose side than there is on the delay second dose side. I haven’t read it yet though, just the abstract and the excerpts posted on MR.

          https://dash.harvard.edu/bitstream/handle/1/37366988/Dose_sparing_evolution.pdf

      • I’ve always liked the ethical framework of “First, do no harm.” It’s nice, because, it says what the primary priority should be right in the framework!

        Seriously, though, from that perspective, we should not be weighing “on-protocol deaths” against “off-protocol deaths,” we should be weighing “allowing deaths” versus “causing deaths.” Sticking with the protocol, until we have evidence that doing otherwise is safe and effective, may result in more deaths–we don’t know–but we won’t have caused them.

        • Michael –

          > I’ve always liked the ethical framework of “First, do no harm.” It’s nice, because, it says what the primary priority should be right in the framework!

          I don’t understand in this context where the “doing harm” is. Is it delaying the 2nd dose or not delaying the 2nd dose? Which is the ethical decision?

    • > if you alter the protocol you run the risk of 0% efficacy, or even worse, is there some likelihood of creating a phenomenon of resistence by allowing a virus to linger and regroup?

      I think this is the framing I hear often, and it seems a little extreme. Sure, we run the risk of 0% efficacy, but we have a lot of information that says we are unlikely approaching 0% efficacy with these delayed doses, so why act as though this is an important possibility we need to avoid and instead build on data and modeling?

      I think there *is* a real fear that if we don’t vaccinate enough we’ll miss the likely now small chance to eradicate the virus (https://www.nature.com/articles/d41586-021-00396-2) and it will evolve. But we’re not exactly coming up with the dosing strategy to avoid virus evolution — if that were our goal we would be taking a much more global approach to vaccination because the goal with vaccination is fewer deaths and herd immunity (only the latter mattering to slowing virus evolution) … and getting to herd immunity faster is part of the argument to delay second doses.

      • If anyone has been operating with the goal of eradicating COVID-19, well I hope they weren’t in a position to seriously gum up the works. That is not a remotely plausible outcome and thinking there is even a “small chance” of doing so should not guide public policy.

        The vast majority of people who wish to receive a COVID-19 vaccine (at least in the developed world) will eventually have a chance to do so. If there’s only one specific dosage and timing that was subjected to a clinical trial it seems obvious that you administer in that manner and not trying to vaccinate more people, sooner, while hoping this ad hoc, untested method also works to some extent.

  3. This reminds me of the 2018 discovery that low dose dexamethasone for myeloma produces much better outcomes than the high doses used for many years. Dex’s severe side effects (e.g. rhabdomyolysis) were first recognized 50+ Years ago but the doctors kept giving the high dose despite severe side effects in myeloma patients. Why? It’s the dose used in the lenolidomide trial that showed improved survival. Why not try a lower dowe? It’s the only dose for which they had data.When a trial of low dose dex was finally run side effects were greatly reduced and survival increased by 1/8. It makes you wonder how much pain and death could be avoided if the uncertainties that come with a trial were more widely recognized.

    • Thank you for this example. It does point out well that we should not automatically stop studying a treatment when one clinical trial has been finished.

      And (having had a sibling die of cancer a few weeks ago), I wonder if it might be worthwhile to try some clinical trials that involve things like giving the patient the option of transferring to hospice care at any time during the clinical trial. That could give useful information on patients’ perspectives on willingness to endure severe side effects of treatments being studied.

  4. Lizzie:

    You raise a good question, and I don’t think enough has been written about the steps from (a) inference about a particular experimental condition to (b) decisions of what to do in the wild. One useful framing could be in terms of treatment interactions. We don’t typically have enough data to estimate interactions of interest, but these interactions are crucial in deciding what flavor of treatment to use.

    • Don’t know if you’ve followed this, Andrew, but the EU is taking interactions very seriously right now, and doing a terrible job of it. That is, they are chasing rumors and small studies that suggest the AstraZeneca vaccine doesn’t work for seniors, particularly in France and Germany.

    • Yes, I assume the scaling hurdle of interactions is much of the issue. But there’s also what happens to data once protocols become the ‘standard of care.’ If data are centralized maybe eventually there are enough consistent deviants for useful inference (but maybe there is no incentive to do those analyses)? Or maybe there is still a strong push to get so many to follow the protocol that you end up with mostly data repeating the trial methods and then you can continue to improve your effect size estimate and that’s the aim? Or maybe there is bias in where the deviations from protocol happen … or we just don’t centralize data.

      And I do find how this uncertainty is communicated to the public and patients less than ideal, but that’s from my very personal perspective.

      • Hey Lizzie and Andrew,

        I can see how interactions can be part of the problem but seems like you’re now entering into observational causal inference territory.

        Seems like the hypothetical you described could be answered if we allow the ‘treatment’ to be a multivariate vector (could include number of times given treatment, time between treatments, dose amount, etc).

        However, in the problem you described then the ‘treatment’ would not be randomly assigned but (at least in part) be self assigned.

        At which point you can use some observational causal methods (propensity scores, matching, etc) to get back at the causal effect of these treatments. But that would sort of mitigate the benefits of doing a random trial in the first place.

        It appears that you can get the benefits of a random treatment assignment is contrary to having people choose they’re treatment.

        • edit last line

          It appears that the benefits of a random treatment assignment is contrary to having people choose they’re treatment.

  5. Strict adherence to the trial guidelines can be motivated by a number of less-than-ideal factors (as is constantly pointed out by Alex Tabarrok on Marginal Revolution). For example, it is safer for your reputation since you can always blame the guideline. In the case of one dose vs. two, it also is relatively easy for risk averse public health officials. Tabarrok makes a strong case that one dose to as many people as possible, before second doses, is the rational policy choice. I am not convinced, but there are plenty of arguments both ways.

    So, the general question you raise should certainly depend on the quality of the evidence we have for deviating from the trial guidelines. And we have some, but not nearly enough, and not clear enough to conclude anything. The biggest problem I have with deviating from the guidelines is that it opens the door to almost anything. In our world of false facts, this seems enormously dangerous to me. But it also seems unwise to be too rigid, and it seems unethical to do so regardless of the individual case (such as the one you describe). I doubt there can be any clear rules applied to such circumstances. The trial guidelines are based on some evidence. As additional evidence is gathered, there may be reasons to consider deviating from those guidelines. But the standards for evidence and analysis seem no different than for anything, including the original trial protocol. There is bad analysis and better analysis (and worse analysis). If we ignore any legal ramifications and vested interests, we are left with the same uncertainty we face in virtually every public and personal decision.

    A bit off the main topic, but along the same lines, there is the issue of “intention to treat.” Virtually all medical studies (as far as I know) use intention to treat to group trial participants – some of whom end up refusing the random treatment they were assigned to. If we use the actual treatment received, the results can differ (I’ve seen some studies that show a big difference). I’ve never found intention to treat very sensible – the actual treatment would seem to me to be a better measure of an intervention’s effectiveness. However, it does raise selection issues.

    • Dale, your second point is interesting. It is now somewhat standard in education research for new interventions to be tested in two phases. The first is an “efficacy” trial, where fidelity may be artificially boosted, and/or the analyses may focus on the “treatment on treated,” in order to show that the intervention works under ideal circumstances. The second is an “effectiveness” or “scale-up” trial that measures the effects in a realistic setting. I believe the main motivation for this model has something to do with the practical and political fights in federal funding of education research in the early 2000’s, when there was a sudden shift from funding lots of development and demonstration projects, to awarding huge grants for very rigorous, large-scale RCTs of the few interventions that had strong correlational or limited experimental evidence already. There appears to have been a compromise, where education researchers have become more quantitative and the agencies fund more opportunities to develop rigorous evidence. Anyway, it’s a different model.

  6. There’s a vast difference between individualizing treatment and revising official treatment protocols. In the public health context, that probably wouldn’t mean literally designing a different treatment protocol for each person, but I’d at least like to see some debate about, maybe, spacing out doses over a period of time based on the risk level of a person or the caseload in a region. But the whole idea of tinkering based on supply-and-demand becomes a lot less palatable if, instead of calling it “delaying second doses,” we call it “denying patients the approved treatment and substituting an unapproved one.” Especially given (my perception that) the biggest obstacle so far has been efficient distribution, not supply.

    • Michael said,
      “Especially given (my perception that) the biggest obstacle so far has been efficient distribution, not supply.”

      I see “efficient distribution” and “supply” as overlapping problems: Inefficient distribution at the top level often creates inadequate supply at lower levels.

      • Not so much overlapping as causal. Fix the distribution, the supply improves. Increase the supply, overwhelm an already precarious distribution system. “The spice must flow!” (Please excuse my irrepressible nerd instincts.)

        • For the specific situation of delaying second doses I’m not sure that tradeoff is true, if I’m understanding you correctly. Like it doesn’t matter administratively, I think, whether a clinic makes two appointments for two people or two appointments for one person. The burden on the system should be the same under both scenarios but in the first you get two people vaccinated versus just one in the latter.

  7. 1. If the radiation trial was only done for 30 days with nobody missing a day, there’s pretty much no chance that is optimal on average, think of how spectacularly lucky that would be.
    2. Even if it were somehow optimal on average, it would still be spectacularly unlikely that it would be optimal for this particular woman.
    3. If the radiation and its side effects weren’t going to kill her, then presumably it’s better to do more days because you kill more cancer cells. And skipping a day could let some cells recover that would otherwise be killed. On the other hand, if skipping a few days let her recover strength so that she could survive another week of radiation (say), this she could potentially come out ahead.
    4. The people who ran the trial presumably do not think there’s an enormous difference between 28, 30 and 32 days for example: if they thought 32 had a decent chance to be much better than 30 then they would have run a trial at 32.

    Putting it all together: I think there’s good reason to tell people to do as many days as they can tolerate (or survive!), up to around 30. 30 days of radiation should be better than 27, when it comes to extending life, all else equal. But all else is not equal, and once someone is having to go to the ER and have various life-saving treatments just to survive the radiation, there has to be at least a decent chance you’re doing more harm than good.

    • I generally agree. On your point 4, I do wonder how often longer treatment regimes run in 5 day intervals (aka, the length of a midweek) and that we lack inference especially the further away you get from n/5. Especially as radiation is mostly about the total dose I can see the rationale to focus studies on varying that instead of length or other treatment decisions.

  8. I am not sure I see an ethics (whatever that means) problem, it is on the contrary a very down-to-earth and material question.

    It is intellectual laziness of the highest order (and borderline religious) to keep repeating you must do exactly as the trials did. It may also be out of a desire to not take responsibility for anything.

