Components of interventions to reduce restrictive practices with children and young people in institutional settings: the Contrast systematic mapping review

Article history

The research reported in this issue of the journal was funded by the HSDR programme or one of its preceding programmes as project number NIHR127281. The contractual start date was in July 2019. The final report began editorial review in October 2020 and was accepted for publication in June 2021. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HSDR editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Copyright statement

Copyright © 2022 Baker et al. This work was produced by Baker et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaption in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library, and the DOI of the publication must be cited.

2022 Baker et al.

Restrictive practices in children and young people’s institutional settings

There are approximately 80,000 CYP living in state and privately run institutions in England alone;1 such institutions include residential children’s homes, residential schools, young offender institutions, secure training centres, secure children’s homes and immigration detention centres, in addition to NHS inpatient settings through Child and Adolescent Mental Health Services (approximately 1140 beds). 2 The CYP in these disparate institutional settings share some characteristics: many have experienced trauma, abuse and loss;3–8 some present serious risks of harm to themselves and/or others;9,10 and some exhibit behavioural and/or psychological difficulties. The health and safety of these CYP and the staff who work with them hinges on the safe and effective avoidance and management of incidents involving violence, aggression and self-harm.

Definition of restrictive practices, rates of use

Staff responses to incidents involving violence, aggression or self-harm may involve the use of potentially harmful restrictive practices (defined by the Department of Health and Social Care as ‘deliberate acts on the part of other person(s) that restrict an individual’s movement, liberty and/or freedom to act independently in order to: take immediate control of a dangerous situation where there is a real possibility of harm to the person or others if no action is undertaken’11,12). Restrictive practices, such as restraint, seclusion and (in health settings) the use of forced medication, are a common occurrence. Rates of use are similar in psychiatric and criminal justice settings. One study calculated that one-quarter of CYP treated in psychiatric settings have had at least one seclusion episode and 29% have had at least one restraint episode,13 whereas the level is estimated at 28% (in 2014) in custodial settings. The rate is estimated to be higher in learning disability services, with more than half of CYP experiencing seclusion, restraint or a harmful incident. 14

Prevalence data are not available from other settings, although individual cases have attracted media attention. 15 Some studies have found that approximately 60–70% of all reported seclusions or restraints in CAMHS can be accounted for by a small minority (7–15%) of all hospitalised CYP. 16,17 Recent UK figures revealed that 17% of girls in CAMHS facilities had been physically restrained, compared with 13% of boys. 18 Face-down restraint was more commonly used on individuals < 18 years old, with > 2500 occurrences in 2014/15, and, again, in particular with girls (> 2300 occurrences), often repeatedly with the same girls. 18

Risks and costs: physical, psychological, financial

Restrictive practices carry high risks of physical and psychological harm. In the UK, 45 CYP died in restraint-related circumstances in inpatient psychiatric facilities in the period 1993–200313,19,20 and two have died in youth custody in the past 15 years. 21,22 In 2015 alone, there were 429 injuries to children resulting from restraint in youth custody. 23

Research has described the negative impact of experiencing restrictive practices in adult service users but little is known about CYP’s experiences. 21,22 It is thought that, as with adults, such practices can have a profoundly detrimental effect on therapeutic relationships between care staff and CYP24 and they are particularly counter-therapeutic for CYP with an abuse history. 25 The subsequent costs of restrictive practices to the NHS are substantial [estimated by the National Institute for Health and Care Excellence (NICE) at £20.5M per year for damage and injury, £88M per year for observations and £6.1M for restraint]. 26 Evidence-based interventions to reduce the use of restrictive practices clearly have the potential to result in significant cost savings.

Concerns regarding overuse

The problem of overuse of restrictive practices within UK state-provided children’s services has become a matter for serious concern. The United Nations raised specific concerns regarding the use of restrictive practices with CYP who have psychosocial disorders, and at the end of 2017 called again for the UK to end all use of restraint in the context of disability, segregation and isolation practices, and any practices that might be considered to be torture or degrading treatment (Section 73a). 27 Voluntary organisations such as Mind (London, UK),24 Article 39 (Nottingham, UK)1 and Agenda (London, UK)18 have ongoing campaigns on the issue.

Legislative frameworks for the use of restrictive practices

Legal provision for restrictive practices varies across these settings. For example, in the UK, pain-inducing restraint techniques remain lawful in Ministry of Justice settings but have been made unlawful by the Departments of Education and Health. Nevertheless, the UK government has sought to reduce restrictive practices across all settings. The Ministry of Justice implemented the Minimising and Managing Physical Restraint (MMPR) programme, although this has been criticised on the grounds that the restraint techniques it authorises are life-threatening. 28 In 2014, the Department of Health and Social Care launched the Positive and Pro-active Care guidance,12 aimed at phasing out face-down restraint and deeming restrictive interventions a ‘last resort’ across health and social care provider organisations. Since then, services’ use of restraint has been subject to inspection by the Care Quality Commission. 29

More recently, in its publication Reducing the Need for Restraint and Restrictive Intervention with Children and Young People with Learning Disabilities, Autistic Spectrum Disorder and Mental Health Difficulties,30 the Department of Health and Social Care and Department for Education has set out core principles for the use of restraint: it should be used only where necessary to prevent risk of serious harm, and not as punishment; with the minimum force necessary; by appropriately trained staff; and should be documented, monitored and reviewed. 30 In 2018, The Mental Health Units (Use of Force) Bill31 – which sought to manage the use of force in mental health services in England and Wales, requiring commitment to a reduction in the use of force and reporting on its use – became law.

Strategies to address reduction of restrictive practices

There is a growing body of research into the reduction of restrictive practices. In the UK, initiatives to reduce restrictive practices in mental health care such as ‘Safewards’,32 ‘Six Core Strategies’33 (6CS) and ‘No Force First’34 have been promoted and adopted by some mental health trusts, including in CAMHS; some of these initiatives have been evaluated and reported in the literature. 32,35 There has been similar research carried out seeking to reduce the use of restraint with people with learning disabilities. 36,37 These have typically aimed to reduce violent and aggressive behaviour by changing staff behaviour to encourage use of de-escalation techniques, supported by various policy and procedural changes.

There is some evidence of interventions that are effective in reducing the use of restrictive measures specifically with CYP in mental health services; however, empirical data are limited13,19,38–41 and often primarily use case studies of single-facility initiatives. 19,42 Although the outcomes of some of these interventions have been the subject of systematic reviews,32 their specific content has not been examined in detail and the causal mechanisms through which they might change behaviour are not fully understood. It remains unclear which components of these interventions have contributed to their effectiveness. Furthermore, it is not known to what extent those interventions that have resulted in reductions in the use of restrictive practices (or other outcomes such as increased staff confidence) have features in common.

The research context for the current study

The existing literature on restrictive practices repeatedly calls for guidance to be based on robust, transparent studies,43,44 and for interventions to be better described and better evaluated. Livingston et al. 45 reviewed training interventions to reduce restrictive practices and highlighted the difficulty of reaching conclusions, as the evaluated interventions comprised ‘different types of aggression management programs, which contain a variety of approaches’ and ‘the focus, curriculum, and duration of the training vary substantially from one program to another’. 45 The NICE guideline on violence and aggression46 calls for research to be carried out into the content and nature of effective de-escalation techniques, together with the most effective and efficient approaches to training professionals in their use. 46 According to the guideline, research is needed that will apply a systematic approach to the description and reporting of de-escalation techniques currently in use. 46 With specific reference to CYP, it notes the ‘lack of research on the nature and efficacy of verbal and non-verbal de-escalation of seriously agitated children and young people with mental health problems’ and recommends research to ‘systematically describe expert practice in adults, develop and test those techniques in aroused children and young people with mental health problems, and develop and test different methods of training staff’46 (NICE. Violence and Aggression: Short-term Management in Mental Health, Health and Community Settings. London: NICE; 2015. © NICE 2015 Violence and Aggression: Short-term Management in Mental Health, Health and Community Settings. Available from www.nice.org.uk/guidance/ng10. All rights reserved. Subject to Notice of Rights. NICE guidance is prepared for the National Health Service in England. All NICE guidance is subject to regular review and may be updated or withdrawn. NICE accepts no responsibility for the use of its content in this product/publication).

The current study is one of a pair addressing NICE’s recommendation to systematically describe restrictive practices with adults and children. The research team’s original study, COMPARE (HSDR 16/53/17),47 fulfilled the first part of NICE’s recommendation by systematically describing practice with adults in mental health inpatient settings. The current review, CONTRAST, is a companion study that took the same approach but reviewed the evidence for interventions to reduce staff use of restrictive practices in child and adolescent institutional settings, including, but not limited to, mental health contexts. It was anticipated that the features of an intervention (its content and delivery) were likely to interact with the delivery context (the target population and setting) and with the features of the target behaviour. 48 Although the target behaviour (use of restrictive practices) was the same as the original study, the context shifted to a range of institutions caring for children and adolescents with different physical, psychological and developmental abilities, employing a wide range of professions, and in which the legality and guidelines for the use of restrictive practices vary. The intention was to compare interventions across these settings to permit exploration of the relationship between intervention features (content and delivery) and context (target population and setting), together with the identification of differences in content, influences on delivery and potential implications for effectiveness.

Addressing the limitations of the evidence base using the behavior change technique taxonomy

The reporting of non-pharmacological trials is challenging because of the absence of a common language with which to describe their components. 49,50 A review51 found that only 39% of interventions were ‘adequately’ described when published. In response to this lack of consensus, the Medical Research Council (MRC) supported the development of a taxonomy of BCTs that can be used across all theory-based interventions aimed at patients or professionals,52 both prospectively in their design and/or to synthesise evidence restrospectively. 52,53 A BCT is defined as ‘an observable, replicable, and irreducible component of a programme designed to alter or redirect causal processes that regulate behaviour’. 52 All interventions to reduce restrictive practices use BCTs. For example, role-playing verbal de-escalation strategies could be coded as rehearsal of relevant skills involving social comparison (BCT 6.2), monitoring of emotional consequences (BCT 5.4) and feedback on behaviour (BCT 2.2). Delivery of information by an expert about risks of restraint could involve information about health consequences delivered by a credible source (BCT 5.1). The taxonomy therefore enables reliable, precise and transparent reporting, replication and comparison of interventions,54 along with more successful implementation with proven effectiveness. 52 It is increasingly used internationally to report interventions,55 synthesise evidence56,57 and reanalyse existing interventions to explore their components. 58 It is also influencing intervention design48 and contributing to the identification of potentially effective BCTs. 52

Aim

The study aim was to identify, standardise and report the effectiveness of components of interventions that seek to reduce restrictive practices in CYP’s institutional settings using the behaviour change taxonomy.

Objectives

The study objectives were to:

• provide an overview of interventions aimed at reducing restrictive practices with CYP

• classify components of those interventions in terms of BCTs and determine their frequency of use

• identify the role of process elements in intervention delivery

• explore the evidence of effectiveness by examining BCTs and intervention outcomes, where possible

• compare the components of interventions in CYP’s settings with those in adult psychiatric inpatient settings47 and identify potential explanations for any differences

• identify and prioritise BCTs that show the most promise of effectiveness and that require testing in future high-quality evaluations.

