bookmark this tweet and we'll have a look in a couple of weeks at how panicked and wrong it was.

meanwhile, here's US ICU capacity (per CDC) as a whole.

it's still quite low, like "we're gonna go chapter 11" kind of low.

this "simpson's paradox" argument is a trope. https://twitter.com/mbeckett/status/1278750662038302720

it presumes that patients are not movable and that you cannot take a patient from one hospital to another and send them where there is capacity (a process so common that "divert status" is published)

hospitals are empty by historical stds and COV % use is 6.3% vs 10.8% peak

normal hospital usage is more like 80-85% with ICU at 85-95%.

note: these %'s diverge from other analyses as CDC uses hospital bed count around 770-800k, which seems low

the AHA puts it at 924k.

but i will use this lower number to be conservative. https://www.aha.org/statistics/fast-facts-us-hospitals

the simpson's paradox idea is not terribly convincing here, especially on cases. testing levels vary by state, cases are a function of that sample rate, and testing levels and % positive are being skewed by data collection issues

it's a flat out bad set https://twitter.com/boriquagato/status/1279097995372269568?s=20

there is no way to look at cases and back out real, comparable numbers anymore because of the ELR issues and you cannot tell if you're looking at real correlation inversion of real relationships or just divergence in data collection. https://twitter.com/boriquagato/status/1274432160682831872?s=20

so trying to run a simpson's paradox analysis on cases is just an exploration in data artifacts and idiosyncrasies.

the border states have a specific issue ongoing right now.

this has been going on for weeks. https://www.cnn.com/2020/06/29/health/border-hospitals-coronavirus/index.html

arizona, with a tiny population and medical system, can be rapidly affected.

they have ~3000 patients hospitalized and ~740 in ICU.

texas could, even now, take the whole lot of them and have plenty of capacity. it would eat ~25% of their base beds and ~60% of baseline ICU

but baseline ICU is misleading. that number can easily be increased 50-100% in the short run. that's how hospitals are set up.

thus far, the rise in covid is not stressing arizona hospitals.

(state data reading a bit higher than CDC, they look to include surge)

nor are TX, FL, or CA stressed. it seems a bit early to be proclaiming catastrophe.

per CDC, not a one of these states is even at 75% hospital use. (again, CDC includes surge)

on ICU, everybody excpet for tiny arizona is about 65% and has been basically flt for a month.

only AZ has been rising but has still not hit 80% of CDC measured capacity.

and there is good news in AZ.

new hospitalizations (reported by real date of admit) peaked about 2 weeks ago and have been declining sharply.

the past few days may be incomplete, but even by june 24th, they were down over 50%.

jun 27 was only 16 vs 83 june 15.

no place else in the US is going to look like NY,NJ. they had some terrible policy choices specific to them that no one is going to emulate.

they forced cov+ patients into nursing homes to "spare hospitals"

they also used "vent early, vent hard" ICU strategy. it was lethal.

cases and hospitalization were dropping everywhere.

they got stirred up by protests and influx from mexico, but both are transitory (and not due to opening)

this is not a real simpson's paradox, it's just idiosyncrasies in subset data and case data becoming basically unreadable

deaths are still the gold(ish) standard. one can argue that they lag, but we sure see no sign of a rise there so far and cases and hospitalizations are trending young.

so i'm taking the other side of this trade.

the next 8 weeks will see nothing look like NYC.

<timestamp>

data note:

all figures here are CDC provided and use CDC definitions. they can vary from state figures and from other national figures due to this.

i tried to use just CDC #'s here to be consistent and comparable.