As promised, I’ll continue to report at the national level from time to time using CDC data. I’m also reporting on NC, my home state.
The CDC data is laborious to work with, and it seems that there is no data oversight or cleansing taking place. Here is just one example:
On March 8th, the CDC reported 81,806 new cases for the state of Missouri. This was totaled into the national numbers, and you’ve all seen the large bump in daily new cases that day in the news. However, it obviously makes no sense. MO has reported less than 600,000 total cases since the inception of COVID tracking. To add to the mystery, the CDC reported 81,474 of these new cases as “probable”. Looking at the state of Missouri’s spreadsheet, they show 1,585 probably cases for the entire week of March 4-10. Suspiciously, Missouri’s spreadsheet shows 82,685 probably cases over the entire past year. I don’t know what MO reported to the CDC, as the various state reports are not public, but it sure seems like a data entry error.
Sadly, Infection2020 still shows nearly 100,000 new cases on March 9th, and Johns Hopkins shows exaggerated numbers for both March 8th and 9th, although by less than 20,000 over the 2 days.
Oh, how I miss The COVID Tracking Project. Their team of analysts would have been all over this. In any event, I did adjust for the Missouri error, but I can only catch the obvious errors – I’m sure there is more that I’m not discovering.
In other news, it seems that mainstream reporting is catching on to the fact that social restrictions were not all that they were hoped to be. Here is an article yesterday from the AP: Virus tolls similar despite governors’ contrasting actions (apnews.com). I did some analysis on this back in May of 2020, comparing the results in VA and NC, coming to a similar conclusion – it’s pinned below.
As of yesterday, about 70 million Americans have received at least the first dose. That’s about 27% of the adult U.S. population. More importantly, about 2/3 of those 65 and over have had at least one dose. By my calculations, we need between 80 and 100 million to reach herd immunity (see my discussion from February 1st below). At the current rate of vaccination, we’ll get there by the end of March. Add another 2 weeks for the vaccinations to reach a level of effectiveness, and we’ll be at a substantial level of herd immunity.
The daily death count continues to fall. I’m modeling that daily deaths will be about 1,045 by the end of the month, down from a peak of over 5,000 daily deaths on February 12th. The latest IHME model (updated March 11th) is now slightly more pessimistic, showing 1,137 daily deaths on March 31st. However, they still show daily deaths dropping to less than 100 per day sometime in June.
Here is the graph of daily new cases. We’ve dropped over 50,000 active cases since my last report. As of yesterday, I’m modeling 352,256. This is a large 80% drop from the peak about 2 months ago.
Below is the national daily death count. The brown line is my daily death model, in which I have a high degree of confidence, at least directionally. As expected, the daily death count has fallen rapidly over the past month. Actual deaths are lower than this, as I haven’t done the work to back out the reported March deaths that are actually occurred in December and January.
Here is the graph of daily deaths per 1,000 known active cases, still remarkably stable, and a good predictor of daily deaths over the next 3 weeks.
It’s beyond the time I have in a day to track multiple states without good raw testing and cases data. I will, however, continue to track North Carolina. Here is what it looks like as of yesterday:
North Carolina is down an amazing 86% from the peak numbers on January 8th. To put it in perspective, NC now has a similar active case count to that of June 2020. It will be nice to watch this continue to fall.
Here is the new reported case graph for NC. Steady drop…
And daily NC deaths. It’s nice to see the numbers improving…
So that’s it for today. I’ll report again in a week or so.
–Shane Chalke, FSA