Basically it comes down to 3 issues:
The accuracy, the amount of data communicated by a qualitative test, and is immunity durable.

here's what I believe

the IFR is 1%

solve for everything else based on this

panamamyers00 said:

---

Where are the videos telling us we shouldn't trust those genome sequences that say it didn't show up in the United States until February?

---

The genome sequences don't tell us that. I don't understand why you think it does. Genome sequences tell us mutation rates and can be used to trace infections back to the source.

We have sequence info the guy who traveled from Wuhan to Seattle area in mid Jan pre travel ban, so we know it was here in January. We have a sequence from someone who got sick about 6 weeks later and can tell from that sequence it was a descendent fo the first person, which is pretty strong evidence of community spread in the Seattle area for about 6 weeks between the two cases.

We didn't sequence virus samples anyone else in the US in Jan or Dec because we didn't identify anyone as having the virus. The lack of genetic sequence information isn't evidence of absence of the virus, and I can't imagine anyone who is remotely knowledgeable said "we don't have sequences from Dec so we know it wasn't here in Dec.

BiochemAg97 said:

---

panamamyers00 said:

---

Where are the videos telling us we shouldn't trust those genome sequences that say it didn't show up in the United States until February?

---

The genome sequences don't tell us that. I don't understand why you think it does. Genome sequences tell us mutation rates and can be used to trace infections back to the source.

We have sequence info the guy who traveled from Wuhan to Seattle area in mid Jan pre travel ban, so we know it was here in January. We have a sequence from someone who got sick about 6 weeks later and can tell from that sequence it was a descendent fo the first person, which is pretty strong evidence of community spread in the Seattle area for about 6 weeks between the two cases.

We didn't sequence virus samples anyone else in the US in Jan or Dec because we didn't identify anyone as having the virus. The lack of genetic sequence information isn't evidence of absence of the virus, and I can't imagine anyone who is remotely knowledgeable said "we don't have sequences from Dec so we know it wasn't here in Dec.

---

To be fair, the genomic study I presume is being referenced, did say with confidence it wasn't "circulating" in the USA in December. Now, it didn't say February arrival, either.

They made that statement with a combination of genetic sequencing and testing of samples from ARDS infections collected going back to October.

Everything lined up for them---the genetics, the samples within January and February, and the samples from 2019.

can you provide the link on the sample bias for the NY test?

I'd like to read the analysis.

TXAggie2011 said:

---

BiochemAg97 said:

---

panamamyers00 said:

---

Where are the videos telling us we shouldn't trust those genome sequences that say it didn't show up in the United States until February?

---

The genome sequences don't tell us that. I don't understand why you think it does. Genome sequences tell us mutation rates and can be used to trace infections back to the source.

We have sequence info the guy who traveled from Wuhan to Seattle area in mid Jan pre travel ban, so we know it was here in January. We have a sequence from someone who got sick about 6 weeks later and can tell from that sequence it was a descendent fo the first person, which is pretty strong evidence of community spread in the Seattle area for about 6 weeks between the two cases.

We didn't sequence virus samples anyone else in the US in Jan or Dec because we didn't identify anyone as having the virus. The lack of genetic sequence information isn't evidence of absence of the virus, and I can't imagine anyone who is remotely knowledgeable said "we don't have sequences from Dec so we know it wasn't here in Dec.

---

To be fair, the genomic study I presume is being referenced, did say with confidence it wasn't "circulating" in the USA in December.

They made that statement with a combination of genetic sequencing and testing of samples from ARDS infections collected going back to October.

Everything lined up for them---the genetics, the samples within January and February, and the samples from 2019.

---

Do you have a link to this study?

Testing ARDS samples going back to October would more likely be the evidence relied upon to show it wasn't here earlier.

Right. Take that "to be fair" as an important choice of language.

This is the Twitter thread that's been passed around a lot:

You can read multiple papers from Bedford at https://bedford.io/papers/

The paper discussing northern California is fascinating. And should get more discussion, and makes the point you're looking to draw regarding genomics.