Design overview

Design and conceptual framework

The study approach was a systematic mapping review. An ‘intervention’ was any documented approach that sought to reduce the use of restrictive practices through BCTs. The literature review focused on ascertaining the range and characteristics of interventions, irrespective of evidence of effectiveness, which involved systematically searching and reviewing all reports of interventions seeking to reduce the use of restrictive practices (Figure 1).

FIGURE 1.

Design of systematic mapping review.

The study design comprised the following six objectives.

Literature search strategy

Data management and review

All potentially eligible records were stored and managed in the reference management software EndNote™ version X9 (Clarivate Analytics, Philadelphia, PA, USA). Two reviewers screened titles. When both reviewers agreed to exclude an article, the reason for exclusion was recorded. When there was disagreement, the full text of the articles was reviewed and any unresolved disagreement was subject to third-party review. When there was agreement between the two reviewers on inclusion, the full-text article was retrieved and independently assessed against the inclusion criteria by the two reviewers and, again, any disagreement was subject to third-party review.

Quality appraisal using the Mixed Methods Appraisal Tool

Because the inclusive search criteria identified very diverse record formats, quality appraisals were used not to exclude papers but to inform the synthesis by identifying study designs and, hence, evaluations. Study quality was assessed using the MMAT. This tool is suitable for appraising studies with diverse designs. 70,71 The characteristics of the MMAT70,71 make it the most suitable tool with which to judge study quality in the context of wide-ranging research methods.

The MMAT was developed for use in complex systematic literature reviews that include quantitative, qualitative and mixed-methods studies (Figure 2). It was developed from theory and a literature review and has been found to have good validity. 76 The MMAT algorithm for selecting study categories is illustrated in Figure 3. Using the MMAT algorithm, reports of milieu change and case studies were categorised as qualitative studies if reporting suggested a primarily qualitative approach.

FIGURE 2.

Mixed Methods Appraisal Tool, version 2018. Adapted from Hong Qn, Pluye P, Fàbregues S, Bartlett G, Boardman F, Cargo M, et al. Mixed Methods Appraisal Tool (MMAT), version 2018. 70 Registration of Copyright (#1148552) Canadian Intellectual Property Office, Industry Canada.

Adapted from Hong Qn, Pluye P, Fàbregues S, Bartlett G, Boardman F, Cargo M, et al. Mixed Methods Appraisal Tool (MMAT), version 2018.65 Registration of Copyright (#1148552) Canadian Intellectual Property Office, Industry Canada).

FIGURE 3.

Mixed Methods Appraisal Tool algorithm for selecting study categories. Adapted with permission from Hong Qn, Pluye P, Fàbregues S, Bartlett G, Boardman F, Cargo M, et al. Mixed Methods Appraisal Tool (MMAT), version 2018. 70 Registration of Copyright (#1148552) Canadian Intellectual Property Office, Industry Canada.

Adapted with permission from Hong Qn, Pluye P, Fàbregues S, Bartlett G, Boardman F, Cargo M, et al. Mixed Methods Appraisal Tool (MMAT), version 2018.65 Registration of Copyright (#1148552) Canadian Intellectual Property Office, Industry Canada).

Using the MMAT, quantitative and qualitative studies are judged against four criteria and mixed-methods studies are judged against three. The quantitative domain is split into three subdomains: randomised controlled, non-randomised and descriptive. As applied in the current study, surveys, case reports, descriptive cross-sectional studies and ecological studies were categorised as ‘quantitative descriptive’ if reporting suggested a primarily quantitative approach. The mixed-methods category included reports of milieu change with substantial quantitative analysis.

Therefore, the tool was used at two levels: (1) to identify records of interventions that had been evaluated to get a sense of the quality of the evidence using the two initial screening questions, and (2) to assess the quality of the evaluation reports. The application of the MMAT to screen and categorise all the records informed the narrative accounts provided in Chapters 4 and 5.

Documented interventions were identified. Data extraction was governed by a pro forma that allowed systematic collection of data relating to the interventions. Analysis of the features of interventions revealed the context of how interventions were delivered (e.g. delivery to groups or individuals; the person delivering the intervention; and the setting, timing and frequency of the interventions). 49 When available, these details were recorded using the WIDER69 checklist. WIDER is a tool for assessing reporting quality. It contains a number of relevant categories that facilitated this process.

Content extraction

The content of interventions was extracted to allow their components to be coded using the BCT taxonomy. Extraction was carried out by two reviewers and any discrepancies were subject to third-party review. Extraction categories were developed from the WIDER69 checklist. Data were extracted about the characteristics of each intervention, including participants, setting, intervention type, outcome measures, fidelity, acceptability, recommendations and quality. When information was available, associated costs were described in terms of training materials, delivery and staff time.

Intervention coding

The researchers were fully trained in the application of the BCT taxonomy. Using the taxonomy and supporting examples, the researchers independently coded the selected interventions. Interventions that were coded for BCT components had information about their outcomes extracted to examine the efficacy of these techniques.

Coding was carried out by importing all intervention materials (published papers, manuals, slides, handbooks) into NVivo 12 (QSR International, Warrington, UK), a flexible qualitative software package that facilitates the coding of multimedia materials for analysis. Each of the 93 items of the BCT taxonomy was turned into a code within NVivo and considered for each intervention. The codes were applied when there was evidence of the BCT being used; for example, when a professional received information about the potentially harmful effects of restraint during a training session, this was coded as BCT 5.1 (information about health consequences). Any assumptions made by the coder were recorded, also within NVivo, in order that discrepancies could be discussed. Once the coding was complete, NVivo was used to generate individual study reports that revealed discrepancies between coders. Each discrepancy was discussed and resolved by the coders and, if necessary, further discussion took place with other expert members of the team to achieve resolution. This discussion consisted of the coder explaining their reasoning as to why they had assigned the code. The individual study reports were compiled to produce a summary of how many of the possible 93 BCTs were found in interventions, how often they occurred and whether or not they were from particular clusters. Study outcome data were extracted and used to explore whether or not there were potential relationships between study outcomes and particular BCTs.

Data synthesis

The approach to data synthesis was designed to suit the diverse set of included records. It was not relevant to apply stringent academic appraisal techniques in a conventional way because the data set included some records that were neither academic publications nor formal reports. For example, a key source of information about the 6CS intervention was a set of workshop slides. 77

Meeting the objectives set out in this chapter involved exploring and categorising the records, identifying intervention evaluations and then conducting a detailed analysis of the available information about the interventions. The purpose of this was to identify BCTs to produce a synthesis of intervention characteristics, components in terms of BCTs, process elements, effectiveness evidence in terms of BCTs and intervention outcomes, and also to compare the results with the results of the companion COMPARE study47 focusing on adult acute mental health settings.

Therefore, data were synthesised by a process of close scrutiny of the included records, and tailored application of the MMAT and WIDER recommendations to understand the scope and quality of the materials and meet the study objectives.

Following extraction, the records were organised into groups according to the intervention or interventions they described. This allowed for a primary focus on the evidence for each intervention, rather than the overall evidence (objectives 4 and 6) in parallel with the classification and analysis of intervention components (objective 2).

Narrative synthesis

Interventions were placed in subgroups according to the setting and type of restrictive practice that they seek to reduce [e.g. p.r.n. (pro re nata) medication, physical and seclusion]. A narrative synthesis across and within each subgroup was carried out exploring and describing the features of the interventions including their theoretical basis, population, outcomes and conclusions. The content of the types of intervention was described in terms of the types and frequency of BCTs that could be identified [e.g. social support, skills practice and modelling (Table 3)]. The outcome data from the interventions were presented in relation to the BCTs present, and hypotheses were formulated around whether specific types of BCTs appeared more frequently, or not at all, in studies reporting certain outcomes.

Introduction

The chapter provides an overview of the literature search results, including a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) figure to indicate the extraction process. The included records are listed in Appendix 2. In this chapter, the results are described in detail and key characteristics of the data set are highlighted. As per objective 1 and in keeping with the mapping approach, a narrative overview of interventions aimed at reducing restrictive practices with CYP is then provided. It describes the characteristics of the interventions identified within the data set of records, including their scope and common features. The description of the evaluations is informed by WIDER reporting recommendations. Figure 4 summarises the study processes.

FIGURE 4.

Summary of study processes.

Search results

As illustrated in the PRISMA figure (Figure 5), the search of academic databases identified 43,494 records and these, as well as 8796 records found in the grey literature (including social media), were entered into Covidence (Melbourne, VIC, Australia) for further analysis. After the removal of duplicates, and accounting for records that were not available, 19,644 records were subjected to title and abstract screening. The final data set consisted of 121 records for extraction. Further details of the search strategy and results are available in Appendices 1 and 2.

FIGURE 5.

The PRISMA figure. Grey literature: non-academic databases and websites, social media and ‘other’ records. ‘Other’ records: discovered using forward citation searches and contact with authors. Excluded because not relevant: records excluded because they report not an intervention but, for instance, a generic policy change or replacement of one restrictive practice with another.

Responses to requests for intervention materials

In addition to the processes described above, attempts were made to request further information about interventions to supplement information found in the 121 included records. This involved sending e-mails to authors and co-authors, and, when appropriate, to organisations, using contact details provided in or otherwise gleaned from the records. Seventy-one e-mail requests were sent (see Appendix 3), resulting in new information about six of the interventions: Six Core Strategies for Reducing Seclusion and Restraint Use,33 Trauma Affect Regulation,78 Neurosequential Model of Therapeutics,79 Milieu Nurse Shift Assignments,80 Crisis Intervention81 and Checklist for Assessing Your Organisation’s Readiness. 82

Study screening

The records were diverse in terms of format and reporting quality. The first two questions of the MMAT70 were applied to screen the 121 records to identify those that were evaluations. In all, there were 76 records that were evaluations and 45 records that were descriptive only and did not contain evaluations. These 45 were used for mapping only and consisted of training resources, blogs, websites and almost all of the reports (e.g. reports to organisations).

Categorising the studies

Some interventions occurred in more than one record, some records reported more than one intervention and some reports were mentioned in more than one record. Overall, the data set contained 107 descriptions of interventions, referring to 82 interventions in total, of which 67 interventions had been evaluated. The data set is summarised in Figure 6.

FIGURE 6.

Summary of the data set.

Categorisation of study (evaluation) design

In view of the widely ranging literature retrieved from the searches, the MMAT was used to categorise all 121 records by study design. As reported above, 76 records were classified as evaluations; the remainder were descriptive only. The 76 evaluations were allocated, where possible, to one of the five MMAT categories:70 qualitative description, randomised controlled trial (RCT), non-randomised trial, quantitative description or mixed-methods study. As summarised in Table 4, the majority of the 76 records of evaluations reported non-randomised designs. Thirty-two evaluation records lacked sufficient detail for categorisation by study design with the MMAT. None was categorised as a RCT. Only 15 of the 45 mapping records provided detail of study design; however, as indicated above, not all mapping records were research reports. Based on the MMAT screening questions, intervention study design was as follows: non randomised, n = 41; quantitative, n = 21; mixed methods, n = 5; qualitative, n = 5; no study design reported, n = 2.