    This is the same issue with Oxford-AstraZeneca for older people: the effect was not tested so it does not exist. We are supposed to believe that, somehow, after 65 years of age, the mechanisms of this vaccine that work for younger people and the mechanisms of other vaccines that work for all ages stop existing.

    But nobody ever wondered if the vaccine work for a 35-year-old redhead woman with an engineering degree having emigrated from Canada to Florida before the age of 12, even though I bet there was none on the trial.

    • I wonder if part of the issue is that the FDA etc. are fairly old organizations, and biological knowledge has advanced *massively* in recent decades. Could this be a factor — IE the way the organization makes decisions implicitly includes the assumption that prior/general biological knowledge is not very useful (because it wasn’t in the 1930s, or even 1962)?

      I do think the avoidance of responsibility for risky decisions (as in any bureaucracy) is at least as important, though?

  9. Being UK based, this is something that has given me some concern since the UK introduced the delay for the second COVID vaccine does. I know my personal preference would be, for example, my elderly mother to get her second dose on the schedule rather than me (who is outside vulnerable groups) get my first one earlier. That this is, inevitably, being trialled on some of the most vulnerable to my mind makes it indefensible.

    However, this is based on a basic presumption. I would have thought that at phase 1 or phase 2, tests were done to determine the best interval between vaccines. I’d imagine these would be low powered due to the small numbers, but I haven’t seen anything data that this was done or what the variability was. If these showed little difference between 4 & 12 weeks I might be somewhat reassured. I do know there is some monitoring going on – a couple of friends who work at a hospital are on the 12 week interval, but do twice weekly swab tests. It does seems a bit late thought. Anyway, anyone seen any data on trials of different intervals, or even know if its been done?

  10. How hard would it be to modify clinical trials (e.g., of vaccines) so that the experimental condition contains a range rather than a single value on timing, dosage, number of shots (for vaccines). It seems that the usual practice is to decide, on the basis of very small samples, on a single value for each parameter and then test just that value on a much larger sample. But some variation around the chosen value is unlikely to weaken the effect very much, but would tell us how likely it is that deviations from the critical value would matter.

    Such variation is done to some extent with patient variables. Happily, they don’t decide to do the large trial on people age 65-70 only, so they get some idea of how effective the vaccine is at different ages.

    When we have no good evidence about deviations from the protocol, it is no more “ethical” than anything else. It is a matter of decision making under uncertainty. But the uncertainty reduction that results from clinical trials is lopsided, providing much information about some variables (vaccine vs. no vaccine) and little about others.

    • A modification like the one you describe would presumably complicate logistics, scheduling, and record-keeping, and would certainly complicate analysis…but I agree with you that this is something that should be considered. In the vaccine trials, if the time between doses in the trial varied between, say, 10 days and 20 days, we’d at least know something about the slope near 14 days. It’s very possible that it’s better to wait longer.

  11. “because we only have data on the people who get the full dose. We don’t know what happens if you don’t take the full dose, or you take a few days off before continuing”

    So how do they decide on the 30 number? Has the trial concluded that on average 30 has better survival than say 28?

    • Most likely the trial concluded that 30 does offers better survival than 0 doses.

      These things can turn into a bit of Catch-22. You do a trial based on 30 doses because you believe (based on pilot studies, theory, whatever) that 30 doses is more likely to produce a benefit than smaller doses would.

      Then once you’ve established that you can save lives with 30 doses while meeting some prescribed safety criteria, it would not be considered ethical to deliberately withhold some portion of the “proven” efficacious dose from some future group of subjects in a reduced-dose trial.

      • So wouldn’t it make sense in the trial itself to incorporate some variation? Eg To try say 25, 30, 35 doses?

        Is there a ethical problem there? Typically for such situations how strong is the doctors original belief in the original metric chosen?

  12. Jon, Rahul:

    You ask why they don’t incorporate some variation in these trials. It would be fine if they were to do so; as Jon points out, you can still estimate an average treatment effect from such a study. The trouble is that a narrow variation will not give you the leverage to estimate interactions with any degree of accuracy, so you won’t be able to use this study to learn anything useful about relative effects of different versions of the treatment. Remember the rule of 16.

    • Is there any way to structure the post trial protocols to incorporate variation to understand these effects?

      Eg following a trial with 30 days should individual docs try variations in a systematic way? Instead of mere post approval surveillance more of a post trial extended trial to tease out variation from a larger sample?

      Or again, would it be ethically unpalatable to try a 25 day treatment on a patient following a 30 day trial?

      Should we pretend we have discovered the optimum ( 30 days) or keep jiggling to see if there’s a better place to be in. I guess it boils down to risk aversion in some sense?

      • Rahul:

        I think it does make sense to incorporate variation in the post-trial protocols. Variation is going to happen anyway, so we might as well make use of it. But to come to strong conclusions, you’ll need lots of data or strong prior information. You can’t go with the classical approach of just reporting statistically significant results.

        Regarding your question about ethics: This depends on your prior information. I assume the 30-day rule comes from some scientific model. It could be that, in this model, 25 days is about the same as 30 so the choice is arbitrary. Sometimes it’s possible to gather intermediate data, biomarkers that can allow the scientific model to be explored and tested.

  13. > what progress have we made addressing this uncertainty from both a bioethics, and data collection and design standpoint?
    Not much. A large part of the problem might be due to most researchers not learning much more than cook bookish clinical trial methods. That is they don’t have a deep understanding of the purposes and limitations of the methods but only a surface level knowledge. Or know who to trust on that. That likely includes the individuals you mention in your post.

    Now, the first question to address in a completed study should be what study should be done next, why and how. That includes evaluating all other similar studies (aka meta-analysis), the relevant clinical background knowledge and an informative cost/benefit analysis. If for ethical or logistic reasons the answer is that no additional study can be done or done in time, then usual clinical trial methods should be replaced with “expert judgement”. That varies and is of variable quality.

    We did succeed here with a small non-randomized study that did change clinical practice by arguing a randomized trial would not be feasible given the possible 70% mortality in previously health patients. It was observed to be 35% with treatment. About 5 to 10 years later a group tried to do an randomized trial but shut in down when they could not enroll patients. https://academic.oup.com/cid/article/28/4/800/401565?login=true

    • One problem is this intersection between science and freedom. When there’s no consensus even within the scientific community about a certain outcome does it make sense for the regulator to mandate that outcome? e.g. Two dose protocol.

      I think we would be better off (for both science and freedom) if in such situations the regulator allowed public choice. If we were more willing to run experiments of this sort perhaps we would gather more data. Which would then allow for greater public consensus?

      I think a large part of the problem is the regulator being forced (or thinking it is) to pick one option. That compels them to behave they are more certain than they actually are!

      • How would you allow “public choice” when there are 10 million doses available and 100 million people who want them?

        Certain or not, the “regulator” has to choose between two undesirable options. No individual member of the public actually has a choice until the allocation policy has been chosen.

        • Not that it solves the two dose problem but they should allow the markets to solve the scarcity problem like we do with most other things!

          Here in India both the vaccine manufacturers are complaining that hey have stopped production since they have too many doses waiting to be picked up by the Government. In the millions.

          And yet here I sit barely a few hundred miles away in Bombay waiting for a dose. Furthermore the Government wants to give it out free whereas my willingness to pay would probably exceed a 1000 USD.

          Hardly an efficient outcome. The obsession with centralized distribution and rationing just isn’t working.

          I would rather they allowed the usual market driven medical distribution chains to work and subsidize it for those who actually cannot afford it.

          A two pronged approach of free markets + public services should work much better than the current mess.

        • Agreed, though here in the US I would suggest to ration about 60% of the doses and sell the 40% in a dutch auction style (declining price, you wait and pay when it gets down to what you are willing). The rationing isn’t terrible here but it certainly doesn’t let people who know they are at high risk of contracting it express this information by any means.

        • It puzzles me that globally there’s very little discussion of this!

          Are there any nations using a significant component of free market strategy for their vaccine distribution?

          What happened to all the libertarians?!

        • I think it just *sounds* ethically iffy to way too many people for money to play a role, even if it actually makes sense in particular situations.

          This is shaped by public perception and PR/”optics” as well as the most efficient way to do things.

        • What’s ironic is that the same nations tolerate nay encourage markets for most other medical interventions, sometimes far more critical too.

          In fact, although covid vaccines have been rationed and centrally distributed gratis, fall sick from covid and bang you are again thrust into the arms of the evil markets for treatment choice.

      • Along with Brent Hutto’s point, what do you do when there is no consensus about anything? I am not sure I can think of a single issue where there is a consensus any more – is there a critical mass of agreement/disagreement that would trigger the public choice mechanism? And, what comprises “the scientific community?” Is it by credentials, reputations, citations,…?

  14. Humans are somewhat diverse and outbred. Clinical trials try to overcome this inherent lumpiness in the biological pudding of people by using randomization to steamroll out the differences. This approach has obvious drawbacks, but it is the only that works reasonably consistently. It routinely incorporates such assumptions as the seven day week; there may be advantages to treating some disorders on a six or every eight day cycle. Doses are often multiples of 50 or 500; maybe the right dose of doxorubicin for many people is 47 or 53 mg per meter square. It is only possible to generate data from a small number of trials, and non-seven day based dosing or doses based on some random number have simply not been explored. There is a cultural imperative to push the treatment in my profession; the expression “He is a very aggressive treater” is a term of praise. That’s reality. Every idea can’t be tested.
    On the other hand going outside the limits of what has been established by the slow and painful processes that we have is also difficult. Many brilliant ideas thought up by highly intelligent people have turned out to be wrong. Most clinical trials are not successful. Many successful trials yield very modest results.
    I have treated people outside the parameters of established trials. I consider myself the Babe Ruth of treatment, 714 homeruns, 488 strikeouts (some objective observers might dispute this self-assessment.)

  15. It really seems like medical ethics has very little to do with ethics and everything to do with prestige-conservation and liability transfer.

    • I think the field is very conservative due to past bad examples.

      Whether one thinks this degree of conservatism is warranted in the 2020s depends, IMO, on whether you think current societal structure & technology is different enough from the 1950s and before to rule out those bad examples being repeated. (I do.)

  16. What I read in what you are writing is that the clinician did not follow the feedback of the body to refine – that is to say, over-ride – the rough generalization he had from the trial data, and the resulting missfit resulted in pain.

    That over-ride may well be definable by statistics- your call; I am primed for casuistry

  17. I think the common mentality is that if your confidence level is 94.99%, you can confidently reject that the vaccine works, and if it is 95.01% then you can be absolutely sure it works perfectly. Whether they have low IQ’s or high, a lot of these people have rigid brains and no understanding of what they’re really doing combined with a religious belief that they’re being “scientific” and superior to people who don’t know about the 5% rule.

