Covid-19 Sanity – Covid-19 papers from bioRxiv/medRxiv searchable and sortable
biomed-sanity.com
biomed-sanity.com
Until we do that, I feel like we have no idea of the actual infection rate and population immunity.
We know for a fact that ICUs are being overloaded with cases and morgues are overwhelmed with the dead. Arguing over exact figures is like quibbling over your exact velocity as you're falling off a cliff.
Finding some prettier numbers won't change the reality of the situation, no matter how convincingly you argue your case.
We’re planning for more than the next three days here.
Yes, it's really bad in a few places right now. It looks like it's going to be really bad in a lot more places in the next 15-20 days. How _long_ will it be really bad for and for how _many_ people will it be really bad are important questions to understand.
Don't let the dire situation make you irrational.
Italy and the UK book the death of anyone who died testing positive as a covid19 death. The UK won't release any figures beyond that. Not even hospitalisation data. Studies in Italy show fewer than 15% of those deaths might be specifically caused by it.
> On re-evaluation by the National Institute of Health, only 12 per cent of death certificates have shown a direct causality from coronavirus, while 88 per cent of patients who have died have at least one pre-morbidity - many had two or three,” he says.^
In fact the average comorbiditities is 2.7. mean age of death is 79.5 years. Median 80.5.
^ https://www.telegraph.co.uk/global-health/science-and-diseas...
Shut down everything except essential services, isolate everyone as much as possible, throw as many resources as possible at your health service. Even then, they will be overwhelmed at some point.
The same actions apply today. What's changed?
If you have random sampling from two weeks ago and today, you might find out that you won't have to go to the same extremes as italy, or you might find out that military should have been on the streets a week ago and putting them on the streets now is a good idea
I certainly don't think the US has done everything it can. Quite the contrary - it's no surprise to me at all that it is now top of the leaderboard in recorded cases. The US had to go to at least the extremes of Italy, and it didn't, so you'll end up with many more deaths than you would otherwise have had.
The US could just have looked at what was actually happening in Europe to establish what it needed to do. Arguing over the numbers is exactly what caused the inaction in the first place.
When should Denmark end the lockdown?
(And if Denmark had a random sampling from two weeks ago, and one from today, would they be better positioned to know when to end the lockdown or not?)
Look at the graphs: https://www.worldometers.info/coronavirus/country/denmark/
Is it perhaps the change in daily new cases that's thrown you? Many countries have that drop and increase again as they change the way they're testing.
(but number of daily deaths have been linear for a week now, I don't know why you'd think that's not stable, but you're just completely proving my point by disagreeing with how stable the situation is in Denmark, how do you ACTUALLY know that it's not?)
I know it's not from the statistics. Note I don't need to know the exact percentage of asymptomatic cases for this, or to know the correct course of action. Which was my point to start with. Perhaps you lost track of it?
Being reactionary to a crisis like this is a very bad idea and will lead to it being worse than it needs to be.
And phaemon is completely correct in saying that US has not done everything it can. It was the opposite - spinning situation in absurdly optimistic way until it became impossible to ignore.
Not everywhere ICUS are being overloaded and morgues are overwhelmed with the dead. And ignorance was the first reason why it happened where that's the case. So anything that could decrease the ignorance should be met with open arms.
That's exactly what I mean. There is no scenario where that is the case. If your numbers tell you that, your numbers are wrong.
The actions are the same regardless: shut down everything except essential services, isolate everyone as much as possible, throw as many resources as possible at your health service. Even then, they will be overwhelmed at some point.
What numbers exactly would lead you to a different course of action, based on what we have already observed?
And, again, that's my point: you're asking why it can't be a ridiculously small number when it should be obvious from the reality of the situation that it isn't!
Second, and that's my point: you can have the same result with two very different scenarios: either a lot of people have/had it but in the vast majority of the cases it goes without symptoms at all. So when a big part of the population gets infected only then you start noticing "ICUs being overwhelmed". The second scenario is: it causes serious problems for a substantial part of the infected so even the small number of those who got the virus will flood the hospitals.
And those two scenarios require different responses.
https://www.eurosurveillance.org/content/10.2807/1560-7917.E...
And the percentage of really asymptotic cases is much lower than the most here expect.
All the tests in all the countries could be seen as the statistical sampling. The speed of the transmission is also known, and surely not fundamentally different than the speed of people needing hospital or dying -- the doubling time is nowhere faster than 2 days, and the typical is 3 days.
In short, it can't be expected that "already everybody has it" in the positive sense of "nothing will happen."
So the true numbers are definitely higher than 0.0005%.
As I'm sure you're aware, the point of the numbers isn't to change the reality of the situation, it's to understand what that reality is. And the people that care are epidemiologists and leaders with decades of experience fighting pandemic after pandemic. You know, people that know what they're talking about? Might be worth trying to spend time actually learning how the people that are spending day and night trying to save your life and the lives of your loved ones do their work before dismissing them as useless.
The virus won't just bounce off your skin after one has got into you. If it is present everywhere in the environment then you will keep getting infected, maybe faster than your immune system can deal with it.
