This title is really clickbait for “All you need to know about COVID false positives,” which is possibly the biggest un-publicised problem of this whole pandemic. I’m writing about the UK, but much the same applies across the world.
1 Mass COVID testing – the test used
What is currently driving UK public policy is the recent large increase in “positive COVID tests.” They occur in an increasingly stressed “test, track and trace” service whose escalating capacity has been outpaced by escalating demand.
The tests involved are almost entirely “PCR” tests, which repeatly amplify RNA fragments from COVID-19 until they are detectable. Potentially, enough amplification cycles can detect even a single virus fragment, which makes the test highly sensitive, but also highly prone to “noise” such as contamination. You will remember all the gowns, gloves and careful technique that scientists use to collect DNA samples, because even one contaminating molecule could convict the wrong person of murder, identify a Pharaoh as a modern European, or tell you that your dinosaur soft tissue came from a dog. The PCR test technique is similar, only for RNA.
On the clothes of the person doing your swab, a few stray molecules could turn an entire batch of negative tests positive, and it appears to have done so sometimes, because entire sports teams who tested positive became negative when the test was repeated.
2 The problem of oversensitivity
Because PCR tests RNA fragments, it can pick up “dead” bits of virus either from previous infections, or acquired from other people with previous infections. Truly infected people shed living viruses only for around 7 days, but researchers have found PCR tests positive up to 78 days after infection. This problem can be lessened by regulating how many amplification cycles the sample undergoes, thus weeding out more “stray fragments.” But the government has not set such a limit, thus making anything up to 7 out of 8 positive tests potentially false.
3 False positives
All clinical tests have false positives and negatives. PCR has both, but for reasons explained in Source [1] only the false positives matter here. As mentioned above, contamination can cause false positives, as can errors in technique in inexperienced (or even experienced) samplers or operators, variations in methods from lab to lab, or in the specific case of COVID-19, the presence of dead viruses in healthy people. So what percentage of tests are false positives?
The answer is that nobody knows, least of all the government scientists, because mass testing for COVID-19 was rushed in so quickly that the necessary research to validate and calibrate the test has not been done. In the government’s own literature, Source [2] on false positives, a prior estimate based on PCR testing for other viruses was 2.5% false postives. Source [1] reviewed the literature from 2004-2019 for other RNA viruses, and found false positive rates from 0 – 16.7%. That is quite a range, but does it matter?
4 What happens if the error rate is bigger than what you’re testing?
If we, for now, accept the UK government’s initial estimate of 2.5% false positives, then in the early stages of the pandemic, when tests were being done on very sick people in hospital, they would give reasonably reliable results. Suppose 70% of 100 patients you test actually have the disease, you’re going to misdiagnose another 2 or 3, who might very well require the same ICU and respirator care anyway.
But when Source [2] was written , in June, only 1.6% of tests were coming back as positive. That is smaller than the originally estimated 2.5% error rate, which means that estimate must have been too high, or else at least 2.5% of tests would have been positive. But if the false positive rate was actually 1.6%, almost 1% better than predicted, in theory all the tests in Britain might be false positives.
Source [2] surmised that the false positive rate must be much lower, and guessed at 0.4%, but that was pure speculation, based on simple faith that most results must be true. But there’s no evidence for that reassurance. In reality it could be much higher, and in any case a test whose error rate may be as high as what it is measuring cannot possibly be deemed reliable. This is obvious – nobody in their right mind would go up in a plane with 10% of fuel in the tank if the the gauge was only accurate to +/-10%.
The problem is that non-existent disease outbreaks based on PCR false positives have happened before. In 2009 PCR testing in the UK for H1N1 Avian Flu created panic about ongoing infections around a year after the virus had more or less disappeared. Such a “casedemic” is equally possible now, since we are using the same sort of test in the absence of significantly increased hospitalisations and deaths from COVID.
Is there any evidence that this is happening? I believe there is. I mentioned that Source [1] found an overall positive rate of 1.6% of tests. The government’s daily data includes total tests, as well as the total of positives. I have been following these for a while, and the positive rate has varied between about 1.2% and 1.7%, with no clear trend. Today’s (18th September) results show 3,395 positives in 236,219 tests. That is 1.43% – lower than the figure in June. The explanation for this can only be either that COVID prevalence is unchanging, but that more people are being tested; or that there is little COVID around, and what we are measuring is a fairly consistent false positive rate.
What it doesn’t appear to validate is that the rate of COVID disease is increasing as we are told, despite the government’s shroud-waving, the Rule of 6, increasing lockdowns and the threat of a cancelled Christmas.
