The Most Complex Boundary in English Healthcare

This is the first article in a series examining the boundary between local authority social care and the NHS — arguably the most constitutionally complex crossing in English healthcare. While our neighbourhood health series examined how NHS organisations struggle to govern the boundaries between them, and our private healthcare series analysed the ungoverned crossings in insured care, this series addresses the hardest version of the problem: two entirely different constitutional domains, sharing responsibility for the same populations, governed by different legislation, different regulators, different data regimes, different funding mechanisms, and different accountability structures.

In 1999, the Court of Appeal heard the case of Pamela Coughlan, a woman left tetraplegic after a 1971 road accident, who had been promised a “home for life” at Mardon House, an NHS facility in Exeter. When the North and East Devon Health Authority decided to close Mardon House and transfer her care to the local authority — which would have meant means-tested social care rather than free NHS provision — she challenged the decision.

The Court of Appeal’s judgment did more than protect Miss Coughlan. It established the legal framework for the boundary between NHS healthcare and local authority social care that remains in force today. The court found that there was an upper limit to the nursing care a local authority could lawfully provide. If the care required went beyond what was “merely incidental or ancillary” to the social care accommodation — if its quantity or quality exceeded what a local authority could reasonably be expected to provide — then the care was the NHS’s responsibility and must be provided free.

This became known as the “quantity and quality” test. It was codified in the National Framework for NHS Continuing Healthcare, developed into the “primary health need” concept, and embedded in Section 22 of the Care Act 2014, which explicitly prohibits local authorities from meeting care needs by providing services that the NHS is required to provide — unless doing so would be “merely incidental or ancillary” to other care the local authority is permitted to deliver.

Twenty-six years after Coughlan, the boundary the court drew between health and social care remains the most consequential line in English public services. On one side: universal, free at the point of need, nationally funded, clinically governed. On the other: means-tested, locally funded, subject to eligibility criteria, governed by assessed social care need. The same person, with the same conditions, can fall on either side of that line depending on how the “primary health need” assessment is completed, by whom, and in which part of the country.

This is not an edge case. It is the boundary that every patient with complex needs crosses — the boundary where the NHS and local government share responsibility for the same person and disagree about where responsibility lies.

NHS vs Local Authority: Two Constitutional Domains

The LA–NHS boundary is not merely an organisational boundary. It is a crossing between constitutional domains — two systems of governance built on fundamentally different legislative foundations, different accountability structures, and different theories of entitlement.

The NHS operates under the NHS Act 2006 and the Health and Care Act 2022. Its duties flow from the Secretary of State through NHS England to Integrated Care Boards. Its funding is national taxation, allocated centrally. Its provision is universal and free at the point of need. Clinical decisions are governed by professional registration, clinical standards, and the Bolam principle — the standard of care expected of a reasonable body of medical opinion. When something goes wrong, the question is whether the clinician acted in accordance with responsible professional practice.

Local authority social care operates under the Care Act 2014, the Children Act 1989, and the Mental Capacity Act 2005. Its duties flow from Parliament through elected councillors to the Director of Adult Social Services. Its funding is a mixture of council tax, government grant, and charges to service users. Its provision is means-tested and subject to national eligibility criteria that local authorities assess and apply. Decisions are governed by statutory duties and Wednesbury reasonableness — the standard that asks whether a reasonable local authority could have reached the decision in question. When something goes wrong, the question is whether the authority acted within its statutory powers and exercised reasonable judgement.

These are not two versions of the same system. They are two systems, built on different premises, for different purposes, under different democratic accountability. The NHS was designed to provide healthcare free at the point of need. Local authority social care was designed to provide a safety net for people whose social care needs could not be met by other means. One is a universal entitlement. The other is a residual service.

When Parliament, regulators, and policy-makers talk about “integration,” they are asking these two systems to operate as one. The Coughlan judgment tells us why they cannot simply merge: the legal boundary between them determines who pays. Cross it in one direction, and the individual receives free NHS care. Cross it in the other, and the individual may pay for their own care until their assets fall below the means-test threshold. The boundary is not administrative. It is financial and constitutional.

