© Crown copyright 2021
This publication is licensed under the terms of the Open Government Licence v3.0 except where otherwise stated. To view this licence, visit nationalarchives.gov.uk/doc/open-government-licence/version/3 or write to the Information Policy Team, The National Archives, Kew, London TW9 4DU, or email: firstname.lastname@example.org.
Where we have identified any third party copyright information you will need to obtain permission from the copyright holders concerned.
This publication is available at https://www.gov.uk/government/publications/lateral-flow-device-performance-data/key-points-summary-asymptomatic-testing-for-sars-cov-2-using-antigen-detecting-lateral-flow-devices-evidence-from-performance-data-october-2020-to-m
Around 1 in 3 people never develop symptoms despite having coronavirus (COVID-19). Testing for those without symptoms supports finding and identifying infectious cases to reduce community transmission.
Infectiousness can be broadly related to viral load (how much virus can be found in a test sample). Those with a high viral load are most likely to be infectious, therefore it is particularly important to identify those with high viral loads. These people are most likely to be in the infectious and transmissible stage of the virus.
Asymptomatic testing (testing for those without symptoms) allows us to find more people with transmissible virus. Because they have no symptoms these people will probably not know they are infectious. By locating these cases these people can be notified, self-isolate and break the chain of transmission.
Lateral flow antigen testing devices (LFDs), sometimes referred to as ‘rapid tests’, are well suited to mass testing for those without symptoms as they require minimal infrastructure and less training than other tests. Validation experiments show these tests have suitable sensitivity (how good the test is at detecting true positive cases) and specificity (how good the test is at not incorrectly giving a positive) for use within the community setting across the range of coronavirus prevalence (measure of the number of positives in the population at a given time) levels likely to exist.
To assess the suitability of coronavirus tests, laboratory-based validation studies and real-world field studies are run across different settings. This helps to better understand how to operationalise testing. Polymerase chain reaction (PCR) tests (tests that need to go to a lab for processing) can be used to measure viral loads and, in the real-world evaluations detailed in this paper (asymptomatic testing: lateral flow antigen testing devices), the viral load was measured and used to provide an indicative link between the LFD result and infectiousness, and to allow the results of different evaluations to be compared with each other.
A summary of the findings in the paper is as follows:
Initial LFD evaluation
The original rigorous clinical evaluations of lateral flow tests by Public Health England (PHE) and Oxford University, published in November 2020, showed that the tests perform best when levels of virus are at their highest, and are accurate and sensitive enough to be used in the community.
In PHE field tests 1 and 2, individuals swabbed themselves and the samples were analysed by an experienced professional to determine the LFD result (whether the test showed as positive, negative or void). These evaluations showed that the sensitivity of the test (how well the test did at detecting positive cases) increased as the viral load and by extension the likely infectiousness of the individual increased:
for people with high viral loads (over 1,000,000 RNA copies per millilitre), 96% (66 of 69) were detected
for individuals with low viral loads (10,000 to 1,000,000 RNA copies per millilitre), 92% (180 of 195) were detected
for those with minimal viral loads (less than 10,000 RNA copies per millilitre, 43% (55 of 127) were detected
Applying the performance of LFD to data obtained from thousands of historic users of Test and Trace, it was also calculated that LFD tests would detect the majority (83% to 89%) of known infected people who themselves had at least one PCR-positive contact. This suggests LFDs would pick up 83% to 89% of infectious individuals.
Further LFD evaluations
A small-scale, university-based evaluation also showed an increasing likelihood of a positive result for the LFD tests with higher ‘infectious’ viral load levels. Similar results were also found in data from the Liverpool mass testing pilot: the proportion of people identified as positive by the LFD increased with viral load. A quality control reappraisal of these results was done where photographs of the results were taken and reassessed by experts and the sensitivity of the tests was shown to be higher as a result of the reappraisal (full results of the reappraisal are shown in table 3 in the paper).
A further 5 evaluations are outlined in table 2: 2 evaluated LFD performance when less experienced but trained users processed the test and a further 3 when the participant (an untrained, inexperienced user) processed the test themselves (full self-use). The sensitivity results across differing viral loads are outlined in table 4. For the Innova test (which uses a throat and nasal swab) the sensitivity for those with high viral loads was 4% higher when administered by the experienced users in field test 1 (85%) when compared to the inexperienced users in field test 3 for inexperienced users (81%). Given the ease and convenience of testing at home this slight decrease is deemed acceptable. For Orient Gene LFDs (which uses a nasal only swab) the sensitivity was the same whether administered by experienced or inexperienced users.
As part of work to improve the tolerability of testing and overall access to testing, different swabbing techniques were assessed by the Department of Health and Social Care as part of the evaluations using Orient Gene LFDs. This evaluation compared throat-and-nose swabbing with nasal-only swabbing. Nasal-only swabbing still had an 88% sensitivity at high viral load when administered by an experienced user and the same sensitivity (88%) when administered by an inexperienced or self-user. Although there was a slight reduction in test sensitivity with nasal-only swabbing compared with the throat and nose swab test, due to the likely improved tolerability, especially for regular testing, this is an acceptable compromise.
Analyses of real-world positivity from Innova LFDs delivered in the community suggest the Innova LFD outperforms initial estimates of specificity, instead estimating a real-world specificity for Innova greater than 99.97%: in every 10,000 people tested we would expect 3 people to receive a ‘false positive’ LFD result. It is possible that an equivalent improvement will be found for Orient Gene once larger sample sizes have been collected. This is because when evaluating the performance of a lateral flow test, we compare the LFD results to the PCR result. However, we know that sometimes the PCR does return incorrect results: while PCR tests are very unlikely to give a positive result when they should not (false positive) they will miss some true positives (for example, due to poor sampling technique).
If we account for the fact that sometimes PCR tests miss true positives then the specificity of the LFD is higher, and by collecting more data from real-world use we are better able to estimate the likely specificity of the LFD.
|Viral load||How much virus can be found in a test sample, can be used to indicate how infectious someone may be.|
|Sensitivity||The proportion of people infected with coronavirus who receive a positive result.|
|Specificity||The proportion of people who receive a positive test result who are genuinely infected with coronavirus.|
|Prevalence||The measure of the number of positives in the population at a given time. That is, we would say there is low prevalence when the number of people with the virus is low.|