Series 4: Summer upticks and beyond…

Data last updated: March 5, 2021 (Figures 4-5, 4-5.b and 4-6 only)

Figure 4-5 (Winter 2021 )is currently the only chart that remains to be updated on this page. All others are archived and presented for comparison purposes. A detailed discussion on the purpose of these charts follows this brief discussion on Figure 4-5.

The span of data for Figure 4-5 extends from December 1, 2020 to March 1, 2021, or basically the winter of 2021. It shows that after a modest increase in the number of new daily tests for COVID-19, the trend of new daily testing dropped until late February.

Prior to approximately January 14, the rate of new daily COVID-19 cases increased slightly over the rate of new testing – that is, the United States was still seeing an increase in winter infections which continued a trend from the autumn, albeit with a lowering rate.

Then in early-mid January, the rate of new testing each day began to drop. So did the rate of new COVID cases – at a much greater rate of drop. So, even though we were testing less people on average daily, we were seeing far fewer new cases on a daily basis.

Recently, it has made news that the rate of a drop in new cases has leveled off. And this chart shows this to be true from about February 21. But the rate of testing has also picked up considerably since February 21 and the rate of new case, though leveling, does not correspond with the rate of new testing increases yet.

****

Figure 4.5.b presented below shows the same data as the chart above. The key difference here is that the y-scale has been cropped to a maximum change of +75% since no y- data point in this period exceeds that value. The reason the y-scale in the figure above extends to +400% is that Figures 4-2 through 4-4 did have y-values that required a higher maximum value y-scale and I wanted to keep it similar for each season for easier comparison views.

Here, in Figure 4.5.b, it is easier to see that the positivity rate (the blue line and bars) has been gradually but steadily dropping since early December from roughly 11% to under 5% in early March.

****

Finally, I’ve added a new chart, Figure 4-6, as we’ve moved into meteorological spring. This chart will likely not extend beyond March 7 since The COVID Tracking Project will no longer update the data source files beyond that date. I could’ve just added the first seven days of March to the tail end of Figure 4.5. But I wanted to rebase the percentage change to zero for February 21 in order to track the recent flattening of the new COVID cases curve.

So, is the recent rise in COVID cases in the U.S. due to another surge. Or might it be mostly due to a recent increase in testing. The data is very limited, but unlike last summer it appears the rise in cases may be due mostly to a corresponding rise in testing.

Look at Figure 4-6. Initially, from February 21 through about February 27, the rise in new cases parallels the rise in new testing. But, after February 27, the rate of testing continues to increase, while new cases have begun to drop again.

Also, look at the positivity bars and curve. They essentially remain flat. If the rise in new cases was caused by a new outbreak rather than new testing, one would expect a subsequent rise in positivity rates as well. So far, that does not appear to be the case. Of course, this is less than two weeks worth of data.

Earlier this summer, around late June or early July there was a strong uptick in COVID-19 cases. Of course, now we can look back at that time from a distance and see it was a second wave of the pandemic spreading, this time more in the South and Southwest rather than the Northeast, which received the harshest of the first wave. And now that we’re in the midst of a third winter wave which is far greater in the number of cases and its spread, much has been forgotten about last summer’s uptick.

At the time, though, there was a minor controversy going on in the nation about whether this second wave was really a wave or not. The administration claimed it was due to an increase in testing, saying the more you test people the more positive cases you’ll see. And to a small extent they were correct – testing had increased during this period, significantly since the previous winter and spring. But hospitals, especially in the South and in Texas were seeing much higher admissions and this increase in patients was not solely due to increased testing.

One way to look into the matter graphically was to chart the percent increase in the amount of testing and compare this to the percent increase in the number of positive cases. And to compare these two lines after rebasing the results to zero (Wong, 2010, [1]). This is a common practice when charting comparisons of stock gains over time whose price for a single share of stock may differ greatly between stocks.

So I did this last summer to compare results. I used a vertical bar chart to compare the daily percent change in testing to the daily percent change in positive test results from a common zeroed-baseline of June 1, 2020. The changes were tracked from June 1 through July 31. I also added 7-day smoothing curves to each data set.

