Just some back of the envelope math for y'all on TEST SENSITIVITY for PCR and covid "rule out". This is important when deciding whether or not PPE are needed for working with a patient, or whether or not an exposed HCW can be cleared to work.

As someone who has always thought PCR fell down on the side of specificity (probability of a NEGATIVE test in someone who DOESN'T have the disease), the SENSITIVITY (probability of a POSITIVE test in someone who DOES have the disease) is less good than one might have hoped.

I don't know why. I'm not a microbiologist or virologist. Perhaps it's because we are sampling the "nasopharynx"* while this is often a lung infection (and the higher sensitivity in lung specimens would support this).

Nasopharynx: the part of your body the milk passes through if someone makes you laugh while drinking

Sensitivity of the test varies by day of infection...the sweet spot seems to be within a few days of developing symptoms (which usually onset 5-6 days after infection). Sensitivity then drops off again.

In studies like this one, in pretty sick people, all of whom had COVID-19 in China, nasal and pharyngeal specimens were not especially sensitive. That may reflect disease state in the population under study. https://jamanetwork.com/journals/jama/fullarticle/2762997

But let's be optimistic and say sensitivity is 75%.
That means our "rule out" test is going to be wrong 1/4 times in people who DO have COVID-19.

How can we improve that? It's pretty simple, you test again (assuming that the negatives and positives are "independent")

and aren't driven by patient-specific features. Which they could be. But let's keep this simple.

In our first sample of infected people 75% tested positive first time around. Let's re-test the 25% who tested negative again in 24 hours. The same sensitivity applies.

75% x 25% = 18.75%. That's the fraction of the total population that tested negative on the first test, but now "converted" to positive on the second. We think they have disease. Add 18.75% + 75% and our new test sensitivity (for 2 tests) is 93.75%.

So now instead of 1/4 tests being falsely negative, 1/16 are falsely negative. A big improvement from that second test.

Note that I'm assuming a perfectly specific test too, which is likely wrong.

This may be helpful for those brave IPAC folks who actually choose to test, knowing they may find individuals with disease (you know who you are/aren't!).

But it also speaks to the Korean reports of reinfection. We do need to know more about the nature and timing...

of testing in these individuals to know that we aren't simply seeing false negative test results superseded by true positives as a result of a coin toss. I don't know and haven't seen data that would allow one to decide what's at play.

here endeth ye tweetorial.

(I know I'm delinquent on the SMR tweetorial...it's coming).