SARS-CoV-2 testing

Covid-19 discussion, bring your own statistics
User avatar
shpalman
Dorkwood
Posts: 1146
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: SARS-CoV-2 testing

Post by shpalman » Sat Apr 18, 2020 8:40 am

https://www.theguardian.com/world/2020/ ... ly-thought
At the time of the [not yet peer reviewed] study, Santa Clara county had 1,094 confirmed cases of Covid-19, resulting in 50 deaths. But based on the rate of people who have antibodies, it is likely that between 48,000 and 81,000 people had been infected in Santa Clara county by early April – a number approximately 50 to 80 times higher.
It's based on samples from 3330 people.

The population of the county is about 1.8 million.

YOU DON'T HAVE HERD IMMUNITY IF NOT EVEN 5% HAVE HAD IT.
molto tricky

User avatar
Woodchopper
Dorkwood
Posts: 1170
Joined: Sat Oct 12, 2019 9:05 am

Re: SARS-CoV-2 testing

Post by Woodchopper » Sat Apr 18, 2020 1:58 pm

shpalman wrote:
Sat Apr 18, 2020 8:40 am
https://www.theguardian.com/world/2020/ ... ly-thought
At the time of the [not yet peer reviewed] study, Santa Clara county had 1,094 confirmed cases of Covid-19, resulting in 50 deaths. But based on the rate of people who have antibodies, it is likely that between 48,000 and 81,000 people had been infected in Santa Clara county by early April – a number approximately 50 to 80 times higher.
It's based on samples from 3330 people.

The population of the county is about 1.8 million.

YOU DON'T HAVE HERD IMMUNITY IF NOT EVEN 5% HAVE HAD IT.
If we take the mid point between 48 000 and 81 000 we get 64 500. 50 deaths gives us a case fatality rate of 0.0775%. Which is lower than influenza.

However, the population of Lombardia is circa 10 million, and 11 851 have died of Covid-19 already, which gives us a fatality rate of 0.1178. So unless there is massive over reporting, the case fatality rate of Covid-19 can't be lower than 0.1178%, and it could only be that low if everyone in the region has been infected already (and that's very very unlikely given that they keep discovering new infections).

I'll assume that the Santa Clara sample wasn't representative of the population as a whole.

ETA the Guardian article states: "Stanford researchers said their findings show a death rate of just 0.12% to 0.2%." But I don't see how they got that from 50 deaths out of between 48 000 and 81 000 cases.

User avatar
shpalman
Dorkwood
Posts: 1146
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: SARS-CoV-2 testing

Post by shpalman » Sat Apr 18, 2020 10:28 pm

According to SciBabe
volunteers were preened from Facebook ads that looked for people who wanted to get tested
molto tricky

User avatar
Woodchopper
Dorkwood
Posts: 1170
Joined: Sat Oct 12, 2019 9:05 am

Re: SARS-CoV-2 testing

Post by Woodchopper » Sun Apr 19, 2020 6:53 am

shpalman wrote:
Sat Apr 18, 2020 10:28 pm
According to SciBabe
volunteers were preened from Facebook ads that looked for people who wanted to get tested
Possible selection bias there. Volunteers might well be people who thought they might have been infected.

EACLucifer
Fuzzable
Posts: 307
Joined: Fri Dec 13, 2019 7:49 am
Location: Behind you

Re: SARS-CoV-2 testing

Post by EACLucifer » Sun Apr 19, 2020 9:25 am

I think that massive selection bias is almost inevitable in such a situation ;)

User avatar
Woodchopper
Dorkwood
Posts: 1170
Joined: Sat Oct 12, 2019 9:05 am

Re: SARS-CoV-2 testing

Post by Woodchopper » Sun Apr 19, 2020 12:18 pm

Recent open access NEJM article on the screening in Iceland.

As part of the sample of 10 787 people who were part of the 'population screening' there were only 87 positives (or 0.8%).

The sample was self selected and a bit under a half had shown symptoms or had traveled to an affected country. So it looks like there wasn't a large number of asymptomatic cases in the general population.

User avatar
Woodchopper
Dorkwood
Posts: 1170
Joined: Sat Oct 12, 2019 9:05 am

Re: SARS-CoV-2 testing

Post by Woodchopper » Mon Apr 20, 2020 8:22 am

Woodchopper wrote:
Sat Apr 18, 2020 1:58 pm
shpalman wrote:
Sat Apr 18, 2020 8:40 am
https://www.theguardian.com/world/2020/ ... ly-thought
At the time of the [not yet peer reviewed] study, Santa Clara county had 1,094 confirmed cases of Covid-19, resulting in 50 deaths. But based on the rate of people who have antibodies, it is likely that between 48,000 and 81,000 people had been infected in Santa Clara county by early April – a number approximately 50 to 80 times higher.
It's based on samples from 3330 people.

