How accurate/useful are rapid Covid tests?
February 1, 2021 9:31 AM   Subscribe

We're told not to use rapid Covid tests to decide that you are "negative" and are free to socialize without risk of spreading Covid to others.

For example this article, entitled "'Worse than flipping a coin': Quidel rapid COVID-19 test fails in asymptomatic people" says that this test has a 40-something accuracy at identifying people with asymptomatic infections. That's pretty bad!

However, later in the article, there is this:

"The Sofia test is much quicker and cheaper than lab-based tests, called PCR tests, but PCR tests are much more accurate. Sofia's instruction manual claims it can detect 96.7 percent of the infections that PCR tests can when used on people in their first five days of illness, the Times reported."

WHAT? But PCR tests are much more accurate than 42%!

David Harris, a stem cell researcher and one of the study's authors, said some of the asymptomatic people may have been missed because they carried too little of the virus to spread it to others.

Okay, I get that. Much has been written, lately, that PCR tests may be TOO accurate: that perhaps many (? some?) people who test positive are actually beyond the point where they are contagious, and, if they're clinically okay, a test isn't even necessary or helpful. It's the SPREAD that we care about.

Michael C. Mina of Harvard, whose mission is to bring rapid (and preferably cheap, and preferably at-home) tests to the public, stresses that this "window of contagion" is far more important than identifying just any amount of virus that's left in somebody who's been infected for awhile -- because, if you're beyond that contagion window, and you're not actually experiencing symptoms (and maybe never have), what's the point of being tested?

(Dr. Mina also says that, if there's not enough virus to produce a positive result on a quick test, that means that there is not enough virus present to infect another person. Is this true? His idea is that the test doesn't haven't to pick up ANY amount of virus; just the RIGHT amount of virus -- enough that the person is contagious (a couple of days after initial infection to maybe 7 days after initial infection) -- so testing too soon is dangerous, and testing too late is unnecessary -- and both are distortions, but the "too soon" distortion is dangerous, whereas the too late distortion is more like a pain in the ass for people who are missing work etc because they're isolating unnecessarily).

At the end of the article I referred to, "Quidel (the test manufacturer) told the Times the test was good for rooting out people who are infectious." What?

They also cite studies that give much higher accuracy stats than the study cited in the article (duh, they're a drug company).

So -- what is the latest reliable information on these tests? when I do searches I get a hodgepodge mass of information from various points in time, all kinds of sources etc etc #the internet

[Sidebar question: if you get TWO of these tests in a row, do you necessarily increase the accuracy? that is, would a false negative always be a function of something external like the swabbing, or the examination of the test by the testing center? or could it be something in the PERSON that is producing the false negative and that is likely to be in effect no matter how many of the tests that person takes in a day or so?]

[Sidebar question 2: will better rapid tests be coming soon, maybe? There's something coming to the pharmacies in the US very soon that you take at home, but I don't think their accuracy is so great either -- amirite?]
posted by DMelanogaster to Health & Fitness (22 answers total) 3 users marked this as a favorite
It's important to be really clear what we are comparing, and I think a lot of these links use unclear or imprecise language.

For example:
Sofia's instruction manual claims it can detect 96.7 percent of the infections that PCR tests can when used on people in their first five days of illness, the Times reported."
If you look at the Sophia instructions, what it actually says is that it can detect 96.7% of infections that PCR also detects within 5 days of onset of symptoms.
posted by muddgirl at 9:47 AM on February 1, 2021 [5 favorites]

Rapid tests have their place. At my institution, rapid tests are generally recommended for people who are already feeling unwell, or otherwise presenting signs of active infection. If you're at the point of having signs (like fever) and symptoms (like a dry cough and loss of sense of smell), you're much more likely to have a viral load that's above the lower limit of detection for the test kit. So, think of it more as a quick option to confirm that someone is suffering from COVID-19 rather than the flu. If the rapid test is negative in someone who feels unwell it shouldn't be the only test given, but if it's positive then there's your answer. Think of the rapid test as an effective but incomplete approach to adding certainty to a likely clinical diagnosis.

If you're asymptomatic, a rapid test is only useful when there is no access to PCR-based assays, especially if it's possible to stagger multiple rapid tests over time. Yes, two negative rapid tests over a 10 day period (for example) gives more confidence in a negative result than a single test, but a rapid test is nevertheless most valuable when the results in question are positive.

Yes, more reliable rapid tests are coming but they will (almost certainly) never approach the level of diagnostic power that a PCR-based assay has.
posted by late afternoon dreaming hotel at 9:52 AM on February 1, 2021 [6 favorites]

My very-layperson understanding is that the rapid tests can be useful at a population level, but not for individuals. So, for instance, schools could implement daily or weekly rapid testing and catch a significant proportion (but not 100%) of infectious individuals before they infect the rest of the school. But a rapid test is not accurate enough to decide if it's safe to visit grandma, because the accuracy is low and the consequences of a wrong answer are high.

