The latest coronavirus antibody test is a lot more accurate (100% accurate at detecting antibodies)

[jberryhill]

10. Stated an odd way

"99.8% accurate at ruling out the presence of antibodies"

A "positive" is "the presence of antibodies".

A "negative" is "the absence of antibodies".

This is something of a conundrum, no?

What we want to know is "has this person been exposed to the virus". If it is 99.8% accurate at saying "this person has not been exposed to the virus" (i.e. negative), then I am confused by the definition of "positive".

Of those who have antibodies, the "100%" claim suggests that anyone with antibodies will correctly test positive with zero false positives.

Of those who do not have antibodies then the "99.8%" claim suggests that 99.8% will correctly test negative, and that .2% will falsely test negative.

Which, in Philadelphia, would mean that 10,000 people would be told they are at risk of infection, when they are not at substantial risk. (assuming we've nailed down, per other news, that the risk of reinfection is very low)

That is an err on the safe side outcome, but would suck to be one of those 10,000 people.