  18. It is kind if a moot point now:

    Findings on B.1.351 are more worrisome in that this variant is not only refractory to neutralization by most NTD mAbs but also by multiple individual mAbs to the receptor-binding motif on RBD, largely owing to an E484K mutation. Moreover, B.1.351 is markedly more resistant to neutralization by convalescent plasma (9.4 fold) and vaccinee sera (10.3-12.4 fold). B.1.351 and emergent variants13,14 with similar spike mutations present new challenges for mAb therapy and threaten the protective efficacy of current vaccines.

    https://www.nature.com/articles/s41586-021-03398-2

    Not looking good for the current batches of vaccines when it comes to reducing transmission. In a couple months it will be all resistant variants. Looks like ~20% of convalescent sera retained its neutralizing activity though (probably due to non-spike antibodies and vs the S2 region that was mutated to stabilize the vaccines):

    It is also interesting to note that cases such as P7, P10, and P18 have neutralizing antibodies that are essentially unperturbed by the multitude of spike mutations found in these two new variants (Fig. 3b). A detailed analysis of their antibody repertoire against the viral spike could be informative.

    Hopefully the t-cell immunity is more robust so symptom severity is less, but there have already been many reports of more severe symptoms upon reinfection.

    • This seems very overly pessimistic – the mRNA vaccines at least are very strong, so reduced neutralization doesn’t *at all* mean “no longer useful”.

      And in the long run (probably even in the medium term – the high-risk population, at least in the US, will be mostly protected very soon) reducing transmission isn’t what’s really important – reducing severe illness/hospitalizations/deaths is.

      And reports of more severe symptoms upon reinfection don’t mean anything by themselves – of course there will be some. The question is whether *average* severity is less.

      • It is the *magnitude* that is concerning. You lose ~10x due to mutations, then 2-5x due to waning after a few months. Then if you limit to one dose that takes another ~10x off the peak value.

        It is obvious that any neutralizing protection will be very short lived. A few months at most.

        And look at the sudden selective pressure in mid-December 2020 here (fig 2): https://www.medrxiv.org/content/10.1101/2021.02.23.21252268v1

        These variants are being driven by pressure to specifically escape the vaccine. This is what happens when you create a monoculture, should have used multi-valent vaccines to begin with.

        • IIRC the evidence now seems to be showing one dose working rather well in practice (not sure how well that correlates to antibody titers, but real-life efficacy is what we mostly care about…)

          Remember antibodies aren’t the only factor in immunity… there are T-cells as well.

          >>These variants are being driven by pressure to specifically escape the vaccine.

          I doubt it. B117 seems to be the one winning out in the US and a number of other places, and it isn’t one of the ones with more vaccine-escape concerns, IIRC.

        • It is anti-protective for the first week. This has now been reported over and over. People should be warned to be much more careful during this week.

        • It could be, but it could also be reduced caution after getting the shot (although for nursing home patients perhaps this is not the case).

          In any case, the advice is IIRC to retain all precautions until at least 2 weeks after second dose (not sure what it is for the J&J vaccine)…

        • It could be, but it could also be reduced caution after getting the shot (although for nursing home patients perhaps this is not the case).

          It is obviously the lymphocytopenia reported in early Pfizer trials and then seen in the RCT. Should I repost this info once again?

        • Earlier you posted:

          “It means people are getting sick and dying in the short term.”

          And I responded:

          “I didn’t see anything about clinical outcomes for those who got a detectable infection during the first two weeks in the Danish study.

          Did I miss that?”

          And your response was:

          “Seems like something worth reporting on.”

          In other words the paper doesn’t report on clinical outcomes, therefore gives no basis for your claim that the one week drop in white blood cell counts is causing people to die.

          Another instance of your citing a paper then making claims unsupported by the paper.

          Tiresome as well as dishonest.

        • I really don’t understand why you are emphasizing any potential doubt/issue with the vaccines. Nothing is 100% perfect, but they seem very good overall – and if the early concerns like antibody-dependent enhancement were real, we’d have seen obvious evidence by now (from natural infection, if nothing else; it’s been a year since the first big spikes, so antibodies should have waned).

        • confused –

          > Nothing is 100% perfect, but they seem very good overall

          I thought this might be relevant:

          -snip-

          Data released by the Health Ministry Monday provided a further indication of the effectiveness of the coronavirus vaccine in Israel: Out of those who were tested for the coronavirus at least a week after their second shot, less than 1 percent tested positive, and less than 0.2% developed COVID-19 symptoms.

          The data shows that out of 3,387,340 vaccinated people who had had more than a week pass after receiving their second vaccine dose, only 4,711 were found to be positive for the virus and of those, only 907 developed symptoms, including fever or respiratory problems.

          Israel has almost exclusively been using the two-shot COVID-19 vaccine developed by Pfizer and BioNTech. On Sunday the Health Ministry released data showing that less than 3% of all seriously ill COVID-19 patients in Israel have been fully vaccinated.

          Of the 6,095 coronavirus patients hospitalized in serious or critical condition since the start of Israel’s vaccination campaign, only 175, or 2.87 percent, had received the second vaccination dose, the figures show.

          At the same time, 4,589 patients, or 75% of those in serious or critical condition, had not received a first dose.

          -snip-

          Now of course, it would be different if they were testing people one week after their first shot. It would be interesting to see a dichotomous evaluation of a random sample one week out vs. no vax. But meaningfully controlling for confounds would be awfully tough.

          It would also be nice to know what % of a comparable sample of non-vaccinated were infected and/or symptomatic, but again getting that kind of sample would be tough. The best studies I’ve seen – where they make a serious attempt to control for confounding variables, shows a quite promising comparison of vax vs. non-vax.

          So I’m going to take some solace in 0.2% symptomatic after vaccination.

          https://www.timesofisrael.com/of-fully-vaccinated-israelis-only-0-2-develop-covid-19-symptoms/

        • Factually correct information can still be misleading, if selectively presented. It seems you are only mentioning potential problems and not the *vastly greater* success stories.

        • @Joshua: Yeah, I think everything I’ve seen is significantly better than just “some solace”, it seems extremely promising.

          I wish they would report the time scale after first/second dose for those who were hospitalized, though…

        • Factually correct information can still be misleading, if selectively presented.

          Correct. Just look at the quote Joshua posted for an example of this.

          at least a week after their second shot… who had had more than a week pass after receiving their second vaccine dose

          They should not leave out that in the first week all evidence and reason (science) now says you are *more susceptible*.

          Then there are the issues of waning and new strains. I personally hope the news just stops talking about covid altogether no matter what happens but it is obvious that people getting infected during that first week are getting sick and incubating resistant strains.

        • Anoneuoid
          Now you have me confused, not the “confused” persona/author. You appear to be saying that the first week after the 2nd dose is an especially vulnerable time. Yet the study you list above concludes “The results were promising regarding the VE both within and beyond seven days of second vaccination with the BNT162b2 mRNA Covid-19 Vaccine currently used in many countries to help mitigate the global SARS-CoV-2 pandemic.” So, are you referring only to the period after the first dose? Or are you disagreeing with the study you cited? I’m just trying to understand and the article you cite and your last comment appear to be contradictory.

        • You appear to be saying that the first week after the 2nd dose is an especially vulnerable time.

          No, there is no evidence for that.

          So, are you referring only to the period after the first dose?

          Yes, there is a lot of evidence for this.

        • “Now you have me confused, not the “confused” persona/author. You appear to be saying that the first week after the 2nd dose is an especially vulnerable time.”

          Nah, he’s talking about the first week after the first dose.

          Note that the protection given afterwards, including after the second dose, swamps whatever minor increase there might be in the IR in that first week.

          Overall, the vaccines work, and work well. The studies he points to don’t argue that they don’t, not at all.

        • Healthy adults 18 to 55 years of age or 65 to 85 years of age were eligible for inclusion.
          […]
          The largest changes from baseline in laboratory values were transient decreases in lymphocyte counts, which resolved within 1 week after vaccination (Fig. S3) and which were not associated with clinical manifestations.

          https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7583697/

          As specified in the protocol, suspected cases of symptomatic COVID-19 that were not PCR-
          confirmed were not recorded as adverse events unless they met regulatory criteria for
          seriousness. Two serious cases of suspected but unconfirmed COVID-19 were reported, both in
          the vaccine group, and narratives were reviewed. In one case, a 36-year-old male with no
          medical comorbidities experienced fever, malaise, nausea, headache and myalgias beginning
          on the day of Dose 2 and was hospitalized 3 days later for further evaluation of apparent
          infiltrates on chest radiograph and treatment of dehydration. A nasopharyngeal PCR test for
          SARS-CoV-2 was negative on the day of admission, and a chest CT was reported as normal.
          The participant was discharged from the hospital 2 days after admission. With chest imaging
          findings that are difficult to reconcile, it is possible that this event represented reactogenicity
          following the second vaccination, a COVID-19 case with false negative test that occurred less
          than 7 days after completion of the vaccination series, or an unrelated infectious process. In the
          other case, a 66-year-old male with no medical comorbidities experienced fever, myalgias, and
          shortness of breath beginning 28 days post-Dose 2 and was hospitalized one day later with
          abnormal chest CT showing a small left-sided consolidation. He was discharged from the
          hospital 2 days later, and multiple nasopharyngeal PCR tests collected over a 10-day period
          beginning 2 days after symptom onset were negative. It is possible, though highly unlikely, that
          this event represents a COVID-19 case with multiple false negative tests that occurred more
          than 7 days after completion of the vaccination regimen, and more likely that it represents an
          unrelated infectious process.

          Among 3410 total cases of suspected but unconfirmed COVID-19 in the overall study
          population, 1594 occurred in the vaccine group vs. 1816 in the placebo group. Suspected
          COVID-19 cases that occurred within 7 days after any vaccination were 409 in the vaccine
          group vs. 287 in the placebo group. It is possible that the imbalance in suspected COVID-19
          cases occurring in the 7 days postvaccination represents vaccine reactogenicity with symptoms
          that overlap with those of COVID-19. Overall though, these data do not raise a concern that
          protocol-specified reporting of suspected, but unconfirmed COVID-19 cases could have masked
          clinically significant adverse events that would not have otherwise been detected.

          https://www.fda.gov/media/144245/download

          The odds of testing positive by interval after vaccination for BNT162b2 compared to those unvaccinated was initially analysed for the full period since the roll-out of the BNT162b2 vaccination programme on 8th December 2020 (supplementary table 2, supplementary figure 2). During the first few days after vaccination (before an immune response would be anticipated), vaccinated individuals had a higher odds of testing positive, suggesting that vaccination was being targeted at those at higher risk of disease.