The population of the county is about 1.8 million.

YOU DON'T HAVE HERD IMMUNITY IF NOT EVEN 5% HAVE HAD IT.
If we take the mid point between 48 000 and 81 000 we get 64 500. 50 deaths gives us a case fatality rate of 0.0775%. Which is lower than influenza.

However, the population of Lombardia is circa 10 million, and 11 851 have died of Covid-19 already, which gives us a fatality rate of 0.1178. So unless there is massive over reporting, the case fatality rate of Covid-19 can't be lower than 0.1178%, and it could only be that low if everyone in the region has been infected already (and that's very very unlikely given that they keep discovering new infections).

I'll assume that the Santa Clara sample wasn't representative of the population as a whole.

ETA the Guardian article states: "Stanford researchers said their findings show a death rate of just 0.12% to 0.2%." But I don't see how they got that from 50 deaths out of between 48 000 and 81 000 cases.
Another estimate here, from the Centre for Evidence-Based Medicine:
Taking account of historical experience, trends in the data, increased number of infections in the population at largest, and potential impact of misclassification of deaths gives a presumed estimate for the COVID-19 IFR [Infection Fatality Rate] somewhere between 0.1% and 0.36%.*
Again, the lower band only applies if the whole of Lombardy was infected.

User avatar
Woodchopper
Dorkwood
Posts: 1170
Joined: Sat Oct 12, 2019 9:05 am

Re: SARS-CoV-2 testing

Post by Woodchopper » Mon Apr 20, 2020 8:49 am

shpalman wrote:
Sat Apr 18, 2020 8:40 am
https://www.theguardian.com/world/2020/ ... ly-thought
At the time of the [not yet peer reviewed] study, Santa Clara county had 1,094 confirmed cases of Covid-19, resulting in 50 deaths. But based on the rate of people who have antibodies, it is likely that between 48,000 and 81,000 people had been infected in Santa Clara county by early April – a number approximately 50 to 80 times higher.
It's based on samples from 3330 people.

The population of the county is about 1.8 million.

YOU DON'T HAVE HERD IMMUNITY IF NOT EVEN 5% HAVE HAD IT.
Andrew Gelman thinks the authors of the study should apologize for wasting everyone's time.

User avatar
shpalman
Dorkwood
Posts: 1146
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: SARS-CoV-2 testing

Post by shpalman » Mon Apr 20, 2020 9:34 am

On the other hand it seems that 49% of the population of 456 people tested in Ortisei have antibodies for the covids, possibly thanks to all the morons from around here who treated the closure of the schools as a skiing holiday. (Two thirds of the people tested said they never had symptoms. The doctor doing the tests did have symptoms, and severe ones at that.)

456 people (ages between 20 and 59) does correspond to about 10% of the population of the town though. I can't obviously find figures for the official number of cases and deaths in that town, but in the whole South Tyrol region (population: half a million) it's something like 2300 cases and 234 deaths.

Bonus points for the use of the expression "partial herd immunity". Just like your car would be "partially watertight" I suppose.

https://www.ildolomiti.it/cronaca/2020/ ... e-e-laives seems to be from about 5 days ago and says there had been 10 deaths in Ortisei.
molto tricky

User avatar
Woodchopper
Dorkwood
Posts: 1170
Joined: Sat Oct 12, 2019 9:05 am

Re: SARS-CoV-2 testing

Post by Woodchopper » Mon Apr 20, 2020 12:12 pm

Woodchopper wrote:
Mon Apr 20, 2020 8:49 am
shpalman wrote:
Sat Apr 18, 2020 8:40 am
https://www.theguardian.com/world/2020/ ... ly-thought
At the time of the [not yet peer reviewed] study, Santa Clara county had 1,094 confirmed cases of Covid-19, resulting in 50 deaths. But based on the rate of people who have antibodies, it is likely that between 48,000 and 81,000 people had been infected in Santa Clara county by early April – a number approximately 50 to 80 times higher.
It's based on samples from 3330 people.

The population of the county is about 1.8 million.

