Evidence over Hysteria – Covid-19
medium.com
medium.com
The article points to South Korea as an example of a success story in the fight against covid-19. South Korea immediately began widespread testing for covid-19 as soon as infections were detected. They trace cases and isolate people suspected of having become infected. South Korea has universal healthcare, ranking first among OECD countries for access. All of this is pretty much the opposite of we've been doing in the US, which is why a lot of people here are worried about what's going to happen in the next few months.
It's reasonable to ask whether school closings or shutting borders are the right policies. But we should be discussing alternative means of containing the virus and we need to be clear what we're up against.
Have previous estimates for other diseases been off by that much? A whole order of magnitude, at this level of study? Citations please. Yes, the number is probably not perfectly accurate, but it's also converging toward what is very likely to be a correct figure. Also, how much does it have to change to affect the broader point? A hundred thousand deaths (i.e. one order of magnitude) is still nothing to take lightly. Such aggressive minimization of danger is still a worse problem than the "hysteria" the OP is supposedly trying to address.
It would require the infection to be very contagious, and it probably would require us to have been pretty unlucky with the timing, but as far as I know none of that is contradicted by the available evidence.
[0]: https://mobile.twitter.com/ARanganathan72/status/12414116867...
Also just reported today, apparently over 99% of fatalities in Italy were folks who had existing medical conditions, but I have no idea what that really tells us.
Anyway, you could be right too, maybe it’s unreasonable to guess that true IFR is an order of magnitude lower.
Not a lot, without knowing the number with existing medical conditions among the infected population. How many over-70s don't have any existing medeical condition?
1.1% of their cases have died, and an additional 0.8% are in serious or critical condition. And they're a best case scenario: the health care system was not overloaded, and their diagnosed cases are skewing disproportionately young.
The Diamond Princess also has had a fatality rate so far of 1.1%. In that case we know with absolute certainty that there is no massive hidden iceberg of infected but undiagnosed people. There might be a hidden iceberg of unreported fatalities though, since once they were off the ship, the cases stopped being tracked as Diamond Princess cases. On the plus side, the demographics obviously skewed very old.
Anyway, there's very good reason to believe that the IFR is at least 0.6%. Suggesting that it's actually <0.06% seems just ludicrous. Is there any evidence at all supporting that, or are you just arbitrarily demanding "absolute certainty" in a time-critical situation?
If enough people have already recovered from mild or asymptomatic cases (it would have to be a pretty hefty percentage, granted) then that, combined with not-total-lockdown conditions, could bring down the R0 value enough to cause the same result. Do you think it’s much more likely that the virus has been totally contained?
> The Diamond Princess also has had a fatality rate so far of 1.1%. In that case we know with absolute certainty that there is no massive hidden iceberg of infected but undiagnosed people.
This might be the strongest evidence against the “mild and everywhere” theory. However, there are still something like five out of six passengers who tested negative but could have recovered before they were tested. Even if they’re not a great representative sample, testing them all for antibodies might be a good move.
They're reporting basically no new cases at this point. For that to happen naturally, they'd need to have gotten like 2/3rds of the population infected. And it's quite hard to see how that could be true.
Korea had the most comprehensive testing (320k tests, available to basically anyone with even mild symptoms). They have so far confirmed infections in 0.02% of the population, and only got positive results in less than 3% of the tests. It's hard to believe that 60-70% of the population got infected, but even in the skewed sample of people who got tested only 3% being positive.
Compare that with something like Switzerland where the number of confirmed cases is at 0.08% of the population despite worse test coverage (basically only get tested if your symptoms warrant hospitalization), and there's no sign of it slowing down yet.
Or Spain which has 3x the cases of Korea on a similar population, has a health care system that's melting down, and where more than 55% of the tests are positive.
If herd immunity provided by all the undetected cases protected Korea, why is it not kicking in anywhere else even once their infection rates are obviously higher?
> Do you think it’s much more likely that the virus has been totally contained?
"Totally contained" is an exaggeration, since they are still finding new cases. But given how long the infection rate has been stable, it does look like they've managed to reach a R0 of almost exactly 1 with the aggressive testing + tracing strategy.
The only alternative that's at all plausible is that the Korean numbers are being fudged somehow, and they actually have a lot more cases than are being reported. But then why aren't the hospitals in a crisis unlike in other affected regions?
It’s like 50-100 new cases every day, looks like. I’m getting out of my depth with what exactly that allows us to rule out, but it sounds like “constantly restarting the chain of transmission” might be the actual reality?
If that drops further, then maybe these new cases are just the ones that incubated longer instead?
Yes, but it's a tiny number of undetected cases restarting the chain, and those chains being detected and lopped off before they have time to go exponential. But it's just not possible that they have tens of thousands of undetected cases that are spreading the virus further. That'd spin out of control quickly.
