"Looking at data from countries with robust testing systems does support the idea that the disease’s mortality rate may be lower than 3.4%. Countries that have tested significant numbers of people are generally reporting lower mortality rates than those, like the U.S., that have tested in far lower numbers and with a stronger focus on severe cases. This suggests that when testing networks are broadened to catch people with less serious illnesses, and case counts then reflect this range of severity, mortality rates go down.
The mortality rate in South Korea, where more than 1,100 tests have been administered per million residents, comes out to just 0.6%, for example. In the U.S., where only seven tests have been administered per million residents, the mortality rate is above 5%." Time magazine
"Just less than .6%"
IOW you have to include; mild cases, asymptomatic cases, and moderate cases in your calculation and not just figure a death rate among the severely afflicted in order to get a valid result.