7 ms·

These are exceedingly rare, like almost never happen.

mike_hearn 3y ago
That's unfortunately not the case. PCR false positives are a massive problem. Scientists liked to claim during COVID that PCR tests don't have false positive because they use an obscure definition that doesn't match how people normally use the term. I wrote about this problem back in 2020 when it started becoming apparent that the public health establishment wasn't attempting to stop or even measure COVID PCR test false positives, because they had simply defined the problem out of existence [1]. Yet dig into the literature a bit and you discover that FPs are a well known problem that have occurred many times before. There are at least three problems and arguably more: 1. Scope limits. Scientists have an extremely myopic view of what counts as an FP somewhat akin to "that's not a bug it's a feature" in the software world. Sample contaminated somewhere along the line from collection to machine? Maybe even when the sample brushed up against the outside of the machine itself, or the skin of a lab worker? Not a false positive in their view, because the PCR test itself is finding the RNA fragment it's looking for. PCR detects virus your immune system already destroyed? Not a false positive, because the test is actually looking for RNA so begins with a lysis stage that breaks open the viral capsid anyway. Results got mixed up in a shoddy IT system? Not a false positive because the "test" is just the machine they operate, not the whole infrastructure the public interacts with. 2. Over-sensitivity. COVID PCR tests were run at thresholds normally considered way too sensitive for normal usage, which is why they were so frequently triggering for weeks after a patient seemed to have fully recovered. Even with CTs ~25 these tests were pre-2020 run in extremely controlled labs with specialized air handling equipment and the like, to try and avoid FPs, but COVID tests were being routinely run in ad-hoc labs with CTs of over 40 and each additional cycle is twice as sensitive as the last! 3. Circular logic. Normally medicine is careful to keep pathogen and disease separate. It's important because some diseases ("collections of symptoms") can be caused by multiple pathogens, or no pathogen, and many people become infected with a pathogen yet never develop disease. In the very early days of COVID the distinction between disease and causative agent was kept properly separated because diagnosis was done by doctors, but this didn't survive contact with the public health establishment. They wanted a mass-scale system in the incorrect belief that they could slow down spread this way (there's no link in the data between testing levels and outcomes). Once mass PCR testing started the definition of COVID became circular: A positive test means you have COVID, but COVID was defined as having a positive test. It's obvious what happens if you define COVID this way: the tests can by definition never have false positives. On what basis can you dispute the accuracy of the test if the test and not disease is defined to be ground truth? Using this approach, false positives always become "asymptomatic cases" that inconveniently never become sick and thus are not a case under classical medical reasoning. This is what led to bizarre claims coming from health authorities, like the claim that the tests had FP rates of zero. But it's not reality. As I document in the essay, we know PCR results have FPs because there have been "pseudo-epidemics" in the past where ordinary coughs and colds were mis-diagnosed as outbreaks of dangerous diseases due to bad PCR results, and because lab challenges have often returned failures i.e. labs were submitted samples known to contain just ordinary rhinovirus or whatever and came back with positive SARS-CoV-2 results. There were also cases where reagent mixups caused false results, or where PCR tests returned results that switched between negative and positive then back to negative from samples taken back to back and so on. None of these were ever really investigated or root caused. And all that's before you get into the systems that surround the labs. I got a PCR certificate for someone else's name at one point, almost certainly a form autofill mistake (I saw workers wrestling with this problem when they lost my wife's results). When I complained I got a new certificate in under 5 minutes for my own name this time but now the results were signed off by a different doctor i.e. the signatures supposedly demonstrating expert supervision of the test were just being randomly picked by software. [1] https://blog.plan99.net/pseudo-epidemics-7603b2da839 https://blog.plan99.net/pseudo-epidemics-7603b2da839
- freshpots 3y ago
  But aren't we talking about rapid tests? There is a big difference between false positives when it comes to PCR vs rapid tests. False positives with rapid tests are known to be far less prevalent.
  - mike_hearn 3y ago
    Other types of test are calibrated against PCR tests. That's why the article says: "[...] finds the largest study to compare home rapid tests with gold-standard PCR tests" PCR is assumed to be the truth to which rapid tests should aspire. > False positives with rapid tests are known to be far less prevalent. That's probably true if you use a normal definition of false positive, but under the intellectual framework public health uses you can't say this because you can't have less prevalent than zero.
- zanfr 3y ago
  no.
