[filleduchaos]

If you're not testing a sizeable enough sample of the population, then whatever number of cases you give has to be taken with a rather large helping of salt.

For example, my country's CDC has the official number of confirmed cases in the low hundreds with a dozen deaths, but anyone wise enough knows that between the ridiculously low number of samples tested (less than ten thousand in a population of a couple hundred million), the poor public health infrastructure, the anaemic government response and general public health illiteracy, the real number of cases is easily an order of magnitude or two higher.

[SpicyLemonZest]

Completely agreed. Sweden is worth observing in terms of number of deaths and how overloaded their hospitals get. But their number of cases, like in a lot of countries, is essentially garbage data that can't possibly help our understanding of the situation to consume.

[seppin]

Right, but in planning for a hospital overrun you can't see it coming if you have no idea how many sick you have, if you don't test.

Are you suggesting that as long as your hospitals aren't overrun, inflection rates aren't worth knowing ?

[SpicyLemonZest]

If your hospitals aren't overrun, and you don't expect them to become overrun in the future, improving your estimate of infection rates doesn't seem like a high-leverage activity. The argument for doing a lot of testing is that it enables good contact tracing, which Sweden doesn't seem to be planning to do anymore.