Thread by @kylamb8

Early anecdotes coming from doctors in South Africa suggested Omicron was mild, or certainly mild(er).

Data there quickly started supporting those anecdotes, and also in places like UK, Denmark, and more recently, Florida.

But now there is hope for the U.S.

A thread.

We'll call this decoupling rate. It's basically an odds ratio type of calculation using a 30d rate of change of three indicators.

A) New hospital admissions 7d ave compared to cases 7d

B) Hospital census 7d compared to new admissions

C) ICU census 7d to hospitalizations

2/

Using these odds ratio, we can see decoupling in the differences of rates of change. We have been seeing it in Florida. Measuring these three together is analogous to measuring ICU (severe illness) as a rate of cases.

Until now, these have predictably moved similarly.

3/

I have grouped states into eight groupings using BEA regions. I only chose that because these eight groups tend to match trends fairly well.

FL (0.10) has the lowest ratio followed by LA (0.13), AL (0.15), HI (0.16), GA (0.18), & TX/CA (0.19).

So what does this mean?

4/

In general, those three ratios I mentioned (A, B & C) would usually give us a product of roughly ~1.00 (A x B x C). Most of these indicators move together between 0.9 and 1.1 though there are exceptions. Hospital census and ICU both track near a 1:1 ratio with cases.

5/

So let's bring back the first graph I posted. You can see each state's ratios using 30d rate of change from Dec. 6, 2021 to Jan. 5, 2022 for A, B and C and the decoupling ratio is the product of the three.

6/

To illustrate the ratios *usually* move together near 1.00, I also calculated the ratios for the same dates one year ago, Dec. 6, 2020 - Jan. 5, 2021.

So look at the right hand columns for each state for last year. See how most ratios tend to hover closer to 1.00?

7/

Now look again at the ratios being seen now. Most states are starting to see significant case decoupling to hospital admissions (A), and many are also seeing shorter stays from admissions (census, B), and less severity (ICU, C).

8/

With exception of Maine, every single state has a ratio lower than last year, most by a great deal. Minnesota is the only other state that does not have a ratio lower than the lowest ratio of ANY state one year ago. We can also see similar patterns by region.

9/

Again, to sum this up, the decoupling ratio is essentially the rate in which cases are turning into ICU admissions, which I'm using for severe illness. We are seeing fewer cases turn into admissions, shorter hospital stays and less severe illness as a product of Omicron.

10/

This is unprecedented for the U.S. Earlier this past summer, the UK had already decoupled cases from ICU and death rather significantly, but we had been waiting for that same trend to be seen here. It is most definitely happening in Florida and now apparently several states.

11/

Whether this will continue remains to be seen. Whether it's merely a product of Omicron is unknown. However, if prior immunity and vaccines alike hold up against severe illness, there is no reason to think this wouldn't be a sign of moving to endemic stage.

/end

Addendum: all hospital data comes from HHS reported patient impact and hospital capacity file.

Case data is from CDC cases and deaths over time file.

Odds Ratio formula is:

(7d ave now / 7d ave 30d ago), divided by the same for the second indicator.