TestingLast updated September 11th 2020, 6:21:53pm
Testing: The Basics
There are two main types of coronavirus tests out there: PCR (polymerase chain reaction) and serology. You’ve probably heard about both of them. Broadly, PCR tests for current infection, and relies on a nasal or throat swab. Serology tests, also known as antibody tests, test for past infection, using a blood draw. Of course, it’s a bit more complicated than that.
There are no formal FDA approved versions of either test yet. But the FDA has approved at least 75 tests so far under the “Emergency Use Authorization” authority, meaning they can be used now even though approval is still pending.
These tests get a lot of discussion. There are those who see the expansion of testing as the way out of the crisis. And there are those who argue that tests are not yet sensitive enough. To understand why there is so much discussion — and so much disagreement — about tests we need to start by understanding the science behind them.
PCR Tests: The Overview
Let’s say you’re showing symptoms of COVID-19, and you need a test. Or, let’s say you run a labor and delivery unit at a hospital, and you want to test laboring moms. In these, and many other situations, the test you’ll rely on is the polymerase chain reaction, or PCR, test. PCR tests are widely used for many kinds of viruses — influenza is the one you are likely most familiar with. So the technology for doing this test is well established.
The first step in this test involves collecting samples from an individual’s nose or mouth using a cotton swab. A healthcare worker (or someone else, but really, you need someone with a lot of training) will carefully extend a long cotton swab into an individual’s nose or throat for a few seconds, allowing it to soak up mucus.
The most comprehensive way to do this is with what is called an NP swab - a very long swab that goes all the way up your nose or way, way down your throat. These swabs are extremely unpleasant, but they are the most likely to catch the virus, since they access the areas that the virus is most likely to live. They are also the most heavily validated; they have the most data suggesting they work.
While these NP swabs are regarded as the gold standard, there are other collection methods that are gaining popularity.
Less invasive swabs, including “mid-turbinate” and “anterior nasal”, are much less unpleasant and have the advantage that people can do it themselves. There are fewer studies investigating the accuracy of these swabs than NP swabs, but recent analyses suggest that these swab types are almost as good as the NP swabs, and other small studies show that they’re as good. However, if you are actively sick, an NP swab still likely makes sense as we have more data showing that this test has low false negative rates. For surveillance purposes — like, for doing widespread population testing — the less invasive swabs are likely to take over.
Saliva tests are also an option. On August 15th, the FDA granted emergency use of a new saliva test for COVID-19 called SalivaDirect. Saliva tests allow people to collect their own sample at home by spitting in a cup that is then returned to a lab for analysis. Like the self-swabbing methods, saliva tests are advantageous in that you can collect your own sample. These tests are expected to reduce time, money, and training required to administer the test. SalivaDirect has so far yielded results with a sensitivity as high as nasal swab tests, setting it apart from past saliva tests that have received FDA emergency use authorization.
Once a sample is produced — be it a mucous-laden swab or saliva cup — it is sent to a lab, where its contents are placed into a PCR machine to do the actual test. The machine looks for the genetic material of the virus. This is hard to do because there is a lot of different genetic material in your nose, like your own genetic material or other stuff you breathe in. Maybe only a little tiny part of it is the coronavirus. (Note: another benefit of SalivaDirect’s new saliva test is that it skips this nucleic acid extraction step and goes straight to converting RNA to DNA).
To find the coronavirus you need to replicate its genetic material so there is more of it and it’s easier to detect. In the case of coronavirus specifically, you cannot do this directly because it is what’s called an “RNA virus”. This means that the genetic material of the virus is RNA, not DNA. Other RNA viruses include the flu, measles, or rabies. RNA is fragile. It’s single-stranded, not double-stranded like DNA. It also has a different backbone structure which makes it degrade more easily; for instance if hair or skin particles flake off onto it, it degrades. So if you try to replicate the RNA genetic material, it will likely break down, and you wouldn’t be able to identify the virus.
