This is a potentially huge problem. I have not yet confirmed the data (we're digging) but given the impacts if true it's worth a quick discussion on testing, reopening and algebra. Lives could be in the balance. Thread: theatlantic.com/health/archive…
1/ Lots of states are evaluating whether to re-open based in part on the % of positive test IDs. To the extent those tests measure the % of people with positive viral loading, a falling % is a good thing.
2/ A falling positive viral test % ID is also a good thing for public health to the extent it means we have expanded testing capacity to the asymptomatic population and can monitor hotspots (rather than just testing folks who are already sick, and more likely to be positive)
3/ In recent weeks, we have indeed seen a falling positive ID rate. Still above the rest of the world, but moving in the right direction: 
4/ We've also seen a significant increase in testing, which would appear to be consistent with our ability to test / isolate / treat people who are positive and to test asymptomatic people for community-based screening. *In theory.*
5/ But if this Atlantic piece is correct that the increased testing includes a meaningful increase in antibody rather than viral testing, all of those conclusions are backwards. theatlantic.com/health/archive…
6/ So now for (simple) algebra. Assume that most of the population has not yet been exposed to COVID. That means most people will test negative for COVID antibodies and are NOT protected if the economy reopens and they are exposed.
7/ By contrast, a falling postive ID rate for viral testing means the rate of infection is falling, all else equal. SO... by adding antibody test results into the positive ID results we are mixing apples:oranges, in potentially deadly ways.
8/ As the rate of positive viral IDs falls, it is gradually safer to reopen. On the other hand, if we are aware that a rising % of the population does not have antibody protection, it is LESS safe to reopen. A + result from one test is the opposite of a + result from the other.
9/ If the CDC has in fact recently started adding antibody tests into viral tests as suggested by the Atlantic article, it would lead to an increase in total reported tests and a falling total positive ID rate exactly as we have seen... but for the wrong reasons.
10/ And if states/cities use this data to accelerate re-opening they are going to put people's lives at risk. Not with intent, but because they got bad data. THIS IS A POTENTIALLY HUGE, REAL-TIME PROBLEM.
11/ To be clear, we want antibody testing. We want viral testing. And we want the number, location and results of both posted publicly to inform public health decisions. But we want them reported separately, not conflated as this article suggests is being done.
12/ So if you are a Governor, Mayor, business owner, public health director or anyone else with authority to re-open some portion of the economy, INSIST on separating that data. And be hyper cautious about using positive ID rates to inform your decision until this is clarified.
13/ A final point: I hope the Atlantic article is wrong. But in this moment we are better erring on the side of caution. Be safe everyone. /fin
