Excess Mortality, Thread:
I have just discovered a major error in the #CoVid data of the NCHS Mortality surveillance data. This is all based on the official NCHSData49.csv as downloaded from here (original downloaded 12/15):
https://www.cdc.gov/flu/weekly/weeklyarchives2019-2020/data/NCHSData49.csv
The data begins in week 40 of 2013 and ends week 49 of 2020. The data looks basically like this:
Yr | Wk | All Deaths | Pneumonia | Influenza | COVID | PIC*
*PIC = Pneumonia, Influenza, or COVID deaths (sum of)
The first COVID death listed was 2020 Week 6. Pneumonia death (P) = 3799. Influenza death (I) = 520. COVID death (C) = 1. PIC = 4320.
Math check: 3799+520+1=4320 ✅
The math adds up fine until 2020 Week 8. P=3699. I=566. C=4. PIC=4268.
Math check: 3699+566+4=4269 ❌
Here's this problem: Deaths are being classified as PIC deaths, as well as P/I/C deaths. So, in 2020 Week 8, someone was classified as a PIC death AND a P or I death, also.
Someone was counted twice.
But, it gets worse. As the COVID deaths accelerate, the double-counting does, too.
2020 Week 9 - 5 fewer PIC than P+I+C
2020 Week 10 - 19 fewer PIC than P+I+C
2020 Week 11 - 29 fewer PIC than P+I+C
2020 Week 12 - 257 fewer PIC than P+I+C
By 2020 Week 13, the NCHS was double counting 1479 excess deaths.
The following weeks vary from a low of 1548 excess deaths to a high of 7550.
By my calculations, there were 121,888 deaths double counted as COVID and as either Pneumonia or Influenza.
So, what do we do with this knowledge?
My assumption here is that few (if any at all) of these deaths were both COVID and Influenza. Likely, these deaths were COVID / Pneumonia. It would seem reasonable that a death could be from Pneumonia caused by COVID.
Or, perhaps some people died of Pneumonia AND had also tested positive for COVID, thereby were counted as both (even if the cause was the Pneumonia, alone).
So we have two likely scenarios, either (or both) could be true.
One, we have 121,888 fewer Pneumonia deaths.
Two, we have 121,888 fewer COVID deaths.
Possibly, we have a split of both fewer Pneumonia and COVID, but without individual case details, that may never be known.
However, I have discovered a strange anomaly elsewhere. In trying to verify numbers, I found the "NCHS Mortality Report for the Week Ending December 5, 2020 (Week 49) Data as of December 10, 2020"
In this chart, they list Total Deaths, Influenza deaths, COVID deaths, and PIC deaths. These numbers all align with the previous chart. But Pneumonia deaths are not listed.
So I check my history to find the original link to the COVID data. It isn't there.
What do we do with this information?
Well, in reviewing the data, the 2020 Week 39 can be safely compared with the Week 39 data from prior years. Since COVID data for those years will be zero, we can make an apples-to-apples comparison of Pneumonia and Influenza.
Week 39 data:
2014 - P=3017; I=6
2015 - P=3057; I=6
2016 - P=2986; I=11
2017 - P=2888; I=19
2018 - P=2814; I=14
2019 - P=2780; I=16
2020 - P=4715; I=3
We can also compare the Week 40 through Week 39 totals of each "Season":
2013-'14 - P=182,691; I=4485
2014-'15 - P=193,237; I=8197
2015-'16 - P=178,002; I=3448
2016-'17 - P=179,621; I=6954
2017-'18 - P=180,137; I=15,620
2018-'19 - P=168,608; I=7175
2019-'20 - P=263,791; I=9415
It appears evident (to me) that we have found our error. Of the 121,888 duplicated deaths, it would seem reasonable that these were additional Pneumonia deaths who ALSO had COVID.
It would appear that we have excess Pneumonia deaths over the standard yearly average.
And it would be reasonable to assume that these additional Pneumonia deaths, who also had COVID, were likely COVID-induced-Pneumonia, rather than stand-alone Pneumonia cases which incidentally had COVID. Nothing else explains the excess mortality of Pneumonia cases.
Additional conclusions from the data -Overall mortality:
Comparing yearly mortality from ALL CAUSES, including PIC, there does seem to be a spike. But is the spike great enough to account for the COVID deaths, which should likely be in excess of the average?
Total "Season" Deaths:
2013-'14 - 2,580,853
2014-'15 - 2,750,884
2015-'16 - 2,697,072
2016-'17 - 2,790,278
2017-'18 - 2,835,734
2018-'19 - 2,831,233
2019-'20 - 3,142,232; COVID=203,899
Average deaths 2013-2019 "Seasons" = 2,747,674
2020 "Season" deaths less COVID = 2,938,333
It seems reasonable to assume that overall mortality is in a reasonable range for the year if you exclude the excess deaths caused by COVID.
What can we say about the effect COVID has had on Influenza deaths?
Some might say the Influenza deaths are down due to COVID.
Some would assume this is due to the mask wearing, social distancing, and hand washing.
Others would assume this is due to nefarious counting.
Well, if you compare the total "Season" cases of Influenza between 2019 Week 40 and 2020 Week 39, you would see we're actually up a little from the average of 7646 (2013-2019) to 9415 for the 2019-2020 "Season".
What I would like to see is if there is any rise in excess death from suicide and homicide to see if it correlates to the COVID lockdowns. And also if those excess deaths would offset the COVID deaths in any fashion.
But, I do not have those numbers. Yet.
source:
https://twitter.com/rvolt24/status/1343975559668367360
Excess Mortality, Thread:
— Señor Hilaria (@rvolt24) December 29, 2020
I have just discovered a major error in the #CoVid data of the NCHS Mortality surveillance data. This is all based on the official NCHSData49.csv as downloaded from here (original downloaded 12/15):https://t.co/VkzgNfDZO6