To avoid this, the first step in the process is to convert the RNA into DNA by introducing a tiny molecular machine that turns the single strand into a double strand. This tiny machine is an enzyme (a chemical) called reverse transcriptase. So (very technically) the PCR process that is used for coronavirus is an “RT-PCR” (reverse transcriptase PCR), since we start with RNA and then transform it into DNA.
Once the DNA is produced, the second step in the PCR is to target the viral sequence for copying. Right now, your genetic material and the virus’s genetic material (if you’re infected) are all mixed together in one big soup. Let’s think of all of the genetic material present as chapters inside a book, and the coronavirus is the epilogue. Scientists will select a short string of words, let’s say 100 words, that they know is only in the epilogue — a string of genetic material that is only in the coronavirus and not in human DNA. If they find these 100 words, they copy this part of the epilogue (the viral DNA) over and over again by a copy machine (an enzyme) that can only recognize this string of 100 words.
That small section of the coronavirus genetic material gets copied over and over again — literally millions of times — which makes the genetic material easier to detect. Ultimately if the virus is there, the PCR machine will be able to detect the copies and produce a positive result. If there is no coronavirus genetic material, nothing will be copied, and the test result will be negative.
The lab process for the PCR test takes just a few hours. However, it can often take days to get test results back, mostly due to the logistics of getting samples to a lab and the data analysis and validation.
These tests are very accurate if the sample is taken correctly, and at the right time. In particular, a PCR can generate a false negative result — telling you that you’re not infected when you really are — if not enough sample is collected, if the sample is collected in the wrong way, or if you test too early (or too late) after infection. As your body fights the virus, there is less and less of it to detect. And once you’ve recovered, there is no virus. So if you’re tested at a time when there is less of the virus in your system, you are more likely to get a false-negative. Similarly, very early in infection the viral load is low, and tests may miss it.
To give you a sense, the graphs below show the accuracy of tests — taken by either nasal or throat swabs — based on days from when symptoms first present. If used right when you first get symptoms, the PCR test works well. As the virus evolves, the probability of detecting the virus decreases, even for those who initially tested positive. With nasal swabs, the accuracy is about 94% on “day zero”, when symptoms first present. However, by day ten, it’s only 65%. The same is true with throat swabs, which are 88% accurate on day zero but only 47% on day ten.
You’ve got questions! We’ve got answers.
What about this machine that does the test in five minutes? What’s that about? Yes, there is such a machine (or several). They work similarly to PCR tests, by amplifying the genetic material of the virus. Their exact mechanisms are proprietary (super) but one thing we do know about them is that the process is an “isothermal nucleic acid amplification” which means you do not need to change the temperature during the test,which you do in PCR.
These tests are fast, and the machine itself is portable. Downsides include their inability to handle a large number of samples at once as well as a 15% false-negative rate.
A rapid molecular test (called N1-STOP-LAMP) that can detect COVID-19 from nasal swab samples in 20 minutes has been recently validated by scientists in Australia. It is not intended to replace traditional PCR testing since it’s not as sensitive, but it does have the ability to detect COVID-19 infections quickly, especially in cases with a high viral load.
This nose jamming business seems uncomfortable — is there any better way? Potentially, yes! In principle, one could use saliva to do the same kind of test. In fact, a saliva-based testing is currently gaining traction: On August 15th, the FDA granted emergency use of a new saliva test for COVID-19 called SalivaDirect. Saliva tests allow people to collect their own sample at home or at a clinic by spitting in a cup that is then returned to a lab for analysis. Saliva testing requires less protective equipment from testers and eliminates the need for nasal swabs, which have been prone to shortages. Additionally, this new test, which was developed by researchers at Yale, doesn’t require nucleic acid extraction--a key step that has caused shortages in chemical reagents and delays in testing in the past. SalivaDirect has so far yielded results with a sensitivity as high as nasal swab tests, setting it apart from past saliva tests that have received FDA approval.
