Statistical analysis of a dataset obtained under FOI by Joel Smalley. In this article I present excess deaths for ICD-10 chapter X (J00 – J99): diseases of the respiratory system.
You can only calculate an excess series for a condition that existed over the period 2015 - 2019. With COVID the excess would simply drop out as the total number of COVID deaths per week, which is not really an excess as such. Think of these series as 'excess over historic'.
But Covid is just another respiratory illness, so shouldn’t it be in this chapter? Do we really think influenza disappeared for two years or was it more likely the tests couldn’t distinguish between them. As Dave says, the negative peaks in mid-winter are suspicious.
The clinical picture is far from clear. Medics I've spoken too say the acute respiratory side is rare, as is hypoxia, respiratory failure and all that scary stuff (confirmed using an A&E record dump), with the second (severe) phase more akin to anaphylaxis. Then there's kidney/liver failure muddying the water (some likely due to pharmacological overload).
Those massive drops do look mighty suspicious but I'm wondering if labelling under COVID is the only consideration. I say this because back when SARS-1 did its thing it was coded primarily as B97.21 SARS-associated coronavirus causing diseases classified elsewhere (chapter I: infectious and parasitic diseases). Secondary coding (if applicable) was J12.81 Pneumonia due to SARS-associated coronavirus (chapter X). When MERS did its thing it was coded under B34.2 Coronavirus infection, unspecified.
But, we can ignore the rules and lump COVID and respiratory together to see what this gives us in terms of excess death - I just need to find the time!
You must feel like the man who spends his whole Saturday hanging shelves in the spare room, pours a cold beer and stands back to admire his work. The wife walks past and says it would have looked nice a couple of centimetres to the left.
Under-reporting in the sense of delays at the coroners end applies to 2022 onward. Those twin peaks (troughs?) you mention paradoxically reveal lower levels of respiratory death during the peak influenza seasons and could indeed be the result of MUSE software forcing respiratory death such as from bacterial pneumonia to be coded as COVID (a single positive PCR result will take precedence), but it could also be due to dominance of SARS-COV-2 over influenza viruses. If I had data going back a few more years we'd be able to see if this paradoxical pattern was typical or a new development.
As per my usual question, could we get the earlier data from the ONS? It seems that Joel was able to get this set of data, why not a request framed the same way, but for an earlier time period?
It would be pretty easy to repeat the request over an extended time period and use crowd funding to drum up the dosh. I'll make enquiries to see what sort of dosh they want and what the lag will be. Last time they indicated 3 - 4 weeks, which will give me time to finish this series.
On a separate topic, does substack give you a dashboard to see who is reading your posts, and if so, does it give enough granularity to see whether people from the ONS are reading your posts? The UK government? Just curious.
Yep, I can see who is reading what and when and whether they bother to open emails etc. Lots more info than you need imho. No official ONS email addresses are lodged but they wouldn't be doing this sort of thing, that would be down to the 77th who organise herds of social media orientated officers that come at you with smiles and niceties!
I greatly appreciate your work and am following this thread carefully, but I do think that rather than writing "the third most popular cause of death" it might be more approriate to talk of "the third most frequent cause of death".
I noticed that the y axis scale changed by an order of magnitude so the respiratory cases are essentially noise here. Is that right? What does the same chart look like for COVID alone? I know, copious free time...
Yep, because of the size of those two extreme peaks then the detail gets compressed. I can't produce excess deaths for COVID alone since the code didn't exist back in 2015 - 2019!
What does it look like with the ICD-10 Chapter XXII: Codes for special purposes added back in?
You can only calculate an excess series for a condition that existed over the period 2015 - 2019. With COVID the excess would simply drop out as the total number of COVID deaths per week, which is not really an excess as such. Think of these series as 'excess over historic'.
But Covid is just another respiratory illness, so shouldn’t it be in this chapter? Do we really think influenza disappeared for two years or was it more likely the tests couldn’t distinguish between them. As Dave says, the negative peaks in mid-winter are suspicious.
