We think that rapid tests can reduce risk a lot, but only if you're careful about what you infer from them. A negative test, properly collected*, can give a lot of confidence that you're not infectious right now.
* including a throat swab, thanks to Omicron
Conversation
Everything ends up hinging on the test sensitivity (false negative rate). We were extremely confused about how to reconcile low headline sensitivity with really high sensitivity in various lab studies.
1
2
Once we thought through a minimal-but-physically plausible model for transmission, we figured it out!
The key ingredient was a minimal-but-physically plausible model for transmission:
1
2
This model resolves the confusion by breaking out test sensitivity by viral load (amount of virus).
We think that high viral loads are both (1) responsible for almost all transmission and (2) very likely to be detected by an antigen test.
1
2
The infuriating thing is that aggregate sensitivity numbers from clinical trials seems to be basically a property of the trial design.
1
2
Sensitivity is usually measured with respect to PCR tests, which remain positive long after the window of infectiousness is over. So, getting a test approved at all seems to be largely about making sure that only sufficiently infectious people show up in your trial. 🤦♂️
1
2
This isn't an original observation. Scientists like Michael Mina have been beating this drum for over a year:
Quote Tweet
2) FDA requires Rapid Ag tests compared to entirely different technology, w entirely different analyte that diminishes in a very different time scale.
This causes MASSIVE slow downs and hurdles with clinical trials for EUA
Compare rapid tests to a gold standard rapid test!!
3/
Show this thread
1
3
Some key takeaways:
- Rapid tests can be really useful.
- Combined nose and throat swabs are probably important for detecting Omicron.
- Negative rapid tests "expire" very quickly. Take them right before you need them!
1
2
And an important caveat:
Quote Tweet
We use a model where transmission rates are linear in viral load. There are lots of physical reasons to think this is about right, and we go through the arguments. But this is central to the whole analysis, so if we've got that wrong ~all of our conclusions fall too.
2/6
Show this thread
1
1
I'm still pretty confused about this stuff, but I feel like I understand enough of it that I know what to do, practically speaking.
More importantly, I know where to "plug in" new facts as they appear. :-)
2
3
I'm less and less worried about Omicron for me, personally. But, I'm glad to know what to do to reduce the chance that high-risk family and friends get sick at a time when help is limited.
Replying to
Second this! Amazing paper!
Quote Tweet
Oh, and I'd like to highlight this awesome paper on droplet production and dynamics by Pöhlker+2021 (arxiv.org/abs/2103.01188). @nschiefer
Show this thread
1