Excess Deaths by Cause, England 2020/w1 – 2022/w46 (part 2)
Statistical analysis of a dataset obtained under FOI by Joel Smalley. In this article I present excess deaths for ICD-10 chapter IX (I00 – I99): diseases of the circulatory system.
Please see part 1 of this series for background detail.
We now come the second most frequent cause of death, this being diseases of the circulatory system. This ICD-10 chapter covers everything from I010 Acute rheumatic pericarditis to I99X Other and unspecified disorders of circulatory system. In plain English we’re talking heart attacks, heart failure, heart flutter and heart valve conditions as well as sudden cardiac death, aortic rupture and bacterial infection.
More Deaths To Come
When viewing these slides please do bear in mind that delays in processing (largely due to involvement of the coroner) mean counts as far back as January 2022 are likely to be under-reported, especially for the younger age groups. Because of the volume of slides I need to present and limits to email delivery I will keep commenting to a minimum.
Keep an eye out for the mysterious death spike of 2020/w15 (w/e 10 April) and a tailing-off of excess during 2022 for the younger age groups. Under ‘normal’ circumstances circulatory death should show as a series of waves flipping from positive excess to negative excess to positive excess. Keep your eyes peeled for any persistent trends or offsets, whether this is characterised by a persistent positive or negative excess. Please do remember that vaccine harm will be one of many factors.
Get that coffee pot on, get a muffin toasting and let’s get stuck straight in…
Circulatory Diseases By Age Band
My eyeballs suggest delays to reporting for 2022 onward but my eyeballs can deceive! If I conjure an indicator variable to split the data into the period 2020 – 2021 and 2022, then we find a mean of -0.06 excess deaths per week for 2020 – 2021 and a mean of -0.33 excess deaths for 2022. Though this difference supports our eyeballs and thus our notion of delayed reporting, it fails to reach statistical significance at the 95% level of confidence owing to the substantial inherent variation (p=0.212; ANOVA).
This time our eyeballs are support by the stats! The overall mean excess for 2020 – 2021 was -0.15 excess deaths per week compared to -1.76 deaths per week for 2022(p<0.001; ANOVA). Take a sheet of paper and hide the points for 2022, then hide the points for 2020 – 2021 and the difference is made obvious. This is an important finding for it means analysts should not be viewing this data as a homogeneous time series. If they do they’ll come to the fallacious conclusion that circulatory deaths for 18 – 29y are decreasing over time. Keep your beady eyes on how ONS and other ‘institutionalised experts’ handle this issue!
Delays to processing of death certificates for this age group has got pretty bleedin’ obvious, innit? And what’s with those spikes during spring 2022 – the results of a booster campaign perhaps? Until we get all certificates in (and correctly coded) we can’t really say much other than note the possibility.
The first feature that caught my attention with this slide is the persistent positive mean excess of +18.41 deaths per week that continues for most of 2022, this excess ramping up from a near zero excess prior to the pandemic. It would be easy to pin the blame on COVID but we’ve got lockdowns, alcoholism and loss of service provision to consider as well as folk with chest pain avoiding calling for an ambulance. FYI back in April 2020 a cardiologist confided that nobody was turning up and they were worried people were dying at home from myocardial infarction for fear of visiting hospital and catching a super deadly virus.
Another example of sustained positive excess following a ramp up during spring 2020, this time settling at an overall mean +12.75 deaths per week for the 60 -69y age group for the period 2020 – 2021. With death certificate processing incomplete it’s tricky to say anything sensible about that whopping great peak during 2022 but what I’ve done is plotted the mean excess of +16.8 deaths per week for the reference period 2020/w16 to 2022/w33 (red dashed line) and added in the three sigma boundaries for this (grey dashed lines). You can now see that the 2022 spike just pushes past the upper three sigma boundary. In betting parlance there is a 1 in 741 chance of this happening as a fluke if we assume a normally-distributed variate (p=0.00135), so we may conclude that something was very likely going on.
It sure would have been wonderful to squint at earlier years to see if this triple-humper is a typical pattern or represents something odd that is happening. I expected something more akin to a random walk, maybe with a touch of seasonality, so this pattern of excess death for the 70 -79y group is somewhat intriguing. Are those humps following vaccination campaigns or am I brimming with bias? I shall have to pursue this further with some ARIMA intervention modelling to see how much excess can be objectively pinned onto the pattern for jabs. It’s worth noting the blip back down during 2020/w15 (w/e/ 10 April).
Crikey, now there’s a triple-humper and a half, and that 2020/w15 (w/e/ 10 April) blip is now a blooper! What sort of event (event?) would cause a well-defined sudden surge in circulatory death at the precise same time we see a surge in neoplasm death (see part 1)?
Our blooper of 2020/w15 (w/e 10 April) is now a monster. This was all about saving granny, was it?
It is interesting that this catch-all slide should bring out the points I’ve made rather than yield a sorry looking mess. We’ve got some interesting spikes whose dates are worth checking against various happenings on the COVID management front, we’ve got a triple-humper and we’ve got the tell tale tail-off of delayed processing. I can’t wait to run multivariate ARIMA over this but that will have to wait until I’ve considered three more major causes of death. Until next time…
It takes 40-50 hours per week to crank the handle as I do and this level of commitment is only made possible through subscription income. I am most grateful for your support.
Don’t kill your granny, says Matt Hancock. Leave that one to me.
Do we have any data on how much is backlogged with coroners?