Discover more from Where are the numbers? by Norman Fenton and Martin Neil
Sensors set to Off
Did Covid-19 quarantines hide the flu?
Quarantines were promoted as measures to reduce spread of SARS-CoV-2 and are claimed to have prevented the spread of flu.
Flu tests are recommended to be administered within 4 days of symptom onset. If they are administered after 4 days, they would likely produce a false negative result for someone with flu (flu tests are rarely administered routinely anyway).
Mandatory Covid-19 tests, run at high cycle thresholds and suffering from cross-reactivity with other pathogens (amongst other operational issues), may well have resulted in false positives for Covid-19, when in fact the pathogen causing symptoms may have been flu.
Therefore, people with flu would have been wrongly categorised as having Covid-19, and as a result quarantined for a period sometimes up to 14 days. Hence any flu test given after quarantine ended would inevitably result in a negative for flu even if that was the causative agent because it was given later than the 4 days needed for the flu test to be accurate.
Compared to Covid-19, diagnosing flu ‘out of season’ is fraught with tricky clinical and bureaucratic barriers, which also served to depress the likelihood of reporting flu cases. Even armed with a positive flu test result the chances of this overruling an all-pervasive prior belief in covid-19 being the cause of all respiratory illnesses, encouraged by powerful incentives directed by a centralised bureaucracy, would have been close to zero.
In combination we propose that these primary mechanisms, rather than ‘viral competition’, could partially or wholly account for the disappearance of flu.
This is our third article in our flu series. In our last article we raised questions about flu testing and viral interference between SARS-CoV-2 and influenza, and in the first we noted that the flu had not completely disappeared in 2020/21 as had been claimed. The focus there was on PCR testing for both viruses, but here we look at the interplay between policy decisions, and in particular quarantines put in place for SARS-CoV-2 and we ask whether these quarantines themselves, in conjunction with testing policy, are enough to explain the disappearance of flu.
Everyone “knows” that flu disappeared in the Winter of 2020-2021. The stark juxtaposition of its absence and its replacement by the novel and deadly SARS-CoV-2 virus has been visually communicated by John Cullen here.
Identification of any virus relies not only on the test itself but also on the protocols, procedures and the public health bureaucracy that govern when the test should be administered and how the test result is validated, interpreted and reported. So, a systematic assessment of the effects of seemingly unrelated policy decisions is needed to determine whether policies were enacted - wittingly or unwittingly - which brought about a particular result as a kind of ‘spooky action at a distance’ that caused flu to appear to vanish from some countries but not others.
Quarantines were promoted as measures to reduce spread of SARS-CoV-2 and are claimed to have prevented the spread of flu. In this article, we investigate the possibility that what they actually did was dramatically reduce the chance of a positive flu test result. If you don’t have a positive flu result, there is little possibility of being diagnosed with flu in the presence of an extremely contradictory explanation. Given that PCR tests for SARS-CoV-2 were mandated (when not enthusiastically and voluntarily performed by a populace terrified by propaganda) there was a very high chance of being diagnosed with covid-19 instead of the flu.
Where are the numbers? by Norman Fenton and Martin Neil is a reader-supported publication. To receive new posts and support our work, consider becoming a free or paid subscriber.
Algorithms and Policy - Context is King
When examining the algorithms that govern the application of any tests we need to focus on the overall context, including:
Pre-conditions that exist that determine whether the test is encouraged or allowed to be done.
Limitations on interpretation of the test result and whether it needs to be validated (i.e., confirmed by a repeat of the same or another type of test).
Official information made available about the test performance specifications which influence the informativeness of the test result given the context of use.
The probability of the patient having the virus given a positive or negative test result and how this might vary according to the background prevalence of infection.
The cognitive dissonance at play when deciding what to do about test results for competing viruses (SARS-CoV-2 and flu)
So, a flu test result is only one part of the decision-making process that determines whether you have flu.
By now we are all pretty familiar with the problems associated with SARS-CoV-2 testing.
But before we look at flu testing let’s remind ourselves of the CDC protocols on SARS-CoV-2 testing and in particular testing in ’broad settings’. Here the assumption is that testing takes place en masse with social distancing, masks, pre-screening, temperature checks and at special screening centres. Within these documents there is a huge amount of information on the context and conditions for testing - creating an impression of a wartime emergency, and thus reinforcing the view that SARS-CoV-2 is deadly.
But nowhere do these documents present an algorithm for deciding if a patient is infected with SARS-CoV-2. It’s absent and entirely implicit. The idea that there is an outbreak is taken for granted. There is no recognition of any competing explanations for symptoms, such as other viruses, and because of this any reference to the possibility of false positives and negatives is entirely absent.
