COVID-19 Testing

Notes on the accuracy of various approaches to COVID-19 testing.

Overview

It's customary to distinguish two aspects of a COVID test's performance:

its sensitivity: how frequently it correctly detects the presence of the virus (test is 'positive' when virus is present)
its specificity: how frequently it correctly detects the absence of the virus (test is 'negative' when the virus is absent)

The relative importance of sensitivity and specificity varies depending on the application and the audience for the test:

From a public health or practitioner standpoint, it's desirable to have a test that is both highly sensitive and highly specific. Low sensitivity ('frequent false negatives') means one misses many cases where the virus is present, leading to increased transmission risk; low specificity ('many false positives') means one incorrectly identifies individuals as positive, leading to unnecessary burden on the healthcare system, unnecessary absences from school or work, etc.
From a individual risk standpoint, a test which is highly specific test may still be very useful, even if it isn't very sensitive. I.e.: if a negative test is very likely to indicate a true 'negative', one can simply postpone a social engagement on this basis, and test again when convenient.

The most accurate overall testing technology currently is 'molecular'. 'PCR' is a type of molecular test, as is 'LAMP'. (For more details on this, see https://www.technologyreview.com/2021/05/04/1024450/at-home-covid-test-review-accuracy-binaxnow-lucira-ellume/.

LAMP tests can be performed at home, and seem to be as specific as PCR. See: https://www.southampton.ac.uk/news/2022/01/lamp-testing-accuracy.page

Also see: https://www.theverge.com/2022/6/3/23141813/molecular-covid-19-test-otc-pcr-fda -- which references Lucira as an 'OTC' test (which simply means 'over the counter').

Also: some info from Lucira on the Positive Percent Agreement (PPA): https://www.fda.gov/media/143809/download