[BostonCard]

In addition, plasma transfusions from recovered patients may be an effective aid to patients currently struggling with the virus.  That would be a great way to help with the most severe cases.

Also, eventually could be used prophylactically to healthcare workers on the front line to give them some protection.

BC

[BostonCard]

In addition, plasma transfusions from recovered patients may be an effective aid to patients currently struggling with the virus.  That would be a great way to help with the most severe cases.

Also, eventually could be used prophylactically to healthcare workers on the front line to give them some protection.

BC

[OutsiderFan]

THIS is a BIG deal!

I wonder how long Abbott Labs has been working on developing this test.

https://www.bloomberg.com/news/articles/...t-anywhere

https://www.youtube.com/watch?v=AAi5ErSjNvY

https://www.youtube.com/watch?v=MqICkkOaNKo

[OutsiderFan]

THIS is a BIG deal!

I wonder how long Abbott Labs has been working on developing this test.

https://www.bloomberg.com/news/articles/...t-anywhere

https://www.youtube.com/watch?v=AAi5ErSjNvY

https://www.youtube.com/watch?v=MqICkkOaNKo

[magnus]

2 minute test (administered like a glucose test) approved by FDA

https://www.reuters.com/article/us-healt...SKBN21I1T7

[magnus]

2 minute test (administered like a glucose test) approved by FDA

https://www.reuters.com/article/us-healt...SKBN21I1T7

[BostonCard]

THIS is a BIG deal!



I wonder how long Abbott Labs has been working on developing this test.

Well, the underlying machines have been around a while; it was just a matter to develop a coronavirus test that worked with the system.  For example, Cepheid's GeneXPert system was most well known as a rapid diagnostic for TB, that was modified to recognize the genetic code of SARS-CoV instead of the genetic code for TB.  I assume Abbott's test is similar.

Designing the kits is a lot easier than designing the machinery to run the tests.

BC

[BostonCard]

THIS is a BIG deal!



I wonder how long Abbott Labs has been working on developing this test.

Well, the underlying machines have been around a while; it was just a matter to develop a coronavirus test that worked with the system.  For example, Cepheid's GeneXPert system was most well known as a rapid diagnostic for TB, that was modified to recognize the genetic code of SARS-CoV instead of the genetic code for TB.  I assume Abbott's test is similar.

Designing the kits is a lot easier than designing the machinery to run the tests.

BC

[Goose]

Take a look at https://www.nytimes.com/2020/04/01/well/...ative.html
The CDC test hast a negative control and a positive control for the overall test run in the PCR machine. Ir also has (at least in theory) a control to make sure the sample didn't get "lost" in the prep. I naturally have no idea what the "Chinese" test controls were. However, it is really, really important to realize the tests we are using, and that everybody else is using, while possibly different, have not been "validated". In the US they are approved on an emergency basis. We don't know what their sensitivity is. We don't know for sure that anybody with the disease will have adequate virus titers in the upper respiratory tract for the test(s) to detect it.
This is nobody's "fault". This is a novel virus. There were no calibration runs. If there is a 30% false negative (or worse) level, thank goodness we are treating everybody that has symptoms as if they were positive.

[Goose]

Take a look at https://www.nytimes.com/2020/04/01/well/...ative.html
The CDC test hast a negative control and a positive control for the overall test run in the PCR machine. Ir also has (at least in theory) a control to make sure the sample didn't get "lost" in the prep. I naturally have no idea what the "Chinese" test controls were. However, it is really, really important to realize the tests we are using, and that everybody else is using, while possibly different, have not been "validated". In the US they are approved on an emergency basis. We don't know what their sensitivity is. We don't know for sure that anybody with the disease will have adequate virus titers in the upper respiratory tract for the test(s) to detect it.
This is nobody's "fault". This is a novel virus. There were no calibration runs. If there is a 30% false negative (or worse) level, thank goodness we are treating everybody that has symptoms as if they were positive.