Consistency and comprehensiveness of intervention reporting

Overall, reporting about interventions lacked consistency and comprehensiveness. The WIDER tool69 that was used to develop the data extraction strategy also informed the appraisal of reporting quality and identified a great deal of missing information about key aspects of interventions. Within the evaluation records, intervention recipient and setting were well reported, but intervention aims and by whom the intervention was delivered were not consistently reported. Most evaluation records did not report on intervention dose, fidelity to the intervention protocol, whether or not modifications were made to the intervention, whether or not intervention protocols were used and whether or not service users were involved in the development of the intervention. Within the mapping records, reporting was weak across the WIDER categories. The detail is presented in Table 5. Evaluation and mapping records are reported separately because of the differing overall characteristics of each subset. The detail provided in Table 5 reflects information as reported directly in the records, rather than inferred or extrapolated.

Publication date and format

Figure 7 illustrates the pattern of publication dates of 118 of the records. Three records were undated training resources in which the context and content indicated that they fell within the inclusion criteria. The figure shows that there was a brief increase in publications in the late 1980s and a sharp increase from the mid-2000s. The latter increase coincides with a US-wide policy response83,84 to a series of newspaper reports published in 1998 in the Hartford Courant newspaper, highlighting deaths related to the use of restraint in mental health and learning disability facilities across the USA. 33,83,84

FIGURE 7.

Publication dates.

Characteristics of records

The 121 records were organised by format type. Study designs of evaluations and mapping records are summarised in Table 6. Over half of the records were published in academic journals (n = 61). Eighty (66.1%) of the 121 records were peer reviewed. The additional peer-reviewed sources included book chapters (n = 3),85–87 dissertations (n = 11)88–98 and conference proceedings (n = 5). 99–103 The other records comprised training resources (n = 12); newsletters (n = 6); professional magazines (n = 4); presentation slides (n = 4); websites (n = 3) and blogs (n = 2); and reports (n = 10), of which seven were for government departments (UK, Wales, USA), two were for training organisations and one was for a US health-care service provider. 104

Peer-reviewed sources

The 11 peer-reviewed sources that featured more than once appear in Table 7. The most frequently occurring single source was the Journal of Child and Adolescent Psychiatric Nursing.

Service setting

Just under half of the records (60/121) came from mental health settings. The other service settings were health and social care, criminal justice and education. Three evaluation records and 14 mapping records reported more than one setting within a single service and were categorised as ‘generic’ (Table 8).

Records and interventions by geographical setting

Reporting of interventions

An ‘intervention event’ indicates an occasion on which an intervention was implemented. For example, an intervention implemented on two separate occasions generated two ‘intervention events’. The same intervention implemented on a single occasion generated one ‘intervention event’, regardless of the number of records reporting it.

Two records seemed to pertain to an ongoing programme,38,105 but there were no other follow-up or replication studies. Several interventions were reported in more than one record, for example Craig88 and Canady,106 including some instances in which the same intervention evaluation was reported in different formats, such as a dissertation and a published paper (e.g. CPS89,107 and the Grafton program88,108).

The intervention for which the most records were identified was the 6CS [n = 12 records, including five evaluations (journal articles) and seven mapping records, comprising one journal article, one magazine, one training resource, one set of presentation slides, two blogs and one implementation tool]. The next largest group of records (n = 9) pertained to CPS. This group consisted of evaluations only, and comprised four dissertations,90–93 one publication from a dissertation107 and four journal articles. 109–112

The eight remaining interventions were comfort versus control (two events), TCI (three events), the Grafton program (two events), CPS (seven events), M-PBIS (three events), TARGET (two events) and SPA (two events). Those interventions that featured more than once are shown in Table 10. All records of interventions in evaluation studies and mapping studies are listed by author in Appendix 4, and further details are given in Appendix 5.

Intervention aims

All interventions aimed to reduce restrictive practices, and most focused on achieving that by changing staff behaviour.

Intervention recipient

When the recipient was reported, all interventions were delivered to staff, with some also aimed at service users and/or introduced within a wider organisation or as milieu change. Seventy-nine intervention events targeted staff only, 13 targeted staff and service users, and two included staff and/or service users in the context of change in milieu (Table 11).

Outcomes reporting

The 82 interventions described in the records reported a total of 228 outcome measures, with the number of measures described per record ranging from 0 to 11. The number of occasions when restraint was used was reported in 63 of the records, and the number of times seclusion was used was reported in 36 of the records. Other outcomes were reported in ≤ 11 records. Outcome measures are listed in Table 12.

Outcomes categories

Outcomes reported were in four broad categories: staff development and activity, use of restrictive methods, resource implications, and patient progression and satisfaction (see Table 12).

Use of standardised outcomes measures

The reporting of standardised measures is shown in Tables 13 and 14. The range of measures reported per record was 0 to 7. In 106 of the records, no standardised measures were reported. One record129 reported the use of seven standardised measures to evaluate an intervention. In total, 22 different standardised outcome measures were reported across the 121 records.

Assumed change process and design principles

Fifty-two records reported mandatory participation in the interventions, including 31 records that described interventions involving a whole system, either across a whole organisation (e.g. a hospital) or in a self-contained unit (e.g. a section of a residential school). Nine records reported voluntary engagement in interventions, and in the remaining 60 records it was unclear whether engagement in the intervention was mandatory or voluntary.

Many studies lacked internal congruence, in that the relationships between the aims, intervention, mechanisms of change and reported outcomes were not necessarily clear. For example, reductions in restraint data occurring after a staff education intervention might be interpreted as an effect of the intervention, with little attention to potential confounding factors or fidelity. This point is noted in the literature. 40,112

Mode of delivery: intervention procedures

The intervention procedures are set out by theme in Table 15. The most common procedures focused on staff training. Other procedures related to guideline or policy change, risk assessment tools, data review, milieu changes and changes to therapeutic approach (e.g. introducing trauma-informed care, and staff involvement in intervention development).

Organisation-focused procedures

In addition, several organisation-focused procedures were identified. These were system-wide structural and cultural changes including making changes to staffing levels85,109,132,170 or the way staffing was organised. 80 Another procedure involved changing therapeutic approaches (e.g. to a trauma-informed approach78,171,172). This theme also included improvements to communication (e.g. Ercole-Fricke et al. 89 and Kalogjera et al. 173), community meetings102 and de-escalation. 161 Further procedures focused on policy change115,174 and leadership, in which senior management tended to be directly involved in meetings and made statements of commitment. 38,78,175

The extraction process highlighted the procedures used by each intervention to address restrictive practices. The maximum number of procedures found in a single intervention was 15. A total of 16 unique procedures were identified from the analyses (Figure 8). The average number of unique procedures reported per record was 4.28 (mean) and 3 (median).

FIGURE 8.

Count of unique procedures per intervention. Unique procedures across interventions, n = 16. Maximum number of unique procedures per intervention, n = 15.

Twenty interventions (24%) used a single procedure only, and the most common single procedure was staff training (Table 16). However, many interventions (n = 62) used more than one procedure.

TABLE 16 - Procedures (frequency)

TIC, trauma-informed care.

Procedure	n	Percentage
Training	88	19.3
Changed approach (e.g. TIC)	58	12.7
Guidelines or policy change	44	9.6
Data review	36	7.9
Care planning changes	33	7.2
Debriefing	33	7.2
Enhanced leadership	30	6.6
Risk assessment tools	21	4.6
Milieu changes	20	4.4
Environmental changes	17	3.7
Staff involvement	17	3.7
CYP involvement	15	3.3
Family involvement	14	3.1
Enhanced staffing	13	2.9
Activities	9	2.0
Sensory approaches	8	1.8
Total procedures reported	456	100.0

Procedures used in interventions

The reporting on the procedures used in interventions was inconsistent and at times limited.

Intervention evaluation

Evaluations were identified by scrutinising each report using the screening questions of the MMAT to ascertain whether or not a research question was described and whether or not the data required to answer the question had been collected. Those reports that passed the screening were then appraised, again using the MMAT. The MMAT prompts an appraisal of if qualitative methods are appropriate; if the data collection methods are adequate, and the findings and their subsequent interpretations are sufficiently reported; and if the study has overall coherence. Evaluations are detailed in Appendix 5.

As seen in Appendix 5, there are more evaluation data about the 6CS and CPS than about any of the other interventions reported or described in the 121 retrieved records. Although the 6CS is more frequently reported, only 5 of the 12 6CS records are evaluations, as defined by the WIDER criteria. In comparison, there are nine separate implementations of CPS, all evaluated. The 6CS evaluations span 10 years, from 2007 to 2017, and the CPS evaluations span 8 years, from 2008 to 2016. Therefore, over a similar timespan, 6CS has been used more but evaluated less than CPS. This suggests that intervention use is not routinely generating evaluation data, and that intervention choice may not be informed by evaluation data.

Year of evaluation

As seen in Figure 9, starting in the mid-1990s, the number of evaluations that were commenced suddenly increased compared with the previous decade. Evaluations published prior to 1989 were not eligible for inclusion in the review. The commencement of evaluations appeared to then decrease steadily from the mid-2000s.

FIGURE 9.

Year evaluation commenced.

Individual behaviour change techniques identified across all interventions

The result of the search strategy was a data set of 121 records, which on analysis was found to report a total of 82 interventions. All 121 records were examined for BCT content. Descriptions of intervention content were usually found within the methods sections of studies, although additional details were occasionally provided in the results or discussion sections.

Thirty-six out of a possible 93 BCTs (39%) were identified across the 121 records. At least one BCT was detected in 78 of the interventions (95%). BCTs were not detected in four interventions because of a lack of content to code: two from mental health settings,103,158 one from a generic setting186 and one from the criminal justice system. 187

Behaviour change techniques found at least once across the interventions are shown in Figure 10. Figure 10 indicates the proportion of interventions in which each BCT was detected. For example, BCT 4.1 (instruction on how to perform the behaviour) was detected in the majority (71%) of interventions, whereas BCT 10.10 [reward (outcome) of the behaviour] was detected in only 1% of interventions. Figure 10 also illustrates which BCTs were the most and the least often detected across interventions. Seven were identified in > 20% of interventions (range 28–71%) and 11 were identified in ≤ 2% of interventions (range 1–2%).

FIGURE 10.

Individual BCTs identified in 82 interventions (% of interventions).

Identification of individual behaviour change techniques by setting

Table 23 illustrates the distribution of individual BCTs across interventions in different children’s settings in the order of frequency identified. Six BCTs (17% of BCTs detected) were detected in interventions from only one setting: four in mental health settings {BCT 14.10 [remove punishment], BCT 2.4 [self-monitoring of outcome(s) of behaviour], BCT 10.10 [reward (outcome)] and BCT 1.1 [goal setting (behaviour)]} and two in health and social care {BCT 1.7 [review outcome goal(s)] and BCT 5.1 [information about emotional consequences]}. These BCTs came from five clusters in total (36% of clusters detected).

Seven BCTs (19% of BCTs detected) were identified in interventions from all five settings. As seen in Figure 11, these BCTs came from five clusters in total (36% of clusters detected). Of these, BCT 4.1 (instruction on how to perform the behaviour) was the most often detected, although at lower rates in criminal justice and generic settings. This BCT was the only one to be detected in all interventions from a single setting (education). BCT 12.2 (restructuring the social environment) and BCT 1.4 (action-planning) were among the most often detected BCTs in all settings except the criminal justice system; both were detected in smaller percentage of interventions from generic settings. Similarly, BCT 13.2 (framing/reframing) was among the most often detected BCTs in all settings except education. BCT 2.7 [feedback on outcome(s) of behaviour] was among the most often detected BCT in three settings and BCT 1.2 (problem-solving) in two settings. BCT 13.1 (identification of self as role model) was among the most often detected BCTs in generic settings only.

