What the NHS can learn from international models — and why the operational layer matters more than the label.
Bec Richmond
Senior Specialist Advisor · International Perspectives & Facilitation
Across the NHS, we talk a lot about integration. But anyone working across primary, community, or acute care knows the truth: integration doesn’t happen because we draw new organisational charts. It happens when teams change how they work together.
That’s why the concept of the “medical neighbourhood” — widely used in health systems I’ve had the privilege of working alongside internationally — deserves our attention. Nobody has it perfect. But there is a practical, operational blueprint emerging from these models that aligns uncomfortably well with what the NHS is trying, and often struggling, to achieve.
The name, frankly, is imperfect. Neighbourhood working isn’t, and shouldn’t be, primarily medical. But the ingredients matter far more than what we call the cake. So, let’s focus on those.
The NHS doesn’t need more structures. It needs practical, proven ways to strengthen the relationships, processes, and shared accountability that make integrated neighbourhood care real — not just a policy ambition.
What a medical neighbourhood actually is
At its core, a medical neighbourhood is a structured way of organising collaboration between primary care, specialist and acute services, community teams, mental health, diagnostics, pharmacy, social care — and the patients and citizens that all the above exist to serve.
It is not a new organisation. It is not a new service. It is a set of agreements, processes, and behaviours that make care safer, faster, and more joined-up. It is the missing operational layer between high-level NHS strategy and the day-to-day reality of patient experience.
The challenges these models have been designed to fix are painfully familiar: unclear handoffs between providers, fragmented communication, patients bouncing between services with no coherent thread of support. The medical neighbourhood tackles these not with more structure, but with clarity, consistency, and accountability.
What to do. Fast.
- Identify the right partners
Start small. Pick specialties with high volume, high variation, or high frustration — dermatology, MSK, cardiology, gastroenterology, mental health. Find the good neighbours who are ready to collaborate, including from the voluntary and community sector. They are out there. They want to help. - Develop care compacts — not just contracts
A care compact is a shared, human agreement that defines what good behaviour looks like: referral standards, communication timelines, what information is needed to make care seamless. This is not about contracts and incentives first — it is about culture and an agreed way of working. The contracts matter, but start here. - Build in continuous quality improvement
Integration is not a one-off project. It is a cycle. Agree upfront how and when evaluation will happen, and what you will measure — for patients and for the system. Then iterate, iterate, iterate.
Five things the evidence keeps telling us
- Integration is a behaviour, not a structure
ICSs can set the vision, but neighbourhoods deliver it. The magic happens in the agreements, workflows, and relationships between teams — not in the organogram. - Start small, scale fast
Pick two or three specialties. Build the model. Prove it works. Then spread it deliberately, not organically. - Standardisation is not bureaucracy
Clear referral criteria, shared templates, and defined responsibilities reduce workload. They don’t add to it. - Digital is an enabler, not the solution
Technology supports the neighbourhood; it doesn’t create it. The real work is cultural and relational. - Shared metrics change behaviour
When inter-agency teams look at the same data, in the same room, with the same goals — everything shifts.
The NHS has the ambition. What it often lacks is the operational architecture to turn that ambition into daily reality. That architecture exists, tested and refined, in health systems across the world. We should be borrowing from it — and building on what already works here.