Post #1347: Omicron, two observations on the U.K. experience so far.

Posted on December 13, 2021


Briefly, those two observations are:

1:  Extremely rapid growth of Omicron cases is occurring, but this is not translating into extremely rapid growth of total cases.  As I explain in excruciating detail, I think the reason boils down to this:

Net new case growth rate = Omicron growth rate – Delta disappearance rate.

2:  The case hospitalization rate for Omicron, so far, is about one-fifth the rate for Delta.

Details follow.

Why it’s hard to pin down the true severity of the average Omicron case.

On one end of the spectrum of severity there are asymptomatic cases.  At the other, deaths.  Varying degrees of morbidity that are harder to characterize lie between those end points.

You might think it would be fairly straightforward to compare Omicron to Delta along that spectrum of severity.  But it’s not.

First, in some countries, genetic sequencing is applied to only a tiny random sample of COVID-19 cases.  You only know the variant for that sample, and you only know the aggregate proportions.  In those countries, when a case is admitted to the hospital for COVID, physicians have no way of knowing what variant it is.  There is no person-level way to assess the severity of one variant against another, from published data, in countries that track the data that way.

Worse, in the U.S., we don’t even know the mix of variants in real time.  Currently (12/13/2021), the CDC variant tracking page shows us the mix of variants for the two weeks ending 11/20/2021.  That might as well be last century, for as fast as Omicron is moving.  Per today’s reporting, in London, Omicron already accounts for 40% of new cases, and that’s expected to top 50% tomorrow.  We should have every expectation that the U.S. is on a similar schedule, just a few days behind.  Which we will know, in hindsight, sometime next year.

At the low end of the severity scale, we just plain cannot count asymptomatic cases.  Sometimes these are identified when individuals are tested for some other cause (exposure, say).  But by their nature, the infected people themselves have no reason to seek a test.

Various intermediate degrees of severity are not tracked by any sort of centralized database, in any country that I know of.  There’s no real-time monitoring of (e.g.) the fraction of diagnosed COVID-19 cases with loss of sense of smell or taste.

We have just two hard-number indicators for the severity of the average Omicron infection:  Case hospitalization rate, and case mortality rate.  That is, total hospital admissions for Omicron and deaths from Omicron, divided by total diagnosed cases of Omicron.

Both of these suffer from being “observational data”.  That is, they are based on whatever we happen to observe, not on any randomized-controlled comparison between groups of patients.  First, those will depend on whatever population happens to have gotten infected.  E.g., if all those infected are healthy young people, we’re not going to see many deaths.  Second, those rates will be affected by (e.g.) the fraction of the population that has been vaccinated, and fraction with booster shots.  Finally, those rates will reflect the health care infrastructure and practices of the country of occurrence.  Some countries, for example, might have a higher propensity to hospitalize people compared to others.

Case mortality rate suffers from being a lagging indicator, but case hospitalization rate does not.  In the U.S., it takes about two weeks for the deaths data roughly to catch up with the new cases.  Hospitalizations, by contrast, are concurrent:  When new cases go up, reported hospitalizations go up with them within a day or two.

An eye on the U.K.

The U.K. does a much better job of tracking variants than the U.S. does.  What’s happening there is the best guide we’ve got to what’s likely to happen here.  With that upbeat look at data quality issues as background, what can we infer from the most recent information from the U.K?

Predictions of rapid spread of Omicron are now verified in Great Britain.

But that’s not translating into extremely rapid growth in total cases.  Yet.  And I think I know why.

It took less than three weeks for Omicron to become the dominant strain:  Per the reporting cited above, the first case of Omicron in England was detected 11/27/2021.  Omicron is expected to account for half of new cases in London as of 12/14/2021.  That’s an elapsed time of 17 days. 

The first detected case of Omicron in the U.S. was announced on 12/1/2021 by the CDC.  Or, just four days later.  There are some differences between the U.S. and Great Britain.  Great Britain is more densely populated than the U.S., has a higher vaccination rate, and currently has a much higher rate of new COVID-19 infections per capita.  Despite these difference, my best guess is that we’re on roughly the same timetable.  Based on the U.K. experience, Omicron should be come the dominant variant in at least a few U.S. cities sometime this week.  Possibly early next.

But I think it’s well worthwhile to note the trend in total cases in the U.K, even as Omicron is taking over.  Scrutinize the right-hand end of the line, for U.K. COVID-19 new cases:

To me, the key feature is that it doesn’t look remarkably different from the pre-existing trend.  So far.  Maybe a touch faster.  But the point is that it doesn’t look like the near-vertical growth curve for South Africa, for this phase of the spread of Omicron there.  We certainly expect the trend to pick up at some point, because Omicron is more contagious than Delta.  But so far, we haven’t seen anything like what occurred in South Africa.

Why not?  Luck?  Lower infectiousness?  Better COVID hygiene?  More vaccination?  All of the above?

None of those explanations makes a lot of sense, because all of those factors would have affected Delta in the U.K. as well.  We’re now taking a situation that already reflected all of those factors, and plowing in a far-more-infectious strain.

Why isn’t the growth rate going vertical, as it did in South Africa, in this phase of the spread of Omicron?  Here’s my best attempt at an answer.  It boils down to this:  The net case increase in Great Britain isn’t the growth of Omicron.  It’s the growth of Omicron, less the demise of Delta.