          […]

          The odds of testing positive among vaccinated individuals increased during the early period up to days 7-9, reaching 1.48 (95%CI 1.23-1.77). The odds ratios then began to decrease from 10-13 days after vaccination, reaching 0.41 (95% CI 0.32-0.54) in the 28-34 day and remaining at a similar level from 35 days onwards.

          https://www.medrxiv.org/content/10.1101/2021.03.01.21252652v1

          Why aren’t people being told to act as if they are immunosuppressed during that first week?

        • “They should not leave out that in the first week all evidence and reason (science) now says you are *more susceptible*.”

          Which means diddly over the long term.

          People are warned to continue precautions until two weeks after their second vaccine, and the fact that there might be an increased risk of infection for the first week after doesn’t change that.

          The Denmark and Israel papers show very high overall efficacy, closely matching the trial results for Pfizer.

        • Which means diddly over the long term.

          It means people are getting sick and dying in the short term. The long term consequence is that they are near perfect incubators for new strains.

          People are warned to continue precautions until two weeks after their second vaccine, and the fact that there might be an increased risk of infection for the first week after doesn’t change that.

          Immunocompromised people need to take extra precautions. People should be told to be extra careful during that first week (even take off work, etc). Basically act like you tested positive for covid.

          So once again we have a simple, safe method of saving lives that is not being used.

        • I doubt incubating new strains will matter at this point, given the vaccination rates and the time it would take for a new strain to become common. The window probably closed a month or two ago for anything to become common before the end of the pandemic.

          (A true vaccine escape strain – not just somewhat-reduced neutralization capacity as in the B.1.351, but actually immunologically naive again, would change that, but I am *extremely* skeptical of that being a thing that can plausibly happen, as it doesn’t seem to happen with other respiratory viruses.)

        • > I tend to agree w @CT_Bergstrom that virus unlikely to run out of evolutionary “space” for antigenic change. This doesn’t happen much for flu or CoV-229E, so doubt it will for SARS-CoV-2. That said, two things that could somewhat slow antigenic evolution in future…

          https://mobile.twitter.com/jbloom_lab/status/1369080278312886273

          > We see high neutralizing activity against all strains tested, including P.1 that is spreading so rapidly in Brazil. While activity is diminished against the B.1.351 strain from South Africa, it is still very impressive—and higher than mean activity against wild type after 1 dose.

          https://mobile.twitter.com/CT_Bergstrom/status/1369123248470827016

        • “It means people are getting sick and dying in the short term.”

          I didn’t see anything about clinical outcomes for those who got a detectable infection during the first two weeks in the Danish study.

          Did I miss that?

        • “(A true vaccine escape strain – not just somewhat-reduced neutralization capacity as in the B.1.351, but actually immunologically naive again, would change that, but I am *extremely* skeptical of that being a thing that can plausibly happen, as it doesn’t seem to happen with other respiratory viruses.)”

          The premise is that the lower white blood cell counts make it easier for people to become infected before the antibody response builds to any reasonable level.

          So it’s hard to see why an antibody-escaping strain would be selected for during that short period of time in the first place. Sure, such a mutation might happen, but unless it was also more efficient at reproducing in an environment where antibodies are absent there’s no reason for it to reproduce in large numbers. If it were less efficient the “normal” virus would out reproduce it anyway.

          And … there’s really no difference in the lack of selection pressure compared to a situation where the rate of infection over the first week or so isn’t increased. A relatively small number of additional people falling ill in an antibody-free environment.

          All this on top of your skepticism.

        • I didn’t see anything about clinical outcomes for those who got a detectable infection during the first two weeks in the Danish study.

          Did I miss that?

          Seems like something worth reporting on.

          So it’s hard to see why an antibody-escaping strain would be selected for during that short period of time in the first place.

          Might be hard for people who don’t use science to see, but that is what happens in immunosuppressed people. It is already happening:
          https://www.medrxiv.org/content/10.1101/2021.02.23.21252268v1

          This really feels like Feb 2020 all over again in every way. But as I said already, there is no way a no variant can be worse than the early intubation, OD-ing on HCQ, and sending infected into nursing homes. So it won’t be that bad. I hope the news just ignores it so we don’t get a repeat of that.

        • > Whatever SARS-CoV-2 variants eventually summit that peak could be a considerably bigger problem for us than any variants that we currently know in that they might have any combinations of increased transmissibility, altered virulence and/or increased capacity to escape population immunity.

          I’m going to try to unread that.

        • “Might be hard for people who don’t use science to see, but that is what happens in immunosuppressed people. It is already happening”

          One of us doesn’t understand how the word “hypothesis” is used in science. As found in the term “chronic-illness-emergence-hypothesis”. The discussion in the paper makes clear the fact that they are proposing this as a hypothesis, nothing more.

          And the hypothesis was developed in regard to chronic covid illness caused by the inability of the immunocompromised individual’s body not being able to clear the virus for months, not a week long suppression of white blood cells followed by a rapid increase in antibody levels, as is seen in those who get their first shot of an mRNA vaccine.

          This presents a very different host environment for the virus than they are discussing, and there is no basis for you to extrapolate their hypothesis to this very different host environment. The timescale is vastly different, the drop in white blood cells is transient, most will have healthy immune systems without even considering the immune response to the vaccine.

          As they say:

          “By allowing the accumulation of larger combinations of epistatically interacting mutations and facilitating better exploration of local fitness landscapes within individual patients, chronic infections could have enabled the serendipitous discovery of new peaks on the global fitness landscape”

          Chronic infections. Chronic. Their hypothesis simply doesn’t apply to the physiological response to the first shot of an mRNA vaccine.

          Go back and read the paper more closely.

          Just between you and me, I actually read that paper when it first showed up …

        • Just between you and me, I actually read that paper when it first showed up …

          Look at figure 2C. What method did they use to choose the date for that big red line with a gap in mutations until mid-December? Think critically.

        • >>> I tend to agree w @CT_Bergstrom that virus unlikely to run out of evolutionary “space” for antigenic change. This doesn’t happen much for flu

          I am not sure they are talking about exactly the same thing, because IIRC the flu pandemics recent enough to have known origins come from *zoonotic* events, not from mutation within humans.

          I would like to see evidence of any disease, anywhere, circulating within humans mutating to cause a new pandemic/epidemic (as opposed to starting as zoonoses, or contact between two previously separated human populations e.g. smallpox in the New World) before I start worrying about that.

        • confused –

          > I am not sure they are talking about exactly the same thing,

          Yes, but I thought it was related, and interesting.

        • >>Yes, but I thought it was related, and interesting.

          Oh, it is, sorry; I thought that was a direct counterargument against my idea that new variants (as opposed to B117, and maybe B.1.351, which had significant time to spread before vaccination was common) won’t make much of a difference.

        • Anon,

          It’s just a function of both groups having high chance of getting infected (HC workers and nursing home).

          There’s nothing special about the vaccine making you more susceptible to infection, compared to not getting vaccinated.

          A lot of those studies are just documenting noise. Remember that in the original trials that ‘magic’ 95% efficacy is derived from 8 vs. 162 infections in vaccine vs. placebo arm.

          Vast majority (heck, almost all) of the participants in both arms didn’t get covid.

          The assumption that participants in both arms somehow had the same chance of exposure for weeks on end is the mother of all assumptions.

          What matters is that immunity takes time to build and one dose might as well be enough given enough time.

          All this dose-response stuff in medicine is quite arbitrary. Various intermediate points never get tested.

          I always get half a dose of Advil if I need, and it works.

          Take it from former big pharma guy:

          https://www.youtube.com/watch?v=lXK2j1Qxb4U

        • There’s nothing special about the vaccine making you more susceptible to infection, compared to not getting vaccinated.

          Yes, there is:

          Healthy adults 18 to 55 years of age or 65 to 85 years of age were eligible for inclusion.
          […]
          The largest changes from baseline in laboratory values were transient decreases in lymphocyte counts, which resolved within 1 week after vaccination (Fig. S3) and which were not associated with clinical manifestations.

          https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7583697/

          I guess now lymphocytopenia does not lead to increased chance of infection? Even though now we have seen increased infection rates multiple times including in the RCT?

          You can’t just ignore decades of science because of covid.

        • “Look at figure 2C. What method did they use to choose the date for that big red line with a gap in mutations until mid-December? Think critically.”

          In other words you can’t defend your claim that their hypothesis is fact, and despite being based on chronic infection it is equally factual in a host environment where immunosuppression consists simply of lower white blood cell counts and only lasts about a week.

          I’d expect someone whose vision is based on science to be above a simple Gish Gallop.

        • I was just checking whether you could go deeper than parroting back what they wrote. I’ve been in many journal clubs filled with that type of shallow thinking mostly driven by the need to “get back to work”. This breeds an anti-scientific authoritarian mode of thinking. Nullius in verba.

          You read the paper but did not understand it.

        • For completeness sake, what’s the counterfactual?

          Someone who sees conspiracies everywhere?

          That the standard behavior I saw all the time when doing medical research continues to be standard. Reviewers telling you to leave out inconvenient data, excuses to include/exclude various points, once a paper is published the interpretation is accepted to be correct by default, p-hacking “because I need to survive”, etc.

          I like how using science and thinking critically is now a “conspiracy theory” though. Western civilization has a bright future if this continues.

          I mean look at the Denmark paper. It is right in your face. Why do they not mention that result in the text? Did they not see it in their own table?

        • Anoneuoid –

          > Reviewers telling you to leave out inconvenient data,…

          OK. Just one more (Andrew, I promise).

          A lot of these are preprints.

        • “I was just checking whether you could go deeper than parroting back what they wrote.

          You read the paper but did not understand it.”

          What you are really saying is that they wrote the paper and didn’t understand it …

          You do that constantly with the papers you cite.

  19. Look again at this:

    Simply look at “0-14 days after first dose” in table 2 from the paper: https://www.medrxiv.org/content/10.1101/2021.03.08.21252200v1

    They saw a statistically significant 2x increase in infections during that first week for healthcare workers and 40% increase for nursing home residents, yet the authors do not mention it in the text. Can you apply critical thinking to offer an explanation for why that would be?

    • Because people would take it out of context to make the vaccine sound dangerous, and because it’s unimportant if people maintain measures until 2 weeks post-second-dose as recommended?

      I just got the first dose today; even if my risk is doubled for the next few days, the fact that I’ve successfully avoided it for a year implies that my base-rate risk of getting it in the next few days is not very high, so doubling it is not that much.

