Improving community health care logistics using Operational Research
Principal Investigator: Dr Carlos Lamas-Fernandez
Team members: Dr Carlos Lamas-Fernandez, (Research Fellow in Operational Research, University of Southampton), Professor Peter Griffiths (Chair of Health Services Research. University of Southampton), Dr Antonio Martinez-Sykora (Associate Prof of Business Analytics. Southampton Business School, University of Southampton), Dr Tom Monks (Associate Professor of Health Data Science, University of Exeter)
Start: 1 October 2019 Ends: 30 September 2021
Partners: University of Southampton, Solent NHS Trust
Operational Research (OR) is the application of computer and mathematical modelling to support decision making. In health services research, OR aims to improve patient outcomes, increase efficiency and enhance health professionals and citizens understanding of how an NHS service achieves good performance. In this study, we will use OR to improve the quality of patient care by supporting community nursing teams organise how they visit people in their own home.
Community nursing teams in England are part of an NHS Community Trust. Everyday each Trust sends nurses out to visit hundreds of citizens in their own home to provide care to meet their needs. The process of planning home visits is largely manual (paper and pencil) and planned a day in advance. Nurses of different bands and even within the same band have different skills, for example some nurses are able to deliver wound care and some nurses are not. Home visits are allocated to nurses in a way that makes sure that patients care needs are met (for example, an insulin injection must be administered between 9am and 11am ), that the cost of delivering care is minimised (for example, by clustering patients by geographic location) and that nurses and patient preferences are met (for example, variation in care activity for nurses, and continuity of care for patients). Much of community nursing work is planned, however, it is common for teams to receive urgent referrals during a working day. This type of planning problem is well known to OR: it is very difficult for manual planning of home visits to achieve all its aims. To make this problem simpler in practice large regions are broken down into smaller regions and planning takes place only for the next day. Such an approach works but is likely to be missing benefits for patients and efficiency savings for the NHS offered by looking at the problem as a whole.
To address this gap, in recent years, academic OR has developed algorithms to automate the solution of what it has called the ‘Home Health Care Routing and Scheduling Problem’. In 2018, Dr Carlos Lamas-Fernandez and Dr Thomas Monks from NIHR CLAHRC Wessex’s Data Science Hub worked with Solent NHS Trust to learn from both academic research and the practice of community nursing to develop a novel decision support tool (DST) to support community teams to improve patient care and efficiency. This study will scale up this work in the following ways:
- The decision support tool will be tested with nursing teams in Portsmouth, Southampton and Dorset
- Evaluate the feasibility to compare a sample of historical patient home visit schedules to those generated by the DST.
- Evaluate the acceptability of schedules generated by the tool to a group of clinical practitioners.
- Identify the adaptations needed to model urgent referrals
- To identify, by modelling, the benefits for NHS trust to centralise their planning across larger regions and to plan further ahead than a single day.
We will achieve these aims by:
- Collecting primary data, to quantify the benefits (and/or disadvantages) of an OR approach over manual planning
- Making use of a Turing Test* framework to explore nurse planners’ views on automated home visit plans
- Using OR mathematical and computational approaches to adapt the DST for new setting
- Developing novel approaches for planners to handle urgent (unplanned) daily referrals
* The Turing Test - a test set by the computer pioneer Alan Turing which challenged programmers to design a computer program that could fool a human being into thinking they were interacting with another human.