COVID-19 Megathread 4: Grandma Got Run Over by the Dow Jones

[GeorgiaModerate]

https://twitter.com/kellyswails/status/1248606674195230720

[Vaccinated Russian Bear]

Taking samples from one region is by definition not random. Even if it was a random subset of people within that region (nothing in what you posted suggests that is true by the way) you need to do a lot of work to ensure that accurately is applicable to other regions. E.g., if there are discrepancies in the way tests are administered or causes of death are assigned between Germany and Spain, then you couldn't generalize these results to Spain unless you had some sort of corrections. It's the same reason why you have limited inference in applying mortality estimates from Lombardy to mortality estimates in Montana.

At best this study gives you the estimate of mortality for sick people in one region of Germany, and that's under ideal conditions of how these people were included in the survey.

What? Germany is also just a "region". Whole Europe as well... So we should stop doing these studies? LMAO.

Obviously, it will vary depending on age, or nations health (US is for instance a younger, but more obese smoking? not running etc nation than Germany).

Unfortunately, the source is not working, but you can find cache in google https://www.land.nrw/sites/default/files/asset/document/zwischenergebnis_covid19_case_study_gangelt.pdf

Hintergrund: Die Gemeinde Gangelt ist in Deutschland einer der am stärksten vonCOVID19 betroffenen Orte Deutschlands. Es wird angenommen, dass dasInfektionsgeschehen auf eine Karnevalssitzung am 15. Februar 2020 zurückzuführenist, da mehrere Personen im Nachgang zu dieser Sitzung SARSCoV2 positiv getestetwurden. Die Karnevalssitzung und das Ausbruchgeschehen der Sitzung wird derzeitnoch genauer untersucht. Es wurde eine repräsentative Stichprobe aus der GemeindeGangelt (12.529 Einwohner) im Kreis Heinsberg gezogen. DieWeltgesundheitsorganisation (WHO) empfiehlt ein Protokoll, bei dem je nach zuerwartender Prävalenz stichprobenartig 100 bis 300 Haushalte untersucht werden.Diese Stichprobe wurde auf ihre Repräsentativität abgestimmt mit Herrn Prof. ManfredGüllner (Forsa).

Ziel: Das Ziel der Studie ist es, den Stand der durchgemachten und noch immerstattfindenden SARS-CoV2 Infektionen (Prozentsatz aller Infizierten) in der GemeindeGangelt zu bestimmen. Zusätzlich soll damit der Stand der derzeitigen SARS-CoV2Immunität ermittelt werden.

Vorgehen: Ein Serienbrief wurde an ca. 600 Haushalte verschickt. Insgesamt nahmenca. 1000 Einwohner aus ca. 400 Haushalten an der Studie teil. Es wurden Fragebögenerhoben, Rachenabstriche genommen und Blut auf das Vorliegen von Antikörper (IgG,IgA) getestet. In diese erste Auswertung gehen die Zwischenergebnisse undRückschlüsse von ca. 500 Personen ein.

Translation from here: https://www.reddit.com/r/medicine/comments/fxqszt/study_of_heinsberg_cluster_in_germany_suggests/fmvw144/

For context: Heinsberg is a municipality in Northrine-Westphalia. On February 15, a man who was unknowingly SARS-CoV-2 positive attended a large carnival event in the town of Gangelt, setting in motion the wide spread of SARS-CoV-2 within Heinsberg and the rest of the state.

Goal: The study aims to uncover the number of current and past SARS-CoV-2 infections in Gangelt with the intention of estimating the current level of immunity within the community.

Methods: A letter was sent to a representative sample of 600 households, asking them to participate in the study. 400 households, totaling 1000 people, chose to do so. Surveys were administered, pharyngeal swabs taken and antibody testing (IgG and IgA) was also done. 500 people were part of the group studied in this pre-publication.

