Safeguarding Adult Reviews, CHC Disputes and the Accountability Gap

This is the sixth article in a series examining the boundary between local authority social care and the NHS. The first article established the constitutional domain gap. The second examined the simultaneous reorganisation of both sides. The third showed the £2.6 billion annual cost of delayed discharge. The fourth revealed the structural asymmetry in data infrastructure. The fifth demonstrated why neighbourhood health cannot work without boundary governance. This article examines what happens when boundary failure causes harm — and why the mechanisms designed to catch it keep finding the same problems without solving them.

In July 2024, the Local Government Association and the Association of Directors of Adult Social Services published the second national analysis of Safeguarding Adult Reviews in England. The analysis examined 652 SARs completed between April 2019 and March 2023, with 229 reviewed in detail. The findings are stark: 72% identified poor coordination between agencies. 70% identified information-sharing failures. Only 24% noted effective cross-agency communication. Only 23% noted effective cross-agency coordination.

Read those numbers again. In nearly three-quarters of the cases serious enough to trigger a statutory review — cases where an adult died or suffered serious harm — the review found that the agencies involved did not coordinate effectively. In seven out of ten, they did not share information effectively. The most dangerous crossing in English public services is between health and social care. The evidence says so, repeatedly, in the reviews commissioned after people are harmed.

This article examines three mechanisms that exist to identify and address boundary failures — Safeguarding Adult Reviews, continuing healthcare disputes, and CQC local authority assessments — and a fourth, the Casey Commission, that is examining the system those mechanisms are embedded in. All four are looking at the same boundary. None is addressing it as an infrastructure problem.

Safeguarding Adult Reviews: 72% Find Poor Coordination at the Boundary

Under Section 44 of the Care Act 2014, Safeguarding Adults Boards must arrange a SAR when an adult with care and support needs dies from abuse or neglect and there is concern that agencies could have worked more effectively. The purpose is to identify lessons about how local professionals and agencies work together to safeguard adults.

The critical phrase is “work together.” SARs are not primarily reviews of individual clinical or social care practice — though they examine those too. They are reviews of how organisations interact. How information crosses between them. How responsibility transfers. How alerts propagate. How outcomes are tracked. They are, by statutory design, reviews of the boundary.

The second national analysis found that management oversight failures appeared in 31% of cases, training gaps in 23%, and poor policies in 28%. But the headline findings — poor coordination at 72%, information-sharing failures at 70% — dwarf everything else. The most common type of failure is not what happens within an organisation. It is what happens between them.

The analysis identified four recurring themes across SARs nationally: mental capacity, making safeguarding personal, application of safeguarding processes, and multi-agency information sharing and communication. Three of these four operate at or across the boundary. Mental capacity assessment involves health professionals and social care professionals applying different professional frameworks to the same individual. Making safeguarding personal requires coordination across the constitutional domain gap. Multi-agency information sharing is, literally, the problem of governed data crossing that the fourth article in this series documented.

The pattern in SAR after SAR is consistent. An NHS professional observes something concerning but does not know the social care context. A social care worker identifies clinical deterioration but cannot reach the right clinical team quickly enough. Information exists in one system but is not available in the other. A safeguarding concern is raised but the alert crosses the boundary slowly, informally, or not at all. Responsibility transfers implicitly rather than explicitly — and in the gap between one organisation’s duty ending and another’s beginning, harm occurs.

These are not failures of individual competence. They are failures of infrastructure. The people involved are usually doing their best within systems that do not support governed crossing between constitutional domains.

NHS Continuing Healthcare Disputes: A Boundary Governance Failure

Continuing healthcare eligibility is the most financially consequential boundary-crossing decision in English public services. If a person’s primary need is for healthcare — as established in the Coughlan case that the first article in this series examined — the NHS funds their entire care package, free at the point of use. If their primary need is for social care, the local authority takes responsibility, the person is means-tested, and those with assets above £23,250 must self-fund.

The stakes could not be higher. Care home fees of £50,000–100,000 per year fall on one side of the boundary or the other. The difference between an NHS-funded placement and a self-funded one is the difference between a family preserving its assets and a family losing everything.

In the 2024–25 financial year, 51,981 assessments were made for CHC funding in England. 80.5% were rejected. As of 31 March 2024, 52,096 people were eligible — 34,055 via the standard route and 18,041 via fast-track. Between January and March 2024, just 21% of people assessed for standard CHC were found eligible — but this ranged from 7.3% in Gloucestershire to 42.5% in Leicester, Leicestershire and Rutland. A sixfold variation in eligibility rates for the same national framework, applied to the same boundary, under the same law.

