There are three types of coronavirus tests, and they each test different things in different ways with different levels of effectiveness. What you need to know about the three types of coronavirus tests now…

RT-PCR tests, known as “nasal swab” tests, search for coronavirus genetic material in cells sampled from the surface of the nasal lining. RT-PCR is the gold standard of coronavirus tests—sensitive enough to identify infection even if the virus is present in only very small amounts, and it is extremely unlikely to produce an inaccurate result. It often takes several days to get results, however. Sometimes called a molecular test, nucleic acid amplification test (NAAT) or PCR test, this test is most accurate when the sample is taken from way up in the back of the nose. The most common causes of false-positives include contamination of a negative sample with a positive one or mislabeling (i.e., your name goes incorrectly onto someone else’s sample).

Warning: While there are RT-PCR tests that use saliva samples, saliva has not yet proved to be an adequate replacement for nasal-swab testing.

Also: A few companies have created tests meant to provide the accuracy of RT-PCR but with fast results—in under an hour. Cepheid’s Xpert Xpress test is fast and accurate but requires an expensive piece of equipment found only in large hospitals. Abbott’s ID Now test also is fast (it uses a different technology from PCR tests called LAMP), but it’s far less reliable than RT-PCR tests. 

New: Lucira’s COVID-19 All-In-One Test Kit is a recently approved at-home kit that uses self-collected nasal swabs and will be available with a prescription for around $50. The manufacturer claims it will match lab PCR test results 94% of the time for positive results and 98% of the time for negative results. You’ll have your results in 30 minutes. However, no test ever works as well in the field as it does in the documentation for approval, since the company’s tests are always done in perfect conditions. This test will be used by people without any training, so it will assuredly perform worse than promised. It would be most accurate in people with symptoms and less accurate in those who are asymptomatic.

Antigen tests (also a nasal swab) examine cells for proteins associated with coronavirus. They tend to be faster, simpler and less expensive than RT-PCR tests. Antigen coronavirus tests even might become available at some point for at-home use. 

Drawback: They are far less sensitive than RT-PCR tests, which amplify the signal from the virus, while antigen tests just detect the amount that is present in the sample. Studies suggest that antigen tests miss 10% to 30% of infected-and-symptomatic people and probably a lot more asymptomatic people. They also have false-positive rates of 1% to 2%, which might not sound high, but it’s enough to make these tests a poor choice for screening asymptomatic people. A 1% false-­positive rate means that 10 people in 1,000 will be told that they are infected even though they’re not. If this test were used to decide who goes into the COVID ward of a nursing home, many healthy people could be put at risk. 

Because of the high error rate, do not put complete faith in the results of a single antigen test’s ability to identify a recent infection. Retesting with another antigen test may not be a good idea, as the source of a false-­positive may do the same thing again. Instead, get retested by RT-PCR as soon as possible. Antigen tests are best used in symptomatic people to confirm the diagnosis. 

Antibody tests examine blood samples to see if the body has had an immune response to the virus. These are very good at determining if you previously had coronavirus but not at telling if you currently have it. The body’s immune response to the virus is so modest in the first few days following infection that this test usually fails to produce a positive response during this time. Even a full week after ­infection, there’s only a 50/50 chance that the test will come back positive. 

Accuracy of an antibody test approaches 100% if administered about three weeks after infection. Very mild or asymptomatic infections may have a lower rate of producing antibodies, and after a few months, those antibodies could be gone. However, it appears that people who have symptomatic ­infections generally have measurable antibodies to the virus for many months at least. We don’t know about greater time scales because the coronavirus hasn’t been around long enough.

Right now, the need to make a diagnosis retrospectively is not terribly clinically important, so it is not being done much. I suspect antibody testing will see far more use in trying to care for patients after the vaccine is available—i.e., to see if antibodies are being made and if the patient is likely to be protected. 

Final Considerations

If you go to a gathering that offers or requires testing for admission, be aware that tests provide only marginal safety over masks, distancing and avoiding crowds. You cannot have a gathering and assume it is safe because guests were tested ahead of time. 

Also understand that labs are incredibly stressed meeting the demand now. Using testing for personal convenience reasons can deprive those who need to know whether they are infected.