3 ms·

You are correct that the Buzzfeed article includes that quote, but I'm extremely doubtful that Ioannidis ever made this prediction. I'm guessing that they are

by nkurz 6y ago

You are correct that the Buzzfeed article includes that quote, but I'm extremely doubtful that Ioannidis ever made this prediction. I'm guessing that they are misinterpreting his "Fiasco in the making" article: https://www.statnews.com/2020/03/17/a-fiasco-in-the-making-as-the-coronavirus-pandemic-takes-hold-we-are-making-decisions-without-reliable-data/ https://www.statnews.com/2020/03/17/a-fiasco-in-the-making-a.... In that, he says: Projecting the Diamond Princess mortality rate onto the age structure of the U.S. population, the death rate among people infected with Covid-19 would be 0.125%. But since this estimate is based on extremely thin data — there were just seven deaths among the 700 infected passengers and crew — the real death rate could stretch from five times lower (0.025%) to five times higher (0.625%). It is also possible that some of the passengers who were infected might die later, and that tourists may have different frequencies of chronic diseases — a risk factor for worse outcomes with SARS-CoV-2 infection — than the general population. Adding these extra sources of uncertainty, reasonable estimates for the case fatality ratio in the general U.S. population vary from 0.05% to 1%. This strikes me as the sort of thing a careful statistician might say. He then goes on to say that if the rate is at the lower end of this, a shutdown would be unwarranted: A population-wide case fatality rate of 0.05% is lower than seasonal influenza. If that is the true rate, locking down the world with potentially tremendous social and financial consequences may be totally irrational. One can definitely disagree with his reasoning, but based on this article I can't see how Buzzfeed would be justified in claiming that he "predicted that roughly 10,000 Americans could die", unless you are interpreting "could" very literally as a lower bound. I'd definitely update my opinion of Ioannidis negatively if you can find evidence that he confidently predicted that such a small number of deaths the most likely outcome. Conversely, I'd hope you might update toward seeing the article as a "hit piece" if you can't find any such evidence.

coldtea 6y ago
What he did was that he used weasel phrasing to ease down fears and advocate policy ("if that's the true rate" -- despite he and everybody already knew well that the true rates in Italy, Spain etc at the time were far higher). There was absolutely no reason (much less a "careful statistician's reason) to write ("A population-wide case fatality rate of 0.05% is lower than seasonal influenza.") except to imply that's a plausible scenario. Especially since he closes with that nugget... And even ignoring that a "careful statistician" (who knows biology) would also not discard long term health effects and just focus on deaths, as if once you've not died from covid it's all a-ok...
jsnell 6y ago
From your own link: > 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 That was the only point estimate he made in that piece, so it seems quite fair to hold him to it. It wasn't being presented as a lower bound, but as a reasonable middle ground.
- nkurz 6y ago
  > That was the only point estimate he made in that piece Although I'd argue that neither is an "estimate", he also says 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. Is the difference for you that in the first case he only says "if" and in the second he also appends "which I do not espouse"? All I can conclude is we must use language differently. I would never interpret a statistician as making an estimate if they give a single worked example from a range of numbers, much less consider this a prediction. I would consider it a reasonable and serious criticism to point out that his range was wrong, though! Was it? Or do the current numbers fall somewhere within his range? My guess (although I don't know the current US numbers well enough to verify) his range is biased very low, but the upper end is still possibly correct.
  - jsnell 6y ago
    Yes, I would indeed say that 40M is not his estimate. It was him ridiculing other people's numbers as unrealistic. (Note also the sleight of hand of switching from US numbers to global ones). While for the 10k number he had used the previous sections to estimate a range, and then chose the middle of that range. Again, it was not phrased in any way as a lower bound, he got to make all the assumptions, and he must have known that this is the number that would get reported in the media. When Ioannidis published that 10k number, the US death toll was at 120 people. Three weeks later, it was at 20k people. Even at that point Ioannidis was still predicting (and this time there can be no ambiguity about it being a prediction): > If I were to make an informed estimate based on the limited testing data we have, I would say that covid-19 will result in fewer than 40,000 deaths this season in the USA. It was an absolutely nonsensical number, there was no possible justification for it. Given the time delays involved, you would have gotten to over 40k deaths even if all further Covid transmission were to stop on dime the moment he made the prediction. This number was exceeded 10 days later. If his "informed estimate" in this situation was 40k, it's quite implausible that the 10k number in much better conditions was just intended as some kind of illustration of what the low end of the range was. Certainly calling him out on the 10k estimate does not make the posted article a hitpiece, like you claimed. > Or do the current numbers fall somewhere within his range? So even if we're charitable and assume he consistently meant IFR when writing CFR, the low end of his range was already known to be impossible at the time of writing. The reality will probably intersect with his range, but that's just because it's so absurdly wide and if we're only concerned with the IFR. But the problem is that what uncertanties remained at the time had an inverse correlation between R0 and IFR. Ioannidis was assuming both to be low. If this article had been all there was to it, one could maybe give Ioannidis the benefit of the doubt. Maybe he really was trying to highlight the amount of uncertainty, and was too naive to realize how the general public would understand the 10k number that he just put in as a meaningless bon mot. But this article wasn't the end of it. He's continued to consistently cherry-pick and misrepresent the data, always in the direction that tries to make Covid appear less serious.

