January 21, 2021

expert reaction to a preprint looking at viral load and lateral flow devices

A preprint, an unpublished non-peer reviewed study, looks at viral load, how infectious an individual is, and detection by lateral flow devices.

Prof Jon Deeks, Professor of Biostatistics and head of the Biostatistics, Evidence Synthesis and Test Evaluation Research Group, University of Birmingham, said:

“This paper marks progress in investigating and characterising which individuals with Covid are more likely to infect others. There have been previous claims that Covid cases can be grouped into those who are infectious and those who are not using particular values of viral load from a PCR test, or the positive or negative results of a lateral flow test. These data make it clear that infectiousness is not that simple, with a continuum of increasing risk with increasing viral load, which is altered by other factors.  Using the track-and-trace data has provided a useful resource, but it must be noted that the individuals will have been symptomatic when tested and therefore most likely close to their peak viral loads, and that as contact tracing is severely incomplete there will be many contacts and secondary cases missed.  This will affect the generalisability of these findings, particularly to asymptomatic mass testing.

“A key part of the paper estimates the detection rates of four different lateral flow tests for identifying infectious cases. The performance data for these four tests on which these predictions are based is not clear, but seems likely to be based on laboratory studies used to establish limits of detection, possibly of spiked rather than real samples.  Thus the predicted sensitivities of lateral flow tests should be considered as “best case estimates”, as they do not account for the 10-20% of swab failures, poorer performance of these tests when used in the real world outside of laboratories, and the different distribution of viral loads in different asymptomatic groups.  Wisely the authors conclude that the validity of these approaches needs to be evaluated in real world settings – a request that many of us have been making since Moonshot proposals were first announced. Until then, whilst the study provides strong evidence that lateral flow tests can be used as a “red light”, as those detected by lateral flow tests are most likely to be infectious, the safety of using these a negative value from these tests to “green light” other activities remains a concern and the emphasis from the MHRA, the Royal College of Pathologists, WHO and others that these tests should not be used as test-to-enable should remain.”

Dr Angela Raffle, Consultant in Public Health and Honorary Senior Lecturer University of Bristol Medical School; Population Health Sciences, said:

“This study, based on analysis of several hundred thousand UK test and trace records, demonstrates a relationship between the amount of viral material in a person’s PCR test sample, and the likelihood that their contacts will subsequently test PCR positive. This provides important evidence that ‘viral load’ is related to infectiousness. The study also found that being a household contact of a case is more likely to lead to transmission compared with household visiting, or outdoor contact, and that children are least likely to transmit infection. Predictions were also made about whether four different Lateral Flow tests would have picked up the most infectious cases. Whether this is based on Lateral Flow performance in ideal circumstances, or on performance when testing is conducted by untrained staff or by self-testing, is not entirely clear*, and the authors note that evaluation in real world settings is needed.”

*NB The DOI for the reference source for the Lateral Flow performance data used in the estimates (reference 15) does not seem to be functioning correctly at present.

Prof Sheila Bird, Formerly Programme Leader, MRC Biostatistics Unit, University of Cambridge, said:

“The title is misnomer as the data which are analysed are ct-values, not viral loads; and are available for only about half the RT-PCR positive cases referred to Test & Trace. The analysed cases had to have at least one contact and had to share contacts with no other case. Whereas 93% of index cases were symptomatic prior to testing PCR-positive, only 51% were symptomatic of 18,291 contacts who tested PCR positive within 2 to 7 days of their case’s PCR-test. The paper concentrates in the ct-values for its 145,973 index cases. I’m more interested in the distributions of ct-values for the symptomatic versus non-symptomatic 18,291 contacts who sought an early PCR-test and were found to be positive. The three distributions may differ markedly.”

Dr Alexander Edwards, Associate Professor in Biomedical Technology, Reading School of Pharmacy, University of Reading, said:

“This extensive analysis of Test and Trace data supports the expectation that at a population level, the more virus detected on a swab, the more likely transmission is. This is incredibly valuable but big policy and public health questions remain. What constitutes a tolerable risk if someone is to be released from self-isolation following a test that is positive but with low levels of virus? How accurately can a single swab of one individual be assessed to provide that individual a precise likelihood that they might transmit virus? In some parts of the world, stringent containment is used to eliminate any risk of onward transmission. At the moment in the UK the aim is to cut overall infection (and death) rates by reducing transmission as much as possible. Increased testing using faster and more portable lateral flow tests may be able to achieve this, but we still aren’t sure how well.

