Mass population screening for SARS-CoV-2 and the problem of false positives. Why Liverpool shows we have an issue.
In your Editor’s Choice article of the 19th of November, "Covid-19: Screening without scrutiny, spending taxpayers’ billions" (1), you highlight a range of issues associated with the UK’s proposed mass testing regime. Despite all the issues discussed in this article, not least the ethical issues such screening efforts raise (2), mass testing is now planned for the UK population.
The idea behind mass population testing for SARS-CoV-2 is that there are a substantial number of asymptomatic or pre-symptomatic individuals within the population who are silently (and unknowingly) spreading the coronavirus infection. By identifying and quarantining such individuals we can more effectively limit the spread of the disease through the population by breaking potential chains of infection.
But as stated in this article “The lateral flow test being used is of doubtful value, with a high false negative rate. Although the false positive rate is small it is still a problem in a low prevalence setting”. The recent mass screening trial in Liverpool demonstrated that only a small proportion of individuals (648 out 92,683 or 0.7%) with no COVID-19 symptoms were positive for the presence of SARS-CoV-2 (3). So, if the Liverpool experience is representative of the UK as a whole, then we have precisely the conditions where false positives become a problem i.e., a low prevalence of infection in a largely negative population.
Because the infection prevalence is low, we can estimate the impact of putative false positive rates on the Liverpool screening results by simply multiplying them by the negative proportion of the population screened (this gives a slight under-estimate of the true number of false positives). For example, a false positive rate of 0.1% (1 in 1000 tests of true negative samples returns a positive result) would mean that about 92 (92,035 negative individuals x 0.1%) of the positive results are false positives or about 14%. We gain another 92 false positives for each 0.1% increase in the false positive rate up to 0.7%, when all the samples are actually false positives.
Clinically we cannot tell the difference between a pre-symptomatic individual, an asymptomatic “carrier” and a false positive result from a healthy individual, because by definition none of these individuals have symptoms. We are wholly reliant on the fidelity of our screening approach to separate the infected individuals from uninfected ones.
Recent data from Wuhan suggest that the risk of asymptomatic individuals may actually be low with little evidence of onward viral transmission from such individuals to their close contacts (4). However, the impacts on asymptomatic individuals and their immediate contacts on receipt of a positive test result are in themselves significant, requiring self-isolation and possibly leading to broader societal impacts like closures of schools and businesses. Screening tens of millions of individuals means that even a low false positive rate translates to a large number of uninfected individuals being potentially told they have “asymptotic COVID-19”. So, rolling out such mass testing in the absence of a thorough real-world understanding of performance is fraught with danger as the testing itself may end up having a greater impact on lives and society than the disease it is designed to protect us from.
As you highlight in this article, the ethical and moral problems of such mass screening are profound more so if, as has been suggested, ongoing repeat re-screening of individuals may become a requirement to live a normal life (5). From a purely technical perspective, repeat re-screening may reduce missing infected individuals because of false negatives but it also increases the likelihood of an individual receiving a false positive result…in fact do enough repeat tests and eventually one of them will be positive.
Finally, we also need to remember that if we do not account for false positives in such screening efforts then rather than an ultimately self-limiting viral pandemic, we will have a never ending “screenidemic” driven by testing artefacts.
1. BMJ 2020;371:m4487: https://www.bmj.com/content/371/bmj.m4487
2. BMJ 2020;371:m4438: https://www.bmj.com/content/371/bmj.m4438
4. Cao, S., Gan, Y., Wang, C. et al. Post-lockdown SARS-CoV-2 nucleic acid screening in nearly ten million residents of Wuhan, China. Nat Commun 11, 5917 (2020). https://doi.org/10.1038/s41467-020-19802-w
Competing interests: No competing interests