FIGURE 11.

Percentage of BCTs detected in multiple settings (n = 36).

Although BCT 2.1 (monitoring of behaviour by others without feedback) was detected in interventions from four settings, it was among the top detected BCTs in criminal justice settings only. Similarly, BCT 7.1 (prompts or cues) was detected in interventions from three settings but was among the top detected BCTs in criminal justice settings only. BCT 2.3 (self-monitoring of behaviour) was another outlier, identified in three settings but among the top detected BCTs in education settings only (Figure 12).

FIGURE 12.

Individual BCTS by setting (% of interventions in that setting).

Individual BCTs detected in one or more settings were not necessarily detected in all settings, as detailed in Appendix 7. Four (11%) BCTs detected elsewhere were not identified in mental health settings, nine (25%) were undetected in health and social care, 24 (67%) in education, 17 (47%) in criminal justice and 26 (72%) in generic settings.

Clusters of behaviour change techniques identified in interventions

The BCT taxonomy52 organises BCTs into 16 clusters; for example, BCT 4.1 (instruction on how to perform the behaviour) belongs to cluster 4 (shaping knowledge). The BCT number preceding the name indicates the cluster. The 36 BCTs identified in the interventions came from 14 of the possible 16 clusters within the BCT taxonomy. The 14 clusters, the content that was coded to BCTs within them and the BCTs that were not identified are described below.

Figure 13 shows BCTs identified in interventions by cluster (%). The majority of the BCTs identified belonged to four clusters and were detected in over half of all interventions: cluster 1 (goals and planning), cluster 12 (antecedents), cluster 4 (shaping knowledge) and cluster 2 (feedback and monitoring). Less than 1% of identified BCTs came from two clusters and were detected in less than 2% of interventions [cluster 9 (comparison of outcomes) and cluster 14 (scheduled consequences)].

FIGURE 13.

Behavior change technique clusters identified (% of interventions).

Five clusters featured only by virtue of one BCT being coded:

1. cluster 7 (associations) – BCT 7.1 (prompts/cues)

2. cluster 8 (repetition and substitution) – BCT 8.1 (behavioural practice/rehearsal)

3. cluster 9 (comparison of outcomes) – BCT 9.1 (credible source)

4. cluster 11 (regulation) – BCT 11.2 (reduce negative emotions)

5. cluster 14 (scheduled consequences) – BCT 14.1 (remove punishment).

Clusters of behaviour change techniques by setting

Figure 14 shows the distribution of BCT clusters across interventions in different children’s settings. Eleven out of the 14 clusters detected were identified in four or five settings. Only one cluster was detected in a single setting [cluster 9 (comparison of outcomes) in the criminal justice setting], one in two settings [cluster 8 (repetition and substitution) in mental health and health and social care settings] and one in three settings [cluster 3 (social support) in mental health, health and social care, and criminal justice settings].

FIGURE 14.

Behavior change technique clusters as percentages of settings and interventions.

Interventions from mental health settings were coded with BCTs from all clusters except cluster 9 (comparison of outcomes). Interventions from health and social care settings were coded with BCTs from all clusters except cluster 9 (comparison of outcomes) and cluster 14 (scheduled consequences). Interventions from criminal justice settings were coded with BCTs from all clusters except cluster 11 (regulation) and cluster 14 (scheduled consequences). Interventions from educational settings were identified in seven clusters, and interventions from generic settings from nine clusters (Figure 15).

FIGURE 15.

Behaviour change technique cluster by settings (percentages of interventions in each setting, containing BCTs from each cluster).

Description of behavior change techniques and behavior change technique clusters identified

The BCTs and BCT clusters identified in the interventions are summarised in Figure 16.

FIGURE 16.

Detected BCTs by cluster (percentage of interventions).

Cluster 1 (goals and planning)

Six out of the nine BCTs in cluster 1 (goals and planning) were identified: BCT 1.4 (action-planning), BCT 1.2 (problem-solving), BCT 1.9 (commitment), BCT 1.1 [goal setting (behaviour)], BCT 1.3 [goal setting (outcome)] and BCT 1.7 [review outcome goal(s)]. These six BCTs accounted for 17% of all identified BCTs (n = 36). BCTs from this cluster were identified in 68% of interventions overall (n = 82) from all five settings. Almost all interventions (91%) from generic settings were coded with BCTs from this cluster, compared with around half (57%) from health and social care settings (Figure 17).

FIGURE 17.

Cluster 1 (goals and planning) by setting (% of interventions in each setting containing BCTs from this cluster).

Action-planning (behavior change technique 1.4)

Prompt, detailed planning of performance of the behavior (must include at least one of context, frequency, duration and intensity). Context may be environmental (physical or social) or internal (physical, emotional or cognitive) (includes ‘Implementation Intentions’).

Michie et al. 52

Behavior change technique 1.4 (action-planning) was one of the most often identified BCTs, found in 41% of interventions and in all five settings. Written action plans were produced for a variety of purposes, including setting out how to support or modify the behaviour of children or staff,30,97,100,102,180,185,188 how to support recovery33 plans for follow-up33,77,113,189 and how to prevent or respond to crises,85,122 as illustrated in Lebel et al. :40

Using the tool, a patient, family member (as appropriate), and staff person collaboratively develop a plan . . . that identifies preferred strategies for de-escalation . . . and restraint preferences to consider if restraint becomes necessary.

Lebel et al. 40

In some interventions such planning formed, or was incorporated into, individuals’ treatment plans:82,95,105,160,162,172,188,190,191

The interdisciplinary treatment team (IDT) met to review the treatment plan, conduct additional assessments (e.g., antecedent analyses and functional analyses), and determine possible etiological factors associated with the patient’s frequent crises. Behavioral and pharmacological treatments were then re-evaluated for possible modification.

Schreiner et al. 41

Plans could be aimed at individuals, but could also require teams or organisations to perform incident reviews or implement restraint reduction plans. 33,77,182,188,193 Foci included crises, behaviour, treatment, restraint reduction,179 incidents requiring the use of restrictive practices23,86,192 and safety:

The action plan was developed by the Executive Director of the region taking into account the feedback gathered by employees with particular focus on providing employees support and reassurance that safety would be the top priority. The action plan was reviewed with employees by the Executive Director.

Sanders108

Plans were often proactive, detailing actions or strategies to prevent scenarios with the potential to lead to restrictive practices162,193 to ‘proactively respond in stressful situations’171 and to ‘proactively identify and address problems in a more productive manner’78 before the situation escalates. 194 Action plans could also detail the steps required in specific situations, for example ‘situations in which the youth is likely to exhibit disruptive or oppositional behaviors’,112 or during crisis and emergency situations, as described by Schreiner et al. :41

Staff members were trained . . . to provide patients with concrete de-escalation steps. Rather than telling a patient to calm down . . . staff might tell a patient to go to his room, sit on his bed, take five deep breaths, and think about a pleasant event.

Schreiner et al. 41

Other situations might include single incidents of violence and/or restrictive practices,101,115,122,174,194,195 children exhibiting particular characteristics or patterns of behaviour156,196 or a pattern of restrictive practice use. 197 A checklist from the National Association of State Mental Health Program Directors (NASMHPD)33 suggests that the S/R reduction team could develop a policy to clarify how this level of review would be triggered, for example a child who receives more than three holds in a single week, an event resulting in an injury or an unusual pattern of use of S/R.

Some action plans were developed with the involvement of children. 171,181,188,192 For example, Finnie91 refers to working collaboratively with children to find mutually satisfactory solutions by ‘articulating the problem to be solved . . . associated concerns, the possible solutions and likely outcomes, and whether outcomes are feasible and will be mutually satisfying’.

Problem-solving (behavior change technique 1.2)

Analyse, or prompt the person to analyse, factors influencing the behavior and generate or select strategies that include overcoming barriers and/or increasing facilitators (includes ‘Relapse Prevention’ and ‘Coping Planning’).

Michie et al. 52

Behavior change technique 1.2 (problem-solving) was another commonly identified BCT and was identified in 32% of interventions from all five settings. As illustrated in Greene et al. ,110 some interventions93,111,125,132,161,162,178,193 explicitly referred to ‘problem-solving’:

In its focus on facilitating adult–child problem-solving . . . the CPS approach differs from models typically employed in many restrictive facilities. [. . .] The manner in which the CPS model teaches adults to solve problems collaboratively with patients also has important implications for staff interactions with each other.

Greene et al. 110

The process of ‘problem-solving’ involved identifying problems and solutions41,174,188 (e.g. Thomann98 and Kaltiala-Heino et al. 198) and identifying barriers to these plans. 115,193 Analyses were conducted in clinical reviews,182,193 debriefings or incident reviews,122,162,192 regular meetings and risk assessments. 77 The purpose was to identify factors that influenced the behaviour of staff and children, including existing plans, strategies and values. 116,190 Just one intervention178 referred to organisational ‘problem-solving’. Often organisational problem-solving involved the analysis of triggers, for example for challenging behaviour:33,77,89,105,110,115,154,165,174,189,193,199

. . . using CPS with a youth . . . generates a list of specific unmet expectations and triggers . . . in which the youth is likely to exhibit disruptive or oppositional behaviors. [For example] . . . unmet academic expectations (e.g., turning in homework), . . . safety expectations (e.g., staying sober), or social triggers (e.g., exclusion during playground games).

Pollastri et al. 112

Behavior change technique 1.2 (problem-solving) was mostly concerned with the individual behaviour (of children), but there were examples that were concerned with broader influences. For instance, the Youth Justice Board for England and Wales stated that assessment should not be limited to describing facts but should aim to understand the situation and behaviour of a young person, including factors relating to diversity. 199

Other examples included risk assessment,77 strategies for ‘stressful situations’ and ‘red flags’,171 ‘warning signs’30 and redesigning environmental conditions:

[A functional behavioral assessment] is used to analyze environmental factors, including any history of trauma (e.g., physical abuse), that contribute to a child’s inappropriate (e.g., disrespect, noncompliance, insubordination, out-of-seat) behaviors.

[These data inform] positive behavioral strategies that emphasize redesigning environmental conditions, which may include changes in staff approaches.

US Department of Education. 197 Reproduced with permission from US Department of Education, Restraint and Seclusion: Resource Document, Washington, DC, 2012

Problem-solving activities resulted in the generation of ‘lessons learned’199 that inform future strategies,88,174,194 the development of (potential) solutions78,120 and improvements in practice. 168,192

Commitment (behavior change technique 1.9)

Ask the person to affirm or reaffirm statements indicating commitment to change the behavior.

Michie et al. 52

As shown in Figure 17, 18% of interventions described organisational-level ‘commitment’ (BCT 1.9) to reduce restrictive practices. For example, Caldwell et al. 115 emphasised the role of leadership in prioritising the reduction of restraint and seclusion within the organisation, with a focus on assessing its use.

This took various forms including a mandate issued by the Chief Executive Officer,108 a ‘statement of beliefs’182 or a declaration as a ‘unit priority’. 41 Reynolds et al. 100 described requiring individual staff to make a commitment to reducing restrictive practices:

Tier 1 strategies included establishing commitment from the staff (of note, 100% of the unit staff voted to implement the program), defined set of positively worded expectations (i.e. be safe, be responsible, be respectful).