Quote:

---

The early date and basal phylogenetic position of the WA1 lineage makes it likely that the direction of dissemination was from Washington State to other states; however that conclusion could change if further genomic sampling in the US revealed substantial levels of virus genetic diversity.

---

TXAggie2011 said:

---

Right. Take that "to be fair" as an important choice of language.

This is the Twitter thread that's been passed around a lot:

---

Fair. I can see where people would interpret that tweet that way.

Bedford is saying it wasn't circulating in the fall enough to have herd immunity. The Santa Clara antibody study that suggested up to 80x as many people have had it still works out to less than 5%, no where near herd immunity. Sweden is still weeks away from herd immunity which was an intentional goal. If they haven't gotten there, it is hard to believe that the US got to herd immunity by social isolation. Also, pretty impossible for most of California to have 50% of the population with antibodies, except this one city in the Bay Area that is at 5%. Nothing suggests the US has herd immunity. Even non-representative populations with a high risk of spread aren't showing herd immunity with antibody testing.

Btw, you can get to 80x with a larger R0 and without an earlier infection, and we know R0 is bigger than originally assumed. The initial guess of R0=2-3 is based on positive test results which were based on people with symptoms. The presence of asymptomatic people clearly shows that R0 is larger. How much larger depends on the proportion of asymptomatic cases.

Bedford also says the conclusion that all the genomic analysis points to a single entry with the Washington case and spread from there could change if additional sequences show something different. I would suspect that extensive NY genetic sequencing would show a path from Europe for some portion of cases.

Bottom line, it is unlikely it infected humans until late Nov or later. Thus it wasn't spreading in Cali in Oct. highly unlikely it was spreading in Cali in Nov, and fairly unlikely but possible it was in Cali in late Dec. We know we had it in the US in Jan, and apparent that someone infected from that initial infection spread it to Cali in late Jan.

BiochemAg97 said:

---

Keller6Ag91 said:

---

Duncan Idaho said:

---

Basically it comes down to 3 issues:
The accuracy, the amount of data communicated by a qualitative test, and is immunity durable.

---

Duncan,

Please explain to me the multiple points of data with Antibody testing that is the same with EVERY region thus far......all are showing a significantly larger community of people who've had the Wuhan virus and been asymptomatic (anywhere from 20x to 50x on average).

This tells me 2 things:
1. The death rate is SUBSTANTIALLY lower than we're been warned by the "experts".
2. This virus was in the US before Feb 20, likely by 2-3 months

---

Wow, you clearly didn't watch the video. The Santa Clara study which showed the 50x number had a number of positive results that was in the range of expected false positives at a 95% confidence level. They also didn't do a random sample. If all the positive results could have been false positives, then the result is meaningless. This is why preprint studies that haven't been peer reviewed aren't the best.


The NY study that Cuomo sighted that the media then extrapolated to a very high number of New Yorkers also was not a random sample. Cuomo himself acknowledged the sample bias. only testing from people going out to the store doesnt really give you numbers that can be scaled to the whole population. Those people that have been more strictly isolated have less exposure and likely less antibodies. Yet the media did exactly what they were warned not to do and took the state population and multiplied by the % of positive results.

---

Cuomo isn't a scientist. Why is it that only the more worrisome possibilities get the focus?

I think you can just as easily argue that someone who recently had COVID19 symptoms is much less likely to be out in public, so that population would be undersampled.

DTP02 said:

---

BiochemAg97 said:

---

Keller6Ag91 said:

---

Duncan Idaho said:

---

Basically it comes down to 3 issues:
The accuracy, the amount of data communicated by a qualitative test, and is immunity durable.

---

Duncan,

Please explain to me the multiple points of data with Antibody testing that is the same with EVERY region thus far......all are showing a significantly larger community of people who've had the Wuhan virus and been asymptomatic (anywhere from 20x to 50x on average).

