I have not been too interested in conspiracy theories about over or under counts of coronavirus deaths. At the same time I know that like everything else in health care, there are often errors in attributing cause of death, that government regulations can provide incentives that influence coding of cause of death and that the epidemic has heightened these issues. I have Minnesota death certificates for the past few years and if I ever get enough time I am going to do some analysis. Preliminary examination of the CV ones is enlightening. I have been expecting more research on this topic and it is starting to arrive. Two new papers are covered below.
Given the financial incentives involved, I suspect there will substantial gaming in coding of hospital stays as related to coronavirus, which may spill over to cause of death certifications. (Medrxiv Paper) Hospitals in the US are subject to a bunch of programs supposedly designed to improve their quality. I write about this frequently in my normal blog. The authors looked at attributed cause of death for diagnoses related to these programs to see how they may have changed during the epidemic. Almost every US coronavirus death, and those around the world, involve at least one comorbidity and in most deaths a comorbidity is listed as a contributing cause of death. Supposedly in the US death certificates are only supposed to list CV if it actually contributed to the death, but the guidance also says that can occur even if there was no positive infection test. The authors describe prior research finding that doctors often make mistakes on death certificates, sometimes under pressure from hospitals to avoid causes that might reduce their scores on quality measures.
The researchers note the importance of accurate reporting partly to ensure that we understand the effect of stay-at-home orders and lockdowns. They then describe some of the financial incentives which might cause a hospital to mis-attribute cause of death. Common causes of death such as heart attack, heart failure, pneumonia and stroke are used in Medicare quality programs used to determine payments to hospitals. Lowering deaths attributable to these causes results in higher payments. They also note that since so many coronavirus deaths were to older people with limited life expectancies, as those deaths occur from the virus, there should be reductions in deaths that would otherwise occur because of the other causes. So the researchers conducted an analysis on cause of death from October 2014 through June 20, 2020.
They identified three epidemic periods, January 19 to March 14, March 15 to May 16, and thereafter. 15 million deaths overall were included and 10.7 million were in the categories studied by the researchers. They found that especially at the peak and later in the epidemic, there was a shift from causes related to quality measurement programs to coronavirus cause of death. These changes occurred after the CDC guidance in April lifting the requirement of an actual positive coronavirus test. They believe over-reporting has exceeded under-reporting of coronavirus deaths beginning May 17. It has likely accelerated since the end of the study period. Noting large increases in prescription drugs for depression and anxiety, they suggested that many unassigned causes of death could be suicides or overdoses.
This paper focuses on the topic of supposed excess deaths, or those above recent years’ averages. (Medrxiv Paper) The authors looked at excess deaths for the 13 causes of death the CDC tracks as significant comorbidities for CV. They found significant excess deaths, excess meaning compared to the average of the previous 5 years, among these causes and attributed these to coronavirus. Supposedly these deaths were somehow missed. They ignore the likelihood that these deaths are attributable to the lockdowns and scaring people into missing health care. Let us look at some of the leading causes and their excess deaths. Alzheimer disease and dementia is number one, 8,608 excess deaths. Many of these people would be in nursing homes or special memory care units. They likely suffered from loss of contacts from bans on visitors and lack of staff attention due to CV distractions. Most of them likely had very short life expectancies and advance directives. They are more likely casualties of the lockdown than the virus. Ischemic heart disease, the cause of heart attacks, number two, 4741 excess deaths. These are definitely lockdown caused. We have multiple reports of massive declines in emergency room visits for heart attack symptoms and more people dying at home from heart attacks. I don’t think the virus is responsible, but the fear-mongering is. Diabetes, number three, 2944 excess deaths. Again, multiple reports of people with symptoms of low or high blood sugar avoiding emergency rooms or routine care. These deaths are due to the lockdown. Now here is an interesting one, over 1100 fewer deaths due to chronic lower respiratory disease–emphysema. Those missing deaths likely all got treated as CV deaths. So I am not buying the undercount of CV deaths BS, in fact I think we have over-attributed deaths to CV and my initial review of Minnesota death certificates suggests that as well.
As I said, we are just at the start of figuring out how many deaths really should be attributed to coronavirus.