Key points

  • In England, deaths from causes other than COVID-19 have been lower than usual for 80% of the pandemic. In January to September 2021, this was equivalent to about 34,000 (or 9%) fewer deaths than we would expect, based on historical mortality patterns. This analysis explores some of the questions arising from this reduction. 
  • Some of this reduction can be explained by ‘displaced mortality’: people who would have died from another cause, but died earlier from COVID-19 instead. But this cannot account for all the difference. If people dying from COVID-19 were typical of their age, around 9% would have died from another cause in the following 12 months – around 8,500 people. Even if these people were more likely to die than typical of their age, this would still not account for anywhere near the 34,000 ‘missing deaths’. 
  • During the pandemic, deaths from eight of the nine leading causes of death in England were lower than usual. The greatest reductions in percentage terms are for influenza/ pneumonia (48%) and chronic lung conditions (25%). As well as displaced mortality, reductions here are largely due to the lower prevalence of infectious diseases as a result of restrictions such as social distancing. The reasons behind reductions in deaths from stroke, heart attacks, angina and cancers are less clear. 
  • Understanding these changes is challenging because multiple factors act to change mortality risk in different ways – with some impacts seen immediately, and others taking time to emerge. To help researchers and policymakers try to make sense of this, we propose a model to illustrate the impact of these different factors.  
  • We apply this framework to analyse deaths from lung cancer, which have been 7% lower than usual. We conclude that this reduction is likely not a ‘true’ reduction (as the risk of developing lung cancer during the pandemic is unlikely to have changed). Instead, these are probably due primarily to the reduced diagnosis of lung cancers, with some reductions due to displaced mortality. In the longer term, we may expect an increase in lung cancer deaths given these delays in diagnosis.  
  • Policymakers and researchers should be wary of drawing quick conclusions about the pandemic’s effect on our health – and should instead, for each condition, unpick the complex threads of displacement, the recording of deaths, and any changes in risks.  
 

Introduction

Despite continuing deaths from COVID-19, in 2021 mortality from other causes has been about 9% lower than usual (in England, January to September 2021). In the midst of a pandemic, how could this have happened? Did the pandemic, and our lockdowns, unexpectedly reduce deaths from other causes? Or have deaths during the pandemic ‘displaced’ mortality that would otherwise occur at a later point in time?

This analysis explores the story behind this 9% figure – breaking down deaths by different conditions to try to understand why deaths from some causes have risen, while others have fallen. We do this by analysing excess deaths (deaths compared to the 5-year average) for different conditions, and propose a model to explain the different factors that have acted to both increase and decrease non-COVID mortality during the pandemic. This will help us to interpret and understand the effect of the pandemic on non-COVID mortality. And, in doing so, suggest what might happen to mortality from key conditions such as cancer in the coming years.

 

Did COVID-19 really reduce mortality from some causes?

Since the start of 2020, around 125,000 people have died of COVID-19 in England (ONS, 2021). But this does not give us the whole picture of mortality associated with the pandemic. There are other ways – both positive and negative – in which the pandemic has affected deaths.

The most obvious way is through non-COVID deaths: deaths caused by the pandemic and lockdown restrictions, for example, due to delays in accessing health care, or because of backlogs in providing ‘usual’ care. But while the pandemic has increased the risk of death from some causes, it may have reduced it from others.

The figure below shows the number of excess deaths since the start of 2020. We calculate this by estimating an ‘expected’ number of deaths for 2020 and 2021, based on the 5-year average, and compare this to the actual number of deaths – split into those caused by COVID-19 and those from other causes. It shows that the number of non-COVID deaths has been lower than the 5-year average for around 80% of the pandemic period. The major exception is a period at the start of the first wave when there was limited testing capacity – meaning COVID-19 deaths were not always recorded as such.

The table below shows the number of excess deaths by COVID-19 and from all other causes from 1 March 2020 (taken as the statistical start of the pandemic) to 31 December 2020, and from 1 January 2021 to 30 September 2021. This shows that, for the 2021 period (for which the recording of COVID-19/non-COVID deaths is more reliable), excluding deaths from COVID-19, there have been 34,000, or 9%, fewer deaths than would normally be expected.

We use the 2021 period for the remainder of this analysis as it means we avoid the recording issues experienced in the first wave of the pandemic. Since COVID-19 is a novel virus, all deaths caused by COVID-19 are classified as ‘excess’ as none have occurred in previous comparator years.

Excess deaths by cause for different periods of the pandemic, England

Period of the pandemic Deaths caused by COVID-19 Excess deaths from other causes All excess deaths
2020: 1 Mar 2020 to 31 Dec 2020 72,923  572  73,495 
2021: 1 Jan 2021 to 30 Sep 2021 51,716  -33,864  17,852 
1 Jan to 31 Mar 2021 44,355  -21,871  22,484 
1 Apr to 30 Jun 2021 1,182  -9,141  -7,959 
1 Jul to 30 Sep 2021 6,179  -2,852  3,327 

Source: Office for National Statistics (table 13)

So why are there 34,000 ‘missing’ non-COVID deaths? An obvious hypothesis is that some of these 34,000 people died from COVID-19 instead – the mortality was therefore displaced, occurring earlier than it otherwise would have. We see a similar pattern after a ‘bad flu year’ – there are often fewer deaths among older people in the post-winter months as some died earlier than would be expected from flu.

