A hold that was banned after he died

A 15-year-old died in a seated restraint at a secure training centre. The technique was suspended afterwards, and two-thirds of the centre's restraints that year were judged potentially lethal.

Youth justice · 2000s · Method: Seated

What happened

On 19 April 2004 Gareth Myatt, 15, was restrained on the ground at Rainsbrook Secure Training Centre in Northamptonshire. The technique used was a seated double embrace. He lost consciousness during it and died.

He had been at the centre for four months.

The outcome

The Crown Prosecution Service decided in December 2005 that no prosecution would follow. An inquest was held afterwards; the sources held for this case do not carry its conclusion, and none is stated here.

An independent investigation found that two-thirds of the restraints used at the centre in the year before his death had been potentially lethal. The technique was suspended after he died.

What went wrong

The finding that matters is the ratio. If two-thirds of the restraints at one centre in one year were potentially lethal, the problem was not a decision taken by one member of staff on one afternoon. It was a technique in routine use, by trained staff, inside a system that had not asked whether it could kill.

The suspension came afterwards. Nothing in the record shows the same test being applied before.

What it teaches

A hold can be in the policy, in the training, and in everyday use, and still be capable of killing the person it is applied to. Lawfulness, approval and familiarity are not the same as safety.

Seated and floor restraints are where the positional risk lives, because the person’s weight and the staff’s weight combine against a chest that has to work harder to breathe. The test is not whether the hold is authorised. It is whether anyone is watching the breathing, and whether the hold has an endpoint other than the person going quiet.

The secondary lesson is about review. A fatal case should trigger an audit of every restraint in the period before it, not a suspension of the one technique in the newspaper. At Rainsbrook, the audit that mattered was the one nobody had run.

Questions for the room

Evidence and limits

Confidence: confirmed on the record, but this page rests on thinner sources than its neighbours, and that is worth saying.

There is no page for this case in the case precis pack. The account here is drawn from the vault record, whose sources are INQUEST’s release on the decision not to prosecute, a Hansard debate in July 2007, and one archive clipping. Those were not re-read for this entry.

Two things follow. The “two-thirds” figure is as the independent investigation recorded it, and its basis is not visible in the sources held. And no inquest conclusion appears in them, so the page makes no finding about how he died beyond the fact of the restraint and the death. If more becomes available, this paragraph and the entry should be revised.

Our restraint reduction training covers positional risk, duration limits and the monitoring that has to go with any hold.

Sources

Reviewed 20 October 2026.

All youth justice cases  ·  Restraint Casebook

Book in a 1-2-1 Meeting

Find Your Solution

What do you need? Use our simple tool to find the most relevant training and support options for your role, sector and challenges.