The rapid test kits most widely used in UK universities, schools, and care homes detect just 48.89% of covid-19 infections in asymptomatic people when compared with a polymerase chain reaction (PCR) test, real world data from the Liverpool pilot have shown. The Innova Lateral Flow SARS-CoV-2 antigen test failed to detect three in 10 cases with the highest viral loads, in preliminary data released from the field evaluation of testing in asymptomatic people.

The University of Liverpool document states that the Liverpool Health Protection Board decided as a result of the findings “to pause plans to use Innova Lateral Flow SARS-CoV-2 Antigen tests (LFT) in test-to-enable visitor access to care home settings due to the accuracy statistics”. Liverpool City Council resumed its pilot scheme in care homes on 3 December but only after requiring visitors to have two lateral flow tests and a confirmatory PCR test.

“Clearly, there is a risk of giving false reassurance to people who get a negative result. You also have to question whether mass screening using a test that performs so poorly is the best use of our limited resources.

A false positive result occurred in two of 2981 PCR negative people—a specificity of 99.93% (99.76% to 99.99%). But lateral flow tests missed 23 of the 45 PCR positive participants, giving a sensitivity of 48.89% (33.70% to 64.23%).

The lateral flow tests missed a third of people with high viral load who had Ct scores (cycle threshold, a measure of virus) below 25—but seven of 15 cases with Ct scores of 20-25 were missed, giving a sensitivity of only 47% in this group (21% to 73%). In people with Ct scores of 25-30, six of seven cases were missed—a sensitivity of 14% (0.4% to 58%). Deeks remarked, “We are seeing that it is really very poor at detecting covid in asymptomatic cases, even those with a higher viral load.”

The government has spent more than £700m (€767m; $933m) buying the Innova kits, which are the lynchpin of its mass testing strategy. Over a million kits have been sent to care homes, and targeted testing is now being offered to secondary school children in parts of London, Kent, and Essex. Rapid lateral flow tests are also being used by most universities to test students before they return home to their families for Christmas, and only Cambridge offers PCR testing.

Covid-19: Tests on students are highly inaccurate, early findings show

The use of the lateral flow test for picking up asymptomatic cases is not one of the manufacturer’s intended uses. Instructions supplied with the test state that it is “intended for the qualitative detection of nucleocapsid antigens from SARS-CoV-2 in human nasal swabs or throat swabs from individuals who are suspected of covid-19 by their healthcare provider within the first five days of the onset of symptoms.” (Ref)

The UK government’s plans for community testing for covid-19 received a further blow this week when early results from students testing at the University of Birmingham and universities in Scotland showed that tests had a sensitivity of just 3% and that 58% of positive test results were false.

Jon Deeks, professor of biostatistics at Birmingham and leader of the Cochrane Collaboration’s covid-19 test evaluation activities, explained the results. “We found two positives in 7189 students, which scales up to 30 per 100 000 and was shocking in itself, as Birmingham has a rate of 250 cases per 100 000,” he said. “These results are especially worrying for schools: the government should not be proceeding with plans for schools testing until they have a proper evaluation of the test.”

Using polymerase chain reaction (PCR) testing, the team retested 10% of the samples that had been negative with the Innova test and found six false negative cases, raising the rate to 60 per 100 000.

Deeks said on Twitter, “We thus estimate that we found 2 cases and will have missed 60 (because we only double tested 10%). We estimate the true prevalence to be 0.86% (95% [confidence interval] 0.40% to 1.86%) which is much more credible than the 0.03% test positive rate. Our estimate of overall sensitivity is 3.2%.”