Certainly, some of the people dying from COVID-19 would have died within the next year. The figure below illustrates this for 1,000 85 year-olds in England (a little above the median age of death from COVID-19). Using historical age-specific mortality rates, it projects how many would die in each subsequent year. It shows that if those dying from COVID-19 were typical of their age, we would have expected 90 of these 1,000 (9%) to have died from other causes in the subsequent 12 months.

If we apply this approach to the people of each age who have died from COVID-19 , we estimate that of the 125,000 people dying from COVID-19 in England, around 8,500 (7% of the total) would have died from other causes in the first 9 months of 2021. Even if those dying from COVID-19 were more likely to die than typical of their age (because the most vulnerable and those with comorbidities typically experience this as a higher risk), this would still not account for anywhere near the 34,000 ‘missing deaths’ from causes others than COVID-19. Some other explanation is needed.

 

Mortality patterns for specific non-COVID conditions

It might be tempting to conclude that measures to control the spread of the pandemic somehow prevented deaths from other causes. That even though access to the NHS was reduced, and waiting times are now longer, this did not result in ‘additional’ non-COVID mortality. In reality, the picture is more complex than this.

Looking at mortality patterns for specific conditions helps us to understand this complexity. The table below shows deaths from January to May 2021 for the main nine causes other than COVID-19 (which is the main cause), compared with the number of expected deaths. It shows that:

  • Deaths from seven of the nine causes are significantly lower than usual, with an overall reduction for all nine causes of 12% (or 24,000 deaths). This accounts for about 70% of the total reduction in deaths from all non-COVID causes.
  • The greatest reductions in percentage terms are deaths from influenza/pneumonia (48%) and chronic lung conditions (25%). These reductions are largely due to lower prevalence of infectious diseases as a result of social distancing and mask wearing.
  • The only cause for which deaths are significantly higher is ‘symptoms, signs and ill-defined conditions’ – a broad category used where doctors are unclear on the exact cause of death.
  • The three cancers in the list have seen mortality reductions of around 4%. The reasons for this are unclear and, looking at a longer list of cancer deaths in 2020, not all cancers have seen reductions.

The nine leading causes of death after COVID-19, January to September 2021

Cause Deaths Share of all non-COVID deaths 5-year average Excess deaths Excess deaths as a share of usual deaths
Dementia and Alzheimer's disease 40,872 11% 45,977 -5,105
 
-11%
 
Heart attacks and angina 38,683 10% 40,319 -1,636 -4%
Stroke 19,903 5% 22,498 -2,595 -12%
Lung cancers 19,508 5% 20,964 -1,456 -7%
Chronic lung conditions 16,860 4% 22,609 -5,749 -25%
Symptoms, signs and ill-defined conditions 10,951 3% 8,994 1,957 22%
Bowel cancers 10,529 3% 10,379 150 1%
Influenza and pneumonia 10,373 3% 20,011 -9,638 -48%
Blood cancers 8,111 2% 8,293 -182 -2%

Source: Office for National Statistics

 

Explaining the changes in mortality

So what is causing these reductions in mortality? The answer is complex, with the effect of the pandemic on non-COVID mortality influenced by four main factors: risks, access to health care, how deaths are recorded, and displaced mortality.

We summarise these factors in the diagram below. This aims to show how different factors act to both increase and decrease mortality during the pandemic. The diagram is illustrative and not based on real data.

  • Bar 1 imagines the total number of people who would die from a particular cause in an average, non-pandemic year.
  • Bars 2 and 3 suggest what might have happened to these deaths in the pandemic year – whether they occurred anyway, occurred but were caused by something else, or were avoided. It also considers what other extra deaths might have happened:
    1. Bar 2 considers the deaths of those same people during the pandemic. They will either have died from the same cause (and been recorded in the data as such), have died from that cause but been recorded as something else (eg because of changes in death coding during the pandemic), or will have died from COVID-19 instead.
    2. Bar 3 considers the additional and avoided deaths that would, or would not have occurred without the pandemic. These happen for two reasons – because the pandemic caused a change in risks (eg making someone less likely to die from flu, but more likely to die from alcohol poisoning), or because delays in treatment meant someone was more likely to die from a particular cause (eg a stroke that may not have killed someone in a non-pandemic year).
    3. The total recorded deaths from a given cause during the pandemic will be the sum of A, D and E, and is shown by the bracketed area 4.
The factors that affect non-COVID mortality during a pandemic
Model showing the factors that affect non-COVID mortality during a pandemic

Below we explain how each of these factors might work. In practice, the same factor can act to both increase and decrease mortality, and the overall change in mortality for a given cause will be the result of interplays between these factors.

