Those numbers aren't terribly useful since so many cases are mild or even asymptomatic. Most organizations are estimating 2-3% mortality, and are quick to point out that this is probably an overestimation due to the lack of testing in asymptomatic individuals.
Basically we are counting every case severe enough to cause death, but only some of the cases that aren't very severe, and that latter group is already a much bigger piece of the pie even among confirmed cases.