Reynolds et al. 100

Goal setting for behaviour (behavior change technique 1.1) and outcome (behavior change technique 1.3)

Set or agree on a goal defined in terms of the behavior to be achieved.

Michie et al. 52

Set or agree on a goal defined in terms of a positive outcome of wanted behavior.

Michie et al. 52

Seven per cent of interventions described ‘goal setting for outcome’ and 2% described ‘goal setting for behaviour’ (see Figure 17). BCT 1.1 [goal setting (behaviour)] was identified in mental health settings only. Considerably fewer interventions described setting goals (BCT 1.1)33,40,81,179,190,191,193 than BCT 1.4 (action-planning). BCT 1.4 (action-planning) requires mention of context, frequency, duration or intensity within the plan. In the case of those that specified the goal,179,191 the goal was a reduction in or the prevention of the use of restrictive practices:

Specific patterns of restraint use were reviewed by the team each meeting, and goals for restraint prevention/reduction were set.

Holstead et al. 191

There was some variation in whether these goals were set for teams, individual staff or children. In Plant179 the goals were set for teams, and in Azeem et al. 193 the goals were set for teams and individual staff, whereas in Holstead et al. 191 the goals were set for staff and children. Both interventions describing goal setting in relation to behaviour (BCTs 1.1 and 1.3) referred to the joint goal setting between staff and youth171 and by staff in collaboration with children and families. 163

Review outcome goal(s) (behavior change technique 1.7)

Review outcome goal(s) jointly with the person and consider modifying goal(s) in light of achievement. This may lead to resetting the same goal, a small change in that goal or setting a new goal instead of, or in addition to the first.

Michie et al. 52

Behaviour change technique 1.7 [review outcome goal(s)] was detected in health and social care settings only. Just 5% of interventions (see Figure 17) described reviewing or evaluating outcome goals:

Progress towards the goals was evaluated monthly and ideas to further reduce restraints were discussed.

Holstead et al. 191

As described, the aim of review was to ensure that goals remained ‘fit for purpose’180 and were revised when necessary. 33

Cluster 2 (feedback and monitoring)

Behaviour change techniques from cluster 2 (feedback and monitoring) accounted for 17% of all identified BCTs. All but one BCT [BCT 2.6 (biofeedback)] from this cluster were identified: BCT 2.1 (monitoring of behavior by others without feedback), BCT 2.5 [monitoring of outcome(s) of behavior without feedback], BCT 2.3 (self-monitoring of behavior), BCT 2.2 (feedback on behavior) and BCT 2.4 [self-monitoring of outcome(s) of behavior]. Feedback and monitoring were detected in relation to both outcomes and behaviour. Monitoring was either self-monitoring or monitoring by others, for example at an organisational level. BCTs from cluster 2 were identified in 52% of interventions overall and at similar proportions across settings, except for generic settings where they were detected in 71% of interventions (Figure 18).

FIGURE 18.

Cluster 2 (feedback and monitoring) by setting (% of interventions in each setting containing BCTs from this cluster).

Self-monitoring of outcome(s) of behaviour (behavior change technique 2.4)

Establish a method for the person to monitor and record the outcome(s) of their behavior as part of a behavior change strategy.

Michie et al. 52

Behavior change technique 2.4 [self-monitoring of outcome(s) of behavior] was identified in 1% of interventions, and all identifications came from mental health settings (see Figure 18). Here, ‘self’ refers to the staff-monitoring incidents that occurred in their area {distinct from the centralised system described under BCT 2.5 [monitoring of outcome(s) of behaviour]}. Facilities can be encouraged to participate in setting goals and monitoring changes. 33

Monitoring could also take the form of supervision174 or debriefing23,82,174,192 but both of these were aimed at generating reflection following incidents. A positive staff debrief would involve active discussion about what happened, what went well, what might be done better, the possible effect on the child and how to avoid restraining the child in the future. 200

Self-monitoring of behaviour (behavior change technique 2.3)

Establish a method for the person to monitor and record their behavior(s) as part of a behavior change strategy.

Michie et al. 52

Behavior change technique 2.3 (self-monitoring of behaviour) (rather than outcomes) was identified in five interventions from three settings. BCT 2.3 (self-monitoring of behaviour) was coded when staff monitored and reflected on ‘near misses’174,193 and ‘what worked’193 when restrictive practices were successfully avoided:

These sessions were expected to provide staff with an opportunity to reflect on practices and offer peer support through encouragement and critical feedback [. . .] while also affording an opportunity for trainers to assess the efficiency of the process and make adjustments as necessary.

Glew92

Holstead et al. 191 actively trained staff ‘to recognize and be more aware of their own internal experiences during these critical moments.’191 Sanders108 emphasised that the purpose of debriefs was to be supportive of staff and not critical.

Feedback on behaviour (behavior change technique 2.2)

Monitor and provide informative or evaluative feedback on performance of the behavior (e.g. form, frequency, duration, intensity).

Michie et al. 52

Behavior change technique 2.2 (feedback on behaviour) was identified in 7% of interventions from two settings. BCT 2.2 (feedback on behaviour) was provided in the same ways as self-monitoring, such as debriefing and supervision, but specifically involved providing feedback to staff. Feedback had a number of functions including helping staff to develop awareness of ‘common precipitants’ and how effective they have been in the past190 and ‘to role model how to identify reactive behaviors’. 128 Jones and Timbers201 described the involvement of children in contributing to evaluations of staff performance.

Feedback on outcome(s) of behaviour (behavior change technique 2.7)

Monitor and provide feedback on the outcome of performance of the behavior.

Michie et al. 52

Behavior change technique 2.7 (feedback on outcomes of behaviour) was one of the most frequently identified BCTs, detectable in 39% of interventions across from all five settings. Interventions monitored data and incidents and gave feedback with a view to understanding what could have been done differently108,114,180,192,193 as well as identifying ‘critical success factors’. 202

Sanders108 described how every restraint incident was carefully scrutinised by an executive team that established whether or not the restraint had been necessary, offered feedback such as recommendations for changes to the environment and reported to a regional body.

Some interventions described having data monitoring systems in place through which rates of restrictive practices could be fed back to staff41,125,127,185 to help identify trends and inform change,180 including ‘when staff members had high rates of using restraint and seclusion’. 114

Feedback could also involve children:101,188,194

Following each hands-on incident, the patient is asked his or her interpretation of the incident. . . . Recommendations or individual preferences offered by the patient at this or any other time are communicated to the treatment team . . . The treatment plan then is reviewed and/or updated.

Visalli et al. 188

Monitoring of behavior by others without feedback (behavior change technique 2.1)

Observe or record behavior with the person’s knowledge as part of a behavior change strategy.

Michie et al. 52

Feedback was not always provided following monitoring. BCT 2.1 (monitoring of behaviour by others without feedback) was detected in 5% of interventions. These interventions, which came from mental health and criminal justice settings, described monitoring behaviour using, for example, ward-based registers,163 video- and audio-recorded footage and information from the debriefing of young people and staff. 192 They also described the collection of a variety of data, as in Finnie,91 in which it was reported that data were collected on an ongoing basis for each variable that had been identified [i.e. counts of episodes of four-point restraint, locked seclusion, staff and patient injuries, involvement of security staff to assist with managing a child’s ‘explosive’ episode’ and the use of p.r.n. medication for agitation].

The behaviours being monitored included aggressive incidents, security support and restraint;163 individual incidents;199 and the use of force. 192 Rather than for providing feedback, the purpose of monitoring was described by the Youth Justice Board for England and Wales as providing a ‘total picture’ of restrictive interventions over a period of time199 and acting as an ‘accurate record’: a way of verifying whether or not correct restraint techniques have been used and to permit investigation of antecedents. 192

Monitoring of outcome(s) of behavior without feedback (behavior change technique 2.5)

Observe or record outcomes of behavior with the person’s knowledge as part of a behaviour change strategy.

Michie et al. 52

Similarly, where monitoring of outcomes of behavior was described, feedback was not always provided as the end point. Indeed, BCT 2.5 (monitoring of outcomes of behaviour without feedback) was more common. It was detected in 10% of interventions (see Figure 18) and in all settings except criminal justice. This type of monitoring was usually undertaken at an organisational level with the systems in place to ensure the central collection and analysis of data:

All seclusion and restraint events were chart audited through the hospital’s EMR [electronic medical record] system. The documented nursing note on each patient placed in seclusion and restraint was reviewed to determine if staff followed the decision making algorithm.

Eblin162

. . . establishments should collect the following information about incidents involving the use of restraint and analyse this information at least monthly: reason for restraint . . . emerging patterns of restraint; restraint ‘hotspots’, for example . . . locations; time restraint incidents occur; which staff . . . were involved; risks in restraint techniques; training gaps identified.

Youth Justice Board for England and Wales199

The Youth Justice Board for England and Wales199 stated that institutions should, at least monthly, collect and analyse restraint data concerning why restraint had been used and any patterns of restraint use, such as locations, timing, staff involved, risks arising from restraint techniques and implications for training needs.

These data might be reviewed internally by medical, nursing or operational directors or others at an executive level105,120 or the data may be made available outside the individual institution to an overarching governing body such as the Youth Justice Board for England and Wales199 or the National Offender Management System. 23

In some interventions, all incidents were recorded161,192 but not necessarily reviewed. Elsewhere, reviews occurred either as the incidents happened87,105 or on a regular basis. 90,120 Two interventions described specific features, such as an individual experiencing multiple incidents in a given period that triggered notification for review by the medical director. 105,203

Cluster 3 (social support)

There was evidence for all three BCTs – that is emotional, practical and unspecified social support – from cluster 3 (social support). These three BCTs amounted to 8% of all BCTs identified (n = 36) and were detected in 37% of interventions overall, but slightly more often in health and social care settings (57%) (Figure 19).

FIGURE 19.

Cluster 3 (social support) by setting (% of interventions in each setting containing BCTs from this cluster).

Social support (practical) (behavior change technique 3.2)

Advise on, arrange, or provide practical help (e.g. from friends, relatives, colleagues, ‘buddies’ or staff) for performance of the behavior.

Michie et al. 52

Behavior change technique 3.2 [social support (practical)] was detected in 28% of interventions from all settings except those that were ‘generic’. It was also one of the most often detected BCTs in education and criminal justice settings. Various forms of practical (as opposed to emotional or unspecified) support were described. Several interventions deployed response teams to provide additional support following incidents115 or during incidents to assist with, for example, conflict resolution or de-escalation to prevent escalation of an incident. 99,105,194,198 This support might be provided via coaching191 or modelling. 108 Another approach evolved organically:

Staff began briefly consulting with each other prior to approaching a patient in crisis. This practice enabled staff to focus on the individual’s treatment plan, brainstorm creative alternatives in the choices offered to patients in crisis, and provide consistent, professional interactions with an individual in crisis.

Witte160

Other types of practical support were also identified, including coaching,108,114,119,174,180,204 staff forums and meetings,40,122,175 peer support and the use of peers and buddies,92,160,174,205 mentoring and modelling,114,180 and supervision. 178,182,183,206

Some interventions set up a committee specifically to perform these functions. 41,125 These functions could also be provided by an external party, such as the intervention developer or trainer:40,111,128,129,178

Each facility was assigned a trainer-mentor to provide on-going consultation, modeling, and coaching in PARS concepts, techniques, and methods.

