COVID-19 saliva tests not as accurate as widespread reporting suggests

Matt Woodley

2/07/2020 5:00:43 PM

While useful, evidence shows a relative lack of sensitivity means there is still a high chance of returning a false-negative.

Man holding bag containing sputum sample.
While less-invasive than PCR tests, the accuracy of saliva testing appears to have been misinterpreted.

With Victoria in the midst of a coronavirus outbreak that has seen it record nearly 440 new cases in the past two weeks alone, the recent arrival of new saliva tests has been welcomed by public health officials and the media alike.
But while the less-invasive tests are a painless alternative to ‘gold standard’ polymerase chain reaction (PCR) tests, there appears to have been some confusion over their sensitivity.
Numerous mainstream media reports have labelled the saliva tests as ‘87% accurate’, based on a report by the Peter Doherty Institute, which developed them.
However, what that report makes clear is that the accuracy rate is comparative to PCR testing only, and does not indicate that it can identify all coronavirus cases with 87% sensitivity.
While PCR tests are incredibly accurate in a laboratory setting, recent research suggests sensitivity in a clinical setting is affected by timing and estimated to be only around 70%.
As such, Melbourne-based GP Dr Vyom Sharma told newsGP that it is ‘very likely’ the sensitivity of the test ‘isn’t anywhere near’ 87%.
‘[The reporting] creates the impression that the saliva test is very good, which might lower the threshold for people to refuse the deep nasal or nasopharyngeal swab,’ he said
‘It’s a very useful test ... but we don’t want a “drift” of people who’d otherwise go for the traditional swab, down to this second-line test.
‘The purpose of the saliva test is to pick up cases that would have been missed by people who are deterred by a traditional nasopharyngeal or deep nasal swab. That’s it.’
Ear nose and throat surgeon Dr Stephen Kleid told newsGP the reduced sensitivity rate of PCR tests in a clinical setting is most likely because the samples have not been collected properly.
‘There might be several reasons for that. One is the patient has just blown the nose and there’s no mucus there at the time or, secondly, the swab wasn’t put in the right place in the first place,’ he said.
‘It’s not easy to get a swab all the way back 7 cm in a patient because they’ll move, it hurts them. Also you have to go horizontally back, which is not how [many] people take swabs.
‘If you stick a swab into somebody’s nose, once you get passed more than 1–2 cm the patient’s going to jump and squint and pull away. It’s very hard for them to stay still.’
Dr Kleid also said the nature of a pandemic, which requires mass testing in unorthodox settings such as car parks, means there is an even higher chance for human error.
‘Who’s taking the swabs and how cooperative the patient is makes a big difference as to how the swab is going to be. Most people won’t let you stick something 4 cm into a nose, and certainly not 7 cm in, unless you put local anaesthetic into their first,’ he said.
‘It seems like there’s a lot of room for human error in this testing method and that’s probably one of the main reasons why we might get so many false-negatives.
‘The problem is, the Government’s relying on a test of uncertain accuracy. We just have to accept the limitations that the swabs aren’t accurate.’
The high chance of human error and false-negatives recently led Associate Professor Mark Morgan to call for more nuanced public health messaging related to coronavirus tests and their results.
‘There seems to be an over-reliance on the value of a negative test,’ he recently told newsGP.
‘I have people contacting me in general practice who are developing flu-like symptoms who are quite convinced their negative COVID-19 test means that they don’t need to self-isolate.
‘In an attempt to encourage more people to get tested, we seem to have over simplified the messaging. Negative tests are only an indicator that makes it less likely that you have COVID-19.’

A study published last month in the New England Journal of Medicine used Bayes’ theorem, which incorporates information about both the person and the accuracy of the test, to calculate how less likely a person is to contract COVID-19 following a negative PCR test.
It estimated someone with early symptoms linked to COVID-19, who had been in a ‘hotspot area’ and was recently exposure to coronavirus, had a 50% probability of being infected before the test.
A negative result from a test with a sensitivity of 95%, such as PCR in lab conditions, would mean that person only has a 5% post-test chance of being infected. However, when sensitivity is reduced to 70%, which more accurately reflects clinical accuracy rates, that probability jumps to essentially one in four cases.
There is also limited information regarding the accuracy of these tests in asymptomatic carriers, which some studies suggest could be as high as 42% of cases.
Dr Sharma said testing is thus mainly useful for prompting people to stop mingling.
‘The concern with recent jumps in community transmission in Victoria is that we’ve had undetected cases circulating for some time now,’ he said.
‘If we merely get people who test positive to isolate, that will not be enough, because we aren’t finding every positive case. This is why physical distancing is absolutely essential.
‘The test doesn’t stop spread. Physical distance does.’
Log in below to join the conversation.

coronavirus COVID-19 nasopharyngeal swabs PCR testing polymerase chain reaction saliva testing

newsGP weekly poll As an international medical graduate, what was your primary reason for wanting to practise in Australia?

newsGP weekly poll As an international medical graduate, what was your primary reason for wanting to practise in Australia?



Login to comment

Dr Horst Paul Herb   4/07/2020 12:41:38 AM

Wyllie et all showed that in real world sampling saliva can actually be more sensitive than the unpopular nasopharyngeal swabbing:
Nasopharyngeal swabbing quality varies highly depending on operator skill and patient collaboration. For mass screenings, salivary samples may indeed yield less false negative results

I agree that the population has to be divested of the notion that a negative PCR test result equals to not being infectious - and likewise has to understand that a positive PCR result does not equal to being infectious either. The current media brouhaha around the Victorian post quarantine management is simply unscientific populism.