Behind the headlines: Counting COVID-19 deaths
Defining a COVID-19 death
Although it might seem straightforward, counting the number of people who have died from COVID-19 related illness is complex.
The infection can lead to death soon after diagnosis, but it may also cause death many weeks later. Someone who tests positive can of course die from another cause such as cancer or heart disease at any time.
A death in someone who has tested positive becomes progressively less likely to be directly due to COVID-19 as time passes and more likely to be due to another cause. However, there is no agreed cut-off after which COVID-19 can be excluded as a likely cause and sadly, we know that some people die from their infection many weeks later. Coronavirus can also contribute to a death without being the main or “underlying” cause.
The World Health Organization (WHO) recognises this complexity and states that:
A COVID-19 death is defined for surveillance purposes as a death resulting from a clinically compatible illness in a probable or confirmed COVID-19 case, unless there is a clear alternative cause of death that cannot be related to COVID-19 disease (e.g. trauma).
This definition therefore requires a clinical assessment of each case.
For several months, the COVID-19 Data Dashboard has been reporting, for England, all deaths in people who have a positive test. This a robust measure as it uses the fact of a positive test and the fact of death to derive the number reported. However, it is only an approximation of the number of people who die from COVID-19 because other causes of death are included and some people who die from COVID-19 never had a positive test. It was decided to adopt this measure in April in order to be sure not to underestimate the number of COVID-19 related deaths. It was always intended to review the approach as the pandemic progressed.
The countries of the UK have been using slightly different methods. Scotland, for example, has only been counting deaths within 28 days of a positive test so that deaths from COVID-19 beyond 28 days are not included.
Subsequent reporting of the cause of death
In the UK, COVID-19 deaths should be identified as such on death certificates, recorded by a registered medical practitioner. Provisional data on numbers of certified deaths from COVID-19 have been reported weekly by the Office for National Statistics (ONS) since 31 March 2020. However, there is an inevitable delay in reporting and publishing deaths based on death certification – for example, the most recent publication available at the time of writing reports deaths to week ending 31 July 2020.
Calculating the excess death rate
Another approach to assessing the impact of COVID-19 is to calculate the excess death rate. This method compares the total number of deaths in a week to the average expected from previous years. This is an excellent method, but it also takes some time for the results to be available.
How does PHE identify deaths in people who have tested positive?
In England, we have collated the details of every person who has had a laboratory-confirmed positive COVID-19 result at any point since the start of the pandemic. This will be a very valuable resource for tracking the impact of COVID-19 on the health of those affected.
We actively look at four sources to identify the death of any of these people should it occur. These are:
- deaths occurring in hospitals, notified to NHS England by NHS trusts using the COVID-19 Patient Notification System
- deaths with a confirmed COVID-19 test, notified to PHE Health Protection Teams during outbreak management (primarily in non-hospital settings) and recorded in an electronic reporting system
- all people with a laboratory-confirmed COVID-19 test identified to have died through tracing against NHS records
- ONS death registrations which can be linked to a laboratory-confirmed COVID-19 test
Does this mean you have overcounted deaths from COVID-19 in England?
The total number of deaths reported in the daily numbers is less than the total number of deaths registered with COVID-19 on the death certificate, so the numbers reported have not generally been an over estimate. However, in recent weeks the numbers of deaths in people who have tested positive have become substantially greater than the numbers of deaths subsequently registered as COVID-19 deaths by the ONS, which is why we are now changing our approach to reporting deaths.
What did the review look at?
Our review considered epidemiological evidence to see how likely it was that COVID-19 was a contributory factor to a death at different points in time after a positive test. We examined all 41,598 deaths in confirmed cases of COVID-19 reported up to 3 August 2020 and found that:
- 88% of deaths occurred within 28 days of a positive COVID-19 test and 96% occurred within 60 days or had COVID-19 on the death certificate
- of those who died 29 to 60 days after their positive test, COVID-19 was included on the death certificate for 64%
- 2,295 (54%) of the 4,219 deaths excluded by a 28-day time limit had COVID-19 on the death certificate
- overall 91% of deaths reported by PHE in confirmed cases up to 3 August had COVID-19 on the death certificate
What is changing?
Two new deaths indicators will now be used by all four nations in the UK to provide a full picture of both recent trends and the longer-term burden of the disease.
The additional indicators which will be used to calculate daily death figures are:
- the number of deaths in people with COVID-19 that occur within 28 days of a first positive laboratory-confirmed test. This is intended to provide a headline indicator of the immediate impact of recent epidemic activity. Deaths that occur more than 28 days after a positive test will not be included in this count.
- the number of deaths that occur within 60 days of a first positive test. Deaths that occur after 60 days will also be added to this figure if COVID-19 appears on the death certificate. This will provide a more complete measure of the burden of the disease over time.
Using these new measures, the total number of deaths in people with laboratory-confirmed infection is reduced by 5,377 if only deaths within 28-day of a test are included, and by 1,668 if including only deaths within 60 days or at any time with COVID-19 mentioned on the death certificate.
This approach has been peer reviewed by external statistical experts.
What do other countries do?
The WHO recommendation is to report on death certification, which England will continue to do through the ONS. For daily reported death figures, there is no international consensus on methods.
Why do the public health figures differ from the ONS numbers?
PHE has been counting all deaths in people who have laboratory-confirmed infection – this is technically robust because it does not require a judgement to be made about cause of death. ONS reports deaths where a doctor suspects COVID-19 as a cause – these data include a clinical assessment as recommended by WHO but are subject to variation in clinical judgement as to the cause of death.
These are two different measures with different strengths and weaknesses. The PHE data series is also available daily, making it more useful for real-time surveillance, whereas the ONS survey only appears once a week and is delayed.