When I assume a reasonable lag time between a case and death, the figures I am seeing (worldometers and elsewhere) seem to be about 5%. Then the difference bet

When I assume a reasonable lag time between a case and death, the figures I am seeing (worldometers and elsewhere) seem to be about 5%. Then the difference between 5% and <1% would presumably be all the mild cases that aren't diagnosed. If you look at the percentage of outcomes in the graph on worldometers' page, which I think you're getting ~20% from, it declined and then increased again; it is clearly not converging (yet) to the number people really want to know.