Ultimately, clinical judgement regarding Covid-19 is delegated from the physician to diagnosis based on PCR-test. Cultured samples aren’t mentioned, and neither is confirmatory testing. In other words: “It’s all about the test”.
Influenza Testing Algorithms
Now let’s look at the CDC algorithms for the flu, the governing protocol for which is documented here (10th August 2020).
In contrast to SARS-CoV-2, flu testing is remarkably straightforward but loaded with qualifying criteria that the physician is instructed to actively consider.
PCR is only recommended for symptomatic hospitalised patients but for symptomatic outpatients the rapid molecular assay test should be considered. Other, less reliable, tests, such as rapid antigen tests are only recommended if other tests are unavailable. Culturing of virus samples is only recommended for the purposes of confirmation. Furthermore, testing is recommended within 4 days of symptom onset, presumably because of the decay in the presence of live virus and the reduced capability of tests to detect virus in diminishingly small quantities.
It is fascinating to learn that flu testing is governed by a single question acting as a pre-condition to the testing process: Is flu in circulation in the community? When there is no epidemiological outbreak present a completely different algorithm applies, compared to when flu is acknowledged by the public health authorities to be in circulation.
This is the flowchart to be used when there is no outbreak and flu is known to not be circulating:
In the event of a positive test the clinician is asked to pause and consider if this is a false positive. Furthermore, they would need to justify any decision to support the positive result and diagnose flu with an assessment of whether there is evidence of an epidemiological link between this case and others (i.e., link to existing circulation in the community).
Likewise, the clinician would also have to consider the signs and symptoms of flu, but given these will heavily overlap with covid, which the authorities are proclaiming as an epidemic, it looks like the cards are stacked against them.
Finally, it calls for the clinician to initiate antiviral treatment as a consequence of the positive test result, thus incurring expense and potential waste.
Here is the flowchart to be used if flu is circulating:
The default here is that influenza is highly likely given the flu test result is positive. There is no call to reconsider the symptoms or to consider if there really is a genuine outbreak. Also, there is no call to consider expensive anti-viral treatment (whether or not there are incentives to use them depends on norms of practice in different locations). All in all, this is an easier call to make than diagnosing flu when it is ‘not in circulation’.
There is an elegant logical circularity at play here that a doctor needs to consider. You need an outbreak and an epidemiological link to help justify a positive influenza test, but surely you only know there is an outbreak, and can determine an epidemiological link, if you and others, in coordination, have already accumulated enough positive test results. It’s a chicken and egg situation. Who determines whether there is an outbreak? None of the CDC documents say. Clearly individual physicians do not have the authority to make such a decision themselves. A hospital’s authorities could maybe do so after liaising with public health authorities, but at a national or international level such a decision would be in the hands of the centralised public health bureaucracy.
So, compared to SARS-CoV-2, diagnosing flu ‘out of season’ appears fraught with tricky clinical and bureaucratic questions.
Accuracy of flu testing
The controversy around the accuracy of the SARS-CoV-2 PCR test is well known, but there is much less discussion about the accuracy of the flu tests. The reported accuracies for the flu rapid molecular assay test are that they have a perfect specificity (!) and high sensitivity (between 90-95%).
The one thing the CDC are very careful to advise about is the time from onset of symptoms to the application of the flu test, which they say should be less than four days. This is the time window for viral shedding. Any upper respiratory tract test will be more reliable the shorter the delay between symptoms and administration of the test. Waiting for longer than four days might lead to the test failing to detect live virus in someone who has suffered or is suffering flu.
The seminal paper on influenza and immunology was written by Nobel laureate Peter Doherty (Doherty et al.) and appeared in nature in 2006, and states:
Influenza viruses grow rapidly in the human respiratory mucosa, allowing transmission to colleagues and family members via respiratory droplet inhalation (coughs and sneezes) even before the development of obvious symptoms.
Exposure to flu for the first time provokes a primary response by the immune system which reduces viral titres to zero within 10-12 days. So, someone young, who contracts flu for the first time in their lives, would largely clear the virus within 10 days. Someone older who has contracted flu more than once in their lives will enjoy a secondary immune response and will clear the virus within five days.
Liu et al report that co-infection rates for flu and SARS-CoV-2 should be much higher than reported simply because of the shorter time window for incubation and viral shedding in flu compared to SARS-CoV-2:
A positive flu test should only occur when there is a sufficient amount of flu virus to detect. For those who are older and have been pre-exposed to the flu this means the test needs to be applied after symptom onset, giving a target time window of 4 days. For younger people, previously unexposed to the flu, the target time window to apply the flu test would be, at most, 10 days. After that the chance of a positive would reduce substantially.
