The authors would like to gratefully acknowledge Anita Charlesworth for her contributions in many ways throughout the course of this work, including as Chair of the Expert Advisory Group (EAG); the research team: Michael Murphy, Marc Luy and Orsola Torrisi; and Veena Raleigh, who acted as adviser to the project. They would also like to thank both Veena Raleigh and Peter Goldblatt for reviewing and commenting on the research report and this Health Foundation report, and for being part of the EAG. Thanks also go to the other EAG members: Justine Fitzpatrick, Chris White, Azeem Majeed, Asim Butt and Jean Marie Robine.

The importance of understanding trends in mortality

The Health Foundation aims to improve the health of people in the UK by learning what works to make people’s lives healthier and improve the health and care system. Mortality rates, and the length of time that people live in good health, are critical measures of how well a country is doing. Mortality rates in a population are driven by a broad range of factors, which are largely outside the control of individuals. The strongest influences on people’s health are the social, economic, commercial and environmental conditions in which they live: the ‘wider determinants of health’. Thus, governments need to be accountable and responsible for creating the conditions that can enable everyone to live healthy lives.

Tracking and understanding trends in mortality sheds light on the impact of national and local policies and can inform future planning. A better understanding of the drivers of mortality and inequalities in life expectancy will point to where action needs to be taken, what should be monitored to help foresee future changes in mortality, and what care needs may arise.

There is growing concern about recent mortality trends in the UK. The improvements seen over many years are slowing and even stalling in some segments of the population. Attempts to understand what is happening have led to extensive debate about the causes, but few widely accepted answers. While this is perhaps not surprising given the complexity of the factors shaping people’s health and mortality rates, it is important to build a fuller analysis of the factors behind these changes and whether these trends are likely to continue in the future.

There is no single lens through which to understand and interpret trends in a nation’s mortality. Changes over time are important, but so too are the absolute and relative differences between countries and groups (as defined by geography, sex, ethnicity or socioeconomic status) within a country.

Given that the UK has a lower life expectancy than many countries considered to be comparable in terms of their population structure and socioeconomic circumstances, getting to grips with what is happening is particularly important (Figure 1). As Figure 1 shows, women in the UK are doing particularly badly, both in terms of their absolute life expectancy of 83.1 years (in 2017), which puts them towards the bottom of the rankings, and that they fell down the rankings between 2000 and 2017 (indicated by the distance between the blue and red dots in Figure 1). Only the USA remain significantly lower than the UK at the end of this period. While Denmark sits below the UK in terms of absolute life expectancy, women in Denmark have experienced a much greater improvement over the 21st century, and have caught up with the UK. The picture for UK men relative to other countries is better, but there is still much room for improvement from their 2017 level of 79.5 years.

In response to the debate about mortality trends, the Health Foundation commissioned a team at the London School of Economics (LSE) and Vienna Institute of Demography (VID), led by Professor Michael Murphy, to carry out a literature review of previous research, plus new analyses into mortality trends. The aim was to provide an in-depth picture of what is happening, and a comprehensive critical assessment of the multiple potential drivers of recent trends. The research is the first in a series of research commissions over the coming years to explore current and underexplored population health issues.

Building understanding about mortality trends and drivers is particularly relevant to two large programmes of work at the Health Foundation, both of which aim to inform a long-term approach to improving health and care in the UK:

• the Health Foundation’s ‘Healthy Lives’ long-term strategy to improve people’s health in the UK, which aims to focus policy attention on the wider determinants of health and support long-term action on these to improve health and address inequalities
• a new specialist unit being set up by the Health Foundation working with academic partners across the UK to provide independent projections, research and analysis to help ensure the long-term sustainability of health and social care in the UK. The aim of the unit is to bring about more evidence-based policymaking and a shift in focus towards long sustainability.

This briefing paper is based on the research by Murphy et al. but also draws on other sources, including the review of mortality trends the Department of Health and Social Care commissioned Public Health England to carry out in 2018, and Health Foundation analysis. The Public Health England review was carried out in parallel with the early stages of this research and is referenced in this briefing and full research report by Murphy et al. This briefing focuses on themes of particular interest to the Health Foundation. The full report is available online from the LSE.

