All Cause Mortality in England & Wales (15 – 44y)

From The Office For Nobbled Statistics

I am trying to find robust estimates of the population of England & Wales by 5-year age band so I may have a stab at producing slides of standardised mortality rates for the period 2010 - present. Whilst I’ve nailed some reasonable figures for the weekly total population of England & Wales using linear interpolation and ARIMA time series modelling of ONS mid-year estimates since 1970, I’ve yet to factor these down into age group.

I’ve a figure of 21,399,050 head of population for the 15 – 44y age band (Statista 2020 MYE for England); assuming the proportion of young adults has remained largely invariant over the last decade we can then produce a first crude estimate for SMR and here it is…

This is a pretty noisy graph, as may be expected from weekly registered deaths that will be subject to administrative artefact as well as seasonal and other blips. Martians, when studying this slide, would invariably ask us what we meant by ‘pandemic’ and the answer is that pandemics are defined by spread rather than severity (if they are defined at all!). We ought to expect sideways looks from said Martians at this point. If forced to guess I reckon Martians would point to the winter of 2010/11, though they’d wonder what was going on at the very beginning of 2021.

I would remind the Martians that a One-sample runs test for randomness did not yield a random walk (p<0.001), so something systematic is going on beneath all that noise. If we get out our flexicurve of yesteryear we can see there is a gentle slope down from 2010 - 2012 and a gentle slope up from 2020 or thereabouts, with a distinctive tail from 2021/w21. That tail could indeed be the start of something ghastly but, to be honest, it is too early to tell whether this is a change in the signal or a quirk of the noise. If I can get hold of some decent population estimates I’ll revise this and see what we can then see. Until then, I suggest all eyes be put on ONS weekly registered death data.