This variation is not explained by differences in population health. It is explained by differences in how the boundary is interpreted, administered, and contested. The Nuffield Trust’s analysis describes CHC as “a window into the stark divide in our system between care that is funded by the NHS and care that isn’t.” The PHSO’s investigations between April 2018 and July 2020 examined 150 cases and found failings in 55 — including cases where families paid tens of thousands of pounds for care the NHS should have funded.

CHC disputes are conventionally understood as clinical disagreements about the nature and severity of a person’s needs. They are not. They are governance disputes about where the boundary falls and who is responsible for what on each side. The Decision Support Tool — the instrument used to determine eligibility — is a boundary-mapping exercise. It asks assessors to determine whether each domain of need (behaviour, cognition, communication, mobility, nutrition, continence, skin integrity, breathing, drug therapies, altered states of consciousness, psychological and emotional needs) is primarily health or primarily social care in character. For each domain, the question is: which side of the constitutional boundary does this need fall on?

The National Framework provides the assessment criteria. But it does not provide the infrastructure for making that assessment governed, transparent, and consistent. As of 31 March 2024, 1,730 referrals were incomplete and delayed beyond 28 days — including 40 delayed by more than six months. In Q4 of 2023/24, there were 596 local resolution requests, of which only 13% resulted in eligibility — meaning one in six decisions is overturned when challenged at even the first stage of appeal.

The incentive structure makes this worse. It is in each ICB’s financial interest to find people ineligible for CHC, because eligibility transfers the entire cost of care from the means-tested local authority system to the NHS. It is in each local authority’s interest to establish eligibility, because it relieves them of financial responsibility entirely. The boundary is not just a governance gap — it is a financial fault line where billions of pounds of care costs are contested annually, and the person at the centre of the dispute is caught between two organisations each trying to assign responsibility to the other.

Every CHC dispute is a boundary failure. Not because the wrong decision is always made, but because the infrastructure for making the decision is not governed at the crossing. The assessment draws on clinical data held in NHS systems and social care data held in local authority systems. The multi-disciplinary team includes health professionals and social care professionals operating under different professional frameworks. The consent to share information across the boundary operates under different lawful bases on each side. The outcome of the assessment — eligibility or ineligibility — determines which constitutional domain takes responsibility. And none of this has infrastructure. It runs on process, negotiation, and dispute resolution — the same mechanisms that the first article identified as procedural rather than infrastructural.

CQC Local Authority Assessments: Nodes Without Edges

In December 2023, the CQC began a new assessment programme for all 153 local authorities in England with adult social care responsibilities, required under the Health and Care Act 2022. By October 2025, the CQC had published 27 assessment reports, with each authority scored 1–4 across nine quality statements covering four themes.

The nine quality statements include “safe systems, pathways and transitions,” “partnerships and communities,” and “safeguarding.” These sound like boundary assessments. They are not. They are assessments of how well a local authority performs its own duties — including its duties to work in partnership. The assessment examines one node in a network. It does not examine the edges that connect nodes together.

The results so far illustrate the distinction. Blackpool and East Riding of Yorkshire have been rated inadequate. East Riding scored 1 in all but one area, achieving only a 2 for “partnerships and communities.” Lancashire was rated requires improvement, with the CQC noting that “partnership working was hindered by strained relationships” and that “some organisations had disengaged.” Staff described electronic records systems as “unreliable and hard to use, hampering timely, person-centred assessments.” On the other hand, Liverpool was rated good, with the CQC praising its “positive culture of continuous learning and improvement” and noting the authority “worked well with partner organisations to deliver intermediate care and reablement services.”

These are useful assessments of individual local authorities. But consider what they cannot assess. When a safeguarding alert crosses from an NHS trust to a local authority, the CQC can assess how the local authority receives and processes it. It cannot assess the crossing itself — the infrastructure (or absence of infrastructure) that governs how the alert travels between the two constitutional domains. When a patient is discharged from hospital, the CQC can assess whether the local authority has adequate processes for receiving discharge referrals. It cannot assess the governance of the handoff — the identity reconciliation, consent transformation, clinical information compression, and responsibility transfer that the third article documented in detail.

The CQC also regulates NHS trusts and inspects care providers. But it does so through separate assessment frameworks, under separate regulatory powers. There is no CQC framework for assessing the boundary itself — the governed infrastructure that connects the organisations it regulates on each side. The State of Care report can describe partnership working in general terms. It cannot measure whether the Seven Flows are governed at any specific crossing.