NHS Continuing Healthcare: The Boundary as Dispute Mechanism

The most revealing feature of the LA–NHS boundary is that it comes with its own legal machinery for resolving disputes about where responsibility lies. The Care and Support (Provision of Health Services) Regulations 2014 require local authorities and ICBs to agree dispute resolution procedures for cases where they disagree about whether an individual’s needs are primarily health needs (NHS responsibility) or social care needs (local authority responsibility). The National Framework for NHS Continuing Healthcare sets out the process: checklist, multidisciplinary assessment, Decision Support Tool, four key indicators (nature, complexity, intensity, unpredictability), and the “primary health need” determination.

The framework is thorough. It is also, by its nature, a mechanism for deciding which constitutional domain a person belongs to — not for governing the crossing between them.

The numbers reveal what this produces. NHS Continuing Healthcare eligibility fell 8.8% between September 2021 and June 2025, from 55,108 to 50,281 eligible individuals. The Nuffield Trust found in September 2025 that access to CHC is “unfair and inconsistent,” with significant regional variations that cannot be explained by differences in population need. The NHS Confederation’s January 2026 report called for a revised CHC framework, observing that disputes between ICBs and local authorities over CHC eligibility undermine local relationships and directly affect access to care for the individuals caught in between.

CHC is the boundary in its purest form. Two organisations, looking at the same person, applying the same framework, and reaching different conclusions about which constitutional domain should hold responsibility. The framework determines which organisation pays. It does not determine what happens to the person while the determination is being made. It does not govern how information crosses between the organisations during the assessment. It does not ensure that the clinical reasoning informing the NHS assessment and the social care reasoning informing the local authority assessment are visible to each other. It does not track what happens to responsibility for the individual while the two organisations negotiate.

The framework governs the allocation of responsibility. It does not govern the transfer of responsibility. And it is in that gap — between allocation and transfer — that harm occurs.

Delayed Discharge: The £2 Billion Boundary Symptom

The most visible and expensive manifestation of the ungoverned LA–NHS boundary is delayed discharge. In September 2025, 12,459 hospital beds were occupied by patients clinically fit for discharge, at an estimated cost of £220 million that month alone — annualising to over £2 billion per year. By November 2025, 291,000 bed days had been lost, 49,688 patients delayed, and 2,654 had been waiting 21 days or more.

The NHS’s own classification of the causes is instructive. 28.2% of delays are attributed to “interface processes” — a clinical euphemism for what happens at the boundary between the NHS and its partner organisations when they are negotiating onward care. Not capacity. Not clinical readiness. The interface itself.

A further 32.8% are attributed to capacity issues, primarily in social care — care home beds, domiciliary care packages, residential placements. These too are boundary problems, even if they present as capacity constraints. The capacity exists or does not exist in the local authority’s commissioned market. The NHS trust cannot create it, cannot commission it, cannot direct it. It can only wait for the other constitutional domain to provide what its own domain cannot.

The Discharge to Assess model, made permanent in 2022, attempted to address the problem by moving the CHC assessment out of the acute hospital and into a community setting. This was a sensible procedural change. It moved the assessment. It did not fix the boundary. The person still crosses from NHS to local authority care. Responsibility still transfers between constitutional domains. The governance of that transfer — who is responsible during the transition, what clinical information survives the crossing, whether the receiving organisation has explicitly accepted responsibility, whether the outcome is communicated back — remains procedurally managed rather than structurally governed.

Where Responsibility Dissolves at the LA–NHS Boundary

The LA–NHS boundary creates a specific pattern of failure that differs from the boundary failures we have analysed in NHS-to-NHS crossings and private healthcare pathways. In those contexts, the organisations on either side of the boundary share a constitutional domain — they operate under the same legislation, the same data governance regime, the same regulatory framework. The boundary is organisational: different entities, same system.

At the LA–NHS boundary, the crossing is constitutional. Different legislation. Different regulator. Different data infrastructure. Different professional standards. Different theories of what “good” looks like. The organisations are not just separate — they are structurally different.