What are the trends in mortality in the UK?

Over the 20th century in the UK, mortality rates fell in both sexes, at all ages. Successive cohorts born over this period experienced improvements in survival (Figure 2, which includes cohorts born in the recent past, for which the dashed portion of the survival curves are projections based wholly on assumptions). Of those born around the beginning of the 20th century (1905 cohort), 62% survived to age 60, increasing to 89% of those born mid-century (1955 cohort). This was due to a combination of reasons. Big gains were made in infant survival in the first half of the 20th century, followed by advances in infectious disease control that helped to increase survival rates throughout adult life. In more recent decades, improvements in cardiovascular disease (CVD) mortality through a combination of health care improvements, reductions in smoking and occupational changes have helped improve the proportion of cohorts reaching older ages, with the majority of further improvements in survival rates for younger cohorts anticipated to continue at older ages in future.

These improvements in mortality led to improvements in life expectancy (derived from mortality data, see definitions in section 2) throughout the 20th century and into the first decade of the 21st (Figure 3). This has been the case in all UK constituent countries, although it is notable that absolute life expectancy differs between countries, England and Wales having the highest life expectancy throughout the period, followed by Northern Ireland, and life expectancy being lowest in Scotland.

Since 2010–2011, mortality improvements have significantly slowed or completely stalled in both sexes and in all constituent countries of the UK, and socioeconomic inequalities in mortality have been reported to be widening. In addition, high mortality in 2015 compared with 2014 caused the largest fall in period life expectancy since the early 1970s. Questions have been raised by academics, policymakers, actuarial analysts and public health experts about the factors that have caused such an apparent deterioration in mortality trends, and about the extent to which the changes are temporary or likely to continue. Finding answers to these questions will have important implications for future policy, pensions and service planning. It will also help to identify action required to foresee and minimise further slowdown or stalling of mortality improvements and widening of inequalities in life expectancy.

What is driving these changes?

There has been insufficient reflection on the complexity of the drivers of mortality trends in the academic and media debate. Conflicting views have been expressed about the relative importance of potential underlying causes. Some contributors to this debate have focused narrowly on the recent changes, without setting the trends in a longer-term or global context. Others have conflated the longer-term trend of stalling improvements in mortality with the observed peaks in mortality, particularly the high levels of excess winter mortality in 2015 and 2018. Most have taken a very narrow view of drivers, addressing them in isolation, and much of the debate on links to austerity has been limited to health and social care cuts. In some cases, attempts have been made to estimate exactly how many deaths may have been attributable to austerity, without always being entirely clear how these estimates were reached or the justification for assumptions made.

The evidence base is therefore limited and inconclusive. The recent Public Health England review of trends in mortality in England concluded that rather than being attributable to any single cause, the slowdown in improvement is likely to be the result of a number of factors operating simultaneously across a wide range of age groups, geographies and causes of death. The Health Foundation concurs with the conclusions of the Public Health England report. A deeper recognition and understanding of the complexity of the factors shaping mortality is now essential if effective policy and action to improve mortality and reduce inequalities are to be identified.

UK and international mortality trends in the 21st century

Rapid improvements in mortality rates seen in the UK followed by stalling

In all constituent countries of the UK, mortality rates for both men and women were improving rapidly, at an average rate of three months a year, during the first decade of the 21st century. Since 2011, these improvements have slowed, and even completely stalled in some segments of the population (discussed later and shown in Figure 11).

Figure 6 shows the actual age-standardised mortality rate in England and Wales for each year 2000–2018 (at the time of writing, data was not available to 2018 for the whole of the UK) and how the trend in improvements changed at the turn of the decade. Trend lines are fitted for the periods 2000–2011 and 2011–2018, and show that during the first period, age-standardised mortality rates fell by 26 deaths per 100,000 of population a year. During the second period, these improvements slowed dramatically to just under 2 deaths per 100,000 of population a year.

