Referral Criteria in Mental Health Services: Balancing Access, Risk and Capacity

Referral criteria are not just an admin filter. They determine access to care, the safety of redirection decisions, and the credibility of a service under commissioner scrutiny. Poorly designed criteria create two predictable failures: people who need specialist support are wrongly diverted, and people who do not meet scope flood the system until triage collapses. This article sits within Access, Referral & Clinical Triage and aligns criteria design with pathway reality in Mental Health Service Models & Care Pathways. The focus is practical: how to balance access, risk and capacity while keeping decisions safe and defensible.

What referral criteria must do (beyond “in scope / out of scope”)

Effective criteria clarify:

• Who the service is commissioned to serve (population, severity, complexity).
• What the service will deliver (interventions, coordination, recovery support, specialist functions).
• What sits elsewhere (primary care, social care, crisis, specialist pathways).
• How risk is handled when someone is redirected or declined.

Criteria that only describe eligibility without defining safe alternatives create unmanaged risk.

Commissioner expectation (explicit)

Commissioner expectation: Commissioners expect eligibility rules to match the commissioned service model and to be applied consistently. They also expect evidence that redirection does not create harm, including robust safety-netting, documented rationale, and monitoring of outcomes after decline or diversion.

Regulator / inspector expectation (explicit)

Regulator / inspector expectation (CQC): Inspectors will look for safe decision-making at the boundary of services. They will expect clear safeguarding escalation routes, evidence that vulnerable people are not inappropriately excluded, and governance oversight of adverse events linked to access or refusal decisions.

Designing criteria around real pathway capacity

Criteria must reflect what is realistically available locally. If the “alternative service” is overstretched, redirection becomes unsafe unless there is active bridging support or escalation.

Operational example 1: Criteria that include “route + safety net”

Context: Staff decline referrals that do not meet threshold, but outcomes are unknown and referrers complain about “bouncing” patients.

Support approach: Rewrite criteria so every decline category includes a recommended route and a safety-net action.

Day-to-day delivery detail:

• Decline letters use standard wording plus case-specific rationale in plain English.
• Each decline template includes: recommended service route, minimum information to include, and urgent escalation instructions.
• Where risk markers exist (self-harm history, safeguarding, rapid deterioration), a duty clinician completes a brief risk screen before decline is confirmed.
• A named contact route is provided for referrers to clarify missing information quickly, reducing repeated inappropriate referrals.

How effectiveness is evidenced: Monthly reporting tracks repeat referrals after decline, complaints relating to access, and any incidents occurring within 30 days of a decline decision. Cases are reviewed for learning and criteria refinement.

Handling “borderline” cases consistently

Borderline cases are where inconsistency appears. Good services define how to handle these situations: either through a short assessment offer, a joint call with another pathway, or a time-limited bridging intervention while signposting to the correct service.

Operational example 2: Borderline pathway review panel

Context: Referrals with mixed presentations (for example, trauma, substance use, housing insecurity) lead to variable outcomes depending on the triaging clinician.

Support approach: Create a brief weekly “borderline review” with senior oversight.

Day-to-day delivery detail:

• Cases are selected using a clear rule: any referral declined twice, any referral involving safeguarding, and any referral where referrer disputes eligibility.
• Panel includes senior clinician and operational lead to align clinical decision with pathway capacity and commissioning scope.
• Outcomes include: accept, redirect with safety-net, or joint pathway plan (for example, coordinated approach with another team).
• Decisions are documented with rationale and used to update staff guidance and examples library.

How effectiveness is evidenced: Reduction in repeated referrals, improved decision consistency in audit samples, and fewer escalations to commissioners about “gatekeeping”.

Equality, inclusion and reasonable adjustments

Referral criteria must be tested for unintended exclusion. Groups at risk include people with learning disabilities, autistic people, people with limited English, people experiencing homelessness, and those with fluctuating engagement capacity. Services should embed reasonable adjustments into the access process (communication format, advocacy involvement, flexible contact methods) and audit for differential decline rates.

Operational example 3: Equality audit and criteria refinement

Context: Concerns emerge that some groups are disproportionately declined or repeatedly re-referred without resolution.

Support approach: Implement an equality and outcomes audit linked to criteria decisions.

Day-to-day delivery detail:

• Access team records protected characteristics where available and flags communication needs.
• Monthly audit reviews decline rates by demographic group and referral source.
• Sampled cases check whether reasonable adjustments were offered and whether safeguarding indicators were recognised.
• Findings lead to practical changes: updated scripts, revised templates, additional training prompts, and clearer escalation routes.

How effectiveness is evidenced: Improved equity in acceptance/decline patterns, fewer “revolving door” referrals, and clearer documentation supporting defensible decisions.

Governance mechanisms that make criteria defensible

Referral criteria should be treated as a governed clinical policy, not a static document. Strong governance includes:

• Version control and review schedule (at least annually, and after significant incidents).
• Staff supervision that tests decision consistency using real cases.
• Audit of documentation quality and safety-net actions.
• Review of complaints, MP/advocate concerns, and referrer feedback themes.

When criteria are linked to real pathways, backed by safety-netting, and actively governed, they balance access and capacity without compromising safety — and remain credible under scrutiny.

---