Excess non-COVID Mortality in England & Wales 1975 - 2021

What killed granny back in Apr 2020? (rev 1.1)

In my newsletter of 10th June we saw something go bump in the night, being an inexplicable and sudden rise in age & gender adjusted non-COVID mortality for England & Wales during Apr 2020 that came to light when we ran some ARIMA modelling over the data series. Here is that result again…

There are a number of ways we could about about figuring what caused this but I opted for looking at excess non-COVID mortality for all 18 age groups by gender. That makes 36 slides to ogle in succession, each being a splatter of 564 tiny little red dots, which is a lot of ogling. All this is subjective but we somehow have to objectively decide if an outlying point is indeed an outlying point, which entails significance testing to determine whether such a stray point could have arisen by chance.

All this seems like hard work so I took the lazy route of letting the implementation of ARIMA within my stats package do all the work for me by automatically identifying the optimum model structure for each of the 36 time series and at the same time churning out a list of statistically significant outlying data points. A few seconds later I was looking at a list of months deemed to be unusual in some way at the 95% level of confidence (p<0.05).

With a cuppa in one hand and a biscuit in the other I squinted at two substantial tables to find the pandemic rearing its ugly head in only two instances over the period Jan 2020 - Dec 2021. Here they both are:

Now if that isn’t darn interesting I’ll eat my Panama hat! Nothing sticks out for males, nothing sticks out for younger females and nothing sticks out for any other pandemically-flavoured month apart from Apr 2020. What on Earth was going on in Apr 2020 to remove older females from the population? These are not small beer either, with an estimated rise of +117.1 deaths per 100k of 80 - 84y females (p<0.001) and a whopping great rise of +576.2 deaths per 100k of 85+ females (p<0.001). What this looks like in the spotty flesh is presented below.

We now see that this sort of thing has happened before to both age groups, so it is not a unique event for these last 52 years. Pre-pandemic problem months for 80 - 84y females have been Feb 1976, Mar 1976, Dec 1989 and Jan 1997, whilst pre-pandemic problem months for 85+y females have been Feb 1976, Mar 1976, Dec 1989, Jan 1997 and Jan 2015 so we’re talking seasonal influenza. If the Apr 2020 non-COVID hike for older females wasn’t seasonal influenza then what was it?

What killed granny back in Apr 2020?