At the time of writing, debate continues to reverberate around the role that mass-population screening may or may not be capable of playing in giving society back its freedoms as we await a vaccine to increase immunity. Whilst there are several high-volume tests being developed, their deployment may not come in time to address the current second wave. Instead, I’d like to advocate a technique called pooled testing. It demands some very tricky maths, but crucially requires no change to critical pieces of established equipment. I see it as a truly practical approach to increasing highly accurate throughput by an order of magnitude. 

Shooting for the moon: how to test 10 million samples for COVID-19.

Population-wide mass screening has some unique requirements. First, it needs to be simple and reliable enough to deploy when depending on rapidly trained staff working in car parks or care homes. Second, it must have a very low false-positive rate. Indeed, a rate of just 1% would lead to more than half a million of the UK’s 65 million population being wrongly diagnosed. The knock-on impact for society could be huge, even if secondary testing was undertaken. Every school, every workplace, every town and village would have to deal with suspected cases, potentially closing us down even more. 

As my colleague Andrew Goulter explained in his own opinion article on testing, the current gold standard utilises real-time reverse transcription-PCR (RT-PCR). With this technique, RNA, a single stranded form of viral genetic material analogous to DNA, samples are amplified by producing millions of copies. The process allows the sample to be studied in detail and provides an established and highly accurate route to a diagnosis. Given the need for combining high accuracy with the lowest possible false-positive rate, RT-PCR is an appealing technique. Especially as it is established in many current clinical test settings. 

Upping throughput for population-wide screening 

But the downsides are that RT-PCR devices are expensive, and the physical process of amplifying RNA samples is slow – typically around an hour. This can be offset to a degree by utilising 48 or 96 individual sample containers, microtitre plates, per cycle effectively reducing the sample preparation time to the order of a minute per sample. But to get close to the throughput required for population-wide screening, this throughput will need to increase by a least another 10 times. 

Pooled testing, a recently published approach, may provide this increase whilst not requiring extra PCR machines. Imagine your sample being added to a single well within a microtitre plate. Following amplification, this will be tested and scored either positive or negative. Now imagine you share the well with seven other people's samples. If any of them are positive the well will score positively. If you then spread your sample over a further seven wells, all also mixed with seven other different people you can imagine a pattern of wells lighting up if you were positive. This pattern is identifiable and unique to you. If this is extended for more positive samples a number of unique patterns will light up.  

The maths starts to get very complicated when you try and untangle further multiple positive samples. But critically, it can be proven to work assuming the percentage of positive tests is below approximately 2%. Population-wide screening should fit comfortably below the 2% positive threshold – providing a potential route to higher throughput without an increase in expensive and limited PCR machines.  

Addressing practical issues 

A number of practical issues need to be addressed to turn this into reality. Reliably mixing multiple samples loaded into specific microtitre plate wells will require an additional sample preparation step. This will probably be implemented as an extra device which could, however, be constructed from existing commonly available parts such as automated pipetting systems and so forth. The real complexity is in the software, especially when it comes to testing a complex mixing and control system which leads to a critical diagnosis. But software is highly scalable in development and multiple teams could develop solutions to reduce development time.  

Interested to find out more?

In summary then, population-wide screening for Covid-19 requires systemic changes to the way in which samples are processed. An obvious route is to increase the quantity of testing devices. This, however, is not always straightforward to implement as such devices are by nature in high demand and often require complex training and specialised staff to use. By pooling samples following a sophisticated but defined mathematical approach, another route opens itself up. It increases the throughput of standard equipment, reducing the gap between that currently achieved and what is forecast to be required. It is a practical, pragmatic and sensible approach to better protect the population against the virus.

Symon Cotton
Associate Director