For the UK and for other health systems, the response to the Covid-19 pandemic has needed to balance actions aimed at reducing deaths from the virus with the negative impact of restricting economic and social activity and reduced access to other health and public services. Understanding how many deaths have resulted from the pandemic is therefore critical – this is the yardstick against which we can assess the success or otherwise of the measures we are taking.
With a second wave of infections underway, we have updated the Q&A we published in May, and set out some of the key questions about how we measure mortality, both now and in the longer term.
How many extra deaths have there been as a result of the pandemic?
We can estimate the overall impact of the pandemic by comparing the number of deaths each week with those we would expect to see based on the average for the last five years.
For the weeks ending 13 March to 9 October, there were 356,450 deaths registered in England and Wales, which is 59,161 above the expected number of 297,289 – indicating there were over 20% more deaths registered over this period.
Over 90% of these excess deaths occurred within the first two months, and at the height of the first wave there were more than double the expected number of deaths in a week.
Where does this data come from?
When a death occurs, a medical practitioner completes a death certificate that enables the person’s family to register the death. Death registrations are collated and the numbers reported by the Office for National Statistics (ONS) on a weekly basis.
There is a time lag between when the death occurred and when the data is published. For example, data published on 20 October is for registrations that occurred during the week ending 9 October, with those deaths having occurred approximately five days prior to registration. Registrations can be delayed when there is a bank holiday – which we can see in the higher number of registrations in the weeks following the May and August bank holidays.
The death certificate will include the immediate cause of death and the underlying disease or injury that led up to the death. Contributory causes of death can also be recorded, where they may have affected the outcome, but illnesses should not be included on the death certificate if they were present but did not contribute to the death.
How significant is this increase in deaths?
Death registrations vary on a week-by-week basis each year. This is due to seasonal factors such as winter flu and cold weather, and also as a result of reporting changes, such as a greater time lag in registering deaths over bank holidays.
However, the number of weekly deaths we have seen during the Covid-19 pandemic is among the highest on record, even with lockdown measures. Previous peaks in deaths have occurred over winters when there were bad outbreaks of flu. During the winter of 1999/2000, there were 18,000 deaths recorded in one week – which would be equivalent to over 21,000 deaths given today’s larger population.
Over the past 20 years, death rates have declined overall, so the increases we are seeing now are very significant.
Are the extra deaths all related to Covid-19, or are some of them the result of increases in mortality for other reasons?
During the first wave of the pandemic, there was a gap between the estimated extra deaths and the number of deaths where Covid-19 was mentioned on the death certificate, with Covid accounting for approximately three-quarters of the number of extra deaths.
The difference could reflect a number of different factors – both clinical and in the reporting.
For patients who haven’t been tested, and more particularly if the patient has tested negatively for the coronavirus, doctors may have been reluctant to include the diagnosis on the death certificate, even though the clinical pattern suggests that Covid is a factor.
In terms of reporting, the ONS data includes any death where Covid-19 is mentioned on the death certificate – either directly, as an underlying cause or as a contributory cause. However, the guidance states that only illnesses that contributed to the death should be listed.
For patients who were already close to the end of their life, doctors may judge that the Covid-19 infection was not a significant factor. Even though the patient died with the disease, they didn’t die from it.
As we move into a second wave, changes in coding may reduce the gap between deaths from Covid-19 and excess deaths. The understanding of how the disease affects different groups of patients has increased, and clinicians have much more experience in identifying and treating the disease. There is also more capacity to test patients for Covid-19, so it is likely to be identified at an earlier stage of their illness.
However, it remains the case that some of the extra deaths may be a result of people not seeking help for other illnesses, or because the NHS stopped planned treatment – which includes care for life-threatening conditions such as cancer, as well as routine check-ups.
In addition, the lockdown response to the pandemic will have affected the risk of death from other causes – from road traffic accidents through to falls and injuries, drug and alcohol related deaths, and suicide. The impact of these wider changes won’t be clear for some time.
