4 ms·

Fun fact: the trial you referenced failed to show any benefit of COVID vaccination in reducing all-cause mortality in the population they studied: https://www.

Fun fact: the trial you referenced failed to show any benefit of COVID vaccination in reducing all-cause mortality in the population they studied: https://www.medrxiv.org/content/10.1101/2021.07.28.21261159v1.full.pdf https://www.medrxiv.org/content/10.1101/2021.07.28.21261159v... > During the blinded, controlled period, 15 BNT162b2 and 14 placebo recipients died; during the open-label period, 3 BNT162b2 and 2 original placebo recipients who received BNT162b2 after unblinding died. None of these deaths were considered related to BNT162b2 by investigators. There was actually one more death in the experimental group, albeit that's not significantly significant. So some combination of the following must be true: (1) The vaccination is effective at reducing COVID mortality, but COVID mortality for people in the trial was such a joke that eliminating 95% of COVID mortality doesn't actually change one's risk of dying in a non-negligible manner (2) The vaccination is effective at reducing COVID mortality, which does spare lives, but it ends up killing just as many from adverse events / side effects (3) The vaccine isn't effective and they doctored the numbers. My money is mostly on (1) with a sprinkling of (2), personally. --- And since people like to be binary thinkers, this is where I mention that I'm not an Ivermectin shill and as a medical nihilist I'm strongly skeptical of treatments in general. And while I haven't looked at the IVM data very much at all, what I have seen is incredibly weak evidence at best, as well as a bunch of really crappy associative arguments from the IVM crowd ("Africa uses IVM and Africa has less COVID mortality than the US!" as if that proves anything) To quote (slightly paraphrased) the great Jacob Stegenga: > If we consider the ubiquity of small effect sizes in medicine, the extent of misleading evidence in medical research, the thin theoretical basis of many interventions, and the malleability of empirical methods, then our confidence in medical interventions ought to be low.

