By Daniel Gartner, Professor of Operational Research at Cardiff University and Aalen University, School of Management, Germany
As demand for healthcare continues to rise, the NHS is under increasing pressure to deliver better outcomes while managing limited resources. Across Wales, operational research (OR) is helping meet this challenge by using data, mathematical modelling and analytical techniques to support more informed decision-making.
Working closely with Welsh NHS Health Boards, researchers at Cardiff University are applying these techniques to improve patient flow, make better use of capacity and support more effective operational decisions[i]. Two recent projects involving Cardiff University and Aneurin Bevan University Health Board, supported by contributors including Dr Izabela Spernaes, Joanne Buchanan, Steve Bonser and Dr Jonathan Clarke, demonstrate what this looks like in practice.
Matching ambulances to real demand
Inter-site patient transfers are a vital part of healthcare, moving patients between hospitals for specialist treatment, ongoing care and rehabilitation. Within Aneurin Bevan University Health Board, this system was under strain. The opening of the Grange University Hospital had increased both the volume and complexity of transfers, but ambulance schedules had not adapted.
Analysis revealed a mismatch between when ambulances were available and when they were needed. Most crews started work at 7am despite demand peaking later in the day. Vehicles were often underused in the morning but heavily stretched in the afternoon, contributing to average waits of 122 minutes from request to patient pick-up.
The problem was compounded by "mission creep", with around a third of ambulances being used for work outside their primary transfer role, including routine patient discharges. Using operational research, a multidisciplinary team analysed around 15,000 transfer records to map demand patterns. They found capacity was too high overnight and mid-morning but insufficient during peak periods.
Rather than adding resources, the team used modelling and optimisation to redesign shift patterns and better align capacity with demand. A staged approach introduced model-informed shift timings before applying advanced optimisation to safely reduce resources.
The impact was significant. Average delays fell from 134 minutes to 110 minutes and then to around 87 minutes, while performance against targets improved from 68 per cent to 83 per cent, exceeding the 80 per cent goal.
These gains were achieved while reducing peak ambulance numbers from 10 to seven, delivering annual savings of around £1.1 million without compromising care. Staff reported more predictable workflows, while patients benefited from faster access to treatment.
Tackling missed appointments using data
The Cardiff University team also examined missed outpatient appointments, known as “Did Not Attends” (DNAs), in the head and neck cancer service at Aneurin Bevan University Health Board, where around 13 per cent of appointments were missed. Missed appointments waste valuable clinical time, reduce capacity and can delay diagnosis or treatment. The team asked a simple question: could data help predict which patients were most likely to miss appointments?
Researchers analysed more than 13,000 appointments and examined over 60 factors, including appointment timing, demographics and previous attendance history. Using operational research techniques, they developed a predictive model to estimate the likelihood of non-attendance.
A logistic regression model was selected because it was transparent, explainable and straightforward to implement within existing NHS systems. The analysis found that previous non-attendance was one of the strongest predictors of future missed appointments, while socioeconomic background and appointment timing also influenced attendance.
The real innovation though was how the model was used. Embedded into the booking system, it ranked patients by risk and helped staff focus interventions where they could have the greatest impact.
For high-risk patients, a simple intervention was introduced: a phone call. Staff checked whether patients had received appointment details, needed support with transport or whether the appointment time remained suitable. These conversations often uncovered practical barriers that could be resolved easily.
The model identified around 2,400 high-risk appointments suitable for intervention, with approximately one in four successfully "saved" through improved attendance or by reallocating the appointment slot. With each intervention costing less than £3 and each avoided no-show saving around £120, the model delivered a return of up to £12.57 for every £1 invested.
Turning insight into improvement
These projects show how operational research can help the NHS solve real-world problems using evidence rather than assumptions. Whether reducing ambulance delays or tackling missed appointments, the starting point is the same, understanding demand more clearly and making better use of available resources.
As NHS teams face growing pressure on services and budgets, the ability to turn data into practical action will become increasingly valuable. The experience in Wales demonstrates that relatively small, targeted changes can improve efficiency, free up capacity and enhance patient care without necessarily requiring additional resources.
Perhaps most importantly, these projects show the value of collaboration. The best results came not from data alone, but from bringing together analysts, clinicians and operational teams to understand the problem and identify workable solutions. When analytical insight is combined with frontline experience, services can be redesigned in ways that benefit both patients and staff.