[magnus]

2 minute test (administered like a glucose test) approved by FDA

https://www.reuters.com/article/us-healt...SKBN21I1T7

Nope,  not authorized. 

https://www.cnn.com/2020/04/02/health/co...index.html

[magnus]

2 minute test (administered like a glucose test) approved by FDA

https://www.reuters.com/article/us-healt...SKBN21I1T7

Nope,  not authorized. 

https://www.cnn.com/2020/04/02/health/co...index.html

[burger]

WSJ has an article today saying that tests being used in the US has a 1/3 false negative rate.  Article is paywalled--I just saw a snippet. https://www.wsj.com/articles/questions-a..._lead_pos3

The Chinese test supposedly has a high false-negative rate, too.  https://www.nytimes.com/2020/04/01/well/...ative.html  A quick search did not turn up any comparable numbers for the WHO test.

I realize that false negatives don't necessarily make a test useless.  But if we are going to get to the containment phase, it will be important to have a good test.

[burger]

WSJ has an article today saying that tests being used in the US has a 1/3 false negative rate.  Article is paywalled--I just saw a snippet. https://www.wsj.com/articles/questions-a..._lead_pos3

The Chinese test supposedly has a high false-negative rate, too.  https://www.nytimes.com/2020/04/01/well/...ative.html  A quick search did not turn up any comparable numbers for the WHO test.

I realize that false negatives don't necessarily make a test useless.  But if we are going to get to the containment phase, it will be important to have a good test.

[Hurlburt88]

via MSN this is not behind paywall (same article as in WSJ)
https://www.msn.com/en-us/news/us/questi...li=BBnb7Kz

[Hurlburt88]

via MSN this is not behind paywall (same article as in WSJ)
https://www.msn.com/en-us/news/us/questi...li=BBnb7Kz

[OutsiderFan]

False negatives have been an issue ever since testing began in China. It probably doesn't matter much for people with shortness of breath, low blood oxygen and CT scans showing lung lesions, but it does for asymptomatic people for sure if you want to stop spread and not need everyone at home killing the economy to do it.

I remember reading some stuff that it sometimes took 2-3 tests to get a positive result. I guess the good news is false positives have not been an issue.

[OutsiderFan]

False negatives have been an issue ever since testing began in China. It probably doesn't matter much for people with shortness of breath, low blood oxygen and CT scans showing lung lesions, but it does for asymptomatic people for sure if you want to stop spread and not need everyone at home killing the economy to do it.

I remember reading some stuff that it sometimes took 2-3 tests to get a positive result. I guess the good news is false positives have not been an issue.

[burger]

An excellent thread on the antibody test--read the whole thing. 

TL,DR: via the magic of Bayesian statistics, the test is going to produce a brobdingnagian number of false positives as long as the actual number of infected people is small.  This could kill people if they let down their guard based on a positive test.

[tweet]https://twitter.com/zbinney_NFLinj/status/1245789672833417217[/tweet]

[burger]

An excellent thread on the antibody test--read the whole thing. 

TL,DR: via the magic of Bayesian statistics, the test is going to produce a brobdingnagian number of false positives as long as the actual number of infected people is small.  This could kill people if they let down their guard based on a positive test.

[tweet]https://twitter.com/zbinney_NFLinj/status/1245789672833417217[/tweet]

[oldalum]

As far as applying the antibody test widely to people who have not had symptoms and no known exposures (and especially who have also practiced social distancing well), the positive predictive value of the test (true positives divided by all positives) will be even worse among than he says because the majority of people who have antibodies will have diagnosed infection or high-risk exposures, thus the prevalence of infection and antibodies among the people with no symptoms and no exposures will be even lower than he assumes in his example). But as he explains a bit later in the thread, the antibody test will be useful in contexts in which the pre-test probability of disease is high enough. Those situations would probably include re-checking people who have symptoms consistent with Covid-19 but a negative antigen test, or people with high-risk exposures but no symptoms such as front-line health care workers (especially with poor PPE), spouses/partners of people with antigen-confirmed infections.

Of course, this Bayesian analysis applies to all medical tests. Patients who push doctors to order various tests when the prior probability is very low, and some doctors who order such tests, may not realize a positive test would probably be a false positive.

