For the last two weeks (ending September 26), there was very modest improvement for all COVID metrics, except new cases. There was a nominal increase in new cases, but as I discuss below, the increase is not a harbinger of a second wave, at least, not yet.
The number of COVID patients hospitalized, in ICU units and on ventilators fell by 4%, 6% and 7%, respectively. Total hospitalizations have really flatlined over the last two weeks.
Average daily fatalities increased slightly in Week 38, before falling again in Week 39. We seem to be stuck at a range of 700-800 daily fatalities. If fatalities continue at this rate, the US will see another 70,000-75,000 deaths by year-end. The widely followed IHME model is projecting more than double that number, which would require fatality rates we have not seen since May. However, they have recently been lowering their projections and I suspect they will continue to do so.
A narrative has emerged in the media in the last week that new cases had begun to rise, which was widely heralded as the beginning of a new wave of COVID infection. Graphs like this one from the Wall Street Journal were widely touted by the media as evidence of the new wave.
But what most of these stories failed to mention was that the increase in new cases on the far-right side of the chart was the result of a drop in testing around Labor Day weekend and then a significant acceleration over the last ten days. Daily tests exceeded one million per day three times in the last eight days. The US had never tested one million people in one day before.
The week before Labor Day, daily testing averaged 788,000. But the week of Labor Day it fell to 712,000 then rebounded the following week back to 788,000. The surge in testing last week took the average to 897,000 per day.
New cases tracked this pattern of testing variation. The week before Labor Day there were 285,000 cases, 240,000 the week of Labor Day and back to 280,000 the next week.
But here is the important point, the positivity rate remained slightly over 5% for the entire period. In order words, the supposed surge in new cases was really driven by a lag in testing over the Labor Day holiday, followed by an acceleration in the following two weeks.
Adding the context of test volume produces this chart, clearly showing news cases and tests moving in concert.
Also, we should always keep in mind that the “new case” metric is, by far, the least reliable indicator. The problems with equating PCR test results as being “new cases” have been well documented. Also, recent antibody tests suggest that we are only diagnosing 10% of the actual cases. When you are looking at a small sliver of the total picture, small errors or anomalies are geometrically magnified.
The last two weeks suggest that we have reached a level state on transmission of the disease, at least for the time being. It is disappointing that the improvement in the metrics did not continue at earlier pace but it is encouraging that there is not yet any sign of a fall surge from Labor Day or schools reopening.