De-escalation

What can be supported about an acute adult response intended to reduce escalating distress and danger, and which findings instead concern a named staff-training course or an organisational package? Schools, youth clinical/residential settings and predominantly adult inpatient wards are considered separately. Immediate incident counts, recipient experience and recovery are different outcomes.

This targeted review includes twelve source records, with the Safewards correction attached to the original trial. They are not twelve independent efficacy studies. Guidance has no comparison group; syntheses may overlap; a published child consultation does not count as Phoenix recipient review. No overall evidence grade is assigned.

• Identity and version: Name the component or programme, its edition or protocol where relevant, and the purpose being evaluated.
• Source role: Separate direct comparative evidence, descriptive implementation findings, guidance, theoretical support and firsthand experience.
• Population and setting: Retain the studied age, communication/access characteristics and setting. State when an intended application lies outside them.
• Intended and actual delivery: Record the planned components and dose separately from what was received, by whom, and with what support or interruptions.
• Comparison and other support: Name the actual comparison and concurrent changes. If there is no comparison group or it is unreported, say so rather than supply an assumed usual-care condition.
• Outcome and reporter: Record the outcome, time point and reporter. Keep the young person’s account separate from proxy judgement, and access or participation separate from task performance.
• Fidelity and adaptation: Record whether the core components were delivered and describe adaptations; distinguish an adapted application from the protocol tested.
• Burden and unwanted effects: Record time, effort, practical access, cost or workload where known, alongside refusal, fatigue, distress, restriction and other unwanted consequences. Unreported burden is not evidence of no burden.

S01 — Department for Education. Restrictive interventions, including use of reasonable force, in schools (April 2026).¹

Identity and version: Effective 1 April 2026; replaces 2013 guidance.

Design and source role: Guidance, not a comparative study.

Population and setting: All schools in England, including maintained nursery schools and pupil referral units; further-education powers are outside its explanatory scope.

Intended and actual delivery: Advice covers prevention, accessible communication, individual planning and support after incidents. No measured intervention dose.

Comparison and other support: No research comparison group.

Outcome and reporter: Significant force and seclusion/non-force restraint have distinct recording/reporting routes; as-soon-as-practicable duties differ from same-day endeavour advice.

Fidelity and adaptation: Local implementation is not measured.

Burden and unwanted effects: The document recognises physical and psychological harm; it supplies no harm-incidence estimate for de-escalation.

Role, access and limits: Official England guidance; not efficacy research. Official full PDF text; substantive sections inspected. Page 3 says page 14 is statutory/rest non-statutory, while page 15 also labels reporting statutory. Preserve that inconsistency; consult the relevant legislation rather than claiming the entire document is statutory or only recording has legal force.

Source locator: pp 3–5,8–17; current GOV.UK listing.

S02 — UK legislation. Schools (Recording and Reporting of Seclusion and Restraint) (No. 2) (England) Regulations 2025, SI 2025/1348.²

Identity and version: Made 17 December 2025; substantive duties commence 1 April 2026; replaces defective SI 2025/1340.

Design and source role: Statutory instrument, not intervention evidence.

Population and setting: Applies to schools in England despite extending to England and Wales.

Intended and actual delivery: Maintained-school duties and corresponding independent/non-maintained-special-school provisions require incident procedures; no clinical dose.

Comparison and other support: Regulation 6 avoids duplication when section 93A use-of-force recording already applies.

Outcome and reporter: Written recording as soon as practicable; parent reporting has age 20+ and significant-harm exceptions with local-authority provision where no parent can safely receive it.

Fidelity and adaptation: Compliance not evaluated here.

Burden and unwanted effects: No effectiveness or safety study.

Role, access and limits: Primary legal source (secondary legislation): scope and reporting duties. Official full eight-page PDF; HTML retrieval failed. Definitions cover non-disciplinary seclusion/restraint, including prevented exit without touch. These regulations do not make every restriction lawful or authorise an individual response. Do not substitute an impact assessment for the instrument.

