This was originally a Facebook post of mine from 2021-1-2, and is archived here as a curiosity. Minor changes to formatting, as well as basic copy-edits, may have been made in the transition from Facebook post to web format.
Time to run the weekly numbers.
We've just gotten through Christmas and New Years, and this messes with the Covid statistics in two main ways. As the people normally reporting the data go home, the data backs up, and gets reported later. First, this causes numbers to drop, so it looks like everything suddenly got much better on Christmas and Christmas Eve. Then, as they come back and catch up, the numbers look extra-high for a few days. Then the cycle repeats over New Year's Eve and Day. This means that numbers right now don't tell us much. Before Christmas, deaths were still rising, cases were levelling off, and hospitalizations were still rising just a bit. Nothing in the fuzzy numbers since have given us much new information.
On the vaccine front, Bloomberg is reporting 3.5 million vaccine doses administered in the US, out of 12.4 million shipped. 3.5 is almost certainly an underestimate, as it takes some time for a local vaccination to find its way into the national totals: first the hospital or institution giving it has to report it to their local health authorities or state, then the state compiles its latest numbers together, then they release those numbers to the public, and then some staffer at Bloomberg adds them to the total. So it's at least 3.5 million so far, and enough vaccines earmarked for nursing homes have been shipped to vaccinate all nursing home residents. On the vaccine front, Youyang Gu, who is generally very good at these things, estimates 5.6 million people have been vaccinated so far, and that in January we'll go from about 400,000 new vaccinations per day to about 800,000 by the end of the month.
So the clock is ticking for coronavirus. By Gu's estimate, we're set to hit herd immunity in mid-June, although of course things could in reality go slower or faster.
The one big wild card out there right now is the new UK strain of the coronavirus, which has been approximately doubling each week under pretty strict lockdown conditions in the UK, and has recently been discovered in the US as well.
So far, I have not seen any media outlet showing its work on any estimate of how many people have gotten Covid-UK in the US, other than a CNN post from around December 20th that just said there may be "hundreds" of cases here.
I think we can do a bit better. Here's my calculation, although there all sorts of reasons it could be wrong. I look forward to seeing better-informed people with real qualifications publishing their own estimates. In the meantime, here's my best guess.
Up until mid-to-late December, the genomics company Helix has been checking 2 million samples of coronavirus for what is known as "S gene dropout". The exact meaning of this term isn't important for our purposes, except that you should know that all Covid-UK cases exhibit S-dropout, while not all S-dropout cases are Covid-UK.
The last week for which data has been released is the one ending December 23rd, and for that whole week, about 0.5% of US cases exhibit S-dropout. Then Helix took 31 of the cases of S-dropout and fully sequenced them, and found that 4 of them were Covid-UK. So if as of around December 20, 0.5% of US cases had S-dropout and 4/31 cases of S-dropout were Covid-UK, then a very rough calculation suggests that for around December 20th 0.065% of cases in the US were Covid-UK.
Given that the prevalence ratio of Covid-UK to Covid-Classic seems to double about every week, that would suggest that right now about 0.2% percent of cases in the US are Covid-UK.
I want to remind you that in college I studied Hebrew, not medicine, so it is entirely likely I am making some bone-headed mistake in calculating those numbers out. I am fully qualified to tell you what a hitpael verb in Hebrew is; I am not qualified to tell you what will happen next in the weird world of Covid. I look forward eagerly toward seeing more qualified people run the numbers themselves, and the moment they do I plan to stop using my own estimates and start using theirs.
Still, 0.2% of cases is the best figure I can come up with for now, and it's not a pretty one. If the prevalence ratio of Covid-UK to Covid-Classic doubles weekly, 0.2% is just nine weeks away from Covid-UK being as big a problem as Covid-Classic, and then additional doublings would fill up all of the hospital beds in the US in March.
Again, the infectiousness of Covid-UK is still being debated, and there is no rule that says the speed of transmission won't change, so this isn't a prediction of what will happen. Hopefully this will still somehow turn out not to be a big deal.
The next thing to watch is the UK. In addition to the US's most recent numbers, I've attached a graph of recent Covid spread in the UK, and it's ugly. The hospitals are under strain. The next few weeks will tell us whether the UK is able to get it under control, or whether the hospitals will simply totally shut down. We'll check in again in a week.
The writing on this page is released under the CC0 1.0 license. The four graphs are from the Covid Tracking Project, which released its work under the CC-BY-4.0 license. The graph of Daily New Cases in the UK is from Worldometers.