  • Changes in risks as a result of differences in behaviours and environmental factors (eg accidents, pollution, alcohol consumption and infectious diseases). These can both decrease and increase mortality. Lockdowns, for example, were instituted to reduce the spread of COVID-19. But they also incidentally reduced the spread of other infectious diseases. From January 2021 to September 2021, there were 14,000 fewer deaths in England and Wales from influenza, pneumonia and chronic lung conditions. Conversely, other risks increased, such as the heavy consumption of alcohol. In 2020 there were 1,400 more deaths from cirrhosis and other diseases of the liver (up by 17%). The latter make up a large proportion of alcohol-specific deaths, which in 2020 reached the highest level for 20 years – up by 20% on the previous year.
  • Changes in people’s access to and use of health or social care. The suspension of routine NHS care discouraged people from using the NHS, suspended some care entirely and has led to much longer waiting times for treatment. We discuss this at length in our long read Waiting for care, which highlighted that 58,000 fewer people than usual had had a hip replacement due to the pandemic; and there were 40,000 missed or delayed diagnoses of diabetes. Not all of these delays will have led to changes in mortality. But some will have – particularly where emergency care or cancer diagnoses (discussed further fellow) were delayed.
  • Changes in how deaths are recorded. For example, during the first wave of the pandemic when testing capacity was low, particularly in care homes and the community, some COVID-19 deaths may have been recorded as deaths from dementia. This is likely to have been addressed during the second wave, when we actually see a decrease in deaths from this cause.
  • Displaced mortality: people who would have died from another cause, but died earlier from COVID-19 instead. Assuming those dying from COVID-19 faced the same non-COVID mortality risk as others of their age, around 8,500 would have died between January and September 2021.

Lung cancer: an example of the complexity

There were 7% fewer lung cancer deaths between January and September 2021, compared with what we would expect. Below we use our framework to unpick this figure – and explore why and how lung cancer mortality has changed.

Our analysis suggests that the apparent reduction in lung cancer deaths is likely not a ‘true’ reduction (as the risk of developing lung cancer during the pandemic is unlikely to have changed). Instead, reductions are probably due primarily to reductions in the diagnosis of lung cancers, with some reductions due to displaced mortality (such as some lung cancer patients dying of COVID-19 before they died of lung cancer). In the longer term, we may expect an increase in lung cancer deaths. This is because those same delays in diagnosis will mean some treatable cancers will not have been diagnosed early enough.

What caused the 7% reduction in lung cancer deaths from January to June 2021 (compared with the 5-year average)?

Change Impact
Risks: No effect. The risk of developing lung cancer over the short term is very unlikely to have changed. It is possible that the risk to people with lung cancer of death from co-morbidities, such as flu, may have decreased.
Reduced access to care: May increase deaths in 2021. Researchers have estimated deaths from lung cancer could increase by more than 6% in the second year of the pandemic as a result of delays in diagnostic and treatment services.
Changes in how deaths are recorded: Likely reduced recorded deaths in 2020. Between April 2020 and March 2021 there were 10% fewer ‘first’ treatments for lung cancer, likely the result of reductions in diagnostic services. Because lung cancer is particularly aggressive, with 60% dying within a year of diagnosis, more people will therefore have died from undiagnosed cancers. In these cases their deaths may have been recorded as from another cause.
Displaced mortality: Likely reduced deaths in 2020. If those developing lung cancer were to have the same age-specific risk as the general population of dying from COVID-19, around 1% would have died from COVID-19 instead. Those with lung cancer may be heightened risk of dying from COVID-19 (if they caught it) so the true percentage may be higher. However, it is still only likely to account for a small proportion of the 7% reduction in deaths.

 

 

Conclusion

The story of mortality from this pandemic does not begin and end with deaths from COVID-19. Mortality across a host of other causes – both infectious and non-infectious – has changed, sometimes dramatically, during the past year and a half. There were almost certainly real decreases in deaths from some causes – notably from other infectious diseases like flu. But for many other conditions, the apparent fall in mortality during the pandemic really masks changes due to a host of other causes, including displaced mortality and changes to recording deaths.

A 7% fall in lung cancer may, at first glance, seem like a positive news story. In reality, it likely foretells an increase in mortality in future years, as delays in diagnosis catch up. These changes in mortality are – to put it mildly – quite perplexing. Policymakers and researchers should, therefore, be wary of drawing quick conclusions about the pandemic’s effect on our health. Instead they should, for each condition, unpick the threads of displacement, recording, and changes in risks. COVID-19 changed the way all of us lived, and the ways many have died – what these changes mean for future mortality is still an open question.

Further reading

Data analysis

Waiting for care

26 August 2021
Data analysis

How has the suspension of routine NHS care affected people’s health and wellbeing? This analysis...

REAL Centre

Discover

REAL Centre

Discover

The Health Foundation’s REAL Centre is here to support better long-term decision making in health...

REAL Centre
Kjell-bubble-diagramArtboard 101 copy

Get social

With general practice facing unprecedented pressures, Fuller’s recommendations will do little to improve the immedi… https://t.co/wABrlIk5dE

Follow us on Twitter
Kjell-bubble-diagramArtboard 101

Work with us

We look for talented and passionate individuals as everyone at the Health Foundation has an important role to play.

View current vacancies
Artboard 101 copy 2

The Q community

Q is an initiative connecting people with improvement expertise across the UK.

Find out more