phonypc 5y ago
>Fun fact: the trial you referenced failed to show any benefit of COVID vaccination in reducing all-cause mortality in the population they studied: Is the implication that we should have expected it to? The study endpoints are clearly "vaccine efficacy (VE) against laboratory-confirmed COVID-19 and safety data".
- __blockcipher__ 5y ago
  I think most people would have expected it to. Whether that's a reasonable expectation or not is up for debate. But yeah, many if not most people in the US were convinced that COVID was a threat so severe that it warranted completely uprooting life as we knew it. A sizeable minority of people are convinced that COVID vaccination is so important that people should be coerced into it by almost any means necessary. So I think understanding that we don't have a single randomized controlled trial that actually shows a reduction in all-cause mortality when given this intervention is pretty important. It's a general principle of medicine (or at least, it used to be) that the important outcomes are the actually clinically relevant outcomes. Did people die less? Did they achieve a better quality of life? When you get into proxy metrics you get to this weird place where you can show that something is insanely effective for proxy metric X, and yet it makes no difference (or is even deleterious) in thing-you-actually-care-about Y.
  - phonypc 5y ago
    This study not finding an effect it wasn't designed to isn't incompatible with the more hyperbolic interpretations of the threat of COVID / the importance of vaccination. Did we need RCTs demonstrating a reduction in all-cause mortality for polio or measles vaccines before it became blindingly obvious that they were a good idea?
j-wags 5y ago
I initially skipped over this post because it appeared to be written in bad faith [1]. But I gave it another shot and read the linked paper, and I agree with the summary. Thanks for pointing this out -- your post has changed my understanding of the vaccine's effectiveness. For skeptical readers - The linked paper is a medrxiv preprint, with authors from several locations. Many of the authors list affiliations with Pfizer, and the last author is an MD at Pfizer. This seems to be a required report from the ~6-month point of a clinical trial. There really are 14 and 15 all-cause deaths reported in the placebo and treatment groups, respectively. The causes of death for each group are in table S4 [2]. The incidence of COVID in the treatment group is much lower than in the placebo group. I'm surprised by these results because I would have expected to see significantly lower mortality in the vaccine group. One possible reason for this surprising result is that only a small fraction of people - even in the placebo group - caught COVID at all. I'm not sure how to account for possible missed infections but Figure 2 of the main text indicates that 0.08% of the placebo group became CASES (which I think means something like "person felt bad enough to see a medical professional, who then diagnosed them with COVID"), and the table in Figure 2 indicates that there were 1034 "occurrences" out of ~22,000 people in the placebo group (which I read as "~5% confirmed infections"). So, it's possible that the beneficial effect of the vaccine is still hidden in the statistical noise since there have been so few infections in the placebo group. It's also possible that some people _were_ harmed by receiving the vaccine (see table S4 [2]), so there's a fixed harm upfront to the vaccine group, which may eventually be surpassed once more of the placebo group gets infected. Thanks for posting this! [1] I really do appreciate you making this post, and it helped me learn something. I would have been more willing to read it initially if it had excluded the phrases "such a joke", "since people like to be binary thinkers", "crappy associative arguments", "to quote the great...". It seems like there's a silent majority of HN lurkers that are interested in all perspectives, and really are open to quality arguments like this, but might dismiss this post before reading it because of the choice of phrases. [2] Link to supp material PDF is on this page: https://www.medrxiv.org/content/10.1101/2021.07.28.21261159v1.supplementary-material https://www.medrxiv.org/content/10.1101/2021.07.28.21261159v...
- __blockcipher__ 5y ago
  [meta: this was supposed to be a quick response and then I ended up getting real longwinded] Glad you got some value out of my comment! Thanks for the feedback on my tone. In particular the references to specific examples was very helpful to me. I’ll [try to] be mindful of its effect in the future. Not that it's necessary but just because it's kind of interesting to psychoanalyze, I think some of the tone comes as a [maladaptive] response to past times on HN where I'd go really deep into analyzing certain studies, etc and then would get downvoted or even flagged because the conclusion was that lockdowns were deleterious or that mandates are a bad idea, etc. The binary thinkers comment specifically (you're right that it will tend to put people off btw) was because in general (not just on HN) it happens all the time that if I point out, say, how we don't even have an RCT that shows a reduction in all-cause mortality from getting COVID-vaccinated, very often I'll get a response to the tune of "Ivermectin doesn't even work you dummy" because most people seem to only have two boxes to put people in (you're either team trump or team biden, team vaccine or team ivermectin, etc). But obviously even if I had done it in a more diplomatic way, making reference to people being binary thinkers is only going to distract from the actual points being made. I am curious of your mentioning of being offput by "to quote the great...". Did it sound like I was being sarcastic, or was it something else about it? Because I intended it in the literal sense, i.e. indicating my respect for Stegenga and his work. --- Back to the actual study: > One possible reason for this surprising result is that only a small fraction of people - even in the placebo group - caught COVID at all. I'm not sure how to account for possible missed infections but Figure 2 of the main text indicates that 0.08% of the placebo group became CASES (which I think means something like "person felt bad enough to see a medical professional, who then diagnosed them with COVID"), and the table in Figure 2 indicates that there were 1034 "occurrences" out of ~22,000 people in the placebo group (which I read as "~5% confirmed infections"). Yes. I suspect that this is partially due to the fact that they actually take some steps to confirm that it really is COVID, whereas in the real world (at least in 2020, but probably still now) they'd run a PCR test with an absurdly high cycle threshold cutoff and then call that COVID even if you were completely asymptomatic (which as an aside is an oxymoron because COVID is supposed to stand for "Coronavirus Infectious Disease", and yet an asymptomatic individual is not diseased (nor are they very infectious btw)). So one interpretation is that when care is taken to actually be somewhat careful about calling something a COVID case, that the prevalence of COVID infection ends up being quite low. Whereas when that care is not taken (whether due to negligence or institutional incentives to drum up case counts / fear in general) you end up with much higher apparent case counts. This is quite speculative, I know. I'd really like to look at a chart of reported COVID cases in the US versus the rate of cases in this trial. I'd be curious if they are similar shape / relative magnitude, or if there's dramatically more cases in the US in general than the laboratory-confirmed cases in the trial. I couldn't actually find an actual case definition (maybe it's in the supplement?) but I found this under the Efficacy subheading in the main text: > BNT162b2 efficacy against laboratory-confirmed COVID-19 with onset ≥7 days post-dose 2 was assessed descriptively in participants without serological or virological evidence of SARS- CoV-2 infection ≤7 days post-dose 2, and in participants with and without evidence of prior infection. Efficacy against severe COVID-19 was also assessed. So I think that means when calculating their overall VE number, they don't "start counting" lab-confirmed COVID-19 until it's been at least a week post second dose. I do know in general in these trials they like to play that game (ignoring infections before the "fully vaccinated" cutoff to get a better-looking VE number). Although Figure 2 seems to provide numbers starting immediately after dose #1 so I suppose those numbers tell us that while VE is quite bad immediately after the first dose, its expected value doesn't seem to be negative (at least in this cohort) As an aside, I'd note that they blew up the control arm after 6 months, so this is basically the only clean data we're ever going to get. Which should be very concerning to people considering the enormous push to mandate vaccination for an intervention that has never been shown to reduce all-cause mortality in the studied populations. > So, it's possible that the beneficial effect of the vaccine is still hidden in the statistical noise since there have been so few infections in the placebo group. It's also possible that some people _were_ harmed by receiving the vaccine (see table S4 [2]), so there's a fixed harm upfront to the vaccine group, which may eventually be surpassed once more of the placebo group gets infected. Unless the data is doctored or has some flaw that I'm not seeing, I agree that a much higher base rate of COVID infection would probably have shown a positive effect on mortality. Which brings us back to #1 of the 3 possible explanations I gave in my original comment. I find it very interesting that actual COVID mortality (or even bad outcomes) were so rare in this cohort of >44,000 people, that a well-over-90-percent-effective-in-reducing-COVID-mortality intervention literally did not make a detectable effect on net mortality. I think from a societal perspective it's quite interesting because during the time this trial was running, people were (and still are btw) getting bombarded by fearmongering, giant red eternally-incrementing "live" death tickers, etc. It doesn't take some grand conspiracy to see why that might be, but I think it's something interesting to reflect on. It's a large part of why COVID took me from kind of libertarian leaning to full-blown anarchocapitalist, because I am personally so disgusted by what the media, the public health "authorities", and "polite society" at large did in terms of catalyzing such tremendous suffering on a worldwide scale (via the hysteria, global supply chain disruption, missed medical appointments, cancelled elective surgeries, and outright authoritarian/totalitarian public policy), for something that for most people, they would literally never notice if not bombarded about its existence.

