Restraint and seclusion: two names

Twenty-five minutes face-down on the floor

David "Rocky" Bennett was held face-down on the floor by nursing staff for about 25 minutes, from 22:58 to 23:25 on 30 October 1998, and was pronounced dead at 00:20 the following morning. Five staff held him for the first two or three minutes, then never more than four — nobody at his head.

The independent inquiry chaired by Sir John Blofeld concluded that institutional racism was present in the NHS, and recommended, at recommendation 9 of 22: "Under no circumstances should any patient be restrained in a prone position for a longer period than three minutes."

Neither was adopted. The Secretary of State for Health declined to make the acknowledgement of institutional racism the inquiry called for. And NICE's 2015 guideline NG10 set a ten-minute benchmark for manual restraint — not Blofeld's three-minute limit on prone restraint.

The inquiry sat under HSG(94)27, with no power to compel evidence, and was not the first: a 1993 inquiry into three deaths at Broadmoor had already made findings the Blofeld panel called "disturbing".

[7] Independent Inquiry into the death of David Bennett, HSG(94)27, Norfolk, Suffolk and Cambridgeshire SHA, December 2003, recommendation 9 · [8] NICE NG10 (2015)

Seven years to the conclusion, eight to the Act

Olaseni "Seni" Lewis, 23, was restrained at the Bethlem Royal Hospital by Metropolitan Police officers — not by nursing staff, with hospital staff present. He collapsed, was transferred to another hospital, and died days later.

On 9 May 2017, seven years after the investigation opened, the inquest concluded that he "died from hypoxic brain injury caused by restraint in association with acute behavioural disturbance". The jury returned a narrative conclusion: the restraint was prolonged, disproportionate and unreasonable, the Trust had failed to meet training targets, and police had failed to follow their training.

The Mental Health Units (Use of Force) Act 2018, known as Seni's Law, followed. It prohibits no restraint technique and imposes no time limit. It requires policies, training, recording and reporting. Twenty years after Bennett, the three-minute limit is still not a rule.

[9] Regulation 28 Report to Prevent Future Deaths, South London Coroner's Court, Senior Coroner Selena Lynch, 28.06.2017, ref 2017-0205 · [10] Mental Health Units (Use of Force) Act 2018 c.27

The number "142" circulates as a statistic. It is not.

The figure of 142 restraint- and seclusion-related deaths over ten years circulates everywhere as data. It is a newspaper's own tally, compiled by the Hartford Courant for its "Deadly Restraint" series of 11–15 October 1998. It was never an official statistic and has never been independently audited.

When the General Accounting Office examined the question the following year, it did not attempt to verify the count. It reported instead that no comprehensive federal reporting system then existed, and that from the fragmentary information available it could identify at least 24 such deaths in fiscal year 1998 — a floor, not a total, with the note that "many more deaths related to restraint or seclusion may occur".

That gap has since been closed. Federal rules made in 1999, now at 42 CFR 482.13(g), require hospitals to report restraint-related deaths. Presenting 1998 as the present state would be inaccurate.

[11] US General Accounting Office, "Mental Health: Improper Restraint or Seclusion Use Places People at Risk", GAO/HEHS-99-176, September 1999, p.1 and footnote 2 · [12] 42 CFR 482.13(g)

Beyond half an hour in restraints, a judge is required

On 24 July 2018 Germany's Federal Constitutional Court ruled on two joined complaints. One complainant had been strapped down repeatedly over several days on a doctor's order alone; the other spent eight hours in restraints — both arms, both legs, abdomen, chest and forehead — during a hospital stay of roughly twelve hours.

The Court set out the rule in the judgment's own headnote: "Both a five-point and a seven-point fixation of more than brief duration constitute a deprivation of liberty within the meaning of Article 104(2) of the Basic Law, which is not covered by a judicial committal order". And it fixed the threshold numerically: a measure counts as brief where it is foreseeably expected to last less than about half an hour.

The consequence is procedural and concrete. The committal order is not enough; the restraint needs its own judicial authorisation. And because that would be a dead letter without a judge available, the Court required a daily judicial standby service covering 6:00 to 21:00.