Quarantines and their effects
So, what could cause a delay in testing and a reduction in the chances of a test detecting flu virus? Policy requirements for SARS-CoV-2 quarantines of course!
From Spring 2020 and throughout 2021 the advice was that a person should not visit a GP or a hospital if they had respiratory symptoms, on the assumption that these symptoms were diagnostic of SARS-CoV-2.
Even if they had no symptoms but had contacted someone who tested SARS-CoV-2 positive they were required to quarantine for up to 14 days and advised to only seek admission to hospital if a respiratory condition developed and then they significantly deteriorated.
They would then have needed a negative covid test to re-enter society, sourced from a centralised centre. Note that these centres only tested for SARS-CoV-2, hence flu was never tested for.
If symptoms were severe and someone was admitted to hospital a SARS-CoV-2 PCR test admission would have been performed immediately. People with symptoms would already have stayed at home, in quarantine, for a period up to two weeks from symptom onset. Thus, crucial time was lost from symptom onset to presentation to hospital or GPs, and this introduced a delay, supposedly to keep society safe.
So, the vast majority of people with respiratory symptoms would have stayed away from all health services for much longer than four days.
After four days the accuracy of the flu test would have deteriorated significantly, and had they been tested for flu in all likelihood the test result would have been negative given it was done up to two weeks after symptoms started.
Likewise, the availability of flu tests was resource limited, yet covid-19 tests were readily available. The effects of this on covid-19 and flu rates were recognised in the UK national influenza report which said:
It is inevitable that influenza activity will be low if there is limited testing for it!
In parallel with this, the ready availability of a mandated covid test, subject to excessively high cycle thresholds (or deliberate gene miscounting) would have increased the chances of a covid false positive test result.
Before the widespread adoption of multi-viral PCR test kits, which test for multiple viruses in parallel, the vast majority of testing done in 2020 and 2021 would have involved sequentially testing people for different respiratory viruses. Here the decision to conduct a subsequent test would only be made following the obtaining of a negative previous test. This process efficiently minimises the amount of testing done and allows the physician to target the most likely pathogen. So, even if a physician was considering flu as a possibility the SARS-CoV-2 PCR test would always have been the first test performed and given the high likelihood of a positive test result subsequent tests for flu would not have taken place. And as we say, if a flu test was performed it would likely show negative for flu even if that was the causative agent because of the time delay introduced by quarantine.
Even if the SARS-CoV-2 test was negative and the flu test was positive the flu test result would likely have been discounted as actually a false positive. Why? Because the CDC flowchart instructs the physician to suspect that it is!
The rationale would go something like this a) there is no outbreak of flu b) there is no epidemiological link to get the result accepted by health authorities c) covid-19 disease is deadlier than the flu so a diagnostic misclassification would be risky and d) the antiviral treatments for flu do not apply to Covid-19.
If a flu test was done and it was positive the physician might have a conundrum if there was a positive SARS-CoV-2 test result for the same patient. How would this be resolved? There may be potential for cognitive dissonance here. Recall that all advice on Covid-19 diagnosis was to assume, even in the absence of a SARS-CoV-2 positive test, that the patient had been infected with SARS-CoV-2. With the weight of coercive pressure on the side of SARS-CoV-2 diagnosis, and with the incentives in place to diagnose SARS-CoV-2, it is inevitable that there was no logical reason for a physician to consider that a flu test result is anything other than a false positive. And even more so given they have been told that flu had disappeared and that there was a SARS-CoV-2 epidemic and ‘flu had been driven out by SARS-CoV-2”.
Imagine going to your boss admitting having done a flu test and presenting them with a positive when the WHO and every health authority in the world has said it has disappeared? It would be professional suicide.
A clinician’s cognitive decision-making process can be simulated by a Bayesian model showing how all of the above factors combine to drive a diagnostic decision that favours Covid-19 and supresses any chance of a true positive for flu. See the appendix at the end of this article for this.
Prior to 2020 some patients would have been picked up by sentinel testing in outpatient settings or at GPs and would have been reported as a flu case if they tested positive simply because they would have been tested in time. But post spring 2020 until winter 2021 the chance of such a positive result from the medical system would have been vanishingly small even if testing was performed, which was extremely rare.
