A couple of days ago, our attention was drawn to the problem of new positive tests for corona virus SARS COV 2 in South Korea (whose response as a country has been exemplary)
A lot of the problem lies in the accuracy of the antibody test. The general consensus is that most tests are about 90% accurate. Sounds pretty good, huh? Sadly, no. Will Bedingfield of Wired gives a masterly insight into why this is so, with the help of Professor Robert West of UCL.
I post the link in true humility below, but here is a scenario they offer
Assume you have a test that picks up 95% positive. And 95% negative ( Kh-And that ain’t bad, folks!)
Assume you have a population of 1000, of whom 5% get Covid 19.
The test which should pick up 50 positives, picks up 48 (Will points out you can’t have 47.5 people)
Two people are told-“you ain’t had it, my old son!”
950 people are really antibody negative. So 903 (or 902 , see above) get a correct negative test
that leaves 47 who receive a positive test when they are really negative
This is a big, hole. But heartfelt thanks to Will Bedingfield and Wired for giving an answer to a question that’s been really bugging me. Great stuff.
https://www.wired.co.uk/article/coronavirus-antibody-tests-uk-accuracy
#WillBedingfield #Wired #Coronavirustests #SARS-Cov-2tests #falsepositive #specificity#sensitivity
Learning, Science and Society 