As discussed above (and shown in Figure 1) this is particularly concerning given the relatively low absolute levels of life expectancy in the UK, in particular for women. In 2017, life expectancy for women was 83.1 years, putting them near the bottom of the list of comparable countries; for men this was 79.5 years, placing them around the middle of the group but with ample room for improvement.

Some slowing had been predicted by experts for a number of reasons including declining gains to be had from recent reductions in CVD mortality, and increasing obesity in the population. What has occurred has happened sooner and faster than anticipated, sparking debate about the reasons and downward revision of models used to project mortality trends for purposes such as population and pension forecasts.

While this slowdown in mortality improvements has been widely reported and validated, much of the previous research and debate has focused on the increase in numbers of deaths between 2014 and 2015, which was substantial and drew much attention. As can be seen in Figure 6, while mortality was high in 2015, 2014 was a year in which there was exceptionally low mortality, which also contributed to the between-year difference.

This briefing is most concerned with the slowdown of the long-term positive trend in the rate of mortality improvement and life expectancy, rather than specific annual peaks or troughs. Of course, Figure 6, in showing the trend for the population as a whole, masks differing patterns among different population subgroups, which have not been highlighted in the previous research into the slowdown. It is particularly important to understand the slowdown or stalling in UK population subgroups, by sex, age and deprivation. This briefing explores recent trends by subgroup, but first considers the population-level trend in the context of longer-term and international experiences of mortality.

The slowdown follows a long period of improvement in the UK

Improvements in mortality in the UK go back many years. Figure 7 shows standardised death rates in constituent countries in the UK from 1950 to 2016 (the period for which data is available in the HMD). Since the 1950s, there have generally been steady improvements in age-standardised mortality rates in both men and women, across all constituent countries of the UK. The period 2000–2009, ahead of the 2010–2011 stalling, showed particularly strong improvements in the context of longer-term changes, with reductions in age standardised mortality rates of 2 to 3% per year.

It is notable that over the full period looked at, while trends have been similar across UK countries, mortality rates have been considerably higher in Scotland than England and Wales, and Northern Ireland.

The slowdown is widespread internationally, but pace and timing vary

Comparing what is happening in the UK with other similar countries can provide insight into what may – or may not – be driving the slowdown. If trends are very similar internationally, factors specific to the UK would be less likely candidates. If trends are similar but not identical, there may be common general drivers, with local factors explaining deviation from a common international pattern.

Improvements in mortality have slowed across most comparable western European countries, and the USA (Figure 8, and section 2.2 for selection of comparable countries). The slowdown has been most rapid in the UK though, with a more gradual decline from historically high rates of improvement seen internationally (with the exception of Ireland). The average annual improvement in the UK was 73% lower in the 2011–2016 period compared to the 2000–2011 period. On average, it was 40% lower for a selection of comparable countries.

Intriguingly, all constituent countries of the UK show a similar pattern, including Northern Ireland, while the Republic of Ireland is closer to the continental western European pattern.

Further analysis by the LSE showed differences between the UK and comparator countries in when the change in trend occurred. The slowdown appears to have happened both later and faster in the UK than in the comparator countries looked at, which generally saw a change in the rate of improvement from around 2005, up to 5 years earlier than the UK. The most similar country to the UK over the full period is the Netherlands: levels of mortality as well as the magnitude and timing of change were all very close (Figure 9).

The large increase in mortality seen in the UK between 2014 and 2015 also coincided with similar fluctuations in most European countries. As in the UK, older people, especially women, were most affected by excess deaths in 2015. This resulted in annual changes in life expectancy at age 65 in 2015 being negative for women in 25/28 EU states, and for men in 16/28.

The rate of further improvements in life expectancy in the UK has fallen faster than most other countries

Annual improvements in life expectancy in the UK since 2011 have been particularly low among comparable countries. This is especially concerning given the UK already had a lower level life expectancy than many similar countries prior to the slowdown, in particular for women (Figure 1). The low rate of improvement in UK life expectancy at birth in 2011–2016 for both men and women is in contrast to the previous period 2006–2011, when UK improvement rates were among the highest among comparable countries, with UK women having the largest improvement of all countries looked at over the period 2006–2011, albeit starting from a particularly low level relative to other countries (2006–2016 is the most recent decade for which comparable international data is available in the HMD). It must also be borne in mind that a slowdown was already being experienced by other countries during the period 2006–2011.

