I have mentioned obtaining the Minnesota death certificates for YTD 2020 and the years 2016, 2017, 2018 and 2019. My primary objective was to examine characteristics of alleged coronavirus deaths and to compare trends in 2020 with those from earlier years. This is a bit of a daunting task, largely because as I suspected, there is enormous variability among the certifiers. That is going to make some of the core analysis I hoped to do, like did these people really die from CV, almost impossible. I obtained the files in mid-July and there had been almost 26,000 deaths in Minnesota at that point. Around 1300 were attributed in some manner to coronavirus. It now seems simplest to me to break this into digestible pieces. Many of the deaths in Minnesota have occurred among nursing home residents. The state, appropriately I think, lumps all congregate care settings into one death number. At this point, out of 1806 deaths, 1331 have occurred among residents of these long-term care settings, or 74%. Nursing home residents are the largest subgroup of these deaths. The deaths certificates don’t use quite the same approach to categorizing where people live, but do have a category of nursing home residents. The table below summarizes deaths by cause of death and place of death among nursing home residents from 2016 to 2020, up to death certificates processed as of mid-July. To compare the earlier years, deaths as of July 16 from those years were used. This likely is a slightly flawed comparison to 2020, because of a lag in reporting.
You can see that there were 710 CV deaths in this group. One of the vagaries of the death certificate reporting is that the certifiers basically decide whether to list something as an immediate cause, one of the intermediate causes, or the ultimate underlying cause. There is a part 2 where the certifier can add other contributing causes, and certifiers often add almost every condition the person had in this space. There are four lines in total, a, or the first one, is supposed to be the immediate cause, b, c and d, are supposed to be the underlying cause, so theoretically all death certificates should have at least two causes listed, even if the immediate and underlying cause are the same, but they don’t sometimes. The prototypical CV certificate, with no contributing pre-existing illnesses would be something like immediate cause, line a, was acute respiratory distress (like you might get from an overdose of fentanyl), line b would be pneumonia, line c would be coronavirus.
If you look at those 710 CV deaths among nursing home residents, you see the scatter. There may be some overlap here–certificates with CV on two lines. You have to go through the certificates almost one-by-one to ascertain this. Other observations include that something must have happened in 2018, there was a big jump in deaths among nursing home residents. I am not sure that 2017-18 was a bad flu season, but that would be the most likely explanation. Consistent with my belief about the role advance directives must be playing, you can also see that the biggest pickup in 2020 in terms of place of death is at home. Many of the coronavirus deaths obviously never went to the hospital. In terms of immediate cause of death, the numbers are fairly stable. The lockdown deaths are not largely occurring among nursing home residents–we will see those in the middle-aged and elderly patients living at home. Thanks to Lisa Jansa for doing most of the analytical work on this. I am an Excel rube.