The clinical picture is far from clear. Medics I've spoken too say the acute respiratory side is rare, as is hypoxia, respiratory failure and all that scary stuff (confirmed using an A&E record dump), with the second (severe) phase more akin to anaphylaxis. Then there's kidney/liver failure muddying the water (some likely due to pharmacological overload).
Those massive drops do look mighty suspicious but I'm wondering if labelling under COVID is the only consideration. I say this because back when SARS-1 did its thing it was coded primarily as B97.21 SARS-associated coronavirus causing diseases classified elsewhere (chapter I: infectious and parasitic diseases). Secondary coding (if applicable) was J12.81 Pneumonia due to SARS-associated coronavirus (chapter X). When MERS did its thing it was coded under B34.2 Coronavirus infection, unspecified.
But, we can ignore the rules and lump COVID and respiratory together to see what this gives us in terms of excess death - I just need to find the time!
Great idea to combine COVID and respiratory to see what pops out. Copious free time...
LOL - I guess I better postpone the next article. We ought to be taking bets on whether COVID will neatly cement over those cracks!
You must feel like the man who spends his whole Saturday hanging shelves in the spare room, pours a cold beer and stands back to admire his work. The wife walks past and says it would have looked nice a couple of centimetres to the left.
Just updated the article with two chapters thrown together...
Just updated the article with two chapters thrown together...
If these deaths are underreported due to being coded as something else (COVID maybe?) then why the twin negative peaks?
Under-reporting in the sense of delays at the coroners end applies to 2022 onward. Those twin peaks (troughs?) you mention paradoxically reveal lower levels of respiratory death during the peak influenza seasons and could indeed be the result of MUSE software forcing respiratory death such as from bacterial pneumonia to be coded as COVID (a single positive PCR result will take precedence), but it could also be due to dominance of SARS-COV-2 over influenza viruses. If I had data going back a few more years we'd be able to see if this paradoxical pattern was typical or a new development.
As per my usual question, could we get the earlier data from the ONS? It seems that Joel was able to get this set of data, why not a request framed the same way, but for an earlier time period?
It would be pretty easy to repeat the request over an extended time period and use crowd funding to drum up the dosh. I'll make enquiries to see what sort of dosh they want and what the lag will be. Last time they indicated 3 - 4 weeks, which will give me time to finish this series.
On a separate topic, does substack give you a dashboard to see who is reading your posts, and if so, does it give enough granularity to see whether people from the ONS are reading your posts? The UK government? Just curious.
Yep, I can see who is reading what and when and whether they bother to open emails etc. Lots more info than you need imho. No official ONS email addresses are lodged but they wouldn't be doing this sort of thing, that would be down to the 77th who organise herds of social media orientated officers that come at you with smiles and niceties!
Just updated the article with two chapters thrown together...
I greatly appreciate your work and am following this thread carefully, but I do think that rather than writing "the third most popular cause of death" it might be more approriate to talk of "the third most frequent cause of death".
Thank you kindly! That is a jolly good suggestion and I have amended the articles accordingly.
Just updated the article with two chapters thrown together. You'll need to read this online because substack doesn't update the emails!
Excellent!
I noticed that the y axis scale changed by an order of magnitude so the respiratory cases are essentially noise here. Is that right? What does the same chart look like for COVID alone? I know, copious free time...
Yep, because of the size of those two extreme peaks then the detail gets compressed. I can't produce excess deaths for COVID alone since the code didn't exist back in 2015 - 2019!
Doh! I wasn't thinking!
But I'm loving the enthusiasm! Take a look at the following article (part 1 of 2) in which I have a go at cracking excess COVID death...
https://jdee.substack.com/p/excess-covid-death-part-1
I reckon all those deaths were all bird fanciers lung, but it simply went out of fashion.
I mean gannet - even geese and swan - populations have been devastated by bird flu: but where is the PCR test for bird-flu?
Now that made me chuckle!