The Court did not prohibit restraint. It kept it lawful, on conditions: prior judicial authorisation, continuous one-to-one supervision, mandatory documentation, and a right to later review. It held the statutes of Baden-Württemberg and Bavaria unconstitutional because they made no provision for the judge, and set a compliance deadline of 30 June 2019.

[67] BVerfG, Second Senate judgment of 24.07.2018, 2 BvR 309/15 and 2 BvR 502/16 — headnotes 1(b) and 3

An injection for acute agitation: 2,754 children's admissions, and which children

Pharmacologic restraint — the study's name for an injection of a short-acting antipsychotic given to a child in acute agitation, when de-escalation fails and there is an imminent risk of harm; the database does not show the indication of each dose — was measured on a national scale in the United States in a study published in December 2023: 91,898 admissions of children aged 5–17 with a psychiatric diagnosis, in 43 tertiary children's hospitals, 2016–2021, on paediatric wards rather than psychiatric units. The researchers' definition was deliberately narrow: only parenteral (intramuscular or intravenous) short-acting antipsychotics — aripiprazole, droperidol, haloperidol, olanzapine, ziprasidone; no tablets, no benzodiazepines, no antihistamines. Finding: 2,754 admissions, 3%, and 1.3% of patient days. The highest rate was among children with autism — 79.4 days with restraint per thousand patient days — and the greatest exposure in long stays: among restraints, 25.7% involved stays of seven days or more against 12.0% in the others, which the authors link to children boarding on paediatric wards for lack of psychiatric beds.

The number in circulation, and what the paper itself says. The finding spread as “a 141% rise in the chemical restraint of children”. The number is true and misleading at the same time: restraint days rose by 141% between 2016 and 2021 because children's mental-health patient days rose by 138%. The overall adjusted rate — days with restraint per thousand patient days — went from 16.2 to 19.7 and the change was not statistically significant (RR 1.2; 95% CI 0.7–2.1; p=0.47). By diagnosis, however, the picture is not flat: the rate rose significantly in four categories — somatic symptom disorders (RR 4.2), eating disorders (2.4), bipolar disorders (2.0), disruptive disorders (1.4) — and fell in two, autism (0.8) and anxiety (0.3). More children entered the hospitals; the practice changed by diagnosis, not overall.

The finding that stands is a different one: which children. Among admissions with pharmacologic restraint, Black children were 29.8% against 17.4% in the others, children on public insurance 61.9% against 46.8%, boys 56.4% against 29.5%, children aged 5–8 14.1% against 5.9% (all p<0.001). Those are unadjusted comparisons, and the authors attribute the sex and age differences in part to the diagnoses that cluster in young boys. The racial difference, however, persists after adjustment: a study in the same issue, with a broader definition (any parenteral drug for acute agitation) across 61,503 admissions in 2018–2022 in 41 hospitals, measured the difference after adjustment for age, sex, insurance, diagnosis and severity: compared with Black children, White children had an adjusted odds ratio of 0.81 (0.72–0.92), Asian children 0.82, others 0.68; among boys the White–Black difference was 0.68, while among girls and among Hispanic children it was not statistically significant.

The authors state the limits themselves: the database does not show the indication for a drug or whether the injection “worked”; the rates are probably underestimated; physical restraint was not measured; only large tertiary hospitals. The first study had no external funding; the second and the commentary carry NIH/AHRQ grants. The federal conditions of participation for hospitals in Medicare have defined since 2007 that a restraint also includes “any drug or medication when it is used as a restriction to manage the individual's behavior … and is not a standard treatment or dosage for their condition” — wording the accreditation body repeated in 2024 for behavioural health settings, effective 2025. Whether every injection in the study met that test, the data do not show.

What this exhibit does not say: that every injection given to a child in crisis is mistreatment — there are situations of immediate danger. It says that the race of the child who receives the injection is not explained by the child's diagnosis.

What children are prescribed outside a crisis — 81.7% of antipsychotics off-label, and the Irish review of the 227 children in Kerry — is on the sister site: Part D of psychdrugs.org.