Source locator: Regulations 1,3–8 and explanatory note.

S03 — National Institute for Health and Care Excellence (NICE). NG10: Violence and aggression—short-term management in mental health, health and community settings.³

Identity and version: NG10; child and young-person provisions retain their specific clinical setting.

Design and source role: Guideline recommendations, not one tested de-escalation package.

Population and setting: Health, mental-health and community services; child/adolescent mental-health staff recommendations are explicit.

Intended and actual delivery: Training, respectful interaction, personal space and a chance to move away; child guidance calls for developmental and communication adaptation.

Comparison and other support: Not an efficacy comparison group.

Outcome and reporter: Intended prevention of escalation and restrictive intervention; no measured outcomes in these recommendation passages.

Fidelity and adaptation: Requires age-appropriate application; an adult protocol is not automatically a child protocol.

Burden and unwanted effects: Care participation and safeguarding remain relevant; the advice is not a hands-on training resource.

Role, access and limits: Official clinical guidance; practice rationale. Current official recommendations HTML inspected. Read as clinical guidance rather than school law. This entry does not reproduce restraint, medication or handling procedures.

Source locator: 1.2.1–1.2.3,1.3.12–1.3.20,1.7.1–1.7.16.

S04 — Hodgkiss, B. & Harding, E. (2023). Reducing physical restraint in educational settings: a systematic literature review. JRSEN 23,265–277.⁴

Identity and version: DOI10.1111/1471-3802.12598; searches April 2021–August 2022.

Design and source role: Review reports 15 studies,1999–2019; methods wording about an additional study is not converted into a new total.

Population and setting: Special-education settings; varied ages and needs.

Intended and actual delivery: School-wide programmes and individual procedures differ; some alter restraint duration rather than preventing its use.

Comparison and other support: Designs/comparisons vary; no pooled generic de-escalation estimate.

Outcome and reporter: Most reports describe reductions, but Table 1 includes a school-wide null and an individual frequency increase under one duration condition. These are review-reported findings, not independently reanalysed primary data.

Fidelity and adaptation: Personalisation, staff development and sustained implementation recur; no universal dose.

Burden and unwanted effects: Recipient views need further study. Restraint-duration procedures are not admitted as practical instructions.

Role, access and limits: School restraint-reduction synthesis; indirect component evidence. Full publisher HTML; methods, Table 1, discussion inspected. Selection is heterogeneous, including dissertations and non-independent longitudinal follow-up. Mainstream transfer is uncertain; the review’s pre 2026 legal discussion is superseded for England. Verret 2019 below overlaps this synthesis.

Source locator: Method; Table 1; Discussion/conclusion.

S05 — Verret, C., Massé, L., Lagacé-Leblanc, J., Delisle, G. & Doyon, J. (2019). The impact of a schoolwide de-escalation intervention plan on the use of seclusion and restraint in a special education school.⁵

Identity and version: Emotional and Behavioural Difficulties 24(4),357–373.

Design and source role: Systematic incident observation across three months within a school year; design details remain unverified.

Population and setting: Specialised primary school for emotional/behavioural difficulties near Montréal. Sample size/ages not imported from another review.

Intended and actual delivery: School-wide plan; actual staff training dose and pupil receipt unavailable in accessed abstract.

Comparison and other support: No comparison group described in accessed abstract; no assumed randomised/control design.

Outcome and reporter: Authors report decreased seclusion/restraint frequency and total duration, and more de-escalation than restrictive measures. Reporter is incident observation; no direct pupil-reported outcome established.

Fidelity and adaptation: Protocol detail and adherence unverified.

Burden and unwanted effects: Harms, workload and acceptability not reported in the accessed abstract.

Role, access and limits: Original school study; limited-access lead. Author-provided abstract via ERIC and ResearchGate; publisher full text failed; full methods not accessed. Do not turn observed change into isolated causal efficacy or assume that self-regulation itself was independently measured. School-specific findings are potentially relevant but cannot fill missing methods. No author contact made.

