So here’s a way to visualize it, I think:
What happens when we calculate a CFR based on current cumulative deaths compared to identified cases as of 4-5 weeks ago. If we do that, I think the evidence points to a likely CFR for current cases closer to 3%.
And what’s interesting (to me) is that the numbers look extremely similar regardless of whether you use a 28-day lag or a 35-day lag. That suggests to me that most of the deaths from a given cohort are likely happening within 28 days.
So I think it’s going to be much closer to 2k than 3k based on current cases.