phonypc 5y ago

>Fun fact: the trial you referenced failed to show any benefit of COVID vaccination in reducing all-cause mortality in the population they studied: Is the implication that we should have expected it to? The study endpoints are clearly "vaccine efficacy (VE) against laboratory-confirmed COVID-19 and safety data".

__blockcipher__ 5y ago
I think most people would have expected it to. Whether that's a reasonable expectation or not is up for debate. But yeah, many if not most people in the US were convinced that COVID was a threat so severe that it warranted completely uprooting life as we knew it. A sizeable minority of people are convinced that COVID vaccination is so important that people should be coerced into it by almost any means necessary. So I think understanding that we don't have a single randomized controlled trial that actually shows a reduction in all-cause mortality when given this intervention is pretty important. It's a general principle of medicine (or at least, it used to be) that the important outcomes are the actually clinically relevant outcomes. Did people die less? Did they achieve a better quality of life? When you get into proxy metrics you get to this weird place where you can show that something is insanely effective for proxy metric X, and yet it makes no difference (or is even deleterious) in thing-you-actually-care-about Y.
- phonypc 5y ago
  This study not finding an effect it wasn't designed to isn't incompatible with the more hyperbolic interpretations of the threat of COVID / the importance of vaccination. Did we need RCTs demonstrating a reduction in all-cause mortality for polio or measles vaccines before it became blindingly obvious that they were a good idea?

