Digital modelling should be at the heart of fixing hospital discharge

The technology to address delayed discharge must be used in a coordinated, system-wide way, write Christos Vasilakis, founding director of the Centre for Healthcare Innovation and Improvement, and Dr Zehra Onen Dumlu, research assistant at the University of Bath

Thousands of NHS hospital beds are occupied by patients who no longer need acute medical care. Yet delays in arranging community or social care mean these patients remain in hospital, contributing to delayed discharge, often referred to as ‘bed blocking’.

This is not simply an operational frustration. It is a visible symptom of sustained pressure across the NHS.

In England, around one in eight general and acute hospital beds are occupied by patients who are medically fit for discharge. For patients, many of whom are older and frail, prolonged hospital stays can lead to deconditioning, loss of independence, and poorer outcomes.

However, delayed discharge is not primarily a hospital issue. It is a whole-system challenge involving acute care, community services, and social care.

System problems require system solutions. The technology to address delayed discharge already exists: analytical tools, data infrastructure, and digital modelling capabilities are more accessible than ever.

What matters now is whether we are prepared to use them in a coordinated, system-wide way

Decisions made in silos

Policy frameworks such as England’s Discharge to Assess (D2A) model were designed to address this challenge.

D2A aims to move patients out of hospital as soon as they are clinically ready and assess longer-term needs in the community.

It operates through three pathways for discharge home with domiciliary care support; short-term bed-based rehabilitation; and complex assessment, often leading to long-term care placement.

Conceptually, the model is straightforward, but operationally it is highly sensitive to capacity constraints. Delays in domiciliary care, shortages of rehabilitation beds, or limited assessment capacity can quickly create bottlenecks.

Without a way to understand how changes in one part of the system affect the whole, we risk constantly firefighting rather than solving the problem

Unfortunately, when one part of the pathway slows, pressure cascades back into acute hospitals. Yet decisions about improving discharge pathways are often made in silos.

Hospitals optimise their internal processes, community teams manage their own pressures, and social care providers face separate workforce and funding constraints.

Without a way to understand how changes in one part of the system affect the whole system, we risk constantly firefighting rather than solving the underlying problem.

Collaboration the starting point

An example of where collaboration can help solve the problem is the ‘Improving patient flow between acute, community, and social care’ (IPACS) project.

The project was funded by Health Data Research UK and brought together researchers and healthcare professionals from the University of Bath, the University of Exeter Medical School, and the Bristol, North Somerset, and South Gloucestershire (BNSSG) Integrated Care Board.

This multidisciplinary partnership approach was crucial to its success. Frontline insight into discharge operations was combined with expertise in operational research and clinical oversight.

The aim was not to produce an abstract academic model but develop a practical tool that is grounded in the realities of the NHS.

IPACS developed a simulation model capturing how patients flow between acute hospitals, community services, and social care.

Six months of patient-level data was analysed to create a baseline representation of the system. The team then explored alternative ‘what if’ scenarios by varying arrival rates, lengths of stay, and pathway capacity.

Built using open-source R software, the model aligns with digital health principles of transparency, reproducibility, and scalability.

Importantly, it is freely available and can be accessed via Github. It’s a tool that any integrated care system can adapt and use.

Real-world testbed

The model was tested in the Bristol, North Somerset and South Gloucestershire (BNSSG) region, which serves around one million people. With its mix of urban and rural communities, this region reflects many of the national challenges the health service faces.

At the time of testing in October 2022, D2A services in the region were under intense pressure, with near-full occupancy and significant discharge delays.

Simulation modelling enabled the team to represent patient movement through D2A pathways over time, test the impact of potential interventions before implementation, identify bottlenecks across services, and account for fluctuating demand and capacity.

The findings were persuasive: better alignment with target pathway splits and modest reductions in length of stay could significantly improve patient flow and reduce delayed discharges.

Model outputs supported a £13 million business case for local D2A system development in BNSSG. This demonstrates that digital modelling is not merely analytical, it can directly inform strategic investment decisions.

Proactive design

No model is perfect. IPACS does not yet incorporate every element of discharge, such as detailed social care inputs, specialist palliative pathways, or longer-term outcomes beyond D2A. Data quality and integration across organisations also remain ongoing challenges.

But these limitations shouldn’t deter adoption. The greater risk lies in continuing to make high-stakes system decisions without robust modelling at all.

If we are serious about tackling delayed discharge, digital modelling should not sit on the periphery of its strategy, it should be embedded at its core

This project shows that operational research and simulation modelling can provide a practical framework for understanding complex discharge pathways.

By combining real data, with clinical expertise and advanced analytics, it shifts the conversation from short-term bed management to long-term system design. This is where the NHS needs to be.

If we are serious about tackling delayed discharge, digital modelling should not sit on the periphery of its strategy, it should be embedded at its core.

The tools already exist. They are transparent, adaptable, and free. What is required now is the willingness to collaborate, share data, and move from reactive crisis response to proactive, evidence-based system improvement.

Delayed discharge is a system problem. It is time we used system-level tools to solve it.