Still Trying to Figure Out the Death Data Dump

Minnesota is far from transparent in releasing epidemic data or explaining how they count cases, hospitalizations and deaths.  The recent dump of 138 newly “confirmed” deaths from last fall had me looking again to see if I could figure out where those deaths came from.  It also ties into the mystery about how the CDC, which relies on Minnesota to send it death certificates, has a different count than the state.  Right now CDC is showing more deaths than the state reports, which seems odd given a reporting lag.  And that is a reversal of last spring and summer, when DOH was showing a higher number.  So far, the CDC number is not reflecting reporting of that 138 deaths as a new event, because the additions to the CDC total in the last couple of weeks have been pretty minor.  So were those deaths already on the certificates sent by the state to the CDC last year?  Does CDC have its own, more liberal, standard for determining that a death certificate shows a CV-19 death?

On the daily situation report page on the DOH website, under deaths there is a note which says:

“Total deaths with laboratory testing are deaths due to COVID-19 with:
A positive PCR test (confirmed case) or antigen test (probable case) for SARS-CoV-2
AND
COVID-19 is listed on the death certificate
OR
Clinical history/autopsy findings that provide evidence that the death is related to COVID-19 without an alternative cause (i.e. drowning, homicide, trauma, etc.).
CDC: NNDSS Surveillance Case Definitions: Coronavirus Disease 2019 (COVID-19)
Total non-laboratory-confirmed COVID-19 deaths have COVID-19 listed on the death certificate, but no documented positive laboratory test for SARS-CoV-2.
On 10/14/20, this category name was changed from “probable deaths” to “non-laboratory-confirmed COVID-19 deaths”.
Deaths are represented by the date reported.
There was no data reported on 7/4/20, 11/26/20, 12/25/20, and 1/1/21.
On 3/9/21, a review of unverified reports identified 138 deaths that should have been previously reported but had not been due to a failure of several private labs to report the information as required by state rule.”

Private labs don’t file death certificates, so what the state is referring to is CV-19 test results.  They appear to be saying, and in the briefing covering this death dump said, that these deaths had no matching positive test results, but were identified as possible CV-19 deaths on the certificate.  If that were the case, the deaths should have shown up in that probable category as  “non-laboratory-confirmed COVID-19 deaths”.  If that were true we should have seen a switch of 138 deaths from that category to the confirmed category, but that has not occurred.  None of this makes any sense.  If these deaths were listed as related to CV-19, they should have either been in the confirmed category or the probable one.  The fact that they weren’t reported at all seems impossible.  And people who have looked at the death certificates have pretty much come up with numbers that match the state’s total.  The most logical explanation is that the death certificates for these 138 deaths didn’t mention CV-19, but the state simply is doing positive test matching to death certificates and if someone died and had a positive CV-19 test at some point, however remote from or unconnected to the death, they are calling it a CV-19 death.  This is just one way the death toll is grotesquely exaggerated.  I have asked for clarification from the state but don’t hold your breath on that.

On a somewhat related topic, Scott Johnson asked the DOH for the week of February 20th, for each death reported on a day, tell us what date the death actually occurred on.  As part of the answer the state gave a convoluted reason for why it doesn’t do date of death reporting, claiming it was less confusing.  I think as usual they are trying to hide some of the nonsense they do.  Note for example, that on February 20th one of the deaths reported occurred 9 months ago, on May 21, 2020.  How does that happen?   Two more deaths reported that day are from October.  So out of 50 cases reported during the week, 8 came from the fall or earlier.  During the rest of the week there are five more deaths coming from October or November.  Here are the results, sorry for the awkward forward but it is a direct copy of the state’s response: