The authors are grateful to REAL Centre colleagues for their input, support and guidance and, in particular, Toby Watt for his analysis of consultations and diagnoses in general practice.

The Strategy Unit

The Strategy Unit provided the detailed analysis and modelling of activity in chapter 4, alongside guidance and comments, with particular thanks to:

Steven Wyatt, Paul Seamer, Peter Spilsbury.

We would like to thank the following:

• The members of the REAL Centre Oversight Board for their comments and advice: Sir Andrew Dilnot, Professor James Buchan, Professor Michael Chernew, Dr Jennifer Dixon, Raj Jain, Branwen Jeffreys, Professor Ashwin Kumar, Gabrielle Mathews, Mark Pearson.
• Ben Gershlick, Rosie Hamilton, James Lewis, Nihar Shembavnekar and Ruth Thorlby who reviewed the report.
• Members of the Health Foundation communications team, particularly: Kate Addison, Sean Agass, Creina Lilburne and Pete Stilwell.

Errors and omissions remain the responsibility of the authors alone.

When referencing this publication please use the following URL: https://doi.org/10.37829/HF-2020-RC10

Health accounts for a large and growing share of public spending

Health care is one of the largest parts of the UK economy, accounting for 10% of GDP. In common with most industrialised nations, the majority of health care spending is publicly funded. In 2018/19, publicly funded health spending in England totalled £127.0bn and represented more than a fifth of total expenditure on services.

The NHS is by far the largest part of the publicly funded health care system in the UK. It is a vital part of the welfare state, providing care to both improve population health and relieve pain and suffering. As well as providing care, the NHS acts as an insurer so that people do not have to plan for potentially high health care costs, removing a source of fear and anxiety and preventing exclusion from treatment. As a universal, free at the point of use service (with charges in only a few areas), the NHS is in the top three of 22 OECD countries in terms of low financial barriers to care.

The impact the NHS has on people’s health extends well beyond its role as a provider of treatment. The NHS has an important role to play in addressing health inequalities. This includes ensuring access to care is based on need rather than the ability to pay, but also in providing prevention services that help to reduce some of the health impacts of socioeconomic disadvantage. As large employers, purchasers, and capital asset holders, health care organisations are well positioned to use their spending power and resources to address the adverse social, economic and environmental factors that widen inequalities and contribute to poor health.

Health care has also been a growing part of the welfare state. With so much spending coming from public funds this impacts public finances. Since 1948 overall public spending as a share of GDP has not increased (remaining at ~40%) so health care spending accounts for an increasing share of overall public spending. This is in part due to reductions in public spending in other areas (such as defence) by successive governments. In recent years concerns have been raised that additional health care spending is ‘crowding out’ spending on other public services – specifically education and social care, which play vital roles in safeguarding health and wellbeing.

For these and other reasons, government decisions about the amounts that are allocated to health care and how the money is spent must be informed by the best evidence and analysis, with careful consideration of trade-offs.

The NHS is shaped by decisions at all levels

Decisions about the NHS are taken at many different levels, and each decision offers a range of choices and trade-offs for decision makers to consider. These can range from national policy and strategic choices about the overall level of funding for the NHS and the goals and priorities the NHS should target (eg better cancer care), to regional and local decisions about the funding and design of different types of services. Decisions are also taken by individuals, such as by clinicians about the treatment options they discuss with patients and the time they spend with them, and by patients themselves (eg whether to attend A&E or another service). Collectively these decisions affect every aspect of how health services are delivered: how much money the country chooses to spend on health and care in comparison to other areas; what we do and do not spend that money on; and who benefits from that spend.  

Decisions also need to take account of the context in which health services are funded and delivered, which changes over time. As the future unfolds, new challenges and opportunities emerge and new information becomes available, meaning that decisions need to be continually revisited and adjusted. It is the role of policymakers to make decisions that enable the NHS to adapt, respond and take advantage of these changes. This means taking account of the long term as well as the short term.

As COVID-19 has vividly shown, uncertainty about the future makes this far from straightforward. But policymakers cannot avoid long-term decisions – it takes more than a decade to train a doctor; and buildings, equipment and the IT infrastructure last for many years and shape the context in which care is provided.

Further, we know that there are a number of cognitive biases that can affect decision making, including in health. For example, the planning fallacy causes consistent underestimation of the time it takes to complete a given task and diminishes the relevance of past experience. There is a need to be aware of the potential for bias in decision making and ensure the assumptions on which decisions are based are supported by evidence.

Long-term and strategic decision making

Policymakers and system leaders need to balance two goals: 

1. ensuring that today’s service is effective and efficient, while 
2. planning and investing to ensure that the service can adapt to potential future changes.

The House of Lord’s Select Committee on the Long-term Sustainability of the NHS argued that despite the enduring nature of the NHS, which is now over 70 years old, it is held back by an excessive policy and leadership focus on the first of these goals – the short term.

Despite enduring institutions and a permanent civil service, myopia is still problematic across many areas of policy. Our current political system, with 4 to 5-year electoral cycles and highly centralised accountability for ministers, is a large driver of this. Recognising the negative consequences of short-termism in economic and fiscal policy, governments have sought to counterbalance this with bodies that are explicitly tasked with a longer term and systemic focus, such as the Office for Budget Responsibility. To date these have been focused on the economic infrastructure. However, the challenge for much of the welfare state infrastructure is no less and, as COVID-19 has shown, they are interconnected. 

Yet while short-termism is a key challenge, it is not the only one. Other areas in which decision making could be improved are:

• The allocation of resources to different aspects of the health and care system. For example, there is a clear and increasing imbalance between spending on health care and social care.
• Alignment between desired service delivery models and the workforce plans, capital investment and funding allocations to achieve them. For example, despite intentions to move care from hospital into the community, the number of district nurses and health visitors have declined in recent years.
• Incorporating public opinion so that the level and quality of care provided by the health care system matches public expectations and reflects what the public (as taxpayers) are prepared to pay for. Public opinion tends to be taken into consideration at a moment of crisis following years of low spending. This then leads to a period of catch-up, compounding the tendency towards ‘feast and famine’.

Improving the evidence base for better decision making

The COVID-19 pandemic has put the relationships between researchers and policymakers, and between evidence and policy, under the spotlight. In some ways the pandemic has shown how well research and policy can come together, rapidly producing and sharing evidence.

However, it also highlighted challenges in ensuring evidence is comprehensive, readily accessible and easy to understand and interpret. For example, it has been difficult to assess the trade-offs in policies that seek to minimise the direct burden of COVID-19 on hospitals, but impose costs on those with other health conditions and on the economy and society more generally.

More broadly, there are systemic weaknesses in: 

• the information base, and depth of understanding of the key demand and supply trends that influence health and social care
• the understanding of the policy and service design implications and choices resulting from those trends 
• the accountability of policymakers for the longer term implications of current decisions on the resourcing and design of health and social care 
• the overall approaches to planning for the long term, specifically how to deal with uncertainty over future scenarios and risk.

The role of evidence and analysis

Although most decisions will involve elements of value judgement – such as the desired balance of spending between younger or older people – high-quality evidence and analysis play a vital role in informing decisions. This often involves comparing the costs and outcomes of different options over different future time horizons and requires two major elements:

1. Understanding the future trends and scenarios shaping the world in the absence of any change. For example, changes in the prevalence of long-term conditions.
2. Understanding the consequences of different policy choices against the backdrop of these trends and scenarios. In some cases, policy choices may affect the trends (eg improved health care could change the prevalence of long-term conditions).

Modelling plays an important role in both. But any model is only as good as the assumptions it is based on. Studying what has happened in the past can help us generate and challenge assumptions and understand more about the underlying dynamics of a situation. For example, the decision about how much outpatient activity is needed to meet population demand could be based on a model that includes: population size and age, levels of morbidity, technological advances and costs, clinician behaviour, and public expectations. However, assumptions about many of these aspects and how they interact are highly uncertain. Even learning from the past may not be useful for all future challenges. It is therefore important to balance learning from the past with robust approaches to planning for the future and dealing with uncertainty.

Learning from trends in NHS activity

NHS activity refers to what the NHS does. It is used as a proxy for the NHS’s role in improving the health and wellbeing of the population, particularly in the absence of a full range of outcome measures. While there are many factors affecting population health, the amount and quality of care the NHS provides is one key factor. Influencing the volume and quality of activity is therefore a crucial way that policymakers seek to influence outcomes.

Measures of activity are widely used throughout the NHS. Activity measures are often used as the basis of contracts between a commissioner and a provider and are used to track the use of services and to identify and monitor trends. Estimates of future activity are also widely used as inputs to many important decisions about the NHS. Decisions on major capital builds, such as new hospitals, are based highly on activity projections, as are decisions about funding and workforce training places.

Although future projections of activity are important and widely used for planning, what drives activity over the medium term is not always well understood. This report aims to help national and local policymakers better understand trends in activity, in the hope this will aid better planning. We outline a framework for understanding health care activity that includes changes in demand, changes in supply, and political and policy choices. We then describe trends in activity and, where possible, we have modelled the proportion of any change that can be explained by population size, age, gender, and proximity to death. We then suggest explanations for the remaining growth using the factors described in our framework, before suggesting lessons for the future.

Key points

• The factors influencing health care activity are wide ranging and complex. We introduce a framework to help illustrate how these factors interact.
• The supply of health care is determined by the amount of health funding and the availability of inputs (such as staff, medicines and diagnostic equipment).
• A range of factors influence demand for health care, including: the size of the population, age structure and health status, the availability of treatments and individual choices about when to seek health care.
• Where a patient’s need for health care is not met, a gap emerges. There is nearly always a gap, but a widening gap creates pressures for action.
• There are several things policymakers and planners can do in response: increase funding, change what funding is spent on, or increase efficiency. The challenge is choosing the right intervention.

Health care activity framework

In the UK most health care is publicly funded and free at the point of access. This is important context in framing how we think about the supply of and demand for health care and the interaction between the two. Even when left to market forces, the market for health care behaves differently to the market for other goods and services. One significance of public funding, however, is that the patient is not aware of the price of care at the point of receiving it, and often neither is the supplier. This is important, as in markets for other goods and services, price acts as a signal to both suppliers and consumers that helps establish a price at which supply and demand balance.

Overall, this means that demand for health care will almost always exceed supply, as price/cost will not constrain demand as it does in a private market. Given a limited budget, governments must use other mechanisms to allocate the finite supply of publicly funded health care. These mechanisms include the scope of services, National Institute for Clinical Excellence (NICE) guidelines, clinical thresholds for care and waiting lists. Government can exert control through policy choices, with one key decision being the amount of public funding allocated to health care. This, in part, determines how much supply of health care the government will fund to meet demand, although other factors are also important. For example, workforce policies: the government may state an intention to increase funding to recruit more nurses, but training, immigration and other workforce policies will influence the potential supply too. When supply for health care meets a demand for health care, we see this as health care activity.

The interactions between policy, funding, supply, demand and activity are captured in the diagram (Figure 1). When taking decisions, policymakers and system leaders must consider how to balance these factors, both to achieve short-term goals and to ensure the longer term resilience and efficiency of the system. Doing both is often impossible and trade-offs need to be made.

Figure 1: Framework for health care activity

4

2

3

1

Funding and supply of health care

Gap

Policy

Health care activity

Demand for health care

[1] Funding determines how much supply (staff, facilities, drugs and technology) can be afforded

In England the amount of NHS funding depends on the overall amount of government spend (mostly raised through taxation) and the proportion allocated to health. Funding is used to purchase inputs to health care activity, including labour (staff), goods and services (drugs, equipment) and capital assets (buildings and machinery). The overall efficiency of the system depends on realising the optimal combination of inputs. The availability of inputs, which helps determine the unit cost (total expenditure incurred to produce one unit of output, such as an appointment with a physiotherapist), depends on long-term decisions such as investment in capital and in the workforce training pipeline.

Other forces help to determine unit costs. For instance, health care is labour intensive, and it competes with other sectors of the economy to recruit staff. Historically, productivity has risen faster in other sectors, such as manufacturing, driving wage increases. The health care sector must in turn match these wage increases in order to recruit enough staff, but it is often unable to make equivalent productivity gains due to the labour intensity of health services. This is known as Baumol’s cost disease, and is posited as one reason for rising health care costs over time.

[2] The demand for health care is driven by a complex combination of factors

Firstly, demographic factors, including the size, age and gender profile of the population. All else being equal, a growing population increases demand for health care, and vice versa. Age profile is also important: almost everyone uses health care services but use generally increases with age because of its association with morbidity and proximity to death – the latter of which independently drives some health care use. Socioeconomic deprivation and ethnicity can also affect demand for services.

Demand is also influenced by changes in the underlying health of a population (changes to disease prevalence and condition specific mortality rates) and by technological and medical advances. A new treatment for a previously untreatable cancer will increase demand for care, but a new vaccine that eliminates a condition will lead eventually to a decrease (eg the polio vaccine). New innovations can also influence the type of services demanded (for example, advances in surgical techniques have led to many operations being performed as day surgeries).

Therefore, a health care need exists where there is an identified health need for which there is an acceptable treatment. This hints at the importance of patient behaviour and knowledge, and societal norms in determining health care demand (eg do I recognise I have a health need?). However, demand for health care goes beyond this definition. Firstly, health care is about more than just treatment and includes activities such as prevention. Secondly, policies and factors outside of health can impact on demand. For example, the wider determinants of health are important. There is therefore a tension between the provision of health care activities that meet a present need and other activities that might reduce future need, but are less urgent and for which there may not be explicit demand.

[3] When demand grows more than activity, there will be a gap – but interactions are sometimes more complex

A demand for health care may be met with a corresponding supply of health care, which in turn may lead to activity and improved health outcomes. However, not all demand for health care is met, resulting in a gap between demand and activity. There will nearly always be some kind of gap as the NHS has finite resources and not all health care needs are treated or identified.

However, the size of the gap may vary over time: in a period where demand is growing faster than supply, it will widen. This is often measured in terms of waiting times, for instance for A&E or elective care, although a gap can also manifest in health outcomes or in access to care. What gets measured will influence how gaps are identified.

