Misinformation 1. PCR testing – the great non-hoax of our times

SUMMARY

Introduction

Sars-Cov2-19 pandemic- is it the greatest hoax ever?

So please excuse my attention grabbing title, but I thought I would take a leaf out of the websites which are claiming the Sars-Cov2-19 pandemic is the greatest hoax ever perpetuated on an unsuspecting public.

Pandemic front line

Headlines that 90% of positive test results are false have circulated widely, including an article in the New York Times, with online “experts” then pronouncing on the greatest hoax of all time. 

Can you imagine it – the pandemic is 90% less of a problem than we thought! 

Wouldn’t that would be a relief! But can it true? 

Of course there is more to this than meets the eye. Fact checkers to the rescue! So, where has this come from?

Conspiracy theorists

There is a plethora of well designed, authoritative looking websites which claim all sorts of pandemic related hoaxes.  It can be easy to get sucked in by them – they look authentic, they appear to be full of references and they play into vulnerable belief systems. It is easy to get driven from healthy scepticism to cynicism, then to believing that everything is so black and white it can only be a well organised global conspiracy. Contrary to their much claimed notions of being persecuted and censored, such websites are plentiful and easy to access.

For example, I’m on the mailing list of the popular Mercola supplement sales platform to sample their view that the pandemic has been intentionally created in a laboratory, that PCR testing has created a false pandemic, that Sars-Cov2-19 is no worse than flu, that Bill Gates is behind it all, and that it’s all about the “Great Reset” of global society into a new era of control of “Us” by “Them”. There are many sites far worse.

It’s the old currant bun effect, small flecks of appealing truth in a big stodgy bun of misleading pseudoscience.

Currant Bun Effect

It needs noting that Mercola’s website has 1.9 million new visitors a month, with many of them returning regularly and so has more visitors than the National Institute of Health, the central library of health science in the USA.

Mercola and others have been creating much confusion about PCR testing with claims varying from valid questions on its accuracy and use, to it being at the heart of a global conspiracy to imprison us all within our Fear – inventing a pandemic if you will. Some even claim the virus doesn’t even exist at all, manipulating an epidemic of distrust in this disturbed and disturbing world.  

Back in the summer the New York times story reported that 90% of positive RT-PCR tests could be wrong – that headline number was corrected to a still significant 60% at the end of the article. In many ways the NYT article was good reporting; it raised an important issue, highlighted the problem and helped define the issue more clearly.  Yet it fuelled the incorrect notion that the pandemic has been overestimated by a factor of 10. 

This is then passed from website to website, each time amplifying the story, hardly surprising given that some consider the virus to be non-existent. Dr Mike Yeadon has played a big part in this online show – however, he as is wrong about PCR as was his prediction that the pandemic would be over in October and that there is no need for a vaccine. 

As we shall see, digging a bit deeper reveals a different picture altogether. First, how does the PCR test actually work?

The RT-PCR test

The real time RT-PCR test by any account is a marvel of science in itself. The stages of the test go like this:

1. The swab is taken – itself a tricky procedure likely to miss cases.

PCR test

2. The swab is stripped of everything except the viral genetic material.

3. It is then mixed with a cocktail including enzymes (biological catalysts) to convert RNA to DNA (RT=reverse transcriptase), and more to unzip and duplicate the DNA (P=Polymerase) in a process which thus magnifies the DNA if it is present.

4. DNA markers (and fluorescing probes) of the virus are included which will tag any genetic material of viral origin, make it fluoresce and hey presto a positive test if the viral RNA was present.

5. The mixture is heated and cooled to enable the DNA to be doubled and doubled again, up to 40 times to enable its detection. (CR = chain reaction). Then a result which can be read according to the cycle which the test became positive, hence “real-time”.

The tests are developing too – most now use three gene targets rather than one. Indeed the new Sars-Cov2-19 variant is being detected because it is negative to the S probe currently used, though positive to the others. (N and ORF). 

PCR problems – the raisins in the bun..

No test is 100% accurate and problems exist because of the considerable number of false negatives, averaging about 16% though higher in some community settings. 

Inaccuracies come into effect from the moment the swab leaves its packet to the moment it is tested in the lab. Most of these are missed cases, and these matter when people infected continue to mix with others, infected staff continue to work, or patients potentially shedding the virus are kept in COVID free parts of a hospital. 

False positives are far less likely, though still cause problems for those who have to isolate as a consequence. Repeating the test in various ways would solve this. 

Mass screening populations without symptoms in an attempt to reduce asymptomatic spread has raised far more questions than answers in Liverpool, although this used a different test and different scenario. Mass screening of people without symptoms is a different matter and increases the proportion of tests which will be false positives. I’ll come to that in a minute. 

Like any test, there are uncertainties and problems, none of them are 100% accurate, they all need context. So a real problem is the under-diagnosing those who are infected, but that is not the question here, so what is actually happening?

