Death rate was under 1% outside of Wuhan and in South Korea. It's over 5% within Wuhan and in Italy.

The way I interpreted that discrepancy is that in the presence of full testing and good ICU care, the death rate from COVID-19 is about 0.7-1%. When the hospitals get overloaded, as they've been in Wuhan and Italy, the death rate skyrockets to 5-10%. This also squares roughly with the number of people who need ICU care in places where the death rate is low. If you need ICU care and don't get it, presumably you'll be dead.

Median death is 14 days after first symptoms with many cases taking significantly longer. That combined with an exponential growth curve looks significantly less deadly. A major discrepancy is the under detection of cases as full population testing has been rare outside of the cruse ships which also have an elderly population.