We’ve reached the part of the year where we’re only going to get glimpses of the actual U.S. trend in new COVID-19 cases. Between now and the second week of January, we will observe a mix of:
- the true long-term trend of new case counts,
- the lack of data reporting on weekends,
- the even greater lack of data reporting on holidays,
- reduction in test-seeking (and case counts) over holidays,
- any actual impact of the holidays on new COVID-19 infections.
Today is a case in point. When I left this last week (Post #1326), there seemed to be a sharp uptick in cases. Now I can see that was temporary.
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 11/23/2021, 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.
The trend is still up, but at a much slower rate than appeared to be the case last week.
And that’s a good way to illustrate the potential data-reporting impacts of the holidays. Holidays? There were no holidays last week. How could holidays have possibly affected this apparent slowdown in new case growth?
In part, the trend reflects a true slowdown in case growth, in states such as Michigan and many of the New England states. But it also appears slower because I goofed: I forgot about Veterans’ Day (11/11). The non-reporting of cases for the Veterans’ day holiday, combined with the use of a seven-day moving average, provides a boost to the apparent growth rate on 11/18. That’s the date on which the under-reported Veterans’ Day case count finally moves outside of the seven-day window, leaving only the additional cases that were reported on the day following Veteran’s Day remaining within the seven-day window.
If you failed to follow that last bit in every detail, that’s fine. Suffice it to say that holidays scramble the numbers. We’re entering the holiday season. And so the numbers are going to be scrambled for the next few weeks.
We’re in the same situation as last year, absent the hindsight that allowed us to fill in the underlying trend after-the-fact. Below is the graph of the U.S. new case rate, from last year’s holiday season. Imagine trying to guess what the actual trend would be, at each step of the way. That’s where we are for the coming holidays. The two large “dips” in the line below — known only after the fact — are the large data reporting artifacts of Thanksgiving and Christmas/New Year’s.
The 2021 winter wave and popular press reporting.
First, I want to continue to highlight the behavior of ID, MT, WY. These were leaders in the 2020 U.S. COVID-19 winter wave. But this year, they appear to have peaked at the end of the summer Delta wave. As yet, there’s no indication that they’ll have a winter wave this year. I don’t know whether that’s a harbinger for the rest of the country, or whether they’re just a little late getting started. Or whether it’s just something unique to those states, this year.
The South seems to be split along temperature lines. States bordering on the Mid-Atlantic region and Midwest are showing slight upticks. But south of Virginia on the Atlantic Coast, right on through the Gulf Coast, there’s no apparent increase in cases. (This isn’t very different from last year, where those regions had late peaks of the winter wave.)
Finally, here’s the U.S., comparing the first and second pandemic years. Aside from being a couple of weeks late, and missing part of the Mountain states, we still seem to be on track for a winter wave.
If you’ve stuck with this post, this far, you’ll probably understand why this graph above — the one that compares this year to the same time last year — is the only one worth looking at for the next six weeks or so. The hope is that this year’s holiday “artifacts” are about the same as last year’s. Which means that the only hope of making sense of this year is to compare it to the same period last year.
I’m guessing that most popular press reporting isn’t going to bother to do that. But if you don’t see the comparison to last year, you really can’t make head-or-tail out of what the most recent trend has been.
For example, we should now expect to see a substantial decline in the new case counts, owing to the Thanksgiving holiday. (Which never generated a “surge” in cases, despite what you may have read to the contrary (Post #1324, Winter Wave Buzzword Bingo)).
By itself, that decline is meaningless. Only in the context of last year can we start making some judgments about where the trend is actually heading. Look for the popular press reporting that does that, and you’ll have put your finger on the people who understand how to make sense of the numbers. Ignore the rest.
