Testing cannot save us from Covid-19

The media has been largely responsible in shifting the public debate about the British Government’s handling of Covid-19 to an obsession with the gross number of people being tested. They have ignored the more important issue that the current PCR (Polymerase Chain Reaction) test has worryingly low levels of sensitivity, thus telling many people that they do not have Covid-19 when they do. The real issue is improving the type of tests available, not ramping up the number of current tests. The PCR test cannot be relied upon for evidence, either to end the lockdown, or as a reliable first step in contact tracing to limit the successive waves of infection.

In the words of Dr Rochelle Walensky, a Vice-Chair at the Infectious Diseases Society of America, “We believe the [RT-PCR] test is pretty good if it gives you a positive result. We think we can trust that to say the person really has the disease. But if it gives you a negative result, the reported sensitivity of that is only about 70 per cent.” By comparison, one of the WHO recommended tuberculosis tests has sensitivity of 88 per cent and the most accurate home pregnancy test in the UK has sensitivity of 95 per cent.

So serious is the issue that a special article was rushed into print in the prestigious medical journal, the Mayo Clinic Proceedings, which sheds light on the inherent problem of relying on tests with a significant number of false-negative results for making both clinical and public health decisions. The Mayo Clinic paper is far from isolated. Similar findings have been reported in a wide variety of medical journals, including an article in Radiology that suggested that PCR testing alone is necessary but not sufficient for clinical diagnosis.

From a public health perspective, false-negative results (a test result which tells someone that they don ’ t have the disease when they actually do) are even more problematic than false-positives (telling someone they do have the disease, when actually they don ’ t). Someone who thinks they don ’ t have the disease, when in fact they do, becomes an unwitting spreader.

The full magnitude of the problem is hard to discern because so far test performance has not been reported consistently or clearly. As the Mayo Clinic article mentions, early data from China already suggested poor sensitivity of initial RT-PCR tests. Sensitivity and specificity are the two key statistical measures for the performance of medical tests. Sensitivity refers to the percentage of people who have a disease and who test positive for it.

In the case of Covid-19, the presumed sensitivity of about 70 per cent still means that 30 per cent of people with the disease would incorrectly test negative, falsely suggesting they do not have the disease. This is clearly an unacceptable figure. To put this in perspective, if 50 per cent of the 9.3 million population of the London Metropolitan area had Covid-19 and were all tested, you would see 1.4 million false-negative results. The infection figures could, of course, be far worse. The mayor of Madrid predicted that 80 per cent of its population would become infected.

These figures matter a great deal, particularly as widespread testing has been suggested as a means of getting quarantined NHS staff back to work by ruling out Covid-19. If we applied the same test sensitivity and infection rate to the 470,000 doctors, nurses and midwives of the NHS, that would mean 70,500 false negatives. That would be an unimaginable number of frontline staff being greenlit to return to work in the mistaken belief that they do not have Covid-19.

This is what makes the synthetic media debate so deplorable. The media ’ s focus on testing led to Matt Hancock’s arbitrary target of 100,000 tests per day. In turn, this has become a crude proxy for those obsessed with skewering the Government with the charge of incompetence. This has driven unfavourable comparisons with Germany and some of the Asian countries that have managed higher levels of testing. In reality, their infection rates are more likely to be related to population density than testing and their PCR testing suffers from the same problems as ours.

When I queried Public Health England as to their understanding of the PCR test’s sensitivity and specificity and their confidence in it, they ignored my specific questions and referred me to a paper from January 2020. That paper, although often cited, is long since out of date and, as the author’s note indicates, was produced in the absence of physical sources of viral genomic nucleic acid. It seems incomprehensible that PHE would not have updated their understanding of the test’s sensitivity and specificity in the light of new data, let alone that they would now be trying to obfuscate the issue.

It is unclear exactly why the PCR test has such low levels of sensitivity, but Rochelle Walensky suggests that the reasons might include “a sampling problem if the clinician doesn’t stick the test swab far enough into someone’s nose… Second, as patients get sicker, the virus migrates from the nasopharynx to the oropharynx and into the respiratory tract. We don’t know, when it migrates down, whether it has left the nasopharynx.” The sampling problem is very likely to increase with a widespread ramping up of testing, where minimally trained staff working in drive-in centres are tasked with taking accurate samples.

The question remains, from a public health perspective, how much better the PCR test really is than the Government’s advice to self-isolate if you are symptomatic. As one senior epidemiologist, who wished to remain anonymous, put it to me: “All one can say is that the tests are not reliable enough.” At best, the PCR test remains part of the arsenal for dealing with post-lockdown flare ups. Some, but by no means all, cases will be picked up and contact tracing will be possible. But this is at best a partial solution, not a release from the terror of the second wave. For that we will have to wait until industry develops both antibody and antigen tests with far greater sensitivity and specificity.