Source locator: Abstract; bibliographic identity checked against UQAM author listing.

S06 — Sellers, D.E. (2023). Increasing Safety in High Need Schools: An Evaluation of Therapeutic Crisis Intervention for Schools. NIJ-funded final report, NCJ309962.⁶

Identity and version: Therapeutic Crisis Intervention for Schools (TCI-S) organisational package, not generic acute de-escalation.

Design and source role: 19 schools randomised 9 immediate/10 delayed; planned two-year implementation disrupted by COVID.

Population and setting: 14 elementary and 5 pre-K–grade 8 schools in one north-eastern US district.

Intended and actual delivery: Three/four-day staff variants; Table 2 reports training saturation 91%/72% and 47/19 consultant visits across cohorts. These are implementation totals, not proof of full delivery.

Comparison and other support: First-year delayed cohort received no TCI-S activities. Other ordinary school provision remained.

Outcome and reporter: Pre-pandemic 2018–19 first-year staff/student survey analyses found no significant intervention impact. Analytic records: 2,078 staff survey responses,199 matched staff,4,266 student responses—not all distinct people.

Fidelity and adaptation: Training attendance preceded incomplete organisational implementation; later student surveys did not resume.

Burden and unwanted effects: Training, consultation and organisational capacity cost time; no comprehensive adverse-effect assessment established.

Role, access and limits: Original school cluster evaluation plus implementation report. Full official 37-page report; design, delivery, analysis/results inspected. Incomplete first-year organisational delivery and later pandemic disruption of the longer trial restrict inference; a null first-year contrast is neither demonstrated programme success nor proof a fully delivered package never works. Positive staff accounts are not pupil testimony or restraint-reduction results.

Source locator: pp 5–16, Table 2, p 16; pp 25–30; limitations.

S07 — Slaatto, A., Mellblom, A.V., Kleppe, L.C., Baugerud, G.A. & Kjøbli, J. (2021). Conflict prevention, de-escalation and restraint in children/youth inpatient and residential facilities: a systematic mapping review.⁷

Identity and version: Children and Youth Services Review 127,106069.

Design and source role: 14 publications: 2 reviews and 12 individual studies; search 2015–November 2020.

Population and setting: Youth inpatient, residential and some juvenile-justice settings.

Intended and actual delivery: Diverse prevention/training/organisational interventions; no shared dose.

Comparison and other support: Mixed designs, including reviews; counts must not be summed as independent trials.

Outcome and reporter: Review distinguishes reductions in restrictive practices from reductions in aggression/conflict, and identifies evidence gaps in de-escalation.

Fidelity and adaptation: No specific programme fidelity can be inferred for a school from this mapping.

Burden and unwanted effects: Negative youth/staff restraint experiences are reported; they are not uniform reactions to one acute technique.

Role, access and limits: Historical youth-facility synthesis lead freshly checked. Author-uploaded full text returned; abstract and targeted synthesis passages inspected; no independent extraction of every included study. Web abstract search total 4,244 differs from author-text abstract 4,698. Do not use that unstable denominator. This older synthesis cannot establish the current absence of controlled studies.

Source locator: Abstract; review scope; results conclusions.

S08 — Moell, A. et al. (2026). Interventions to prevent and reduce coercive measures in psychiatric and residential care for youth: a mixed-methods systematic review. Lancet Psychiatry 13,699–710.⁸

Identity and version: Online 4 June 2026; August 2026 issue.

Design and source role: 26 studies,13 quantitative/13 qualitative, low-to-moderate risk of bias as assessed by authors; searches 2005–19 February 2025.

Population and setting: Psychiatric, residential or juvenile-justice settings; people under 25.

Intended and actual delivery: Organisational/staff-focused and child-focused interventions distinguished; study-level doses inaccessible here.

