Hacker News Reader

> Are there things we need to look at? Yes. But if someone doesn’t save your life then you won’t be around to argue about how they could have done it better. It’s also pretty rich to claim the numbers don’t add up and substitute numbers that don’t add up either.

Using this argument, we can justify anything that was done in the past on the basis of "well, we're here to argue it over now, so it must have been the right thing".

> Currently we list 604 thousand Americans dead from COVID, that’s 0.18% of the entire population - not the infected population. Brazil is at 0.23%, and rising fast. Worldwide it’s almost 4 million out of 181M reported cases. That’s over 2% IFR.

Regardless of the exact number, isn't it a bit disturbing to you that, per the article, "Americans also estimated that the share of COVID-19 deaths for people between 18 and 24 was 8%"? The fact that we're so misinformed can't be a good thing, can it?

For my money, it's plausible that we massively overreacted (or misreacted) to covid. The opposite is also plausible though. It's a good thing that we're having this conversation and it's mildy creepy that we haven't been able to have it until now.

Finally, there's a question that always comes up with this topic: is it permissible for public employees to lie to the public in order to get them to do the right thing (or whatever they think the right thing is)? A lot of the anger over the government's handling of covid has to do with that. It isn't unreasonable to believe (as I do) that, consequences be damned, the government is obligated to tell the truth and should not lie to get people to act a certain way.

I don’t think it is weird we are talking about it now. The last year was mostly about getting through it, and only now can we start putting data in context.

I can only say what is needed : a medicine is number one. Making society resilient against airborne viruses and we need absolute data on what tools we have to stop a virus in a room with people.

But last but not least it’s also a behavioral problem (which you allude to from a different angle) where people are making decisions that allow the virus to spread. I talked to a red cross doctor and he said that distancing and masks in a hospital setting is pretty much enough to kill transmission. It was his opinion that truly honoring the guidelines would have stopped it.

> Americans also estimated that the share of COVID-19 deaths for people between 18 and 24 was 8%"? The fact that we're so misinformed can't be a good thing, can it?

The correct numbers have been available on the CDC website the entire time, unfortunately a lot of people either don’t care or don’t find it useful to memorize those statistics. Not great to see but I think it track’s with the public usually mangling other statistical numbers as well when surveyed.

> A lot of the anger over the government's handling of covid has to do with that. It isn't unreasonable to believe (as I do) that, consequences be damned, the government is obligated to tell the truth and should not lie to get people to act a certain way.

Agree, the fact that so many government leaders spent a year downplaying the danger of the virus and lying to their followers to try to win an election or for other political reasons should make people angry and upset. It’s always sad seeing the stories of someone who was led astray by those leaders and subsequently even refused the vaccine and died of Covid-19 because they had been convinced it was just a common cold that only killed nursing home patients.

Couple quick points here:

1)I'm no mathematician or statistician, but as I understand it the IFR also counts an estimate of those presumed to have had an asymptomatic case. A published study earlier this month estimates that the reported number of non-hospitalized cases between February and September of 2020 accounts for 1 out of every 7 cases of reported infection because of reasons like the symptoms weren't typical or didn't appear at all, amongst others. I don't know if that puts the IFR as low as they are saying, but it's still pretty low.

https://academic.oup.com/cid/article/72/12/e1010/6000389

2) The lockdowns have not shown any statistical data to the effect of controlling the transmission of the disease. As an aside, they actually have some support that they may have created a negative impact overall as people were forced to reside with family members who contracted and passed on the disease. There are plenty of anecdotal pieces of evidence on both sides of the argument on the subject of lockdowns. That in and of itself should be an indicator for caution and moderation when calling for unilateral action.

3) Even if one subscribes to the idea that the lockdowns did have a positive effect in controlling the rate of infection, the lockdowns were never billed as protecting the vulnerable. That was masking. The lockdowns were billed as an effort to stop hospitals from being overwhelmed. Sure, while there were hotspots and times of higher activity, I have not personally come across data that shows that these hospitals were ever in danger of being over-whelmed. However, I have personally seen examples of people who have lost their jobs in the medical industry because the shift to a laser focus on COVID-19 caused someone to lose their job due to lack of the necessary funding.

I don't know that any numbers are really going to add up for quite some time. I do think that the imposition of unilateral action like lockdowns was a terrible idea and the cost of doing so will continue to show itself over the next several years. Data will continue to change and be updated just like it is for all major medical events. It appears to me that perhaps the best course of action is making recommendations to people and letting them chose their own path. The history of this country and trying to impose something by force is not exactly stellar.

My memory from the time of the initial lockdowns in the US was that we saw hospitals in other countries being overwhelmed, that scared the crap out of our public health experts, and that drove our response. I don’t know how close any areas in the US eventually came to that scenario.

In some ways this is similar to the response to the issues around Y2K. Did we need to spend as much as we did on that? We don’t have an alternate universe to compare with.

No, dividing the number of reported cases by the number of reported deaths absolutely does not give the infection fatality rate. For example, the UK has one of the most widespread testing programs in the world and also monitors the actual infection rate via sampling, and we're still not detecting most Covid infections and reporting them as cases. I think for a while the government was even publishing estimates of what proportion of infections were being found, and it wasn't much, can't find them off-hand though. Covid-related deaths, though, are pretty well reported since pretty much anyone touching the medical system gets tested. (Some countries have been massively underdetecting deaths but they're making an even worse job of reporting mild cases...)