“Note this preprint uses prior accuracy data regarding lateral flow antigen tests from previous studies. The discussion of possible “performance” of lateral flow testing as a replacement/alternative for PCR testing is also a valuable calculation, but more complicated than it seems at first sight. We already knew that lateral flow antigen tests for COVID-19 do appear more accurate with patients that have more virus present. It follows that they may be better suited to spotting “spreaders” than identifying everyone infected. The question of how to manage false negative results remains- i.e. are people who receive a negative test result ‘safe’ or ‘safer’ than they were before they were tested?

“There remains plenty of complexity to the testing process, for example this study analyses symptomatic testing and it’s still not clear directly how these findings will translate into asymptomatic screening. Asymptomatic screening using lateral flow antigen tests has been widely discussed recently, remaining much needed but still hard to achieve. We already know that test accuracy can vary a great deal in different settings and for different people. This variation in accuracy was only briefly considered in the present preprint.

“It remains vital to keep working hard to make best use of all options including lateral flow tests, but we must continue to evaluate how well tests work in different settings, and to make sure new testing systems are checked to confirm they do reduce transmission. Needless to say, this would be more feasible with far lower rates of infection, as the workload and cost of testing would be far lower releasing resource to properly support and validate testing services.”

Prof Paul Hunter, Professor in Medicine, The Norwich School of Medicine, University of East Anglia, said:

“The preprint by Lee and colleagues from the University of Oxford essentially looks at data from the national SARS-CoV-2 PCR and contact tracing data from the United Kingdom and investigated the relationship between viral load and the probability of transmitting infection to contacts. This part of the study was, in my view, well done and showed convincingly that the more virus being shed by a case increased their chance of spreading the infection. Whilst not particularly surprising, it is always good to have what we believe proven or disproven by well-conducted studies. Indeed this observed relationship could in part explain the increased infectiousness of the English variant of concern which has already been noted tends to be associated with lower ct values in PCR and therefore higher viral loads.

“The authors went onto then use the results of this analysis to model the value of various lateral flow test devices (LFTD) in detecting cases during mass testing. The estimated that of the four LFTDs modelled each would detect between 80 and 90% of individuals who had infected a contact. In doing this modelling the authors used their own estimates of the sensitivity of LFTD.

“There has already been quite a lot of debate in the media about whether or not the high estimates of the sensitivity of LFTD produced by this group can be replicated in the real world. In the report of the Liverpool pilot study it says ‘The Innova SARS-CoV-2 antigen lateral flow device sensitivity was lower than expected (based on the preceding validation studies) at 40% but identified two thirds of cases with higher viral loads (~Ct<25)’. Other scientists found even poorer sensitivity with Birmingham suggesting just 3% (https://www.newscientist.com/article/2263746-test-caught-just-3-per-cent-of-students-with-covid-19-at-uk-university/). So the conclusions from this study about the sensitivity of lateral flow tests to detect the most infectious individuals depend on whether you accept the initial high sensitivity of LTFD produced by this group or the reports of their sensitivity being reported by people undertaking these tests in in-use settings. 

“I think as a minimum the authors should also run their models with data from the Liverpool pilot study, before any firm conclusions should be drawn.”

Preprint (not a published paper): ‘An observational study of SARS-CoV-2 infectivity by viral load and 2 demographic factors and the utility lateral flow devices to prevent 3 transmission’ by Lee at al.

Declared interests

Prof Jon Deeks: “Jon Deeks is Professor of Biostatistics at the University of Birmingham and fully funded by the University of Birmingham.  He leads the international Cochrane COVID-19 Diagnostic Test Accuracy Reviews team summarising the evidence of the accuracy of tests for Covid-19; he is a member of the Royal Statistical Society (RSS) Covid-19 taskforce steering group, and co-chair of the RSS Diagnostic Test Advisory Group; he is a consultant adviser to the WHO Essential Diagnostic List; and he receives payment from the BMJ as their Chief Statistical advisor.”

None others received.