This tells me 2 things:
1. The death rate is SUBSTANTIALLY lower than we're been warned by the "experts".
2. This virus was in the US before Feb 20, likely by 2-3 months

---

Wow, you clearly didn't watch the video. The Santa Clara study which showed the 50x number had a number of positive results that was in the range of expected false positives at a 95% confidence level. They also didn't do a random sample. If all the positive results could have been false positives, then the result is meaningless. This is why preprint studies that haven't been peer reviewed aren't the best.


The NY study that Cuomo sighted that the media then extrapolated to a very high number of New Yorkers also was not a random sample. Cuomo himself acknowledged the sample bias. only testing from people going out to the store doesnt really give you numbers that can be scaled to the whole population. Those people that have been more strictly isolated have less exposure and likely less antibodies. Yet the media did exactly what they were warned not to do and took the state population and multiplied by the % of positive results.

---

Cuomo isn't a scientist. Why is it that only the more worrisome possibilities get the focus?

I think you can just as easily argue that someone who recently had COVID19 symptoms is much less likely to be out in public, so that population would be undersampled.

---

Yeah, don't really think Cuomo thought up those limitations on his own. Probably read it from the report.

Ok, let's divide the population up into groups.
1) stayed home, haven't gone out, very little likelihood of exposure.
2) generally stay home, but still go out to the grocery store, have some exposure.
3) recently had symptoms and is now staying home for 14 days awaiting test results (or to be tested)

Do you really believe that more people are in group 3 than in group 1? Or even anywhere close?

BiochemAg97 said:

---

DTP02 said:

---

BiochemAg97 said:

---

Keller6Ag91 said:

---

Duncan Idaho said:

---

Basically it comes down to 3 issues:
The accuracy, the amount of data communicated by a qualitative test, and is immunity durable.

---

Duncan,

Please explain to me the multiple points of data with Antibody testing that is the same with EVERY region thus far......all are showing a significantly larger community of people who've had the Wuhan virus and been asymptomatic (anywhere from 20x to 50x on average).

This tells me 2 things:
1. The death rate is SUBSTANTIALLY lower than we're been warned by the "experts".
2. This virus was in the US before Feb 20, likely by 2-3 months

---

Wow, you clearly didn't watch the video. The Santa Clara study which showed the 50x number had a number of positive results that was in the range of expected false positives at a 95% confidence level. They also didn't do a random sample. If all the positive results could have been false positives, then the result is meaningless. This is why preprint studies that haven't been peer reviewed aren't the best.


The NY study that Cuomo sighted that the media then extrapolated to a very high number of New Yorkers also was not a random sample. Cuomo himself acknowledged the sample bias. only testing from people going out to the store doesnt really give you numbers that can be scaled to the whole population. Those people that have been more strictly isolated have less exposure and likely less antibodies. Yet the media did exactly what they were warned not to do and took the state population and multiplied by the % of positive results.

---

Cuomo isn't a scientist. Why is it that only the more worrisome possibilities get the focus?

I think you can just as easily argue that someone who recently had COVID19 symptoms is much less likely to be out in public, so that population would be undersampled.

---

Yeah, don't really think Cuomo thought up those limitations on his own. Probably read it from the report.

Ok, let's divide the population up into groups.
1) stayed home, haven't gone out, very little likelihood of exposure.
2) generally stay home, but still go out to the grocery store, have some exposure.
3) recently had symptoms and is now staying home for 14 days awaiting test results (or to be tested)

Do you really believe that more people are in group 3 than in group 1? Or even anywhere close?

---

When you add in that people under 18 were not sampled, I would guess that the numbers are quite close. It would either offset the oversampling of the "more active" (presumably less at-risk) population to a large extent, or possibly even surpass it.

The population of NY under 18 is larger than the population over 65. If you didn't sample anyone over 65 (unlikely) then not sampling anyone under 18 would more than wash that out. And the over 65 group is most of your lockdown quarantine crowd.