[oldalum]

As far as applying the antibody test widely to people who have not had symptoms and no known exposures (and especially who have also practiced social distancing well), the positive predictive value of the test (true positives divided by all positives) will be even worse among than he says because the majority of people who have antibodies will have diagnosed infection or high-risk exposures, thus the prevalence of infection and antibodies among the people with no symptoms and no exposures will be even lower than he assumes in his example). But as he explains a bit later in the thread, the antibody test will be useful in contexts in which the pre-test probability of disease is high enough. Those situations would probably include re-checking people who have symptoms consistent with Covid-19 but a negative antigen test, or people with high-risk exposures but no symptoms such as front-line health care workers (especially with poor PPE), spouses/partners of people with antigen-confirmed infections.

Of course, this Bayesian analysis applies to all medical tests. Patients who push doctors to order various tests when the prior probability is very low, and some doctors who order such tests, may not realize a positive test would probably be a false positive.

[Goose]

This link has recent info on the state of testing. There are lots of problems that were not anticipated by the people involved in testing, particularly problems with the "extraction kits". Turns out this is a much bigger problem logistically and in terms of "What swabs actually work".

https://www.npr.org/sections/health-shot...-up-with-d

[Goose]

This link has recent info on the state of testing. There are lots of problems that were not anticipated by the people involved in testing, particularly problems with the "extraction kits". Turns out this is a much bigger problem logistically and in terms of "What swabs actually work".

https://www.npr.org/sections/health-shot...-up-with-d

[Goose]

https://cmr.asm.org/content/29/4/773
This article about Ebola testing makes some important points about the problems with the PCR tests for Ebola. There are lots of issues regarding sensitivity that are not totally worked out as of 2015. It is not surprising that there are several unknowns in the current testing worldwide. It is also clear that in at least the Ebola testing, some things once thought to be so turned out not to be.

[Goose]

https://cmr.asm.org/content/29/4/773
This article about Ebola testing makes some important points about the problems with the PCR tests for Ebola. There are lots of issues regarding sensitivity that are not totally worked out as of 2015. It is not surprising that there are several unknowns in the current testing worldwide. It is also clear that in at least the Ebola testing, some things once thought to be so turned out not to be.

[magnus]

WSJ has an article today saying that tests being used in the US has a 1/3 false negative rate.  Article is paywalled--I just saw a snippet. https://www.wsj.com/articles/questions-a..._lead_pos3

The Chinese test supposedly has a high false-negative rate, too.  https://www.nytimes.com/2020/04/01/well/...ative.html  A quick search did not turn up any comparable numbers for the WHO test.

I realize that false negatives don't necessarily make a test useless.  But if we are going to get to the containment phase, it will be important to have a good test.

Sounds like the tests these days aren't as bad as 1/3 false negative.   However, Abbott's 5-13 minute fast test apparently has a (industry leading?) 84% false negative rate when samples are combined with certain virus transport media.

https://www.cnn.com/2020/04/21/health/ab...index.html

[magnus]

WSJ has an article today saying that tests being used in the US has a 1/3 false negative rate.  Article is paywalled--I just saw a snippet. https://www.wsj.com/articles/questions-a..._lead_pos3

The Chinese test supposedly has a high false-negative rate, too.  https://www.nytimes.com/2020/04/01/well/...ative.html  A quick search did not turn up any comparable numbers for the WHO test.

I realize that false negatives don't necessarily make a test useless.  But if we are going to get to the containment phase, it will be important to have a good test.

Sounds like the tests these days aren't as bad as 1/3 false negative.   However, Abbott's 5-13 minute fast test apparently has a (industry leading?) 84% false negative rate when samples are combined with certain virus transport media.

https://www.cnn.com/2020/04/21/health/ab...index.html

[Goose]

However, Abbott's 5-13 minute fast test apparently has a (industry leading?) 84% false negative rate when samples are combined with certain virus transport media.


https://www.cnn.com/2020/04/21/health/ab...index.html

I don't think that is what the article says. I think it asys the false negative rate was (100 - 84.4), or 15.6%. Not great, but not as bad as an 84% false negative would be.

