Consider adding options for rapid tests

[winternewt]

Suggestion is to add selectors or drop-down menus for "Nearby people" in case activity attendance assumes obligatory checks for all participants. Those selectors are meant to apply multiplicators

1. For vaccination status, i.e if only vaccinated are allowed on premise (already proposed in in #853 )
2. For rapid tests: 1x for "no test" vs average false-negative rate for express tests: https://www.cochrane.org/CD013705/INFECTN_how-accurate-are-rapid-tests-diagnosing-covid-19

[beshaya]

We already have an option for specifying that people are vaccinated (only available in regions where we have local vaccine data, and only for "random person" profile).

For rapid tests, 42% FNR vs asymptomatic is... a lot. Add that to the fact that timing is really important and you really have to do the test at the door to get any signal out.

[neelnanda-io]

I think FNR of 42% significantly under-rates lateral flow tests. The thing I care about is not being infected, infectiousness increases with viral load, and lateral flow tests are way more accurate with high viral load. I haven't found any good studies looking at this directly, but based on studies that indirectly look at this my best guess is something like a 5x risk reduction in infection.

[jonasvollmer]

I agree that LFTs seem pretty helpful. False negative rates are high, but taking into account infectiousness, the tests plausibly lower the risk ~3x or so during the ~4–8 hours after taking the test.

Some discussion and relevant sources: https://forum.effectivealtruism.org/posts/YeCeBfNSfkY2kHKSb/coronavirus-is-it-a-good-idea-to-meet-people-indoors-if

[beshaya]

I haven't found any good studies looking at this directly, but based on studies that indirectly look at this my best guess is something like a 5x risk reduction in infection.

Link?

Here's the ladder of studies that I use that makes me wary of these tests:

• Binax NOW Certification:

Compared with real-time RT-PCR testing, the BinaxNOW antigen test had a sensitivity of 64.2% for specimens from symptomatic persons and 35.8% for specimens from asymptomatic persons, with near 100% specificity in specimens from both groups.

• https://www.webmd.com/lung/coronavirus-incubation-period#1 (could probably dig up an actual study but lazy rn)

Researchers estimate that people who get infected with the coronavirus can spread it to others 2 to 3 days before symptoms start and are most contagious 1 to 2 days before they feel sick.

So the rapid tests are inaccurate before symptoms start and people are most contagious before symptoms start. I acknowledge that this does not rule out the following situation:

• The LFT is actually accurate in the 1-2 days before people get sick.
• The low reported sensitivity prior to symptoms occurs because tests are taken 2-4 days before symptoms.

That said, this seems unlikely to me because the sensitivity is relative to PCR tests, which themselves suffer from low accuracy prior to symptom start. So if PCR tests are mediocre at detecting pre-symptomatic COVID, which includes the peak contagiousness window, and LFT's are worse at detecting pre-symptomatic COVID, what does that say about LFT's?

This also violates my intuition. But given my background (not a respiratory virologist), there's no particular reason for the virus to obey my intuition. Perhaps virus in mucus lags virus in aerosols.

I'd love actual data on transmission dynamics after tests.

[beshaya]

Noodling on it more, the LFT's 44% asymptomatic false negative rate is vs RT-PCR, which means the person is already shedding detectable amounts of virus. If the person is just starting to shed virus, we should expect them to be maximally contagious within 12-24 hours. I'd finger-in-the-wind guess that a negative LFT suggests reduced risk by 40% for the next 24 hours, perhaps more than that for the next 2 hours. 4-8 seems like a gamble.

This is a completely different sort of false negative rate than RT-PCR's list, which is getting a negative test result followed by a positive test result a few days later, in which case "the person wasn't shedding virus yet" is a plausible explanation.

There are rapid tests on the market that seem likely to have better performance at low viral loads - Lucira and Cue tests are both RT-LAMP tests, which is an alternative way of amplifying viral RNA. Ironically, neither was tested thoroughly against asymptomatic individuals.

However, these tests do provide LoD in terms of viral particles per mL. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7302192/ suggests: Abbot M2000 detects down to 1e2 genome copies / ml (cp/ml) CDC's test and GenMark EUA detects down to 1e3 cp/ml Labcorp EUA detects 2e4 cp/ml Sofia 2 AG test EUA detects 2e7 (this is an antigen test, not PCR) https://www.fda.gov/media/143808/download suggests that Lucira's rapid tests detects down to 9e2 cp/ml https://www.fda.gov/media/146470/download suggests that Cue's rapid test detects 1.3e3 cp/ml https://journals.asm.org/doi/10.1128/JCM.02880-20 suggests that the Binax Now rapid test detects 8e5 cp/ml

[beshaya]

https://www.fda.gov/media/138096/download limit of detection for the Aptima® SARS-CoV-2 Assay is 16 cp/ml https://www.fda.gov/media/136049/download same for Roche Cobas.

These two seem to be the gold standard for clinical trials for new tests (Lucira tested against Roche, Cue was tested against Aptima).

[beshaya]

https://journals.asm.org/doi/10.1128/JCM.02880-20 Further mentions that Binax Now's test is equivalent to RT pcr with a CT of 30 and live virus can be cultured down to a CT of 34. My understanding is that each Ct corresponds to a factor of 2 of RNA, so you can test negative on Binax Now while having 16x more viral load in your nose than necessary to culture the virus.

Which to me is pretty compelling evidence that you can be mildly contagious while testing negative on an LFT, which in turn means you could be pretty contagious within a few hours.

[blanchardjeremy]

Just got another email asking about how home tests effects their risk profile. (They're using BinaxNOW)

[neurodynamic]

@beshaya Your reasoning sounds good to me, but even if at-the-door testing only lowers risk by about half (per the microcovid blog post about rapid testing), I'd still like to be able to include that in risk calculations.