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?
However, the point stands that fatality rates will likely go up a lot if medical resources are overwhelmed. Long-term lung damage is also another risk for younger people and its rate will also go up.
Among the ~16-19% who are hospitalized and recover, you may suffer through weeks of pneumonia and your lungs could be damaged long-term.
Even if you're lucky to be among the 80% who survive without hospitalization, you could still be the vector who directly or indirectly infect someone you care about.