The government’s "game changer" antibody tests are inaccurate. What next?

Why is it so hard to develop a relaible home test for coronavirus antibodies and what does that mean for our fight against the disease?

Sign Up

Get the New Statesman's Morning Call email.

The coronavirus antibody tests that Boris Johnson hailed as a potential “game changer” have, it turns out, changed nothing at all. The government had ordered millions of pregnancy test-style home kits, which have now been found to be ineffective, Public Health England’s Professor John Newton told the Times on 5 April.

What next? Professor Newton is “optimistic” about finding reliable tests within months and scientists from the University of Oxford are working with manufacturers to improve results. For ministers, good news can’t come soon enough. Antibody testing, which detects who has previously been infected with Covid-19, is central to the government’s strategy for tackling the virus. Once proven in the lab, these tests could “potentially be done at home with a finger prick” and deliver results within 20 minutes, according to Health Secretary Matt Hancock’s five-point coronavirus testing plan unveiled 2 April. In time, test results could guide the government’s social distancing policy. 

But developing accurate home antibody test kits is fraught with challenges, and some experts predict the government may need to initially roll out kits with high error rates while officials try to source better tests. 

The difficulty comes not from the chemistry of the virus itself, but in translating a laboratory process into a simple at-home test in record time, explains Professor Danny Altmann of Imperial College London. “If you’re doing laboratory medicine, you can take a tube of blood, spin it down, collect the serum, pipette it into a plate and collect very fancy data. That’s been very doable for this virus, and gives very clear answers,” he says. “All of the problems come when you try to do a very rapid translation of that into… the pregnancy-test style yes or no answer, which normally would be evolved by industry through a lot of iterations over a few years.”

Additionally, scientists still do not have a full picture of the timing of the antibody peaks and troughs in infected people – a fact further complicated by differences in every individual’s immune response to the virus, says Professor Lawrence Young, virologist and pro-dean at the University of Warwick. “You have to work out what we call the antibody titre: how much antibody is there, and is the test going to be sensitive enough so that for those folks a month post-infection, six weeks post-infection, the test can still detect that antibody?” 

Dr Andrew Preston, reader in microbial pathogenesis at the University of Bath, says scientists developing the kits must accommodate groups with vastly different antibody profiles, which involves testing all age groups as well as individuals who were infected but were asymptomatic (so far, antibody tests have only proved effective in patients who had suffered from severe Covid-19). 

And even after this “extensive” process, antibody tests may only be as good as our understanding of coronavirus immunity, which remains patchy.  On 6 April, South Korea’s Centers for Disease Control and Prevention identified 51 patients who tested positive for coronavirus despite previously recovering from Covid-19. Officials said it was more likely the virus had been “reactivated” than that the patients had been reinfected. “We still don’t know for sure that if you have these antibodies that you’re immune,” Dr Preston says. “In the vast majority of infectious diseases you’ve got at least a window of immunity where you’re protected against going down with the disease again. But there are plenty of situations whereby you might not go down with disease again, but you might still be able to carry the virus and transmit it to others.”

Identifying a test suitable for mass rollout will involve evaluating the more than 100 commercial antibody tests currently available. Professor Marion Koopmans, head of virology at the Erasmus University Medical Centre in Rotterdam, told the Guardian that the tests her lab had analysed were only identifying around 50-60 per cent of infections where the individual only has mild symptoms. “Based on what I’ve seen so far I’d be hesitant to give you an immunity passport based on a rapid test result,” she said. Matt Hancock has said that “no test is better than a bad test”. But how good is good enough?

It depends entirely on what the government hopes to do with the results. Here, experts believe there are two options: first, results of home tests could help build a clearer picture of just how widespread coronavirus is, and second, data could inform policy decisions, including who can safely return to work. If tests are used for data purposes, a wider margin of error is acceptable, Professor Altmann says. “If you wanted to get a broad brush epidemiology picture of how big our iceberg is, how much herd immunity we have, I wouldn’t care if we had a 10 or 20 per cent wobble in the answer, because I can get a ballpark picture.”

But if the government uses the test to inform its social distancing policy, as it eventually intends, antibody tests will need to be more accurate. Most of the experts cite targets of 90-95 per cent or above for both the test’s sensitivity – the proportion of real infections the test detects – and for specificity, which is the proportion of healthy patients correctly identified. A false negative tells an individual they lack the antibodies when, in reality, they may be immune. A false positive is potentially more damaging: it could lead to a person returning to work thinking they are immune when in fact they can still catch and spread the virus. 

“What we’re really looking at is using a test as a proxy for whether you have immunity or not,” says Dr Preston. “So that changes the picture somewhat: you might need much higher sensitivity for that if it’s going to be the basis for saying to a health worker: ‘Return to the front line.’”

Doubt remains over whether antibody tests will ever reach the 90-95 per cent mark. Professor Koopmans told the Guardian that she “wouldn’t be surprised if you end up around 50-60 per cent” sensitivity. Professor Tim Spector, head of the twin research and genetic epidemiology department at King’s College London, says that the government may use a “first version” test that’s “not very good” while work to improve testing continues. But even a test that’s only 50 per cent accurate may be useful if combined with data about people’s symptoms, such as that collected via Professor Spector’s Covid Symptom Tracker app, which is currently being used by nearly two million people in the UK.

“We could wait another six months for a really good antibody test, but if there’s one out there that’s 50 per cent sensitive, and if you put that in an algorithm with all these other symptoms, you can make it 90 per cent, why wouldn’t you do that? It’s cheap.”

Ultimately, deciding when to roll out an antibody home test will involve balancing risks, says Dr Preston. “You’re trying to balance this with the need for us to feel we’re moving forward, that we have some clear way of exiting the lockdown, and knowing what the potential risk of flare-ups might be with whatever exit strategy we have.” 

Officials will come under mounting pressure to make progress in the coming weeks – but it’s the risk of flare-ups that will give them pause for thought. “If we’re going to go through this for two months,” says Dr Preston, “the last thing we want to do is come out off the back of false data, and have a week of freedom for another two months lockdown, with goodness knows how many more infections.”

Samuel Horti is a freelance journalist who covers politics, culture and gaming
Free trial CSS