And, even if you argue that the population under 18 is less likely to be infected due to school closures, which is a questionable argument given likely social interactions, the application of the antibody data to IFR projections isn't going to change at all since the under 18 group has almost no fatalities.

I wish people would look as hard at both sides when trying to find holes in the data.

There are similar holes to be poked in the death projection numbers I frequently see thrown around.

DTP02 said:

---

BiochemAg97 said:

---

DTP02 said:

---

BiochemAg97 said:

---

Keller6Ag91 said:

---

Duncan Idaho said:

---

Basically it comes down to 3 issues:
The accuracy, the amount of data communicated by a qualitative test, and is immunity durable.

---

Duncan,

Please explain to me the multiple points of data with Antibody testing that is the same with EVERY region thus far......all are showing a significantly larger community of people who've had the Wuhan virus and been asymptomatic (anywhere from 20x to 50x on average).

This tells me 2 things:
1. The death rate is SUBSTANTIALLY lower than we're been warned by the "experts".
2. This virus was in the US before Feb 20, likely by 2-3 months

---

Wow, you clearly didn't watch the video. The Santa Clara study which showed the 50x number had a number of positive results that was in the range of expected false positives at a 95% confidence level. They also didn't do a random sample. If all the positive results could have been false positives, then the result is meaningless. This is why preprint studies that haven't been peer reviewed aren't the best.


The NY study that Cuomo sighted that the media then extrapolated to a very high number of New Yorkers also was not a random sample. Cuomo himself acknowledged the sample bias. only testing from people going out to the store doesnt really give you numbers that can be scaled to the whole population. Those people that have been more strictly isolated have less exposure and likely less antibodies. Yet the media did exactly what they were warned not to do and took the state population and multiplied by the % of positive results.

---

Cuomo isn't a scientist. Why is it that only the more worrisome possibilities get the focus?

I think you can just as easily argue that someone who recently had COVID19 symptoms is much less likely to be out in public, so that population would be undersampled.

---

Yeah, don't really think Cuomo thought up those limitations on his own. Probably read it from the report.

Ok, let's divide the population up into groups.
1) stayed home, haven't gone out, very little likelihood of exposure.
2) generally stay home, but still go out to the grocery store, have some exposure.
3) recently had symptoms and is now staying home for 14 days awaiting test results (or to be tested)

Do you really believe that more people are in group 3 than in group 1? Or even anywhere close?

---

When you add in that people under 18 were not sampled, I would guess that the numbers are quite close. It would either offset the oversampling of the "more active" (presumably less at-risk) population to a large extent, or possibly even surpass it.

The population of NY under 18 is larger than the population over 65. If you didn't sample anyone over 65 (unlikely) then not sampling anyone under 18 would more than wash that out. And the over 65 group is most of your lockdown quarantine crowd.

And, even if you argue that the population under 18 is less likely to be infected due to school closures, which is a questionable argument given likely social interactions, the application of the antibody data to IFR projections isn't going to change at all since the under 18 group has almost no fatalities.

I wish people would look as hard at both sides when trying to find holes in the data.

There are similar holes to be poked in the death projection numbers I frequently see thrown around.

---

Most kids live in family groups with parents. We would expect rapid spread from child to parent or parent to child in the same household. Also, we would expect parents that are hyper isolating to also isolate their kids, whereas parents more likely to go to the store are more likely to let their kids out of the house. Bottom line, we would expect kids to mirror their parents in antibody rates. It isn't like the kids have all been infected but somehow their parents we never exposed.

Over 65 often live separately in the US. In multigenerational households where kid, parents, grandparents live together, we would expect the over 65 to mirror the under 18. But when over 65 live separately from kids and working age population and the over 65 are isolating, we expect them to have lower exposure.

So if 18 mirrors the test age group, and the test group is biased to people with more risk of exposure, then the undercounted low risk group can still be a significant factor.