YOU DON'T HAVE HERD IMMUNITY IF NOT EVEN 5% HAVE HAD IT.
Andrew Gelman thinks the authors of the study should apologize for wasting everyone's time.
And here's the link: https://statmodeling.stat.columbia.edu/ ... revalence/

User avatar
Woodchopper
Dorkwood
Posts: 1170
Joined: Sat Oct 12, 2019 9:05 am

Re: SARS-CoV-2 testing

Post by Woodchopper » Mon Apr 20, 2020 12:36 pm

Looks like the CDC is doing a proper study on the prevalence of Covid. Hopefully the results will be out soon.
https://www.cdc.gov/coronavirus/2019-nc ... sting.html

User avatar
Woodchopper
Dorkwood
Posts: 1170
Joined: Sat Oct 12, 2019 9:05 am

Re: SARS-CoV-2 testing

Post by Woodchopper » Mon Apr 20, 2020 3:01 pm

Looks like a proper random sample from Austria: https://www.sora.at/uploads/media/Austr ... ersion.pdf
Estimate of period prevalence
Prevalence is the occurrence of a disease in relation to an entire population. This study makes it possible to estimate the prevalence of acute infections with COVID-19 (“Corona Virus”) among non-hospitalized people living in Austria for the period earlyApril 2020. The proportion of positively tested in the weighted sample is 0.33%.This proportion represents about 28,500 people among the population.

Confidence interval(95%)If a sample survey draws a conclusion on a population, the confidence interval must always be observed. For this study, the generally accepted principle was appliedthat results should be within the stated interval with 95% certainty.Applying the Clopper-Pearson interval method, we find that the prevalence of COVID-19 in Austrian households is 95% likely to be between 0.12 and 0.76%.In absolute terms: In addition to the patients in hospitals, there were between 10,200 and 67,400 people acutely infected with COVID-19 in the period April1-6.
ETA testing was between 1-6 April, and on 6 April there had been 12 206 cases identified via testing. So it looks like testing identified about half of the infections (if we take the 28 500 figure).

User avatar
Woodchopper
Dorkwood
Posts: 1170
Joined: Sat Oct 12, 2019 9:05 am

Re: SARS-CoV-2 testing

Post by Woodchopper » Tue Apr 21, 2020 5:01 am

Not a random sample but better than a Facebook ad.
About 3 percent of blood donors in the Netherlands have developed antibodies against Covid-19, the respiratory disease caused by the coronavirus, RIVM director Jaap van Dissel told the Tweede Kamer based on an as-yet-unpublished study by blood bank Sanquin, NU.nl reports.
https://nltimes.nl/2020/04/16/3-dutch-b ... antibodies

User avatar
Woodchopper
Dorkwood
Posts: 1170
Joined: Sat Oct 12, 2019 9:05 am

Re: SARS-CoV-2 testing

Post by Woodchopper » Wed Apr 22, 2020 9:45 am

An odd survey from Sweden, which states that between 26 March and 3 April 2.5% of people were currently infected, based upon self reported symptoms (including very mild symptoms).
https://www.folkhalsomyndigheten.se/nyh ... stockholm/ (in Swedish)

However, there must be lots of false positives, given that people will also be getting colds, flu and allergic reactions.

User avatar
Woodchopper
Dorkwood
Posts: 1170
Joined: Sat Oct 12, 2019 9:05 am

Re: SARS-CoV-2 testing

Post by Woodchopper » Wed Apr 22, 2020 9:53 am

Woodchopper wrote:
Wed Apr 22, 2020 9:45 am
An odd survey from Sweden, which states that between 26 March and 3 April 2.5% of people were currently infected, based upon self reported symptoms (including very mild symptoms).
https://www.folkhalsomyndigheten.se/nyh ... stockholm/ (in Swedish)

However, there must be lots of false positives, given that people will also be getting colds, flu and allergic reactions.
The survey is used as the basis for this piece of modelling though
https://www.folkhalsomyndigheten.se/pub ... pril-2020/ (also in Swedish).

User avatar
shpalman
Dorkwood
Posts: 1146
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: SARS-CoV-2 testing

Post by shpalman » Wed Apr 22, 2020 4:39 pm

Como is going to get a drive-in swab test centre for health workers and anyone else for whom it's deemed necessary.