That's 180' from what I believe and have experienced repeatedly in industry settings. Transforming the numbers to meaningful data is always fundamental to success. For example here the number of cases confirmed is determined by the testing approach and process - you must know that to be able to make meaningful comparisons (there are many other aspects to epi data but that's just one).
410 This post is under investigation or was found in violation of the Medium Rules.
--
It is archived here: http://archive.md/yuaUq
Well yes, it fell afoul of the conspiracy. :eyeroll: Also it's crap.
I swear you could go into any HN comment thread related to this outbreak and it breaks down to; Person A saying X is going to happen and then Person B saying Y is going to happen and then Person C saying that No! Z is going to happen!
How did everyone get so opinionated about something so unpredictable and unprecedented? In any case I think we should all calm down regardless of how dangerous this situation is. Stay sane, everyone!
There is no reason to post news articles or blogs from random Joes in HN when you can post directly from experts.
I’d rather judge them on the merits of their contribution than some meta signals.
This assumes that you can do that. If you are not an expert you can't possibly have the background to put things in the perspective.
This is not the issue of the ability to read and study, it's about missing variables and weighting evidence. If you don't have the background, you must spend weeks studying to get some hold of issues.
Argument from authority is fallacy of reasoning but it's not fallacy when learning and gathering relevant information. We you can reason only after you have learned the subject.
Social media is full of people who learned basic math behind SIR model and are building layman theories based on them.
But death toll is counted in absolute numbers. No one will think 1M deaths is victory just because this is 1% of whatever. People are right to be hysterical, we have tough times ahead.
I’m not saying we shouldn’t care about deadly illnesses. But the world is a dangerous place, and we’ve been pretty bad about changing that, historically.
Also, if we ruin 5 million lives to save (some percent of) 1 million, is that worth it? How about ruining 50 million lives? We can save lives with drastic measures, but drastic measures will have bad consequences too, and it seems like nobody has any perspective on that.
1. "New thing X puts us at risk of losing Y lives every day."
2. "We lose Y lives to car accidents every day".
3. "Ergo, X is harmless!"
You now have 2Y lives lost every day.
Since this is an exponential phenomenon, I don't buy that "per-capita" matters (on top of the fact that he seems to be pulling in numbers as a snapshot in absolute time rather than windowed from initial cases).
If the numbers double every 5 days, the US can have 16 times the population of some other smaller country and be only 3 weeks from the same per capita effects.
> driven by data from medical professionals and academic articles
...as filtered through one person's understanding and biases. Often this results in a very different picture than perusal of the original source material would.
> I’m most known for popularizing the “growth hacking movement” in Silicon Valley
That's an anti-qualification.
> Rank ordering based on the total number of cases shows that the US on a per-capita basis is significantly lower than the top six nations
This is right after he got done explaining why total number of cases of useless. So why is it here? Because it suits the overall theme of minimizing the threat.
> Daily growth rates declined over time across all countries regardless of particular policy solutions
Sounds compelling, but is it true? Note that the first chart is from SoberLook, which is definitely not one of the top-tier sources cited at the beginning of the article and not even one of those supposedly cited at the bottom (Bloomberg, Johns Hopkins). Bit deceptive, that. I'll leave it to you to figure out what kind of source SoberLook is. Oh, and we're using the "losing metric" of total reported cases again. Doesn't everyone know by now how fraught that is? The author certainly does.
The second chart already shows a very different story. There's no legend (which is suspicious) but it's clear that many of the lines do not flatten out as would be necessary to support the author's claim.
> Both the CDC and WHO are optimizing virality and healthcare utilization, while ignoring the economic shock to our system.
We're starting to see a hint of where the author wanted to go with this, probably before one word was written.
> According to WHO’s COVID-19 lead Maria Van Kerkhove, true community based spreading is very rare.
Hm. Where did van Kerkhove say this? Why is there no specific citation, as for more convenient factoids?
> The data from China shows that community-based spread was only a very small handful of cases.
Same as above. No citation. And does the data actually show that?
https://www.cdc.gov/coronavirus/2019-ncov/prepare/transmissi... "The virus that causes COVID-19 seems to be spreading easily and sustainably in the community"
https://www.who.int/docs/default-source/coronaviruse/situati... "Community transmissionis evidenced by the inability to relate confirmed cases through chains of transmission for a large number of cases,"
Don't let the digression about aerosols (which nobody has claimed to be a primary mechanism for spread of COVID-19) fool you; the experts, including those included in the author's appeal to authority at the beginning, seem to have very different beliefs than those he presents.
> Released on March 10th, one study mapped COVID-19 virality capability by high temperature and high humidity.
One study. Minimizing again. It's a very pretty temperature graph, which doesn't actually prove anything. If you follow the link, you'll see that it was actually limited to China, which might not tell us much about anywhere else. What we actually know about the temperature sensitivity of coronaviruses is that they're remarkably stable up to about 37°C, and infectivity only starts to decline above that.
https://www.ncbi.nlm.nih.gov/pubmed/14631830
> Children and Teens aren’t at risk
This does seem to be true, or close to it, for once. But why only talk about the lowest-risk groups? Why not people in their 20s, or the high-risk folks at the other end of the spectrum? Minimizing again.