Despite the earlier start to the slowdown in other comparable European countries looked at, for the majority (14/15, the exception being Norway), life expectancy at birth still increased more slowly between 2011–2016 than it did between 2006–2011. This highlights the widespread and ongoing nature of the slowdown during this second decade of the 21st century. None of the comparator countries experienced the sustained period of strong improvement lasting throughout the 2000s that was seen in the UK, and while a reduction in rates of improvement has been widespread, a consequence of the particularly high initial rates of improvement means that the UK exhibits a more marked change of trend.

There is room for improvement: the UK has a relatively low life expectancy

While comparison with other western European countries may provide evidence against the case for UK exceptionalism that has previously been made, it is no argument for complacency. In terms of life expectancy at birth, the UK is doing worse than many other comparable countries, in particular for women (Figure 1).

Sex-specific trends in mortality

The life expectancy gap between men and women in the UK has narrowed

Women have historically had a longer life expectancy than men in the UK, but the gap has changed over time (Figure 10). This female advantage widened until around 1970, and then narrowed to the present day, with men improving faster than women over the last 50 years to ‘catch up’. The narrowing of the gap between 1970 and the 1990s has been associated with improvements in CVD mortality, which historically had a greater impact on men, and – not unrelated to CVD – much higher levels of smoking among men in the early period, followed by greater reduction in smoking in men than in women, plus declining numbers of men in high-mortality occupations. Analysis of more recent trends, driven by an ageing population with multiple conditions, requires a more sophisticated understanding of diagnosis and coding, and of the interrelationship between the condition or event causing a death, and the underlying contributors that ultimately led to the death.

The LSE research found that there has been little, if any, improvement in mortality among women in the period 2011–2016, while men have continued to improve at a slow rate (Figure 11). This has been in the context of male life expectancy that is comparable with European peers, but a female life expectancy that is among the lowest, meaning that UK women are falling further behind from an already low starting position over this period.

As shown in Figure 10, the male–female gap in life expectancy has been narrowing since around 1970, due to male life expectancy increasing faster than female. Since 2011, and the population-wide slowdown in the UK, this has been reflected in very slow improvements for men and no improvement for women. Certain subgroups have fared particularly badly, and gaining a better understanding of what is happening requires detailed exploration by age, sex and socioeconomic group.

The narrowing male–female gap in life expectancy is widespread

As in the UK, the improvement in male life expectancy at birth was higher than that for females over recent years in comparable countries. As a result, the male–female gap in life expectancy at birth in the UK narrowed by just under 1 year between 2006–2016 (Figure 10). While there was variation between countries in the extent to which this male–female gap closed over this period, there does not appear to be a markedly different pattern between the sexes emerging in recent years. Life expectancy improvements for women have approached zero, ahead of men, as they had been improving at a lower rate ahead of the slowdown. It must also be noted again though that UK women are doing particularly badly internationally in terms of both their absolute level of life expectancy and their current (zero) rate of improvement.

Age-specific mortality trends

The slowdown in improvement in mortality rates at older ages has dominated trends, but they are not the only group affected

In the UK, as in comparable high-income countries, recent trends in mortality at the population level have been determined primarily by changing mortality patterns in older age groups, where most deaths now occur. LSE analysis shows that all age/sex groups showed improvements between 2006–2011, but particularly the 70–84 age group (Figure 12). In the later period, 2011–2016, mortality improvements were reduced and low for all groups, but particularly women aged 85 and above, who experienced some worsening, and men and women aged below 50, in which zero improvements occurred.