      (And do we really know this is not at least partially due to reduced carefulness?)

      • IE… double risk of getting it for one week, then vastly reduced risk for the next year (and maybe reduced risk of severity forever, T-cells may persist for ages as they did for SARS-1), is still an incredibly clear win. Total risk is still vastly reduced.

        • And the rate of vaccination is increasing as the rate of new infections is steadily decreasing.

          Where I live the new case rate is lower than it’s been since May. About 30 cases per day in a county of 434,000. Obviously there are more cases than those that are confirmed, but still, the risk of exposure is low during that first week after shot one even if one does nothing to protect themselves. And those who get vaccinated are those most likely to be protecting themselves, so seriously, the risk is low.

          Then we have the problem that he’s conflating a host environment where there’s a possible short-term decrease in protection followed by a steep increase in antibody production with a host environment associated with chronic covid caused by long-term immunosuppression. He is claiming that they are evolutionarily equivalent. And worse, he takes a proposed hypothesis as being an established fact.

        • Yes, exactly. The decline may be slowing, but the numbers in the US over the last six weeks have really been better than I hoped – two months ago I’d expected the B117 variant to be having much more of an impact by now (nearly two weeks into March) than it seems to have had.

      • Because people would take it out of context to make the vaccine sound dangerous, and because it’s unimportant if people maintain measures until 2 weeks post-second-dose as recommended?

        Yes, one possibility is the scientific literature is being censored for the “greater good.” Who is making the decision about what it is ok to publish now though? I don’t remember ever hearing about such a person/committee being appointed to this task. How can we know whether there is a conflict of interest here?

        • Anoneuoid –

          > Yes, one possibility is the scientific literature is being censored for the “greater good.”

          Ive stayed out, but you persist in doing this over and over. I’ll make this one comment here.

          This is arguing from personal incredulity. It’s considered a fallacy for a reason.

          You’re obviously a very smart and knowledgeable person, but there are many reasons that things aren’t the way that you think they should be other than malice and conspiracies and laziness and incompetence and coverups.

          You’re constantly saying, “What other explanation could there be other than a conspiracy to lie to the public and cause many unnecessary deaths?” And then you supply a reason (usually under a thin blanket of plausible deniability) as if you actually things you can’t possibly know. How many medical researchers have you accused in just the last week of deliberately producing misleading research on vaccinees, or failing to consider obvious problems because they didn’t “like” them, or other such attributions of motivation to people you don’t even know?

          You have to consider the plausibility that so many people would be sociopaths.

        • Why are you responding to me rather than confused? They are the one who supposed this info was left out on purpose.

        • It wasn’t left out – the numbers are after all there. Your complaint seems to be that *special attention wasn’t called to it* in the text – but this is only warranted if it is a) much more significant than I think it actually is and b) unlikely to increase vaccine hesitancy (and I think the latter is already disproved by the fact you are pointing to it…)

        • Mentioning your statistically significant result and discussing possible interpretations is normal attention in the medical literature, it is not special.

          Failure to do so is special.

        • Anyway, it now seems simply accurately describing your data is now a bad thing. And all it indicates is to tell people to be more careful in the first week after the first dose. It doesn’t mean do not get vaccinated.

          If you really care about saving lives you would just tell people to be more careful when they are immunosuppressed. The fact this is even an issue is highly disturbing.

        • It may just have been considered unimportant – my “avoiding vaccine hesitancy” line is pure speculation.

          However…

          >>If you really care about saving lives you would just tell people to be more careful when they are immunosuppressed.

          I disagree. The effect (if really due to immunosuppression) is transient and within a period when the vaccine is not yet expected to be “working” and measures are still recommended, therefore not very important, so it is very plausible that the harm done by giving anti-vaccine types something to point to is greater than the benefit of pointing it out.

          If anything, the public health messaging has IMO been somewhat problematic in this respect e.g. with regard to variants and post-vaccination behavior recommendations – the vaccine’s efficacy has been consistently “understated”, which is not good for uptake.

          (Similarly, I think there ought to have been a stronger emphasis on the fundamental advantages of the mRNA technology, to counter claims that it was ‘rushed’.)

        • measures are still recommended

          Different measures are recommended based on whether you are currently immunosuppressed or not. This is for good reasons that everyone understands except when it comes to these vaccines apparently.

        • Sure, but does this effect really rise to the level of “immunosuppression” as that term is usually meant?

          How does one untangle what part of the increase is genuinely biological/immune system based and what part is reduction of caution before the vaccine’s benefit kicks in?

        • Well, the text actually says “Short-lived decreases in postvaccination lymphocyte counts had no associated clinical effect, were observed across the age groups, and probably reflect a temporary redistribution of lymphocytes from the bloodstream to lymphoid tissues as a functional response to immune stimulation by the vaccine”

          This implies that the lymphocytes are not really gone, just redirected.

          Is this, in 25% of people, really enough to say that someone post-vaccination can be considered immunosuppressed?

        • Is this, in 25% of people, really enough to say that someone post-vaccination can be considered immunosuppressed?

          Nope, but it deserved further monitoring but was never published again. And now we see over and over more infections recently after vaccination. And this is just of sars-2, what about everything else? There is a total lack of data on that.

          So if you want to save lives, you want to investigate this issue. People who don’t want to do that, well…

        • >>Nope, but it deserved further monitoring

          Maybe, but apparently the authors of that paper didn’t think so.

          But a couple days ago on here you were arguing *censorship* or *suppression* of the information. That didn’t happen. The data is presented, it is discussed, and the conclusion is “this isn’t a problem”.

          That conclusion could be wrong, but that is *not* the same thing as the information being suppressed!

          >>So if you want to save lives, you want to investigate this issue.

          How can you rule out the possibility that it was investigated, but the answer was “yeah, it’s no problem, just as the original paper said”? How publishable would that kind of “yeah, they were right” observation be?

        • > So if you want to save lives, you want to investigate this issue. People who don’t want to do that….

          Yeah. All those people who don’t want to save lives. Really a problem.

        • How can you rule out the possibility that it was investigated, but the answer was “yeah, it’s no problem, just as the original paper said”? How publishable would that kind of “yeah, they were right” observation be?

          There is every incentive to publish no problem and every anti-incentive to publish the opposite. Or else this would be mentioned in the paper.

          This anti-science approach is not helping anyone.

        • >> Or else this would be mentioned in the paper.

          You seem to be simply assuming that the problem is in fact important, therefore greater attention not being called to it must be a sign that something is wrong re: the publication process. (Rather than it being an insignificant issue which receives little attention since little is warranted.)

          I don’t know your knowledge base; maybe you do know better than the paper’s authors. But to me (or likely any other random reader of this comments section) that doesn’t seem the more likely possibility.

          I see no point in discussing this any further, however.

        • How many medical researchers have you accused in just the last week of deliberately producing misleading research on vaccinees, or failing to consider obvious problems because they didn’t “like” them, or other such attributions of motivation to people you don’t even know?

          To be clear, I quit medical research because this type of behaviour was standard. Or even enforced, it is very difficult to *not* produce misleading research given the statistical significance culture. They are just applying what they do on everything else to vaccines.

        • I am not suggesting censorship of any sort. (The data was in fact published, or you wouldn’t have it to point to.)

          I am suggesting that authors (not censors) saw no reason to call special attention to that data (in the text) as it would be more likely to cause harm than benefit, but published the numbers themselves (as good scientists should).

        • I am suggesting that authors (not censors) saw no reason to call special attention to that data (in the text) as it would be more likely to cause harm than benefit

          Then the authors are the censors. If you are right, we can expect more of this type of selective reporting in the future.

        • I don’t think this is even unusually selective, much less censorship. Describing numerical data in text inherently involves decisions (selection, if you will) about what to emphasize. The difference here IMO is that you think this data is particularly significant and the authors did not.

          I’m not even sure it was necessarily to avoid vaccine hesitancy, although I think that’s very plausible. It might just have been considered too transient an effect to matter much (especially as the vaccine isn’t said to “work” until 2 weeks after second dose or whatever).

        • I don’t think this is even unusually selective

          I have read tens of thousands of medical research papers. Ignoring a statistically significant result is very unusual and would typically be considered a big red flag.

        • Another thing you have repeated multiple times (maybe it was in a different thread) is that the UK strain is spreading faster despite little drop in neutralizing activity. That is true, however still misleading.

          The antibodies and receptor compete for binding to the virus. So this needs to be considered relative to ACE2 binding, which is 2x higher for that strain:
          https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7924271/

          In general you can’t use affinity as a proxy for neutralizing activity, so it is difficult to say quantitatively what effect this would have. But ceteris paribus if you have the same antibody interaction but 2x higher ACE2 then this variant will be selected and spread more just as if there was a drop in antibody affinity.

  20. Anoneuoid –

    This guy seems to be in line with your thinking:

    https://youtu.be/YtHfI00D_s4

    What I don’t understand is why he thinks that the pressure for mutation under partial immunity is higher with vaccines than (potentially vastly more) asymptomatic infection w/ no vaccine (particularly if, indeed as evidence is now showing, the vaccines reduce as the level of asymptomatic spread).

    Wouldn’t high rates asymptomatic infection mutating all over the place lead to a high probability of a “super strain” irrespective of a widespread partial immunity with vaccines? (I realize that the partial immunity with vaccines might be more dangerous in the sense that it presents more of a “threat” to the virus, but still, given the # of infections there are already and the high % of asymptomatic infection…. doesn’t asymptomatic infection always imply a partial immunity?)

    Of course, his argument seems to me a bit like that we are worse off for the development of antibiotics because of the potential for antibiotics to result in super duper bacteria.

    • >>What I don’t understand is why he thinks that the pressure for mutation

      Well, it doesn’t really work that way in the first place! Selective pressure doesn’t *cause* mutations, it just changes which mutations persist and become predominant.

      Mutations and selective pressure are two different forces, and may pull in opposite directions.

      More infections mean more mutations; vaccination means fewer infections, thus fewer mutations, but possibly more selective pressure. (But only possibly – if natural immunity is about as good as vaccine immunity then it probably makes no real difference. I don’t think reinfection rates are high enough to suggest that this will actually happen in practice.)

      So, it’s not a completely crazy idea, but I see no reason to worry about it in reality.

      >>Wouldn’t high rates asymptomatic infection mutating all over the place lead to a high probability of a “super strain”

      Well, I’d say no, because I don’t believe there is any scenario that would give a *high* probability of a “super strain”, if by that we mean something qualitatively worse than the existing variants, e.g. true immune escape [not just reduced neutralization, but “prior infection/vaccination essentially makes no difference”].

