This is a staggeringly difficult question. By definition, the future is uncertain. And planning healthcare means making judgements about factors such as population size, age and healthiness—as well as how likely needs will differ for different services.

And yet answers are needed. Better answers will lead to better outcomes; worse answers will cause needless suffering.

Successfully navigating this challenge requires high-quality modelling. And so the Strategy Unit has produced this interactive tool.

Local population ‘shape’

The tool shows how aging and other demographic changes affect future need for different types of hospital services. Crucially, it is locally specific. The tool recognises the radically different demographic starting points of, for example, coastal towns and large cities.

By showing how populations are composed—and how they are expected to change—population pyramids bring local differences to life. Contrast the Christmas tree-shape of the Barking and Dagenham population with the rectangular-shape of Worcestershire.

The shape of each population pyramid is determined by trends in births, deaths, and migration. To help with decision-making, the Office for National Statistics (ONS) publish population projections. Because they are projections, the ONS also constructs alternative scenarios. For example, what if international migration declined? Or what if women had fewer children than in the past?

To arrive at a tailored projection, the tool allows you to choose from 10 variants and watch how your local population pyramid changes to 2043.

Age and healthcare use

As we age, we tend to use health services more. This is especially true for treatment of physical health conditions provided in hospitals.

The relationship between age and healthcare use is best shown by ‘age-related activity profiles’. These plot the average number of treatments (e.g. hospital stays) per head of population against age.

The age of the population affects demand for some services far more than others. So the tool allows you to see local activity profiles for eleven service types. These cover the three main settings for hospital care (A&E, inpatients, and outpatients).

How healthy might we be in the future?

Need is mainly determined by the size and age of the population. So it is reasonable to take current patterns of healthcare use, and combine them with expected changes in the population.

But age is not the cause of healthcare use. It is possible for a population to both age and become healthier. Future gains in life expectancy could be spent in good health. Following the popular press stories, 60 really could become the new 50.

Conversely, we might not get healthier. We might improve average lifespan while also increasing the prevalence of chronic diseases and disability.

Which scenario is most likely? The evidence is mixed. But the question remains, and answers are of great significance to those involved in planning care.

In the absence of research evidence, we can use expert opinion. And, in October 2022, a group of independent experts were asked for their view on the future health of the population in England.

Broadly, the consensus was that the share of remaining life spent free of limiting long-standing illness at age 65 years would fall slightly by 2035.

But there was considerable uncertainty around this central view. The group also saw more chance that the decline in health status might not be as bad as their central view—or that it might even improve—than that it might be worse than they expected.

The tool takes account of this, using the probability distribution from the group to adjust estimates of future healthcare activity.

Adding this together, the tool therefore allows you to see results for your local area across eleven different service types, covering the three main settings for hospital care, in five year steps up to 2040.

The charts also show the uncertainty as to how health status might evolve. The progression from purple (more probable) to yellow (less probable) stripes portrays the frequency of outcomes from our simulation.

Uncertainty associated with future trends in fertility, mortality and migration can also be explored by selecting one of 10 variant population projections.

Other work in this area

Other UK-based research teams are also working to understand what effect our aging population and changes in patterns of population ill health will have for future health care demand. Colleagues at NHS Bristol, North Somerset and South Gloucestershire ICB, working with the University of Bath, have developed a dynamic population model to forecast the long-term health needs and resource requirements in their local area. The Health Foundation's REAL Centre, in partnership with the University of Liverpool, has used a microsimulation model to look at projected patterns of illness in England to 2040. Please check out their work.

Contributors

Design and implementation

Paul Seamer, Strategy Unit, was responsible for the overall design and implementation, including conceptualisation, data curation, methods, analysis, visualisations, and drafting the explanatory text. Steven Wyatt, Strategy Unit, supervised the work. Tim Brock, Jumping Rivers, built the app, taking the code from proof of concept to a production ready web application. Fraser Battye, Strategy Unit, crafted the published version of the explanatory text.

User testing

We are grateful to the following people for reviewing and providing helpful feedback on a beta version of the app: Matthew Eves, Derbyshire Community Health Services NHS Foundation Trust, Ivan Dale, NHS Midlands and Lancashire, Tracey Genus, Health Innovation East Midlands, Helen Harvey, Shropshire Council, Gabriel Hobro, NHS England, Julie Pugh, Shropshire Council, Helena Robinson, Countess of Chester Hospital NHS Foundation Trust, Dr Chenyu Shang, NHS Northamptonshire ICB, Emma Smith, Shropshire Council, Emma Washbrook, NHS Midlands and Lancashire, Justin Wiltshire, Strategy Unit.

Attribution

The animated population pyramid was inspired by an interactive data visualisation published by the Federal Statistical Office of Germany. Many national statistical agencies publish animated pyramids, but the design and implentation by the Statistisches Bundesamt is by some distance the best I have come across. We adapted the original source code to fit our needs; this included a major refactoring of the code ...