The last post on why under current definitions COVID will never end picked up some interest. Other people have started writing articles about it too in recent days.
It wasn’t the most focused writing I’ve ever done, to be honest. Here I’ll try to simply summarise the key points about COVID PCR testing false positives, instead of using a narrative format. New information is provided and all claims are sourced.
- Many scientists and governments consider COVID testing to be highly accurate, without any meaningful number of false positives. For example Switzerland’s COVID task force page says “The specificity of PCR tests is very high, leading to essentially no false positive results which could bias our estimates”. When announcing emergency approval of tests, the WHO said, “The move should help increase access to quality-assured, accurate tests for the disease”. This article by a molecular biologist says, “RT-PCR is the gold standard due to its accuracy”. This position has become standard orthodoxy throughout public health systems. That’s why a single positive test is sufficient to trigger weeks of quarantine.
- But in 2006 a PCR test for whooping cough had a 14% false positive rate. In fact every single positive test result was incorrect (error rate = 100%). The doctors involved warned afterwards that mass testing can create “pseudo-epidemics”: outbreaks that appear to be real but are really just bad tests combined with mis-diagnosed symptoms.
- In 2015 PCR testing labs were challenged with samples they were told had the MERS coronavirus but which actually had cold viruses. 8.1% of labs returned false positives.
- In 2003 a pseudo-epidemic of SARS-CoV-1 occurred in British Columbia. Both PCR testing and antibody testing indicated the SARS coronavirus was present, although in reality it was just a common cold virus.
- A new study re-tested COVID positive samples with a more advanced test and concluded there may be a 30% false positive rate. The author concluded, “False-positive laboratory test reports can easily create unnecessary panic resulting in negative impacts on local economies”.
- A recent letter to the BMJ (British Medical Journal) said, “Of the 26 single gene low level positive results, 19 were repeated and all 19 were negative on repeat testing … The harm afforded by false positive results should not be ignored … Large volume screening at a time of low prevalence has the potential to do more harm than good and some of these strategies should be temporarily suspended”.
- PCR tests routinely flip when the same patient is immediately re-tested. For examples see diagram 1D from this paper, or Ohio governor Mike DeWine, or this US sportsman.
- In a random testing survey published on July 7th the UK statistical agency stated that: “Even in a purely hypothetical situation that the virus is not circulating, a test specificity of 99.9% would be associated with an expected number of positive tests that is approximately equal to what we observed over the entire study period”. In other words, PCR testing would need to have specificity significantly higher than 99.9% to avoid reported UK COVID case numbers for June being meaningless. If it is merely 99.9% accurate, COVID in the UK since the end of April would have been entirely false positives.
- British government rules state that 95% specificity for PCR tests is “acceptable” and 99%+ is “desirable”. Thus a “desirable” test would still be so inaccurate that every current reported COVID case in Britain would be false. The survey report claimed a scenario in which the virus is not circulating is “purely hypothetical”, but by their own numbers this situation is by now actually very likely.
- The only systematic testing programme for COVID-19 PCR labs shows that whilst many tests didn’t report any false positives, some had specificity of lower than 98% and a couple were even at 90% or lower. This is much too low (i.e. worst case FP rate of 10%). In addition the uncertainty intervals for many tests are very high, due to insufficient sample size. Because false positives are generated by some tests/labs and not others this could yield spontaneous localised “outbreaks” as samples collected in one place get sent to a nearby lab, even with randomised testing.
- According to Professor Carl Henegehan of Oxford University, in the UK “the ONS [Office of National Statistics] currently cannot estimate prevalence because it does not now know what the false positive rate of PCR testing is …. We’re now in a spiral of bad data”.
- In the UK 5000 people “un-died” from COVID-19 after it was discovered the health agency was defining a COVID death as anyone who had ever tested PCR positive and later died, of any reason, at any time i.e. they defined COVID as a terminal disease that is 100% fatal. After the definition was changed UK death figures for the first days of August dropped 90%. However, even the new definition still does not track people who actually died of COVID. Instead it’s assumed that anyone who dies within a month of testing positive is a COVID death. This is not a UK specific problem: it’s a common international definition.
- The New York Times devoted a front page cover to 1000 names of people whose obituaries or death notices said they died of COVID. In fact the 6th person on the list had been murdered.
- The government of Norway reported that, “Given today’s contagion situation in Norway, health professionals must test around 12,000 random people to find one positive case of COVID-19. There will be about 15 positive test responses, but 14 of these will be false positives”. However this is not a correct measurement of FP rates because they are calibrating the test against itself. As I discussed previously this is not logically or scientifically valid.
- In the USA a widely reported “recommended minimum number of tests per day” comes from the Harvard Global Health Institute. As documented by Daniel Horowitz, the moment their initial goal was met it was increased by 700%, from 500k per day to 3.5 million per day. The justification was that there were more COVID cases so there had to be more testing. But in any test with a >0% false positive rate more testing will automatically create more COVID cases, because a case is defined as someone testing positive, not having symptoms. So as Horowitz observes Harvard University has created a feedback loop, perhaps obliviously.
- After this occurred CNN did an interview with a US government official. Challenged on testing numbers the official said, “I’ve talked to modellers all over the place, and they throw up these numbers with very little data support for it and they change tenfold over a period of time”. The CNN presenter then interrupted him and claimed “it’s a straw man to claim Harvard has been changing the numbers … they’ve been pretty consistent for the last three or four months”. In fact CNN was wrong and the official was correct (albeit it was a seven-fold increase rather than ten-fold).
- What about false negative rates, i.e. the test doesn’t tell you you’re infected when it should? “A systematic review of the accuracy of covid-19 tests reported false negative rates of between 2% and 29% (equating to sensitivity of 71–98%), based on negative RT-PCR tests which were positive on repeat testing”. Defining a false negative this way is not logically valid (a test that tells you both yes and no in quick succession cannot be used to conclude anything), but this again shows how volatile PCR results are: they routinely give different answers for the same person when repeated.
- According to the testimony of a Swedish doctor, “The next day all those patients were gone and the only thing coming in to the hospital was Covid. Practically everyone who was tested had Covid, regardless of what the presenting symptom was. People came in with a nose bleed and they had covid. They came in with stomach pain and they had Covid”. In other words, test results don’t select for respiratory problems, which is what actually matters.
- Spanish researchers claimed in June to have RT-PCR positive test results from sewage in Barcelona sampled in March 2019. Either this is a false positive or the virus has been circulating for a year before the epidemic supposedly started.
Conclusion. Given current levels of testing, overall low numbers of true cases and widely varying accuracy rates at COVID testing labs, localised outbreaks of COVID as of August 2020 may not in fact exist at all.