Excess Death Figures: Further Considerations (part 3)
Excess death is use to assess the impact of COVID, government policies and COVID therapies but this method is brimming with issues. In this article I conjure projected non-COVID death.
In part 2 of this series I introduced something I’m calling mean mortality, this being the arithmetical average of mortalities derived for each of ten quasi decadal age bands (under 1 year, 1 – 17, 18 – 29, 30 – 39, 40 – 49, 50 – 59, 60 - 69, 70 – 79, 80 – 89, +90). This may seem a strange thing to do because it provides each age band with an equal say in how the final time series turns out, this implying a population with a rectangular age distribution. Obviously this approach isn’t going to produce an overall mortality rate commensurate with that derived from the distribution of the real population, but it is going to provide a rather useful idealised population that is invariant over time. In plain English, it means we can more easily spot trends rather than artefact brought about by the shifting age profile.
Means vs Projections
I tied the technique out on the five biggie causes of death and mumbled something about using modelled projections as a baseline instead of classic five-year prior means. The latter is how the Office for National Statistics (ONS) go about the business of calculating excess death and the former is how bods at Our World In Data go about calculating excess death. The beef with using a 5-year mean is that this ignores any long-term changes that may be occurring, which may lead to under- or over-estimation of excess death. The solution is to resort to statistical techniques to model the variation inherent in a data series (both seasonal and long-term) which, on paper, should produce more robust estimates. The downside is that we’re relying on an extra large raft of assumptions that underpin all modelling, and models have the nasty habit of going all linear and ignoring real world complexities that give rise to saturation effects. In plain English if the mortality rate has been decreasing for a decade we can’t assume it will continue to do so for the next decade. In fact, if modellers came clean about things they’d have to admit that modelling can produce a serious amount of rubbish; more rubbish, in fact, than using something simple like a five-year prior mean.
To illustrate this I’ve concocted five different ways of going about the derivation of excess death. To keep it simple I pooled all ICD-10 chapters bar chapter XXII (codes for special purposes) and all age groups, this giving me the grand weekly total for non-COVID deaths by date of death. I guess we better start by eyeballing this time series:
There we go. We observe strong seasonality apart from the mysterious winter period of 2021/22, and we also observe a rather peculiar out of season spike during spring of 2020. Yes indeed, that’s the infamous CHEC death spike (Catastrophic Health Collapse), for which a summary may be found here. Any reasonable person will ask why we saw such a significant spike in non-COVID deaths and any reasonable government would do its best to investigate the matter. As it is we are going to have to resort to legal action and serious levels of activism to get anywhere near the truth. I’m hoping that official inaction and legacy media silence is fuelling a revolution in how people view government and all institutions that feed into the narrative, and especially those rather vocal experts who will invariably find an excuse not to consider any action whatsoever.
TBH I can’t see anybody in a position of authority owning up but there’s a truckload of whistle-blowers waiting to chirrup (herewith one such example).