No, it's worse than that. These studies - particularly the santa Clara study - are a hot pile of garbage. Don't kid yourself - this changes nothing and does not indicate, even a little bit, that things are better than they seem.
They recruited their participants from "targeted Facebook ads". This is not how you perform epidemiological surveillance. You get a recruitment bias - people who think they've had covid would have been more likely to sign up. It's also come out that the wife of the lead investigator of the santa Clara study emailed friends and family encouraging them to participate so they could get tested for their peace of mind. Totally ethically and methodologically unsound.
The false positive rate of these tests. They used lateral flow antibody tests. You may have heard in the news that initial excitement over these has dimmed as its become clear that they're unreliable. Nevertheless, instead of a more reliable ELISA test, that's what they use here, and they claim that they're 99% accurate. This is not in keeping with everybody else's observations of these tests. Additionally, their "80x higher infection" or whatever it was they claimed, is based on finding that 60 participants out of 3500 had a positive result. Hardly a huge number. More to the point, they actually acknowledge in the paper that if their tests were only 98% accurate, they'd have no interesting results at all.
There are good reasons epidemiologists are shitting all over these papers. I'm a biomed scientist, and my lab mates and I spent yesterday taking it apart in our group chat.
Do not fall into the trap of "some studies say x, other studies say y, therefore the truth is somewhere in the middle". These studies are worthless.
No comments:
Post a Comment