Excess deaths – a new methodology and better understanding

The COVID-19 pandemic heightened interest in patterns and levels of ‘excess’ deaths, typically defined as deaths over the number that might be expected to occur in an ‘average’ year. But with different organisations using different methods to calculate excess deaths, it can be difficult to build a clear picture. Working across government and the devolved nations, we have now agreed a common UK-wide approach to producing national estimates of excess mortality. As Julie Stanborough explains, this new methodology will give us a better understanding in this complex area.

During the peak-pandemic years, we saw dramatic increases in mortality rates. It was a stark example of a public health threat resulting in many more deaths than we would have expected in a ‘normal’ year. Data on excess mortality provide valuable insight for policymakers in planning health interventions, seeing patterns of need and identifying at-risk groups. In the short term, excess deaths measures can be used to identify emerging threats. However, it’s important to note that excess deaths estimates are just that – estimates. They cannot be counted on an individual basis, as can be done for death registrations. They are estimated using statistical techniques and, as a result, there is no single “true” measure of excess deaths.

As discussed in a previous ONS blog, fundamental to any method for estimating excess deaths is the question, how many deaths do we expect there to be in normal conditions; in other words, what would “normal” mortality levels look like? The current approach used by ONS and the devolved administrations provides a comparison between the number of deaths registered in the current year and the average number over a recent five-year period. For example, excess mortality in 2019 was estimated from data covering 2014 to 2018. 2020 was excluded from subsequent calculations to avoid distortion due to the extremely high number of death registrations, particularly during the first wave of the COVID-19 pandemic. For 2020 and 2021, the average was calculated over 2015 to 2019, and for 2022, it was calculated over 2016 to 2019 plus 2021.

The weakness of this approach is that it doesn’t take into account the ageing and growing population of the UK (all else being equal, more people means more deaths, particularly if a greater share of the population are elderly); nor does it reflect recent trends in population mortality rates, which were generally falling until 2011 before levelling off until the onset of the pandemic.

Finding a new way forward

The UK-wide excess mortality technical working group benefitted from expertise spanning many organisations and fields, with representatives from different areas of ONS, Office for Health Improvement and Disparities (OHID), UK Health Security Agency, Public Health Wales, Welsh Government, National Records of Scotland (NRS), Northern Ireland Statistics and Research Agency (NISRA) and members of the actuarial profession.

We explored several different methodologies for estimating excess deaths, including five-year averages, relative age-standardised mortality rates, segmented regression analysis, time series models, neural networks, the approaches used in the CMI mortality projections and CMI pandemic monitor, EuroMOMO, the World Health Organization’s method for estimating excess mortality during the pandemic, the UK Health Security Agency’s daily mortality method and OHID’s excess death model. A review of the strengths and limitations of these approaches was discussed by the UK Statistics Authority’s Methodological Assurance Review Panel (MARP) last year.

The new approach

The chosen methodology uses statistical models to obtain the expected number of deaths in each period. Importantly, this approach moves away from averages drawn from raw numbers and instead uses age-specific mortality rates. This means when we ask that first question – how many deaths would we expect there to be? – we take into account how the population has grown and aged over time. The models also account for trends and seasonality in population mortality rates, and allow for estimates of excess deaths to be broken down by age group, sex, and constituent countries of the UK and English region.

This approach provides a method for routine monitoring of excess deaths on an ongoing basis. It was not designed specifically for measuring the impact of the COVID-19 pandemic on mortality – though it can be used to shed light on this.

The new methodology is closely aligned with that used by OHID to estimate excess deaths in English local authorities and other population subgroups, and will be used by NRS and NISRA to estimate excess deaths in Scotland and Northern Ireland, respectively. By introducing a consistent method for estimating excess deaths across the UK, we intend to increase coherence and bring improved understanding to this important topic.

Going forward

This month, ONS will publish a methodological article detailing the new approach to estimating excess deaths. This will contain estimates for periods before and during the pandemic using the new approach, and will compare these with estimates from the current methodology. After the new methodology has been published, excess deaths estimated using the new approach will be available in our Deaths registered weekly in England and Wales publication, as well as in mortality outputs produced by NRS and NISRA. In the future, we plan to include estimates for each of the four countries in ONS releases, allowing users to draw comparisons and interpret trends across the nations, as well as for the UK as a whole.

In the spirit of continuous improvement, we will regularly review estimates of excess deaths produced by the new methodology, with further refinements being undertaken if necessary. As such, the new estimates will be labelled as Official Statistics in Development while further review, testing and development work is undertaken.

Julie Stanborough is Deputy Director for Data & Analysis for Social Care & Health

Julie Stanborough is Deputy Director for Data & Analysis for Social Care & Health