I'd take that with a bit of a grain of salt. Cross reactivity with other coronavirus antibodies means that COVID antibody tests can have a small false positive rate. On an individual level, it's tiny, so if you get a reliable test and it shows you being positive, chances are it's correct. But when you're testing 24,000 samples, even a test with a very high specificity rating will produce false positives.
So in this specific example, they used tests which are rated to be 99.5% and 99.7% specific, which would imply 70-120 false positives in a sample that size. To get around that limitation, they did sequential testing with two different antibody tests, so if a sample was positive in one of them, then they tested it on the other and if both showed a positive result then they counted it. Apparently they tested 1,000 known negative samples and none of them tested positive on both tests, so it seems like a pretty robust method.
Still though, they had 147 positives on one of the tests, and then 9 of those tested positive on the 2nd test, so I think there's still a chance of most or all of those being false positives. Particularly given that they have no evidence of travel history among those people or their close contacts and no real evidence of early outbreaks in the areas where these people lived.
|