Error rates and CoVID-19 antibody tests

rslade — Thu, 16 Apr 2020 01:07:22 GMT

We, in security, know that there are errors that are false positives, and errors that are false negatives, and that both can create problems.

At the moment, everybody is eagerly looking forward to serology tests for CoVID-19. These are tests (usually blood tests) that determine if you have antigens or antibodies related to defence against the SARS-CoV-2 virus.

At least, they try to determine that. Because, well, errors.

A good article on this is available at NPR. If you want the tl:dr version:

If the test has 99% specificity, and you live in an area where only 1% of the population is actually infected, then when you get a "positive" test, and are reassured that you are immune, you actually only have a 50/50 chance that you encountered the virus, and do have any defence. (In BC, where I live, the infection rate is about .03%, so the chance that a positive test is of any use at all is far worse.)

topic Error rates and CoVID-19 antibody tests in Tech Talk

Error rates and CoVID-19 antibody tests