And another thing…on false positives

Sometimes the full story can’t fit into a media article. A lot of words can be said during an interview with a journalist but sometimes it’s just really hard to convey the full story. So here, I add some of the missing content to a recent piece I was quoted in about false positive test results.

Testing for current, active, right now, SARS-CoV-2 infection relies on detection other viruses genetic material. This doesn’t have to be present in an intact virus, because the test can detect virus even if its is not longer able to replicate or when there’s likely to be too little to infect someone else.

The article entitled “Coronavirus false-positive test results rare, experts say β€” so how do they get COVID-19 wrong?” sought to explain where a false-positive laboratory test result comes from. The ABC has been providing extraordinarily comprehensive coverage of every aspect of the COVID-19 pandemic. This was a good question to ask since media worldwide have been talking about false positives and false negatives throughout, but often without much detail to help out the public. I was happy to chew the ear of an expert journalist.

We don’t throttle our PCRs

First off, this comment was a bit wrong so let’s delve more deeply.

“Scientists could adjust the test to find smaller amounts of the virus RNA but that would increase the chance of a positive result where the virus wasn’t actually present.”

Scientists and companies create and optimize these sorts of virus detection tests – called PCRs (lots to read about these here, here, and here) to be as sensitive as possible, right out of the gate.

There is no adjustment.

And with these tests – which don’t guarantee detection of infectious virus, just bits of their genetic material – that high level of sensitivity has been a problem sometimes. Finding traces of the virus in people who are well can really confuse matters.

How can a false positive occur?

The most common way they happen is usually as per this quote…

University of Queensland virologist Ian Mackay said false positives were rare and usually only caused by contamination of the test once it was back in the lab.

There was another part to this which didn’t make the final cut. That’s around the nature of the test. Sometimes, a test might cross-react with other, related viruses. This can be both a design issue or a lack of pre-release testing.

Sometimes one “test” is actually a combination of multiple tests. These are called multiplexed tests and they aim to detect two or more different parts of the virus genome, sometimes with other targets included ensuring the test is working. A multiplex test can behave a bit differently sometimes.

All of these “sometimes'” can be mitigated, overcome – or at least predetermined – by doing lots of pre-release testing of:

  • different sample types (throat swabs, nose swabs, saliva, sputum, blood, urine – you name it. Put that test through the hoops!),
  • different viruses (rhinoviruses, other coronaviruses, or other things you might find at the sampling sites)

…and looking for unexpected results.

This is the process of validating the test. Validation is a kind of pressure testing and it’s essential.

In Australia, validation required of tests for human use (see here and here for example). Australian labs are well aware of all this and constantly work hard to address all those needs to deliver the best results for the community (of which they are also a part!).

It’s not possible to “fix” a commercial assay if problems are found during its evaluation – kits are sold “as is” and are more of a closed box. One hopes they have been thoroughly validated. Sometimes they might not have.

Tests made by labs (“in-house tests”) can be fixed – or thrown out and redeisgned – if problems are seen during their development.

But it’s important to note that whether commercial or in-house or well-validated or not – sometimes a test may just throw up extremely rare problems not seen until used on thousands of tests.

It might be the sample not the test

Another cause – mentioned in the article – is the contamination of the sample at the point of collection. This scenario could happen if another infectious person’s sample got into a negative person’s sample tube in some way. Both samples could test positive by RT-PCR.

False negatives

This bit was confusing.

If a sample was taken too soon after someone was exposed to COVID-19, there may not be enough of the virus in the nose or throat to be detected.

This is actually about how a false negative result might arise. Not a false positive.

The same thing can happen at the other end of illness; a person is well and has so little virus left from their resolving illness, that it’s right at the limit of the test’s capability to detect.

Sick weeks ago, healthy now, still testing positive

Here we may see one sample positive, the next negative and so on in a random-looking pattern. We’ve all read a lot of these examples. They are happening in this pandemic because so many labs all over the world are concurrently hunting the same viral beast using RT-PCR – the scrutiny on results is more intense and rapidly communicated than ever before. We’re also testing samples from people who are basically well (looking to satisfy discharge criteria), and whom we wouldn’t normally test.

We’re asking and expecting an awful lot even from RT-PCR, the most sensitive virus test we’ve ever known.

A well contact example

As part of screening contacts, for example, a person may test positive even when they are currently well. From a medical point of view, this could look like a false positive. But it may be that after delving into this person’s story, they were sick, but weeks ago. We know from lots of experience now that some people shed viral genetic material weeks even when an infectious virus cannot be found using methods to try and amplify or “grow” it and the patient looks, and perhaps feels, clinically well.

As Prof Bill Rawlinson noted in the piece..

Our lab and other labs are doing thousands and thousands of tests every day and we are only seeing a couple of false positives.

It’s really a rare event

The main point

Can we rely on our lab results? Absolutely. Australian labs are every bit as good – and often better than labs the world over.

This is the point I really wanted to stress in this interview…

Our labs are good, and have good processes, so you won’t see many false positives.

…and we don’t.

The fact that we know this should tell you something about how expert the labs in Australia are and how well they work with doctors and public health authorities to each check other and resolve in “funny” results through communication.

