Why mass testing for coronavirus isn’t an easy solution

The Covid-19 test is producing too many false negatives. 

Sign Up

Get the New Statesman's Morning Call email.

Change.org petition to test frontline NHS staff for Covid-19 as a priority has received over a million signatures this week. “We need some common sense here,” writes the petitioner, a junior doctor. Given the comments from signatories, you would think common sense was as scarce in the Department of Health and Social Care as a coronavirus vaccine. If only it was so simple. 

There are high hopes for a new 15-minute “fingerprick” test for Covid-19, which will soon be made available in the UK. By testing for antibodies to the virus, it will hopefully tell you whether you have been infected. But it will not show whether patients are no longer infectious, and it may take three days or more for an individual to mount an antibody response that would show up in the test. Matt Hancock, the Health Secretary, has said that 25,000 tests will be carried out every day. But just as the workhorse of the Battle of Britain was the Hurricane aircraft, and not the flashier Spitfire, the standard genetic test already in use will do most of the heavy lifting.

In the language of researchers, the genetic test for the coronavirus is highly “specific” but not very “sensitive”. This means it hardly ever confuses other viruses for Covid-19, so false positives are rare, but false negatives are quite common. 

A study of 1,000 patients in Wuhan province, China, found lung damage characteristic of Covid-19 infection was visible on chest CT scans in 15-25 per cent of those who registered negative in their first genetic test. As “fit to work” assessments go, this is a failure rate that would make Atos blush. Compare it with home pregnancy tests — which commonly advertise an error rate of only one per cent. Even then, women are advised to repeat the test several days later to obtain a reliable result. 

False negatives for pregnancy are mostly due to variance in urine samples, rather than any defects in the testing kit. Likewise with testing for the coronavirus with the complication that good samples are harder to obtain. The genetic test for Covid-19 works by detecting viral genetic material in a sample of sputum or mucus. If it doesn’t identify the virus, that could be because you don’t have it or perhaps there isn’t enough of a viral load in your sample. A swab from the mouth or throat may not be enough. Sputum from the lungs is not easily extracted from someone with a dry cough, let alone from someone without symptoms. The best samples are obtained from the lungs using invasive procedures. But while it’s possible to perform such a procedure on selected unconscious patients in intensive care, this test cannot easily be extended to 150,000 doctors and 320,000 nurses and midwives across the UK.

To mitigate the risk of sending contagious NHS workers back to the wards, doctors say they would want to conduct two tests at least 24 hours apart. The testing process takes 48 hours (sometimes longer), as the sample has to be sent to a lab, meaning a total waiting period of at least five days. While modest improvements count, this is not vastly better than the recommended seven days of isolation for suspected cases.

False negatives aside, the tally of positive diagnoses will be useful for demonstrating how the struggle against coronavirus is progressing. But if the primary purpose of testing is to screen NHS staff, we may be disappointed. 

Boris Johnson complained a week ago that countries were taking measures for which there was little scientific backing because they wish to be seen to be doing something. But now the UK public need some good news, so the NHS will “massively ramp up” testing. If this is a war, then perhaps it is worth being seen to attack the enemy even when the damage to him is debatable.

Nicholas Fearn is the author of Zeno and the Tortoise: How to think like a philosopher 

Free trial CSS