PCR reality – the bun of nonsense..

The skeptics direct their fire at the false positive rate, which though far lower than the false negative rate, does cause problems too. False positives matter when indiscriminate testing results in restrictions for non-infections people. 

So how did the 90% headline come to be: It is true if you follow this logic – If the prevalence of COVID19 in the community is 0.01% (as in the middle of the summer) and the false positive rate of the test is 1%, then testing 1000 random people will result in 10 positive results, though with only 1 being real – hence 90% false positive – headlines follow! Global fraud exposed!!

True or false??

However, the tests are not being done on random people. They are being targeted at people who have symptoms or have been in contact with cases, (Pillar 2) or at health professions and carers looking after vulnerable people. (Pillar 1)  

As of Jan1st, the prevalence in the total tested population is 13%. This means that in a population of 1000 people getting tested, 130 are testing positive and 9 will be false positive.  Not such a dramatic headline in the real world.

Another real world fact is to consider the test context – for instance a pre test probability of 20% (new cough and fever for example), a test sensitivity of 80% and 99% specificity – in this scenario for every 16 positives there is one false positive. The more sensible the context of the test,  and the more common the infection, the less important false positives become. 

Further, if most of the tests were false positives, their rate would stay the same with time and location – they don’t. There are big differences between positivity rate in different parts of the country and at different times. There are lots of nails to hammer into the coffin of the notion of a false pandemic. 

Bear in mind too that a true negative test (the vast majority) is really useful to many people wishing to continue to offer care for the vulnerable and unwell, or just go about their business. 

The NYT story has been amply dissected by the Huffington Post and brought into the real world by the BMJ whose COVID risk calculator you can explore, indeed, play with here. These have not been picked up by the misinformation grinders above.

Suffice it to say the the combination of symptoms, situation, timing and test result work together to give the a pretty good decision making framework, both for us as individuals as well as a community. So it’s benefits outweighs it’s harms. Also, its the best test we have.

Cycle threshold cut off’s – the second greatest non-hoax

More scandalising from Mercola and others revolves around the PCR cut off threshold of 40 cycles which results in billions of copies (2 to the power of 40) of the original RNA and raises concerns that we are catching too many ‘small fish’ as the net it too small. 

Yet, the cycle cut off is not as big a deal as they state. It has to be set somewhere and in some ways the higher the cut off the better the more likely it is to capture infections accurately – the sensitivity of the test needs to be really high.

One way of looking at this is to compare the number of PCR positives confirmed by viral culture at these different cycle thresholds. This study has done just that. 

Clealry the lower the cycle threshold at which the test becomes positive, the more the viral load in that sample. The brown bars in the graph below show the positive tests at different cycle thresholds, the grey bars represents negative viral cultures. 

Following the theme of tests never being 100% accurate, viral cultures themselves may also miss infections and viral culture is no longer used routinely for diagnostics due to the technical  difficulties involved. 

With that caveat, it is correct that the percentage of tests which become positive according to viral culture at Ct of 35 is small, but these only represent 1.9% of the total positive tests in this sample. Indeed, this data shows that above a cycle threshold of 35, there are very few new PCR positives at all. That is why the majority of tests are true negatives – you can magnify nothing as many times as you like, the answer will still be nothing.

This graph shows that if the cut off were to be reduced to 30, as some have suggested, an unacceptable 8% of cases where viral culture is positive would be missed. At a cycle threshold of 35, only 0.7% of samples are PCR positive in any case. In other words, the Ct cut of 40 will generate false positives, but will also capture the majority of the real positives, critical for the sensitivity of the test

So why not give cycle threshold result?

It might seem this could be addressed by publishing the results of the cycle threshold at which the individual sample became positive. However, this is beset with technical problems, the need for internal controls and delays in results.  

Different PCR tests from different providers in the highly privatised sector can use different genetic markers and testing kit. Many of the ones in used right now use three, (spike, envelope and ORF) and Cycle thresholds can differ between them. 

You can imagine the scenario – a 55 year old with early diabetes has a cough, feels unwell and a positive PCR at a Ct of 30. What is the resultant advice – it is binary – either to isolate or not to isolate – there is little value in having a theoretical value of say 12% likelihood of that swab culturing the virus when the person needs to know what to do. 

PHE has addressed this in advice to retest those results at the limit of detection and further defined ways this can be addressed, including repeat tests using different genetic markers on the same sample, repeating the test altogether and of course, looking carefully at the clinical context. 

They say:

“A positive result at the limit of detection from the repeat or retested sample is suggestive of the late stage cycle of infection, but should be interpreted in the context of the clinical presentation. Isolation and contact tracing should be determined based on whether testing is thought to reflect early or late stage infection”

A little trust is needed

Trust can be helpful

So for me, the RT-PCR test is a marvel of science, and like all marvels of science, needs to be used well.