Here's a good summary from Dear Pandemic. In general, I have found them to be a reliable source of information that's geared towards the general public.

Additionally, as muddgirl points out, the rapid tests are almost as good as PCR at detecting symptomatic cases, but much worse at detecting asymptomatic (no symptoms) or presymptomatic (no symptoms YET) cases, and those are the ones you really want to find for the visiting grandma use case. (Presumably, if you're bothering to test before visiting, a cough or fever will also stop you from visiting, even without a test.)

Re: sidebar #2 -- the soon-to-come over the counter Ellume test can detect 91% of asymptomatic cases identified by PCR, so it's better than other rapid tests, but still not as good as PCR. (NYT, via Dear Pandemic)
posted by natabat at 9:56 AM on February 1, 2021 [12 favorites]

Echoing late afternoon dreaming hotel's point: my experience with the rapid test has been when I or a family member have been symptomatic for something and need to show that it's not covid. The state I live in has approved some brands of rapid test for this specific use case (kid has symptoms, needs to show it's not covid to go back to school.)
posted by damayanti at 9:56 AM on February 1, 2021

Another place I've heard of rapid tests being used effectively is regular (like, daily) in congregate settings where a lot of people are likely to get infected at once - if you detect one such person, everyone else needs to quarantine. But yeah, they're not good for knowing whether you, personally, have asymptomatic COVID. What natabat said, basically.
posted by mskyle at 9:58 AM on February 1, 2021 [1 favorite]

In other words, the Sophia test (and other similar rapid tests) can answer the question: "this patient has recently developed flu symptoms, is it Covid?"

For economic reasons, people want to use it to answer the question "Is this person infectious?" This is the current debate.

On preview what others said.
posted by muddgirl at 9:58 AM on February 1, 2021 [2 favorites]

Read this: Antigen Testing: Guest Post with Michael Mina. The tl;dr is that current rapid tests do a very good job of telling you if you are currently infectious, but PCR tests tell you if you have Covid. In other words, rapid tests are not less accurate. They are more specific.

Whether a covid+ person testing negative on a rapid test is a false negative depends on your perspective. If you're a doctor wondering how to treat someone then you would consider it a false negative, and you should use the PCR test. But if you're a public health official (or you run an institution such as a school) a rapid test is probably better because it tests for the thing you actually care about, which is "is this person infectious today". It's a bonus that it does this quickly and cheaply, but it's real selling point is it tests for the thing you care about, not the "the thing you care about plus a bunch of other stuff".
posted by caek at 10:15 AM on February 1, 2021 [1 favorite]

Response by poster: caek:

re: " this person infectious today"?

If a rapid test can answer that question with pretty great accuracy (according to Dr. Mina), then the people who are using those tests to answer the question, "Can I go visit my parents today?" seem to be using the test appropriately! (as opposed to those who say the tests are only good at a population/crush-the-pandemic level)

THIS is what is confusing to me! (i.e., Which is it?) Is the point that, on a pandemic/population level, Mina and others are willing to accept that some people are going to slip through, but they're making progress on a population level, and that's how they think about things?

-- as opposed to telling my son to get a quick test and, if he tests negative, say, "Sure, come on over!" -- which many people I know have been doing for months now, BUT which I have NOT been willing to do (because I am 70 and an individual person with a heavy dose of self-preservation -- not a data point in a pandemic study -- but -- am I wrong?). NOTE: will be fully vaccinated on Thursday (!) and wonder if, after a few weeks, the entire situation re: visitation changes - but that's another question for another week! THANK for the great responses.
posted by DMelanogaster at 10:35 AM on February 1, 2021

muddgirl wrote, It's important to be really clear what we are comparing, and I think a lot of these links use unclear or imprecise language. I think so, too; your question links to Becker Hospital Review ["up-to-date business and legal news and analysis relating to hospitals"] for an article written by a music journalist.
posted by Iris Gambol at 10:51 AM on February 1, 2021 [1 favorite]

A person can have an infection without having an illness. This encompasses the people who are asymptomatic carriers that never develop symptoms. Similarly, the illness is considered to begin when the signs and symptoms are present.

So if you don't have enough virus to be coughing and/or fevered, you may still have enough virus to spread but not enough to get back a positive rapid result.

The rapid tests say that they can detect the virus in the first five days of illness are basically saying "if you're symptomatic, we'll have a Really Good chance of finding the virus" and they are not commenting on their success at asymptomatic detection. For folks who aren't accustomed to thinking about the difference between infection and illness, this distinction is not intuitive.