[Goose]

However, Abbott's 5-13 minute fast test apparently has a (industry leading?) 84% false negative rate when samples are combined with certain virus transport media.


https://www.cnn.com/2020/04/21/health/ab...index.html

I don't think that is what the article says. I think it asys the false negative rate was (100 - 84.4), or 15.6%. Not great, but not as bad as an 84% false negative would be.

[magnus]

However, Abbott's 5-13 minute fast test apparently has a (industry leading?) 84% false negative rate when samples are combined with certain virus transport media.


https://www.cnn.com/2020/04/21/health/ab...index.html

I don't think that is what the article says. I think it asys the false negative rate was (100 - 84.4), or 15.6%. Not great, but not as bad as an 84% false negative would be.

Whoops.  I miswrote.  I meant as you corrected.   16% false negative.   Thanks (I'll leave original for posterity)

[magnus]

However, Abbott's 5-13 minute fast test apparently has a (industry leading?) 84% false negative rate when samples are combined with certain virus transport media.


https://www.cnn.com/2020/04/21/health/ab...index.html

I don't think that is what the article says. I think it asys the false negative rate was (100 - 84.4), or 15.6%. Not great, but not as bad as an 84% false negative would be.

Whoops.  I miswrote.  I meant as you corrected.   16% false negative.   Thanks (I'll leave original for posterity)

[magnus]

So I'm a little confused with these false negative numbers.   In a different thread,  MT pointed out how PCR tests aren't close to 100% on negatives even with symptomatic cases.   I also saw somewhere that depending on where you got the sample from a person, the chances of a representative sample varied. 

So is the Abbott 16% false negative on top of the say 70% hit rate for a nasal swab?  In other words,  using abbott's quick test on nasal swabs would yield over 40% false negatives?

Yikes!  
Makes you wonder about all those hospitals relaxing PPE requirements on negative resulted patients.

[magnus]

So I'm a little confused with these false negative numbers.   In a different thread,  MT pointed out how PCR tests aren't close to 100% on negatives even with symptomatic cases.   I also saw somewhere that depending on where you got the sample from a person, the chances of a representative sample varied. 

So is the Abbott 16% false negative on top of the say 70% hit rate for a nasal swab?  In other words,  using abbott's quick test on nasal swabs would yield over 40% false negatives?

Yikes!  
Makes you wonder about all those hospitals relaxing PPE requirements on negative resulted patients.

[BostonCard]

So I'm a little confused with these false negative numbers.   In a different thread,  MT pointed out how PCR tests aren't close to 100% on negatives even with symptomatic cases.   I also saw somewhere that depending on where you got the sample from a person, the chances of a representative sample varied. 



So is the Abbott 16% false negative on top of the say 70% hit rate for a nasal swab?  In other words,  using abbott's quick test on nasal swabs would yield over 40% false negatives?



Yikes!  

Makes you wonder about all those hospitals relaxing PPE requirements on negative resulted patients.

I don't think so.  In an ideal world, the sensitivity and specificity of a test is determined by comparing it to a "gold standard" that is 100% or nearly 100% accurate.  In the real world, since we don't know how good the "gold standard" is, the measures of sensitivity and specificity are determined independently in a artificially constructed cohort.  It is possible to get samples that you know for certain are negative, without having to rely on a gold standard test to rule them out for disease; you pull stored samples from before the pandemic began.  Nobody should have antibodies for SARS-CoV-2 before November 2019, so you can be sure that those are true negatives.  Any positive serology result in a sample from pre-November-2019 can be considered a false positive.

For known positives, you can be fairly sure that anyone who had a positive viral RNA test was a true positive.  A false positive viral PCR test would be exceedingly rare (it would have to involve something like a sample mix-up or contamination, as PCR is extremely specific), so you can be fairly sure that anyone who tested positive for COVID-19 with conventional tests are true positives.

But yes, otherwise, if you are comparing a test to an imperfect gold standard, there are some issues, and that can be adjusted for to some extent.

BC

[BostonCard]

So I'm a little confused with these false negative numbers.   In a different thread,  MT pointed out how PCR tests aren't close to 100% on negatives even with symptomatic cases.   I also saw somewhere that depending on where you got the sample from a person, the chances of a representative sample varied. 