This is getting long, so I'll stop dissecting each little piece. You get the idea. Let's skip to the end, where the agenda becomes clear.
> Local governments and politicians are inflicting massive harm and disruption with little evidence to support their draconian edicts.
Very objective choice of words there. Inflicting. Draconian.
> Every local government is in a mimetic race to one-up each other in authoritarian city ordinances
Attributing motive.
> The COVID-19 hysteria is pushing aside our protections as individual citizens and permanently harming our free, tolerant, open civil society
More appeal to emotion, and expressive of the author's real priorities. Don't get me wrong, politicians taking advantage of the situation to extend authoritarian power is a real and valid concern. But no way in hell should it be our first concern. It certainly doesn't justify twisting the facts and minimizing the threat that disease poses to the most vulnerable - not just the old but also the homeless, the already ill, etc. Those people are not expendable to "Support business and productivity" as the author puts it in a heading.
“This virus is not circulating in the community, even in the highest incidence areas across China,” Van Kerkhove insisted.
What’s the difference between spread among close contacts and community spread, you might wonder? Van Kerkhove said the data the mission saw in China pointed to the virus finding its way into households and transmitting there. One family member gets infected and infects others. The “secondary attack rate” — the percentage of people in a household who got infected after someone brought the virus into the home — was between 3% and 10%.
Van Kerkhove said true community spread involves transmission where people get infected in a movie theater, on the subway, or walking down the street. There’s no way to trace back the source of infection because there’s no connection between the infected person and the person he or she infects. That’s not what the Chinese data show, she said.
Marc Lipsitch, an infectious diseases epidemiologist at the Harvard School of Public Health, found this claim puzzling. “I have reached out to the World Health Organization to understand the basis of some of those statements. My perception is that there is significant community transmission, especially when you aren’t aware that someone is sick, because there’s not enough testing,” he said.
https://www.statnews.com/2020/03/06/were-learning-a-lot-abou...
If it turns out this is indeed an overreaction, they’re going to be totally pissed off, and I’m probably going to agree with them. You’re free to laugh all you want, but it’s a privilege that you probably don’t appreciate.
It's a no-win situation for a lot of people with a certain mindset. If we take extreme measures and the infection is worse than predicted, then they'll say that everyone should have done more.
If we take extreme measures and 'flatten the curve' enough so that the hospitals aren't overwhelmed and the number of deaths is lower than expected, then they'll say that all those measures were an overreaction to something that turned out to not be that bad. Even though that was the desired outcome of taking those extreme measures in the first place.
There is also a worryingly-large number of people that will insist that everything is overblown, everything is fine, et cetera, until they (or someone they know personally) is impacted. Until then all these things are happening to 'someone else' and can be ignored.
This is the possibility I would be super pissed about, not the other two. Unfortunately, in either “not that bad” outcome it’s going to be very hard to prove which one it is, which of course won’t stop a bunch of folks from parading around and bragging about how they saved the world.
I’ve seen this pattern more than once in my short time alive. But assuming what we all hope, that one of the “not that bad” scenarios plays out, the difference this time will be that many millions of lives will have been directly ruined because of those extreme measures.
Emergency funding will only go so far and once we get past a month of extraordinary measures the serious pain hitting millions of people will be a guarantee. Even in countries with far wider safety nets than the US, none of them are designed to handle this sort of unprecedented event.
Then we see Hong Kong announcing 50 new cases in a single day [1] after getting down to zero for an extended period. The only certainty is that there will be no easy answers. Which I think the OP posts downplays in their ‘data driven’ recommendations.
[1] https://www.scmp.com/news/hong-kong/health-environment/artic...
https://twitter.com/jburnmurdoch/status/1241092187577819137?...
I'll probably do myself a favour and stop clicking on coronavirus medium links
This crap shouldn’t be on HN. This forum has become next to useless, slowly, then all at once in the last few weeks. I would assert that this community could do better, but at this point maybe it can’t.
I am not on board with everything in this article, though. I am unsure about his proposition on reopening schools, for example. But I'm not throwing it all away over that one part.
Yes, he got as Farr as noticing that it’s roughly bell-shaped. But just because it goes up and then down it’s not actually a normal distribution.
At the beginning, when no one is immune, infections obviously grow exponentially. But the bell curve, even though it looks similar (sigmoid) doesn’t rise exponentially.
So it looks similar, but it’s not. If you observe infections and somehow try to fit a normal distribution, you’ll get widely wrong predictions, especially at the beginning.
Farr came up with this in 1840. That’s before Pasteur formulated germ theory as we know it today, so it was probably close enough for its time. Today it’s not.