Antibody tests might also arrive here in about a week.
molto tricky

User avatar
Woodchopper
Dorkwood
Posts: 1170
Joined: Sat Oct 12, 2019 9:05 am

Re: SARS-CoV-2 testing

Post by Woodchopper » Thu Apr 23, 2020 10:25 pm

Preliminary results of statewide antibody testing in New York: https://twitter.com/nygovcuomo/status/1 ... 38338?s=21

User avatar
shpalman
Dorkwood
Posts: 1146
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: SARS-CoV-2 testing

Post by shpalman » Thu Apr 23, 2020 11:09 pm

Woodchopper wrote:
Thu Apr 23, 2020 10:25 pm
Preliminary results of statewide antibody testing in New York: https://twitter.com/nygovcuomo/status/1 ... 38338?s=21
The results of almost all of these are (a) it's much worse than you thought and (b) but not enough to bring down the relevant fatality rate to 'flu-like levels or give any useful herd immunity.

(The infection rate looks like about ten times higher than the official number of cases.)
molto tricky

User avatar
jimbob
Catbabel
Posts: 843
Joined: Mon Nov 11, 2019 4:04 pm
Location: High Peak/Manchester

Re: SARS-CoV-2 testing

Post by jimbob » Fri Apr 24, 2020 10:07 am

shpalman wrote:
Thu Apr 23, 2020 11:09 pm
Woodchopper wrote:
Thu Apr 23, 2020 10:25 pm
Preliminary results of statewide antibody testing in New York: https://twitter.com/nygovcuomo/status/1 ... 38338?s=21
The results of almost all of these are (a) it's much worse than you thought and (b) but not enough to bring down the relevant fatality rate to 'flu-like levels or give any useful herd immunity.

(The infection rate looks like about ten times higher than the official number of cases.)
Yup, and has been consistent for a couple of months at least
Have you considered stupidity as an explanation

User avatar
Woodchopper
Dorkwood
Posts: 1170
Joined: Sat Oct 12, 2019 9:05 am

Re: SARS-CoV-2 testing

Post by Woodchopper » Fri Apr 24, 2020 3:53 pm

shpalman wrote:
Thu Apr 23, 2020 11:09 pm
Woodchopper wrote:
Thu Apr 23, 2020 10:25 pm
Preliminary results of statewide antibody testing in New York: https://twitter.com/nygovcuomo/status/1 ... 38338?s=21
The results of almost all of these are (a) it's much worse than you thought and (b) but not enough to bring down the relevant fatality rate to 'flu-like levels or give any useful herd immunity.

(The infection rate looks like about ten times higher than the official number of cases.)
Antibody testing gives an infection fatality rate of 1-2%
https://twitter.com/bendygardiner/statu ... 28864?s=21

That’s similar to my back of the envelope number for New York City based on the above.

Enormously more dangerous than influenza: https://twitter.com/trvrb/status/125343 ... 05793?s=21

User avatar
shpalman
Dorkwood
Posts: 1146
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: SARS-CoV-2 testing

Post by shpalman » Fri Apr 24, 2020 4:30 pm

It is of course vital to have good data, but it makes you realize how SARS-CoV-2 is such a perfect storm of bad news: it's deadly enough to show up in the mortality data of entire countries but can still sneak around being extremely contagious in mild, asymptomatic, or pre-symptomatic patients. And then patients with wrecked lungs don't even notice that they have less and less oxygen in their blood until suddenly they can't get up.

If 15% of the population have had it then there's still potential for the situation to get about 4-5 times worse than it currently is.

If there were fewer undetected cases, it would be much harder to randomly catch it.

If there were more undetected cases, we would have all had it by now already and the pandemic would peak on its own.

If it were less deadly it wouldn't be overwhelming health services.

If it were more deadly it wouldn't have seemed like "just the 'flu" at the beginning.

It's not obvious to me that it should evolve to get milder over time or that it should be seasonal.

Antibody testing (and regular swabbing) will be essential in any phase 2 but we're not even sure how much long term immunity you'll have even after you recover.
molto tricky

bmforre
Clardic Fug
Posts: 194
Joined: Mon Nov 11, 2019 1:15 pm
Location: Trondheim

Re: SARS-CoV-2 testing

Post by bmforre » Sat Apr 25, 2020 1:16 pm

Norway will soon test almost everyone
Report for general readership, in Norwegian.

Our university NTNU recently rolled out a fast and efficient test method and have already produced reagents enough for large scale testing and started this. Production is being stepped up and infrastructure is being prepared so testing can take place in local environments all over Norway and soon be available to everyone. Planned ca. end of May.

Leader of the team that did the very rapid development is Magnar Bjoras.

