>the death rate will surge from 0.5-4% (no exact figure at the moment) to 10-20% like earlier in Wuhan

Those percentages come from different samples altogether and cannot be compared. 10-20% is heavily biased toward people who were admitted before there were even reliable tests for the virus or a recognition that it's novel and serious. Now you're comparing that to 0.5-4% which includes a bunch (though by no means close to all) of mild cases who got picked up by the tests but may not have required a hospital visit at all. What's the point of mentioning two numbers that aren't just unreliable and biased, but unreliable and biased in different ways?