Moreover, the relationship between the supply of and demand for health care is complex and dynamic. Increasing supply may not always reduce the size of the gap – it may also increase demand. For example, making it easier for patients to access a service by introducing a telephone helpline, may lead more patients to seek care than previously. This is sometimes referred to as supplier-induced demand: the provision of health care that is led by supply factors and not necessarily in response to an explicit demand.

[4] Governments can try to influence the size of the gap through their policy choices

Action by policymakers or planners can affect the size of the gap between health care demand and supply. Very broadly, there are three potential areas of action: change the amount of resources; change the allocation of resources; change the efficiency with which resources are used. What the government does will depend on the context and its understanding of the nature of the gap.

If a widening gap is due to a sustained period where activity growth has not kept pace with demand and, for example, waiting times for treatment increase, there will be political pressure to increase activity. This generally leads to calls to increase funding. However, spending more means raising taxes or reducing spending on other services, neither of which may be desirable or popular.

Simply increasing activity may also be insufficient if the answer is not ‘more activity’ but ‘different activity’. Governments may then try to reallocate resources within the system to meet a specific area of demand. Or they may look to moderate demand itself, perhaps through preventative programmes (eg smoking cessation) or by shifting expectations of services (eg campaigns to reduce A&E use). Not all these decisions are taken centrally, with local commissioners setting priorities for their areas.

Governments may also look to increase productivity (or efficiency): to do more with the same resources. This may be by reducing costs (eg by using generic medications rather than branded), or reducing spare capacity (eg by maximising bed occupancy). Further options include rationing the availability of services through higher thresholds for care, and the systematic elimination of activities that are not deemed cost effective. However, the options for making efficiency gains are limited and, in some cases, have natural restrictions as to what they can achieve (eg there is a limit to how far the number of beds can be reduced).

The challenge for both national and local decision makers is to choose the right intervention at the right time.

Summary

Activity measures are used across the NHS. Measures of recent and past activity are used in monitoring against plans and contracts between commissioners and providers. Estimates of potential future activity – needed to achieve specific ambitions under different future scenarios – are used to inform decisions about future levels of funding and staffing. And local service planners, working within an overall budget, will estimate changes in activity levels in different settings resulting from changes to policy or services – such as opening a new minor injuries unit, or investing in a falls prevention service.

Over the past two decades the amount of care the NHS has provided has, in many cases, substantially exceeded the amount expected based on changes in demographic factors. However, this is not universal and within that global increase the pattern varies considerably for different services.

Overall growth

Overall, the NHS has seen significant growth in total activity over the past two decades. The amount of care provided by the NHS increased by 114% between 2000/01 and 2017/18, an annual average of 4.6% per year. We estimate that hospital activity increased 4.7% per year, with planned care (6.1%) growing by almost 2.5 times the rate of emergency care (2.6%).

The expansion in activity means the population receives significantly more health care today, which has contributed to lower mortality rates for a range of conditions and, up until 2010, increasing life expectancy. However, improvements and increases in care have not been uniform over the past two decades. Between 2000/01 and 2010/11 activity rose by 5.3% a year, quality of care improved, and life expectancy grew by nearly 3 months a year for women and almost 4 months a year for men. As funding slowed after 2010 so activity slowed; between 2010/11 and 2017/18 activity grew by 3.6% a year, quality gains slowed and life expectancy has stalled overall and for some groups (notably women younger than 50 in deprived areas) declined.

The extent of the growth in health care activity differs across services, meaning there have also been major shifts in the composition of output and spend by the NHS (Figure 31).

Figure 31: Annual growth by service: projected by demographics, explained by other factors and observed

Source: Various, see Annex, *full model

The pattern here is one of a shift from community and primary care to hospital care and, in particular, to diagnostics and elective procedures. This is in spite of explicit policy ambitions to shift care from hospitals to the community.

This has been accompanied, and possibly enabled, by large reductions in the average unit costs of diagnostic and elective procedures. For elective care, this has been achieved by two changes:

• a shift away from care with an overnight stay (average cost £3,894 in 2017/18) – in 2000/01, 33% of elective procedures were inpatient; in 2017/18 the figure was 5%; and
• the rapid growth in much cheaper (average cost £140) outpatient procedures – they now account for 62% of all planned procedures.

Improvements in technology allowing more outpatient procedures and day cases is one factor leading to the increase in elective procedures. Likewise, the falling price of certain drugs helps explain the rapid volume growth in community prescribing as drugs become relatively better value for money (compared to other uses of the budget).

As such, shifts in spend have been less dramatic than shifts in activity. However, this may also reflect the difficulty in reallocating spend across services. This may have contributed to ongoing challenges in shifting care from hospitals to our communities.

