"WHO Issues Warning of False Positive Covid Tests (97%)" Topic

WHO Information Notice for IVD Users

Nucleic acid testing (NAT) technologies that use real-time polymerase chain reaction (RT-PCR) for detection of SARS-CoV-2

14 December 2020 Medical product alert Geneva Reading time: 2 min (554 words)

Description of the problem: WHO has received user feedback on an elevated risk for false SARS-CoV-2 results when testing specimens using RT-PCR reagents on open systems.

Users of RT-PCR reagents should read the IFU carefully to determine if manual adjustment of the PCR positivity threshold is necessary to account for any background noise which may lead to a specimen with a high cycle threshold (Ct) value result being interpreted as a positive result. The design principle of RT-PCR means that for patients with high levels of circulating virus (viral load), relatively few cycles will be needed to detect virus and so the Ct value will be low. Conversely, when specimens return a high Ct value, it means that many cycles were required to detect virus.

link

In case of virus detection, >35 cycles only detects signals which do not correlate with infectious virus as determined by isolation in cell culture [reviewed in 2]; if someone is tested by PCR as positive when a threshold of 35 cycles or higher is used (as is the case in most laboratories in Europe & the US), the probability that said person is actually infected is less than 3%, the probability that said result is a false positive is 97%.

cormandrostenreview.com/report

the probability that said result is a false positive is 97%.

This does not match what the title says.

It doesn't say 97% of COVID tests are false positives. It says that of a subset of the tests (RT-PCR) have another subset (ones done not strictly following the protocol) where the false positive rate is 97%.

To put that into wargaming perspective:

97% of the lips on female figures I have (~15% of total) where I do a rush job to finish the details (10% time for the the 15%) look crappy.

I don't have a report with peer reviewed numbers on distribution of different test methods. RT-PCR requires more steps done by more people with higher skill levels than shallow nasal or throat swab "rapid" tests. You do the math on what those ratios look like in a world (not nation) where people love jeans made in labour conditions they would not accept, but cost less.

The danger is not that 97% of the tests are giving false positives or that 97% of positives are wrong (two different things). The problem is that people who can't think beyond 120 characters at a time are pushing for stupid conditions for testing, and people willing to do a shoddy job are perfectly willing to comply, just like they have for the last 10,000 years of human history.

So Borger, et al. are wrong in their findings?

No, they are not.

You are not representing their findings properly.

It's 97% of only RT-PCR tests and then only the ones that are done not according to the IFU, not 97% of COVID tests.

Basically, they found that there's a strong correlation between screwing up one specific test and getting a false positive result.

You can't apply that finding to the general population of tests without data on the distribution of tests being done (which could be collected) and the number of people not following the directions, for which you don't have data.

Also, their findings are only valid for a subset of specific ways of screwing the test up.

So when they state:

In case of virus detection, >35 cycles only detects signals which do not correlate with infectious virus as determined by isolation in cell culture [reviewed in 2]; if someone is tested by PCR as positive when a threshold of 35 cycles or higher is used (as is the case in most laboratories in Europe & the US), the probability that said person is actually infected is less than 3%, the probability that said result is a false positive is 97%.

They mean that a false positive rate of 97% is achieved using North American and European testing standards.

This bit

(as is the case in most laboratories in Europe & the US)

lacks a data reference. Also, if you read the IFU for the test, there is no standard threshold for cycles. It is situational based on other factors.

You would have to have all the factor data for each of the individual tests to determine if that was an appropriate procedure or not.

n case of virus detection, >35 cycles only detects signals which do not correlate with infectious virus as determined by isolation in cell culture

This is just a mathematically incorrect statement. Every data set correlates with every other dataset. Correlation is not a yes/no, it is a degree.

[reviewed in 2]

Reference 2 is a private email between researchers, so it cannot be reviewed. What 3% they are talking about cannot be determined.

I spoke with a lab tech personally and she confirmed labs here do use a minimum of 35 cycles (and was told that would only yield 100% accurate results).

From Pieter Borger on Dec. 16:

WHO Backtracking…

Although there is no action over at Eurosurveillance, who could have easily retracted the Corman-Drosten paper by now, the WHO has understood our message:

link

I spoke with a lab tech personally and she confirmed labs here do use a minimum of 35 cycles (and was told that would only yield 100% accurate results).

Ah, so papers are good when they say what you want. When you lack that, "somebody said". However, I would like to meet the lab tech intimately familiar with the procedures at a couple thousand labs and can speak for all of them.

Was she told that or did she read that in the procedure? Ask her to point out where in the IFU it says that.

And you're side-stepping the two up front facts:

1) RT-PCR is not the only test, so the error rate in those tests does not apply across all tests, as you said in the title. PCR is one of three broad categories of test, RT-PCR is one type of PCR.

2) The Borger paper refers to a subset of RT-PCR tests, however it does not tell you what subset it is talking about. There are 307 IFUs for the EUA for COVID testing:

link

I assume your reference is an expert in all of them in addition to all the test sites.

There are six to ten major variables in how the test is set up. So if we go with just six and set ten levels of sensitivity (the dial only goes one to ten), that's one million configurations for each of the 300+ different test units.

I have never bothered to dispute the 97% number because its 97% of a teeny-tiny percentage of the tests that behave like that, and, again, only if you don't follow the IFU.

WHO Backtracking…

First, this is continually evolving, so this is one of thousands of change notifications.

Did you read it? It actually tells you not that the test is bad, but to read the IFU carefully. That is, if you do the test wrong, you will not get a reliable results. Groundbreaking.

So does this mean that 97% of the 300,000+ dead in this country aren't really dead?

It purports that 97% of tests done at this level were false positives.

As for deaths, Stats Canada, for instance, has shown the mortality rates in Canada have gone down this year compared to previous years. ( TMP link )

Not sure about the USA, but you would want to check mortality rates in the USA to see if they are any higher this year, the same or lower.

Well I for one am very glad to know that at least one of the five relatives and friends I know who died from covid are not really dead according to you.

Yes,sarcasm.

I'm not sure who you are referring to but the paper says nothing about mortality or mortality rates.

It is about accuracy of testing.