Wisdom et al. 114

Social support (emotional) (behavior change technique 3.3)

Advise on, arrange, or provide emotional social support (e.g. from friends, relatives, colleagues, ‘buddies’ or staff) for performance of the behavior.

Michie et al. 52

Less often, BCT 3.3 [social support (emotional)] was detected (in 2% of interventions from two settings). These interventions described offering emotional (as distinct from practical) social support193 via events for the purpose of reinforcing staff ‘well-being through self-care’,105 culture change183 and debriefing. 30,122,180 Advice from Holden et al. 122 was to ensure a positive outcome from a crisis. This could be facilitated by recognising its emotional impact on staff, and allowing them to process that by conducting an incident review once the urgent safety concerns had been dealt with.

Provision of emotional support was also an item in two intervention checklists. 33,207

Social support unspecified (behavior change technique 3.1)

Advise on, arrange or provide social support (e.g. from friends, relatives, colleagues,’ buddies’ or staff) or noncontingent praise or reward for performance of the behavior. It includes encouragement and counselling, but only when it is directed at the behavior.

Michie et al. 52

Behavior change technique 3.1 [social support (unspecified)] was also identified, but in just 4% of interventions from two settings. 201,204,208,209

Cluster 4 (shaping knowledge)

Two out of four possible BCTs were identified from cluster 4 (shaping knowledge). These were BCT 4.2 (information about antecedents) and BCT 4.1 (instruction on how to perform the behaviour). These two BCTs amounted to 6% of all BCTs identified (n = 36). This cluster was detected in 84% of interventions overall, but was present in all interventions from the education setting (Figure 20).

FIGURE 20.

Cluster 4 (shaping knowledge) by setting (% of interventions in each setting containing BCTs from this cluster).

Instruction on how to perform the behaviour (behavior change technique 4.1)

Advise or agree on how to perform the behaviour (includes ‘Skills training’).

Michie et al. 52

Behavior change technique 4.1 (instruction on how to perform the behaviour) was the most common BCT, detectable in 71% of interventions. This BCT was usually evident when interventions described training of some sort. The type and duration of training varied widely, as did descriptions of its content and delivery. Some interventions described a training curriculum, whereby training was ongoing, substantial and multifaceted. 30,33,82,92,100,107,113,114,125,175,178–180,191,207,210 In one children’s centre, a training department was set up with two professional staff trainers who were employed full time. 211

The RRCs [Restraint Reduction Committee] implemented staff training that was centered on less restrictive interventions (LRIs) for aggressiveness and covered assessment of the client and situation, behavior management, and verbal de-escalation strategies. Although training included quarterly refresher courses, training was not limited to single, discrete sessions.

Miller et al. 182

These were in contrast to one-off or time-limited training sessions (which also ranged in depth and duration) described by others. 93,161,172,177 Several referred to training resulting in accreditation. 169,170,174 Some interventions specified that training was aimed at all staff in the setting:128,159,183

All detention center personnel (including food service, clerical, maintenance and janitorial, educator, administrator, and clinician staff) received an introductory training and periodic refresher trainings from one of two TARGET consultants.

Ford et al. 128

This BCT was also coded when instructions were provided in a manual89,165,209 or multimedia sources:79,89

A comprehensive training manual was supplied to all participants. The training manual . . . mirrored the training.

Bobier et al. 165

Training content included how to implement the intervention,91,125,193,204 how to use a tool,154,165 how to perform a specific skill such as de-escalation,23,92,101,105,115,119,159,202 and models and theories underpinning interventions such as trauma-informed care105,115,120,129,183 or sensory modulation. 155,156,164 In a Crisis Prevention Institute blog, Rettmann119 explained that Nonviolent Crisis training was delivered to staff involved in direct care and supervisors. Although, in most cases, the staff had received the training previously, on this occasion there was greater emphasis on the importance of developing professional relationships with the young people, and proper use of the Nonviolent Crisis Intervention de-escalation techniques.

Information about antecedents (behavior change technique 4.2)

Provide information about antecedents (e.g. social and environmental situations and events, emotions, cognitions) that reliably predict performance of the behaviour.

Michie et al. 52

Evidence of BCT 4.2 (information about antecedents) could be found in 15% of interventions from two settings. Several interventions made specific reference to antecedents:85,111,182,191,194,197,199,202,207

CPS seeks to identify pertinent social and cognitive pathway impairments and precipitating antecedent events.

Martin et al. 111

Antecedents were also referred to as ‘warning signs’. 30,114,174,194 Familiarity with an individual’s early warning signs, for example pacing, avoiding eye contact or going silent, was identified as important knowledge that would help practitioners in preventing service users from becoming over-agitated. 174

This BCT was also detected when interventions provide information about triggers in debriefings, on admission (from children and families), risk assessment or in training:85,98,114,115,122,132,163,164,189,191,194,198,202,203

All staff, including a number of non-agency temporary staff during the first two years, received skills-oriented, criteria-based training in the GBT [Girls and Boys Town] Psychoeducational Treatment Model, including teaching youth behavioral and cognitive techniques for recognizing antecedents and triggers.

O’Brien202

Cluster 5 (natural consequences)

The four BCTs [i.e. BCT 5.6 (information about emotional consequences), BCT 5.1 (information about health consequences), BCT 5.4 (monitoring of emotional consequences) and BCT 5.2 (salience of consequences)] from cluster 5 (natural consequences) accounted for just 8% of identified BCTs. BCTs from this cluster were identified in 12% of interventions overall. As illustrated in Figure 21, BCTs from this cluster were detected at varying rates in each setting.

FIGURE 21.

Cluster 5 (natural consequences) by setting (% of interventions in each setting containing BCTs from this cluster).

Information about health consequences (behavior change technique 5.1)

Provide information (e.g. written, verbal, visual) about health consequences of performing the behavior.

Michie et al. 52

Behavior change technique 5.1 (information about health consequences) was detected in 4% of interventions from two settings. Interventions provided staff with information about the high-risk nature of restrictive practices as well as the negative impact on the physical health of staff and children, which may include injury and death:33,122,170,194

Investigations have shown that the causes of restraint- or seclusion-related death include suffocation, heart complications, drug overdoses or interactions, blunt trauma, strangulation or choking, fire or smoke inhalation, and aspiration.

World Health Organization (WHO). 194

Reproduced with permission. © World Health Organization 2019. Some rights reserved. This work is available under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 IGO licence (CC BY-NC-SA 3.0 IGO; https://creativecommons.org/licenses/by-nc-sa/3.0/igo)

Staff are taught and acknowledge that emergency physical interventions are risky for both the child and the staff and that such techniques should be used only when the risk of intervening is outweighed by the risk of not intervening.

Paccione-Dyszleski et al. 170

Salience of consequences (behavior change technique 5.2)

Use methods specifically designed to emphasise the consequences of performing the behaviour with the aim of making them more memorable (goes beyond informing about consequences).

Michie et al. 52

Behavior change technique 5.2 (salience of consequences) was detected in 2% of interventions from two settings in the representation of the seriousness of restrictive practices as ‘human rights violations’ and ‘violent adverse events’,194 and in the form of testimonies of staff and children who had experienced the consequences of restrictive practices:115,174,191,211

Training content should include contributions from people with lived experience of having restraint or other restrictive practices used on them. It is important that practitioners who apply restraint have an understanding of the personal and often traumatic impact it can have.

Welsh Government. 180

Reproduced with permission. © Crown copyright. Contains public sector information licensed under the Open Government Licence v3.0

Monitoring of emotional consequences (behavior change technique 5.4)

Prompt assessment of feelings after attempts at performing the behavior.

Michie et al. 52

This BCT was detected in recommendations for immediate post-incident review,33,199 apparent in just 2% of interventions from two settings.

Information about emotional consequences (behavior change technique 5.6)

Provide information (e.g. written, verbal, visual) about emotional consequences of performing the behaviour. Note: consequences can be related to emotional health disorders (e.g. depression, anxiety) and/or states of mind (e.g. low mood, stress).

Michie et al. 52

Just 1% of interventions provided information about the emotional consequences (BCT 5.6) of restrictive practices, which may include trauma. 122,180,194 This BCT was detected in health and social care settings only.

Cluster 6 (comparison of behavior)

Two BCTs [i.e. BCT 6.1 (demonstration of the behaviour) and BCT 6.2 (social comparison)] out of the three BCTs in cluster 6 (comparison of behavior) were identified. This constituted just 6% of all identified BCTs. Cluster 6 was identified in 10% of all interventions but at varying rates (range 7–22%) in the different settings. It was not identified at all in education or generic settings (Figure 22).

FIGURE 22.

Behavior change techniques in cluster 6 (comparison of behavior) by setting (% of interventions in each setting containing BCTs from this cluster).

Demonstration of the behavior (behavior change technique 6.1)

Provide an observable sample of the performance of the behaviour, directly in person or indirectly e.g. via film, pictures, for the person to aspire to or imitate (includes ‘Modelling’).

Michie et al. 52

The BCT demonstration of the behavior (BCT 6.1) was identified in 4% of interventions from two settings. Examples included interventions demonstrating the desired skills or a tool to staff204 or providing example scripts of interactions. 91 For example, one training programme explained that a skill should be modelled in a way that illustrates the behavioral steps clearly and unambiguously, and, furthermore, than the skill will be viewed as important and effective where the trainer is viewed as having high-level skills, competence and relatable experience, and as friendly, helpful and successful in their career. 122

Social comparison (behavior change technique 6.2)

Draw attention to others’ performance to allow comparison with the person’s own performance.

Michie et al. 52

Social comparison (BCT 6.2) was identified in 6% of interventions (see Figure 22), for example where they recommended or implemented visits or contact with other organisations,105,116 attending presentations,40 field trips or conferences. 102 The purpose of these was to make comparisons,175 foster healthy competition33,193 and to learn from the successful implementation of an intervention. 116 BCT 6.2 (social comparison) could be identified within183 or across organisations105,116 or both. 33

Cluster 7 (associations)

Behavior change technique 7.1 (prompts or cues) was the only BCT from a possible eight to be identified from cluster 7 (associations). This amounted to 3% of total BCTs identified. Nevertheless, this BCT (and, therefore, the cluster) was identified in 23% of interventions overall. Rates of detection varied across settings, from 9% in generic settings to 33% in criminal justice (Figure 23).

FIGURE 23.

Cluster 7 (associations) by setting (% of interventions in each setting containing BCTs from this cluster).

Prompts or cues (behavior change technique 7.1)

Introduce or define environmental or social stimulus with the purpose of prompting or cueing the behavior. The prompt or cue would normally occur at the time or place of performance.

Michie et al. 52

The BCT prompts or cues (BCT 7.1) was identified in 23% of interventions from all settings except generic (see Figure 23). It was also one of the most often detected BCTs in education and criminal justice settings. A prompt or cue could take the form of a tool (i.e. a safety planning tool, a risk assessment tool, prevention tools or a risk management tool33,77,114,156). Examples of such tools included the personal safety assessment tool165 and the ‘moment by moment assessment’ tool. 78 Tools could be used to prompt preventative behaviour (i.e. through assessment and planning) to assist advanced decision-making and advance directives30,33,77 or to aid responsive decision-making:

The decision-making algorithm with an agitated patient was to be used before the initiation of seclusion and restraints and included the following steps: verbal de-escalation, assessing effectiveness, notifying the physician for emergency medication, then assessing again for effectiveness.