Comparison and other support: Heterogeneity prevented meta-analysis; no common comparison group.

Outcome and reporter: Abstract associates organisational/staff approaches with larger coercive-measure reductions and describes relational/collaborative themes. This is not a pooled effect or evidence for all schools.

Fidelity and adaptation: Sustained institutional support highlighted; source-level delivery not independently verified.

Burden and unwanted effects: Qualitative material present; direct/proxy composition of each included sample remains unverified.

Role, access and limits: Current youth-care synthesis; access-limited context. Primary indexed abstract and metadata only; publisher full text retrieval unsuccessful. No ranking of specific programmes, certainty grade, study counts beyond abstract, or independent overlap calculation is imported. The abstract’s stronger concluding language is not adopted as a causal guarantee.

Source locator: PubMed abstract; DOI10.1016/S2215-0366(26)00056-8.

S09 — Nielson, S., Bray, L., Carter, B. & Kiernan, J. (2026). “In the Midst of a Thunderstorm”: Young People’s Experiences of Physical Restraint in Inpatient Mental Health Services in the UK.⁹

Identity and version: International Journal of Mental Health Nursing 35, e70209.

Design and source role: Interpretative phenomenological interviews.

Population and setting: Eight children,5 boys/3 girls,10–13 years, three England inpatient wards; near planned discharge. Diagnoses withheld for anonymity.

Intended and actual delivery: Face-to-face interviews with optional drawing/timeline and parent presence; separate patient and public involvement (PPI) consultation 9 young people+8 relatives is not the interview sample.

Comparison and other support: No intervention or control group.

Outcome and reporter: Children described perceived threat before touch, loss of control, noise, fear and lingering distress; several valued later discussion. Findings concern their own restraint experiences.

Fidelity and adaptation: Accessible methods supported participation, but clinical exclusions and parental consent shape whose experience was heard.

Burden and unwanted effects: Sensitive recollection required support arrangements; no quantitative harm prevalence or comparative de-escalation result.

Role, access and limits: Direct child experience; not an efficacy trial. Full publisher HTML; Methods 4.3–4.6 and Results 5 inspected. Accounts challenge interpreting silence as wellbeing. They do not show how every child experiences an incident or prove that one response would have prevented restraint. This published consultation is not consultation for Phoenix.

Source locator: Recruitment; Results The Gathering, Thunderstorm, Aftermath; PPI Table 1.

S10 — Celofiga, A. et al. (2022). Effectiveness of De-Escalation in Reducing Aggression and Coercion in Acute Psychiatric Units: A Cluster Randomized Study.¹⁰

Identity and version: Frontiers in Psychiatry 13,856153.

Design and source role: Six hospitals/12 wards; hospital randomisation; five-month 2018 baseline and corresponding 2019 intervention period.

Population and setting: 6,401 treatment episodes,5,615 people; mean age about 48–49; table ranges include 14–99, without youth-specific estimates.

Intended and actual delivery: 16-hour verbal/non-verbal course with workshops; all experimental staff teams reportedly attended. Actual patient-level receipt not measured.

Comparison and other support: Usual treatment in controls; annual coercion training in both groups; pre-incident medication data unavailable.

Outcome and reporter: Staff/record outcomes favoured training for aggression and aggression-related restraint; no reduction for non-aggression restraint or mean restraint-episode duration.

Fidelity and adaptation: No inter-rater reliability assessment; training attendance does not demonstrate faithful real-world use.

Burden and unwanted effects: Recipient experience and comprehensive adverse-effect surveillance not established; 16 hours requires staff time.

Role, access and limits: Comparative training evidence in predominantly adult acute wards. Full publisher HTML; Methods, Tables 2/3, Results, limitations inspected. Only 6 clusters, case-mix imbalance and observer bias limit inference. Aggression incidence rate ratio (IRR) upper confidence interval (CI) is .342 in abstract versus .324 in Table 3; disputed interval not promoted. Do not describe this as a child/school trial or say there are no controlled de-escalation studies.

