Coronavirus: why antibody tests may miss a quarter of cases

A test for Covid-19 antibodies that health officials have described as “100% accurate” may be giving incorrect results for as many as a quarter of people who have had the virus, an independent analysis suggests.

The UK government ordered ten million of the tests, made by Swiss company Roche, last month after Public Health England (PHE) vouched for their accuracy.

“However, an independent scientific evaluation of the PHE tests has raised concerns about the methods used to assess them and how well they are likely to work,” The Times reports.

Why are antibody tests important?

As far back as March, Boris Johnson was describing the prospect of a reliable test for Covid-19 antibodies as a game changer.

“The great thing about having a test to see whether you’ve had it,” he said, “is suddenly a green light goes on above your head and you can go back to work safe and confident in the knowledge that you are most unlikely to get it again”.

However, scientists have been more cautious. “If you test positive for antibodies, it’s likely you have a degree of immunity,” Dr Ron Daniels of University Hospitals Birmingham NHS Foundation Trust told The Telegraph.

“We’re not sure for how long, and how much, so you shouldn’t stop [social] distancing, but best guess it is likely to be partially protective for at least a few months.”

What were the findings of the PHE research?

“It was widely reported in May that a new antibody test - which can detect if someone has had Covid-19 - is ‘100% accurate’,” says fact-checking website Full Fact. “However, this is not quite correct.”

After carrying out an evaluation in government science labs at Porton Down, PHE concluded that the Roche tests had “100% specificity, which means it accurately identified all negative samples as negative”, says Sky News.

This is important, as it means no one will be told they have immunity when they don’t.

Where the test was less successful was in its “sensitivity” - which refers to detecting positive cases. “It missed 16% of the samples of people who had Covid-19,” Sky News reports.

And the conclusions of the new analysis?

A review by the Test Evaluation Research Group at the University of Birmingham found that the small sample size used in the Porton Down validation tests has resulted in a significant margin of error. “This could mean that the test may miss between 8% and 25% of actual past Covid infections,” says The Times.

More encouragingly, the review did not find any evidence of false positives.

–––––���–––––––––––––––––––––––––For a round-up of the most important stories from around the world - and a concise, refreshing and balanced take on the week’s news agenda - try The Week magazine. Start your trial subscription today –––––––––––––––––––––––––––––––

What about other types of antibody test?

While imperfect, the test made by Roche offers higher levels of accuracy than its predecessors.

“Previous tests have not been as accurate, meaning they detect antibodies against other coronaviruses such as the common cold,” says Sky News.

In March, the UK government placed orders for 17.5 million antibody tests, none of which turned out to be effective. An evaluation of their accuracy found that all were giving significant false negatives and false positives.

In April, Downing Street admitted that “no test so far has been proved to be good enough to be used”.

Why are they so hard to make?

Antibody tests “detect the body’s immune response to the infection caused by the virus rather than detecting the virus itself”, explains the US Food and Drug Administration.

Since people’s bodies react in different ways, depending on the health of their immune system and the seriousness of their infection - and since different viruses can lead to the production of similar antibodies - developing a universal test is tricky.

“To be considered accurate, devices must be able to distinguish the presence in someone’s blood of antibodies specific to this coronavirus, and also be able to identify antibodies in people who have had relatively mild strains of the disease,” says the Financial Times.

Even tests that look promising in early research can turn out to be wildly inaccurate when put into widespread use, according to Professor Paul Hunter of the University of East Anglia.

“When you don’t have enough samples you can be misled - 100 tests might look good, but after 20,000 they might not,” he says.

–––––––––––––––––––––––––––––––For a round-up of the most important stories from around the world - and a concise, refreshing and balanced take on the week’s news agenda - try The Week magazine. Start your trial subscription today –––––––––––––––––––––––––––––––