Eblin162

Moments in time were also coded as prompts or cues. For example, in several interventions a specific process, such as carrying out risk assessments30,77,162,169,176 or person-centred assessments180 or the formulation of crisis management strategies,189,198 would be prompted by preadmission, admission or an incident. Prompts or cues also appeared in visual displays (i.e. signs, posters and displays):183,184

One program created an ‘On Track Action’ wall . . . to reinforce positive steps that each client could take . . . The goal of the wall was to provide clients with weekly and monthly incentives for positive behaviors and to provide staff members with a systematic tool for consistently giving praise.

Hodgdon et al. 183

In some interventions, prompts or cues occurred in the form of cards to be carried or t-shirts displaying the curriculum logo to be worn:183,184

They also had all staff and residents make safety plan cards (written plans of how to handle triggers and other safety measures) and made sure that everyone wore them.

Elwyn et al. 178

Staff members were also instructed to carry a pocket-sized tool step guide (i.e., that listed all of the steps for each tool) at all times.

Crosland et al. 204

Cluster 8 (repetition and substitution)

Just one BCT, behavioral practice/rehearsal (BCT 8.1), was identified from the seven available BCTs in cluster 8 (repetition and substitution), representing 3% of all identified BCTs; however, this BCT was detected in 17% of all interventions. It was not detected in generic settings, and rates of detection ranged from 7% in mental health settings to 44% in criminal justice settings (Figure 24).

FIGURE 24.

Cluster 8 (repetition and substitution) by setting (% of interventions in each setting containing BCTs from this cluster).

Behavioral practice/rehearsal (behavior change technique 8.1)

Prompt practice or rehearsal of the performance of the behavior one or more times in a context or at a time when the performance may not be necessary, in order to increase habit and skill.

Michie et al. 52

From cluster 8 (repetition and substitution), the single BCT behavioral practice/rehearsal (BCT 8.1) was identified. This BCT was detected in 17% of interventions from all settings except generic (see Figure 24). It was also one of the most often detected BCTs in education and criminal justice settings. This BCT involved practising various techniques,92,93,108,122,128,183,204 sometimes with supervision. For example, a resource document from the US Department of Education197 specified that, with regard to the prevention of restrictive practices, staff should practise and review their approaches regularly and frequently, under supervision, as they would a fire drill.

Techniques reported included role-play40,85,107,174 and other exercises:184

All staff members practiced and rehearsed the procedures in the training sessions. The program concluded with the staff members applying the material learned in situational role plays and discussing post intervention techniques.

Ryan and Rigby212

Skill practice exercises for monitoring and enhancing appropriate staff-to-youth communication in areas such as body language, voice tone, facial expressions.

O’Brien202

Cluster 9 (comparison of outcomes)

The single BCT credible source (BCT 9.1) was identified from cluster 9 (comparison of outcomes) and constituted < 3% of all identified BCTs. This cluster was one of the least often detected, appearing in only 3% of interventions. This BCT was detected in 22% of criminal justice setting interventions but only 2% of mental health settings (Figure 25).

FIGURE 25.

Cluster 9 (comparison of outcomes) by setting (% of interventions in each setting containing BCTs from this cluster).

Credible source (behavior change technique 9.1)

Present verbal or visual communication from a credible source in favour of or against the behavior.

Michie et al. 52

Behavior change technique 9.1 (credible source) was identified in criminal justice settings204 and mental health settings. 122 Crosland et al. 204 described training being given by certified behaviour analysts, whereas in Holden et al. ’s study122 staff were asked to provide credibility by describing their experience of working with CYP.

Cluster 10 (reward and threat)

Three BCTs out of a possible 11 from cluster 10 (reward and threat) were identified {BCT 10.10 [reward (outcome)], BCT 10.5 [social incentive] and BCT 10.4 [social reward]}. These three BCTs accounted for 8% of all identified BCTs and were detected in 8% of all interventions, but at a rate of only 2% in mental health settings, compared with 22% in health and social care settings (Figure 26).

FIGURE 26.

Cluster 10 (reward and threat) by setting (% of interventions in each setting containing BCTs from this cluster).

Social incentive (behavior change technique 10.5)

Inform that a verbal or non-verbal reward will be delivered if and only if there has been effort and/or progress in performing the behavior (includes ‘Positive reinforcement’).

Michie et al. 52

The BCT social incentive (BCT 10.5) was identified in 4% of interventions (see Figure 26). Although social incentive for staff could take the form of individual recognition,178 for Schreiner et al. 41 the incentive was the potential improvement to the workplace:

Staff were enticed to participate in the project through potential benefits (e.g., a more pleasant patient living environment, a more pleasant work environment for staff, and decreased risk of patient and staff injuries).

Schreiner et al. 41

Social reward (behavior change technique 10.4)

Arrange verbal or non-verbal reward if and only if there has been effort and/or progress in performing the behavior (includes ‘Positive reinforcement’).

Michie et al. 52

The BCT social reward (BCT 10.4) was detected in all settings except generic, amounting to 8% of interventions (see Figure 26). Examples included recognition of best practices, celebration of successes,180,193 awards,119 weekly peer nominations100,125 and praise. 41,185 The example below from Hodgdon et al. 183 illustrates how social incentives and rewards could be identified in conjunction:

[Staff] encouraged each other . . . by providing reinforcement (giving a fellow staff member a chip) when they observed a skill being used with a client. At the end of a 2-week period, the staff with the most chips was recognized in staff meeting and also earned an incentive.

Hodgdon et al. 183

Reward (outcome) (behavior change technique 10.10)

Arrange for the delivery of a reward if and only if there has been effort and/or progress in achieving the behavioral outcome (includes ‘Positive reinforcement’).

Michie et al. 52

Similarly, reward (outcome) (BCT 10.10) was manifested as staff recognition164 in just 1% of interventions (see Figure 26). This BCT was detected in mental health settings only.

Cluster 11 (regulation)

The only BCT (of four possible) identified from cluster 11 (regulation) was BCT 11.2 (reduce negative emotions). This amounted to 3% of all BCTs identified. This BCT (and, therefore, cluster) was detected in 7% of interventions, but rates varied from 2% in mental health settings to 18% in generic settings (Figure 27).

FIGURE 27.

Cluster 11 (regulation) by setting (% of interventions in each setting containing BCTs from this cluster).

Reduce negative emotions (behavior change technique 11.2)

Advise on ways of reducing negative emotions to facilitate performance of the behavior (includes ‘Stress Management’).

Michie et al. 52

The BCT reduce negative emotions (BCT 11.2) was detected in 7% of interventions (Figure 27). Interventions offering post-incident support to involved staff and witnesses to reduce trauma and other negative outcomes were coded with this BCT:105,122,180,207

Organisations should have a person-centred policy for providing both immediate and longer-term support after any use of restrictive practices. The policy should indicate future options for accessing longer term support or counselling if needed. This may include individual and/or group supervision/debriefing and individual psychological therapy delivered by trained professionals.

Welsh Government. 180

Reproduced with permission. © Crown copyright. Contains public sector information licensed under the Open Government Licence v3.0

This BCT was identified where interventions encouraged reflection30,122 and self-regulation. 122,183 A secondary purpose of reducing negative emotions was preventing ‘workplace stress’ from adversely affecting practice. 180

Cluster 12 (antecedents)

Behaviour change techniques identified from cluster 12 (antecedents) accounted for 16% of all BCTs identified. Three BCTs were identified from this cluster: BCT 12.5 (adding objects to the environment), BCT 12.1 (restructuring the physical environment) and BCT 12.2 (restructuring the social environment). This cluster was the second most commonly detected across all settings, at a rate of 74%. BCTs from this cluster were identified in almost all interventions (93%) from the health and social care setting but in less than half (43%) in the education settings (Figure 28).

FIGURE 28.

Cluster 12 (antecedents) by setting (% of interventions in each setting containing BCTs from this cluster).

Restructuring the physical environment (behavior change technique 12.1)

Change, or advise to change the physical environment in order to facilitate performance of the wanted behavior or create barriers to the unwanted behavior (other than prompts/cues, rewards and punishments).

Michie et al. 52

Examples of restructuring the physical environment were found in 23% of interventions (see Figure 28). Multiple examples of changes to the physical environment were detected, many of which involved the installation of a sensory or comfort room or zone:33,77,114,129,156,164,165,183,194

The Unit’s sensory room contained a variety of sensory equipment, including a rocking chair, weighted blankets, fidget toys, scented oils, candy and teas, pictograph cards (i.e. flashcards depicting pleasant or calming images), music and projected images (e.g. bubbles floating or rivers running).

West et al. 164

The introduction or improvement of recreational spaces, such as playgrounds, was another physical change. 114,166 Some interventions also reported making physical changes to create a low-stimulation environment (i.e. with low noise and lighting)85,122 and decorative changes to improve the appearance and maintenance of the setting82,102,122,168,181,202 and personal spaces, such as bedrooms:122,172

The therapeutic environment intervention involved making inexpensive physical changes, including repainting walls with warm colors, placement of decorative throw rugs and plants, and rearrangement of furniture [. . .] The second intervention included replacing worn-out furniture.

Borckardt et al. 181

Others removed locks and improved free access to outdoor spaces to facilitate independence. 168,174,180

Restructuring the social environment (behavior change technique 12.2)

Change, or advise to change the social environment in order to facilitate performance of the wanted behaviour or create barriers to the unwanted behaviour (other than prompts/cues, rewards and punishments).

Michie et al. 52

Restructuring the social environment (BCT 12.2) was the second most common BCT detected, appearing in 69% of interventions (see Figure 28) and in three out of the five settings. The frequency with which this BCT was detected is partly attributable to its broad definition and how it was operationalised. The social environment was conceptualised in three ways: (1) milieu and culture, (2) communication and social interaction and (3) staffing and peer support.

Various interventions referred to changing the culture,89,120,174 for example ‘creating a culture of caring’,116 organisational culture change82,116 and creating a supportive and therapeutic milieu:82,110,183

Assessing for distress or threat within the therapeutic milieu of residential programs is [an important innovation]. Residential-based TST teams now have a system and format for regularly discussing [team functioning] and ascertaining whether individual team members (or the team as a whole) are effecting a distressed or threatening environment.

Brown et al. 78

These broad changes often included revising rules,82,178,181 clarifying expected behaviours,163,168,199 increasing freedoms33,89,176,203 and introducing restrictive practice reduction policies:210

The rules and language intervention included the establishment of a team for each unit that was tasked with reviewing and modifying unit rules and policies to be less restrictive to patients or eliminating unit rules that were too restrictive.

Borckardt et al. 181

This BCT was also coded where interventions described making changes to improve communication or social interaction. Some changes involved opportunities for children to socialise with each other89 or with their families:105,115

Youth and family commitment was a key to [. . .] success [. . .] The hospital has implemented many youth and family activities, which have enhanced their participation together at the facility. Many families were reimbursed for . . . travel to family meetings, or medical cabs were accessed. Visitation by families has been encouraged.