Source locator: Methods; Table 2; Table 3; restraint outcomes; Strengths and Limitations.

S11 — Bowers, L. et al. (2015), with 2016 corrigendum. Reducing conflict and containment rates on acute psychiatric wards: the Safewards cluster randomised controlled trial.¹¹

Identity and version: IJNS 52,1412–1422; correction IJNS 58,102, DOI10.1016/j.ijnurstu.2015.12.006.

Design and source role: 31 wards/15 hospitals: 16 Safewards,15 active control.

Population and setting: Adult acute wards; 564 consenting staff; no patient-level research interviews/outcomes collected.

Intended and actual delivery: Ten-component package; eight-week baseline, implementation and outcome phases; researcher visits 2–3 times/week.

Comparison and other support: Staff physical-health package controls attention/expectancy.

Outcome and reporter: For shifts with events, corrected reductions 15% conflict (interval 5.7–23.7%) and 23.2% containment (9.9–35.5%); no significant difference in zero-event shifts.

Fidelity and adaptation: Observed visible-evidence fidelity 38% versus staff-report 89%; neither alone measures all interactions.

Burden and unwanted effects: Staff workload and over 50% missing outcome-phase shift reports; no direct patient-experience outcome.

Role, access and limits: Adult package cluster trial; corrected outcome record. Full original repository PDF and two-page accepted corrigendum; methods/results/fidelity inspected. Event-count composite includes more than restraint; not individual risk reduction. Methods use Bayesian credible intervals; corrigendum calls them 95% CI. Correction is same trial, not another study; effects do not isolate de-escalation or transport automatically to schools.

Source locator: Sections 1.2–1.8,2.1–2.4; Table 3; corrigendum p 2.

S12 — Price, O. et al. (2024). Development and evaluation of a de-escalation training intervention in adult acute and forensic units: the EDITION study of de-escalation training. HTA 28(3).¹²

Identity and version: DOI10.3310/FGGW6874; ISRCTN12826685.

Design and source role: Uncontrolled 10-ward evaluation; 24 weeks: 8 baseline,8 embedding,8 follow-up.

Population and setting: Adult acute/intensive-care/forensic wards; one withdrew. 283 patient data collections are repeated observations, not 283 unique patients.

Intended and actual delivery: 276 staff trained overall,214 ward staff in de-escalation; multiple professional/environmental/reflection components. Training hours not independently extracted.

Comparison and other support: Own baseline; no concurrent control.

Outcome and reporter: Conflict/containment fell over phases; staff secondary outcomes showed no detectable improvement and patient-secondary changes were not sustained at follow-up.

Fidelity and adaptation: Some components implemented in only 2/10 wards. Primary shift-report completion 62% overall,68% excluding remote collection.

Burden and unwanted effects: Role-play assessment abandoned after only 7 baseline participants; time/capacity burdens. Abstract reports no intervention-related adverse events, not no adverse events of any cause.

Role, access and limits: Adult feasibility evaluation with implementation and patient reports. Full official chapter 4 and relevant chapter 6 passages; abstract/metadata; chapter 3 retrieval failed. Feasibility signal cannot establish causality. Patient experience is measured separately from staff acceptability; the whole intervention is not a school script.

Source locator: Chapter 4 recruitment, adherence, within-group outcomes; chapter 6 burden.

The strongest school claim here is that specified programmes and plans have been studied, with different designs, delivery and outcomes. An uncontrolled school change, a pre-pandemic first-year null cluster contrast with incomplete organisational delivery and a qualitative account cannot be combined into a generic treatment effect. School-specific uncertainty remains even when an adult ward study is favourable.

The adult evidence also resists a single summary. Safewards compares packages and measures shifts; Celofiga evaluates a course across a small number of hospital clusters; EDITION is uncontrolled and measures partly different components and recipient outcomes. The inclusion of some adolescents in a predominantly adult sample does not create a child estimate. Training reported as delivered does not establish what each patient actually received.