So is the Abbott 16% false negative on top of the say 70% hit rate for a nasal swab?  In other words,  using abbott's quick test on nasal swabs would yield over 40% false negatives?



Yikes!  

Makes you wonder about all those hospitals relaxing PPE requirements on negative resulted patients.

I don't think so.  In an ideal world, the sensitivity and specificity of a test is determined by comparing it to a "gold standard" that is 100% or nearly 100% accurate.  In the real world, since we don't know how good the "gold standard" is, the measures of sensitivity and specificity are determined independently in a artificially constructed cohort.  It is possible to get samples that you know for certain are negative, without having to rely on a gold standard test to rule them out for disease; you pull stored samples from before the pandemic began.  Nobody should have antibodies for SARS-CoV-2 before November 2019, so you can be sure that those are true negatives.  Any positive serology result in a sample from pre-November-2019 can be considered a false positive.

For known positives, you can be fairly sure that anyone who had a positive viral RNA test was a true positive.  A false positive viral PCR test would be exceedingly rare (it would have to involve something like a sample mix-up or contamination, as PCR is extremely specific), so you can be fairly sure that anyone who tested positive for COVID-19 with conventional tests are true positives.

But yes, otherwise, if you are comparing a test to an imperfect gold standard, there are some issues, and that can be adjusted for to some extent.

BC

[magnus]

So I'm a little confused with these false negative numbers.   In a different thread,  MT pointed out how PCR tests aren't close to 100% on negatives even with symptomatic cases.   I also saw somewhere that depending on where you got the sample from a person, the chances of a representative sample varied. 



So is the Abbott 16% false negative on top of the say 70% hit rate for a nasal swab?  In other words,  using abbott's quick test on nasal swabs would yield over 40% false negatives?



Yikes!  

Makes you wonder about all those hospitals relaxing PPE requirements on negative resulted patients.

I don't think so.  In an ideal world, the sensitivity and specificity of a test is determined by comparing it to a "gold standard" that is 100% or nearly 100% accurate.  In the real world, since we don't know how good the "gold standard" is, the measures of sensitivity and specificity are determined independently in a artificially constructed cohort.  It is possible to get samples that you know for certain are negative, without having to rely on a gold standard test to rule them out for disease; you pull stored samples from before the pandemic began.  Nobody should have antibodies for SARS-CoV-2 before November 2019, so you can be sure that those are true negatives.  Any positive serology result in a sample from pre-November-2019 can be considered a false positive.

For known positives, you can be fairly sure that anyone who had a positive viral RNA test was a true positive.  A false positive viral PCR test would be exceedingly rare (it would have to involve something like a sample mix-up or contamination, as PCR is extremely specific), so you can be fairly sure that anyone who tested positive for COVID-19 with conventional tests are true positives.

But yes, otherwise, if you are comparing a test to an imperfect gold standard, there are some issues, and that can be adjusted for to some extent.

BC

Sorry to request a further explanation, but I'm not sure what your answer is to my question about what the true false negative rating is for say a nasal swab being tested by 84% accurate (16% false negative) abbott's fast test. (the comments about false positives, I'm not disputing)

In other words, if you nasal swabbed and abbott tested 100 positive patients, how many results would end up positive (statistically).  At least 16% for sure, but would it be 40% if nasal swabs are only 70% capable of grabbing enough sample to trigger a positive result?

Thanks

[magnus]

So I'm a little confused with these false negative numbers.   In a different thread,  MT pointed out how PCR tests aren't close to 100% on negatives even with symptomatic cases.   I also saw somewhere that depending on where you got the sample from a person, the chances of a representative sample varied. 



So is the Abbott 16% false negative on top of the say 70% hit rate for a nasal swab?  In other words,  using abbott's quick test on nasal swabs would yield over 40% false negatives?



Yikes!  

Makes you wonder about all those hospitals relaxing PPE requirements on negative resulted patients.

