Why is it so difficult to estimate expenditure on health and care at the end of life?

Today our new analysis of public expenditure on people in the final year of life is published. The research was undertaken by the Nuffield Trust and the Health Economics Unit, supported by the end of life care charity Marie Curie.

Drawing on a wide range of data sources, we estimate that at least £22 billion of public funds was spent on people in the last year of life in the UK in 2022. This amounts to almost £34,000 per person who died. Over half of this expenditure was on health care (almost £12 billion), 22% on social care (almost £5 billion) and 25% on social security (£5.5 billion). 

Of the money spent on health care, 81% of public funds spent on health care for people in their last year of life was spent in hospital, and 56% was spent on emergency hospital care. Put another way, £4 in every £5 was spent in hospital, including £2.80 on emergency hospital care. Strikingly just 11% of health expenditure for people in the last year of life is spent on primary and community care.

This report is the latest from a number of projects we have undertaken about end of life care. In the last three years we have reported on care in hospices, working with Hospice UK; studied the impact of the Covid-19 pandemic on care for people who died at home; and tracked trends in use of services and inequalities in care.

And 10 years ago we estimated costs of health and social care at the end of life (focusing on a smaller number of services, and covering a smaller population). We highlighted at the time that “acquiring good-quality data surrounding costs for palliative care is challenging, especially as care services for those at the end of life are fragmented across many different providers”.

In that report, we noted that “given the finite resources available for health care, it is important that there is an understanding of the costs of end-of-life care”. This remains just as true today, as the NHS grapples with significant cost pressures, and as much-needed reform of social care is some way off.

Data challenges have not gone away

So why is this the first major update to the cost of end of life care since 2014, and what were the gaps in data that have made it challenging?

Three factors stand out.

There remains a dearth of routine data published about critical areas of care for people at the end of life, in particular community health services and social care. There is a community health services dataset, but data collection is incomplete, and the level of detail within it is insufficient to reliably identify people receiving end of life care. For social care, we had to base our estimate on data from our study from 10 years earlier (adjusted for change over time). Although a new client-level dataset for adult social care is being rolled out, data from it is not routinely published.

Secondly, the gold standard for identifying service use for people at the end of life is to link service use data with information on death registration at the individual level: our original analysis and our recent work was able to use linked data. But such data is not routinely available. In the report published today, we drew on such linked studies where possible, but we often needed to make estimates of activity using routine data not specific to people at the end of life, or estimates of service use from a survey with individuals’ families.

Thirdly, information on cost of care for people at the end of life was often not available. For example, we used standard NHS cost data but this was not specific to people at the end of life. While we adjusted for longer stays in hospital, we could not fully account for differences in complexity of care.

Our analysis this time was wider than the study 10 years ago – encompassing social security expenditure as well as health and social care – and also undertaken for all four countries of the UK.

Learning from across the UK

Identifying data sources for each UK country was a particular challenge. Most of the research data we found applied to England, although we also made use of some information from Scotland. Data from the Welsh end of life care dashboard was extremely valuable. It also allowed us to test the validity of our main UK results, by assessing whether there was variation between countries.

Information from each UK country could also provide insights to enable each country to learn from others. For example, Wales appears to have lower average numbers of A&E attendances and emergency admissions than England and Scotland – understanding the reasons for this could inform opportunities to improve care across the UK.

We hope that our new analysis will inform discussions and decisions on funding and improving services for people at the end of life, in each country and by health and social care organisations at a local level. And in another 10 years’ time, we also hope that improvements in data collection and linkage will mean that much of this information is routinely available.