On November 15th, this blog presented three possible scenarios for the progression of the 3rd wave. This analysis looked at the two previous waves and observed three separate modes in the line graph of the numbers of daily new cases, with the general idea being that the U.S. would begin getting the third wave under control when decreasing numbers of daily new cases were observed. The method used was analysis of the slopes of the daily new cases curves. This blog follows up on that analysis.
Daily New Cases and 7-Day Average
At the time of that analysis, it was hoped that the latest data available at that time, through Nov 13, would be the end of the Phase 1 mode of the line slope - the period during which the slope was increasing. This turned out to be exactly right. A peak increase in the change of the daily new cases number occurred on Nov 13 (5,724), and the change in the daily new cases numbers decreased thereafter. The case numbers were still going up, but the steepness of the increase was not as great.
Slope of New Daily Cases Line To generate this chart, the raw numbers of daily new cases were collected from covidtracking.com on Dec 12, 2020. Then a 7-Day average of those numbers was computed. The difference in that 7-day average from one day to the next was computed - this is the steepness of the line: when it is greater than 0, the daily cases are increasing; when it is 0, the daily cases are accumulating, but by the same number each day instead of more each day; when it is negative, the number of new daily cases is decreasing each day. The blue line here is the change in the 7-Day average of daily new cases; the orange line is the 7-day average of that change--the average slope over the past week.
The holiday period of Thanksgiving interrupted the flow of data on the actual progress of the disease. We can see that disruption in the chart above, as the slope briefly goes negative, followed by a very large increase to 'catch up' on previously unreported data, observed on Dec 3. The original analysis of scenarios presented here in the Nov 15th post did not take into account the Thanksgiving disruption. Now, two weeks after Thanksgiving, we can see that the Thanksgiving period delayed but did not prevent the continued spread of the COVID-19 third wave. The 'Minimal Scenario', considered highly optimistic even as it was posted, would have seen a peak on Dec 1. Even with a delay from Thanksgiving accounted for, that scenario has been decisively disproven by the data since then. The middle scenario hoped for a peak by Dec 11. If we add a delay of a week to 10 days caused by the Thanksgiving disruption, that scenario may still be possible. Here we would hope for a peak, a slope of 0 in the daily new cases, sometime between Dec 18 and Dec 21. Our present question remains: What has happened post-Thanksgiving? We can definitely say that the pandemic shows NO signs of coming under control up to the present time (data complete through Dec 11). We ignore the one-day very high blip of data on Dec 3, but even discounting that data point, we see that the average slope of daily new cases has reached an all-time high, on Dec 9 (6,169). This is consistent with a look at the actual data tables, where we see raw daily numbers > 200,000 on each of the past four days, and a 7-day average of daily numbers > 200,000 on each of the past four days: The all-time record high for both of these measures occurred on the most recently available day.
DATE DAILY 7-DAY AVG OF DAILY
From this, I conclude that the progress of the third wave has undergone a restart: for about the past week, we have re-entered Phase 1, during which the slope of the line is *increasing*, meaning that COVID-19 is spreading more quickly on each day. This is unfortunately consistent with the clear and persistent warnings of epidemiologists and medical professionals about the consequences of failing to observe safe practices during Thanksgiving. It remains to be seen how this will or will not be reflected in hospitalization. This blog published a timeline of calendar dates to watch, as indicators of the effects of a Thanksgiving surge; our next deadline arrives on Dec 14.
We can compare our most recent period, showing a high rate of increase, to a period *before* the Thanksgiving disruption, to see whether our current surge is a continuation of the previous rates of spread, or is instead even worse. The latter case would be consistent with a worsening of the pandemic in the U.S. attributable to Thanksgiving-period spreader events. The following table of data shows this comparison.
Our periods of comparison are BEFORE Thanksgiving, Nov 18 - Nov 24, to AFTER Thanksgiving, for the most recent 7-day period. Highlighted in orange we also calculate a broader average, going all the way back to Nov 1. The green highlighted lines show the numbers to compare. The week before Thanksgiving was very tough, but the most recent week was even worse. Also notice that in the entire period since Nov 1 BEFORE Thanksgiving, the highest 7-day average increase (steepest slope) was 5,724. But now we have two days AFTER Thanksgiving with even higher numbers, on Dec 8 and 9. Clearly, this is our worst week ever, on multiple measures. We definitely have not peaked. We have restarted Phase 1 of the third wave, the period of accelerating spread. It is still possible, as new control measures are instituted, that we could bring the slope to 0 by Dec 18 - Dec 21. This would be a reset of the 'Middle Scenario', taking into account the Thanksgiving holiday data disruption.
In order for that to be sustained, our aggregate behavior during the Christmas holidays would need to be very different from that observed over Thanksgiving. An additional spreader event and a post-Christmas surge would bring us back into Phase 1 for yet a third time, guaranteeing that the hospitalizations and deaths to follow would be even higher than we are expecting right now.
All opinions expressed in this blog are solely those of the authors, and do not reflect those of University of Texas Rio Grande Valley, or any organizations of which either is a member.