Preliminary results: Present immunity was determined to be around 14% (anti-SARS-CoV-2 IgG positive, Specificity 99%). Roughly 2% of participants showed an ongoing infection as determined by PCR. The combined percentage of past and present infections was found to be 15%. The case fatality rate relative to the total number of infected detected in this study was 0.37%. Lethality for Germany as a whole as determined by John-Hopkins University currently stands at a number five times higher: 1.98%. Mortality relative to the total population of Gangelt was found to be 0.06%. [This last sentence previously said 0.15%, by which they probably meant mortality from all causes for whatever reason. The number has been updated in the link in the comments.]

They mention in German the WHO's recommendation about at least 300 randomly chosen households.

It's not perfect, obviously, it is small, it is preliminary, it is [insert other caveats], but it is the (only) random study we have. And, as WaPo mentions, it is "at the lower end of the 0.3 to 1 percent range that has been estimated by scientists", so the results are NOT sensational, but inside previously estimated interval.

[Vaccinated Russian Bear]

https://twitter.com/NateSilver538/status/1248618725990739970
https://twitter.com/NateSilver538/status/1248618728192786433

Thanks to Trump, US has got a 2 weeks of valuable time!
LEARN FROM EUROPE! 🇺🇸🇺🇸🇺🇸

[RI]

Apple and Google are building a coronavirus tracking system into iOS and Android

[RI]

https://twitter.com/kellyswails/status/1248606674195230720

Here's the thing, though: it could be either case (or somewhere in between), and we won't know with any degree of certainty until well after this is all over, which is a terrible position for a policymaker to be in.

[Skill and Chance]

https://twitter.com/kellyswails/status/1248606674195230720

Here's the thing, though: it could be either case (or somewhere in between), and we won't know with any degree of certainty until well after this is all over, which is a terrible position for a policymaker to be in.

You have to overreact to things that scale exponentially.   I'll bet Blockbuster wishes it had "overreacted" to Netflix when Netflix was a tiny startup.

[It’s so Joever]

https://twitter.com/kellyswails/status/1248606674195230720

Here's the thing, though: it could be either case (or somewhere in between), and we won't know with any degree of certainty until well after this is all over, which is a terrible position for a policymaker to be in.

You aren’t wrong, but given how the virus played out in the past and what is happening now in NYC, I think we still need to err on the side of caution.

[DINGO Joe]

https://twitter.com/kellyswails/status/1248606674195230720

Here's the thing, though: it could be either case (or somewhere in between), and we won't know with any degree of certainty until well after this is all over, which is a terrible position for a policymaker to be in.

Right now, there seems to be ample evidence that societies that had the ability/will to test, trace and isolate outperform those lacked it by wide margins.

[💥💥 brandon bro (he/him/his)]

Taking samples from one region is by definition not random. Even if it was a random subset of people within that region (nothing in what you posted suggests that is true by the way) you need to do a lot of work to ensure that accurately is applicable to other regions. E.g., if there are discrepancies in the way tests are administered or causes of death are assigned between Germany and Spain, then you couldn't generalize these results to Spain unless you had some sort of corrections. It's the same reason why you have limited inference in applying mortality estimates from Lombardy to mortality estimates in Montana.

At best this study gives you the estimate of mortality for sick people in one region of Germany, and that's under ideal conditions of how these people were included in the survey.

What? Germany is also just a "region". Whole Europe as well... So we should stop doing these studies? LMAO.

Nobody said we should stop doing these studies. I was saying unless you're limiting your scope of inference to the one city this survey was conducted in then this isn't a random sample. Of course these studies are worth doing; trying to generalize the results in one city to the US is just not good statistical practice. It fails for the same reason that trying to generalize the mortality data from NYC to the rest of the US would fail.

