If you believe the survival rate is 99.8%, then that implies you think that ***110%*** of people in New York have been infected. https://twitter.com/PoliticalSock/status/1321543846698897410

This is actually a nice chance to talk about COVID IFRs.

New York is the hardest hit place in America in terms of excess deaths per million people, and top 5 in terms of official COVID deaths per million.

I've shown a bajillion times that excess deaths are the right metric, but I'll compromise a bit: New York has had 45,000 excess deaths and about 24,000 official COVID deaths, with a population of about 19 million. Let's call it 40,000 COVID deaths.

At a 99.8% survival rate, that implies a total of 20,000,000 cases in New York to generate those 40,000 deaths.

Whoopsie!

That's more than the total population of New York!

That IFR must be incorrect somehow!

Folks, even at a 0.5% death rate you get a pretty wild conclusion that ***42%*** of New Yorkers have been infected.

That seems kinda high to me!

Now, look, I get it. You might think the IFR has fallen since March.

So what about recently-hit states? How would, say, Arizona look?

At a 0.2% fatality rate, that's implying that ***67%*** of Arizonans have been infected.

If instead you use serology studies to try to calibrate this stuff, the usual IFR implications range from 0.5-1.5%. As they have since April.

IFRs estimate from Spanish excess mortality and serology data suggest a Spanish IFR in the same range as the US! https://twitter.com/Roadants/status/1321548443580571653

This study puts the all-groups IFR at 0.64%. Which is what I just said. https://twitter.com/wigglin/status/1321548957999271936

Anyways, suppose IFR really HAS fallen.

Why? What caused that fall?

Viral mutation? That'd be the most optimistic reason. Alas, not much evidence for it.

Better care? Maybe! But what kind of care?

Did we find a highly effective drug? Well, we HAVE found a variety of drugs that can help.

But they're not take-home. They're all stuff you get IN THE HOSPITAL.

The big problem with this is that means you're gonna have tons of people running up huge healthcare costs and crowding hospitals. Getting treatment for everyone will be hard and expensive.

There are some things that can improve survival rates without hospitalization. But so far there really is *not* evidence of a massive decline in IFRs; certainly not enough to justify letting things ride.

The best read of things I think is that IFRs in the spring were around 0.8%-1%, and today are around 0.5%-0.75%.

That's a big improvement.

But it still implies that less than 20% of Americans have gotten COVID.

That implies that deaths between now and a vaccine from COVID could run up to ANOTHER few hundred thousand, maybe more.

If we have another 10% of the population exposed before a vaccine with 0.5% IFR, that's another 165,000 deaths.

If we can prevent that. If we can prevent HALF of that, or a QUARTER of it, we should do it!

I mean if we knew that Iran was going to set off a bomb that was going to kill 165,000 American civilians, I just don't think we'd really be dithering about our response in this way.

More bluntly, if it actually were the case that China engineered the virus and introduced it as a weapon (this is not true, but, IF it were), I think we'd all understand that we're at war and a lot of our civil liberties will have to take a vacation for a bit.

But because it's mother nature, we just throw up our hands and say, well, rats, guess there's nothing we can do!