Views: 5174

25 thoughts on “And another thing…on false positives”

  1. Crank up the number of amplifications… Viola! another ‘confirmed case.’
    How about 1) actually count how many ‘viruses’ are in circulation to see if there is actually an infection? Or 2) just label it ‘confirmed’ and move on?
    Always choose #2, says every govt public health agency.

      1. Then explain how it’s in any way appropriate to set the threshold at 40 instead of maybe 35 when lab studies show any meaningful infection or contagiousness stops at 32 cycles. This seems important. Please don’t hide behind your 3 letters. Actually explain it.

      2. Well can you please explain the NYT article that says the cycle threshold has been set too high and a large % of positive cases were likely never contagious?

  2. Thanks for filling in the gaps. (Not) surprising to see how challenging it is to find anything usefully written about false positives when it comes to COVID testing.

    Do you have any insight into how things are going in the USA? We’ve got Fox News telling us the false positive rate is 96% and everyone who dies in a car crash gets labeled as COVID-positive, and anyone who is actually trained to understand the science gets drowned out in the noise.

    1. If you test positive (and that test was done in a reliable laboratory) and that result is in an appropriate clinical and epidemiological context, you are a “case”.
      You can play games with words as some have done, and say that you tested positive then died fo a car crash. However, this is not death due to COVID-19 – this is a car accident. And that’s where the result + clinical signs and symptoms and epidemiological context all matter…a lot! Try not to listen to media that seeks not to inform but to inflame and cause fear. They are like an immune response running haywire; in the end, it will only cause pain and damage.
      Trust your (clearly good) instincts and seek out actual expertise.

    2. People have died from the flu or some other illness ever since man evolved. This whole covid -19 pandemic is in my opinion is just alot of blown out of proportion propaganda. I understand its there and i understand it may kill people but i also believe alot of deaths are just being attributed to the virus because that had to be the case. I don’t buy it and really dont believe in these tests i dont care if these scientist say false positives are rare, because their not they happen all the time and yet still they get added to their positive test polls. To make this germ and pandemic for some political reason one way or another drag on. I work in the hotel industry and clean rms for i dont know how many different people and who ever they let in their rms to spread their germ everywhere and i can honestly say i have not been sick while all around me here in the Knoxville, TN area is supposed to be covered up by sick people with this germ. Look nobody at the hotel i work at has been sick and died and i cant think of anyone i know even having this covid 19 but you still here poeple loosing jobs and businesses going bankrupt because of all the fear somebody or poeple are telling the world. Im just saying i find it odd i dont know of a single person who actually has been diagnosed with covid 19 oh i should say i have known poeple say they tested positive but three days later say it was false positive on more occasion than ounce so. I have no reason to believe any of this but keep listening to the experts they dont even know what to say about it because every time they get asked about it they just dance around the questions asked.

  3. How could a person test positive using a RT/PCR test one day then negative using the same test and lab the following day?

    1. They were clearly recovering and their virus load dropped below the limit of detection of the test being used.

  4. can a patient diagnosed with COPD who is asymptomatic from COVID 19 TEST POSITIVE and 6 days later test NEGATIVE, Please make me understand

  5. How can I test negative and two days later my husband tested positive for sars 2 ( after being quarantined waiting for a surgery) ? Now he’s surgery is Postponed.

    1. He, unfortunately, acquired the infection from somewhere. Perhaps not from you or perhaps the test you used was insensitive. Hard for me to say I’m afraid.

  6. Mechanistically, are you essentially of the opinion that false positives for RT-qPCR COVID tests are coming from contamination? Most seem to target RdRp IP2 IP4, and E E_Sarbeco as far as I can tell. You seem to strongly imply no potential cross reactivity here, so mechanistically what other processes could be causing false positives? Reagent mix ups, etc? Thinking particularly of the Spanish pre-print study that “detected” COVID19 in Spanish sewage from March 2019 ( “prompted us to analyze some
    archival WWTP samples from January 2018 to December 2019 (Figure 2). All samples
    came out to be negative for the presence of SARS-CoV-2 genomes with the exception
    of March 12, 2019, in which both IP2 and IP4 target assays were positive”.

  7. Mechanistically is it your opinion the vast bulk, if not entirety of false positives are coming from contamination then (at source or in the lab)?

    You treat briefly other confounders. In terms of false positives mechanisms you strongly rule out cross reactivity, so could you elaborate on what else could cause a false positive? Most tests target RdRp IP2 and IP4, Sarbecovirus specific E gene, N2 region in the N gene etc., with these multiplexed tests you mention they can “behave differently sometimes”, can you elaborate on that? If cross reactivity is not plausible, have you any comment (other than contamination) on the Spanish sewage study that had positive PCRs for sewage from March 2019 (see “prompted us to analyze some archival WWTP samples from January 2018 to December 2019 (Figure 2). All samples
    came out to be negative for the presence of SARS-CoV-2 genomes with the exception of March 12, 2019, in which both IP2 and IP4 target assays were positive. This striking finding indicates circulation of the virus in Barcelona long before the report of any
    COVID-19 case worldwide.”?

    1. What is the rate of false positives? In other words, what is the size of the problem do you think? I think it is vanishingly small when testing is conducted by professional and experienced labs.

  8. How do you find the time to deal with these morons? I was lucky enough to catch a special lecture at Uni by Nick Eyre on the subject and he backs you up and so do I. Keep up the great work! It can be difficult to explain science to the brainwashed.

Comments are closed.