Will the PCR test detect infection if I do not have symptoms? Yes. If you are infected with SARS-CoV-2 this test will detect the presence of the virus even if you are not experiencing symptoms.
If I’m doing something risky where I may be exposed, should I get tested right afterwards to confirm I didn’t get it at the event? Probably not. If you head straight from a crowded wedding to getting tested, there isn’t enough time for the virus to amplify and infect you. Two or three days is probably better, but keep in mind that it’s possible that you are contagious after two days and it’s also possible that it takes much longer for a test to detect the virus. False negatives are more likely when the virus is present in smaller amounts earlier on in the infection.
Will the PCR detect past infection? No. This is a test for active infection. For past infection, we need serology (see below).
Is PCR the same as antigen testing? While PCR and antigen tests both detect current infection using samples collected via swab, they are not exactly the same. Antigen tests generally look for proteins associated with the virus rather than at the genetic material. These tests are not as widely used for detecting SARS-CoV-2 infections yet. The FDA has recently granted emergency use authorization to an antigen test intended for patients who have developed COVID-19 symptoms in the past 12 days. This is the third antigen test that the FDA has authorized. Antigen tests are faster than the typical PCR tests, but tend to miss more infections. The new test, developed by a British company, reportedly gives results in 12 minutes. It’s important to note that antigen tests have a higher rate of false negatives than PCR tests, so even if you get a negative antigen test result, you might still have the virus (see our article about antigen testing for more about the pros and cons of antigen tests).
Can I get this test? If you have symptoms and live in a place with good testing, probably. This is the kind of test that most mobile testing sites are doing. Hospitals certainly do a lot of them. At the moment it’s still the case that most tests are reserved for people who are showing symptoms, but that may change.
Do I need this test? If you have symptoms of COVID-19, you should try to be tested, so you can avoid spreading to others. If you’ve been exposed to the virus, you may want to be tested, although generally this is not recommended unless you develop symptoms. If you do not have symptoms or exposure, there is no need to be tested and given how unpleasant it is, you probably will not want to be.
Serology Tests: The Overview
The other approach to testing is a serological test, perhaps more commonly known as an antibody test. The big advantage of this test is that it can identify previous infection even after you’ve recovered. If you want to know the answer to the question “What share of people have had the coronavirus and recovered?”, you’ll need to test for antibodies.
These are the tests that can tell us if we are moving towards herd immunity or if a significant share of the population is still susceptible. In the case of coronavirus, these tests are valuable in part because a lot of people have the virus without symptoms or with very mild symptoms. These people would probably never reach out to a doctor or hospital to detect infection. Without widespread serology testing, it will be hard to figure out how much the virus has gotten around.
The disadvantage of antibody tests is that they don’t detect active infection. If you want to identify people early in their illness so you can treat or (more likely) isolate them, this test will not help.
The reason for this lies in how serology tests work — namely, by testing for “antibodies.” Antibodies (in general) are proteins your body makes that stick to and kill pathogens (like coronavirus). Generally, about a week after you become infected you begin to produce antibodies. These antibodies are specific to COVID-19; they are designed to fight this particular virus.
They come in five classes of “immunoglobulins”: IgA, IgD, IgE, IgG, and IgM. The “Ig” stands for immunoglobulin and the last letter is based on their amino acid makeup. All of these antibodies are aimed at killing the virus, although they all have slightly different roles, and their timing differs. For most viruses, IgM is produced first, followed by IgG and IgA. Less is known about IgE and IgD; IgE is generally involved in allergic reactions rather than pathogen immunity. In the case of coronavirus, some evidence shows that IgM and IgA are produced first, then followed more slowly by IgG.
Over time, as someone recovers from the virus, the antibodies “win” and the virus is defeated.
