Pooled pcr testing for covid-19 - worth considering?

major Update - 5 April 2020

This methodology has been proven effective by German researchers!


Congratulations and thanks to the German Red Cross Blood Donor Service, Professor Erhard Seifried, Professor Sandra Ciesek, Goethe University, and all involved.

I imagine that these scientists and institutes conceptualised this breakthrough independently... Though for the sake of my own pride, below is what I proposed on the 3rd of March, 2020.

Testing of possible COVID-19 patients is slowed by the challenges faced in diagnosing cases of any new pathogen. Scientists around the world have done excellent work, making many excellent tests available rapidly. Though the provision and availability of sufficient quantities of these to clinical pathology laboratories is a challenge; this is entirely understandable! 


The most-useful tests at this time seem to be those based on the polymerase chain reaction (PCR) - the same reaction that allows extremely tiny specks of bodily material left at crime scenes to be used as DNA evidence in court cases. The reaction produces an exponentially expanding amount of the same DNA within the sample. PCR tests are desirable for a number of other reasons too - for one, the samples of a suspected virus case can be inactivated, making them far safer to transport, yet still providing useful diagnostic data.


So what if there was a way to test multiple people for CoV-19 infection in just one test? It’s possible that there is! And we might already have it!


Veterinary medicine pools samples from multiple herd animals for diagnostic testing often - it’s faster, cheaper, and in many instances just as accurate as testing individually. 


When one seeks to verify an entire herd of animals is free of a disease, for several diseases, it is not necessary to test each individual animal separately. Rather, samples from the animals (or a representative sample of a herd) are collected (either individually, or in some cases - for example, bulk milk samples - collectively). Then, at diagnostic laboratories, individual samples are combined, and tested as one pool. If the pooled sample is entirely negative, the herd can be cleared. This is a brief over-simplification and intentionally vague example (honestly, it’s a really boring subject).

There’s no scientific reason the same cannot be considered for use on human samples. Though for clarity, obviously the samples themselves should be taken individually. After all, we are, as Katie Hopkins demonstrates continually, actually just relatively simple apes (often comparatively racist, stupid apes; but apes nonetheless). 

Any test’s diagnostic utility is determined by it’s sensitivity and it's specificity. Sensitivity is the proportion of true positives that are correctly identified as such. Specificity is the inverse; the proportion of true negatives that are identified as being truly negative. 


No test is perfect; in the case of COVID-19 PCR tests, it seems likely these have low sensitivity but high (close to 100%) specificity - it is prone to false-negatives, but not to false-positives… Though theoretically, only a small amount of initial viral RNA is necessary for the PCR to replicate it to a point where its detection is possible. 


Theoretically, the PCR media in which this takes place can have samples from multiple people placed into it… If none of these samples contain COVID-19 virus, the test would return a negative result. Whereas if one out of, say, ten people tested in parallel like this did shed virus into the sample swab, this will multiply in the PCR media just as it would if the media only contained a sample from one person. 


Accordingly, it may be wise to take multiple diagnostic specimens from potentially infected patients; submit all of these to a laboratory, where a collection of [x] patients’ samples are pooled, and the pooled sample subjected to testing. If a positive result is returned, further testing on an individual basis can occur. If a negative result is returned, a far greater number of possibly infected patients can be cleared at a time. This would allow more people to be tested, faster, with theoretically limited impact on sensitivity, and - after follow-up individual testing of positive pooled samples - an even further limited (possibly no) reduction in specificity.


The increased availability of testing may have a substantial reassuring impact on concerned patients who have respiratory disease symptoms; they can be tested, without quite so strict case definition requirements to justify testing; and most will be cleared to at least some extent. Furthermore, it may vastly increase the rate at which people who are true positives are accurately diagnosed within a rapid timeframe - we’ll simply be able to do far more testing!


Note that this is entirely speculative, based on scientific insight and guessing - I do not know what impact such a regime would have on sensitivity or specificity, nor if it is possible or practicable for the current tests. This is just an idea that seemed like it needed to be put forward!

Update 31/3 - This might work. If it does, it can save many lives, around the whole planet.

Note; I do not know if this will work, nor if it is feasible. Though I can see no particular reason why it would not be possible to try this. But if it is feasible to try, it's worth trying even if there's only a 1% chance it might work. I believe the chances of it working are substantially greater than 1%!


See Real-Time Reverse Transcription–Polymerase Chain Reaction Assay for SARS-associated Coronavirus (opens a new window) - Emery, et al; Emerging Infectious Diseases; 2004 Feb; 10(2): 311–316.


Riveting reading I know.

Here's the key part... Note that this is for the first SARS, from 16-17 years ago...

"...To assess the efficiency of amplification of the RNA transcripts in the presence of exogenous nucleic acid and potential RT-PCR inhibitors, 10-fold serial dilutions of the RNA transcripts were prepared in water and pooled total nucleic acid extract from 20 SARS-CoV–negative human respiratory specimens (nasopharyngeal aspirates, bronchial washes, sputum, naso- and oropharyngeal swabs, and lung tissue)...
Exogenous nucleic acid had no discernible effect on amplification efficiency of the SARS1 and SARS3 primer/probe sets 

...as demonstrated by the similarity in linear regression slopes and endpoint detection limits in the presence and absence of specimen extract (Table 3 [new window]).
In contrast, the standard curve for SARS2 had a more efficient slope (–3.21) in water than in the presence of spiked extract (–3.48) and with greater variation in the CT values at 20 target copies or lower, suggesting that the amplification reaction was less efficient in the presence of the specimen extract. This observation was confirmed on two additional repetitions of the same experiment."
In layperson's terms, diluting a sample taken from a known-positive case of SARS-CoV-1, by a massive amount - ten-fold serial dilution - then mixing with twenty samples taken from a variety of other known negative patients, prior to the 'amplification' step of the Polymerase Chain Reaction, for two of the three SARS genetic sequences (SARS1 and SARS3) had no impact on the amplification efficiency of the still-detectable, and discrete/discernible, positive sample. 
For the SARS2 marker, these experiments found that there was an impact on the efficiency of detection, and greater variation on results, when combined with the known-negative samples. 



OK great I'm really bored now. What does this all mean?
The SARS of the early 2000's is the same type of virus as COVID19. The tests we're using have the exact same principles behind them; they are simply targeting new, specific-to-COVID19 sequences of nucleic acid.
So this means, this is worth testing. If this works, it massively changes the rate at which entire clusters of people can be screened. 
Conveniently, for the forseeable future, Aotearoan New Zealanders are all tucked neatly inside their self-isolation bubbles. 
Perhaps better still, it may be possible to take only swabs from the oropharynx of people prior to pooled testing... This is because; if similar results as obtained by Emery are achieved; while people are clustered as part of The Great New Zealand Lock-down; a single positive is less likely to occur within a cluster (bubble), compared to the likelihood of multiple cases existing there. Thus if only one in the cluster sheds COVID19 genomic material into the pool, it - theoretically - would still test positive for the whole pool... Follow-up testing via nasal swabs and individual testing could then occur.
For negative pools, if the possibilities I have outlined are indeed accurate, it may be possible to eliminate groups as large as twenty people in one test
That is worth pursuing. This might save lives; not only here, but globally.


Dr J Schofield


(I normally sell brake pads, and obsess over the Battle of Britain and Sir Keith Park... But this whole diseases/herd health thing is where a lot of my formal training lies!)