In South Africa, they started from a baseline of few COVID-19 cases.  In Great Britain, by contrast, they are starting from the middle of its winter wave.  There are lots of Delta cases already in circulation.

And that raises an interesting point, and one that I have never understood over the entire course of this pandemic.  New strains displace older strains, they don’t simply add cases on top of the older strains.

Empirically, this was clearly the case as Alpha displaced the native (Wuhan) strain, and as Delta displaced Alpha.  In both cases, it’s not as if the absolute number of cases or trend in cases for the prior strain was maintained, and the new strain added on top of that.  Instead, the trend of new cases with the prior strain went to zero.  The new strain not only takes over percentage-wise, with both strains co-existing.  The new strain drives the old strain out of existence.  Or nearly.  Cases with the new strain displace (almost) all the cases with the prior strains.

A rambling but important aside:  I guess that’s obvious to some people, but to me that makes no sense.

I mean, think about it, in the last battle-of-the-strains, where Delta displaced Alpha.  I understand how Delta spread faster than Alpha.  So I can see how Delta would rise as a fraction of all cases.  But how did Delta suppress the spread of Alpha?  How did it, in effect, kill off Alpha as it spread?  

For that to happen, Delta must have been able to find the people who otherwise would have been infected with Alpha, and get them infected with Delta instead. That implies that there must be a highly constrained set of people who are open to being infected, at any one time.  Just-so-many opportunities for infection.  And whichever strain gets to those people first, wins.  Many of the “wins” for the new strain are therefore losses for the old strain.  That’s how Delta managed to kill off Alpha.

For some reason, I just have a hard time getting my mind around that.  But that’s the only mechanism I can figure that would result in Delta killing off Alpha.  The implication is that the same people who got infected with Delta, would have gotten infected with Alpha, if Delta hadn’t been around?  Some of those Delta cases have to be cases that would otherwise have been Alpha.

I have a real point here, I think.  In South Africa, there were almost no existing cases of COVID-19 in circulation.  The net growth in new cases was more-or-less the growth rate of Omicron.  In Great Britain, they already have a heavy case load of Delta cases.  In Great Britain, the net growth of cases is the growth rate of Omicron, less the disappearance rate of Delta. 

My gut tells me that’s why cases aren’t skyrocketing in Great Britain yet.  Not the way they did in South Africa.  It’s not because Omicron is less infectious in Great Britain.  It’s because some of the growth of Omicron comes at the expense of Delta.  So the net growth rate isn’t the native rate of spread of Omicron.  It’s that, less the rate of shrinkage of what would otherwise be Delta cases.

Net new case growth rate = Omicron growth rate – Delta disappearance rate.

In South Africa, that second term was zero.  In Great Britain, where the country is saturated with Delta cases now, it’s large.  If anyone has explained this point in the popular press before, it has escaped my notice.  But I think it makes sense.  It may even be true.

For the U.S., in the middle of a weak winter wave, the same should hold true, but to a lesser degree than in Great Britain, where they were already having a particularly bad Delta wave.   Especially in those states that appear to be saturated with Delta cases already.  It there really is just some finite pool of “targets for infection”, and Omicron grows in large part by eating Delta’s market share, then the net increase in cases we will see won’t be like South Africa.  It’ll be more like Great Britain.  It’ll be the faster growth rate of Omicron, less the disappearance rate of Delta.

So, yes, Omicron is spreading like wildfire in Great Britain.  But that isn’t translating into a vertical spike in cases yet.  Because, so far, much of what Omicron appears to be doing is displacing what would otherwise have been new Delta cases.  Or, at least, that’s my explanation of it.

So far, Omicron has around one-fifth of the historical hospitalization rate of Delta in Great Britain.

The other little tidbit from Great Britain is the extremely low hospitalization rate for Omicron, relative to the historical rate for Delta.

Per this reporting:

“The UK has so far reported 3,137 cases of the Omicron variant, though the true number is likely to be higher. Javid said “about 10 people” are in hospital with the new variant. Overall, the country’s seven-day rolling average of Covid-19 cases has surpassed 50,000 a day.”

“In hospital” isn’t the same as “ever were admitted to the hospital”, but it’s close enough.  My math tells me that 10/3137 = 0.3% case hospitalization rate, so far, for Omicron.

How does that compare to Delta?

As of 12/1/2021, Great Britain was admitting about 800 patients per day for COVID-19, per government sources.  At that same time, they had about 48,000 new cases per day (same source).  Or a 1.7 percent current case hospitalization rate for Delta.

From which I conclude, with all the usual caveats, that so far, in Great Britain, Omicron has about one-fifth of the case hospitalization rate of Delta. 

Given how scant the data are, I’d say that’s in the same ballpark as my one-tenth estimate from a couple of postings back.

No matter how you slice it, it’s a lot lower.  Materially lower. 

We’ll get final proof of that Great Britain extends these graph further into the Omicron era.  (Currently they end on 12/7/2021).  Both are from the government of the U.K. 

The top graph is new cases of COVID-19.  The bottom graph is new hospital admissions for COVID-19.  If I’m right, the bottom one is going to fall as the top one rises.  As Omicron displaces Delta, if it’s true that the case hospitalization rate for Omicron really is something like one-fifth that of Delta, it will be completely and totally obvious from this graph.  Once Great Britain updates it.