      Such things just don’t seem to happen in practice. I don’t doubt that B117 for example is genuinely more contagious – but I do doubt that it makes terribly much difference long-term; even in an environment without vaccines, you’d hit whatever final % infected faster and then the pandemic would end.

    • confused –

      > Well, it doesn’t really work that way in the first place! Selective pressure doesn’t *cause* mutations, it just changes which mutations persist and become predominant.

      Sure. That’s why I put “threat” in quotes below.

      > More infections mean more mutations; vaccination means fewer infections

      Not sure if you watched any part of the clip. What I don’t get is his logic in assuming that vaccinations won’t decrease the number of infections. Originally people were very hesitant about making any such projection but the data seem to be showing that they markedly reduce even asymptomatic indecftions.

      > I don’t believe there is any scenario that would give a *high* probability of a “super strain”,

      Sure. I meant relatively high.

      > but I do doubt that it makes terribly much difference long-term

      Sure. But increased virulence is also a possible outcome of mutations – which would obviously make a difference.

      • >>What I don’t get is his logic in assuming that vaccinations won’t decrease the number of infections

        This is simply wrong.

        >>Originally people were very hesitant about making any such projection but the data seem to be showing that they markedly reduce even asymptomatic

        Yep. (I think even the original hesitation was irrational – we have plenty of other vaccines to extrapolate from, “it hasn’t actually been demonstrated yet” does not mean “there is no way to know”.)

        >>But increased virulence is also a possible outcome of mutations – which would obviously make a difference.

        Yeah. I just think that the in-practice limitations on this are probably a lot stronger than we might think. I am not a virologist, but the historical evidence (essentially all major new outbreaks/pandemics are zoonoses or previously-disconnected human populations coming into contact, e.g. smallpox in America, measles in Polynesia) IMO pretty strongly suggests that viruses in a single human population mutating toward greater virulence doesn’t generally happen.

        Probably they hit a “local maximum” on the “fitness landscape” pretty quick.

        • So is there a contrasting situation? Ie under what situation would we expect / know viruses to mutate towards higher virulance?

        • Well (and again, I am not an expert on viruses, but in general evolutionary terms) a species in a new / radically changed environment is going to evolve much faster.

          A species already well adapted to its existing environment – if it is already well adapted, changes will tend to reduce fitness, therefore be selected against (“stabilizing selection”).

        • Ie under what situation would we expect / know viruses to mutate towards higher virulance?

          Leaky vaccines:

          There is a theoretical expectation that some types of vaccines could prompt the evolution of more virulent (“hotter”) pathogens. This idea follows from the notion that natural selection removes pathogen strains that are so “hot” that they kill their hosts and, therefore, themselves. Vaccines that let the hosts survive but do not prevent the spread of the pathogen relax this selection, allowing the evolution of hotter pathogens to occur. This type of vaccine is often called a leaky vaccine.

          https://journals.plos.org/plosbiology/article/info:doi/10.1371/journal.pbio.1002198

        • This is theoretically possible in some situations, but sounds irrelevant for COVID, because:

          >>This idea follows from the notion that natural selection removes pathogen strains that are so “hot” that they kill their hosts and, therefore, themselves

          COVID mortality is not high enough to seriously limit its spread; the overall IFR is almost certainly below 1%, and the highest mortality is found among people who are not “out and about” a lot (long-term-care facility residents) while actively socializing young adults are at *much* lower risk than the population-average IFR.

          (This scenario would also seem to require that that deadly-enough-to-kill-itself-off pathogen was nonetheless common enough to have a vaccine developed against it and distributed. Given that vaccine trials require the disease to actually be present in the population, this would seem to be an incredibly narrow window.)

        • I disagree.

          >>natural selection removes pathogen strains that are so “hot” that they kill their hosts and, therefore, themselves.

          COVID is not such a strain (it sometimes kills hosts, but not often enough to meaningfully limit its spread, especially as those at highest risk of death are not the most social) therefore this will not happen with COVID, though it might happen with some other hypothetical disease.

        • Oh, I think I see the distinction: that if some ultra-deadly COVID strain (not COVID as we now know it) evolved it might be suppressed by natural selection in the absence of vaccines, but not in the presence of them.

          I mean, maybe in theory, but:

          – natural immunity is probably fairly similar (maybe not quite as good, but still…), and it won’t be long until most people on Earth have either natural or vaccine-induced immunity; vaccines don’t really change the overall immunological picture in the grand scheme of things, they just get us there with a lot fewer deaths;

          – this doesn’t happen with other diseases we vaccinate against;

          – more broadly, evolution of “super-strains” of existing diseases within a human population (as opposed to zoonotic cross-species jumps, which is where new deadly diseases actually do come from) doesn’t seem to happen.

          I’m not saying it’s *impossible*… but I would imagine it’s a much smaller risk than an equivalent disease originating separately, zoonotically (given that we have many examples of that and zero of this… I am not convinced domestic chickens are sufficiently analogous to humans for this paper’s example to count as demonstration…)

        • @Anoneuid

          No one yet alerted me to the risk that an errant extra terristrial object may smash my house.

          So what? There’s a ton of low risk tail events that we all face.

          It’s no reason not to build houses!

        • I’ve heard of such things speculatively before you mentioned them in this comment thread, yes; I never took them at all seriously.

          And I still don’t – I remain unconvinced that the risk is real (=that this can actually happen under the conditions actually occurring in a human population with a virus of this type; I don’t doubt that it’s *theoretically* possible, only that the conditions required for it to occur are realistic).

          And I am absolutely convinced that the risk is too uncertain, and too unlikely even if real (given that no such thing has ever been observed in humans, despite tons of large-scale vaccination programs) to be relevant compared to the enormous benefits/prevented risks of vaccination.

        • No one yet alerted me to the risk that an errant extra terristrial object may smash my house.

          So what? There’s a ton of low risk tail events that we all face.

          It’s no reason not to build houses!

          Point is the benefits are being constantly overstated (you can already see the goal posts move from herd immunity to less severe symptoms), and risks understated. This misinforms people and leads to non-optimal decisions.

          Here is another thing still going ignored, see supplementary figure 1:

          https://www.nejm.org/doi/full/10.1056/NEJMc2102179

          No mention that the neutralization went negative in in 6/8 people vaccinated with the moderna vaccine when tested against the South Africa (B.1.351) variant. They just stop the y-axis at zero so we can’t see the magnitude.

          Between trying to not report on what happens weeks 1-2 after first dose (when antibodies are weak and people are immunosuppressed) and what will happen when antibodies later wane (also when antibodies are weak), a huge risk is being ignored.

          This is not how scientists behave.

        • > you can already see the goal posts move from herd immunity to less severe symptoms

          Classic bad faith characterization.

        • 18 March 2021:

          As COVID-19 vaccination rates pick up around the world, people have reasonably begun to ask: how much longer will this pandemic last? It’s an issue surrounded with uncertainties. But the once-popular idea that enough people will eventually gain immunity to SARS-CoV-2 to block most transmission — a ‘herd-immunity threshold’ — is starting to look unlikely.

          https://www.nature.com/articles/d41586-021-00728-2

        • And much of the reasoning in your cite has to do with human behavior, not the effectiveness of the vaccines themselves.

          The article mentions other things. Gu misunderstood “herd immunity” to lead to eradication, rather than the disease becoming endemic. I’ve been following him for months. I can’t think of any epidemiologist claiming that we’d eradicate covid-19. Driving it to an endemic state is the most optimistic expectation.

          Then we see things like this subhead:

          “Immunity might not last forever”

          No one has said it will. Not the vaccine manufacturers, for sure. If anyone might be accused of that, it would be folks like Ioannidis or Battachyara who confidently insisted that naturally acquired immunity would lead to and end to the epidemic in the US, at least, after no more than perhaps 30-40K dead.

          If no one has claimed it would last forever, how do statements saying immunity might not last forever represent a goalpost move?

          “New variants change the herd-immunity equation”

          This hardly qualifies as a goalpost move. It matches reality.

        • I think this is less “moving the goal posts” and more “people misusing the term ‘herd immunity’ so consistently that the real meaning has become horribly confused”.

          I think it’s been clear for a long time that true herd immunity, where R remains permanently below 1 until the virus disappears (as for smallpox, or polio or – before antivaccination became semi-popular – measles in the US) almost certainly won’t be achieved.

          But, what many if not most of the people *saying* “herd immunity” actually mean – the end of the pandemic – will be. That’s what you’d get if natural infection was allowed to run free — vaccination will just get us there with lots fewer deaths. And this is how past pandemics have ended – flu pandemics convert to seasonal flu viruses. The 1890 ‘Russian flu’ may have been the jump-to-humans of one of the circulating “common-cold” coronaviruses.

          (The FDA said 50% efficacy was enough to approve a vaccine. 50%, combined with less than 100% uptake, wouldn’t be enough to achieve herd immunity. So clearly that was never core to the plan, even if people have consistently misused the term.)

        • Anoneuoid –

          So now you double down with the “moving goal posts” bad faith accusation by conflating updating views based on emerging evidence with “moving goal posts?”

          Your logic often rests on an ability to mind-probe. To my knowledge no one has ever established such an ability. But maybe you can provide evidence otherwise?

        • @dhogaza

          >>Gu misunderstood “herd immunity” to lead to eradication, rather than the disease becoming endemic.

          Technically speaking, I think this is the correct meaning of herd immunity (as seen for polio in much of the world, measles in the US at least pre-anti-vaccinationism, and in smallpox eradication) – R remains below 1 until prevalence goes to zero, and when the disease is reintroduced from outside, new outbreaks are squelched by immunity.

          It is not, however, how past pandemics have ended (even 2009, when vaccination was introduced rather quickly – though still after the second wave had peaked). In those cases, the disease has gone endemic, yes.

        • “Technically speaking, I think this is the correct meaning of herd immunity ”

          Well, yes, over a long enough period of time in theory it should. I should’ve made clear that Gu’s epiphany about herd immunity was that it wasn’t going to eliminate covid in the US in anything like the timeframe he was modeling.

          As I understand it, at least, the herd immunity threshold is where Rt (or effective R) drops below one, and at that point it can take a very, very long time before you reach no cases. The time it takes depends on how far below one you drop …

        • So now you double down with the “moving goal posts” bad faith accusation by conflating updating views based on emerging evidence with “moving goal posts?”

          Everyone informed on the topic knew there was going to be waning, new resistant variants, some would spread to animals, and more severe reinfections. There was more chance of the sun failing to rise tomorrow than avoiding all that once “flatten the curve” was implemented.