These charts were originally published on this site blog around August 1 under the former numbering schema as Figure 40 and Figure 41. The original plots are shown below in updated Figure 4-1.

These two plots showed:

1. yes, daily testing for COVID-19 had increased during this period by about 70% from a baseline on June 1;

2. new daily cases of COVID-19 had increased by over 130% over the same period; and

3. the surge in COVID-19 cases began roughly around June 16. This is where the slopes of the two curves really begin to diverge.

Therefore, for the administration to insist consistently that increase in daily case rates were simply due to increased testing as they were doing at the time was wrong, at best.

Figure 4-1. The original New Testing vs. New Cases Rebased-to-Zero curves published in August 2020.

An even great uptick this Autumn; and now Winter…

Since these original rebased curves were published, I’ve decided to update them into four distinct periods: Spring 2020 (Mar-May); Summer 2020 (Jun-Aug); Autumn 2020 (Sep-Nov); and Winter 2021 (Dec-Feb).

The first three of these are shown below in Figures 4-2, 4-3 and 4-4. The fourth seasonal rebased curve is just beginning to bear fruit and it is shown in Figure 4-5.

..

Figure 4-5 was last updated on March 1, 2021. It shows since the start of December, the rate of new positive cases to the rate of new testing has remained roughly steady until about January 12, 2021. This is the point where new daily positive cases have begun to drop quicker than any corresponding change in rate of daily testing. This is a small, albeit possibly temporary, bit of good news.

Since the first of December, the amount of daily testing has risen and fallen slightly several times over the past 2-½ months. The level of daily testing now is roughly the same as on December 1, 2020. But the number of positive test results has dropped by quite a bit. It is currently about half as what it was on December 1.

Click on any chart to enlarge it.

Testing Rates and New Cases: 2020 Holiday Season (Figures 4-7a thru 4-7c)

Three charts are shown below: Figures 4-7.a, 4-7.b and 4-7.c. Each chart has had its respective bar chart and smoothing curves rebased to zero on October 1, 2020. Each chart’s timeline currently extend to January 20, 2021. Data for these charts is not expected to be updated beyond January 20.

• Figure 4-7.a shows the percentage change since October 1 for two data sets. The bars show the daily percentage change from October 1 for both new testing and new cases. The line plots for each set of data bars are smoothing curves based on moving 7-day averages.

• Figure 4-7.b shows the percentage change since October 1 for two data sets. The bars show the daily percentage change from October 1 for both new testing and new cases and is the same as those bars in Figure 4-7.a. The line plots for each set of data bars are smoothing curves based on moving 15-day averages.

• Figure 4-7.c shows the percentage change since October 1 for two data sets. The bars show the daily percentage change from October 1 for both new testing and new cases and is the same as those bars in Figures 4-7.a and Figures 4-7.b. The line plots for each set of data bars are smoothing curves based on moving 30-day averages.

Observations:

1. Taking a step back to the start of October, well before the holiday season, the percentage of new cases has been increasing at a much higher rate of change than the rate of new testing.

2. Figures 4-7.a and 4-7.b show a second bump in the increase of positive cases over the increase in testing beginning about a week after Thanksgiving. The rate of increase in new cases is clearly rising more quickly than the rate of new testing.

3. Figure 4-7.c uses a 30-day smoothing curve and a post-Thanksgiving bump in new cases is less obvious. The curve appears to be more of a continuum of the 4-6 weeks preceding Thanksgiving rather than a second ‘bump’ in positive cases.

4. All three curves show a leveling of the rate of testing and new cases from approximately mid-December, though at the tail-end of December the rate of new cases seems to be outpacing the rate of new testing. This my be due to (a) a brief lull in testing during the holiday break combined with (b) latent positive test results from earlier in the week still coming in and being accounted for.

covid, covid-19, new testing, new cases, USA, rate of change, seasonal, positivity, Thanksgiving, Christmas

Reference:

[1] The Wall Street Journal Guide to Information Graphics; W.W. Norton & Company, Inc.; Wong, Dona; 2010; pp. 104-105.