My USA Omicron Predictions, so far so Good

Omicron IS proving to be a clusterfsck, exactly as predicted

In the beginning of December, I developed a publicly-accessible open source Omicron Pandemic Model. I put some special effort into refining everything I knew about Omicron based on earlier affected countries like South Africa and Denmark.

Since then, as of a couple of weeks ago I stopped changing it as I decided to let it run and see how right , or wrong, I was. I posted the graphs a few times already.

My official prediction in “The Fog of Omicron” was 600,000-2,000,000 dead. As I mentioned, the actual fatalities is the hardest thing to predict. Also, due to possibly delayed effects of Omicron which I will write about, the mortality may only show up finally in excess mortality statistics for December-March, which is hard to decipher due to additional booster-caused mortality. As time goes on, hopefully we will get better grasp on the death statistics. I promise to be honest, dogged and skeptical.

As I mentioned in the Fog of Omicron, the number of “cases” was quickly going to become unreliable. I saw some mile long vehicle queues for PCR testing last week, and with the availability of rapid tests and general well founded fatalism, many cases will not be reported. I did not bother reporting a case in my own family, because why.

However, hospitalizations are a much more reliable statistic as anyone admitted to a hospital “with” or “due to” Covid is counted, at least so I hope.

What the two paragraphs above imply, accidentally, is that the observed case fatality rate for Covid will actually be far above actual, due to cases not reported. This is purely a data observation issue, but it makes actual numbers much less reliable.

Anyway, let’s look at hospitalization. Per Our World in Data, as of Jan 7, there were 125,921 Covid hospitalizations in USA. This includes Covid Delta hospitalizations, while my model includes only Omicron.

My prediction for January 7 was 76,757 hospitalizations for Omicron. This is actually relatively in line with Our World in Data, if we assume that Delta hospitalizations still account for about 40-50,000 persons hospitalized.

All of these estimates are very sensitive to a time shift of even 1-3 days, so it is not essential that they are perfectly correct, as just randomly happened, as that they are roughly in line.

So far, Omicron hospitalizations generally follow my predictions.

Again, as for deaths, Omicron deaths cannot be decoupled from Delta deaths until they finally greatly exceed Delta deaths. This has not started happening yet, but it is going to start happening this week so the deaths will be significant and meaningful in 7 days or so.

Note that even if Omicron deaths right now are less than 0.8% of cases, which was a low confidence guess in my prediction, that ratio will rise due to hospitals being unable to provide the “standard of care” as more and more patients are admitted and as vaxxed personnel is out sick.

Again, deaths so far were minor compared to Delta, so it is hard to separate them, but as of this week they will far exceed Delta and will become very meaningful in the USA.

Near Future

According to my model, cases will crest mid-January at 6,000,000 cases per day. Not all of them will be reported. Hospitalizations will crest at around Jan 25 at 600,000 persons hospitalized. Deaths will crest around Feb 2. I hope that they crest at a much lower number than my predicted 40,000 on the worst day, my model gives out this number but even I cannot make myself believe it.

My model predicts 1,200,000 deaths and I greatly hope that the actual number will be much lower. My official range is 600,000-2,000,000 deaths and I also greatly hope that my model is very mistaken, even though so far a lot of things I said were correct.

The predicted “cases” are actual incidences of people sick, not the official counts of official positive tests at test centers.

The “Quarantined” column is for 10 days off work on average, which may be reduced due to latest CDC guidance. I decided NOT to alter this column just to keep everything as written. Not everyone quarantines and I know persons who did not quarantine at all.

Watch out hospitalizations, which may be the most robust indication of things for the next 1-2 weeks. After that, deaths will finally be significant (and hopefully will be lower than my prediction) and will obviously be more important than hospitalizations, because hospitalizations are not an endpoint.

The brief take from this article is “January and February will be a clusterfsck”.

January and February would also be a great time for China to attack Taiwan, or for a large central Eurasian country to attack whoever it wants, due to general disorganization and specifically likely crippled vaccinated Western navies.

Delayed Deaths

Ever since my “Omicron may be a bioweapon” article, I paid attention to the possibility that Omicron causes delayed deaths. It does seem that it might be the case.

I monitor Twitter for mentions of “Omicron DVT” and “Omicron Stroke” and it seems that something is going on, but not certain enough yet. If I do find anything I will let you know.

Courtesy of this Twitter post by Dr Abe