- tripletao 3y ago
  The subtitle of your Jul 2020 essay criticizing PCR tests is "Why COVID-19 is guaranteed to never end". But in fact, judging by confirmed (by PCR, or by other tests grounded in PCR as you say) COVID-19 deaths, the pandemic has ended--the daily count has decreased by a factor of ~50 from the peak. Does that not suggest to you that maybe you were wrong? It's not just the decrease in testing, since test positivity has decreased from the peaks even as total test count plummeted. There is no evidence that PCR tests greatly overcounted SARS-CoV-2 infections: 1/2. The specificity of commonly-used PCR tests for SARS-CoV-2 is clearly very high, >99.9%, since there have been regions (e.g. Australia) where total positivity was <0.1% for months at a time. It's true that other PCR tests (like your whooping cough tests) may have much worse specificity, especially in the early days before primer design was as well-understood; but we know empirically that these ones doesn't. Sloppy lab work could result in false positives by cross-contamination, but contamination has to come from a true positive sample; so this is a problem only when the true positivity is high. Sloppy clerical work could result in any possible incorrect result, but you've presented no evidence that this is a significant problem in aggregate. Lots of papers studied the relationship between PCR positivity and culturable virus, for example https://academic.oup.com/cid/article/72/11/e921/5912603?login=false https://academic.oup.com/cid/article/72/11/e921/5912603?logi... It's true that at high Ct, the cultures are often negative; but (a) the relationship between Ct and culturable virus may vary dramatically with small changes in PCR protocol, (b) a patient early in their infection may have high Ct now but lower Ct later, and (c) it's not clear whether all naturally infectious patients would be positive by artificial culture anyways. Based on these factors, most public health authorities chose not to use an artificially lower Ct. So a positive PCR test doesn't necessarily mean that a patient is infectious now, but it does mean that they are or were recently (weeks to months) infected. For emphasis, it is absolutely expected that some samples will be positive by PCR but negative by cell culture--the PCR is testing for any RNA matching the primers, but the culture is testing for replication-competent whole virus. Presence of the latter implies the former, but the converse isn't true. The public messaging on that wasn't very good, and perhaps they initially misinformed you; but no one with any background in molecular biology was confused here. 3. Your idea that the PCR test is simply assumed to be accurate is completely false. The point of papers like the one I linked above is to compare two test methods, find the places where they disagree, and try to judge--from the experimental results and from our understanding of the biochemistry of the test methods--the reasons for those disagreements. They're not assuming either method to be perfect. Based on many such comparisons--against cell culture, against electron microscopy, against antibody tests, against excess mortality once that got big enough to distinguish statistically (which it mostly wasn't by Jul 2020), etc.--public health authorities judged PCR to be the best available gold standard, and good enough for many statistical purposes. The public messaging was again pretty bad there, and if you looked only to that then I understand why you'd feel misled. The actual scientific literature is there for anyone who wishes to read it, though.

That's unfortunately not the case. PCR false positives are a massive problem. Scientists liked to claim during COVID that PCR tests don't have false positive because they use an obscure definition that doesn't match how people normally use the term. I wrote about this problem back in 2020 when it started becoming apparent that the public health establishment wasn't attempting to stop or even measure COVID PCR test false positives, because they had simply defined the problem out of existence [1]. Yet dig into the literature a bit and you discover that FPs are a well known problem that have occurred many times before. There are at least three problems and arguably more: 1. Scope limits. Scientists have an extremely myopic view of what counts as an FP somewhat akin to "that's not a bug it's a feature" in the software world. Sample contaminated somewhere along the line from collection to machine? Maybe even when the sample brushed up against the outside of the machine itself, or the skin of a lab worker? Not a false positive in their view, because the PCR test itself is finding the RNA fragment it's looking for. PCR detects virus your immune system already destroyed? Not a false positive, because the test is actually looking for RNA so begins with a lysis stage that breaks open the viral capsid anyway. Results got mixed up in a shoddy IT system? Not a false positive because the "test" is just the machine they operate, not the whole infrastructure the public interacts with. 2. Over-sensitivity. COVID PCR tests were run at thresholds normally considered way too sensitive for normal usage, which is why they were so frequently triggering for weeks after a patient seemed to have fully recovered. Even with CTs ~25 these tests were pre-2020 run in extremely controlled labs with specialized air handling equipment and the like, to try and avoid FPs, but COVID tests were being routinely run in ad-hoc labs with CTs of over 40 and each additional cycle is twice as sensitive as the last! 3. Circular logic. Normally medicine is careful to keep pathogen and disease separate. It's important because some diseases ("collections of symptoms") can be caused by multiple pathogens, or no pathogen, and many people become infected with a pathogen yet never develop disease. In the very early days of COVID the distinction between disease and causative agent was kept properly separated because diagnosis was done by doctors, but this didn't survive contact with the public health establishment. They wanted a mass-scale system in the incorrect belief that they could slow down spread this way (there's no link in the data between testing levels and outcomes). Once mass PCR testing started the definition