Understanding the data about COVID-19 related deaths
23 April 2020
The coronavirus (COVID-19) pandemic represents an unprecedented challenge not only to our health and care services but also to our economy and society. An important part of assessing the consequences of the outbreak is understanding the number of lives lost to COVID-19. Whether this is directly, where the virus has directly contributed to the cause of death or indirectly, such as where resources have been diverted from other areas.
Appropriate analysis of the data, as it becomes available, will help us to understand the scale, shape and progress of the current outbreak. In turn this can inform future policy action and health and care practice. This includes understanding the effectiveness of social distancing measures and how far different groups are at greater risk.
There are currently two main sources of data about the number of deaths related to COVID-19. Both are helpful and provide important information. However, because they tell us different information and are published to different timescales, they can be confusing to navigate. Here we set out what the key sources of data are and the differences between them. We also provide an overview of data related to COVID-19 related deaths published on 21 April to compare these sources and highlight the reasons for differences.
What are the two main sources of information about COVID-19 related deaths?
- Department of Health and Social Care (DHSC) data: This is more timely and provides a count of deaths of people with COVID-19 in hospitals published by DHSC on a daily basis. It is the figure announced by government each day ahead of the daily press briefing.
- ONS data: This is more comprehensive data but less timely, published in the longstanding Deaths registered weekly publication. It captures all officially registered deaths on a weekly basis. This includes those who have died in the community as well as hospitals.
What are the key differences?
There are several key differences between the two sources that are important to understand when interpreting the data. It is worth noting that in recent weeks the UK’s health and statistical agencies have been making welcome improvements to the detail, timeliness and coordination of the data that are released to provide more information during the current crisis.
Key differences between data sources on deaths related to COVID-19
| Data source | When is data released? | Whose deaths are included? | Where did these deaths occur? | When are deaths reported/registered? | When did these deaths occur? | What parts of the UK do these data cover? |
| DHSC | 2pm daily | People testing positive for COVID-19 (though it is not necessarily the cause of death) | In hospital* | Deaths are reported in 24hrs up to 5pm on the previous day | Majority occur at least 2 days before publication** | UK wide |
| ONS | Every Tuesday at 9.30am | All those where COVID-19 is mentioned on the death certificate, whether the main cause of death or not | Any location (including care homes) | Deaths formally registered in the week ending 11 days before publication | Deaths typically take 5 days to be registered | England and Wales (with limited data for the rest of the UK) |
*small number outside of hospital may be included due to differing reporting practices across devolved nations
**for instance of the deaths reported on 13 April, 70% of deaths announced occurred in the preceding 2–4 days, only 18% occurred the day before
Comparing the data sources for COVID-19 related deaths
- These estimates are published at 2pm every day. They capture the number of deaths of people testing positive for COVID-19 that have been reported in the 24-hour period up to 5pm on the previous day.
- These are deaths where the person has tested positive for COVID-19, although the virus is not necessarily the cause of death, and may not have been officially registered with the register office by the time it is reported.
- Deaths almost entirely occurred in hospitals (a small number may have occurred outside of hospitals due to reporting differences across the four UK countries).
- The majority of these deaths tend to have occurred in the 2–4 day period before they are announced (eg of the deaths announced on 14 April, 16% occurred the day before, 43% 2 days before and 18% 3 days before, while the remainder occurred earlier).
- Data relates to the UK as whole but is collected separately by the health service in each UK nation.
- There is some day-to-day variation in estimates, particularly with lower counts on a Sunday and Monday, reflecting delays in the reporting of deaths over the weekend.
- A supplementary release published by NHSE provides a breakdown of when the deaths reported each day actually occurred.
- This more comprehensive data is published each Tuesday at 9.30am. The release provides a provisional weekly count (ie may be revised with further data collection) of all officially registered deaths in the week up to 11 days before publication. COVID-19 related deaths are counted separately within this weekly release.
