Recently, I wrote about COVID19 Vaccine Deaths & Risks. After writing the original article, I was informed one of my premises were inaccurate and needed correction.

The issue came about from a lack of data. I appreciated those on Twitter who were willing to share their data. The additional data enabled discovery of the error I made. It was not initially apparent that the deaths in the CDC dataset: Weekly Provisional Counts of Deaths by State and Select Causes would change over time. Particularly, the deaths in the ICD10 R00-R99 code range (which are effectively “undetermined cause of death(s)”). The documentation is here, which I had not read prior to writing the prior article and suggests using the ICD10 R99 code until COVID19 tests are preformed.

With that in mind, the data we’ve collected is interesting and valuable in its own right. It is now possible to see how the CDC has been updating / changing the underlying cause of death numbers! It’s important to note, the underlying cause of death is likely updated after investigation or tests are completed.

Methods to Monitor Data Changes

Before analyzing the data it’s important to understand how the data and charts are representing said data.

The datasets we have contain multiple snapshots or data pulls from the CDC dataset: Weekly Provisional Counts of Deaths by State and Select Causes. The underlying causes of death are updated weekly and are tracked via ICD10 codes. This particular dataset only contains a select few causes of death (ICD10 codes) and omits causes of deaths such as suicides. It’s important to note, there can also be multiple causes of death.

In way of an example, we can pull the deaths from All Causes from January 2, 2021 through January 30, 2021 on two dates (June 16, 2021 and June 9, 2021) and the count will differ slightly.

Week Ending Date	Deaths, All Causes Pulled 6/16/2021	Deaths, All Causes Pulled 6/09/2021	Deaths, All Causes Difference Between Pulls
1/2/2021	81018	86661	-5643
1/9/2021	80795	86158	-5363
1/16/2021	80717	86000	-5283
1/23/2021	77058	82118	-5060
1/30/2021	73435	78480	-5045

What this tells us is that between the two data pulls there was a reduction in the deaths from “All Causes“. In a chart, this is shown as a negative number of deaths.

Deaths could have been double counted in June 9, 2021 data pull, but not in the June 16, 2021 data pull. Or some other reason, there’s no way to determine without contacting the data owner.

ICD10 R99 Conflation

Monitoring data changes will help explain how the R00-R99 codes are being re-categorized. This is important, as the code is both used for truly unidentified reasons of death, but also per the CDC the R99 code should be used until a COVID-19 test is performed.

If “pending COVID-19 testing” is reported on the death certificate, this would be considered a pending record. In this scenario, NCHS would expect to receive an updated record, since the code will likely result in R99. In this case, NCHS will ask the states to follow up to verify if test results are confirmed that the decedent had COVID-19.

Unfortunately, the final death counts may not be determined for 6-12 months, with most counts being reasonable accurate / stable after 12-16 weeks. This implies we can’t actually determine if all the R99 codes are being attributed to COVID-19 or not. For instance, the tests can come back negative and a different underlying cause is reported. COVID-19 Deaths appear to be very delayed in terms of reporting, as it is.

COVID-19 Multiple vs Underlying, Cause of Death

One of the most interesting items is between May 19, 2021 and June 16, 2021 the dataset appears to have relabeled thousands of deaths labeled “COVID-19 as Underlying Cause of Death” (removed 47,784 deaths, ~10% adjustment) to “COVID-19 as One of Multiple Causes of Death” (added 60,299 deaths, ~10% adjustment). This is relabeling data all the way back to the beginning of the pandemic.

What does this mean? It means that it’s possible there was a significant number of deaths incorrectly attributed solely to COVID-19. I’m not sure why this occurred, but I suspect it’s due to some change in reporting.

COVID-19 should be reported on the death certificate for all decedents where the disease caused or is assumed to have caused or contributed to death
– New ICD code introduced for COVID-19 deaths – Alert-2-New-ICD-code-introduced-for-COVID-19-deaths

At time of writing, I cannot find why this update has occurred, but it’s a fairly large adjustment to the numbers.

Symptoms, signs and abnormal clinical and laboratory findings, not elsewhere classified (R00-R99)

The deaths associated with ICD10 Codes R00-R99 as noted previously, peak with every data pull. It also appears the magnitude of R00-R99 ICD10 codes related deaths appear to be increasing over time.

Adding in a selection of underlying causes you can see the R00-R99 ICD10 codes peak closer to the current date. At the same time, other assorted underlying causes for death show a decrease (and a delay in reporting):

Where do abnormal clinical and laboratory findings (R00-R99) Go?

In terms of answering the question “where are these deaths reclassified to?” We can look at data from September 26, 2020 through November 14, 2020. This data is both pre-vaccine and at least six months old.

Even visually, it’s clear deaths associated with R00 – R99 ICD10 codes are being reclassified primarily as COVID-19 and diseases of heart. If we look at the more recent data, we want to only use data pulls from at least 12-16 weeks after the deaths, as such we will only use data pulls after March 3, 2021.

Unfortunately, in this data, it’s not clear where the deaths associated with R00-R99 ICD10 codes are being re-categorized as. I appears the leading causes are COVID-19 and diseases of heart, but as deaths are still being added it’s not clear which categories they’re being reclassified as. By reviewing only later update(s) from May 19, 2021 and June 16, 2021, we can get a single snapshot (as opposed to a sum of all differences).

In this chart, it is clear there are a significant number of deaths associated with R00-R99 ICD10 codes are being re-categorized as diseases of heart. The number of COVID-19 deaths also don’t appear to be changing in this update, in fact R00-R99 related deaths appear to be primarily being changed from R00-R99 related deaths to diseases of heart.

Implications for Reclassifications

If we review only at the June 16, 2021 data pull and that trend continues, that would indicate an increase in diseases of heart.

At this point, we will still have another three to four weeks to identify exactly what is going on here. January 2021 appears to have a seasonal increase due to COVID-19, correlating with a rise in all causes of death. The question, is whether or not the increase in diseases of heart continues to be high.

Currently, fatalities related to diseases of heart are running one to three thousand higher than in the summer and early fall of 2020. If the deaths associated with ICD10 R00-R99 continue to be changed to diseases of heart (seems likely, given the decreasing COVID19 cases) that would be somewhat alarming.

Conclusions

In terms of conclusions, I believe the following to be true:

1. We cannot make any absolute claims
2. It takes 6-12 months for the underlying causes of deaths to be finalized
3. The number of weekly provisional counts of deaths are roughly stable around 12-16 weeks, after the given date
4. June 16, 2021 the CDC re-categorized 10% of the COVID-19 deaths (~10% from sole cause to multiple causes)
5. Prior to December 2020, many of the R00-R99 ICD10 code associated deaths were changed to COVID-19 (#1) and diseases of heart (#2) related deaths
6. When comparing the latest data pull June 16, 2021 to May 19, 2021 it appears most of the R00-R99 categorized deaths are primarily being changed to diseases of heart (#1) and COVID-19 (#2)

Hopefully, in the coming weeks we can continue to monitor this data. Assuming the data is accurate, it should be possible to determine if there are any strange abnormalities in the data.

If you wish to review the data, I have copied and added to the data provided by @piercedgeek on twitter.

Related Content

• COVID19 Vaccine Deaths & Risks
• Delta Variant Less Dangerous Than Flu
• COVID19 Delta Variant to Become Dominant Strain … Lockdowns?
• Firearms by The Number
• On Committing Suicide