I don't think so.  In an ideal world, the sensitivity and specificity of a test is determined by comparing it to a "gold standard" that is 100% or nearly 100% accurate.  In the real world, since we don't know how good the "gold standard" is, the measures of sensitivity and specificity are determined independently in a artificially constructed cohort.  It is possible to get samples that you know for certain are negative, without having to rely on a gold standard test to rule them out for disease; you pull stored samples from before the pandemic began.  Nobody should have antibodies for SARS-CoV-2 before November 2019, so you can be sure that those are true negatives.  Any positive serology result in a sample from pre-November-2019 can be considered a false positive.

For known positives, you can be fairly sure that anyone who had a positive viral RNA test was a true positive.  A false positive viral PCR test would be exceedingly rare (it would have to involve something like a sample mix-up or contamination, as PCR is extremely specific), so you can be fairly sure that anyone who tested positive for COVID-19 with conventional tests are true positives.

But yes, otherwise, if you are comparing a test to an imperfect gold standard, there are some issues, and that can be adjusted for to some extent.

BC

Sorry to request a further explanation, but I'm not sure what your answer is to my question about what the true false negative rating is for say a nasal swab being tested by 84% accurate (16% false negative) abbott's fast test. (the comments about false positives, I'm not disputing)

In other words, if you nasal swabbed and abbott tested 100 positive patients, how many results would end up positive (statistically).  At least 16% for sure, but would it be 40% if nasal swabs are only 70% capable of grabbing enough sample to trigger a positive result?

Thanks

[BostonCard]

Sorry to request a further explanation, but I'm not sure what your answer is to my question about what the true false negative rating is for say a nasal swab being tested by 84% accurate (16% false negative) abbott's fast test. (the comments about false positives, I'm not disputing)



In other words, if you nasal swabbed and abbott tested 100 positive patients, how many results would end up positive (statistically).  At least 16% for sure, but would it be 40% if nasal swabs are only 70% capable of grabbing enough sample to trigger a positive result?



Thanks

The problem is that "false negative rate" is an ambiguous term.  It could represent the sensitivity (technically 1 - the sensitivity).  The sensitivity isis the proportion of patients who are known to have COVID-19 and test positive [in other words, true positives/(true positives plus false negatives); so the "false negative rate" would be false negatives/(true positives plus false negatives)].  However, it could represent the negative predictive value (again, technically 1- NPV); NPV is the proportion of patients who test negative who don't COVID-19 [in other words true negatives/(true negatives plus false negatives), so the "false negative rate" here would be false neagtives/(true negatives plus false negatives).

I don't know how the tests were conducted.  My sense is that they are quoting the sensitivity, and so it depends on what your gold standard being used is.  If you have an imperfect gold standard, the sensitivity of the Abbott test would be chained to the gold standard, if it failed to test positive in false negatives in the gold standard.  It is possible that the Abbott test would catch some proportion of the 30% of false negatives through conventional tests.

BC

[BostonCard]

Sorry to request a further explanation, but I'm not sure what your answer is to my question about what the true false negative rating is for say a nasal swab being tested by 84% accurate (16% false negative) abbott's fast test. (the comments about false positives, I'm not disputing)



In other words, if you nasal swabbed and abbott tested 100 positive patients, how many results would end up positive (statistically).  At least 16% for sure, but would it be 40% if nasal swabs are only 70% capable of grabbing enough sample to trigger a positive result?



Thanks

The problem is that "false negative rate" is an ambiguous term.  It could represent the sensitivity (technically 1 - the sensitivity).  The sensitivity isis the proportion of patients who are known to have COVID-19 and test positive [in other words, true positives/(true positives plus false negatives); so the "false negative rate" would be false negatives/(true positives plus false negatives)].  However, it could represent the negative predictive value (again, technically 1- NPV); NPV is the proportion of patients who test negative who don't COVID-19 [in other words true negatives/(true negatives plus false negatives), so the "false negative rate" here would be false neagtives/(true negatives plus false negatives).

I don't know how the tests were conducted.  My sense is that they are quoting the sensitivity, and so it depends on what your gold standard being used is.  If you have an imperfect gold standard, the sensitivity of the Abbott test would be chained to the gold standard, if it failed to test positive in false negatives in the gold standard.  It is possible that the Abbott test would catch some proportion of the 30% of false negatives through conventional tests.

BC