[74] Masserano B et al. Pediatrics 2024;153(1):e2023062784 · [75] Wolf RM et al. Pediatrics 2024;153(1):e2023061353 · [76] Dalton EM, Doupnik SK. Pediatrics 2024;153(1):e2023064054 · [77] The Joint Commission, R3 Report 44, 20.06.2024 · [78] 42 CFR §482.13(e)(1)(i)(B)

Three state inspectors, 2022–2026: 1,050 days of seclusion, 76 days in straps

The cases above are British, and in the past. Whether restraint is controlled today, where a law exists for it, is answered by the state inspectors of three countries — and all three reports are public.

Ireland. The Mental Health Commission, the regulator of psychiatric facilities, records in its 2025 annual report (26 June 2026) 758 episodes of seclusion involving 454 people in 26 approved centres, 1.88% more than in 2024; the shortest lasted five minutes, the longest 25,204 hours — 1,050 days, in the National Forensic Mental Health Service, from November 2022 to September 2025. Physical restraint: 1,851 episodes involving 785 people in 52 centres, 11.5% fewer than in 2024. Compliance with the rule on seclusion rose from 28.57% in 2024 to 33.33% — it "remains low", the report itself writes. Six admissions of children to five adult units. The previous report, for 2024, counted 744 seclusions with a longest episode of 124 days — the 1,050-day seclusion did not appear there, because it had not ended — and the Inspector of Mental Health Services, Professor Jim Lucey, wrote that in a minority of centres "breaches at a critical level persist despite assurances given year-on-year"; in 2025 critical non-compliances fell, and we pass that on.

Norway. The Parliamentary Ombud (Sivilombudet) examined in 2022 the 59 decisions on appeals against mechanical restraint — straps — sent to it, for 2021, by 22 of the 54 control commissions, and found that in 28 of the 59 it does not appear that the commission considered whether the restraint was "unavoidably necessary" for its whole duration — the condition the law sets. Duration in the decisions: "from 6 minutes to 76 days"; 16 decisions concerned more than 24 hours. In June 2026, after a visit to the Nordland hospital, the same institution recorded two patients in straps for 34 and 38 hours without a documented reassessment, 39 of 42 decisions taken by trainee doctors or a student instead of the responsible specialist, and spit hoods in the room with the restraint bed — "unlawful means of coercion", which "are not permitted in mental health care". The scale of the system, from the 2019 law commission, with 2017 data: about 5,500 people in involuntary care (7,700 admissions), and nearly 2,200 with a decision on coercive means — for half of them, only brief manual holding.

Ontario, Canada. The provincial Ombudsman, Paul Dubé, in the report "Lost in Transition" of 25 November 2025, on people with developmental disabilities who stayed in hospitals because there was nowhere else to go: one man, "Sean", remained in a psychiatric hospital for more than five years, with mechanical restraints "at times up to 16–20 hours a day"; a young man, "Noah", was in hospital for almost two and a half years from August 2021 and from his eighth month was "tethered to his bed through all of his waking hours, in a spreadeagle position using either three or four-point restraints", and chemically restrained on top. The ministries accepted all 24 recommendations.

What they show together, and what they do not. They are not death statistics, and they cannot be compared with each other — different definition, different law, different method of recording. They show something narrower and more consistent: where restraint has a rule and an inspector, the inspector finds the rule not being kept — in seven out of ten Irish centres to which it applies, in 28 of 59 Norwegian decisions, in people who stayed in hospital because there was no home and not because they needed a hospital.

[86] Mental Health Commission (Ireland), Annual Report 2024 and Report of the Inspector, 27.06.2025 · [111] Mental Health Commission, Annual Report 2025, 26.06.2026 · [87] Sivilombudet, report on the control of mechanical restraint, 14.11.2022 · [88] Sivilombudet, visit report Nordlandssykehuset, 17.06.2026 · [89] NOU 2019:14, ch. 9 · [90] Ombudsman Ontario, "Lost in Transition", 25.11.2025