          If you needed to wait for the evidence you weren’t using science.

          And the current policies are:

          1) Creating an immunity monoculture all towards the exact same spike protein sequence

          2) Not warning people to be careful when immunosuppressed after vaccination

          3) Delaying the second dose so immunity wanes faster

          We are going to be seeing much worse variants soon. But even if it is twice as bad as the original that still isn’t very bad. Especially as long as the heathcare system doesn’t mistreat the patients like last year. Best thing that can happen is the news starts ignoring it so the hysteria ends.

        • “Everyone informed on the topic knew there was going to be waning, new resistant variants, some would spread to animals, and more severe reinfections. There was more chance of the sun failing to rise tomorrow than avoiding all that once “flatten the curve” was implemented.”

          If the data’s correct, the most successful variant when it comes to reinfection is found in Brazil, where there has been very little effort to “flatten the curve”. Look at Manaus.

          Letting the epidemic run rampant wasn’t going to “avoid all that”, either.

          Meanwhile, those who successfully managed the epidemic seem to be doing just fine. Imagine where China, Australia, New Zealand, South Korea and other countries would be if they followed your preferred “let it run rampant until (temporary) herd immunity sets in” strategy.

        • your preferred “let it run rampant until (temporary) herd immunity sets in” strategy.

          Once again with this strawman. You have repeated this multiple times and ignore it when I correct you.

          Try to focus on saving lives by making decisions based off accurate information instead of coming up with strawmen.

        • It’s thought that WWI caused the Spanish Flu pandemic to mutate towards higher virulence. The mechanism was that those who had mild cases stayed in the trenches fighting and those who had more severe cases were brought to hospitals where they infected wounded soldiers and nurses and ambulance drivers etc. So armed conflict can be an example mechanism for increased virulence

        • I have read this too, and it sounds at-first-glance plausible, but I wonder how well established it actually is*, given the limitations of the data.

          I’ve also read that the Spanish flu was (unlike the seasonal flu) very deadly to young adults because death was often caused by the patient’s own immune system overreacting (and young people have stronger immune systems than older ones) – but that’s also true of COVID deaths, and COVID mortality very strongly increases with age.

          I wonder how different our picture of that pandemic would be if they’d had modern tools to study it.

          *IE, how certain is it that the virus itself was deadlier due to evolution in the (admittedly more deadly) third wave, vs. some difference in environmental conditions (seasonality?) or which human beings it was hitting?

      • Also, I didn’t mean to be simply pedantic in the “selective pressure doesn’t cause mutations” thing. I think it’s an important distinction – the virus doesn’t necessarily get the mutations it “needs” to “succeed” in a particular situation, if you will.

        And there may be practical constraints, as there generally are in evolution. If the vaccine protects against the Spike protein, and it needs the Spike protein to enter human cells – how much can it mutate the Spike protein before its being “less effective at doing its job” counteracts “better at evading the vaccine antibodies”?

        • > This is simply wrong.

          Yeah. That’s what I find confusing. Looks like he’s got tons o’ credentials. So my inclination if he seems to be saying something that seems obviously wrong is that there’s something I’m not understanding (or misinterpreting)…

          >Also, I didn’t mean to be simply pedantic

          No worries. I meant to be more careful about using language that didn’t imply any kind of teleological or orthogenetic trajectory but got careless.

        • >>So my inclination if he seems to be saying something that seems obviously wrong is that there’s something I’m not understanding (or misinterpreting)…

          Eh, I’ve seen quite a few ridiculously wrong (much more clearly wrong than this) things said about COVID by people with pretty good credentials…

          >> I meant to be more careful about using language that didn’t imply any kind of teleological or orthogenetic trajectory but got careless.

          Oh, my complaint isn’t about teleological language, it’s about constraints. You have to have both the right genetic “options” (mutations, in this case) and the right selective pressure to get the result.

        • confused –

          > Yep. (I think even the original hesitation was irrational – we have plenty of other vaccines to extrapolate from, “it hasn’t actually been demonstrated yet” does not mean “there is no way to know”.)

          My guess is that they are very aware of the downside of over-promising as a “fat tail” risk, with potential to undermine confidence in vaccines.

          I encountered someone online recently, a younger person w/o comorbidities, who said he didn’t want to get vaccinated because he was concerned it would not affect or even decrease his chances of having an asymptomatic infection, which would in turn increase his likelihood of passing on an infection to more vulnerable people. With the uncertainty about whether vaccines would reduce asymptomatic infections, at first I thought the rationale kind of made sense, and could actually be a reason why it might be questionable as to whether younger people should get vaxxed even if for only the reason the risk outweighed the benefit w/r/t probabilities and his INDIVIDUAL HEALTH (with no meaningful community benefit). But with the data that’s coming out in Israel, showing such a dramatic drop in even asymptomatic infections among the vaccinated…

        • >>My guess is that they are very aware of the downside of over-promising as a “fat tail” risk, with potential to undermine confidence in vaccines.

          Quite possibly. But I think there is at least as much risk the other way (sounding like they are not confident in the vaccines has its own issues with undermining) as it was IMO pretty clear even from a non-specifically-expert general-biology perspective what the expected result would be (of course it will reduce total infections*.

          I have a similar concern with the language being used about the variants re: vaccine escape. In general the vaccines are IMO being “undersold”.

          A lot of things being said about COVID have *greatly* reduced my confidence that public health authorities understand messaging/public relations.

          *Although there is a subtlety here depending on what you mean by “reduce asymptomatic infections”; if it prevents N infections but turns N infections that would otherwise be symptomatic into asymptomatic ones…

        • Joshua, Confused,

          I believe you are not correct regarding a few things. In one of the TWIV episodes just the other day Vincent mentioned that most vaccines actually do not prevent infections (no sterilization of the virus). Their job is to prevent bad outcomes and death. He’s one of the top virologists I believe. He also mentions that natural infection is more complete. Apparently it’s not just counting antibodies after vaccine, it’s their variety that natural infection gives you.
          Regarding the data from Israel, there are tons of fully vaccinated infectious people, far more than was observed in the original Phizer study. (0.004 vs. 6% in Israel data so far). So yes, those nervous Nellies that got vaccinated are now spreading it more efficiently than anti-maskers a few weeks ago (because of silent symptoms).

          However, I think that the Dutch expert makes a huge leap when comparing antibiotic resistance to viral resistance and then builds on that story. Two different animals. Bacteria don’t need hosts, viruses do, among other things. Besides, resistant bacteria and such show only in overuse of anti-biotics I believe. But he’s onto something still, I believe but only very specialized experts in a few fields could really give a good answer and refute his claims. Reality is that as of now there is a lack of long-term data.

          Personally, I think there is so much unknown in virology/immunology that our bodies will find the way to correct any issues vaccines may initiate. They do it on daily basis anyway, covid or not.

          Anyway, we’ll find out soon a lot more. There are ongoing Pfizer and Moderna trials and vaccines data are increasing rapidly.

        • >>most vaccines actually do not prevent infections

          I wonder if we are arguing about the same thing. I am talking about reduction of the number infected, not anything like 100% protection. This is certainly true for things like polio, measles, and smallpox, where vaccination has drastically reduced the circulation of the virus. Things like flu vaccines are vastly less effective, but that doesn’t mean 0% effect on infection rate.

        • Navigator –

          > So yes, those nervous Nellies that got vaccinated are now spreading it more efficiently than anti-maskers a few weeks ago (because of silent symptoms).

          I’m confused by what you’re saying there. Are you saying that – even with things like # of social contacts being equal – a cohort of vaccinated people would spread more virus than a cohort of un-vaccinated people? How do you square that with the evidence (e.g. from Israel) that vaccination significantly reduces infection (symptomatic and asymptomatic)? Seems to me a cohort less likely to get infected would be a cohort less likely to spread infection, even if it would have differentially more “silent spreaders.”

        • Also, as confused points to,

          > that most vaccines actually do not prevent infections (no sterilization of the virus).

          The point is that there’s evidence that he VOVID vaccines reduce rates of infection.

          > Regarding the data from Israel, there are tons of fully vaccinated infectious people…

          Again, it’s the differential rate that matters.

        • confused –

          > A lot of things being said about COVID have *greatly* reduced my confidence that public health authorities understand messaging/public relations.

          I dunno. My take away is that a lot of people have unreasonable standards for the communication of public health officials.

        • I don’t know, my standards may be high, but I don’t think they are *unreasonably* high for agencies publicly funded to deal with this stuff. If CDC doesn’t do very much useful in a crisis like this, what is it even *for*?

          Covid Tracking Project did a better job with presenting what was really going on with US data than CDC did until very recently.

          Things have improved greatly, but I don’t think we should just excuse the errors made in February-April because of that … WHO taking much too long to declare the outbreak a pandemic; CDC claiming masks don’t work to keep people from panic-buying them; the WHO ‘we don’t know whether antibodies do anything’ tweet back in April…

        • Oh, and probably worst of all, the FDA test-kit approvals issue that’s largely responsible for us not realizing how much the virus was already spreading in the US in February and the first week of March…

        • confused –

          It’s easy as an observer looking at sub-optimal results, after the fact, to then reverse engineer to find cauality in the failures of public health officials. We can always do that – it’s inevitable that there will be sub-optimal outcomes at some level.

          It’s like as they sayc: Public health officials will be found at fault for acting too slowly if there are bad outcomes and being alarmists if there aren’t bad outcomes.

          I come at this after years of watching people find fault with climate scientists for the high level of resistance in the US to mitigating aCO2 emissions. Supposedly, they are the ones responsible for “skepticism” because they’ve been too “alarmist” and too tribal (by using expressions like “deniers”). Or others even say they’ve been too passive and haven’t inspired people to take it seriously enough, or they used ineffective communication strategies that rely on the ineffective “information deficit model.”

          But while sure, sub-optimal outcomes have manifest, resistence to aCO2 emission is a politically/ideologicaly driven phenomenon (at least in the US). Maybe alternative communication strategies might have been marginally better, but they weren’t going to move the needle either way because the cauality around opinion formation isn’t a function of what climate scientists say or in their control in any meaningful sense (imo).

          Same with vaccine resistence or vaccine hesitancy or anti-vaxism or testing and tracing policies or reactions to NPIs or even mask-wearing guidance. Maybe if they had done everything perfect in their communication strategies there would have been marginal improvement – but I doubt the pandemic’s trajectory would have been substantively different.

          It might be a mistake to go from your reaction to communication strategies to thinking strategies you would have liked better would have had a very different outcome. It’s a mistake to try to extrapolate yourself – as in this domain, simply by virtue of being here and participating as you do, you’re an outlier.