This creates three distinct failure modes.

First, responsibility changes legal character as it crosses. A clinician’s duty of care under common law and professional regulation is not the same as a local authority’s statutory duty under the Care Act. When a patient moves from NHS care to local authority social care, the nature of the obligation changes — from clinical duty to statutory duty, from Bolam standard to Wednesbury reasonableness. But the person’s needs do not change character. They are the same needs, governed by different legal frameworks on either side of the crossing. No infrastructure marks the transition point. No system records the moment when the legal basis of the obligation shifts.

Second, information degrades at the constitutional boundary. The NHS operates on clinical terminology (SNOMED CT), clinical record structures (FHIR), and clinical data flows (GP Connect, Transfer of Care, MESH). Local authorities operate on social care assessment frameworks, different record systems, and — for most — the Digital Social Care Records programme, which had achieved approximately 63% adoption against an 80% target. There is no Transfer of Care FHIR equivalent for the NHS-to-social-care crossing. A discharge summary sent from an NHS trust to a social care provider may arrive as a PDF, a letter, or a fax. The clinical reasoning is present. The structured data is not. The receiving organisation must manually interpret clinical information and translate it into a social care assessment framework. Context is lost in translation.

Third, consent silently changes purpose. In the NHS, the lawful basis for processing health data is typically established under UK GDPR Article 6(1)(e) (public task) and Article 9(2)(h) (health or social care purposes). When the same data crosses to a local authority, the lawful basis may shift — the local authority processes it under its own statutory duties, its own privacy notices, its own data sharing agreements. The Data (Use and Access) Act 2025 has clarified that organisations cannot rely on another body’s statutory functions to legitimise their own processing. Yet the patient is rarely informed that their clinical data has crossed a constitutional boundary and is now being processed under a different legal basis by a different type of organisation.

These are not theoretical risks. They are the operational reality of every complex discharge, every CHC assessment, every safeguarding referral, every transition between health and social care. And they share a common structural feature: responsibility does not vanish because someone is negligent. It dissolves because no infrastructure makes the crossing explicit, time-bounded, or reconstructable.

MVRT: What Safe Responsibility Transfer Requires

At Inference Clinical, we describe the minimum structural requirements for responsibility to move safely between organisations as Minimum Viable Responsibility Transfer — MVRT. The concept is drawn from how accountability works in industries that solved cross-boundary responsibility transfer decades ago.

Payment systems require the receiving institution to explicitly acknowledge a transaction, with timestamps that determine liability to the millisecond. If acknowledgement does not arrive within a defined window, the transaction fails. Silence is failure, not consent.

Aviation requires the receiving controller to explicitly accept jurisdiction before the handing-off controller is released. Until acceptance is registered, the original controller remains responsible and must continue to act. The handoff is a discrete, evidenced, time-bounded event.

Healthcare has absorbed neither lesson. At the LA–NHS boundary, a discharge can be “completed” without the receiving social care provider explicitly confirming they have accepted responsibility for the individual. A referral can be “sent” without evidence that the receiving organisation understood the clinical reasoning behind it. A CHC determination can be “communicated” without the individual or their representative confirming they understand what it means for their care and their financial liability.

MVRT asks five questions of every boundary crossing:

Who is responsible now? What exactly are they responsible for? For how long? Under what conditions does that responsibility end, escalate, or re-route? And what evidence proves this was understood by both parties at the time?

If a system cannot answer those questions explicitly, in real time, at the point of crossing, responsibility has not transferred. It has been assumed. And assumptions do not survive contact with pressure — clinical deterioration, organisational disagreement, or the slow accumulation of risk in the gap between two organisations that each believe the other is acting.

Constitutional Binding: Governing Cross-Domain Data

Cross-constitutional crossings require something more than MVRT alone. When responsibility transfers within the NHS — between two trusts, or between an ICB and a provider — the legal framework does not change. Both sides operate under the same legislation. The data governance regime is shared. Professional standards are consistent.