Hintergrund: Die Gemeinde Gangelt ist in Deutschland einer der am stärksten vonCOVID19 betroffenen Orte Deutschlands. Es wird angenommen, dass dasInfektionsgeschehen auf eine Karnevalssitzung am 15. Februar 2020 zurückzuführenist, da mehrere Personen im Nachgang zu dieser Sitzung SARSCoV2 positiv getestetwurden. Die Karnevalssitzung und das Ausbruchgeschehen der Sitzung wird derzeitnoch genauer untersucht. Es wurde eine repräsentative Stichprobe aus der GemeindeGangelt (12.529 Einwohner) im Kreis Heinsberg gezogen. DieWeltgesundheitsorganisation (WHO) empfiehlt ein Protokoll, bei dem je nach zuerwartender Prävalenz stichprobenartig 100 bis 300 Haushalte untersucht werden.Diese Stichprobe wurde auf ihre Repräsentativität abgestimmt mit Herrn Prof. ManfredGüllner (Forsa).

Ziel: Das Ziel der Studie ist es, den Stand der durchgemachten und noch immerstattfindenden SARS-CoV2 Infektionen (Prozentsatz aller Infizierten) in der GemeindeGangelt zu bestimmen. Zusätzlich soll damit der Stand der derzeitigen SARS-CoV2Immunität ermittelt werden.

Vorgehen: Ein Serienbrief wurde an ca. 600 Haushalte verschickt. Insgesamt nahmenca. 1000 Einwohner aus ca. 400 Haushalten an der Studie teil. Es wurden Fragebögenerhoben, Rachenabstriche genommen und Blut auf das Vorliegen von Antikörper (IgG,IgA) getestet. In diese erste Auswertung gehen die Zwischenergebnisse undRückschlüsse von ca. 500 Personen ein.

Translation from here: https://www.reddit.com/r/medicine/comments/fxqszt/study_of_heinsberg_cluster_in_germany_suggests/fmvw144/

For context: Heinsberg is a municipality in Northrine-Westphalia. On February 15, a man who was unknowingly SARS-CoV-2 positive attended a large carnival event in the town of Gangelt, setting in motion the wide spread of SARS-CoV-2 within Heinsberg and the rest of the state.

Goal: The study aims to uncover the number of current and past SARS-CoV-2 infections in Gangelt with the intention of estimating the current level of immunity within the community.

Methods: A letter was sent to a representative sample of 600 households, asking them to participate in the study. 400 households, totaling 1000 people, chose to do so. Surveys were administered, pharyngeal swabs taken and antibody testing (IgG and IgA) was also done. 500 people were part of the group studied in this pre-publication.

Preliminary results: Present immunity was determined to be around 14% (anti-SARS-CoV-2 IgG positive, Specificity 99%). Roughly 2% of participants showed an ongoing infection as determined by PCR. The combined percentage of past and present infections was found to be 15%. The case fatality rate relative to the total number of infected detected in this study was 0.37%. Lethality for Germany as a whole as determined by John-Hopkins University currently stands at a number five times higher: 1.98%. Mortality relative to the total population of Gangelt was found to be 0.06%. [This last sentence previously said 0.15%, by which they probably meant mortality from all causes for whatever reason. The number has been updated in the link in the comments.]

They mention in German the WHO's recommendation about at least 300 randomly chosen households.

It's not perfect, obviously, it is small, it is preliminary, it is [insert other caveats], but it is the (only) random study we have. And, as WaPo mentions, it is "at the lower end of the 0.3 to 1 percent range that has been estimated by scientists", so the results are NOT sensational, but inside previously estimated interval.

Okay. I have read the translation you posted. Thoughts:

- There is absolutely no information about how the 1000 households the survey was sent to were selected. Claiming this is "random" without any further information is abusive and misleading.
- You have to assume the probability of responding to the survey is independent of all other factors, e.g., age, family size, and most importantly, infection status. This is the same reason why many political polls conducted via phone are biased.
- There is no detail about how averaging was done here. Was this done at the household level or at the individual level? If it was done at the individual level, then it would be way too confident about the uncertainty around the estimate because of the non-independence of people within a household, not to mention the fact that there could be differences among households.
- 500 is a decent but not excellent sample size; the effective sample size is smaller if you consider the groupings within households.
- With 500 survey participants and an infection rate of 15%, you have ~75 infected people. This is a tiny, tiny sample size for determining mortality. An estimated mortality rate of 0.34%, you likely had only one or two infected survey respondents die. The uncertainty there is going to be massive.
- The estimate is also incomplete because 2% of survey respondents are still infected
- There's no estimate of uncertainty around this 0.34% number! For a sample this small it's going to be tiny.