The end point to aim for in terms of immunity is that your immune system can deal with any level of viral load in the future without you becoming ill. Maybe combined with prophylactic drugs that help prevent the virus reproducing as fast within you.
https://www.nature.com/articles/d41587-020-00010-2
Don't know if this would show immunity though.
Heh, well that's great news. It's an antibody test so it would tell you who has had antibodies against the virus and thus who is probably immune to it.
I understand the need for a proper study, just surprised we're not seeing more on this.
The original report from Friday, auf Deutsch: https://www.spiegel.de/wissenschaft/medizin/coronavirus-gros...
However I want to caution anyone to expect a large proportion of the population to test positive in these tests.
We know that South Korea was able to contain the spread before a significant proportion of the population was infected. A important part of the strategy was contact tracing. For contact tracing to work, you need to detect at least 1-1/R of cases. If we assume that other measures such as masks and limited social distancing pushed R down to 2 and assume they were able to trace 100% of contacts, then we still have only twice as many that are infected. If nobody of the cases that are still ongoing dies, we look at a lower bound of the IFR of 0.8%.
Serological tests will tell you something about the early outbreak and how many cases where not detected and how high the infected fatality ratio really is.
I have seen many people that still believe that the IFR is in the range of an average flu. With the data from South Korea, we can put a number at what we can expect the IFR to be if we make some optimistic assumptions.
That is why I say that you should not expect that serological tests show that a large part of the population already had covid19 and that the IFR is much lower than we expect. They will show where exactly the IFR is.
Edit: And HN-style comments might make it easier to discuss the papers.
I'd put my finger that even random sorting would bring more value.
(Disclosure: I'm a volunteer working on Plaudit.)
Another commenter mentioned a projected call Plaudit, which seems to use the 'ORCiD' system to decide who is worth taking seriously. [0] Can anyone explain how that really works?
R00 that I typically use is about 2.1-2.6, however relating this to the 30-60% case increase per day is a difficult task.
If you know more information about where I could find these I would appreciate it.
On the days where I was still energetic enough to try and debunk some of the pseudosciences, I quickly discovered that one of the primary rules is absolutely never to trust one single article, however "scientific", and however reputable the publisher of the journal/conference is.
Otherwise one quickly becomes overwhelmed by the amount of crap published by otherwise "respectable" publishers (e.g. Elsevier's Journal of Homeopathy). Let's not even get into non-reviewed articles...
The only things that _may_ have any value whatsoever are the ones published in utmost respectable _journals_ themselves (e.g. Nature, but absolutely not on any of their spinoffs), and/or when there are very big labs/organizations in the authors list. Very big. I mean usually national pharmaceutical associations (obviously not from all states) or huge corporations (who will generally avoid putting their name on the line either way).
Meta-analyses most importantly do NOT have any intrinsinc value either _just because_ they are meta-analyses.
By reading bioXiv directly, you are most likely doing yourself a disservice. Unless you are a state agency or huge corporation having huge resources backing your which allows you to internally replicate whatever they claim.
If scientific papers did not contain any information then scientists would not read them.
But that is just the _very minimum_ you can expect in an article. Incredibly enough, by itself, that is already a huge filter. But sadly that is nowhere near enough to give any value whatsoever to the article. i.e. the article can be perfect and still just wrong. Even assuming good faith from the authors, they could just be the lucky 5% who got a positive out of an impossible thing and decided to publish. In a world with a general pressure to publish positive results and specially in a highly-hyped topic such as this one, you have a very high chance of seeing such a result.
At the end of the day this is the only argument I have left to debunk some homeopathy papers.
Scientists who are in a position where they have the resources to _reproduce_ the articles should and do read them. "Mortals" are not and reading these articles is a waste of time and potentially a disservice.
There are tens of journals claiming _utter non-sense_ being published every single month.
Using un-reviewed papers to estimate "consensus" is also pointless. Using publication numbers / citations / impact factors is slightly better but also extremely dangerous. If anything pseudo-science chaps are good at is at publishing large numbers of astonishing crap.
The sad truth is that filtering good science vs not is out of reach for the average layman, which leaves you with unscientific stuff such as "reputation" as your only tool.
>Scientists who are in a position where they have the resources to _reproduce_ the articles should and do read them.
It is typical for a scientist to read hundreds and hundreds of papers every year, or sometimes even every month. Are you suggesting that they reproduce every one of them? You may be dismayed to find out that typically, they would reproduce none of them, and when the system is working its best they might reproduce one of them. Under usual circumstances, you and they have exactly the same replication ability: the ability to search the literature for a replication.
In my opinion, you have the Science/Nature thing vs their spinoff exactly the other way around. Science/Nature only publish flashy things more for their "memeness" value than their actual scientific content. The first driver leading you to those journal is how much the paper can be talked about and picked up by other media outlets. In fact, the most important thing you need to publish in Science/Nature is NOT a good experiment/result, but being a coauthor of somebody who already has a lot of Science/Nature papers. Their spinoffs, on the other hand, being less driven by popularity, often offer more solid content.