Coronavirus testing has been a central part of many countries’ response to the pandemic. The UK government in particular has been prepared to spend up to £100 billion on testing (although this plan has since been abandoned) and recently announced plans to enable all citizens to perform up to two rapid tests per week.
Coronavirus tests can help to prevent the virus spreading by identifying infected people and their contacts, who can be asked – or told – to isolate. This means a positive test result can have severe effects, potentially closing schools or businesses and causing whole families or groups to go into legally enforced isolation.
But is focusing on positive test results and using them as justification for curtailing people’s liberties really the best approach for tackling the virus? Is it also the best use of the money?
Pursuing the right goal
If our goal is to stop an international pandemic and return our lives to something approximating normal, we must understand how the measure of positive coronavirus tests can be used to inform progress towards this goal. It would be a serious mistake to confuse reducing the number of people testing positive for coronavirus with the overall aim of getting back to normal.
This is because the SARS-CoV-2 virus is likely to be around for a very long time. It is unlikely that we will ever be able to eradicate it, and similarly new variants may appear that increase infections. Much as we do with influenza, we need to learn how to live with frequent and possibly high levels of coronavirus infections.
This is not as much of a problem as it sounds if we can work out how to protect the most vulnerable, prevent deaths and ensure our health systems are able to function despite the coronavirus threat. Although the number of positive coronavirus tests is related to these other health indicators, it is not quite the same measurement.
This is not to say that we should stop coronavirus testing altogether. But we need to recognise that our overall goal is more complex than just trying to drive down the number of positive test results.
So what are we testing for?
There are three main coronavirus tests in use at the moment, each measuring different things. Antibody tests look for the presence of coronavirus-fighting antibodies in blood samples. Lateral flow tests look for the presence of viral proteins (antigens) in mouth and nose swab samples. And polymerase chain reaction (PCR) tests identify the presence of viral-related nucleic acids (DNA or RNA) in swab samples.
Antibody tests tell us how many people have recently had a coronavirus infection. However, these tests are currently not widespread, because while they are interesting from an epidemiological or clinical perspective, they measure infections after they have happened.
Lateral flow tests, on the other hand, are being recommended for widespread and frequent use. These tests tell us (with a few caveats depending on how they are made) whether individuals have active virus in their bodies.
However, while their advantage is that they are cheap and quick to perform, there is also a one in 1,000 false positive rate. Although this number sounds very low, if the aim is to ask all 67 million people in the UK to use the test twice weekly, this could lead to over 100,000 false positives each week.
Due to this error rate, the latest advice is now to follow up each positive lateral flow test with a confirmatory PCR test. PCR tests are significantly more sensitive than lateral flow tests and work by detecting DNA or RNA in swab samples.
They can tell us both whether a person has COVID and, with some minor modifications, which strain of virus they have. This is more relevant to our overall aim of learning to live with the virus, because the results can help to inform the development of treatments (like vaccines) targeted at the most dangerous strains, along with providing a reliable indicator as to when significant local outbreaks may be occurring.
But critically, to get this key information we do not need to get everyone in the population performing lateral flow tests followed by confirmatory PCR tests as frequently as is currently being recommended. Instead, it is perfectly possible for statisticians and scientists to design sampling strategies that will test far fewer individuals, at regular intervals, chosen to represent both geographic areas and specific at-risk communities. Such strategies have already been proposed for use in coronavirus monitoring, and the UK’s Office for National Statistics is currently using sampling successfully to calculate the rate of transmission (the R number).
Sampling can also be combined with other techniques to keep costs down. Rather than running tests on swabs individually, models have shown that by pooling them together to test them in groups, ten times fewer tests are needed, costs can be reduced by almost 70 per cent, and 20 times as many people can be cleared from isolation for each test run.
Sampling a much smaller number of people, in an intelligent way, will give us early indicators of when we need to take action on new variants or possible spikes in hospital admission rates. Critically, we will receive these indicators without curtailing the freedom of many families and workers, because instead of using tests to justify taking away people’s liberties, we will be using them to measure progress towards goals that really matter.
Importantly, testing in this way does not confuse the potentially impossible target of trying to reduce the number of positive coronavirus tests with the more important (and hopefully achievable) target of learning to live safely with the coronavirus.