There were excess deaths in December 2020

Public Health England: “This data should be interpreted with caution”.

Claims on social media and elsewhere say there were no excess deaths in England in December. This seems to be misunderstanding Public Health England all-cause mortality surveillance report.

In short

  • The model seeks to account for reporting delays. The figures are subject to revision.
  • After revisions, the report shows excess deaths throughout December. The Office for National Statistics records excess deaths with a different calculation method.

Models and revisions

For example, Laura Dodsworth (The Critic) wrote:

The two most recent Public Health England All-Cause Mortality Surveillance reports (31 December and 7 January) show no statistically significant excess all-cause mortality. Looking at excess deaths on a graph over previous years, this winter looks fairly typical.

Excess deaths are deaths above a baseline value. The choice of that baseline matters.

For its baseline, the Office for National Statistics uses the average of the last five years. The weekly reports count when the registration weeks of death certificates were. There are also graphs by date of death.

In its reporting, Public Health England use which weeks deaths occurred in. The baseline is a model from EuroMOMO which assumes:

  • The pattern is a Poisson-distributed time series following a trend or annual cycle.
  • The main pattern of mortality follows from Spring and Autumn levels.

In effect, this modelled baseline is an estimate of mortality during a ‘good’ Summer and Winter. This serves a different purpose to the ONS method. EuroMOMO’s goal is to identify severe influenza outbreaks or other deadly events.

Due to delays between deaths and their registrations, PHE apply a correcting model. This is the “EuroMOMO algorithm” the reports refer to. That provides estimated numbers by week of death. These statistics are subject to revision, when we know the true number of occurred deaths.

In week 53 (2020), the All-Cause Mortality Surveillance report highlights reporting delay:

This data should be interpreted with caution due to delays in reporting over the Christmas period.

For week 52, there was “no statistically significant excess mortality by week of death”.

Image for post
Image for post
Note this is ‘corrected’ deaths. Their model may not cope well with public holidays. (Image: Public Health England)

Two weeks later, those recent points underwent major revision:

Image for post
Image for post
Those are major revisions. (Image: Public Health England)

In the week 2 (2021) report, there was excess mortality in weeks 52, 53, and the first week of 2021.

The model appears to struggle with public holidays, leading to major underestimates. We should reflect on how best to express uncertainty when revisions are likely. Graphical representations may aid readers. Urging caution at the start did not seem to prevent misinterpretation.

Once revised, the all-cause mortality surveillance report showed excess deaths.

There were excess deaths in December 2020

If you compare mortality in December of 2020 with average December mortality over the the last five years, there doesn’t as far as I can see appear to be any increase at all.

This is false. Against the past five-year average, there were excess deaths in December 2020. According to the Office for National Statistics:

In December 2020, there were 52,676 deaths registered in England, 10,594 deaths (25.2%) more than the five-year average (2015 to 2019) for December; in Wales, there were 3,941 deaths registered, 1,075 deaths (37.5%) more than the five-year average for December.

In December 2020, COVID-19 was the leading cause of death in both England and Wales.

We can also look age-standardised mortality rates. These figures account for changes in a population’s size and age structure.

Image for post
Image for post
There were major increases in December death registrations. (Image: Office for National Statistics)

These figures are for December death registrations. In England, the standardised rate (1,124 deaths per 100,000 people) was higher than every year back to 2010. In Wales, the standardised rate (1,374 deaths per 100,000 people) was higher than every year back to 2008.

This blog looks at the use of statistics in Britain and beyond. It is written by RSS Statistical Ambassador and Chartered Statistician @anthonybmasters.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store