The gap is wide, and the law is changing.
We claim only what the evidence supports, and we measure everything else from the first person we support.
Support falls away at the riskiest moments.
Nearly 4×
Black people had nearly four times the recorded detention rate of white people in England in 2024–25.
262.4 detentions per 100,000 Black people, against 65.8 per 100,000 white people.
Source: NHS England, Mental Health Act Statistics, Annual Figures 2024-25(opens in a new tab)2×+
In Greater Manchester, Black people were more than twice as likely as white people to be detained under the Mental Health Act in 2024–25.
216.2 per 100,000 Black people, against 91.7 per 100,000 white people.
Source: Wigan Today, Greater Manchester detention figures(opens in a new tab)74%
of children released from custody in England and Wales reoffended within a year, in the 2022–23 cohort.
Against 37% of adults in the same cohort. Support falls away at exactly the wrong moment.
Source: Sentencing Academy, sentencing children(opens in a new tab)22%
of children in custody were Black, and over a quarter of those remanded.
2024-25.
Source: YJLC, Youth Justice Board statistics 2024-25(opens in a new tab)36.7%
Black children had the highest reoffending rate of any ethnic group.
2024-25.
Source: YJLC, Youth Justice Board statistics 2024-25(opens in a new tab)52%
of children in care had a criminal conviction by 24, against 13% of those never in care.
2022 analysis.
Source: Children's Commissioner, children in care and the justice system(opens in a new tab)Why Ownsay, why now.

A new Mental Health Act
The Mental Health Act 2025 became law on 18 December 2025. Advance choice documents and choosing a nominated person roll out in stages from 2027. New rights only help people who know about them and trust the system enough to use them.
Source: RCPsych, Mental Health Act 2025 receives Royal Assent(opens in a new tab)PCREF is now mandatory
Every NHS mental health trust must show progress on race equality under the Patient and Carer Race Equality Framework, which is part of how the Care Quality Commission assesses them. Trusts need routes to hear from racialised communities.
Source: Mersey Care, PCREF overview(opens in a new tab)Planning ahead works
Crisis-planning interventions lowered the risk of compulsory admission by about a quarter, in a review of five trials involving adults with psychotic illness or bipolar disorder. Peer support evidence is more mixed, which is why we measure from day one.
Source: Molyneaux et al. 2019, BJPsych Open(opens in a new tab)
What should change, and how strong the evidence is.
| Service | Short-term change | Long-term change | Evidence |
|---|---|---|---|
| Peer support and planning ahead | Trust, engagement, plans written and shared | Fewer compulsory admissions; care that follows people’s wishes | Moderate for crisis planning; mixed for peer support |
| Mentoring | Sessions kept, goals set, steps into education or work | Less violence and reoffending; more young people in education or work | Moderate |
| Transition support | A plan in place, contact kept in the first weeks | Fewer returns to custody or hospital; stable housing, education or work | Problem well evidenced; our model still to be tested |
| Community voice | Trusts hear from communities they struggle to reach | Fairer services | Required under PCREF; effect not yet proven |
| Training | Skills and confidence | Better practice; access to work | To be measured |
What our pilot can show
- Whether people from these communities engage with trusted peers.
- Whether plans get written, shared and used.
- Whether safety holds: every concern handled within policy timescales.
- Changes in wellbeing and trust in services, for the people we support.
What it can’t prove yet
With 15 to 20 people and no comparison group, our pilot can’t prove that Ownsay reduces detentions. We will agree a design that can test that claim with our university evaluation partner, before we make it.

Talk to someone who gets it.
Message us on WhatsApp or fill in a short form. A real person from our team replies in working hours.