What are lateral flow tests, and why do the numbers of false positives keep changing?
PCR tests have been the main test used for Covid-19 during the early part of the pandemic. They can detect Covid-19 very well, but need to be processed in a laboratory. Therefore, they have mainly been used in symptomatic people or people who have been exposed to a known case of Covid-19 and are already self isolating whilst they wait for the result.
Lateral flow devices are a newer type of test. They are easy to do and give fast results on the spot. They can therefore be used for mass testing of people with no symptoms. They are not always as sensitive as PCR tests at picking up potential cases of Covid-19, and so may give false reassurance to infected people. There are also times when they can give incorrect positive results, and so if they are being very widely used, particularly in people with no symptoms, they may make some people think they are infected when they’re not.
Because of this, there has been debate around whether lateral flow tests should be used for mass screening, and whether a positive lateral flow test in this context should prompt another test by PCR to confirm the result.
Some experts have said that large numbers of people are being asked to unnecessarily self-isolate, based on calculations using estimates for the accuracy of the tests and the background levels of Covid-19, whilst others worry that the tests may miss a lot of cases and provide false reassurance if used inappropriately. The Department of Health and Social Care (DHSC) has defended its use of the tests.
What are LFDs?
Lateral flow tests, or lateral flow devices (LFDs), for Covid-19 are rapid tests that do not need to be processed by a laboratory. The tests work in a similar way to a pregnancy test, but measure different substances in the sample. We have previously written about lateral flow tests before.
To do a lateral flow test, a sample is taken from the nose or mouth using a swab. The swab is put into a specially formulated solution to transfer any viral material that is present. The solution is then dropped onto the lateral flow device, where it is drawn along an absorbent strip. If viral proteins (antigens) specific to SARS-CoV-2 (the virus that causes Covid-19) are present, they react with immune proteins on the strip called antibodies. These are tagged with a marker so that if the viral proteins are present, the test shows two stripes (one control stripe plus another which appears with a positive result).
It is important to remember that lateral flow and PCR tests are used in slightly different ways. PCR tests are generally recommended if you are symptomatic or have had close contact with somebody who has tested positive. For asymptomatic testing (for example regular testing at home), a lateral flow test is suggested.
At the time of writing, if you have a positive or inconclusive lateral flow test result, you should request a confirmatory PCR test. This advice has changed recently, due to changes in background levels of Covid-19 in the community, and the need to get further information about variants that are spreading.
How accurate are they and why do the figures keep changing?
- how successful it is in correctly detecting people with a particular condition (how many “true positives” it detects, or how “sensitive” it is),
- how successful it is in correctly detecting people without the condition (how many “true negatives” it detects, or the “specificity”).
This is important because a test may be very sensitive (pick up a lot of actual cases of a disease), but if it is not very specific it will also incorrectly pick up a lot of people who don’t have the disease. This can cause distress, repeated testing, and in the case of Covid-19, unnecessary isolating and contacting tracing.
Similarly, a test that cannot reliably identify cases could mean that people with Covid-19 are wrongly told that they have tested negative. They therefore may not self-isolate and could spread infection.
Research from Public Health England (PHE) for example, shows that when lab scientists performed a lateral flow test they picked up 79% of cases, versus 58% of cases if the swab was performed by self-trained members of the public.
There is also the question of how to test the accuracy of lateral flow devices. Most studies compare how good lateral flow tests are at picking up positive cases of Covid-19 against PCR testing. This is sometimes critiqued as an unfair comparison because although PCR tests are extremely good at picking up cases of Covid-19, some of the cases that they detect are no longer infectious—although there is debate around how many. These cases may still be useful to be aware of, in case of complications related to Covid-19, and for contact tracing and epidemiological purposes.
Lateral flow tests are used in asymptomatic people who would otherwise not be tested at all. Therefore they are being used to serve a slightly different role to PCR tests, assisting in the early detection of asymptomatic people. The aim is to spot infectious people, who would otherwise be missed.
How good are lateral flow tests at detecting cases of Covid-19?
Estimates of how sensitive lateral flow tests are have varied. PHE evaluations at Porton Down using community samples, for example, successfully detected real Covid-19 infections 78-94% of the time when the samples were compared with corresponding PCR results. The number of correctly detected infections was higher than this for groups of samples that had a lower cycle threshold, which is related to a higher concentration of virus. This is important because these may be the people who are most infectious. We have written more about this here .
Data from community testing in Liverpool, showed lateral flow tests correctly identified positive Covid-19 cases 40% of the time, and an analysis of asymptomatic community testing by the DHSC found that the figure was 50%.
In a review of 64 studies published in March, the average value for how well the tests were able to pick up positive Covid-19 cases was 69%. This varied greatly however. For example, if patients were asymptomatic the average was 58%, whereas if they were symptomatic it was 72%.
Lateral flow tests are often described as a “red light” tests rather than “green light tests”. Talking in the BMJ, professor of paediatrics at the University of Bristol Adam Finn, said: “If they come up positive that means you are potentially infectious to others and must self-isolate. They are not ‘green light’ tests: you cannot be sure that if the test is negative you are not infectious, and you must continue to take the usual precautions.” This is also in keeping with the authorisation issued from the Medicines and Healthcare products Regulatory Agency (MHRA).
The DHSC told Full Fact that lateral flow tests remain important because they allow early detection of Covid-19 positive individuals, when they may be most infectious. A spokesperson said: “With around one in three people not showing symptoms of Covid-19, regular, rapid testing is an essential tool to control the spread of the virus as restrictions ease by picking up cases that would not otherwise have been detected.”
What is the specificity of LFDs?
Initial information from PHE evaluations which were released in November, showed that lateral flow tests are 99.68% specific, which means they successfully identify people who are not infected 99.68% of the time.
In March, the DHSC released their data from 7,546 participants which showed that LFDs had a specificity of 99.72%—and perhaps as high as 100% when compared with PCR testing. In their data from asymptomatic NHS testing sites between November 2020 and January 2021, this was 93.7%. Data from the March Cochrane review showed an average specificity of 99.6%, and community testing pilots in Liverpool, showed a specificity of 99.9%.
What about false positives?
Through reported leaks and calculations by prominent statisticians on social media, there have been concerns about very high numbers of false positives from lateral flow tests during times when there are lower rates of Covid-19. These are theoretical calculations, which use assumptions about particular measurements to create their estimates. These paint scenarios that are possible, but only give one version of the picture, because the same calculations with different assumptions can produce very different results.
For example, using some of the government's estimates, with a background level of Covid-19 at 1 in 200 people, roughly half of the positive lateral flow tests might correctly find someone who has Covid-19—although this depends how specific the tests truly are. When the background level of Covid-19 is higher, say a level of 10 in 200 in the population, at least nine out of 10 positive lateral flow tests will relate to a true case of Covid-19.
The Department of Health told Full Fact: “Rapid testing detects cases quickly, meaning positive cases can isolate immediately, and figures show that for every 1,000 lateral flow tests carried out, there is fewer than one false positive result.”
It also said that at times of low prevalence it has advised confirmatory PCR tests to mitigate the risk of people being asked to self-isolate unnecessarily.