This is the nodes-and-edges problem. English health and social care regulation assesses the nodes thoroughly — individual organisations, their internal processes, their staffing, their leadership, their cultures. It does not assess the edges — the governed connections between organisations across which responsibility, information, clinical intent, and accountability must transfer. SARs assess the edges, but only retrospectively, after harm has occurred. CHC disputes assess one specific type of edge — the financial boundary — but through adversarial process rather than governed infrastructure. And the CQC’s new local authority assessments, for all their thoroughness, assess one side of the boundary at a time.

Casey Commission on Social Care: The Boundary as Infrastructure Problem

In January 2025, the government announced an independent commission into adult social care, chaired by Baroness Louise Casey. The Commission started work in May 2025, with Phase 1 reporting in 2026 and Phase 2 by 2028. Its mandate is to set out a vision for a “national care service,” with recommended measures and a roadmap for delivery.

The terms of reference are broad. Phase 1 should start a “national conversation” about what adult social care should deliver, conduct a “data-driven deep-dive” into how existing resources are being used, and produce “tangible, pragmatic solutions” that can be implemented over a decade. Phase 2 should consider “the model of care needed to address demographic change” and “alternative models that could be considered in future.”

The terms of reference ask the Commission to consider older people’s care and working-age disabled adults separately. They ask it to consider how services must be organised and how they should be funded. They ask it to build cross-party consensus. What they do not ask — and what the Commission will inevitably encounter — is what infrastructure is required for the boundary between health and social care to function as a governed crossing rather than an ungoverned gap.

This matters because any recommendation the Casey Commission makes will run into the boundary. If it recommends a national care service, the boundary between that service and the NHS must be governed. If it recommends deeper integration, the constitutional domains being integrated must have infrastructure for governed crossing. If it recommends reformed funding — whether a Dilnot-style cap, a social insurance model, or full public funding — the financial boundary between NHS and social care must have governed infrastructure for determining which side pays. If it recommends better data sharing, the data governance chasm documented in the fourth article must have governed crossing infrastructure, not just legal authority to share.

Every previous attempt to reform adult social care has run into the same problem. The Dilnot Commission recommended a cap on care costs. The 2014 Care Act legislated for one. It was never implemented. The Health and Social Care Act 2012 created Health and Wellbeing Boards to join up health and social care. They have no power to govern the boundary between the organisations they convene. The Better Care Fund pools approximately £9 billion of NHS and local authority funding. But pooled funding does not create governed crossings — it creates shared budgets administered through the same ungoverned boundary mechanisms.

ADASS responded to the Casey Commission’s timescales with a direct warning: “Unfortunately, the timescales announced are too long and mean there won’t be tangible changes until 2028. We worry that continuing to tread water until an independent commission concludes will be at the detriment of people’s health and wellbeing.” The concern is legitimate. But the deeper problem is not timescales. It is that the Commission’s terms of reference frame the challenge as a social care problem that interfaces with the NHS, rather than a boundary infrastructure problem that spans both.

The Pattern: Same Boundary, Same Failures, No Infrastructure

Four accountability mechanisms. Each examining the same boundary. Each finding the same failures. None building the infrastructure to prevent them.

SARs find that agencies do not coordinate and do not share information — in 72% and 70% of cases respectively. Their recommendations call for better communication protocols, improved training, and stronger partnership working. These are procedural fixes for an infrastructural problem. The next SAR will find the same failures, because the infrastructure has not changed.

CHC disputes find that the boundary between health and social care is contested, inconsistent, and financially adversarial. The National Framework provides criteria but not infrastructure. The sixfold variation in eligibility rates — 7.3% to 42.5% — is not a calibration error. It is the inevitable consequence of an ungoverned boundary where the financial incentives of each side point in opposite directions.

CQC local authority assessments find that some authorities work well in partnership and others do not. But they assess partnership as an attribute of individual organisations rather than a property of the infrastructure connecting them. An authority can score well on “partnerships and communities” despite having no governed infrastructure for crossing the boundary with NHS organisations — as long as the relationships are currently good and the individuals currently in post are effective collaborators.

The Casey Commission will examine the whole system. Its recommendations will be important. But unless it engages with the boundary as an infrastructure problem — not just a policy problem, a funding problem, or a culture problem — its recommendations will join the long list of reforms that run aground on the same ungoverned crossing.

What Boundary Accountability Infrastructure Requires: Seven Flows

The Minimum Viable Responsibility Transfer — the five questions that the first article introduced — applies to every accountability mechanism examined in this article:

For every safeguarding alert that crosses the boundary: who is responsible now, for what clinical or social care duty, for how long, under what conditions does that responsibility escalate or end, and what evidence proves that the receiving organisation understood and accepted the transfer?

