Here's a detailed critique of this paper: https://twitter.com/AtomsksSanakan/status/137594206799050342...

Basically it sounds like Ioannidis extrapolated from a lot of statistically non-representative seroprevalence samples to make his case.