Excess deaths methodology and the inclusion of pandemic years

FOI Ref: FOI/2023/4804

You asked

​In your registered death figures, you refer to 2021 as a pandemic year. The year of the "pandemic" was actually 2020.

So why has the year showing the lowest number of deaths in the last 5 years (we know 2018 and 2019 had a high number of "flu" deaths) been removed from the 5 year average?

Are the figures being massaged?

We said

Thank you for your request.

ONS is a Non-Ministerial Department reporting to Parliament, which means that we are part of the Civil Service but independent of political decisions. We decide what to publish about any subject based on a combination of factors: our legal obligations, the availability of data, what has been published before, our knowledge of the subject, and contacts with numerous stakeholders including academic researchers, government and local authority policymakers, Members of Parliament, the media, and interested members of the public. Sometimes government departments ask us to look into particular questions, but in the compilation and production of official statistics, the ONS operates impartially and to internationally-recognised standards of best practice. We have no links to, or funding from, private individuals.

You can read the Code of Practice for Statistics on the website of the UK Statistics Authority, which will provide more information on this topic.

Understanding excess deaths and the five-year average

The years 2015 to 2019 are still the most recent years which were not impacted by COVID-19, however the further we move away from 2015 the less helpful this measure becomes due to the changing structure of the population.

Including 2020 in the five-year average would greatly increase the number of deaths we would class as 'expected' each week, because of the peaks of the pandemic, and therefore decrease the number of 'excess' deaths counted.

The decision was made to move to an average of the following five years: 2016, 2017, 2018, 2019 and 2021. This moves our five-year average along by a year but does not include the exceptionally high number of deaths seen in 2020. It allows deaths in 2022 to be compared with a five-year average that is as up to date as possible while still being close to representing a 'normal' year. This methodological change has been reflected in the "measuring the data" section from Week 1 2022 of the weekly deaths bulletin.

Although 2021 was still impacted by the pandemic, this wasn't to the same extent as 2020. Comparing the average of 2015 to 2019 and 2016 to 2019 and 2021 we see that 37 weeks do show a higher five-year average and 15 show a lower five-year average. On average, the number of deaths in the five-year average has increased by 213 deaths (a 2% increase).

If you wish to use a different five year average, the data is fully available on our website in either Deaths registered weekly in England and Wales, provisional releases or the annual Deaths Registered series of publications.

More information on Understanding excess deaths in a pandemic and the five year average can be found in our blog.

Annual deaths for England and Wales

We publish annual figures of deaths registered in England and Wales as part of the following publication: Deaths registered in England and Wales series. Annual deaths registered from 1838 to 2021 are included in this publication.

Number of deaths in England and Wales registered 2016 to 2021, taken from Deaths registered in England and Wales:

There were 73,766 deaths due to COVID-19 in England and Wales in 2020, removing this from the annual figure of 607,922 leaves 534,156 deaths registered which would not make it the lowest year in the 2016 to 2020 5-year average if we had used that measure.

Further information on mortality statistics is available in our User guide to mortality statistics.

If you would like to discuss further please contact Health.Data@ons.gov.uk.