For every CHC assessment: who holds responsibility during the assessment period, for what package of care, under what authority, what happens when the assessment is delayed beyond 28 days, and what evidence proves that no gap in care existed during the determination?

For every boundary crossing that the CQC might assess: is the crossing governed by infrastructure that records the transfer, tracks the responsibility, and provides evidence of what happened — or is it governed by relationships, phone calls, and “local arrangements”?

Each of these accountability challenges maps to the Seven Flows:

Identity — The person at the centre of a safeguarding concern, a CHC assessment, or a CQC-assessed pathway must be identifiable across both constitutional domains. The NHS number is not universally used in social care. The local authority client ID is not known to the NHS. Every accountability mechanism starts with an identity reconciliation problem.

Consent — Sharing information for a safeguarding enquiry operates under different lawful bases than sharing information for a CHC assessment, which operates under different lawful bases than sharing information for routine care coordination. Each crossing requires governed consent infrastructure, not ad hoc agreements.

Provenance — When a SAR examines what went wrong, it needs to trace who created what information, under what authority, and who had access to it. When a CHC panel weighs clinical and social care evidence, it needs to know the provenance of each data element. Without provenance tracking across the boundary, accountability is reconstructed retrospectively from fragmented records.

Clinical Intent — A safeguarding concern involves clinical and social care judgements that may point in different directions. A CHC assessment requires the multi-disciplinary team to weigh clinical need against social care need. Without governed infrastructure for recording and reconciling cross-domain clinical intent, these judgements are made through negotiation rather than through evidence.

Alert and Responsibility — The most critical flow for safeguarding. When an NHS professional identifies a concern, the local authority is the lead agency under the Care Act. The alert must cross the constitutional boundary with explicit responsibility transfer — not an email that may or may not be acknowledged, not a phone call that may or may not be returned, but governed infrastructure that records the moment responsibility transfers, who accepted it, and what action was required.

Service Routing — CHC is fundamentally a routing decision: does this person’s care package route through the NHS (free, universal, clinically governed) or through the local authority (means-tested, locally administered, Care Act governed)? Without governed routing infrastructure, the decision is made through adversarial process rather than transparent assessment.

Outcome — Does the sending organisation know what happened after the crossing? Does the hospital know whether the safeguarding alert was acted on? Does the referring clinician know the result of the CHC assessment? Does anyone measure whether the person’s outcomes improved after the boundary was crossed? In most cases, the answer is no. The outcome loop does not close across the boundary.

At Inference Clinical, the Seven Flows audit methodology assesses each of these conditions at the specific crossings where accountability failures occur. For safeguarding, the audit maps every alert pathway across the boundary and identifies where governed infrastructure exists and where the crossing relies on informal process. For CHC, the audit examines the assessment pathway as a boundary crossing rather than a clinical determination, identifying where governance infrastructure would reduce variation, delays, and disputes. For CQC-assessed pathways, the audit provides the edge assessment that complements the CQC’s node assessment — the infrastructure layer between the organisations that the CQC regulates separately.

What the Casey Commission Must Recommend: Infrastructure Before Reform

The Casey Commission’s Phase 1 report, due in 2026, will recommend measures for medium-term reform. The evidence in this article — and across this series — suggests that one recommendation is foundational to all others: the boundary between health and social care needs governed infrastructure, and that infrastructure needs to be built before, not after, the reforms that depend on it.

Neighbourhood health cannot work without it. Delayed discharge cannot be solved without it. Safeguarding cannot be reliable without it. CHC cannot be consistent without it. CQC cannot assess it. And no amount of policy reform, funding reform, or structural reorganisation will create it as a byproduct.

The boundary is not the problem. The absence of infrastructure at the boundary is the problem. Every accountability mechanism in this article confirms it. What none of them provides is the infrastructure to fix it.

A Boundary Risk Assessment provides the prospective assessment that SARs, CHC panels, and CQC frameworks cannot — mapping where governed infrastructure exists at the crossing and where informal process masks structural gaps. The Boundary Risk Score quantifies this for every crossing type, giving safeguarding leads and commissioning teams a measurable baseline before the next accountability review finds the same failures again.

Julian Bradder

Founder & CEO, Inference Clinical

Building infrastructure for inter-organisation responsibility transfer in healthcare. 30 years’ experience in digital transformation across government, FTSE 100, and healthcare.