j-wags 5y ago

I initially skipped over this post because it appeared to be written in bad faith [1]. But I gave it another shot and read the linked paper, and I agree with the summary. Thanks for pointing this out -- your post has changed my understanding of the vaccine's effectiveness. For skeptical readers - The linked paper is a medrxiv preprint, with authors from several locations. Many of the authors list affiliations with Pfizer, and the last author is an MD at Pfizer. This seems to be a required report from the ~6-month point of a clinical trial. There really are 14 and 15 all-cause deaths reported in the placebo and treatment groups, respectively. The causes of death for each group are in table S4 [2]. The incidence of COVID in the treatment group is much lower than in the placebo group. I'm surprised by these results because I would have expected to see significantly lower mortality in the vaccine group. One possible reason for this surprising result is that only a small fraction of people - even in the placebo group - caught COVID at all. I'm not sure how to account for possible missed infections but Figure 2 of the main text indicates that 0.08% of the placebo group became CASES (which I think means something like "person felt bad enough to see a medical professional, who then diagnosed them with COVID"), and the table in Figure 2 indicates that there were 1034 "occurrences" out of ~22,000 people in the placebo group (which I read as "~5% confirmed infections"). So, it's possible that the beneficial effect of the vaccine is still hidden in the statistical noise since there have been so few infections in the placebo group. It's also possible that some people _were_ harmed by receiving the vaccine (see table S4 [2]), so there's a fixed harm upfront to the vaccine group, which may eventually be surpassed once more of the placebo group gets infected. Thanks for posting this! [1] I really do appreciate you making this post, and it helped me learn something. I would have been more willing to read it initially if it had excluded the phrases "such a joke", "since people like to be binary thinkers", "crappy associative arguments", "to quote the great...". It seems like there's a silent majority of HN lurkers that are interested in all perspectives, and really are open to quality arguments like this, but might dismiss this post before reading it because of the choice of phrases. [2] Link to supp material PDF is on this page: https://www.medrxiv.org/content/10.1101/2021.07.28.21261159v1.supplementary-material https://www.medrxiv.org/content/10.1101/2021.07.28.21261159v...

