(This is a summarized excerpt from our upcoming book “The Covid Survival Guide“.)
Please see down to the bottom for a dramatic update and proof of the inadequacy of all testing procedures.
The good news is that Covid-19 testing is becoming more commonplace, and new types of tests are less unpleasant to experience and offer faster results.
But do the new tests (or the older tests, for that matter, too) actually work? Are they accurate? Let’s take a quick look at testing, what it is, and what it isn’t.
There are three different types of Covid-19 test. One is very different to the other two – it is a serologic test that looks for antibodies created by your immune system. This type of test tells you if you have been previously infected by the virus and have developed some degree of antibodies. It takes at least five days and sometimes more than ten days between when you are first infected and when antibodies start to appear in one of these tests, so it is not helpful as a way of testing for a current infection.
Discovering you previously did have the virus might be helpful for other purposes – it can be helpful to you to know, and to hope that as a result, you now might have some period of immunity. It is also helpful fpr public health officials to get a retrospective understanding of the evolving spread of the virus in the community.
The other two tests are both designed to see if you are currently infected. One type of test has been around since the start of the virus problems, the other is relatively new.
The longstanding test is called a PCR (Polymerase Chain Reaction) test. This usually involves taking a nasal sample by way of sticking what looks like a very long Qtip way up further in your nose than feels at all comfortable. The swabbed sample is then usually sent away to an officially certified lab somewhere to be worked on and to generate a test outcome.
PCR testing can detect amazingly low levels of virus particles. That is a good thing. But a bad aspect of this is that it can not distinguish between a living threatening virus particle and a dead inert virus particle that just happens to be there – perhaps because of a previous infection, or perhaps it is one that entered your system but failed to create a viable colony and an active infection. It can give a false positive result for as long as three months after you’ve had an infection.
It also requires specialized equipment (that is probably a euphemism for “expensive”) and takes about six hours for the test to be run, once the test sample has reached the laboratory (and assuming no backlog of other samples waiting to be tested).
So while widespread (due to the lack of other types of testing that have official certification), the PCR testing is cumbersome, costly, and unpleasant to experience.
The other test is called an antigen test and detects certain proteins present in the virus. These are slightly less sensitive than a PCR test, which means that if you’ve been recently infected and currently have only a low level of virus in your system, an antigen test might miss your infection while a PCR test might pick it up.
The good news about antigen tests is they are fast, and the results can be processed, sometimes without needing any laboratory equipment (or laboratory personnel) at all. Some will show results in 15 minutes, and some are aiming to provide results in closer to 5 minutes. There’s one other very important feature of antigen tests. They are not very expensive. The best known provider of them currently is Abbott Laboratories, to the point that sometimes they are referred to as “the Abbott test”.
A PCR test costs anything up to $2,315, and usually seems to end up around the $1,000 mark (even if not paid by you directly). Antigen tests can cost as little as $5, and more commonly in the $15+ range, depending on the test itself and the service that is doing the test for you. There are even at-home type antigen tests where you can obtain your own sample, and place it in a credit card sort of sized unit a bit like a home pregnancy test and wait to see if the “You have an infection” line appears in the result window.
Another type of test in development might end up as simple as a breathalyser – you just blow into a tube and get an almost immediate reading for your virus status.
This combination of low cost and fast result makes antigen testing practical as a realtime way of determining who is “safe” and who is “dangerous”. Or, at least, you’d think/hope so.
There are a couple of remaining problems. The most obvious one is that it takes several days between when you first get an infection and when you reach the point where your infection has become strong enough to register on a test. How long is several days? Well, different experts give markedly different answers to that. This article, which is almost certainly talking about PCR testing, says 3 – 5 days. But this article – also probably talking only about PCR testing, reports that eight days after being infected, 20% of people still don’t register as having the virus when tested. A third article observes that because the virus itself typically takes 2 – 14 days to make itself felt in an infected person, testing could need a similar lead time to detect it.
Antigen testing probably requires a day or two of extra time to detect the virus than PCR testing.
How long does it take until a person becomes infectious? Therein lies the problem – a person might be infectious at about the same time or possibly even slightly earlier than when a test will detect their infection. There’s not really any advance warning that a test will provide.
