A Winter in Excess

How will the COVID-19 pandemic affect excess winter mortality?

Statistics offices often compare deaths in winter months, against those in other months.

This article looks at the impacts of the COVID-19 pandemic on excess winter deaths. Those impacts highlight problems in using this measure as a proxy for deaths due to cold weather.

Excess winter deaths

Excess winter deaths are a constructed statistical measure. The calculation counts deaths in December to March — ‘winter’ months. We then subtract half of the deaths in August to November and half in April to July. These deaths are from all causes. There is also an index: excess winter deaths divided by average non-winter deaths.

The Office for National Statistics uses deaths occurring in those months. There is an estimation of late registrations, for complicated deaths. Their statistics covers England and Wales. National Records Scotland uses deaths by registration date, for deaths in Scotland.

We can see this calculation in the latest NRS report:

The NRS report uses death records by registration date. (Image: National Records Scotland)

The excess winter death statistic is:

  • 21,392 − (18,973 + 22,711)/2 = 21,392 − 20,842 = 550

Whilst there were a few deaths involving in March, the initial peak was in April. NRS registered more deaths in April to July 2020 than in December 2019 to March 2020.

By the end of September, week 16 was the high point for COVID-19 deaths. (Image: ONS)

The higher death toll in April to July reduces the excess winter mortality figure. Excluding deaths with COVID-19 increases that figure, though changes its meaning:

Excess winter deaths— excluding COVID-19 deaths — was 2,400. (Image: NRS)

The problems of proxies

Researchers may use excess winter mortality as a proxy for cold weather deaths. That proxy has problems, and could mislead readers.

In 2003, Johnson and Griffiths wrote an ONS review of this measure. That review looked at different months for the calculation:

A standard method of calculating EWM can have drawbacks when a pre-determined winter period does not refl ect the size and timing of a winter peak in a particular year. However, over the last 50 years there has been little change in the months in which mortality is above average (December to March) and those where it is below average (May to October).

As Gasparrini and Hajat (LSHTM) discuss in a 2016 Epidemiology article:

During this period, there is no annual correlation between excess winter deaths and winter weather. However, winter weather and specific cold impacts are highly correlated.

The paper looks at deaths in London in 1 August 1997–31 July 2012. (Image: Epidemiology)

Seasonal factors do not follow a strict calendar. Cold weather occurs outside the ‘winter’ months. If a cold snap in October causes deaths, that would reduce the excess measure.

High winter deaths can have other causes. Flu epidemics can peak during the winter months. There may be other factors unrelated to weather. These factors vary from year to year.

Other factors can distort this measure. There may be heat-waves contributing to deaths outside December to March. For the UK, deaths in a global pandemic peaked in April 2020.

This is deaths by day in London during 1997–2012. The lines show two models. (Image: Epidemiology)

The COVID-19 pandemic reduces excess winter deaths by increasing deaths in comparative months. That issue highlights difficulties in interpreting this crude measure.

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