Minnesota’s drop of 17 new deaths reported yesterday prompts this quick post to demonstrate the misperceptions that causes. I have a fuller chart and table coming soon based on the ongoing comparison of what date of death looks like versus date of report. The CDC reports deaths by week they occurred. Every weekday they update those numbers by state, by week of death. The CDC data is based on death certificates they receive from the states, so there is a lag. But if you go regularly, and I tend to go on Friday afternoon or Saturday, after the full week’s updates are completed, you can see changes in death reporting by when the deaths occurred and when they were reported. So the last pull I used in my analysis was as of the end of the week of July 31. Then this past weekend I looked at the data as of August 14. Here are some examples of how far back the changes go. For weeks ending in the month of April, on the August 14th pull versus the July 31st one, two coronavirus deaths were added, and one was deleted. April! In weeks ending in May, 5 deaths were added. Six deaths were added to June weeks. A significant number of all deaths added between the two CDC reporting dates were deaths that occurred over a month ago.
The average lag from the state getting a certificate and sending it to the CDC should be a few days at most and then it should only take the CDC a couple of days to process the certificate and get it in the data set. So what is happening at the state level that we are still counting deaths from April and May? Does it really take that long for a death certificate to get from the certifying physician to the state? How many of these early epidemic but recently added deaths are changed death certificates? Is the state actively hunting to make every possible death a coronavirus death? Transparency from the state would eliminate suspicions. What is the big deal about refusing to report this way. The certificates very plainly say on them the exact date of death. So why doesn’t the state tell us the actual date of death numbers. This would be simple. For example, yesterday, the state could have said: “17 new deaths from coronavirus were reported to us today. Two of these deaths occurred on August 16, three occurred on August 14,” on so on, whatever the actual dates are. How hard would that be?
And it does matter because people think when the state says 17 deaths were reported today, that 17 people died today and we had a big uptick. They didn’t, and we won’t know how many people died from coronavirus today for a couple of weeks at the earliest, and based on the apparent processing lags, it may be months.