At the LA–NHS boundary, the framework changes. An observation made by an NHS clinician under clinical authority becomes part of a social care assessment under Care Act statutory duty. The data is the same. The governing authority is different. And in today’s systems, that transition happens silently — no marker, no record of the jurisdictional shift, no constraint on what can happen to the information once it crosses.

We describe this as the need for Constitutional Binding — the principle that every action, decision, and data element in a cross-constitutional care context must be explicitly bound to the constitutional authority under which it was created, and that binding must travel with the data and constrain what can happen next.

Constitutional Binding has three properties. It must be explicit: an NHS clinical observation processed under NHS data governance rules does not silently become a social care record when it enters a local authority system. The crossing must be marked. It must be evidenced: the binding points to the statutory basis, the professional authority, and the lawful basis for processing. And it must be consequential: the binding determines what can happen next. An NHS-bound clinical observation can flow within NHS data governance rules. The moment it is used in a Care Act assessment, that is a boundary crossing that requires its own lawful basis, its own consent event, its own evidence trail.

Integration is not a binding. Integration is what happens when two explicit bindings interact under governed conditions. The moment you treat “integrated care” as a single undifferentiated domain, you lose the ability to reconstruct which authority was governing which action at which point. And reconstruction — the ability to prove, after the fact, who was responsible, under what authority, with what evidence — is precisely what the LA–NHS boundary lacks.

The Coughlan Judgment Legacy and the Seven Flows

The Coughlan judgment was right. It identified that the boundary between health and social care is a constitutional line with financial consequences, and that the line must be drawn with legal precision. The CHC framework that followed was right. It provided a structured process for determining which side of the line an individual falls on.

What neither provided — because neither was designed to — is governance of the crossing itself. Coughlan drew the line. The CHC framework determines which side of the line a person belongs on. But the moment of crossing — when the person moves from one constitutional domain to the other, when responsibility transfers, when clinical information enters a social care governance regime, when the legal character of the obligation changes — that moment remains structurally ungoverned.

The mechanisms that exist are financial (the Better Care Fund, Section 75 agreements) or procedural (Safeguarding Adults Boards, discharge protocols, joint assessment processes). These are important. They are not infrastructure. A Section 75 agreement pools budgets between an NHS body and a local authority. It does not govern what happens to clinical information at the crossing. A Safeguarding Adults Board convenes partners to review failures. It does not provide the information infrastructure that would prevent those failures recurring.

The Seven Flows framework — Identity, Consent, Provenance, Clinical Intent, Alert and Responsibility, Service Routing, and Outcome — provides the structural requirements for any organisational boundary crossing. At the LA–NHS boundary, each flow encounters the constitutional gap:

Identity: the NHS uses the NHS number as a universal identifier. Local authorities use their own client identifiers. Reconciliation between the two is manual and inconsistent. Consent: different lawful bases on each side of the boundary, with the transition rarely marked or explained to the individual. Provenance: clinical records in FHIR and SNOMED CT crossing into social care assessment frameworks with no structured interoperability. Clinical Intent: a clinician’s reasoning about ongoing care needs may not survive translation into a social care eligibility assessment. Alert and Responsibility: no mechanism for the receiving organisation to explicitly accept responsibility with a timestamp and evidence trail. Service Routing: the CHC framework routes individuals to a constitutional domain, but does not govern what happens within or between the services that domain provides. Outcome: the NHS trust discharges the patient. Whether the social care provider received the person, accepted responsibility, and achieved the intended outcome may never flow back.

These are not technology problems waiting for a platform to solve them. They are governance gaps — the absence of structural requirements at the most consequential boundary in English public services. Inference Clinical’s Boundary Risk Assessment methodology applies these seven governance invariants to real organisational boundaries, producing a structured Boundary Risk Score that quantifies where the gaps are widest.

Julian Bradder

Founder & CEO, Inference Clinical

Julian leads Inference Clinical’s work on governance infrastructure for clinical handover. He has 30 years’ experience in digital transformation across government, FTSE 100, and healthcare. Connect on LinkedIn