Post #941: No peak yet in new COVID-19 cases.

Posted on January 7, 2021

Source:  Calculated from NY Times Github COVID-19 data repository, data reported through 1/6/2020.  Thin lines are six broad geographic areas, thick blue line is the US average.  Circled areas are artifacts of Thanksgiving and Christmas/New Year’s holidays.  

At this point, enough of the data reporting anomalies from the holidays have (arguably) passed that I can now start to talk about trends again.  That’s not 100% certain, but the remaining data reporting issues are probably small enough that they don’t obscure the underlying trend.

And, as near as I can tell, the trend remains up.  Somewhat.  There is no peak, yet, for the U.S. coronavirus third wave.  Right or wrong, the continuation of the upward trend, marked by a question mark above, will be referred to as the post-Christmas surge.  But the underlying picture is complex.

Data issues remain, muddying the picture.  Not to belabor it, but it looks like several states decided to “clean up their books” with the new year, resulting in large one-day upward adjustments to the case counts.  Most of these occurred circa 1/3/2021.  Those adjustments will be affecting the estimated trends, somewhat, for the next week or so, due to the use of a seven-day moving average to smooth out the data.

The largest of these occurred in New Jersey, where the case counts in the NY Time data repository jumped about 10% on 1/4/2020.  Those counts no longer match the official NJ counts, and there is no notice as to why on either the NJ state dashboard or the NY Times data repository.  Best guess, NJ probably decided to add cases identified via antigen testing to their counts, similar to what ND and some other states have done (Post #912).

Several other states (e.g., Arizona) have large one-day case count increases on or about the same time.   Those are of a size that you can’t tell whether that’s some sort of one-time adjustment, or whether that’s just catching up with the state health department offices being closed for some portion of the holidays.

I say all of this because a) this is a post about the case counts, b) you’ll see some implausible changes in the graphs below, and c) those are arguably large enough, in aggregate, to have a visible effect on estimated national trends.

Regional analysis.

Above:  California, going sideways.  California has not peaked, under the criteria I’ve been using.  Instead, they’re just sort of going sideways.  Daily new case counts remain right at 100 cases/ 100,000/ day.  No telling what’s going to happen next, but at least it’s not going up.

Above:  Midwest:  Still in lockstep, but no longer improving.  If there ever were an argument for weather playing a dominant role in the evolution of this pandemic,  the Midwest states would be exhibit 1.  Plus or minus a little raggedness in the data reporting, every one of them shows an upturn — but just a slight upturn — following the holidays.

Let me emphasize that this clearly isn’t the result of policy, or the result of a huge mingling of these populations across states.  But for some reason, they remain in lock-step.

A lot of these states plausibly have large portions of the population that have already survived a coronavirus infection.  Given that, it doesn’t seem likely that they could return to their previous peak new case/day levels.  That said, they have ceased their rapid reductions in new cases/ 100,000/day, for the time being.

Above:  Mountain states break up.  Arizona (top line), Utah, Nevada show sharp increases in new case counts.  The rest of them look like the Midwest — the prior downward trend in cases counts has stopped.

Above:  South Central, South Atlantic, and Northeast states mostly show modest upward trends.  Spike on the NE chart is New Jersey data reporting anomaly, discussed above.