Vaccines & Death (part 8)

Further investigation into delayed death following vaccination

In part 7 of this series I introduced the potential of T4253H smoothing to handle outliers in the data originating from year end and holiday period certificate processing artefacts. I rumbled on about issues with the derivation of excess deaths using a 5-year means baseline (the method favoured by the Office for National Statistics) and pointed out the nonsense that will arise if disease of any sort comes along in an asynchronous manner. I also grunted on about unreliable certification of COVID death that will lead to wonky weekly figures and suggested we start looking at all cause death to avoid the mess. All cause death brings up the spectre of confounding factors and I so suggested judicious use of covariates to level the playing field. I ended by mentioning I had managed to secure weekly dose 4 and dose 5 data for NHS England by trawling though archives which may be found here.

This morning as I tackled the washing-up I realised that smoothing, though utterly magical, can lead to problems down the line with any form of time series analysis. An analogy is climbing up a telegraph pole that provides little iron stepping grips as opposed to a smooth pole that somebody has greased. Time series analysis is all about traction! Thus, I set about creating three (event) indicator variables that marked end of year periods, holiday periods and the remains of the monster 2020/21 death peak that coincided with the start of the vaccination programme in 2020/w50. These three indicators absorb the excessive variance brought about by administrative hiccoughs as well as mark a period of elevated death (COVID or otherwise).

But What About The Virus?

In addition to these three indicators of great cunning we need to concoct a covariate that represents what the virus was genuinely doing from 2020/w50 onward. I say ‘genuinely’ because a zillion people poking their nose tells us absolutely nothing about changes in disease prevalence, this being the rate of infection within the population. For this task I elected to use my trusty case detection rate (CDR) time series, this being COVID cases per 100 viral tests. I first introduced this handy variable in Hunting for Vaccine Benefit (part 2), though it gained fame in Vaccines & Death (part 3).

Bolting It Together

With three indicator variables and one covariate all ready for use in the prediction of all cause death across England for the period 2020/w50 - 2022/w33, all that remained was to crank the handle on the ARIMA machine to obtain a decent baseline model prior to introduction of the time series for vaccine dosing. It was then I accidentally pressed the wrong button and came face to face with an enigma or three…

An Enigma

It’s always best to ponder over the raw data series before embarking on ARIMA time series modelling, and especially the autocorrelation and partial autocorrelation plots. I like to go one further and eyeball the cross-correlation plot between the two key variables of interest. Thus it was that I ran out a CCF plot of my newly established combined weekly vaccine doses (doses 1 - 5) with all cause death across the fair nation of England for the period 2020/w50 - 2022/w33. Here it is:

What we are interested in here are any red bars sticking above the dashed grey boundary denoting the 95% confidence interval; anything sitting below these could easily be random garbage. We observe a significant positive bar at a lag of zero weeks and two significant negative bars at lags of +1 week and +23 weeks. If we were to summarise the situation in plain English we’d say there is evidence of instant vaccine harm along with evidence of vaccine benefit one week and 23 weeks later. That’s pretty much an enigma right there, but it gets deeper…

The mRNA vaccines are VERY specific therapeutic beasts are they not? They’ve been designed to tackle the spike protein of an early strain have they not? So why is it that we see an identical pattern of instant harm and delayed benefit for non-COVID death? Have a look at this:

These are magical products indeed to solve the myriad of problems ranging from falling off ladders to being trampled by horses. With one eyebrow raised I took a look at the CCF plot for combined weekly doses with certified COVID death, taking care to reproduce the same scaling for the y-axis (vertical):

Where has all the noise gone? By ‘noise’ I mean that motley collection of red bars that didn’t make the grade and push beyond the confidence boundary. The bars have shrunk, that is for certain, and it’s worth pondering why - what is so special about certified COVID death that it slinks away from the harsh light of CCF?

Some analysts might argue that the variance inherent in the time series for certified COVID death is bound to be less than that for all cause and non-COVID death. A fair point indeed, but the table below reveals this is not the case, with the standard deviation of the series for certified COVID death exceeding that of non-COVID death, along with the range:

Yet we digress, for the most important point of points is that the above CCF slide reveals total lack of vaccine benefit for COVID cases even after a lag of 26 weeks - something I shall be pursuing in future articles. That’s another enigma right there, but it gets deeper…

By complete accident I included the 5-year prior means baseline time series, as derived by the ONS, these being mean weekly deaths for the period 2015-2019. Now I’m pretty sure that the vaccine wasn’t around in 2015-2019 and neither was the novel SARS-COV-2 virus so we’d expect the cross-correlation with combined weekly dosing to generate random garbage. Except it doesn’t:

How about that for a statistical shocker? Here we have a near identical CCF plot of instant vaccine harm and 1 week benefit for a time series that never knew COVID! Wergh is the word! We can also see that week 23 benefit trying to nudge down.

So what does this all mean?

It means we have been caught out by a grand illusion that stems from the fact that vaccine roll-out took place over time at such a pace that what appears to be harm and/or benefit is artefact arising from confusion with the periodicity of natural seasonality of things both COVID and not COVID. In plain English it’s a flippin’ fluke! An analogy is to twang two guitar strings that are nearly in tune - that wobbling sound you hear is a beat frequency and an example of wave interference. Think of the wave interference peaks as ‘harm’ and troughs as ‘benefit’ and there you have it. Vaccines are the harm-benefit nonsense machine par excellence.

It also means I need to go bake a large cake and not come back until it’s all eaten with plenty of cogitation, cognac and coffee. I may be gone for some time…

Kettle On!