At the end of the day, the random sample puts a ceiling on the number with antibodies. You still haven't suggested a significant source of bias to the downside. At best you have suggested a balancing. And we could still have a significant population (over 65, plus the under 65 who have chosen complete isolation). I doubt it is half the population that is under-sampled and would guess maybe 25% at most could achieve a high level of isolation. That said, I know a lot of people who are isolating anything that comes into the house until hopefully,the virus dies.

I have suggested a significant source of bias to the downside, but you don't seem to want to turn the same critical eye in that direction. You have exclusively turned your analysis toward things which might indicate a bias to the upside, where every assumption is made in favor of an oversampling of positives (e.g. that a significant number of households with children are getting by without trips to the grocery store; that all infected kids have infected their parents).

Regarding those nursing homes being isolated, when they are now saying ~3,500 deaths in NY just among nursing home residents, with people still claiming that's an undercount, I'm not sure how isolated we can conclude they have been. Definitely hasn't been anything close to complete isolation.

Moreover, the primary discussion we are having relative to the antibody tests are the implications for the IFR, and undersampling minors absolutely exerts downward pressure to the true IFR vs the estimates being made from that test.

DTP02 said:

---

I have suggested a significant source of bias to the downside, but you don't seem to want to turn the same critical eye in that direction. You have exclusively turned your analysis toward things which might indicate a bias to the upside, where every assumption is made in favor of an oversampling of positives (e.g. that a significant number of households with children are getting by without trips to the grocery store; that all infected kids have infected their parents).

Regarding those nursing homes being isolated, when they are now saying ~3,500 deaths in NY just among nursing home residents, with people still claiming that's an undercount, I'm not sure how isolated we can conclude they have been. Definitely hasn't been anything close to complete isolation.

Moreover, the primary discussion we are having relative to the antibody tests are the implications for the IFR, and undersampling minors absolutely exerts downward pressure to the true IFR vs the estimates being made from that test.

---

I didn't say that there was a significant number of households with kids getting by without going to the grocery store. And given 1) kids typically don't have symptoms, and 2) parents living in close proximity are likely to be exposed, it is a little hard to figure how there would be this vast group of infected kids that didn't expose their parents.

And yes, some nursing homes have been hard hit and others have not. But there are also old people,who don't live in nursing homes that could be doing a good job of self isolation. It is possible that overall, the nursing home population is more exposed to the virus than the general population, if a significant proportion of nursing homes have all been hit. It would be interesting to know what portion of elderly in nursing homes in NY were in nursing homes that were infected. Then we could make an educated guess that those in infected nursing homes were exposed and those in nursing homes without COVID patients have likely not been exposed.

Turning to IFR, we are likely undercounting deaths. Also, given the delay between infection and death, there are infected people in the hospital that haven't died yet, but ultimately will, which is a further source of undercounting deaths.

If the deaths are an undercount and the antibody test is accurate or an over-count, then the 0.5 IFR reported based on the antibody study is a floor, and the IFR could be a little higher. I doubt it would be significantly higher, certainly not the 2-3 that was initially believed.

NY started counting home deaths as COVID if they had symptoms rather than requiring a positive test result. That could lead to an over-count of deaths. And if so, that could drive the IFR lower. While possible, I think the have not died might still be offsetting to the died of other causes but were mislabeled. As for undercounting the antibody positive cases, that would assume a large portion of the population that was more likely to have it but less likely to be at the grocery store. Those could be healthcare workers and people in infected nursing homes.

And the 0.5 number assumes the kids have it in line with the antibody testing. If we still had kids in school, I think it would be likely the kids had a higher exposure risk, but NY hasn't had kids in school in a while.

Bottom line, 0.5 is probably close to floor with the actual number possibly being slightly higher.

Also, as doctors learn how to better treat the virus and if some of the drug trials show significant improvements in outcomes, you could see the IFR drop significantly as more people who would have died are saved by medical intervention.