This page originally written 2020-12-30. Last major edit 2020-1-2.
This page is a place to collect bits and pieces of news and information as I follow the spread of the new coronavirus variant originally found in the UK: Variant of Concern 202012/01.
I'm interested in keeping a rough timeline of its spread worldwide as well as its spread in the United States.
This information is taken mostly from the Wikipedia page referenced above, except for the United States, where the date is common knowledge and reported by many news sources.
As of January 2, 2020, I have not yet seen an estimate of how many actual UK strain cases there are in the United States, except for a vague CNN headline alluding to "hundreds" around December 20th.
I've come up with an extremely quick-and-dirty estimate based on information provided by a genomics company known as Helix, which I intend to use as my best estimate until further information is available.
On December 23rd, Helix reported that approximately 0.5% of US coronavirus cases in the most recent week were returning evidence of "S gene dropout". S-dropout is a feature not only of the UK strain, but of other lineages as well. So this would indicate that the total number of Covid-UK cases as of mid-to-late December was some subset of 0.5% of total cases. (Source.)
On December 31st, Helix reported that testing a subset of 31 of those S-dropout cases showed that four of them were the new UK strain. (Source.)
Multiplying 0.5% by four and dividing by thirty-one yields an estimate that 0.065% of US cases are the new UK strain. If this data is going on two weeks old, then, given UK data indicating that the prevalence ratio of Covid-UK to Covid-Classic seems to double about weekly, then perhaps 0.2% of more of US Covid cases as of January 2 are Covid-UK. On the other hand, if the data is newer, perhaps the number is currently closer to 0.1%.
My best guess, then, as of January 2, is that 0.1% to 0.2% of US cases are of the new UK strain.
In further extrapolation from rough numbers, Youyang Gu's model suggests that the United States has perhaps 460,000 new cases of Covid per day, compared to more like 630,000 in the IHME model.
Taken together, these numbers suggest that perhaps approximately 1,000 new cases of the UK variant spreading each day in the United States is on the correct order of magnitude, though any kind of accurate reading will have to await further research.
If it is the case that the US is currently (Jan 2, 2020) experiencing 1,000 cases per day, then what can we say about the path of Covid-UK in the US going forward? This will depend on many variables, but perhaps we can chart out something resembling the worst possible plausible scenario just for whatever value the exercise might have.
There has been some data from the UK suggesting that the prevalence ratio of Covid-UK to Covid-Classic doubles approximately weekly. If that were to be the case here, then we are approximately nine weeks from Covid-UK becoming just as big a problem as Covid-Classic in the United States. The date when there are approximately 500,000 Covid-UK cases per day in the US would be about March 6th.
Note that currently there are approximatley 125,000 Americans hospitalized with Covid-19. If by March 6th there were 125,000 Americans hospitalized with Covid-UK, and if that number were to continue to double weekly, then we would be facing some extraordinary numbers pretty rapidly:
Statista estimates that there are less than one million hospital beds in the United States, so in the scenario just calculated, the US would have its capacity overwhelmed by the end of March.
There are all sorts of reasons that the US might not have its capacity overwhelmed by the end of March. The estimates are sketchy about how many people have Covid-UK now and about how fast the new strain has been spreading. It is difficult to predict how fast the new variant will spread. It is difficult to predict how much of an effect vaccine rollout and other future counter-measures might have on the future spread of the virus.
"Hospitals will have no open beds by the end of March" is not a prediction I'm confidently making. It's just the outcome of one set of assumptions. The assumptions are not good enough to treat them as a crystal ball, but they are plausible enough to suggest that the UK strain is worth paying very careful attention to over the coming days.
For both my current (Jan 2, 2020) guesses about Covid-UK prevalance and my sketchy calculations about possible future spread, I am very interested to see additional data that might help clarify things.
One source of useful future data will be additional surveillance of S-dropout and additional sequencing data. I expect that the quality of both will improve over the coming weeks. We will have a much better idea of how many cases there are and how fast they are increasing.
Another pressing question is how the US will respond to and come with a massive increase in Covid-UK cases, if that increase does indeed come. While we won't know for certain until we're there, one possible indicator to watch is the UK's situation. The UK is already experiencing strain on its hospitals, and the UK strain is already dominant there.
Here are the number of daily cases, pulled from Worldometers, for selected dates in the UK:
In other words, we've seen a rapid escalation from a fairly stable situation to increasing cases as the Covid-UK strain has become dominant. It will be educational to see whether the UK manages to turn the tide and re-establish some kind of equilibrium without most of the country catching the virus. It will also be informative to see whether the healthcare system collapses -- and what "collapse" would mean in this context. It is a vague term, after all.
Taking my initial wildly-unreliable guess of 0.1-0.2% of US infections being B.1.1.7, and also the possibly alarmist assumption of the virus doubling its prevalence weekly, we would expect to see about 0.2-0.4% of US cases being Covid-UK as of about January 9th. I still haven't seen anything to clearly indicate whether this is too high or too low, though I haven't researched much.
I do see a CNN story announcing that as of January 6th, at least 52 cases of Covid-UK have been found in the US. The same story also implies that the CDC is sequencing about 3,250 (or thereabouts) virus genomes per week. This suggests that in the week preceding the 6th, something like 49 new cases were found. 49 divided by 3,250 yields about 1.5% of the cases sequenced being Covid-UK. What I haven't seen is a claim that this is a representative sample. If the CDC is deliberately looking for Covid-UK, then it's entirely possible that less than 1.5% of the infections are B.1.1.7. Perhaps we could say that 1.5% is something like an upper bound -- there are most likely less than 1.5% of cases coming back with Covid-UK, especially when we consider Helix Research's numbers suggesting that as of late December B.1.1.7 was a small fraction of a percent of national cases.
On January 7th, NPR published a piece in which Trevor Bedford, a computational biologist, floated the figure of perhaps 1 in 1000 US cases being Covid-UK. Because it was a short radio piece, it did not contain a detailed justification for that number. So I went and found Trevor Bedford's Twitter and found this, in which Bedford indicates agreement with numbers produced by Alexandre Bolze, a member of the Helix research team. Clicking through leads on to this, in which Bolze implies that multiplies 4/31 by 0.5% is the source of the number. In other words, the same computation I did January 2nd, but a day earlier.
So it looks like the raw number is 0.065%, which has been rounded to 0.1%, or perhaps 0.1% is an increase to account for time elapsed. Now, my previous math, on January 2nd, was based on the assumption that the 0.065% figure was for perhaps December 20th or so. But Bolze indicates that the samples involved were from "early December". This is a bit earlier.
Meanwhile, Zvi Mowshowitz seems to have concluded that the best way to model the growth of the UK strain is to estimate 50% greater reproduction per five-day period. He's better informed generally than I am on coronavirus, so for now I figure I'll borrow his figure.
And as of today, those numbers would suggest we're at about 0.37% of the nations cases are Covid-UK.
If we plug in the "50% more infectious over a 5 year period" used in Zvi's toy model, then, unless I'm mistaken, we'd be looking at about 18 cycles from December 8th for Covid-UK to equal Covid-Classic in prevalence. That's 90 days, so calculating from the guess of December 8th as the starting point, March 8th becomes about the day Covid-UK overtakes Covid-Classic in the US.
How prepared are we for that day, if it comes? The UK is continuing to fail to contain Covid-UK. They've gone so far from about 53,000 cases per day at the beginning of the year to 68,000 as of yesterday. However, by March 8, Youyang Gu's model, as of right now, suggests that about 39% of Americans will have already become immune to the virus. One might hope that would give us some breathing room, but I don't know too much about these things.
This page is released under the CC0 1.0 license.