The contribution of demographics

Demographic factors appear in our framework as a determinant of the demand for health care. Where possible in our analysis we have accounted for population size, age and gender – factors that can be modelled and are commonly used by NHS planners to project forward activity. In addition, we model proximity to death, because for some forms of health care it is not age itself that is the biggest determinant of use, but how close the person is to death.

Overall, our analysis shows that demographic changes explain only a small part of the 114% growth in the total amount of care the NHS has delivered over the past few decades.

In planned hospital care this is particularly striking, with over 90% of the growth unexplained by demographic factors. This is not unique to England. Research has found that across the OECD ageing accounted for just above one-tenth of the increase in government health expenditure per capita between 1995 and 2009. For the UK, ageing explained an even smaller proportion of the growth in expenditure. Real health spending per person grew by 4.6% a year over this 15-year period, with ageing accounting for growth of just 0.2% a year.

However, there are several services that do not follow this pattern. Population growth and ageing explain a relatively high proportion of activity growth in a few services (A&E attendances, consultations in general practice and community health services). This is not because these factors have a larger absolute impact in these areas, but because growth as a result of the other factors in our framework is much lower.

In terms of proximity to death, the number of people in the last two years of life is expected to rise over the next 20 years. This will impact on overall NHS resources and the mix of services. Planning for reductions in emergency admissions against this background is unrealistic, particularly given the growth in admissions seen during a more benign period. If the NHS systematically underestimates the amount of emergency care it needs, this has a potential impact on the quality of emergency care but also knock-on consequences for planned care. This is because operations will have to be delayed or cancelled if capacity is constrained and beds and operating theatres are full.

What explains the growth?

The fact that demographic factors alone do not explain the growth has consistently caused us to ask: what explains the growth? Within each area of care, we have explored this using our framework, outlining major policy changes and supply and demand side factors that may explain the remaining growth.

Policy and political choices

Many of the drivers in our framework, particularly those driving the supply of health care, result directly from political and policy choices. The level of funding is a major choice. It is linked to decisions about pay for NHS staff, which determines how much of any funding increase is available for increases in activity.

The way in which providers are paid is an example of how policy affects supply, highlighted by the OECD in a study of the role of institutions and policies as a driver of health care spend. The NHS operates many different provider payment schemes. Broadly, GPs are paid on a capitation basis, with payment dependent on how many patients they have registered, rather than the amount of activity they do. However, many hospital services – such as elective care – are paid for on the basis of the amount of activity that is carried out. This creates a mechanism to fund increasing activity, eg more procedures, and may be part of the explanation for the rapid growth in hospital planned care.

Government sets priorities for the NHS. These shifted over the two decades and also play a part in explaining some of the trends. In the early 2000s there was a big focus on reducing waiting times for planned care. This was to be achieved through additional NHS funding, the target of a maximum 18-week wait from referral-to-treatment and the introduction of payment systems that helped fund increased elective activity. Overall, this was successful: elective procedures increased by 9.6% a year through the period. By contrast, although a move from hospital to primary and community care was advocated, it was accompanied neither by increased funding or waiting time targets, likely contributing to the much smaller growth in consultations in general practice and community care.

In the second decade, marked by slow growth in funding, policymakers’ emphasis was on efficiency and productivity improvements. Significant cost savings/efficiencies were achieved through a public sector pay freeze which held down the wages of NHS staff. Combined with productivity improvements, this ensured that, although growth in activity was slower than in the first decade, it did not slow as much as the growth in funding. Another strand of the policy drive to improve efficiency were efforts to reduce growth in ‘unnecessary activity’, such as emergency admissions to hospital for conditions that could be treated elsewhere. Despite various initiatives to reduce emergency admissions, such as attempts to make care more integrated through new models of care, these saw consistent growth across the period.

Supply

Evidence suggests that supply side factors and policy may be important determinants of changes in health care expenditure. Supply side drivers are more amenable to influence by policy – particularly over the short term – than demand sides drivers, particularly population size and age structure. The OECD has looked at the impact of supply-side policies and institutions on public health spending. Supply-side decisions can affect the unit costs of health care, while others directly affect the amount of health care activity.

A major factor identified is the overall funding for the health system and the stringency of any budget cap. In England, the NHS budget is predetermined and fixed. This determines the level of inputs (staff and equipment) to the NHS, which in turn influences the growth in activity. Beyond simply the level of inputs, decisions on different aspects of workforce and hospital supply set capacity and can thereby suppress or stimulate demand. We see many examples of this in our analysis.

Workforce may explain why there has been growth in some areas and not others. The number of specialty doctors in the NHS grew by an average of 8.2% a year over this period, well above the 2.0% per year seen for the overall workforce. This meant that there was more capacity for doctors to refer patients and to undertake outpatient procedures as technology advanced. Conversely, the plateauing number of GPs since 2010 may explain the slow growth in consultations, below the levels expected from population growth and ageing.

