Supportive Health Intervention and Escalation for Life Decisions
We want to improve unscheduled care in the last year of life. To do that we first want to understand how people use unscheduled care in this last year of life. We will then develop prediction models which identify future use by specific patient groups and develop a tailor-made coordination system that improves care pathways, patient experience and reduces the need for unscheduled care.
%
of population are in their last year of life¹
%
use unscheduled care in their last year of life²
%
of emergency hospital admissions account from people in their last year of life³
hospital bed days in the last year life (on average)⁴
%
of NHS budget for people in their last year of life⁵
OUR GOALS
To Improve Care in the Last Year of Life
Understand
How people use unscheduled care in their last year of life.
We will speak to patients, families, and healthcare teams to understand their needs and co-design solutions that improve patient experiences. We want to be able to address the diverse ways patients use healthcare and improve quality of life for them and their caregivers.
Develop
Prediction model to identify future care use by distinct patient groups.
We aim to ensure that patients in their last year of life can access compassionate, responsive, and dignified care. This care needs to address their needs and meet their preferences no matter where they live. Testing these solutions will help identify better care pathways that reduce distress and improve consistency in the last stages of life.
Identify
And refer patients to bespoke care pathways that improve patient care.
We want to reduce the demand for unscheduled care while improving patient experiences. By identifying patients early with risk prediction tools, we can match them to the right support from the most appropriate person at the right place at the right time.
OUR PLAN
Project Structure and Research Plan
Using a step-by-step approach, we will combine data analysis, patient insights, and collaboration with stakeholders to develop and test practical solutions. Our focus is on creating meaningful changes in care for some of the most vulnerable individuals, addressing gaps in the current system with innovative, evidence-based interventions to reduce reliance on unscheduled care. We will answer the following research questions step by step.
Identifying Populations & Their Unscheduled Care Use
How do people at the end of life currently use unscheduled care services, and are there differences in patterns of use by different patient groups?
To answer these questions, we will employ a mixed-methods study with an explanatory sequential design. We start with an initial quantitative analysis to identify and characterise patients in their last year of life by how they are using unscheduled care.
Risk Prediction Modelling
Using artificial intelligence, including machine learning models, can we identify who may be in their last year of life and predict their future use of unscheduled care?
To do that we will develop prediction models to identify those patients who may benefit from alternative care pathways. We will evaluate these new interventions, including the undertaking of economic costing.
Understanding Why Patients Use Unscheduled Care
Why do patients use unscheduled care in the way that they do, what do they think of the care provided, and would different ways of providing care be acceptable to them?
Qualitative work will elucidate these initial findings and explore how to place the prediction models within care pathways.
Co-Design Interventions with Stakeholders
Can we develop a bespoke, system-wide, care coordination intervention for people in their last year of life in Scotland, to improve their pathways, experiences and outcomes and reduce unscheduled care utilisation?
Co-design interventions with patients, caregivers, and healthcare professionals.
Feasibility Studies in Regional Test Beds
Can we implement and evaluate a complex care coordination intervention NHS Fife and Highland, exploring comparative service model differences and demographic variation to provide evidence of intervention feasibility in preparation for larger scale implementation across Scotland?
After implementation, we want to evaluate our intervention in using data from the Health Informatics Center (HIC) and NHS Highland. By that we want to provide evidence of its feasibility/suitability for a large-scale implementation across Scotland.
OUR DATA
Study data and population
We will utilise data held by the national electronic Data Research and Innovation Service (eDRIS) linking hospital admission records (General Acute Inpatient and Day Case and Mental Health Inpatient and Day Case), National Records of Scotland deaths data, the Prescribing Information Service (PIS) dataset, Scottish Cancer Registry, GP records and the Scottish Unscheduled Care Datamart (UCD). UCD includes contacts with the NHS24 telephone advice line, primary care out-of-hours services, the Scottish Ambulance Service and Accident and Emergency attendances. It also contains anticipatory care planning and palliative care information, including Emergency Care Summaries and Key Information Summaries.
We will request pseudo-anonymised data on all patients who died in the period 2017-2022 and ask for data going back to 2010 for all other datasets to ascertain medical history. We expect to see over 320,000 deaths, with around 30% of these likely to be cancer-related.
OUR DATA SECURITY
Anonymised Data and Safe Haven
All research is carried out within Trusted Research Environments (TREs), such as the Scottish National Safe Haven, using data provisioned by eDRIS. eDRIS follows robust information governance procedures that balance data protection with the need for research that delivers public benefit. All research proposals are reviewed by independent panels that assess the potential benefits and privacy risks. Researchers must complete information governance training and sign the eDRIS User Agreement, which outlines responsibilities and penalties for non-compliance.
All individual-level data are anonymised or pseudonymised, with personal identifiers stored separately from content data. Data linkages are created using privacy-preserving protocols, including probabilistic matching and clerical review where needed. Secure file transfer protocols are used for all data exchanges between providers and the Safe Haven.
Only approved researchers can access data, and only within secure environments. TREs provide secure access points or remote access where appropriate. All outputs, including tables and figures, are reviewed using Statistical Disclosure Control methods before release to prevent the identification of individuals. eDRIS uses state-of-the-art infrastructure to store, back up and protect data, ensuring a high level of data security throughout the research process.
OUR SOURCES
¹ Thomas, K., & Gray, S. M. (2018). Population-based, person-centred end-of-life care: Time for a rethink. British Journal of General Practice, 68(668), 116–117. https://doi.org/10.3399/bjgp18X694925
² Mason, B., Kerssens, J. J., Stoddart, A., Murray, S. A., Moine, S., Finucane, A. M., & Boyd, K. (2020). Unscheduled and out-of-hours care for people in their last year of life: A retrospective cohort analysis of national datasets. BMJ Open, 10(11), e041888. https://doi.org/10.1136/bmjopen-2020-041888
³ Public Health England. (2025, March 4). Patterns of care at the end of life: Factsheet 2023. Fingertips. https://fingertips.phe.org.uk/documents/peolc_patterns_of_care_factsheet_2023.html
⁴ Diernberger, K., Luta, X., Bowden, J., Fallon, M., Droney, J., Lemmon, E., Gray, E., Marti, J., & Hall, P. (2024). Healthcare use and costs in the last year of life: A national population data linkage study. BMJ Supportive & Palliative Care, 14(e1), e885–e892. https://doi.org/10.1136/bmjspcare-2020-002708
⁵ tb