5 What about increasing hospitalisations? The problem of the pandefinition
Students of the government’s daily figures (essential reading – don’t rely on the BBC in this or anything else) will see that, as the government feared, hospital admissions are increasing again. In Spain (though not really in other European countries) the death rate, too, appears to be creeping up.
What is not intuitively obvious is that false positive tests can largely account for these phenomena too. And that is largely because of the actual definitions of vague, and misleading, terms like “cases,” “COVID hospitalisations” and “COVID deaths.” they do not mean what you’d assume they mean.
In the proper medicine I was taught, a “case” is a patient who displays clear symptoms, signs, and test results of a particular condition. From the start, partly because it was a new disease, the diagnostic criteria for COVID-19 were not clear. But as we all now know, or should know, a dry persistent cough, fever over 38 deg C and loss of smell are the commonest symptoms – though they’re vague enough to be confused with other infections, such as flu. A positive PCR test would confirm such symptoms fairly securely, despite a 1.6% false positive rate, though it would still give a misleading false positive on a few clinically doubtful cases.
But the government (like most other governments, presumably on someone’s initiative) have decided that a positive PCR trumps all other clinical criteria. In effect, the definition of a “case” is “a single positive PCR test,” even in the absence of any symptoms, or with different symptoms or, astonishingly, if there is also a negative PCR test. Unbelievably, if you have no symptoms, and a repeat negative test shows your previous test was a false positive, you are still officially counted as a “case” and added to the stats.
This absolute faith in PCR tests extends to admissions and deaths, too. The definition of a “COVID admission” is any admission (for any cause) with a positive PCR in the previous 14 days, or during admission. There is no criterion of symptoms of COVID. All hospital admissions are tested routinely. So if you go in for a hernia operation (if the NHS were still doing them!) and test positive, you are a COVID admission. If, then, there were no COVID in the country at all, and all the test results were due to a 1.6% false positive rate, then automatically 1.6 % of admissions would be COVID admissions. And the higher the number of patients tested for 14 days before admission, the higher the number of COVID admissions will inevitably be. Test more, and COVID admissions necessarily rise proportionately.
The same phenomenon is true for “COVID deaths,” because the UK government defines these as any death occurring within 28 days of a positive COVID test. If, once more, we speculate that there is no COVID at all (which of course is not the case), then false positives would still dictate that 1.6% of deaths were counted as COVID deaths. Oddly enough, the actual percentage of COVID deaths compared to total deaths is in that range.
But what about the rise we are beginning to see in the number of patients on ICUs and respirators with COVID? Surely they indicate a second wave that justifies the restrictions? Now, this rise appears to have begun at the beginning of this month, though in a miniscule way compared to the epidemic in March. Two things may be said here, and both depend on routine seasonal changes, and not on failed lockdowns, partying teenagers or masks wearing out.
Look at the relevant graph on the ONS page on winter mortality and you’ll see that deaths always begin to increase around the beginning of September, as they are now. So we expect to see more people dying of respiratory infections as we head towards winter. Given the way most Coronaviruses behave, one would expect some of these to be due to COVID – it is still around, and likes the colder, damper, weather. But given what I have shown before, and that what is being recorded are simply all patients on ventilators who have tested positive, some of those patients will be false positives, getting into ICU because of a whole variety of common winter causes. Remember that ventilators were originally stockpiled for flu pandemics, not for COVID. And they’re not testing for flu virus.
6 What can be done about false positives?
This is actually the strangest and most disturbing part of the problem. Because of some decidedly peculiar decisions about PCR testing for COVID-19, and the breakdown of a hugely ambitious testing program, it is unlikely that anything can be done without a courageous ditching of both.
Source [1] shows that, because of the known problems with PCR testing over many years, the WHO and the US CDC recommended for all the previous epidemics this century – SARS1, MERS, Ebola, and Zika – that:
(1) PCR should only be used in patients with a high suspicion of having the disease or being exposed to it.
(2) The test must be confirmed by a different form of test, or at least by a repeat test from a different laboratory.
Furthermore Single test results were considered insufficient for public health decision making.
Now, oddly enough the UK government very quietly issued some advice a little while ago suggesting a confirmatory test is desirable. This advice, however, did not specifically overturn their previous instruction that in patients with both a positive and a negative test, only the positive should be accepted!
