The U.S. remains at about 9 new COVID-19 cases per 100K population per day, roughly unchanged from a week (or two or three) ago. The regions continue to diverge, with the Northeast showing a steady growth in daily new cases.
Data source for this and other graphs of new case counts: Calculated from The New York Times. (2021). Coronavirus (Covid-19) Data in the United States. Retrieved 4/7/2022, from https://github.com/nytimes/covid-19-data.” The NY Times U.S. tracking page may be found at https://www.nytimes.com/interactive/2020/us/coronavirus-us-cases.html
Do they even have curve-balls in cricket?
Based on the count of persons who have newly tested positive for COVID-19, the U.K. seems to be recovering from its secondary Omicron peak.
Source: Johns Hopkins data via Google search.
I have yet to see anyone offer any explanation as to why. The BA.2 strain accounts for almost all cases there. Virtually all restrictions have been removed. My brother (living in London) reports that mask use is almost non-existent. None of that would explain why things appear to be getting better.
Sure, I’ve seen some nonsense. The same trick used to hype the new case counts in America was recently recycled in the U.K media. Note that sometime around March, the U.K. went to reporting their data five days a week, instead of seven. The Monday number now contains three days’ worth of new cases. Any fool can plainly see that Monday number is disproportionately large because of that. Any fool can grasp the reason: It’s not a one-day count, it’s a cumulative three-day count.
And so, of course, when one of those three-day Monday numbers exceeded the prior highest single-day count, the British press could not stop themselves from writing headlines about it. Single largest daily case count.
But that nonsense aside, I now see a lot of serious reporting that COVID-19 infection rates in the U.K. are the highest they have ever been. And for the life of me, I can’t look at the chart above and figure out how that could possibly be so.
Here’s a look at typical headlines. All the stories attribute this to the U.K.Office of National Statistics.
Source: Google news, accessed 4/7/2022
For today’s task, I decided to try to track that down. How can the U.K. Office of National Statistics go from the chart above, to the conclusion that more people are currently infected with COVID-19 in the U.K. than at any time in the past.
How in the world did they manage to go from the current seven-day moving average of about 78K newly-diagnosed cases per day, to an estimated 5M individuals estimated to have an active COVID-19 infection?
Here’s the interesting part: It’s not due to bad data for the high infection rate. It’s not due to some weird extrapolation mechanism. It’s because they have a completely separate source of data on the fraction of the population that has an active COVID-19 infection.
So, how does the U.K. government know what the actual infection rate is, as opposed to the count of individuals who have had a positive test reported to some official public health agency?
Answer: They literally test a random sample of individuals every week. Not people with symptoms, but a random sample. They do surveillance COVID-19 testing of their population.
You have to go to reporting by the BBC to get that part of the story.
Source: The BBC
Could you imagine trying to do that in the U.S.? Have the Federal government actively monitor the true infection rate by randomized testing of the population. The Republican nutters would have a stroke. But apparently this sort of thing is accepted as normal, rational health care policy in the U.K. Here, we only accept that sort of thing within private-sector organizations. E.g., my daughter’s college had randomized prevalence testing in place during the first year of the pandemic.
The results of the survey look like this, again from the BBC
Source: The BBC
For sure, those numbers appear to be a lot more consistent with the trend in hospitalizations, compared to the count of persons with a reported positive test:
Source: Johns Hopkins data via Google search
There’s maybe an order-of-magnitude difference between the count of persons with reported positive tests (very top graph in this section) and the UK ONS survey-based estimate of persons with an active infection (second graph). If the average infected person is capable of triggering a positive on a test for maybe seven days, the top graph would imply just over 0.5M persons with a detectable active infection, at any given time. By contrast, the direct measurement via surveillance testing shows about 5M persons.
Here’s the kicker. This huge disagreement between the positive-reported-test series and the surveillance (random-sample) testing series only appeared in the past couple of months. There has been a sea-change in the behavior of the two series. Here they are, lined up to the best of my (poor) graphics ability. This is just pieces of the first two charts of this section, lined up as to timing:
At least that explains why we’ve never heard of this U.K. surveillance testing, up to now. Historically, all it did was validate the trends that were apparent in the count of daily positive tests. Only since the Omicron peak have the two time series given qualitatively different trends.
I really don’t quite know what to make of that, or what lessons that has to offer for the U.S.
I do know that despite that apparently tiny number of new U.S. cases, per the count of new tests, we’ve had a lot of high-profile outbreaks. These include outbreaks from political gatherings, and both movie and broadway show casts.
My wife’s observation is that these outbreaks are being uncovered among populations that have daily testing, regardless of symptoms. And I think that’s key, because that’s exactly where the two U.K. series diverge: One is people who chose to get tested (probably for symptoms), one is surveillance testing of a defined sub-population (national politicians, Broadway show casts, and so on).
Appealing to Occam’s Razor, one simple explanation would be an increase in the fraction of new COVID-19 cases that are asymptomatic. Thus, we get a larger pool of persons actually infected, but never formally tested. A second simple explanation is that behavior toward testing is rapidly changing, now the governments are all-but-announcing the end of the pandemic, and “everybody knows” that it’s all over. Even with (mild) symptoms, people might now be more inclined to shrug it off as a cold or flu. Except those who are sick enough to end up hospitalized. Hence the hospitalization curve in the U.K. more-or-less splitting the difference between the test-positive curve and the surveillance-random-sample-survey curve.
In any case, I think the lesson for the U.S. is not the U.K. has a lot of true cases, it’s that they wouldn’t know that except for their surveillance testing. Which we do not have and never will have, as explained in the footnote above.
I’m not really sure what to make of this. Maybe COVID-19 is mutating to become even milder, with a higher proportion of asymptomatic cases. Maybe people are just ignoring their minor symptoms. Maybe the true incidence of infection in the U.S. is being revealed by these outbreaks among populations with mandatory testing.
That’s all speculation.
The closest thing we have to the U.K.’s surveillance testing is the CDC’s sero-prevalence surveys. (I.e., the fraction of blood samples, in some sample-of-convenience, that show infection-generated antibodies for COVID). But that’s always months late (the most current data are from January), and that gives some count of the ever-infected, not the currently-infected. That said, if we are undergoing the same sort of transition as the U.K. appears to be, at some point a few months from now, the seroprevalence series will begin to diverge from the count of persons with positive tests.
In the meantime, all this really means is that the U.K. data have now thrown us a curve-ball. That’s interesting, but it’s not really actionable. It just adds an extra bit of uncertainty to an already uncertain U.S. picture.
