Managing Performance and Outcomes Across Mental Health System Partnerships
In integrated mental health systems, performance problems rarely appear first as “missed KPIs”. They appear as system friction: referral bounce, inconsistent thresholds, repeated crises, stalled safeguarding actions, and people falling between services. Commissioners therefore want performance management that explains what is happening operationally and shows how issues are corrected and verified. Good providers combine quantitative oversight with file-level evidence trails and joint learning with partners. This article sits within working with ICBs in mental health and connects to mental health service models and pathways, focusing on practical performance and outcomes management across system partnerships.
What “good performance” means to commissioners in system working
ICBs and system commissioners generally want performance management that achieves three things:
- Credibility: reported performance reconciles to operational logs and case records, so sampling matches the dashboard.
- Control of variation: differences across localities, cohorts and teams are identified early and explained with evidence.
- Improvement that sticks: when issues are found, actions are implemented and verified through re-audit or sampling.
Performance reporting that is “busy” but not traceable usually increases scrutiny rather than reducing it.
Designing a performance framework that supports partnership delivery
1) Choose KPIs that reflect pathway reality
KPIs should reflect the service model and interfaces, not just internal activity. Practical measures often include: time-to-first-contact; referral acceptance/decline reasons; escalation rates and timeliness; safeguarding action completion; repeat escalation patterns; restriction duration where relevant; and outcome indicators linked to routine review records. Measures should be defined consistently so all parties interpret them the same way.
2) Build reconciliation and data quality checks into the rhythm
Commissioners will often test whether your numbers are “real” by sampling cases. Providers should run their own reconciliation checks: does the dataset match the records? Are fields completed consistently? Are there duplicate sources (“shadow spreadsheets”) that create inconsistent versions of truth? A small monthly data quality sample often prevents bigger credibility problems later.
3) Use “assurance traces” to link metrics to real practice
Assurance traces are short, anonymised evidence chains that show how the service model works: referral → assessment → plan → escalation → review decision → outcome evidence. These traces help explain performance variation and make contract meetings more constructive because discussions are grounded in reality.
4) Manage variation as a governance issue, not a blame issue
Variation is normal in community delivery, but unmanaged variation suggests weak control. Providers should be able to explain variation (for example, different referral sources, deprivation patterns, or partner access routes) and show what they are doing to reduce avoidable variation (standard templates, supervision prompts, interface improvements).
Operational examples (performance management that commissioners trust)
Example 1: Reducing referral bounce and improving time-to-start
Context: The provider experiences high referral volumes and frequent declines due to incomplete information and scope mismatch. Time-to-first-contact suffers and Trust partners complain about inconsistency.
Support approach: The provider tracks decline reasons using a structured taxonomy and agrees a referral minimum dataset with partners. A weekly short triage alignment call addresses edge cases and updates guidance. The KPI set includes time-to-first-contact and “avoidable decline rate”.
Day-to-day delivery detail: Triage staff record decline reasons consistently. Team leads review the top decline themes weekly and use supervision to correct inconsistent eligibility decisions. Monthly governance reviews trends and agrees actions (referrer briefings, referral form prompts, template adjustments). A small monthly sample checks whether acceptance/decline decisions are documented clearly in the record.
How effectiveness/change is evidenced: Reduced avoidable declines, improved time-to-first-contact, and more consistent decision-making. Evidence includes referral logs, sampled records and trend dashboards that reconcile to operational data.
Example 2: Performance control for repeat escalations and crisis pressure
Context: A cohort generates repeated escalations, increasing system pressure and creating concern about late intervention. The ICB wants evidence that the provider is recognising deterioration early and taking timely action.
Support approach: The provider introduces a repeat-escalation dashboard: number of escalations per person, time between early warning indicator and escalation, and post-escalation plan update completion. A manager review is required for repeat escalations, and learning themes are tracked.
Day-to-day delivery detail: Staff record early warning indicators and actions at each contact. When escalation occurs, the standard summary is used and recorded. Managers review repeat escalations weekly and ensure the plan is updated with learning (contact frequency changes, safeguarding actions, interface escalations where required). Governance reviews cohort trends monthly and verifies changes through sampling.
How effectiveness/change is evidenced: Improved timeliness, fewer late-stage crises, and clearer evidence trails of decision-making and post-crisis planning. Evidence includes escalation logs, sample cases and governance action tracking with verification.
Example 3: Outcomes reporting that does not collapse into “activity reporting”
Context: The ICB requests outcomes evidence, but reporting risks becoming vague narrative or purely activity counts. Commissioners challenge whether outcomes are credible and consistent across teams.
Support approach: The provider defines a small set of outcome domains aligned to the pathway (stability, safety, daily functioning, social connection, self-management) and embeds them into routine review records using observable indicators. A data quality audit checks completion and consistency.
Day-to-day delivery detail: At reviews, staff record baseline and progress using the same indicators, and document the decision logic behind changes. Managers sample reviews monthly to confirm that outcomes are linked to real evidence (notes, safeguarding actions, reduction in escalation, step-down decisions). Governance uses re-audit to verify that improvements in completion are sustained and not a one-off push.
How effectiveness/change is evidenced: More consistent outcomes evidence across teams, reduced reliance on narrative-only claims, and easier commissioner sampling because outcomes are traceable to routine records.
Explicit expectations that must be met
Commissioner expectation
Commissioners expect performance and outcomes to be traceable, comparable and improvement-led. They will look for stable KPI definitions, reconciliation to records, clear explanations of variation, and evidence that issues lead to verified improvement. They also expect providers to measure what matters at interfaces (referral bounce, escalation, safeguarding follow-through), not just internal activity.
Regulator / Inspector expectation (e.g. CQC)
CQC expects governance and performance oversight to translate into safer, more consistent care. Inspectors will test whether performance claims match file evidence and staff practice, whether safeguarding and escalation are timely, and whether learning leads to sustained change. They will triangulate dashboards, staff understanding and record quality.
Keeping performance management proportionate
Performance systems become unsustainable when they add parallel data entry. Mature providers keep the dataset small, embed fields into routine recording, and use sampling to verify quality rather than increasing reporting volume. The result is fewer surprises in contract monitoring because the evidence trail is stable and auditable.