        • >>It’s like as they say: Public health officials will be found at fault for acting too slowly if there are bad outcomes and being alarmists if there aren’t bad outcomes.

          Sure. This certainly applies to the level of “fear/alarm” implied in messaging, and to mandatory measures vs. not. But these are not really the things I am criticizing.

          My main criticisms (outside the specific FDA issue with COVID tests in February-March time period) are bad presentation of data and poor communication of uncertainties.

          The fact that it took the CDC something like 9-10 months to present data as good as a non-funded private organization did is IMO far less excusable than anything done on the measures side.

          >>I come at this after years of watching people find fault with climate scientists for the high level of resistance in the US to mitigating aCO2 emissions.

          Oh, I definitely don’t believe that.

          But I think there is a critical difference in that this is an issue which has built up slowly over a couple decades, and has been politicized the whole way, whereas no one had heard of COVID 3 months before it became the #1 global issue – there’s a “first impressions” aspect which IMO makes the situation very different from climate change.

          >> but I doubt the pandemic’s trajectory would have been substantively different.

          I generally agree with this, actually! I think the reasons the US had higher death rates than, say, Germany or Canada are more fundamental than anything COVID-specific. I’m thinking more about loss of whatever public confidence previously existed, and maybe erosion of confidence in science in general, which is worrying in the long term.

        • confused –

          > My main criticisms (outside the specific FDA issue with COVID tests in February-March time period) are bad presentation of data and poor communication of uncertainties.

          But again, your reaction and the more general public reaction aren’t the same thing and the translation between your reaction to how data were presented and how uncertainties were communicated doesn’t explain how the public reacted to the policies.

          I’m not defending what they did against your criticisms.

          > But I think there is a critical difference in that this is an issue which has built up slowly over a couple decades, and has been politicized the whole way, whereas no one had heard of COVID 3 months before it became the #1 global issue…

          But the political framework in how this has played out is almost identical. Go to any climate “skeptic” website and you will find a striking alignment on COVID. You could basically just swap out the one issue for the other.

          > – there’s a “first impressions” aspect which IMO makes the situation very different from climate change.

          Climate “skeptics” almost uniformly say that their impressions led to their views on clinate change in pretty much the same way that COVID “skeptics” describe their opinion formation. They say they lost confidence in what scientists were saying because of the flaws in how the material was being presented – as if the political alignment is just purely by coincidence.

          > I think the reasons the US had higher death rates than, say, Germany or Canada are more fundamental than anything COVID-specific.

          And i agree there.

          > I’m thinking more about loss of whatever public confidence previously existed,

          The stage was set for that to happen by political/ideological structures.

          > and maybe erosion of confidence in science in general, which is worrying in the long term.

          That to us part of a long-standing trend, where “confidence in science” has been generally pretty stable for a while but dropping pretty notably among Tea Party types and the religious right. It was fairly predictable that trend line would run right through COVID – especially with the specific focus of Trump on exploiting support from that group.

        • >>But again, your reaction and the more general public reaction aren’t the same thing

          Absolutely – but I think there have been a number of objectively misleading statements and cases of objectively poor data presentation, not just my own reaction.

          >>doesn’t explain how the public reacted to the policies.

          I’m not mostly talking about policies or reaction to them, though, but about giving an accurate picture of the real state of the pandemic in the US (within the existing data and its uncertainties).

          >>But the political framework in how this has played out is almost identical.

          True! But I think there was potentially a narrow window in which that *didn’t have to happen*. This would have been difficult, yes, especially given who was President at the time. But I think it was predictable as a potential problem, and worth attempting to avoid.

        • Sure, 2020 was already going to be a highly politically polarized environment, because of Trump and an election year, etc.

          I also hold CDC etc. to higher standards because CDC’s job isn’t just the science, dealing with the public is core to their role and responsibility. I don’t blame academic scientists for not communicating uncertainty in the way the general public can understand — but an agency like CDC should be able to do that “translation”.

  21. More or less relevant to the uncertainty and covid theme, today is the anniversary of the the “A fiasco in the making?” article.

    “Some worry that the 68 deaths from Covid-19 in the U.S. as of March 16 will increase exponentially to 680, 6,800, 68,000, 680,000 … along with similar catastrophic patterns around the globe. Is that a realistic scenario, or bad science fiction?”

    “If we assume that case fatality rate among individuals infected by SARS-CoV-2 is 0.3% in the general population — a mid-range guess from my Diamond Princess analysis — and that 1% of the U.S. population gets infected (about 3.3 million people), this would translate to about 10,000 deaths. This sounds like a huge number, but it is buried within the noise of the estimate of deaths from “influenza-like illness.”

      • Given that about 20% of the passengers and crew of the Diamond Princess became ill with covid-19 it could not have been through extrapolating that data.

        And given that 2% of those that did become ill it is hard to see where his 0.3% CFR estimate comes from, too.

        • The 0.3% IFR (not CFR) was based on correcting for the average age of the Diamond Princess patients. It turned out to be wrong — I think that’s because of the failure to account for unresolved cases (those who died after this was published).

          IIRC the number of deaths on the ship roughly doubled, so the 0.3% would become 0.6%, which is probably about right (CDC best estimate used to be 0.65%, and CDC’s total infections through December estimate of 83 million depending on which reported deaths are due to infections in December and before).

          1%, however, is pretty crazy. I think flu pandemics tend to be somewhere around 20%-40% (2009 H1N1 is estimated at about 20% in the US, CDC’s estimate of 500 million worldwide for 1918 H1N1 is about 30% of the world population at the time).

          I suppose one could argue that everywhere that had seen serious COVID problems in mid-March had extremely dense populations by US standards, so it was still maybe plausible that it wouldn’t spread effectively through much of the US, but 1% would be ridiculous even then (too much of the US is in fairly dense metro areas like the Northeast megalopolis, LA, Chicago, etc.)

        • Oops – that should say “CDC’s total infections through December estimate of 83 million *probably implies something between 0.5% and 0.6%*, depending on which reported deaths are due to infections in December and before.”

          (CDC’s provisional deaths page has ~378k COVID deaths for 2020, and by mid-March that’s probably fairly complete. That would suggest below 0.5% — but January had a ton of deaths, and those would almost all have been December infections. If we assume ~3000/day for January, or ~90k more deaths, ~480k deaths would mean ~0.58%).

        • He wrote “case fatality rate” in his piece. I suppose you could argue that he probably believed that all those infected had been detected in which case IFR and CFR are the same. 0.3% for IFR is only off by about a factor of two, as you say.

          The 1% figure for how much of the US would be come infected, especially considering that he was arguing for not taking drastic measures, does indeed appear to be pulled out of his rear.

        • dhogaza –

          > The 1% figure for how much of the US would be come infected, especially considering that he was arguing for not taking drastic measures, does indeed appear to be pulled out of his rear.

          Not that I’m particularly inclined to defend Ioannidis…but I’ve read it argued that there’s wiggle room in that he was only using the 1% as an unlikely hypothetical example and not to suggest it would be that low…

          Later in the article he does say this:

          > In the most pessimistic scenario, which I do not espouse, if the new coronavirus infects 60% of the global population and 1% of the infected people die, that will translate into more than 40 million deaths globally, matching the 1918 influenza pandemic.

        • Yes, the 1% was clearly meant as a low end, and he did give a high-end too.

          But I think 1% was not very believable even as a low end. If he’d said 5% or 10%, that I could argue as a halfway-believable low end given that we didn’t know how much of COVID spread was “superspreader” driven; if COVID had turned out to only be able to spread effectively in high-density Old-World-like cities, maybe only 5% to 10% (1/4 to 1/2 of maybe 20% of the US population) would have been infected.

          But 1% just seems ridiculous given that Italy etc. had already disproven a “Ebola/SARS-like” model. I think there’s not much in between a very limited spread outside the source region (e.g. Ebola, SARS) and going global.

        • Well, 60% was much more defensible than 1% given that early R0 numbers were in the range 2.5-3. Nothing in the Diamond Princess data suggested an R0 just barely over one. In fact, nothing since supports the notion that R0 is outside the range 2.5-3 across sizable populations.

          Reading closely, it is clear he thought that 1% was more likely than 60%.

        • Also, the 1% is for the US and the 60% is for the global population, which is a bit odd – they are not really high and low end of the same thing.

        • Yeah, it’s pretty crazy. The only way I can see that as halfway making any sense is the idea that the R estimates at the time, IIRC, were for places that were *super* dense by US standards and had atypical patterns (more mass transit, etc.) for US cities.

          But even using that line of thinking, I can’t see 1% as even halfway reasonable. At least 20% of the US population, I think, is in more “Europe-like” density areas (Northeast megalopolis, Chicago metro, maybe LA) – half of that is still 10%.

          So even with the most “leaning over backwards to be favorable” thinking, still an order of magnitude higher.

          (Not that I believed 60% either, then. I was thinking 20%-40%, as seen in the last few flu pandemics.)

        • >>I suppose you could argue that he probably believed that all those infected had been detected in which case IFR and CFR are the same

          Yes, and that’s probably close to true if not true for the Diamond Princess, actually.

          Yeah, 0.3% IFR was I think well within the range of reasonable estimates at the time, given how far off estimates early in an outbreak can be. 1%, not so much.

        • Early serological studies in Europe came up with 1%+.

          Of course, at that point, treatment was of the three-blind-mice variety and hospitals in some of those areas had been overrun.

          0.5%-1% was a much more reasonable range at that point in time than a point estimate of 0.3%. Over the entire range.

        • 0.5-1% was certainly a reasonable range, but IMO it wasn’t the *entire* reasonable range at that point. There were enough uncertainties that a lot of things were IMO possible (how good were those early serological studies? how comparable are the populations to the US? how much of Italy’s problems were due to multigenerational households leading to very high infection among the oldest population?)

          I agree 0.3% was low, even then, but I don’t think it was *unreasonable* in mid-March.

          The 1% infected estimate was pretty absurd even as a low end, though.

        • Sorry. I was thinking that the US CFR was clearly not 0.3%, but much higher (but the IFR could reasonably be expected to be much lower). But yeah, the Diamond Princess numbers are actually CFR.

        • Yes, that’s what I was unclearly trying to say, sorry… Ioannidis was trying to extrapolate a US IFR (even though he called it “case fatality rate”, that’s actually what he was calculating, given that he was talking % infected, not % tested) from the (higher) Diamond Princess CFR, given the very different age distribution.

Leave a Reply to Michael J Cancel reply

Your email address will not be published. Required fields are marked *