Coronavirus: New antibody test ‘with 99 per cent accuracy’ approved for use across Europe

This is good news, especially for the researchers. The sample size mentioned in the article sounds a little low to be confident of that 99% accuracy, but I am sure those who approved it have looked into it more thoroughly than I can.

There is a paragraph "Dr Clarke added the test was useful to authorities because it showed “how the virus has moved through the population” but added “for the general public it of no immediate utility.” I would echo that on what we know so far, because most people do not understand the relationship between test accuracy and the number of false results it gives. We do not know how widespread the virus is in the population, but if 5% of the population have it, a back of the envelope calculation would be that the test gives an incorrect result about 19% of the time. As we don't really know what the percentage of infection is, I have made an estimate based on the number of tests and the number of those that tested positive, but it is very rough and ready, because most asymptomatic people would not get tested, meaning the actual percentage who have had the virus could be a lot higher.

If we were getting the 20% false results, those medics involved in serious testing will understand that and treat it accordingly, but I am far from convinced the general public will.