of COVID became circular: A positive test means you have COVID, but COVID was defined as having a positive test. It's obvious what happens if you define COVID this way: the tests can by definition never have false positives. On what basis can you dispute the accuracy of the test if the test and not disease is defined to be ground truth? Using this approach, false positives always become "asymptomatic cases" that inconveniently never become sick and thus are not a case under classical medical reasoning. This is what led to bizarre claims coming from health authorities, like the claim that the tests had FP rates of zero. But it's not reality. As I document in the essay, we know PCR results have FPs because there have been "pseudo-epidemics" in the past where ordinary coughs and colds were mis-diagnosed as outbreaks of dangerous diseases due to bad PCR results, and because lab challenges have often returned failures i.e. labs were submitted samples known to contain just ordinary rhinovirus or whatever and came back with positive SARS-CoV-2 results. There were also cases where reagent mixups caused false results, or where PCR tests returned results that switched between negative and positive then back to negative from samples taken back to back and so on. None of these were ever really investigated or root caused. And all that's before you get into the systems that surround the labs. I got a PCR certificate for someone else's name at one point, almost certainly a form autofill mistake (I saw workers wrestling with this problem when they lost my wife's results). When I complained I got a new certificate in under 5 minutes for my own name this time but now the results were signed off by a different doctor i.e. the signatures supposedly demonstrating expert supervision of the test were just being randomly picked by software. [1] https://blog.plan99.net/pseudo-epidemics-7603b2da839 https://blog.plan99.net/pseudo-epidemics-7603b2da839

freshpots 3y ago
But aren't we talking about rapid tests? There is a big difference between false positives when it comes to PCR vs rapid tests. False positives with rapid tests are known to be far less prevalent.
- mike_hearn 3y ago
  Other types of test are calibrated against PCR tests. That's why the article says: "[...] finds the largest study to compare home rapid tests with gold-standard PCR tests" PCR is assumed to be the truth to which rapid tests should aspire. > False positives with rapid tests are known to be far less prevalent. That's probably true if you use a normal definition of false positive, but under the intellectual framework public health uses you can't say this because you can't have less prevalent than zero.
zanfr 3y ago
no.
tripletao 3y ago
The subtitle of your Jul 2020 essay criticizing PCR tests is "Why COVID-19 is guaranteed to never end". But in fact, judging by confirmed (by PCR, or by other tests grounded in PCR as you say) COVID-19 deaths, the pandemic has ended--the daily count has decreased by a factor of ~50 from the peak. Does that not suggest to you that maybe you were wrong? It's not just the decrease in testing, since test positivity has decreased from the peaks even as total test count plummeted. There is no evidence that PCR tests greatly overcounted SARS-CoV-2 infections: 1/2. The specificity of commonly-used PCR tests for SARS-CoV-2 is clearly very high, >99.9%, since there have been regions (e.g. Australia) where total positivity was <0.1% for months at a time. It's true that other PCR tests (like your whooping cough tests) may have much worse specificity, especially in the early days before primer design was as well-understood; but we know empirically that these ones doesn't. Sloppy lab work could result in false positives by cross-contamination, but contamination has to come from a true positive sample; so this is a problem only when the true positivity is high. Sloppy clerical work could result in any possible incorrect result, but you've presented no evidence that this is a significant problem in aggregate. Lots of papers studied the relationship between PCR positivity and culturable virus, for example https://academic.oup.com/cid/article/72/11/e921/5912603?login=false https://academic.oup.com/cid/article/72/11/e921/5912603?logi... It's true that at high Ct, the cultures are often negative; but (a) the relationship between Ct and culturable virus may vary dramatically with small changes in PCR protocol, (b) a patient early in their infection may have high Ct now but lower Ct later, and (c) it's not clear whether all naturally infectious patients would be positive by artificial culture anyways. Based on these factors, most public health authorities chose not to use an artificially lower Ct. So a positive PCR test doesn't necessarily mean that a patient is infectious now, but it does mean that they are or were recently (weeks to months) infected. For emphasis, it is absolutely expected that some samples will be positive by PCR but negative by cell culture--the PCR is testing for any RNA matching the primers, but the culture is testing for replication-competent whole virus. Presence of the latter implies the former, but the converse isn't true. The public messaging on that wasn't very good, and perhaps they initially misinformed you; but no one with any background in molecular biology was confused here. 3. Your idea that the PCR test is simply assumed to be accurate is completely false. The point of papers like the one I linked above is to compare two test methods, find the places where they disagree, and try to judge--from the experimental results and from our understanding of the biochemistry of the test methods--the reasons for those disagreements. They're not assuming either method to be perfect. Based on many such comparisons--against cell culture, against electron microscopy, against antibody tests, against excess mortality once that got big enough to distinguish statistically (which it mostly wasn't by Jul 2020), etc.--public health authorities judged PCR to be the best available gold standard, and good enough for many statistical purposes. The public messaging was again pretty bad there, and if you looked only to that then I understand why you'd feel misled. The actual scientific literature is there for anyone who wishes to read it, though.