The 2023 school review includes Verret 2019. The two youth syntheses cover potentially overlapping literature, and their study totals are not additive. A corrigendum updates one trial. None of these overlaps is counted as replication. Study-level overlap between the inaccessible 2026 review and earlier youth reviews remains unresolved.

Direct child accounts are particularly relevant to defining acceptable outcomes and asking about the lead-up and aftermath. They do not supply a prevalence estimate or establish counterfactual prevention. Proxy judgements, professional acceptability and the young person’s own account remain distinct.

Verret 2019 and Moell 2026 remain abstract-only for purposes of this profile. Search results offering unrelated or unverified copies were not used to invent methods; no paywall bypass or author contact occurred. Department for Education (DfE)’s internal statutory-page description, Celofiga’s abstract/Table 3 interval and Slaatto’s search-total mismatch are recorded explicitly. They are not silently harmonised.

Further review should prioritise full school methods, the 2026 synthesis, actual child receipt and experience, setting-specific safety governance, and the practical burden of sustaining adult changes. There is no validated generic script, fixed dose or universal recovery timetable established here. Human specialist/recipient review and adoption remain pending.

Read the complete A028 working article and claim-by-claim ledger (internal editorial record). Encyclopaedia direct 01/06 and conditional 03/05 are content decisions; all three reading pairs are documented in paired-text reviews (internal editorial record).

1. 1.

   Department for Education. Restrictive interventions, including use of reasonable force, in schools (April 2026).

   ↩1
2. 2.

   UK legislation. Schools (Recording and Reporting of Seclusion and Restraint) (No. 2) (England) Regulations 2025, SI 2025/1348.

   ↩1
3. 3.

   NICE. NG10: Violence and aggression—short-term management in mental health, health and community settings.

   ↩1
4. 4.

   Hodgkiss, B. & Harding, E. (2023). Reducing physical restraint in educational settings: a systematic literature review. JRSEN 23,265–277.

   ↩1
5. 5.

   Verret, C., Massé, L., Lagacé-Leblanc, J., Delisle, G. & Doyon, J. (2019). The impact of a schoolwide de-escalation intervention plan on the use of seclusion and restraint in a special education school.; accessed author-provided abstract

   ↩1
6. 6.

   Sellers, D.E. (2023). Increasing Safety in High Need Schools: An Evaluation of Therapeutic Crisis Intervention for Schools. NIJ-funded final report, NCJ309962.

   ↩1
7. 7.

   Slaatto, A., Mellblom, A.V., Kleppe, L.C., Baugerud, G.A. & Kjøbli, J. (2021). Conflict prevention, de-escalation and restraint in children/youth inpatient and residential facilities: a systematic mapping review.; author-uploaded access copy

   ↩1
8. 8.

   Moell, A. et al. (2026). Interventions to prevent and reduce coercive measures in psychiatric and residential care for youth: a mixed-methods systematic review. Lancet Psychiatry 13,699–710.

   ↩1
9. 9.

   Nielson, S., Bray, L., Carter, B. & Kiernan, J. (2026). “In the Midst of a Thunderstorm”: Young People’s Experiences of Physical Restraint in Inpatient Mental Health Services in the UK.

   ↩1
10. 10.

    Celofiga, A. et al. (2022). Effectiveness of De-Escalation in Reducing Aggression and Coercion in Acute Psychiatric Units: A Cluster Randomized Study.

    ↩1
11. 11.

    Bowers, L. et al. (2015), with 2016 corrigendum. Reducing conflict and containment rates on acute psychiatric wards: the Safewards cluster randomised controlled trial.; 2016 corrigendum

    ↩1
12. 12.

    Price, O. et al. (2024). Development and evaluation of a de-escalation training intervention in adult acute and forensic units: EDITION. HTA 28(3).

    ↩1