That is percentage of confirmed cases that wind up with hospitalization.

However we mostly don't try to confirm cases if it won't affect the care we give. We just tell them to self-isolate. We therefore will not have good statistics on what percentage of cases are confirmed until after the immediate crisis is over and we test to see how many have antibodies.

So your figure is, "Of young people who have it severely enough to get tested, 15-20% get hospitalized." This says nothing about what fraction of cases are severe enough to get tested.

Also note that we undercount deaths. If a person dies and was never confirmed to have COVID-19, they are also not counted as a COVID-19 death. Nobody knows how big this undercount factor is.

All good points. But paulsutter's message remains: young people are at significant risk. Yes, the risk for older people and those with pre-existing conditions in higher. But the parent poster's characterization of the risk for young people as "very little" is dangerous.

This is also why I don't really like that some propose various "herd immunity" schemes - we simply don't know what the virus actually does to people in practice, even if it does not outright kill them. So it seems very irresponsible to propose schemes like that at this point in time.

Exactly. I know I'm waiting until they have herd immunity available for everyone.

The 15-20% number is very stale. The current numbers are under 3% across that cohort: https://www.thelancet.com/journals/laninf/article/PIIS1473-3...

0–9 years 0 13 0·00% (0·00–0·00)

10–19 years 1 50 0·0408% (0·0243–0·0832)

20–29 years 49 437 1·04% (0·622–2·13)

30–39 years 124 733 3·43% (2·04–7·00)

40–49 years 154 743 4·25% (2·53–8·68)

50–59 years 222 790 8·16% (4·86–16·7)

60–69 years 201 560 11·8% (7·01–24·0)

70–79 years 133 263 16·6% (9·87–33·8)

≥80 years 51 76 18·4% (11·0–37·6)

More detail on above: published March 30th, the percentages are expected hospitalisations based on total population of infected (including mild/asymptomatic usually untested).

Data is from Table 3 near end of paper, table title: “Estimates of the proportion of all infections that would lead to hospitalisation, obtained from a subset of cases reported in mainland China“

I'm not sure I buy it as we don't have any idea how many people actually have the disease. I have to assume the denominator is of those tested, X% required hospitalisation. If it is those confirmed to have the disease, and because clearly the whole population hasn't been tested the results are skewed as we only test people with severe symptoms.

Right those population level samples only started on a limited basis in a few areas last week. For now the actual denominator is a total unknown; the range of upper and lower bounds is so wide as to be effectively useless for planning.