May 13, 2020 Update - Testing

Dear All,

Today I wanted to give an update on testing with a focus on reviewing what tests are available and how to interpret them as I suspect several of these tests will be available in your communities in the near future.


We’re making strides nationally from a testing perspective. The big news on the testing front is the potential of using saliva instead of a nasopharyngeal swab to look for virus, the increasing availability of antibody tests, and the release of a new antigen test in addition to the existing PCR test that we’ve been using to this point. To quickly recap these 3 forms of testing and how they’re used:

At the highest level, PCR and antigen testing look for active infection. Both of these are most commonly done via nasopharyngeal or nasal swab and are asking the question: Is this person actively infected with COVID-19. There are a few distinctions between these tests:

PCR Tests – These are the gold-standard for confirming someone is actively infected with COVID-19. It is the most common test out there and if you’ve been tested thus far, it was almost certainly a PCR test. Reported accuracy of these tests has varied widely, and we know it isn’t perfect. I’ve seen false-negative rates range from 5-20% depending on the lab and who collected the specimen, but the bottom line is it will miss sometimes. We know it isn’t perfect by seeing the recent positive infections in the White House inner circle. Despite using best-in-class PCR testing, it is likely that someone (either a staffer or guest) tested negative who was actually positive before entering the White House. Illustrates the point that no test will be perfect and always wise to take basic precautions (masks, etc) indoors no matter if testing is available.

The big development on the PCR test is the approval of using saliva to look for virus as opposed to using a nasopharyngeal swab. A lab in New Jersey recently received FDA approval for an at-home COVID-19 saliva test where a patient can collect their own saliva sample, ship it back to the lab, and have it tested to look for COVID-19. The lab can currently run 10,000 samples per day and anticipates ramping up to 40,000 daily in the near future.

Is saliva as reliable as a nasopharyngeal swab?

Hard to say since it’s so new but the initial data looks like they’re close to equivalent. Per this article in Modern Healthcare, the New Jersey lab found 100% concordance among a sample of 60 patents with a positive nasopharyngeal swab. The article mentions that Yale has also been studying saliva and has seen encouraging results.

 Clearly, this is subject to change with new data, but if saliva can be used routinely instead of nasopharyngeal swabs, this would enable people to collect their own specimen which would drastically reduce the number of healthcare workers needed to obtain samples, and therefore, reduce PPE needs too. Plus, these could be sent to patients at their home and is far less painful than having the nasopharyngeal swab done (trust me on this). Still a big if on whether saliva will prove to be as reliable over time as swabs but I’m cautiously optimistic.

Antigen tests - There is now an antigen test approved by the FDA that gives us another way to screen for active infection besides PCR. Compared to PCR testing, this test will be cheaper to manufacture and faster to get results. It works the same way as when you get a rapid flu test or rapid strep test in your doctor’s office today. Similar to the strep test, when the result is positive, it is highly probable (~99%)  you have COVID-19. Unfortunately, when the test comes back negative, there’s an ~85% chance you’re actually negative for the virus. This high false-negative rate means that relying solely on this test would be risky. Thus, what will likely happen is that those with a negative rapid antigen test will have their sample sent for PCR testing to confirm you are truly negative.  This is the same process we do today for rapid strep tests where it’s possible your rapid test in the office comes back negative but you’re called 1-2 days later that your confirmatory test is, in fact, positive.

Why use these antigen tests at all if they’re so unreliable?

It’s better than nothing (and PCR isn’t perfect either). They come back really fast, are cheaper, and more scalable than PCR testing. The value of this test is that when someone tests positive, you can say with a high degree of certainty they have COVID-19 and should isolate. One could imagine a rapid test such as this being very helpful to rapidly screen people entering a nursing home, hospital, or other crowded indoor space (like a meatpacking plant). A positive test would keep actively infected folks from entering those spaces and a negative test would be no different than if no test were done at all. Everyone entering with a negative test should continue to wear masks, keep distance, etc, since a negative test could be wrong, but it still may keep some infected folks from entering high-risk areas.

The other reason is that we’re still very far behind in testing in this country. Reports estimate we’re running 300,000 tests per day when we’d probably want to be running 3-4 million per day when we’re up to scale. Thus, we’ll need all forms of testing to reach that type of capacity.

Antibody Testing – This looks for prior infection as opposed to active infection. Availability and accuracy of these tests seem to be improving by the week. The key test characteristic for antibody tests is the false-positive rate. This means that your test comes back positive suggesting you have antibodies when, in fact, you don’t. This can be a major problem by falsely reassuring someone they’ve already had COVID-19 when they really haven’t. Scott Gottlieb suggests that if your first test is positive to get a repeat antibody test as ‘the predictive value of two consecutive positive tests is high enough that you can be confident antibodies are present.’

 I’ve spoken with friends in multiple cities who have been able to get antibody-tested relatively easy. I would imagine other cities will have increasing capabilities to do antibody testing in the next month as well.

The challenge is that most people will probably test negative. You’ll recall from the graph yesterday that in New York City, only ~20% of the population has antibodies with the seroprevalence being closer to ~14% statewide. It is likely that in areas where there’s been far less infection, the seroprevalence will be 5% or less.

But, if you think you were exposed to the virus or had COVID-19 to date, it may be worth getting tested for antibodies when testing becomes available. You should wait at least 3 weeks after your symptoms started to get tested for antibodies. Testing for antibodies any sooner than this will be of limited utility.

While we still do not have concrete evidence that the presence of antibodies prevents a repeat infection, I am increasingly optimistic that we’ll find that those with antibodies are at far less risk of infection as compared to those without antibodies. And, it’s not just me who is optimistic. This article from the NY Times is a nice summary of where the latest research stands with antibody research. Key takeaways from it:

  • Antibody test now available with <1% false positive rate.

  • Almost everyone who gets COVID-19 (even those with mild illness) are making antibodies.

  • Those who make antibodies (even small amounts) seem to have enough antibody protection to neutralize the virus.

I wouldn’t call any of the above bulletproof evidence and is subject to change with new data, but I would say it’s reason for optimism that those with antibodies will have some protection against catching COVID-19 a 2nd time. What we don’t know is how long this protection will last.

To summarize:

We use PCR and antigen tests to look for active infection. A positive test is highly reliable that you are actively infected with COVID-19 and should isolate. A negative test isn’t as helpful since it could be a ‘false negative.’ PCR tests should be more reliable than antigen tests. You should be able to ask the lab that is running the test what their false negative rate is to get an idea of how reliable the test you’re getting is. But, even the best of tests will occasionally miss and you may test negative despite truly having the virus (hence always reasonable to take basic precautions indoors).

Antibody tests look for previous infection and takes at least 3 weeks after initial symptoms start to test positive. The key number to ask about when you get an antibody test is what the ‘false-positive rate’ is of the test. There are now tests with <1% false-positive rates but they may not be widely available yet. If you test positive, I would get re-tested to confirm a 2nd test is positive before feeling confident you have antibodies. And, even if you do have antibodies, I’d continue taking standard precautions until more is known.

I’d suspect you’ll be seeing varying forms of these tests become more readily available in healthcare facilities but also potentially office buildings, airports, or other indoor settings so staying informed as they come to market and being familiar with what type of test is being run and how to interpret may be of use to you going forward.

This is a rapidly evolving topic so be sure to stay informed as new information becomes available over the next month.

For those who want to read more, A nice summary from NPR can be found here

I will likely get to contact tracing over the weekend.

Until then be well and stay safe,


Click here to subscribe to future updates if you have not already.