Improvements in mortality rates are also stalling at younger ages in the UK

Mortality improvements are stalling at younger ages in the UK. In recent years, both men and women aged 35–50 in England have experienced lower mortality improvements than in the 2000s and in some cases mortality rates have actually increased. For instance, mortality rates fell for 45–49-year-olds by an annual average of 5.4 people per 100,000 of population between 2011 and 2006, but increased by an annual average of 1.1 people per 100,000 of population between 2001 and 2016. This is highly concerning and, while this group makes only a limited contribution to overall trends, needs further attention.

The negative trends in this age group have been attributed to increases in accidental poisoning arising from drug misuse, alcohol consumption and suicide (termed ‘external causes’ of death). This pattern of causes looks similar to those in the USA that are affecting young adults, to the extent that they have impacted on life expectancy in the population. In the USA, deaths due to accidents, misuse of opioids and risky behaviours – termed ‘deaths of despair’ – have led to an increase in mortality in young adults. In the USA, following some years of declining improvements in mortality, the first fall in life expectancy in the population since the 1993 HIV/AIDS epidemic was seen in 2015, followed by further falls in 2016 and 2017. The USA now sits at the bottom of rankings of comparable countries for both male and female life expectancy, and this is primarily as a result of increasing mortality among middle-aged adults, rather than at older ages.

The UK pattern for deaths in younger adults is divergent from mainland Europe

As in the UK, deaths at younger ages account for only a small fraction of all deaths in Europe. As deaths tend to be concentrated around age 80, the trends in the oldest age groups dominate overall mortality trends in the population, but detailed international comparisons by age group are worrying. While the slowdown in the population overall and at older ages in the UK is similar to its closest comparators, the Netherlands and France, the stalling of improvements in younger adults (up to age 50) that is discussed above is unique to the UK among these countries. Mortality rates in the under 50s continue to improve year-on-year in the other countries shown (Figure 13).

This highlights how looking solely at the population overall can mask important trends in subpopulations that do not contribute high numbers of deaths overall. Identifying subgroup-specific trends is important in elucidating what may be driving the trends in that group, which may be distinct from other drivers of population trends.

Socioeconomic differences in mortality trends

Socioeconomic differences in mortality have increased since around 2010, following a period of decreasing inequalities

Wide socioeconomic inequalities exist in the UK in both life expectancy and healthy life expectancy (the number of years that someone can expect to live in good health). In England, life expectancy for women in the lowest decile for deprivation is 7.5 years lower than for those in the least deprived decile (78.7 years vs 86.2). The gap is even greater for healthy life expectancy, at 18.4 years (52.0 vs 70.4), meaning that women in the most deprived decile can expect not only to live shorter lives, but to live a greater proportion of their life in poor health.

Studies assessing the impact of deprivation in mortality improvements over recent years, using differing datasets and methodologies (both ecological analyses using small area-level indices of deprivation, and analyses of individual-level data held for pension and insurance schemes), generally point in the same direction – that socioeconomic differences in mortality have increased since around 2010, following a period of decreasing inequalities (Figure 14).

Improvements in mortality among lower socioeconomic groups are lagging behind those in higher socioeconomic groups

Since around 2011, women in the most deprived areas have experienced decreases in life expectancy, while life expectancy for men in the most deprived areas has stalled. For both sexes, life expectancy has continued to improve in the least deprived areas (Figure 14). Failure of the most deprived groups to match the ongoing (albeit slowing) improvements of the higher socioeconomic groups has widened inequalities in life expectancy that had previously been reducing. This life expectancy advantage in the least deprived is in large part due to later onset of multimorbidity, and subsequent longer survival, compared with the least well off. These differences are only in part explainable by socioeconomic differences in smoking prevalence. Other factors that contribute to the development of multimorbidity, including obesity and smoking, also show wide socioeconomic inequalities.

Trends in the leading causes of death in the UK

Leading causes of death in the UK

The leading causes of death in the UK are age-dependent. Leading causes in the population overall are therefore driven by those affecting older age groups, who contribute most deaths. Given the concerning trends in young as well as older adults, understanding what people in the different age groups are dying from is important. The leading causes of death in England and Wales in 2017 by age group (20+) are shown in Figure 15.