I've been tested and was not infected.

Herainestold
Fuzzable
Posts: 313
Joined: Mon Nov 25, 2019 1:23 pm

Re: SARS-CoV-2 testing

Post by Herainestold » Sat Apr 25, 2020 2:36 pm

bmforre wrote:
Sat Apr 25, 2020 1:16 pm
Norway will soon test almost everyone
Report for general readership, in Norwegian.

Our university NTNU recently rolled out a fast and efficient test method and have already produced reagents enough for large scale testing and started this. Production is being stepped up and infrastructure is being prepared so testing can take place in local environments all over Norway and soon be available to everyone. Planned ca. end of May.

Leader of the team that did the very rapid development is Magnar Bjoras.

I've been tested and was not infected.
Very good. What kind of test is it?

User avatar
Pucksoppet
Snowbonk
Posts: 599
Joined: Mon Nov 11, 2019 8:13 pm
Location: Girdling the Earth

Re: SARS-CoV-2 testing

Post by Pucksoppet » Sat Apr 25, 2020 9:14 pm

Herainestold wrote:
Sat Apr 25, 2020 2:36 pm
bmforre wrote:
Sat Apr 25, 2020 1:16 pm
Norway will soon test almost everyone
Report for general readership, in Norwegian.

Our university NTNU recently rolled out a fast and efficient test method and have already produced reagents enough for large scale testing and started this. Production is being stepped up and infrastructure is being prepared so testing can take place in local environments all over Norway and soon be available to everyone. Planned ca. end of May.

Leader of the team that did the very rapid development is Magnar Bjoras.

I've been tested and was not infected.
Very good. What kind of test is it?
Detection of viral RNA
Testen skal gi svar på om vi har viruset i oss. Det er ikke snakk om å finne ut hvem som har immunitet, men hvem som har viruset.
Loosely translated:
The test shall give an answer on whether we have virus in us. It is not saying whether to find out who that has immunity, but who that has the virus.
Etter at en prøve er tatt fra pasientens nese eller svelg, sendes den til et laboratorium for analyse. Det er her totrinnsmetoden til NTNU starter.

Første trinn er å blande prøven med en kjemisk løsning. Da blir overflaten av viruset ødelagt. Slik frigjøres arvestoffet og blir liggende og sveve fritt i væsken.
Andre trinn er å skille ut arvestoffet, slik at det kan analyseres. Noen ørsmå magnetiske kuler har blitt utstyrt med et tynt lag av et stoff som binder seg til arvestoffet. Disse tas så opp i løsningen. Etter en stund kan de ved hjelp av en magnet på utsiden av reagensglasset, trekke ut kulene med arvestoffet.
Loosely translated:
After a sample is taken from the patients nose or ?saliva, it is sent to a laboratory for analysis. This is here the two-stage method used by NTNU starts.
The first stage is to mix the sample with a chemical solution. In this way is the surface of the virus damaged/destroyed. In this manner the RNA is released and lies and ?floats freely i the liquid.
The second stage is to rinse out the RNA, so that it can be analysed. Some very small magnetic balls have been fitted with a thin layer of a material that binds itself to RNA. These takes it up in the solution. After a while can one, with help of a magnet outside the reagent glassware, pull out the balls with the RNA

User avatar
Woodchopper
Dorkwood
Posts: 1170
Joined: Sat Oct 12, 2019 9:05 am

Re: SARS-CoV-2 testing

Post by Woodchopper » Wed Apr 29, 2020 7:32 am

New Imperial College report on testing as part of Covid control
https://www.imperial.ac.uk/mrc-global-i ... 6-testing/
The World Health Organization has called for increased molecular testing in response to the COVID-19 pandemic, but different countries have taken very different approaches. We used a simple mathematical model to investigate the potential effectiveness of alternative testing strategies for COVID-19 control. Weekly screening of healthcare workers (HCWs) and other at-risk groups using PCR or point-of-care tests for infection irrespective of symptoms is estimated to reduce their contribution to transmission by 25-33%, on top of reductions achieved by self-isolation following symptoms. Widespread PCR testing in the general population is unlikely to limit transmission more than contact-tracing and quarantine based on symptoms alone, but could allow earlier release of contacts from quarantine. Immunity passports based on tests for antibody or infection could support return to work but face significant technical, legal and ethical challenges. Testing is essential for pandemic surveillance but its direct contribution to the prevention of transmission is likely to be limited to patients, HCWs and other high-risk groups.

Post Reply