BREAKING
Today, the US will have a new record of cases.
Deaths will pass 1,000 for the first time in July.
And they will keep going up.
Thread.
Deaths & cases are both lowest on Sunday & Monday. Probably a result of weekend reporting.
Then they start going up on Tuesdays
Today, the US will have a new record of cases.
Deaths will pass 1,000 for the first time in July.
And they will keep going up.
Thread.
Deaths & cases are both lowest on Sunday & Monday. Probably a result of weekend reporting.
Then they start going up on Tuesdays
We know that cases will keep going up because there are plenty of states that aren't controlling them appropriately. The coronavirus is spreading like wildfire there. [2/17]
Thankfully, so far deaths haven't kept up with cases.
But why? [3/17]
But why? [3/17]
We can see that, in April, deaths trailed cases by one week.
Correcting for this (and a statistical report from NJ), we can clearly see how deaths and cases have been decoupling [4/17]
Correcting for this (and a statistical report from NJ), we can clearly see how deaths and cases have been decoupling [4/17]
This decoupling is phenomenal. Since April 15th, there's a full 85% fewer deaths per case! [5/17]
We can clearly see that in the Case Fatality Rate (CFR), which has been dropping. Can it keep going down? [6/17]
Unlikely. It's already really low. [7/17]
So what's causing such a low CFR?
It's not testing. Testing is pretty good, but nowhere as good as in many other countries. [8/17]
It's not testing. Testing is pretty good, but nowhere as good as in many other countries. [8/17]
Also, testing is an artifact. It's driven. mostly by the states that are doing a good job at controlling the virus, whereas places where cases are exploding aren't keeping up. [9/17]
One of the keys is age.
The CDC doesn't provide the ages of cases, but we can find some proxies. For example, the age of hospitalized patients from COVID-NET.
They show substantially fewer old people as a share of hospitalizations. [10/17]
The CDC doesn't provide the ages of cases, but we can find some proxies. For example, the age of hospitalized patients from COVID-NET.
They show substantially fewer old people as a share of hospitalizations. [10/17]
If people above 65 go from 60% to 20%, that's right there a drop in ~2/3 of deaths vs cases. But remember, the drop has not been 65%, but ~85%. So where is the rest coming from? [11/17]
There are some treatments now: proning, dimethicone, remdesivir, maybe HC, blood thinners, anti-platelet medications... Some work better than others. We're better than in April. But not enough to justify all the remaining gap. [12/17]
It's likely that the last gap is due to... timing.
Let's look at Texas, which reopened sooner than most, so it has already seen a long uptick in cases... and deaths followed.
It took 2 additional weeks.
[13/17]
Let's look at Texas, which reopened sooner than most, so it has already seen a long uptick in cases... and deaths followed.
It took 2 additional weeks.
[13/17]
It's likely that better testing and contact tracing is identifying cases much earlier than when they arrive at the hospital. That helps with treatment. It also means that more time passes btw case identification and death.
If it was 1 week before, now it's 3w in Texas [14/17]
If it was 1 week before, now it's 3w in Texas [14/17]
In other states, the delay has also grown. Eg, in CA by one additional week, while in NV so far by 3 more weeks. [15/17]
If we apply an additional 2-week delay to deaths vs. cases, it looks like the increase in deaths we've seen over the last week is not a fluke. It's the beginning of the ride up to follow cases. [16/17]
If you don't want to miss up these analyses, sign up for my newsletter.
mailchi.mp/b44a5f38b4f7/c…
mailchi.mp/b44a5f38b4f7/c…