Mortality improvements occurred in most leading causes of death prior to the slowdown

Over the first decade of the 21st century, when mortality rates were improving rapidly in the UK population, mortality improvements occurred in almost all leading causes. Improvements in CVD were particularly significant. This was due to a combination of factors: improvements in surgery, increased use of medication to lower blood pressure and cholesterol, and lower rates of smoking. Since 2011 however, CVD improvements have slowed, due to lessening impact of improvements in these contributory factors plus, most likely, increases in other risk factors in the population including poor diet, obesity and type 2 diabetes. Reductions in the pace of improvement of cancer – the other main set of causes of death – also occurred, but at a much lower rate than CVD.

Much attention has been paid to the role of both influenza (flu) and dementia over the period of the slowdown. With both, the challenges of establishing and assigning causation of death in older individuals who commonly have multiple conditions must be borne in mind.

Flu is rarely recorded on death certificates as an underlying cause. Instead, models for estimating the scale and spread of flu outbreaks use multiple sources of data. Dementia is now the most commonly coded leading cause of death in older people. While this is in part due to population ageing and improvements in treatment for other conditions, changes in diagnosis, definitions and coding practices have also contributed to the increase over the past 5 years. The increase in dementia coding as a leading cause may also be leading to underestimation of what is happening with CVD and respiratory causes over recent years, as deaths that may previously have been coded with these as a leading cause may have been attributed primarily to dementia (Figure 16).

Drug-related deaths and suicide have been increasing in younger adults

Drug-related mortality rates (and numbers of deaths) – which include deaths from prescribable drugs – have increased greatly in the UK in recent years, most markedly at ages 30–49. However, the rate differs markedly across the constituent countries and the regions of England. The rate of drug-related deaths in Scotland in 2018 was the highest it has ever been, making it – as widely reported in the media – the highest in Europe and on a par with the USA (at 218 deaths per million population vs 217 per million population in the USA – 2017 data – for the populations overall). England and Wales have much lower rates, but 2018 saw both the highest level and greatest annual increase (16%) since records began in 1993. Although rates are far lower in England and Wales than Scotland, there is significant variation within these. The rate was highest in the North East of England, at 142.7 deaths per million, and lowest in London, at 56.5 per million. Rates were also high in Wales, the North West, and Yorkshire and the Humber.

The leading cause of death in younger adults is suicide. Suicide disproportionately affects men, with rates three times higher than women in the UK. While the overall male suicide rate decreased to the lowest level in over 30 years in 2017, this masks an increase in suicide rates in younger men across the UK, with an increase of 7.4% between 2016–2017 in men aged 45–49. It should be noted that there are issues inherent in the coding of suicide vs accidental poisoning, in particular with drug-related deaths and, for some groups of the population, looking at rates and change in rates for suicide mortality is not reliable due to relatively low numbers of deaths.

Are current patterns indicative of future trends?

From the analysis of UK mortality trends, it still is not known whether current trends represent a long-term shift from improvement, or a short-term deviation from a norm of continuing improvement, and caution must be applied in drawing conclusions based on short-term changes. Recent data indicate some rapid and unexpected improvement in mortality in England from mid-2018, continuing into early 2019. These are very recent – and very short-term – changes, coming after the unusual period of slow improvements, and do not provide a good basis for drawing conclusions about underlying trends, or what may happen in future.

The last period of sustained low improvements or even reversals in the UK was in the 1970s, and this was followed by a long period of improvement. Examples from other countries provide some cause for optimism. Japan showed little improvement in mortality in the first decade of the 21st century, when other countries at similarly high levels of development were improving rapidly, and there was a view that maximum life expectancy had been reached. However, in the most recent period Japan has shown very high rates of mortality improvement.

Examples of sustained poor performance are rare, although the USA experienced this in relation both to earlier periods and other high-income countries. Life expectancy in the USA has fallen in each of the last 3 years for which data is available (2015–2017). This has only happened once in the past century, in the exceptional period between 1915–1918 (World War I). There are no other documented cases of long-lasting increases in mortality in developed countries, other than the Soviet Union between 1960–1995 (this is discussed in detail by Murphy et al.).