When it is defeated, the virus itself is eliminated from the body. This means you no longer need PCR tests to detect viral genetic material. But, the antibodies stick around (to fight another day!), hanging out in your blood. This is important (for you) because this is how your body has immunity (see immunity explainer here). The upshot is that we can figure out if someone was infected by looking for these antibodies in their blood.
The first step in any serology test is to get a blood sample. The simplest form of these tests are called “lateral flow assays”. These work kind of similar similarly to a pregnancy test, but with blood, instead of urine. In this kind of test, you take just a finger prick of blood. The blood flows over an absorbent pad, and regions of the pad are designed to catch and hold antibodies that are associated with the coronavirus.
Check out the picture below! Just like with a pregnancy test there is a “control” line, and then antibody lines. We’ve visualized a test that would show you the presence of both IgM and IgG antibodies. Some only detect one or the other. Since IgM shows up first in the body’s response, this would be the most common place to start.
As the blood flows over the pad, any SARS-CoV-2 antibodies present will stick to the pad. Everything else is flushed away. In the end, you can look for the region that captures the antibodies to be colored or not. And there you go. These tests are simple to run and do not require a lab. They take 15 minutes. Theoretically, they can be done at home although no at-home testing has yet been approved. So, that’s good for rapid, on the spot testing (like the population-based tests done in NY).
However: these tests are not always so accurate and they are limited in what they can do even if they are accurate. They can detect only the presence of some antibodies, not how many or which type. A positive result doesn’t tell you everything about immunity. These tests have fairly high false-positive rates. Such false-positive rates are a problem for using these tests to (say) let people back to work. In addition, in cases where only a small share of the population is infected, even a small false-positive rate can really mess up your understanding of population-level infection. See our explainer on false positives here!
These tests are convenient, but not ideal to really test for immunity. To do better, and to get closer to figuring out if people are immune, we need to test quantitatively for antibodies. We can do this using a technique called an enzyme-linked immunosorbent assay or ELISA.
The ELISA test requires more blood, although not always much more (some labs are doing it with a fingerstick). It also requires a lab for processing and trained scientists to run the samples. Bu it can tell us a lot more.
In this test, blood is sent to the lab. There, the blood sample is added onto a tray that contains several tiny wells covered with proteins specific to the virus that these antibodies can stick to. The wells are then washed and the antibodies stay stuck while other stuff is washed away. Following this washing chemical and biological reagents that can uniquely stick to each of the various types of antibodies are added into the wells. The result of this is that each antibody is marked by a different reagent. Different from the quick serology test, an individual ELISA must be performed to detect each one of the five classes of “immunoglobulins” in each sample (IgA, IgD, IgE, IgG, and IgM).
Each well has a chemical reagent added to it. Based on the color change of the dilutions of blood in the wells you can observe the presence of antibodies and the quantity of them. So, for example, say there is a well where the reagent targets IgA. Based on the color change in that well , you can figure out whether the person has IgA antibodies, and how much they have. Like PCR, the in-lab time for this testing is a few hours , although transporting the samples themselves takes time.
The result of all this is that the ELISA test can determine not just the presence but the quantity of each antibody. This is huge, since it can tell us which antibodies people have and how much, which tells us a lot more about whether they are immune.
So, if I have antibodies am I immune? As noted above, the presence of any antibodies at all does not guarantee immunity. On the other side, most people are reasonably confident that some level of antibodies (specific to the coronavirus) is sufficient to give immunity for some time. An area of active research is how high these antibody levels need to be and which antibodies could be a better asset to have in the fight against the virus in case we get infected again in the future. Once we settle that, then the results of the ELISA test can better answer this question.
Blood sucks. Any chance for saliva here? At the moment, no. Generally, ELISA tests do use blood. Saliva is not out of the question in the longer term, but likely not soon.
When can serology tests pick up infections? In principle as soon as you start making antibodies, which is shortly into a viral course. But the accuracy of this test increases after symptom onset as your immune response kicks in and your body makes more and more antibodies. This means that serology could be used for diagnosis, although it would be a late diagnosis.