How do the weekly deaths reported by the ONS relate to the daily reported deaths?
Daily reported deaths are collated on the government’s Covid-19 dashboard. This data is gathered from a number of sources, and includes anyone with laboratory-confirmed Covid-19, including people who died outside of hospital.
Since 12 August, two measures have been published. Firstly, the number of people who died within 28 days of their positive test. Secondly, the number of people who either died within 60 days of a first positive test, or who died more than 60 days after a positive test but had Covid-19 mentioned on the death certificate.
These figures don’t therefore include the apparently large number of excess deaths in the ONS weekly data that have not been linked to the coronavirus.
The ONS have shown, by analysing the data retrospectively, that the number of daily hospital deaths matches closely the number of daily deaths occurring that are registered with a mention of Covid-19.
On the day that deaths are published, however, the count will include deaths that actually occurred in previous days. The hospital needs to validate that the death was Covid-19 related, and also inform relatives. This process can take a number of days, and is slower over the weekend.
As a result, death counts are sometimes lower on Saturday and Sunday, and more deaths will be reported midweek once recording catches up.
How does mortality vary between groups of people?
The numbers of people who have died of Covid-19 and mortality rates are higher in older age groups and higher in men than women. These differences are in line with mortality from other causes, but are nonetheless very striking. In April, at the height of the pandemic, one in eight people over 90 died of Covid-19, compared with just one in 50,000 aged between 15 and 19.
Covid-19 also has a greater impact on those groups in the population who experience inequalities in health from other causes. Mortality rates are higher in black ethnic groups and are higher in more deprived areas compared with those who are better off.
Where are Covid-19 deaths occurring?
Deaths registered from the week ending 13 March to 9 October were 40% higher than the five-year average in private homes and care homes, and 4% higher in hospitals. Deaths where Covid-19 was mentioned on the death certificate accounted for 24% of hospital deaths, and 18% of care home deaths, but only 3% of deaths in private homes.
The number of deaths in care homes returned to pre-pandemic levels by early June. It is possible that part of the increase for care home deaths during the first wave was that people were not being transferred to hospital shortly before death, as might have happened previously, and the effect of people being discharged from hospital in the early stages of the pandemic, to free up capacity in hospitals. In these cases, the death would still have occurred, but in the hospital rather than a care home.
In relation to care homes, we know that the mortality rate from Covid-19 is much higher in older age groups, and is higher still among older people in care homes than those of the same age in the general population. We need to take into account that residents are more likely to be frail or have multiple underlying health conditions than those living independently. However, rates of Covid-19 infection were higher in larger care homes and those where staff worked across multiple sites – factors that are addressed in updated guidance for care homes.
In contrast to deaths in care homes, deaths at home have continued to be higher than average since the start of the pandemic, with a third more deaths at home than expected each week from June to September. It is likely that in recent months many of these deaths would otherwise have occurred in hospital or care homes, but further research is needed to understand the reasons for the shift, and the impact on patients and families.
Will there be a long-term impact on mortality rates?
So far in 2020, there have been 54,453 more deaths than expected. That is slightly lower than excess deaths since mid-March – reflecting lower mortality than average in January and February this year. From the year so far, and with the number of Covid-19 deaths rising again, we can expect that mortality in 2020 will be well above average. This would reverse the long-standing decline in mortality, and could return mortality rates to those of a decade or more ago.
Over the longer term, it is clear that analysis of just those deaths recorded as being Covid-19 related will not give the full picture. The impact of the pandemic on health and treatment of other illnesses is not yet clear. As well as having a better understanding of how the recording of cause of death differs by setting, more detailed analysis will need to consider the overall number of extra deaths and the causes recorded for these, along with other factors such as age, gender and place of death.
*This explainer was originally published on 5 May, was then updated on 26 May, and has now been updated further to reflect the latest data from October.
Scobie S (2020) “Measuring mortality during Covid-19: a Q&A”, Nuffield Trust comment.