Additional capacity (hospital beds, theatres and diagnostic equipment) can inadvertently increase activity, even if this is not intended. The Any Qualified Provider policy increased capacity by allowing private providers to bid for contracts. The aim was to use competitive pressure to improve quality by allowing consumers to choose between providers. However, a study showed that private sector hospitals created demand: their entry into an area increased the number of publicly funded hip replacements by 12% but did not reduce volumes at incumbent public hospitals, and had no impact on readmission rates.

The reduction in hospital beds – a fall of 31% since 2000 – has also had an impact on the composition of activity. Pressure to reduce beds has led to (or been facilitated by) a reduction in length of stay and a shift away from inpatient admissions to day cases which, as previously described, are much cheaper. These reductions can have mixed consequences: improving efficiency in some areas but contributing to an increase in bed occupancy levels – from 87.1% in 2010 to 90.7% in 2019. This has meant hospitals have struggled to find beds for patients, putting pressure on A&E waiting times.

The final driver we have identified is technological progress. Progress and technological change lie behind much of the growth in outpatient procedures, which increased fivefold over the period. The number of MRI scans – and diagnostic tests in general – has also increased as the price of scanners has reduced. The increase in testing may have led to greater identification of patients with conditions for which treatment is beneficial, leading to additional demand in other parts of the health system.

Research and development spend (public and non-public) drive where technological progress takes place, as do the areas of specialisation within the medical profession where new research is taking place. The priorities for this spend may not match the needs of the population.

Although technological progress itself is largely outside the influence of NHS policymakers, it is shaped by decisions about investment and research. How technology is adopted within the NHS can be influenced. NICE, established in 1999, has a clear role in this area, publishing guidance on the use of new health technologies, such as new and existing medicines, treatments and procedures. However, much innovation, such as in surgery, lies outside the scope of NICE appraisals. Patient Reported Outcome Measures (PROMs) are another way of assessing the contribution of some high-volume hospital procedures, such as hip replacements, but only limited amounts of this information are routinely available to service planners or commissioners, and it is unclear how much is used in planning.

Demand

Over the past two decades a greater proportion of people at a given age are living with long-term health conditions, and healthy life expectancy has not kept pace with improvements in overall life expectancy. This is widening existing inequalities: there is an 18-year gap in healthy life expectancy between the least and most socioeconomically deprived populations. As a result, more people are living more of their lives with long-term health problems. This could be one factor contributing to the growth in elective procedures which, as described, have risen by considerably more than explained by population growth and ageing.

Changes in the health of the population and prevalence of conditions will affect different types of activity in different ways. As described in our analysis, over the past two decades planned care grew at much faster rates than primary and community care. This may reflect the way changes in patterns of ill health combine with changes in technology. Recent projections for England suggest that between 2015 and 2035 multimorbidity will increase, with more of the gains in life expectancy being spent in poor health with multiple long-term conditions. Understanding the potential impact this may have on overall demand for NHS services and the way this will vary for different types of care is clearly very important.

The diseases that contribute most to health care activity are changing. For the leading causes of morbidity most people will rely on the care of their GP, supported by outpatient specialists, diagnostic investigations and procedures, and some planned surgery. If these conditions are well managed, they should not result in emergency hospital admissions. New technologies are available to ensure conditions can be better diagnosed and managed (for example, advances in imaging for heart failure). However, the relatively meagre growth in primary and community care may well be contributing to the consistently high growth in emergency admissions. Investment in public health and prevention could go further and improve the underlying health of the population.

Over time, what the public expects from the NHS also changes. There is significantly more information available to patients, contributing to a growing awareness and perceptions of what health care and treatments are available.

More people with a condition may approach health services to seek treatment. An example here is the growth seen in certain areas of mental health care. Over the past two decades, service use has also increased, with an estimated one in three people with a common mental health problem using mental health services in 2014, compared to one in four in 2000. This growth may reflect genuine change in the disease burden. However, it may also reflect an increased public willingness to disclose a mental health problem and seek treatment, triggered in part by decreased stigma around mental health.

Summary

Over the past two decades the quantity of what the NHS does – the amount of activity – has doubled and this increase has outpaced the growth in demand that arises from demographic factors. Overall, our analysis suggests that the amount of care provided by the NHS since 2000/01 would have needed to grow by around 1% a year to meet the demand pressures arising from these demographic changes. This means that more than three-quarters of the growth in NHS care since 2000/01 is the result of other factors relating to demand, supply, political priorities and health care policies.

The growth in excess of that explained by demographic change varies across different services, showing that the drivers affect different services in different ways. For example, changes in morbidity affect demand differently – a reduction in the prevalence of injuries amongst younger people reduces the demand for A&E, but not to the same extent demand for outpatient services. Similarly, some services are more affected by technological progress – eg prescriptions of drugs (new drugs become available and the price of existing drugs reduces). Variation between different time periods is often the result of prevailing policy priorities (eg the push to reduce waiting times in the first decade).

