Thread by @ElonBachman

1\ Seeing this study get passed around as evidence of a high Coronavirus CFR, often with the wrong numbers being quoted

Let's discuss what it does and doesn't say

cmmid.github.io/topics/covid19…

2\ The authors estimate a CFR of 0.2 to 1.2 with a mean of 0.5

But that is an estimate for Hubei province *only*

It is not an estimate for other parts of China or other countries

3\ The estimate is for Hubei province *only* before February 11, which is when containment measures started

So, 0.5% is their estimate for CFR in a population of high poverty, high smoking patients with an unprepared medical system

4\ For their estimate of asymptomatic cases being overlooked, they use age-adjusted data from the Diamond Princess

The Princess was notable for the low percentage of asymptomatic patients (about half), particularly among the young.

5\ The authors mention other data sets showing much higher asymptomatic rates (80%+), but use the Princess data instead

6\ Grafting unusual boat data onto pre-containment data from a poor province with lots of smoking and no medical preparedness makes a 0.5% CFR feels like a ceiling to me

Still worse than flu, yes

But let's see what the Grand Princess data says this week about asymptomatic cases