That sounds like an extraordinarily high estimate. How could we have an accurate estimate of hospitalisation proportions without even knowing how many infections are out in the community?
It seems very, very likely that we are missing the vast majority of cases given the tests are pretty inaccurate, and we only bother administering them to people who are either extremely sick already, or who have been in an "affected region" (realistically, every region is "affected" at this point).