Looking at the overall changes in the volume and pattern of care provided over the past two decades, it is clear that some of the changes have been despite – rather than because of – the stated policy goals of successive governments and system leaders. Indeed, we have seen things happen that fundamentally counter the stated intentions.

This leads us to suggest six lessons for the future.

1. In the long term, what the health care system does is largely determined by choices. Clear population health goals are needed to help steer these choices.

The contribution of demographic factors to growth is relatively small. Other drivers of change are not well understood, but are within the influence of policymakers if we consider the full range of policy instruments available, especially over a longer time horizon.

We have seen significant shifts in the composition of services and in what the NHS does. But it is not clear whether the changes observed have maximised the health impact of the NHS in England. Decisions have increasingly focused on processes and the organisation of the system, while leaving the overall goals of the system, such as reducing inequalities and increasing life expectancy, less clear. We need clearly stated population health goals against which to judge the decisions taken.

2. A better understanding of the contribution of different services to population health, and what this means for the future composition of NHS services, is required.

The shifts in activity we have observed have not necessarily been in line with those needed to meet the changing patterns of the population, or those that would make the biggest contribution to population health. For example, the growth in the prevalence of multiple long-term conditions suggests that there should have been a relative shift in activity towards prevention and primary care services; this has been the policy rhetoric too.

If the NHS is to maximise its contribution to population health, there is a need to better understand what composition of services the NHS should have going forward. This means developing a better understanding of the changes in patterns of morbidity.

It also means understanding the contribution of different services to improving health. Over the period covered here we have seen a rapid growth in hospital services. This growth started with the aim of aligning the NHS with similar countries internationally, in terms of access to hospital services and care outcomes. However, this has led to hospital activity crowding out other forms of activity. We therefore need a better understanding of the value of additional hospital activity compared to the value of other services in addressing the challenges of the future. These include contributions to improving healthy life expectancy and reducing inequalities.

3. A systematic national projection of future trends in morbidity is needed to inform national and local NHS planning.

Research on the sustainability of health care systems consistently shows that the future path of morbidity is a key determinant of future health care demand and funding requirements.

Previous projections by the Health Foundation and Institute for Fiscal Studies (IFS) showed that for 18 conditions the probability of admission standardised for age had increased by 57% between 2007/08 and 2017/18, with the greatest increase for people with multiple conditions – a growing group in the population. The analysis found that if these trends continued for the next decade, long-term conditions would have as great an impact on hospital activity as ageing.

Currently, there is no systematic national projection of future trends in morbidity undertaken to inform national or local NHS planning. This is a clear gap. If changes to the composition of NHS services are to meet the future health needs of the population, we need good projections of those needs. As the types of services needed by different groups of patients differ – from diabetes to cancer – projections need to cover different forms of morbidity, and, increasingly, multi-morbidity.

4. There is a need for more sophisticated models of future activity and health outcomes, incorporating other important drivers of growth as well as demographic factors. 

We have discussed a number of other drivers of growth within the broad categories of demand, supply and policy. Factors as diverse as workforce planning, morbidity, patient expectations and drugs and technology can all have an impact.

Currently, the exact role these factors play in activity growth is not well understood. Consequently, these are rarely included in future projections of activity. There is a need for more sophisticated models to project future activity that include other drivers of changes, as well as the underlying demographic drivers.

The REAL Centre will create models incorporating these factors, with an initial focus on modelling the workforce (starting with nursing supply and the social care workforce), and the impact of changes in risk factors on morbidity and subsequent health care use.

There is growing analytical capability within the NHS, but further investment is needed to ensure analysts have the skills they need (eg capability in systems dynamics modelling or marginal analysis). This also includes the space and cultural environment where this type of work and research is valued and supported.

5. Technology is one of the most powerful factors shaping both demand (a new treatment makes it possible to address an unmet need) and supply (drug therapies replacing complex surgery).

The UK is a world leader in research and the NHS makes a vital contribution supporting clinical trials. Innovation is inherently unpredictable and complex, but that makes it more important to try to understand the forces shaping decision making.

New technology has been a key driver of activity, but it may not always have provided value for money. How technology is funded and adopted is very much under the influence of policymakers, though this influence could be extended and strengthened. NICE, for example, appraises new technologies and recommends whether they should be funded by the NHS.

6. Shifts in activity take time and there is a need to be realistic about what can be achieved.

Many of these shifts have evolved gradually over time. Some trends have been the result of decisions taken decades earlier. Once allocations in spend are set, it is difficult to make major changes during an annual budget review. Changes are easier at the margins (when allocating additional funding); divesting from services is much more challenging.

