Please note, some figures have been amended in the below article since first publication.
When John Williams joined the Health Service Modelling Associates (HSMA) programme – supported by PenARC – he had never written a line of code. Fifteen months later, John, a Stroke Performance Analyst at Maidstone and Tunbridge Wells NHS Trust, has developed a powerful simulation model that could revolutionise stroke care and save the Trust up to £1.2 million every year.
Using real-world hospital data and a technique called Discrete Event Simulation (DES), John’s model shows how expanding Same Day Emergency Care (SDEC) services and improving access to CT Perfusion (CTP) scans could transform outcomes for patients while significantly reducing costs.
We spoke to John about the problem he set out to solve, what he learned through the HSMA programme, and how his work is already shaping decision-making in stroke care.
Meeting Two Pressures at Once: More Patients, Less Funding
Stroke incidence in the UK is rising faster than the pace of population ageing. Between 2015 and 2035, the number of strokes is expected to increase by as much as 60% year on year. With this looming demand, NHS stroke teams are under immense pressure to treat more patients with fewer resources – a challenge John describes as “daunting”.
To help address this, John focused his HSMA project on building a simulation model to test service changes in a virtual environment. The aim: identify improvements that benefit both patients and finances before implementation in the real world.
Poorer outcomes for stroke patients are not only distressing on a human level, but also place a significant financial burden on the health service. Delays in treatment can lead to increased levels of disability, which in turn result in greater long-term care needs. Patients with more severe impairments often require more intensive and prolonged rehabilitation, additional community and social care support, and are more likely to remain in hospital longer – occupying stroke beds that are already in high demand. By improving timely access to care through models like John’s, the system can reduce these costs substantially while improving lives.
The Power of Discrete Event Simulation (DES)
Using skills developed through the HSMA programme, John created a Discrete Event Simulation model to replicate and analyse the Hyper Acute and Acute Stroke Pathways. This technique, commonly used in systems engineering, allows users to model how processes interact over time.
“DES modelling is an extremely powerful tool for modelling pathways and systems, in this case allowing the user to see the impact of changes to the stroke pathway on finances and patient care,” explained John.
The model drew on a variety of data sources, including the Sentinel Stroke National Audit Programme (SSNAP), locally collected data from Maidstone Hospital, and published stroke care research.
It focused on two key components of the stroke pathway and the potential impact these have on patient care and cost savings:
- Same Day Emergency Care (SDEC): A faster, outpatient-focused stroke pathway in operation at Maidstone and Tunbridge Wells Trust since 2020. All patients first go through the SDEC. Those needing more care are admitted to the stroke unit, while others are discharged the same day for clinic or community therapy support. This helps the system run more efficiently.
- CT Perfusion (CTP) scanning: A diagnostic tool that identifies more patients eligible for time-sensitive treatments like thrombolysis, which improves outcomes. These patients also often spend less time in hospital, helping to lower bed occupancy costs.
John’s model tested whether the cost savings from avoided admissions to the stroke ward could outweigh the operational cost of extending SDEC opening hours.
However, this, along with extended CTP scanning hours, require upfront investment, which can be a barrier to adoption.
By using DES, John was able to simulate both financial and clinical outcomes, offering compelling evidence to support business cases and service redesign.
Key Findings: Faster Care, Better Outcomes, Lower Costs
John’s model produced some striking results:
- Reduced Bed Occupancy: With full utilisation of both SDEC and CTP, the average stroke unit bed occupancy dropped from 49 patients (a full ward) to under 30 patients.
- Capacity Planning: If stroke admissions rise by 50%, the hospital would need over 80 beds, an expansion that would be financially and spatially unviable. Instead, 24/7 SDEC operation could manage this increase more efficiently and at significantly lower cost.
- Savings Over Cost: Even after factoring in setup costs, the model showed that, regarding extending SDEC hours, the savings are two times greater than the running costs. CTP savings are even greater.
“This model has illustrated that despite the upfront cost of both CTP and the stroke SDEC, the savings that are exhibited as a direct result from their use far exceeds the cost,” said John.
Driving Change: Real-World Impact at Maidstone and Beyond
John’s model is already shaping service decisions:
- CTP Scanner: There is now a significant push to upgrade the CPT scanner for 24/7 use, with maintenance and repairs prioritised.
- SDEC Expansion: The model supports a business case to extend SDEC opening hours, potentially to round-the-clock coverage.
- Therapy Team Engagement: The stroke therapy team is exploring how fully staffing the SDEC could improve patient flow and outcomes, guided by insights from the model.
- Trust-Wide Awareness: Plans are underway to present the model to senior Trust leaders, showcasing its potential as a template for simulation modelling in other areas of the hospital.
Improving Care for Thousands
While still early in implementation, there are signs that John’s model is already improving care. National audit data is showing an increase in thrombolysis rates thanks to early changes in CTP scanning.
Maidstone’s stroke team assesses over 3,000 suspected stroke patients a year – and about half are admitted. Enhancing scanning and SDEC pathways could improve care for every one of them.
“Be that directly through receiving the most advanced scanning or being on the fastest SDEC pathway, or indirectly in the form of better staff to patient ratios and increased investment to the unit through the increasing savings,” said John.
Although the current model is tailored to stroke services, the approach is highly transferable. John’s work has sparked interest from stroke teams across the UK, many of whom attended his presentation at the HSMA 6 Showcase.
“John’s completion of the HSMA course…has been pivotal in shaping innovation and approaches to data analytics,” said Victoria Williams, Lead Nurse for Stroke at the Trust. “His standout achievement, the development of a simulation model, has already sparked significant interest within the Trust. I look forward to seeing him present and promote this work further internally. It marks a strong step forward in our use of data-driven insight.”
From Excel to Python: A Personal Transformation
“Before the (HSMA) programme I had no coding experience at all. Now… I have managed to create a discrete event simulation model in Python and have a working understanding of many coding and data science principals” explained John.
John came to HSMA with experience in Excel, but saw the potential of Python and simulation after attending an HSMA open day. Since then, the shift in his skills and perspective has been transformational, for him and his team.
“Comparing working in Excel, as I was before, to using Python and the skills I learned on HSMA is the same as comparing working with your hands bound to them being completely free. It has been a real step change in how our team at Maidstone can work with data.”
He credits course trainers Dan and Sammi for making complex topics accessible and engaging, and describes the HSMA community as “invaluable”.
Explore More
- 🔗 GitHub Repository: https://github.com/jfwilliams4/des_stroke_project
- 🌐 Web App: Coming soon
- 💡 Learn about the HSMA programme.
Authors
Dr Daniel Chalk
Senior Research Fellow in Applied Healthcare Modelling and Analysis