Also bear in mind the sheer hysteria around Covid-19 in 2020 and 2021 would have affected clinical decision making. Key variables would also have heavily influenced decision making, including unfounded assertions from governments and the mainstream media like:
Symptoms previously associated with the flu were now firmly associated with SARS-CoV-2.
Asymptomatic people were told to assume they had the SARS-CoV-2 virus.
SARS-CoV-2 was recognised as more infectious and deadlier than the flu.
In conclusion the delays built into the quarantine system may have created a substantially depressed likelihood of detecting and diagnosing flu, if someone was lucky enough to get a flu test. Similarly, there would have been a hugely increased chance of a positive being overridden by a covid false positive or being dismissed as a false positive anyway, given there an epidemic of SARS-CoV-2 had been declared and flu was said to have disappeared.
Where are the numbers? by Norman Fenton and Martin Neil is a reader-supported publication. To receive new posts and support our work, consider becoming a free or paid subscriber.
Appendix - Bayesian model
Let’s assume the SARS-CoV-2 test and flu tests have the same very high accuracy (sensitivity 90%, near perfect specificity of 99.99%) and that they share the same symptoms. Let’s also assume that if the flu test is taken late (after 4 days) that the sensitivity drops to 10% after that point - so even if you have flu the test only has a 10% of showing a positive. For the SARS-CoV-2 PCR test we assume a high cycle threshold is applied with a 99.9% chance of producing a positive result whether SARS-CoV-2 is present or not.
Also assume the prior probability of someone contracting flu during a flu outbreak is 10% and we have the same probability for someone contracting SARS-CoV-2 during a SARS-CoV-2 outbreak. Of course, in practice the prevalence would fluctuate.
We also know that the presence of the most common symptoms alone cannot discriminate between flu and covid-19, but let’s assume we have a patient with symptoms that could be attributed to either (the model could be varied by including the so-called novel symptoms supposedly attributable to SARS-CoV-2 but not flu).
Finally, let’s assume that we believe there is a 90% probability there is a SARS-CoV-2 outbreak and a complementary 10% of there being a flu outbreak at the same time (this is an extremely generous assumption given we were all told flu had disappeared entirely).
We can build a Bayesian network model representing these assumptions and probabilities.
Let’s consider some cases discussed previously:
Case 1: Flu test taken after a period of quarantine, which is negative coupled with a SARS-CoV-2 test, run at a high threshold, which is positive.
This case represents patients who would present with flu symptoms and were subject to both flu and SARS-CoV-2 tests (itself unlikely). You can see that the posterior belief that the patient has flu is very low, at just under 10%, given the flu result is negative.
In case 1 a covid-19 diagnosis would be concluded.
Case 2: Flu test taken after a period of quarantine but is positive coupled with a SARS-CoV-2 test, run at a high threshold, which is negative.
Case 2 represents the ‘best chance’ scenario of a positive flu test to influence the diagnostic conclusion, but even in the face of a negative SARS-CoV-2 test, because the flu test has been applied late, and given the high prior belief in a SARS-CoV-2 outbreak the posterior belief in flu is approx. 18% and for covid-19 it is approx. 82%.
In case 2 a covid-19 diagnosis would be concluded, despite a negative test for SARS-CoV-2. This is entirely explained by the high prior belief in a SARS-CoV-2 outbreak and that flu has ‘disappeared’.
With this model both of these cases clearly show that with some quite reasonable assumptions patients suffering from flu would be diagnosed as having been infected by SARS-CoV-2. The majority of such people would not have been subject to a flu test, having been given a SARS-CoV-2 test as a matter of priority, and even if a flu test was administered, a positive result would have been disregarded as a false positive on the justification that the flu had disappeared, the risks of covid-19 are so much greater and that administration of flu anti-virals, and perhaps more importantly the administration of antibiotics, would not have been justified by covid-19 treatment protocols.
But what people don’t realise is that in Japan, New Zealand, Jamaica and South Korea it disappeared in 2021/22 as well as in 2020/21 whereas in most Western countries, including Australia, it disappeared only in 2020/21. This can be easily verified using FluNet.
In the UK, annual flu numbers are estimated using modeling techniques, with the most important data used being that obtained from “sentinel testing” sites, which are specific GP surgeries who are asked to perform flu testing on anyone with symptoms suggestive of flu, for public health planning purposes. There is no easily accessible UK information equivalent to that issued by the CDC in the USA, but it can probably be assumed the general principles are the same.
These are CDC numbers. Values from research papers can vary wildly, sometimes down to around 60%. Also, specificity numbers are usually not given but implicitly assumed to be 100% based on a very small number of challenge samples.