Governance and Accountability When Working With ICBs and Mental Health Trusts
When system working fails, it usually fails on accountability: who is leading, who decides, and what happens when thresholds are contested or partner response is slow. Good governance makes accountability explicit and repeatable, so safety does not depend on individual relationships. For ICBs, governance is also the mechanism for assurance: can a provider evidence control of risk, performance and learning across interfaces? This article sits within working with ICBs in mental health and connects to mental health service models and pathways, setting out practical governance structures that commissioners recognise as mature and that support safe, auditable delivery.
The accountability problem in integrated delivery
In integrated systems, accountability becomes blurred for predictable reasons:
- Different services have different thresholds, operating hours and response routes.
- Risk changes quickly, and “who is leading” can shift during escalation.
- Information quality varies, creating disagreement about the urgency of response.
- Performance pressure encourages scope drift and informal workarounds.
Governance must therefore do more than meet contract requirements; it must control interface risk, clarify leadership, and provide a fair mechanism for challenge and escalation.
Core governance components for working with ICBs and Trusts
1) Role clarity and delegated authority
Commissioners expect clarity on who can make operational decisions (acceptance, step-up, discharge), who owns safeguarding escalation, and who holds clinical oversight. Providers should document delegated authority internally so staff do not “freeze” at the point of risk escalation or rely on informal guidance.
2) Escalation ladders and dispute resolution routes
Effective governance defines escalation ladders: front-line escalation, manager escalation, senior escalation. It also defines a dispute resolution route for contested thresholds (for example, disagreement about whether a person meets crisis criteria). Importantly, escalation should be recorded and tracked, creating an evidence trail and supporting learning.
3) Shared risk registers that focus on interface risk
A shared risk register is only useful if it captures the risks that actually cause harm: delayed escalation, unclear referral thresholds, safeguarding drift, long-running restrictions, information-sharing failures, and workforce instability at interface points. Risks should have owners, mitigations, review dates and evidence of monitoring.
4) Assurance rhythms that combine data and sampling
Governance should combine quantitative oversight (activity, timeliness, repeat escalations) with sampling and “assurance traces” that demonstrate file-level reality. This is often what distinguishes mature providers: they can show both trends and evidence chains.
Operational examples (governance that prevents repeat failures)
Example 1: Managing contested thresholds without unsafe delay
Context: A provider escalates a deteriorating case to a Trust team, but thresholds are contested and response is delayed. Staff feel stuck and uncertainty grows. The ICB wants assurance that escalation is managed safely and that disputes do not lead to unmanaged risk.
Support approach: The provider implements a contested-threshold protocol: a standard escalation summary, a manager escalation requirement if no response within an agreed timeframe, and a senior-to-senior dispute route. The provider also adds a “temporary safety plan” requirement while disputes are resolved.
Day-to-day delivery detail: Front-line staff complete the standard summary and record time of escalation. If response is delayed, managers escalate using the agreed ladder and record partner responses. Meanwhile, staff implement a temporary safety plan (increased contact frequency, safeguarding actions where relevant, and clear emergency instructions) documented in the case record. Governance reviews contested cases monthly and identifies patterns (time-of-day issues, unclear criteria, information gaps) and agrees corrective actions with partners.
How effectiveness/change is evidenced: Faster escalation decisions, fewer unsafe delays, and clear audit trails of escalation steps and interim safety planning. Evidence includes escalation logs, sampled cases and governance minutes showing action and verification.
Example 2: Shared risk register reduces repeat interface failures
Context: The system experiences repeated issues: referral bounce, delayed crisis access, and variable safeguarding response across localities. Each issue is discussed, but nothing changes sustainably.
Support approach: The provider creates an interface risk register with structured risks, owners, mitigations, and monitoring measures (for example, time-to-first-contact, referral decline reasons, safeguarding action completion, repeat escalation rates). The register is reviewed monthly with the ICB and Trust partner leads.
Day-to-day delivery detail: Operational leads update the register with evidence: trend data, audit samples and assurance traces. Where mitigations are agreed (referral checklist changes, escalation route updates, training on thresholds), actions are tracked with deadlines and re-audit dates. The provider runs small monthly samples to verify that changes are visible in case files and staff practice.
How effectiveness/change is evidenced: Reduced repeat failures, improved consistency across teams, and clearer commissioner confidence because risks are actively managed rather than repeatedly discussed. Evidence includes the register, action tracking and re-audit results.
Example 3: Governance controls restrictive practice at interfaces
Context: Following safeguarding concerns, teams apply restrictions and increased monitoring. As cases move across services, restrictions persist without review and become embedded, creating rights risks and potential CQC concern.
Support approach: Governance introduces a restrictions standard: rationale, least restrictive alternatives, time limit and review date. Restrictions are reviewed monthly and included as an interface risk on the shared risk register, with quarterly senior sampling.
Day-to-day delivery detail: Supervisors require explicit review decisions in the record: continue, modify, or step down. Where cases are handed over, the minimum dataset includes restrictions and review dates so the receiving team is clear on expectations. Governance reviews long-running restrictions and ensures safeguarding actions and positive risk-taking plans are active, preventing restriction from becoming “default containment”.
How effectiveness/change is evidenced: Reduced duration of restrictions, clearer proportionality documentation, and improved step-down decisions. Evidence includes the restrictions register, case sampling and governance action tracking.
Explicit expectations that must be met
Commissioner expectation
Commissioners expect accountable governance with auditable escalation and controlled interface risk. They will look for clear roles, escalation ladders, shared risk management, and evidence that governance decisions translate into file-level practice. They also expect verification: re-audit, sampling and trend improvement, not just meeting minutes.
Regulator / Inspector expectation (e.g. CQC)
CQC expects leadership and governance to deliver safe, consistent practice and protect rights. Inspectors will test whether accountability is clear at points of risk, whether safeguarding is timely and effective, and whether restrictive practice is least restrictive and reviewed. They will also triangulate governance claims against staff understanding and case record evidence.
What to present in contract and quality meetings
In contract and quality meetings, mature providers bring: a short set of interface assurance traces, trend data linked to operational logs, a live risk register with owners and mitigations, and evidence of verification (re-audit). This makes accountability visible and reduces the risk of partnership working becoming a debate about intentions rather than an assessment of operational control.