In reply to Postmanpat:
> I maybe being dumb but can you show me your arithmetic on that?
Define: IFR = (number of deaths) / (number of infections)
Rearrange: (number of infections) = (number of deaths) / IFR
Plug in numbers: (number of infections) = (38,489) / (0.0001) = 384,890,000
With one in ten thousand people dying from an infection, to get to ~ 40k deaths, ~400m infections would have to have occurred. It just doesn't stand up to any reasonable scrutiny.
If we assume everyone has been infected, a bound for the IFR can be estimated by dividing the deaths by the populations and this gives an IFR of ~ 0.0006, or 6x higher than the bound given by Gupta. As people are still dying we know that our estimate for the bound is low, and that the real number is higher.