The rapid tests do catch many infections where a person has a fever and is wondering "is this covid, or allergies, or a cold?" and it can catch some infections where there are no symptoms but lots of virus replication happening.

The rapid tests are good for helping you make decisions about treatment for symptoms, but the standards to not gather in groups, not gather in enclosed areas, and wear masks remains the same if you have a negative test.

If you get a positive on a rapid test when you have early symptoms it is a clear indicator to be more aggressive in treatment than you would be for a regular cold, so that you have better odds of staying out of the hospital.

Rapid tests with negative results are not a permission to make out with strangers/visit your grandma/go bowling.
posted by bilabial at 11:04 AM on February 1, 2021 [1 favorite]

Rapid tests do have "real" false negatives, i.e. they sometimes fail to catch a currently infectious person. The details of that rate will depend on the specifics of the test, and the situation is changing rapidly, so I'm reluctant to put a number on it. And I should be clear that rapid tests are most sensitive to symptomatic people. Symptomatic people are much more infectious than non-symptomatic people, so it's a reasonable approximation to say rapid tests detect "infectious" people. But if your own life is at stake that's probably not the right way to think about it.

I don't think you're wrong to be uncomfortable with the risk of false negatives. Personally, if I were you, and I lived somewhere I could expect to get the vaccine in the next couple of months. I would wait. One thing to consider though is, assuming you live somewhere where the rate of covid is going down (e.g. most of the US) then that risk is getting smaller. (False negatives never happen if nobody has covid!). So the risk might not be small enough for you, but it's probably going down, for what that's worth.

Certainly the way you think about false negatives will be different depending on who you are. A public health official might be OK with "I guess my measurement of X will be wrong by ~1%" if the alternative is "I do not know what X is, and I'm flying blind". Someone trying to open a school so kids can be less miserable and parents can work more easily might be OK with the risk of an outbreak among a relatively non-vulnerable population. I think that's Mina's perspective. YMMV (and it should!)
posted by caek at 11:12 AM on February 1, 2021 [2 favorites]

Your two points are, indeed, totally different things. I'm not an epidemiologist, but I do have enough of a science background that I can describe how we've interpreted our tests in the context of being individuals who want to visit with family.

First, we quarantined for 5-7 days before testing. Based on what I've read, that's the length of time by which any symptoms should have shown themselves, and when we'd be likeliest to test positive if we were asymptomatic.

Second, all 3 people in my household got tested early enough to get results before meeting, and did an honest assessment of ourselves as having had no symptoms while quarantining. While any one test isn't a guarantee, three negatives among three people quarantining together in close quarters is a lot closer to one. This was enough for me to feel that the chances of killing my mother by sharing a meal were minimal. It probably wouldn't have been enough for my comfort if we'd been visiting my 90+-year-old relatives; in that case I might have done two tests, at 5 and 7 days of quarantining, to halve the chances again.

Locally, rapid tests are also being used in public schools as a general tool for confirming that community positives are where we expect. Again, even if any one positive person has a 50/50 chance of testing positive, a school that has 0 positive tests indicates that there are very few or zero positive cases that are getting missed by all the other tools we are using to prevent COVID-positive people from being in the school.
posted by tchemgrrl at 11:17 AM on February 1, 2021 [1 favorite]

Okay, so I'm still confused here, the gold standard here still that no matter what if you test negative, you may still have the virus/being infectious and it just hasn't showed up yet? So just because you test negative today, you can and may be still infected and passing it to others so you still can't see Grandma? So testing negative means you are never, ever 100% clear of being infected anyway?

Or is it, "tested negative, quarantined for 14 days, tested negative again" the only gold standard to be in the clear?

Or are you just never, ever in the clear until 28 days after your second vaccination?

God, I'm sick of Schroedinger's Virus.
posted by jenfullmoon at 11:27 AM on February 1, 2021 [4 favorites]

I can't comment on testing but re: vaccination - public health advice here is that even if you're vaccinated they don't know if you can pass on the virus so you still have to mask and social distance while there's community spread.

For me, I think it would be in the clear at home if both parties, myself and grandma, had been vaccinated twice with 28 days after the second shot. In other words, mask up while out but at home an "all parties are vaccinated" would make the risk tolerable.