But since that new, ignored, advice the testing service has proved incapable even of handling current demand, let alone the government’s silly “Moonshot” ambition to test 10 million people a day by next year. There is no realistic way they are going to obviate all tests not confirmed by a repeat, especially if there is a chance that expensive, and time-consuming. exercise will show that most, or all, of their draconian measures have been unnecessary and destructive of every aspect of our lives.
But in any case Source [1] points out that for this particular pandemic, all that good advice has simply been abandoned across the world. A single PCR test has been deemed sufficient to make a diagnosis, despite all I have discussed above, by “the World Health Organisation, the US Centers for Disease Control and Prevention, the European Centre for Disease Prevention and Control, Public Healt England, the Public Health Agency of Canada, the Pan American Health Organization, and South Korea’s Centers for Disease Conrol and Prevention.”
If we could trace all these decisions back to a top-down directive to ignorant governments from the WHO, it might tend to confirm the ideas of conspiracy theorists suspicious of the UN and the Chinese Communist Party’s undue influence on it. But Dr Fauci, for example, has been in his role since the 1980s, and one assumes that many of the named organisations have staff who have weathered many such scares, and already knew all about the shortcomings of PCR testing. Was there nobody in any of them to question why a confirmatory test with known, and well-advertised, problems had suddenly become a reliable screening test on which to base public policy?
I know Groupthink is possible, guys, but this failure of policy across most administrations (Sweden being a notable exception, as we know), taken together with the new, bizarre and predictably catastrophic policy of shutting down the whole world for a virus less deadly than some flu strains, taken together with the suppression of dissident views on management and policy, taken together with the mysterious linkage of COVID-19 to Wuhan research labs funded by western governments and wealthy “philanthropists”… this failure of policy smacks more of the kind of universal conspiracy theories involving lizards, Rosicrucians and the whole works.
The only alternative I can think of, as someone who spent a whole career in medicine, is that “Those whom the gods wish to destroy they first make mad,” and that the end of the world is nigh for unconnected divine reasons.
Source [2]: S0519_Impact_of_false_positives_and_negatives.pdf
And here’s a prediction. Back in the Spring, the lockdown appeared to work, although in fact fatal infections peaked a good week before it started, and the Gompertz curve matches every other northern country with or without lockdown.
If the government locks down now, it will (a) be because of widespread, badly conducted testing and (b) be at the very start of the winter excess death period. So I predict that this time lockdown will not appear to reduce infections, which will match test results plus a small percentage, and hospitalisations/deaths will not peak until midwinter. The lockdown will drag on with no visible result.
I also predict that nobody in authority will change the narrative one jot, but will blame the public for giving up on compliance, which certainly a lot more people will this time around.
Jon,
It is indeed depressing, as I lamented in my last contribution, that no-one who should be listening appears to be listening. And I agree with your prediction.
Last night on Question Time, Sunetra Gupta (an infectious disease epidemiologist and a professor of theoretical epidemiology at Oxford University) disagreed with the political and scientific ‘consensus’, and recommended that testing was stopped in low-risk groups, e.g. schoolchildren. She was robustly supported by John Caudwell (a British billionaire businessman).
Shadow Health Secretary Jon Ashworth’s eyes glazed over – he clearly didn’t want to engage with or even hear their contributions.
I would have thought that Labour would have seized on the opportunity to run with a different narrative to the Government’s, and do some real opposing, instead of their lame token support.
Under discussion at work today: A family of 5 may not see their 2 grandparents in the safety of their own home. But they may meet them at the pub.
It’s the Catholic families I’m concerned about – with five kids one of them has to stay at the pub permanently!
Another factor for you to consider. Matt Hancock has insisted that only people with suspected COVID symptoms should be tested. This is not happening, given requirements for testing for work, contact tracing of asymptomatic individuals, etc.
But if we suppose 50% of tests are of people with suspected COVID, and note that, even today, the total positive rate is unchanged at 1.67%, that means that even if the test were 100% reliable, then 49 times out of 50, apparent COVID symptoms are NOT due to COVID, but to other bugs.
The figures for “cases” clearly show, therefore, that COVID cannot be diagnosed clinically in the community, at least until a sick patient has the characteristic complications. But the latter are not criteria for stats for admissions, for ICU care or even for death – only one positive test result.
There was a positive test in a primary school in our town last week, and the whole school year is consequently in quarantine and out of education for a fortnight. Fear and trembling across the community – but what likelihood the child actually caught COVID-19 in the middle of a low-incidence area? No way of knowing, as the pressure on testing means they’re seldom repeated.
Pingback: GESZEFT LEPSZY OD SZCZEPIONEK