__blockcipher__ 5y ago
[meta: this was supposed to be a quick response and then I ended up getting real longwinded] Glad you got some value out of my comment! Thanks for the feedback on my tone. In particular the references to specific examples was very helpful to me. I’ll [try to] be mindful of its effect in the future. Not that it's necessary but just because it's kind of interesting to psychoanalyze, I think some of the tone comes as a [maladaptive] response to past times on HN where I'd go really deep into analyzing certain studies, etc and then would get downvoted or even flagged because the conclusion was that lockdowns were deleterious or that mandates are a bad idea, etc. The binary thinkers comment specifically (you're right that it will tend to put people off btw) was because in general (not just on HN) it happens all the time that if I point out, say, how we don't even have an RCT that shows a reduction in all-cause mortality from getting COVID-vaccinated, very often I'll get a response to the tune of "Ivermectin doesn't even work you dummy" because most people seem to only have two boxes to put people in (you're either team trump or team biden, team vaccine or team ivermectin, etc). But obviously even if I had done it in a more diplomatic way, making reference to people being binary thinkers is only going to distract from the actual points being made. I am curious of your mentioning of being offput by "to quote the great...". Did it sound like I was being sarcastic, or was it something else about it? Because I intended it in the literal sense, i.e. indicating my respect for Stegenga and his work. --- Back to the actual study: > One possible reason for this surprising result is that only a small fraction of people - even in the placebo group - caught COVID at all. I'm not sure how to account for possible missed infections but Figure 2 of the main text indicates that 0.08% of the placebo group became CASES (which I think means something like "person felt bad enough to see a medical professional, who then diagnosed them with COVID"), and the table in Figure 2 indicates that there were 1034 "occurrences" out of ~22,000 people in the placebo group (which I read as "~5% confirmed infections"). Yes. I suspect that this is partially due to the fact that they actually take some steps to confirm that it really is COVID, whereas in the real world (at least in 2020, but probably still now) they'd run a PCR test with an absurdly high cycle threshold cutoff and then call that COVID even if you were completely asymptomatic (which as an aside is an oxymoron because COVID is supposed to stand for "Coronavirus Infectious Disease", and yet an asymptomatic individual is not diseased (nor are they very infectious btw)). So one interpretation is that when care is taken to actually be somewhat careful about calling something a COVID case, that the prevalence of COVID infection ends up being quite low. Whereas when that care is not taken (whether due to negligence or institutional incentives to drum up case counts / fear in general) you end up with much higher apparent case counts. This is quite speculative, I know. I'd really like to look at a chart of reported COVID cases in the US versus the rate of cases in this trial. I'd be curious if they are similar shape / relative magnitude, or if there's dramatically more cases in the US in general than the laboratory-confirmed cases in the trial. I couldn't actually find an actual case definition (maybe it's in the supplement?) but I found this under the Efficacy subheading in the main text: > BNT162b2 efficacy against laboratory-confirmed COVID-19 with onset ≥7 days post-dose 2 was assessed descriptively in participants without serological or virological evidence of SARS- CoV-2 infection ≤7 days post-dose 2, and in participants with and without evidence of prior infection. Efficacy against severe COVID-19 was also assessed. So I think that means when calculating their overall VE number, they don't "start counting" lab-confirmed COVID-19 until it's been at least a week post second dose. I do know in general in these trials they like to play that game (ignoring infections before the "fully vaccinated" cutoff to get a better-looking VE number). Although Figure 2 seems to provide numbers starting immediately after dose #1 so I suppose those numbers tell us that while VE is quite bad immediately after the first dose, its expected value doesn't seem to be negative (at least in this cohort) As an aside, I'd note that they blew up the control arm after 6 months, so this is basically the only clean data we're ever going to get. Which should be very concerning to people considering the enormous push to mandate vaccination for an intervention that has never been shown to reduce all-cause mortality in the studied populations. > So, it's possible that the beneficial effect of the vaccine is still hidden in the statistical noise since there have been so few infections in the placebo group. It's also possible that some people _were_ harmed by receiving the vaccine (see table S4 [2]), so there's a fixed harm upfront to the vaccine group, which may eventually be surpassed once more of the placebo group gets infected. Unless the data is doctored or has some flaw that I'm not seeing, I agree that a much higher base rate of COVID infection would probably have shown a positive effect on mortality. Which brings us back to #1 of the 3 possible explanations I gave in my original comment. I find it very interesting that actual COVID mortality (or even bad outcomes) were so rare in this cohort of >44,000 people, that a well-over-90-percent-effective-in-reducing-COVID-mortality intervention literally did not make a detectable effect on net mortality. I think from a societal perspective it's quite interesting because during the time this trial was running, people were (and still are btw) getting bombarded by fearmongering, giant red eternally-incrementing "live" death tickers, etc. It doesn't take some grand conspiracy to see why that might be, but I think it's something interesting to reflect on. It's a large part of why COVID took me from kind of libertarian leaning to full-blown anarchocapitalist, because I am personally so disgusted by what the media, the public health "authorities", and "polite society" at large did in terms of catalyzing such tremendous suffering on a worldwide scale (via the hysteria, global supply chain disruption, missed medical appointments, cancelled elective surgeries, and outright authoritarian/totalitarian public policy), for something that for most people, they would literally never notice if not bombarded about its existence.