Subtitles section Play video
It seems intuitive that a more sensitive COVID test is better than a less sensitive one. And yet,
you may also have seen news stories suggesting that some COVID tests are too sensitive,
or that cheaper, faster (and even less sensitive?) tests might be better. But
why would they be better? And better for what? According to public health experts at Harvard
and Brown University, in the fall of 2020 the US needed to do 8x more testing to successfully
suppress COVID-19, India needed to do 4X more testing, Brazil needed to do 66X more testing,
Bolivia needed to do 140X more testing, and Mexico needed to do over 400X more testing.
The idea behind saying “COVID tests are too expensive and too sensitive” is that part
of what's kept countries from doing enough testing is the kind of tests being used.
The current gold-standard for COVID testing is a nasal swab PCR [polymerase chain reaction] test.
This test is highly sensitive, which means it's able to detect the presence of the virus
even at very low levels of viral load. That's great for a doctor who wants to know whether
their patient was infected with COVID-19. But if you're screening an entire community for COVID
outbreaks, then PCR tests have a few downsides. First, PCR tests are expensive. To avoid breaking
the bank, many places are focusing on testing only people who already have COVID symptoms or who were
recently exposed to someone with COVID. Using only expensive tests means that we don't test
(and therefore, don't catch!) most people who are contagious without showing symptoms – and
we know that a significant proportion of COVID transmission comes from these asymptomatic people.
Second, PCR tests take time: usually a couple of days for results to come back from the lab. Here's
what the typical viral load in an infected person looks like over time – they're most infectious
around the peak and their infectiousness declines afterwards. So if test results take a few days to
come back, chances are you miss this peak – which means missing the window of time during which the
person is most likely to infect other people. Essentially, although PCR is very sensitive,
it's also expensive and slow, so while it's great for diagnosis, it may not be the best test for
screening a community. And the cheaper, faster testing alternatives are – no surprise – not as
sensitive as PCR. BUT - and this is a subtle point and maybe wouldn't matter if PCR were cheap and
fast, BUT when it comes to screening a community, PCR's high sensitivity could itself be a downside.
PCR tests can detect very low viral loads, which is great if you happen to test somebody very
early on in an infection, or very late and you want to confirm whether they were infected. But
this sensitivity also means that a person can test positive for the infection when they're
basically no longer infectious – and in fact, PCR is so sensitive that it can detect viral fragments
and give a positive result even when there are no viable viruses left. So it's likely that by
using PCR tests we're quarantining & contact tracing some people when we don't need to.
It's not that the PCR test is a bad test, in fact it's a great test for what it's designed to do,
which is to accurately and reliably detect even very low viral loads. It's just that it's more
expensive, slower, and perhaps even more sensitive than would be ideal for screening a community.
A less sensitive test could catch just as many highly infectious people as PCR,
but it would miss people who are post-infectious, which you could argue is a good thing for
screening purposes since those people are no longer a significant threat. The
downside is that you'd also miss people who are pre-infectious, which is undoubtedly a bad thing.
But the math works out that if you have a test that's just a little bit less sensitive than PCR,
but you use it a lot more frequently, you're actually more likely to detect an
infection compared to infrequent PCR testing. So can you identify more infections overall.
All of this suggests that if you're trying to screen a community for COVID outbreaks,
you can get more testing bang for your buck with rapid testing than PCR testing;
that is, you can identify more infectious people more quickly for less money.
As long as the rapid tests are only slightly less sensitive than PCR tests but a lot cheaper.
Rapid testing is not a replacement for PCR testing in a clinical setting,
nor is it a replacement for masks, physical distancing, handwashing,
and so on. And if PCR testing cost 1/10th of what it does and took 1/10th the time,
we probably wouldn't be having this discussion. But if we're going to test widely and frequently
enough to suppress COVID (and we really should), then we can get more for our money
with tests that are cheaper, faster, and - perhaps counter-intuitively - a little bit less sensitive.
Ok, so that's the argument for rapid testing, and here are the caveats.
No medical tests are perfect, and a negative test result isn't a foolproof guarantee of not being
infected. Because rapid tests are less sensitive than PCR tests, this makes it more likely that
infected people will slip through the cracks of a test. So as long as COVID is a problem,
it's important to still take precautions like wearing a mask and social distancing
even if you test negative. And for people who were likely exposed to COVID,
health authorities recommend that they quarantine for 14 days regardless of a negative test result,
and that their negative rapid test results are verified by a PCR test.
When a rapid test is just a little less sensitive than PCR, then the downsides of
missing some infected people can be balanced out by the benefits of more widespread and frequent
testing. But this only works up to a point. If the sensitivity of the rapid test is too low,
the cost of missing many infections will outweigh the benefits of widespread testing.
The graph we showed for viral load was based on measurements of patients' viral loads,
and while all else being equal, a higher viral load tends to mean you're more infectious,
there are a lot of other things that can affect how infectious you are
We didn't really mention what happens when you aren't infected but the test gives a positive
result anyway – this isn't likely on any given test, BUT if you're testing a huge number of
people frequently, then just by statistics alone you can end up with a large number of false
positive results. The rate of false positives is comparable between PCR tests and rapid tests,
but matters more with cheap rapid tests since the whole idea is to do way more tests.
Finally, vaccines. So won't all of this be irrelevant once vaccines come out?
Well, vaccines will certainly help, but unfortunately, testing will probably still
play a major role in suppressing COVID. It's really a topic for another video.
A very big thankyou to Brilliant for sponsoring this episode of
MinutePhysics - with MinutePhysics, I try to bring you deeper level explanations,
while Brilliant engages you to solve fascinating, challenging problems yourself,
guiding you to a deeper level understanding of science and mathematics. Like their course
on knowledge and uncertainty - what's the most efficient way to zero in on the true criminal
from a lineup of suspects? Or, how worried should you be if you know a medical test is 90%
sensitive and you get a positive result? To improve your knowledge and reduce your
uncertainty, go to brilliant.org/MinutePhysics/ and sign up for free. The first 200 people will
get 20% off the annual Premium subscription, with full access to all of Brilliant's courses,
puzzles, & daily challenges. Again, thanks to Brilliant for their support.