This becomes even more of a problem when organizations are requiring a test “within 72 hours” of allowing a person access to somewhere. There’s a measurable chance that the test which suggested a person was clear 72 hours ago failed to detect an infection the person had picked up a day or two, or maybe even a week prior to the test, and even if the person wasn’t infectious at the time of the test, maybe they started being infectious shortly thereafter. This means that someone shows a negative test result and is cleared to travel somewhere or enter some place or whatever else, even though they not only are infected, but are also actively infectious.
It isn’t all bad, though. This is a “corner case” – it is an unlikely rather than certain outcome. Either PCR or antigen tests will pick up somewhere between many and most infected people, and that greatly shifts the odds in everyone’s favor, making it harder for the virus to continue to spread and leading to a steady reduction in cases.
The extra day or two that an antigen test might sometimes need to detect an infection is balanced by its immediate result. And, being inexpensive and easy/convenient, it encourages more testing, and makes it feasible for airlines and destinations to require testing “at the gate” or at the immigration booth.
It is important to appreciate the limitations of these tests. But the solution is not to give up on testing, but rather to test more often. And with nearly instantaneous inexpensive antigen testing, that is becoming more practical, and hopefully soon, instead of being allowed to show a three day old test result that was inconvenient, costly, and also meaningless; people will instead be able to do a real-time test with no delay factor at all. It still won’t detect every infected person, but it is much better than a three day old test, and very much better than no test at all.
Update, later on Thursday
News has just broken, first that Presidential advisor Hope Hicks has been diagnosed with the Covid virus, and then a short while later, both the President and First Lady advised they have tested positive for the virus too.
It is not known, but is generally being guessed, that the President probably caught the virus from Ms Hicks. They had been in close contact several times this week, including in Air Force One and Marine One. Everyone who gets close to the President needs to be tested for the virus every day.
We don’t know what type of testing has been done on Ms Hicks every day, but clearly the testing failed to “beat the virus” and she became infectious before a test picked up her infection.
That’s not to say all testing is useless. Not at all. But it surely does dramatically confirm our earlier observation that there can be a dangerous gap with people becoming infectious before a test shows their infection.
Please click here for a listing of all our Covid-19 articles.
4 thoughts on “Covid Testing – What it Is, and What it Does and Doesn’t Do”
A concern with PCR testing is that no one reports how many “cycles” it takes to become positive. The more cycles run, the longer it takes to get results. There has been some commentary that current testing is too sensitive and may detect not only live virus but genetic fragments. No one but the lab knows what cycle threshold is used since they don’t report that. But many use 37 to 40 cycles as the threshold for positivity. Some virologists think a more reasonable cut off would be 30-35 cycles. CDC calculations suggest that it is extremely difficult to detect any live virus in a sample above a threshold of 33 cycles. In New York, their lab identified 872 positive tests based on a threshold of 40 cycles in July. With a cutoff of 35, about 43 percent of those tests would have no longer qualified as positive. 63 percent would not have been judged positive if the cycles were limited to 30. Likewise in Massachusetts, from 85 to 90 percent of people who tested positive in July with a cycle threshold of 40 would have been deemed negative if the threshold were 30 cycles. This information is from a NY Times article “Your Coronavirus Test Is Positive. Maybe It Shouldn’t Be.” current version of 9/8/20.
This is not to say that people shouldn’t get tested, but could lead one to the counterintuitive conclusion that that the antigen test may actually be more useful than PCR to find actually infectious patients, due to it’s lower sensitivity combined with rapid turnaround.
Food for thought.
Thanks for this excellent information, which is in line with my understanding too.
You make an interesting point. “Positive” and “Negative” results are not like positive and negative results from a pregnancy test. It is not a binary choice. Whereas you can’t be “a little bit pregnant” you can be a little bit positive.
This is just one of the thicket of complexities and subjective settings that intrude on what simplistically seems as a simple process.
I totally agree that the cycle count should be disclosed.
You mention that PCR testing detects very small amounts of virus, including “dead” virus fragments. The fact that it is a sensitive test is good, but I have seen reports that the false positive rate is quite high. What have you seen in this regard?
You are quite correct about high false positive rates. It depends on how many cycles are run to “amplify” the sample. See the excellent comment from James Mowry (it wasn’t released when you sent in your question, but has been now and probably gives a good answer to your question).