Using ARIMA To Investigate COVID Death (part 4)
Cranking the handle on the latest daily data sitting in the UK GOV coronavirus dashboard
In part 3 of this series I presented the results of a predictive model for daily certified COVID death across England for the period Sep 2020 - May 2021, this being a busy period which saw the second and third wave of cases, viral variants and vaccination roll-out. I concluded that the initial jab simply didn’t provide any measurable benefit or significant harm and promised to run the ARIMA model again to assess the impact of the second jab. Herewith the results.
Before We Begin
Before we begin it’s worth plotting out the two time series of interest in a dual plot:
There’s a flurry of early second dosing back in early January 2021 that pops up around 2 weeks before the seasonal peal death toll. Whilst this is interesting, and we may speculate about vaccine harm vs. seasonal effects and dosing of the most vulnerable members of society, we need to put one eye on the surge in second dose uptake that isn’t reflected in the COVID death count.
ARIMA Run Results
We might guess from the above slide that dose 2 is not going to feature as a statistically significant predictor of COVID death over this time period, and we’d be right - here’s the ARIMA model output:
Down at the bottom of the lower table we see dose 2 sitting with a p-value of p=0.598.
Now some may say, “but you need to incorporate lags of at least 14 days”. Modelling with various lags was indeed undertaken but the insignificance of dose 2 could not be overcome.
All this, of course, is modelling at the population level so if the population as a whole isn’t responding to dosing for better or for worse then nothing is going to show in the stats. This isn’t to say nothing is going on - all manner may be going on at the case and sub-population level - but we’re not likely to pick this up at the population level, and especially so during a time frame when seasonal respiratory death is going to dominate the figures.
At this juncture I suggest we grab a plate of cookies and nibble them whilst waiting for results for the next pandemic phase of 23 May 2021 - 12 Jul 2022; this could be a very interesting period indeed!
It takes an investment of 40-50 hours per week to crank the handle and this is made possible through voluntary donation. Your support is greatly appreciated.