by nkurz 6y ago

coldtea 6y ago

What he did was that he used weasel phrasing to ease down fears and advocate policy ("if that's the true rate" -- despite he and everybody already knew well that the true rates in Italy, Spain etc at the time were far higher). There was absolutely no reason (much less a "careful statistician's reason) to write ("A population-wide case fatality rate of 0.05% is lower than seasonal influenza.") except to imply that's a plausible scenario. Especially since he closes with that nugget... And even ignoring that a "careful statistician" (who knows biology) would also not discard long term health effects and just focus on deaths, as if once you've not died from covid it's all a-ok...

jsnell 6y ago

From your own link: > 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 That was the only point estimate he made in that piece, so it seems quite fair to hold him to it. It wasn't being presented as a lower bound, but as a reasonable middle ground.

nkurz 6y ago
> That was the only point estimate he made in that piece Although I'd argue that neither is an "estimate", he also says 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. Is the difference for you that in the first case he only says "if" and in the second he also appends "which I do not espouse"? All I can conclude is we must use language differently. I would never interpret a statistician as making an estimate if they give a single worked example from a range of numbers, much less consider this a prediction. I would consider it a reasonable and serious criticism to point out that his range was wrong, though! Was it? Or do the current numbers fall somewhere within his range? My guess (although I don't know the current US numbers well enough to verify) his range is biased very low, but the upper end is still possibly correct.
- jsnell 6y ago
  Yes, I would indeed say that 40M is not his estimate. It was him ridiculing other people's numbers as unrealistic. (Note also the sleight of hand of switching from US numbers to global ones). While for the 10k number he had used the previous sections to estimate a range, and then chose the middle of that range. Again, it was not phrased in any way as a lower bound, he got to make all the assumptions, and he must have known that this is the number that would get reported in the media. When Ioannidis published that 10k number, the US death toll was at 120 people. Three weeks later, it was at 20k people. Even at that point Ioannidis was still predicting (and this time there can be no ambiguity about it being a prediction): > If I were to make an informed estimate based on the limited testing data we have, I would say that covid-19 will result in fewer than 40,000 deaths this season in the USA. It was an absolutely nonsensical number, there was no possible justification for it. Given the time delays involved, you would have gotten to over 40k deaths even if all further Covid transmission were to stop on dime the moment he made the prediction. This number was exceeded 10 days later. If his "informed estimate" in this situation was 40k, it's quite implausible that the 10k number in much better conditions was just intended as some kind of illustration of what the low end of the range was. Certainly calling him out on the 10k estimate does not make the posted article a hitpiece, like you claimed. > Or do the current numbers fall somewhere within his range? So even if we're charitable and assume he consistently meant IFR when writing CFR, the low end of his range was already known to be impossible at the time of writing. The reality will probably intersect with his range, but that's just because it's so absurdly wide and if we're only concerned with the IFR. But the problem is that what uncertanties remained at the time had an inverse correlation between R0 and IFR. Ioannidis was assuming both to be low. If this article had been all there was to it, one could maybe give Ioannidis the benefit of the doubt. Maybe he really was trying to highlight the amount of uncertainty, and was too naive to realize how the general public would understand the 10k number that he just put in as a meaningless bon mot. But this article wasn't the end of it. He's continued to consistently cherry-pick and misrepresent the data, always in the direction that tries to make Covid appear less serious.