So even if the survey respondents were chosen independently (there is no evidence this is true), and even if the people who responded to the survey who are infected are reflective of a random sample of people in this city, the sample size is laughably small and uncertainty in the estimate would be massive.

[GP270watch]

https://twitter.com/LucasFoxNews/status/1247931506250723328

That's crazy, especially since effective social distancing measures are happening. This tracks very much with reports out of France, Italy, and Spain where they're having an inceased number of deaths that can only be explained by Covid-19.

[Donald Trump’s Toupée]

I think that most experts are overestimating the amount of social distancing going on, too. Of course with a partial shutdown social distancing is going to naturally happen, but I really don’t think it’s social distancing when everyone is panicking together in the supermarket.

Additionally, Cuomo and Blasio were gravely underprepared.

[President Punxsutawney Phil]

If the number of fatalities in people below 25 is that low, then from a policy perspective its functionally 0.

Not if you're a governor deciding whether to re-open your state's schools...

Only in a small, small state. 5 is a very tiny number compared to how many people attend the school system.

[GeorgiaModerate]

https://twitter.com/AGHamilton29/status/1248737943130451971

[It’s so Joever]

If the number of fatalities in people below 25 is that low, then from a policy perspective its functionally 0.

Not if you're a governor deciding whether to re-open your state's schools...

Only in a small, small state. 5 is a very tiny number compared to how many people attend the school system.

If you really think only five kids would die if we reopen schools, I have an ambulance-filled bridge in Brooklyn to sell you.

[#TheShadowyAbyss]

Multiple counties here in Florida are now mandating face masks, and we're now doing rapid testing in my city

[Landslide Lyndon]

OMFG! The guy is an absolute retard.

https://twitter.com/joshtpm/status/1248699275355328512

[GeorgiaModerate]

OMFG! The guy is an absolute retard.

https://twitter.com/joshtpm/status/1248699275355328512

I can just picture some of the FNC talking heads now: "Why don't we TRY antibiotics against a virus?  How do we KNOW they don't work?  What do we have to lose?"

[#TheShadowyAbyss]

Today has been the deadliest day in the U.S with 2,013 deaths recorded so far today

[GP270watch]

 Trump is dumb, I've been saying this. It's pretty amazing how the dumb label hasn't stuck to Trump.

[Progressive Pessimist]

OMFG! The guy is an absolute retard.

https://twitter.com/joshtpm/status/1248699275355328512

How does a well-known germaphobe not know this!? You don't need to be a medical professional to understand the basic difference between pathogens and the means of addressing them!

We've always been f***ed with this nincompoop in charge of our executive branch, but it somehow just keeps getting even more ridiculous and even more concerning. When will we hit the bedrock?! Is it even possible?!

[It’s so Joever]

To be honest, I am just glad he doesn’t think that antibiotics can destroy the virus. (Which would be far, far, far worse than this moment of stupidity)
My standards are so damn low at this point because I have resigned to the fact that we have a dumbf*** in office.

[It’s so Joever]

Also, Trump is a quarter right, in the fact that antibiotic resistant bacteria are a real problem. If Trump wants to promote more long term antibiotic research (when this mega disaster is over) I would be all for it.

[GeorgiaModerate]

Italy has extended its nationwide lockdown until May 3.

[brucejoel99]

Italy has extended its nationwide lockdown until May 3.

Given recent news, this may be the last extension.

[emailking]

https://twitter.com/kellyswails/status/1248606674195230720

It's like the people who complained when nothing happened on Y2K.