Since 'here' is Ontario it's still a pretty moot point as we're out of vaccines for a while. Sigh.
posted by warriorqueen at 11:38 AM on February 1, 2021 [1 favorite]

I don't want to derail the thread here, but since we've wandered in the epistemological question of whether we can be certain: no, vaccines to not confer 100% immunity and no, tests are not 100% accurate. But don't make the "evolution is just a theory" mistake and conclude that the absence of logical certainty means you don't know anything and we need to continue to hibernate forever. Faster, cheaper and more accurate (and, yes, imperfect!) tests and (imperfect!) vaccines will change everything. If you have access to these things then you can do things wouldn't otherwise do. If everyone gets access we can all do things we wouldn't otherwise do. Exactly what those new things are is going to be a local and personal question.
posted by caek at 11:51 AM on February 1, 2021 [3 favorites]

The only experience I had with the rapid test was a friend who has quarantined from the start of all this; works from home; and has a very small bubble of friends/significant other. She was very much wanting to do Christmas with her significant other, yet had to attend a funeral for family member. She scheduled her rapid test for three days after the funeral and a week before Christmas.

Her rapid test came back positive and she threw herself into hysterics. She had no symptoms at all. She scheduled a regular test for two days after that, and was found to be negative. So I'm not sure I have faith in rapid tests.
posted by annieb at 12:40 PM on February 1, 2021

To this point: " If a rapid test can answer that question with pretty great accuracy (according to Dr. Mina), then the people who are using those tests to answer the question, "Can I go visit my parents today?" seem to be using the test appropriately! (as opposed to those who say the tests are only good at a population/crush-the-pandemic level)" there are social variables here to consider. Mainly, what does "visit my parents today" mean. Many people have been relying (ill-advisedly) on rapid test results to give them the okay (or false sense of security) that we're all good and can gather (for example) for a week with a dozen other folks in a vacation home. We all got negative results on rapid tests, so we're good, right? Well, as outlined extensively above, if anyone did have an infection but was pre-symptomatic, the tests aren't telling the whole story.

Could a rapid test be part of the decision making factor to help ease minds in a more careful meeting? Sure. Are you able to stay outside/distanced for the visit? Wear masks? Limit your time together in proximity. All the other things that follow the "Swiss Cheese" model.
posted by leastlikelycowgirl at 1:24 PM on February 1, 2021 [1 favorite]

Response by poster: "Until everyone is vaccinated" -- sigh.
posted by DMelanogaster at 1:42 PM on February 1, 2021 [1 favorite]

Best answer: Rapid tests are good at ruling people IN; a positive test is very likely to mean that you have covid and need to quarantine.

They are not great at ruling people OUT; a negative test doesn’t necessarily mean you don’t have covid, which is why you can’t use it (or any other test) to justify an indoor no-mask gathering with loved ones.

No one knows for sure how much virus you need to have in your body (AKA your “quantitative viral load”) before you become infectious to others. It is *definitely* not as simple as saying that if you don’t have enough virus to trigger a positive test result, you can be 100% certain you won’t spread it to anyone else.

Testing is and has always been a tool for identifying positive people in order to immediately isolate them and prevent them from transmitting the virus to others.

It has never been, and should never be, used as permission for people testing negative to avoid following necessary public health measures like masks and physical distancing.
posted by jesourie at 2:05 PM on February 1, 2021 [6 favorites]

Oh, and the answer to this:

“... is the gold standard here still that no matter what if you test negative, you may still have the virus/being infectious and it just hasn't showed up yet? So just because you test negative today, you can and may be still infected and passing it to others so you still can't see Grandma?”

is yes.
posted by jesourie at 2:30 PM on February 1, 2021 [4 favorites]

On a related note: "Chappelle and his team have been conducting daily Covid-19 tests throughout the pandemic. Audiences at the Stubb’s residency receive rapid antigen tests upon entry. Rogan, who’s been providing rapid testing for his podcast guests since April, wrote on Instagram two days after Chappelle’s diagnosis that he “tested negative every day this week.
“They are approved for use when someone has COVID-19 symptoms, so using them to screen asymptomatic people who want to gather at a club is not necessarily going to prevent an infected person from slipping through,” she told us. Antigen tests like the ones Rogan and Chappelle provide at their Stubb’s residency require a relatively large amount of virus to turn up positive; people with lower amounts of viral particles—perhaps because they’re early in their infection—may receive a false negative even though they’re perfectly capable of spreading the virus.”

posted by jenfullmoon at 2:38 PM on February 1, 2021 [1 favorite]

Rapid tests (of any sort, not just covid) are optimized for sensitivity and its cousin, positive predictive value. A positive test = you should quarantine.

A confirmatory test (of any sort) is optimized for specificity and negative predictive value. That's what you want for the use case of "is it safe to see my parents." Rapid tests are not designed to answer that question.

This is one of the trickiest bits of epidemiology. Further explanation here.
posted by basalganglia at 5:57 PM on February 1, 2021 [4 favorites]

« Older Gift for new Torontonian?   |   Better public speaking [difficulty level:... Newer »
This thread is closed to new comments.