COVID-19 April 26, 2020

Daily new cases reported on Friday departed substantially from the trend.  Yesterday's numbers are lower, but still substantially different.  I am inclined to think that a good amount of that is due to states adjusting their numbers and an increase in testing.  This is just a hunch, but as an example, the number of tests given each day in Colorado is shown below.  In the last two days, the numbers have increased by more than 2.5 times the average.  Not surprisingly then, the number of positives have increased proportionally.   I can't say whether this is true for other states, but the numbers of many other states do look suspicious.

At the same time, I do think that some states are about to undergo an increase in the rate of infections.  In principle, this increase will be balanced by a decrease in other states that have peaked.  In principle.  If too many states take their foot off the break too quickly, then it is almost inevitable there will be an acceleration of cases again.

The best metric continues to be the number of deaths.  Even these numbers, however, can be subject to uncertainty as states go back and reevaluate the cause of some deaths.  It would be acceptable if historic deaths were logged in on the day they occurred, but many states lump them into the daily statistics.  As long as those numbers aren't too great, it's not a big problem. But if New York goes back and decides that there were 8,000 additional deaths that should be attributed to COVID, and if they lump them all into next Tuesday's numbers, that's clearly going to cause problems.

Below are the graphs for today.  I've readded the exponential best fit line on the cumulative cases.  I want to see if the data begins to take off on an exponential trajectory again.  That would be bad.