Update 27May2020: the CDC has issued new guidelines for the use of antibody tests. It begins with the admonition that “Data that will inform serologic testing guidance is rapidly evolving”

There are now a number of different kinds of tests for COVID-19. If you want to be tested, you should be aware of the strengths and limitations of the various testing methodologies.


Most of the testing done so far has been using polymerase chain reaction, or PCR testing. PCR detects fragments of the virus’s genetic material (RNA in the case of SARS-CoV-2, the virus responsible for COVID-19). Because PCR can detect very small fragments of RNA, it is extremely sensitive. No new technology had to be developed for PCR; everything was already in place the moment that the SARS-CoV-2 genome was sequenced.

Strengths: It is widely available and required no new equipment or procedures (delays in making widespread testing available were leadership failures, not limits imposed by the science or technology). PCR testing is quite sensitive, though it can miss early infections (see below). It is also quite specific; if it’s positive, there indeed was viral RNA present.

Limitations: PCR requires a real lab. It’s not all that complicated (shockingly simple, in fact, considering what it does) but because it’s so sensitive, it has to be performed under strictly controlled conditions because an infinitesimal amount of contamination with viral RNA can turn a whole batch of negative samples into falsely positive ones. There will never be bedside or at-home PCR tests. PCR is often misued or misinterpreted, too. When you see headlines scream “CORONAVIRUS DETECTED ON DOORKNOBS 30 DAYS AFTER EXPOSURE!!” you need to see if they used PCR. Finding a tiny RNA fragment is not the same as finding the virus. By itself, SARS-CoV-2 RNA can’t infect you. At the other end of the scale, PCR won’t detect virus immediately after exposure, and can have a significant false negative rate even after symptoms appear, so just getting a negative test doesn’t guarantee that you can’t infect someone else.

cell culture

If you really want to know if you have viable virus, you have to inoculate some human cells and see if the virus can infect them. (If you wonder where we get those human cells and are willing to invest in a book-length read, I recommend reading The Immortal Life of Henrietta Lacks by Rebecca Skloot, the fascinating and well-written story of the woman whose cancer cells revolutionized cellular medicine.) If someone tries to tell you that your Amazon packages could be infectious a month after they’ve been delivered, make sure that the source of their information used cell culture, and not PCR, to assess for viable virus.

Strengths: Certainty that the sample contained actual, infectious virus.

Limitations: Difficult to obtain; I don’t believe it’s available as a routine clinical test. Culturing is a manual, labor-intensive process. Takes significant time (days) to get results.


Antibody tests (also known as serological tests) look for the immune globulins (antibodies) that the immune system of an infected person produces to fight an infection. The FDA has not approved antibody tests, they have simply authorized some of them for emergency use in order to make them available quickly. Immediately, I started receiving spam emails offering to sell me bedside antibody tests, most of which weren’t even FDA authorized. Unfortunately, we still know far too little to say if antibody tests are of any use; guidelines on the use of antibody testing are rapidly evolving. We don’t have good data on which patients produce which antibodies and when: many people with confirmed infection and recovery have failed to produce antibodies that are detected by the tests. It’s unlikely that these patients didn’t produce antibodies at all, and more likely that the tests just aren’t adequate. We know exactly what viral RNA looks like to make PCR tests, but we have very little idea exactly which antibodies any given person might produce in response to any given variant of SARS-CoV-2.

Further, people seem to want antibody testing “so I know if I’m immune.” Antibody tests don’t currently indicate immunity! That thinking is downright dangerous until we get a lot more knowledge of how our immune system responses are related to COVID-19 immunity.

Strengths: cheap, easy-to-use bedside and clinical laboratory tests are available. Can detect if you’ve been exposed even after you have recovered from an infection (or even if you never had symptoms). We don’t yet know, however, how long someone will remain antibody positive after exposure, though it’s likely to be months or years.

Limitations: Antibodies might not appear in detectable amounts for weeks or months after an acute infection, so a negative antibody test means very little. We have no idea if the presence of antibodies confers any kind of immunity (and, no, it’s not at all reasonable to assume that it does; we all have lots of coronavirus antibodies floating around in our systems yet they don’t confer immunity to SARS-CoV-2). So a positive antibody test only means that you were exposed (if the test is valid, and we have little data to support the validity of the current tests). It doesn’t mean you’re immune. It doesn’t mean you’re not currently contagious. So even a positive antibody test is currently of limited value.


Antigen testing looks for proteins produced by the virus. An antigen test (the first) was just authorized (not approved) by the FDA last Friday (08May2020). Antigen tests can be much faster and cheaper than PCR testing, but still verify actual infection like PCR. While I’m a little skeptical of this test just because it is so new, antigen testing will likely overtake PCR as our main way of testing for COVID-19.

Strengths: cheap, easy-to-use test that can definitively detect infection.

Limitations: Not as sensitive as PCR (will miss a small number of actual infections). There will likely be a somewhat higher latency between the time of infection and detectable antigens (a few days, I’m guessing; I haven’t seen the clinical trial results yet).