- COVID-19 related deaths include all those where COVID-19 is mentioned on the death certificate, whether the main cause of death or not. In some circumstances COVID-19 is ‘suspected’ or ‘presumed’ rather than confirmed via testing. Medical practitioners are required to certify causes of death ‘to the best of their knowledge and belief’. Without diagnostic proof, if appropriate, and to avoid delay, medical practitioners can apply clinical judgement in the absence of testing capability, or report COVID-19 while waiting for the test result.
- All registered deaths are included, not just those occurring in hospitals. A breakdown by type of location of death is also provided.
- The headline estimate relates to deaths in the week they are registered, which is not necessarily the week they occurred. Deaths typically take 5 days to be registered, almost three-quarters of deaths in 2018 (74%) were registered within 7 days of occurrence.
- The data predominantly covers England and Wales and is collected from the General Register Office, with some supplementary data for the rest of the UK. The Northern Ireland Statistics and Research Agency and Scottish National Records publish weekly estimates to a similar timeframe.
What does the data tell us about the number of deaths relating to COVID-19?
Here we use data published by ONS on 21 April to help illustrate differences in the data. The data shows a total of 10,335 deaths where COVID-19 is mentioned on the death certificate and registered by the week ending 10 April 2020 in England and Wales. The number of COVID-19 related deaths registered each week has risen from 5 in week ending 20 March to 6,213 in the week ending 10 April.
The count published by DHSC on 21 April shows a total of 17,337 COVID-19 deaths were reported by 5pm on 20 April. The number is much higher than published by ONS because data from DHSC is more timely and relates to deaths across the whole of the UK. The number of COVID-19 deaths for England and Wales in the same release by DHSC was reported as 16,215.
How does similar information from each source compare?
Supplementary data provided with each data release shows when COVID-19 related deaths occurred. This is an important distinction to the headline figures in each release that relate to deaths reported or registered by a given point in time, rather than when the deaths occurred.
To understand how the timing of death occurrence relates to death registration or reporting, the chart below compares data from ONS, for England only, showing when COVID-19 related deaths were registered and actually occurred on a daily basis. It also contrasts the equivalent estimates published daily by DHSC (date reported) and NHSE (date of occurrence) for England only. Deaths by date of occurrence in the most recent period are likely to be subject to significant upward revision as more deaths are registered.
What else can the ONS data tell us?
The strength of the ONS data is that it provides information about all deaths, not just those related to COVID-19 or occurring in hospital. There is also greater detail about age and sex, causes of death, place of residence, and whether deaths occurred at home, hospital, care homes or other locations. Historic weekly data is also available that can be used to compare the pattern of deaths during the current pandemic to historical trends. One limitation of the ONS data is the availability of testing for cases outside of hospitals, which can lead to reporting errors in the numbers of deaths relating to COVID-19.
The extent to which anything meaningful can be said about changes in historical patterns of death depends on how quickly and how much detail is available from ONS. The availability of this data in turn depends on collection and processing times, which in some instances have been shortened to provide more timely data. As more data becomes available, we can begin to understand whether deaths are:
- directly caused by COVID-19
- if COVID-19 was present but not the main cause of death
- indirectly caused, where deaths would not have occurred if the health care system resource were not under pressure
- deaths that may have been anticipated this year, but occurred earlier due to the virus.
It will also be important to understand how far some of the increase in deaths related to COVID may relate to changes in reporting practice or increased awareness of possible COVID-19 symptoms.
The Health Foundation is conducting a programme of work to understand the impact of COVID-19 on the health care system, people’s health outcomes, and the factors that determine people’s health, and how that experience may vary for different groups. Further analysis and research will be published on our website in due course, including further analysis about the impact of COVID-19 on deaths.
David Finch (@davidfinchrf) is a Senior Fellow at the Health Foundation.
Further reading
Work with us
We look for talented and passionate individuals as everyone at the Health Foundation has an important role to play.
View current vacanciesThe Q community
Q is an initiative connecting people with improvement expertise across the UK.
Find out more