Likewise, it takes time for individuals working at the NHS front line to respond to policy frameworks and incentives. Some initiatives, such as targets, may be politically difficult to remove. Although decisions may make sense when seen locally and at a particular point in time, from a global perspective and in future years they may not be seen as optimal.

A better understanding of the factors we have discussed combined, with better approaches to dealing with uncertainty and the dynamics of the system, will help policymakers recognise the limits of what can be changed, to what extent and how quickly.

In summary, what is needed is a strategy for investing in the NHS based on evidence, analysis and modelling. This needs to able to be adapted regionally to allow for differences in population health and health care system.

Conclusion

The past two decades have seen significant shifts in what the NHS does. While activity overall has doubled, growth has not been spread equally between services.

In hospital care, particularly in planned care, we have seen growth far beyond what we would expect due to demographic changes alone. In some ways this demonstrates the impact policymakers can have. An issue was identified: long waiting times for a number of planned procedures, linked to adverse health outcomes. Funding was increased, specialised staff recruited, and high-profile, politically important targets set, supported by technological advances. In many ways this is a policy success – activity grew, waiting times for routine treatment went down.

However, this focus on hospital care has had consequences. Although there has also been an ambition to move care into the community, primary and community care services have not seen the same growth and are struggling to meet the demands of a population with an increasing number of long-term, complex conditions.

This illustrates the importance of considering and, where possible, modelling the consequences of different policy choices – and in taking a whole system view. In this report we have outlined a framework for examining the potential factors that influence activity. Sophisticated models take time to build, but considering these factors when new policies are introduced could illuminate potential consequences.

Our work suggests that while understanding demographic trends is important for planning health services, it is a relatively small driver of the growth in care over recent years – explaining less than a quarter of the 4.6% a year average growth in NHS activity. The NHS therefore needs a much better understanding of the demand and supply side factors that collectively contribute to the volume and pattern of care. To do this we need to fill the evidence gaps with more and better research. We need to understand the potential impact of trends in demand and supply, and how policy can influence and shape them. Last but not least, the NHS needs frameworks and approaches to using this evidence, which help it to make better decisions given the inherent uncertainty and complexity of the factors shaping the health system.

The NHS faces a number of important decisions as it emerges from the COVID-19 crisis, posing an opportunity to align future investment more closely with the changing health needs of the population. The REAL Centre will work to provide evidence and analysis to inform these decisions and build on our understanding of the factors influencing growth.

We have relied on a small number of key data sources when analysing health care output and activity. For total NHS output, we used the public service productivity estimates for health care in England produced by the ONS. NHS reference costs data were used to understand the average unit cost and activity delivered by NHS Trust and Foundation Trusts by service.

Methodology

We model demographic factors that are readily available and relatively easy for planners to project into the future – and therefore useful in planning. These include: size of the population, age, gender and proximity to death. For proximity to death, we use successive years of HES or CPRD data supplemented with information regarding an individual’s death from linked ONS mortality data.

We use a nested modelling approach to estimate the relative scale of the effect of changes in population size, the age and gender makeup of the population, and proximity to death (as a proxy for levels of morbidity) on activity.

We start with an initial model – with time as the sole explanatory variable – that estimates the absolute increase in attendances. This model takes no account of changes in population size, the age and gender makeup of the population or levels of morbidity. We then expand the model, adding one factor at a time until all potential explanatory factors are included. At each stage, we use the model to generate a prediction for the overall number of units of activity (attendances etc).

In effect, we create a counterfactual estimate of attendance numbers by applying the modelled relationships between our explanatory factors and attendances to the population in the final year. If the number of attendances predicted by the final model (including all potential explanatory factors) were only marginally different to the number of attendances in the final year, then we might conclude that these factors are sufficient to explain the observed trend.

It is possible that some of the explanatory factors listed above will be collinear. As such, the order in which the factors are introduced into the model will influence their reported effect on attendance numbers. The change in predicted attendances when a factor is introduced represents the additional change that this variable produces having taken account of all other variables in higher nest levels. For example, at nest level 3, the model adjusts for age, population size and the fixed effect of time.

Time period

As far as possible, we have sought to model the 20 years from 2000/01. In many instances, the latest available data are 2017/18. Where data were unavailable for the whole time series – for instance, HES data are incomplete for A&E attendances prior to 2010/11 – we project the contribution of model parameters from available data over the longest available time span. See the further notes in data definitions.

Results

The full set of results from our model can be found in the table below. This breaks out demographics into its composite factors: population size, age, gender and proximity to death.

Data definitions and sources

When modelling changes in activity, we rely on a range of sources, most often hospital episode statistics (HES). Note, this contains data for all NHS funded care, including that delivered by the non-NHS providers, which is missing in NHS reference costs. Importantly, this is a patient level dataset, meaning we can understand the profile of those using services, which informs our model. A full list of sources and definitions used in the modelling for chapter 4 are provided in the table that follows.