The articles gets things right as well. Cuomo's decisions were atrocious and clearly wrong without the need for hindsight. School closures in the US are odd, and varied to the point of following no science and in large part being politically driven. And the cost of lockdowns is fairly well known, and deserves critique.

The articles's tone is distinct and specific, and going to rub some people the wrong way, but tone shouldn't be confused with the correctness or incorrectness of an argument.

I'd say unfortunately, because tone is far too prevalent in online media, but the horse seems to have bolted on keeping the tone neutral, and I'd wager a strong tone likely leads to higher "engagement", so people have collectively decided with their actions how they feel about tone.

You're confusing case fatality rate with infection fatality rate. Due to limited testing actual infection numbers are far higher than case numbers. The current best estimate of worldwide IFR is about 0.15%, with a wide variance between countries.

https://doi.org/10.1111/eci.13554

There's a whole story about this.

The figure I most often cite is 0.68% from the study by Gideon Meyerowitz-Katz and Lea Merone[1]. That paper is skeptical of serosurveys that show high incidence (see the discussion of reference 61, which is by Ioannidis).

Ioannidis responded with the paper you cited, the preprint of which contained a personal and unprofessional attack against Meyerowitz-Katz, removed from the published version, though a scientific critique remains. Meyerowitz-Katz responded on Twitter[2], mostly to the substantive critiques. There's more drama around this, with a number of prominent virologists taking M-K's side (see the Twitter threads).

Also see the PubPeer reviews for this paper[3], which contain more serious criticism of Ioannidis.

I go into detail here to make a few points. The main one is that, especially when there's legitimate scientific controversy, citing a single paper can really fail to give the whole picture. Another is that "the science" isn't one simple monolithic set of beliefs, even on such a seemingly basic statistical question, but is actually more of a conversation.

[1]: https://www.medrxiv.org/content/10.1101/2020.05.03.20089854v...

[2]: https://twitter.com/GidMK/status/1376304539897237508

[3]: https://pubpeer.com/publications/9FE142DA15F748FA283605C7201...

Hardly.

If you are calculating the bounds on a ratio, then you use the smallest denominator you have to establish the upper bound, and the largest denominator to set the lower bound. That’s reported cases and the total population respectively.

The article claims that 0.15% have died. Basic math on the published statistics say at least 0.18% have died in the US, and 0.23% in Brazild, which last I heard is still in the grip of it. Sometimes you should put down your citations and look out the window.

The argument of the article is that people fucked up and we should get to the bottom of it. That The IFR of 0.9% was too high, it’s really 1/6th of that, which sounds pretty damning. It is not an accident when you bring up that actuals are multiples of predicted. The discrepancy is your hook, and a small error does not demand action (ie, let it go).

Hooks that lie are called manipulative.

The author is claiming a factor of 6 overestimation, and by the most cynical reading of the death statistics it cannot possibly be greater than 4, and in all likelihood is more like 2. Ie, let it go, and stop attacking disaster science.

No you're just making up numbers and picking outlier countries. If there is a more accurate estimate for worldwide IFR then please provide a citation.

For people who are curious about this and want to read further than the original Ioannidis and the Meyerowitz-Katz papers, here are two citations I found enlightening. The first is a frequentist analysis that is age-stratified, so is likely to control for the rather different age structures in different countries[1]. The second is a Bayesian approach[2]. Both seem pretty reasonable to me, and don't display obvious axes to grind.

[1]: https://pubmed.ncbi.nlm.nih.gov/33289900/

[2]: https://www.medrxiv.org/content/10.1101/2021.05.12.21256975v...

1. GP is arguing that even assuming that the entire USA had been infected (ie, using the largest possible denominator), there were enough deaths for the IFR to exceed 0.15%.

2. 0.15% is not "the current best estimate", but an estimate by John P. A. Ioannidis who has published some remarkable sophistry at the beginning of the pandemic, which was (while not technically wrong) absurdly misleading IMHO.

I'd say best estimates for the IFR now a) acknowledge that it depends enormously on age (looks like a pretty exponential relationship), thus on the age structure of a country/state, and b) put it, for the US, say, between 0.2% and >1%.

> Currently we list 604 thousand Americans dead from COVID, that’s 0.18% of the entire population - not the infected population. Brazil is at 0.23%, and rising fast. Worldwide it’s almost 4 million out of 181M reported cases. That’s over 2% IFR.

If one were to assume that everyone in these countries was infected and we are just really awful at reporting, that number is still wrong. But the US is currently reporting a hair over 1:10 people as having been infected, and Brazil has not crossed that threshold yet, so 0.15% is already empirically - to use the technical term - horse shit [1].

Your analysis is severely flawed from the fact that you are equating the number of "confirmed cases" with "number of infections," but many infections are never confirmed with a test (and indeed never even become symptomatic).

The article links to a source to support its 0.15% number, but some others are below:

Nature, estimates about 0.5% for USA: https://www.nature.com/articles/s41586-020-2918-0

World Health Organization, median IFR of 0.27% (0.05% for those under 70): https://www.who.int/bulletin/online_first/BLT.20.265892.pdf

Did you ignore the GP's first sentence? For 0.15% to be right either the death #s have to be significantly overinflated, or more than 100% of the U.S. population must have been infected with COVID.

0.5 and 0.27 are both larger #s than 0.15, so that supports the GP's point.

The 0.15% IFR estimate is a worldwide number. There is a significant variance between countries. Unfortunately Americans on average are older, sicker, and fatter so they tend to die more.