Azeem et al. 105

Others involved opportunities for staff to socialise with children, such as at mealtimes. 89,105,115,168 Often in conjunction with such social opportunities were improvements to communications with children and their families, such as via meetings or information-sharing:82,89,105,168,174,178,181

Other Sanctuary Model tools such as community meetings, where the ‘community’ of staff and girls would gather regularly, and red-flag meetings, arranged to defuse a potentially serious incident, were held consistently. Sanctuary was also incorporated into after-care planning and in work with the girls’ families.

Elwyn et al. 178

Azeem et al. 105 also mentioned providing access to interpreters.

A key aspect of improved communication as a form of restructuring of the social environment was the active involvement of children and their families in planning and setting goals in meetings,82,85,115,168,176,193,207 training sessions and evaluations. 89,116 Similarly, the involvement of front-line staff by management in planning and achieving change was coded to this BCT:89,116,178,211

Involve staff. Staff asked to make this change must be part of the process; otherwise early buy-in and support for the new direction will be compromised. Staff often know far better where the operational obstacles lie and have pragmatic ideas about how to address them.

Caldwell and LeBel116

Staffing changes, such as increased staffing during critical periods, and task-sharing were also coded as restructuring the social environment. 82,85,168,178,207 Colton82 reported that it was possible for organisations to influence the implementation of seclusion and restraint measures, provided that staffing levels were adequate at critical points in the day such as transitions, shift changes and evenings.

Changes to staffing and organisational culture were reinforced by managerial support and leadership,30,82,88,105,115,116,178,207 peer support,105,178 team building115 and the establishment of working groups, task forces and implementation teams:85,168,183

Development of strategies to improve interdisciplinary communication. A subcommittee of the task force set up multidisciplinary mini-meetings, where the work of the aggression-free task was discussed and feedback was encouraged.

Goren et al. 168

Adding objects to the environment (behavior change technique 12.5)

Add objects to the environment in order to facilitate performance of the behavior.

Michie et al. 52

Adding objects to the environment (BCT 12.5) was detected in 11% of interventions (see Figure 28), such as where sensory equipment33,39,105,114,129,156,164,165,194 and recreational/occupational equipment105,114,115 were provided as part of the intervention. Kaltiala-Heino et al. 198 also described the requirement for staff to carry an alarm device.

Cluster 13 (identity)

Two out of the five BCTs in cluster 13 (identity) were identified, accounting for 6% of all BCTs identified. These BCTs were BCT 13.1 (framing or reframing) and BCT 13.2 (identification of self as role model) and they were identified in 48% of interventions overall (although the rate ranged from 29% to 64%) (Figure 29).

FIGURE 29.

Cluster 13: Identity by setting (% of interventions in each setting containing BCTs from this cluster).

Identification of self as role model (behavior change technique 13.1)

Inform that one’s own behavior may be an example to others.

Michie et al. 52

The BCT identification of self as role model (BCT 13.1) was detected in 24% of the interventions (see Figure 29). Several interventions used the concept of a role model or role modelling by managers,108,160,178 team leaders174 and other key staff:41,183

Managers modelled responses to challenging situations in ways that did not include the use of physical restraints.

Sanders108

In some interventions, trained staff took on the role of internal expert, trainer or champion:115,168–170,178,194

Develop a cadre of staff who can share their expertise by providing training to their colleagues, thus expanding our still-limited training resources.

Girelli210

Often this drew on a train-the-trainer model79,118,172,176,183 or mentoring. 128,180,213

Framing/reframing (behavior change technique 13.2)

Suggest the deliberate adoption of a perspective or new perspective on behavior (e.g. its purpose) in order to change cognitions or emotions about performing the behavior (includes ‘Cognitive structuring’).

Michie et al. 52

Framing/reframing (BCT 13.2) was another commonly identified BCT, detectable in 42% of interventions (see Figure 29). Many interventions sought to encourage staff using framing/reframing and often this was informed by a specified model or approach, such as strengths-based care,40,105,116,214 trauma-informed care,78,85,105,114,116,120,122,171,172,177,178,181,183,190,193,207 recovery-oriented care,77,99,188,193 CPS92,112 and positive behavioural support:100,126

For trauma-informed care, all unit staff attended a half-day standardized training seminar on the nature of trauma and its effects on patients’ experiences, physiology, and psychological processes, along with instructions on how to minimize engaging in behaviors that could exacerbate trauma related reactions from patients.

Borckardt et al. 181

Within or in addition to these approaches, staff were encouraged to reframe their views of the children they worked with, such as the origins of their challenging behaviour174 or their expectations of their abilities:

Through debriefings, case processing, and team meetings, we found that staff interventions were based on the patient’s chronological and/or physical age rather than on the patient’s cognitive/social/emotional age. [. . .] We educated staff . . . for assessing and reassessing a patient’s developmental ages and discussed appropriate developmental age expectations and interventions.

Fralick176

Similarly, interventions encouraged framing/reframing of staff views of children’s behaviour and its meaning. 110,168,180 Finally, staff were encouraged to reframe their beliefs about restraint, for example as a ‘last resort’33 and as harmful and not therapeutic. 194 They were also encouraged to view their relationships with the children as a therapeutic tool. 170,210 Framing/reframing was usually initiated through training. 33,39,111 At the organisational level, framing/reframing could be detected in descriptions of culture change89,108,193 and paradigm shifts. 91,115

Cluster 14 (scheduled consequences)

The single BCT identified from the 10 in cluster 14 (scheduled consequences) accounted for < 3% of all BCTs identified. Cluster 14 was the least often detected (2% of interventions). It was not detected at all in education, criminal justice or generic settings, and at very low rates in mental health and health and social care settings (Figure 30).

FIGURE 30.

Cluster 14 (scheduled consequences) by setting (% of interventions in each setting containing BCTs from this cluster).

Remove punishment (behavior change technique 14.10)

Arrange for removal of an unpleasant consequence contingent on performance of the wanted behavior (includes ‘Negative reinforcement’).

Michie et al. 52

Only one BCT from cluster 14 (scheduled consequences) was identified: remove punishment (BCT 14.10). This BCT was detected in interventions from mental health settings only, amounting to 2% of all interventions (see Figure 30). This BCT captures the avoidance of undesirable consequences if the desired behaviour is performed. Three interventions described creating a non-punitive environment and debriefing sessions being conducted in a non-punitive and supportive manner, without attributing blame. 33,174,193

Undetected behavior change techniques and behavior change technique clusters in interventions

Behavior change techniques from cluster 15 (self-belief) and cluster 16 (covert learning) were not detected in any of the interventions.

A total of 56 (60%) individual BCTs were not identified in any of the interventions (see Appendix 7). It should be noted that for a BCT to be identified in an intervention, evidence of its presence needed to be documented in the intervention materials. Therefore, there may be instances in which a BCT remained unidentified because of a lack of evidence.

Conclusion

Application of the BCT taxonomy to 82 interventions that sought to reduce restrictive practices in children’s settings identified 36 out of a possible 93 BCTs. The most frequently BCT identified was BCT 4.1 (instruction on how to perform the behaviour), reflecting the high use of training within interventions.

Intervention procedures

The majority of the interventions used more than one procedure (e.g. staff training and data review), and, as such, it was not possible to identify which BCTs were found within individual procedures. However, 20 of the interventions used only one procedure (six different procedures were used only once); therefore, it was possible to look at what BCTs were used in relation to these six procedures. The six different procedures and the number of interventions they were used in are shown in Table 24.

Training was the most frequently used single procedure and was used in nine interventions. The most frequently identified BCTs were BCT 4.1 (instruction on how to perform the behaviour), BCT 12.2 (restructuring the social environment), BCT 1.1 (problem-solving), BCT 1.4 (action-planning), BCT 2.7 (feedback on outcomes of behaviour) and BCT 13.2 framing/reframing.

The rest of the procedures were used in only a small number of interventions. Data review, risk assessment and guidelines or policy change were all identified as lone procedures in three interventions each. The most frequently identified BCTs in data review studies were BCT 4.1 (instruction on how to perform the behavior), BCT 12.2 (restructuring the social environment), BCT 2.7 [feedback on outcome(s) of behavior], BCT 1.4 (action-planning) and BCT 1.2 (problem-solving). The BCTs associated with the introduction of risk assessment were BCT 4.1 (instruction on how to perform the behavior), BCT 12.2 (restructuring the social environment), BCT 2.7 [feedback on outcome(s) of behavior], BCT 1.4 (action-planning), BCT 1.2 (problem-solving) and BCT 13.2 (framing/reframing). Three studies used guidelines and policy change as a lone procedure. The BCTs most frequently identified in these three studies were BCT 2.7 [feedback on outcome(s) of behaviour], BCT 1.4 (action-planning), BCT 1.2 (problem-solving) and BCT 13.2 (framing/reframing).

Milieu changes and staff involvement alone were used in one study each. Apart from the restructuring and instruction, the most frequent BCT related to milieu change was BCT 13.2 (framing/reframing). The BCTs related to staff involvement were BCT 13.1 (identification of self as role model), BCT 3.2 [social support (practical)], BCT 1.4 (action-planning), BCT 2.7 [feedback on outcome(s)] and BCT 13.2 (framing/reframing).

When BCTs used in all interventions, regardless of the number of procedures, were collected, all had BCT 4.1 (instruction on how to perform the behavior) and BCT 12.2 (restructuring the social environment) as the two most commonly identified. When the remainder of the interventions are looked at by the procedures, some differences can be seen. (It was not possible to disaggregate any of these procedures from the others used simultaneously, so these BCTs must be viewed in that context.)

Figure 31 shows the BCTs identified in ≥ 50% of the interventions according to the procedures they contained. The only other BCT used in > 50% of interventions using sensory approaches was BCT 7.1 (prompts/cues). Similarly, only BCT 13.2 (framing/reframing) was commonly used in interventions using milieu changes. Over half of interventions using staffing changes used BCT 13.2 (framing/reframing) and BCT 13.1 (identification of self as role model). BCT 1.4 (action-planning) was found across all the interventions except these three. The interventions that used environmental changes and promoted involvement of CYP and their families used a higher number of BCTs than those that did not.

FIGURE 31.

Different BCTs identified in interventions using specific procedures.

Behaviour change techniques and outcomes

One objective of the study was to identify and prioritise BCTs that show promise of effectiveness for testing in future interventions. Sixty-seven of the interventions had been subject to evaluation and, of these, 29 had reported statistically significant positive findings and 38 had not. Table 25 shows the different BCT content of interventions that had reported positive findings and those that did not.

The ranking of the BCTs is very similar. The two most frequent are the same, although the percentage of interventions with positive outcomes is higher for BCT 4.1 (instruction on how to perform the behavior) than for BCT 12.2 (restructuring the social environment), both found in > 60% of all interventions. Those interventions without positive findings featured BCT 1.4 (action-planning) third most frequently, used in 66% of those without positive findings compared with only 33% of those with positive findings. In those interventions with positive findings, the most frequent BCTs were BCT 2.7 [feedback on outcome(s) of behavior] and BCT 1.2 (problem-solving). Eighteen per cent more of the unsuccessful interventions used BCT 2.7 than those reporting positive outcomes.

The evaluations used a broad range of outcome measures, as described in Chapter 4. This makes it difficult to compare interventions on their specific outcomes. However, the two most frequently reported outcomes were (1) a reduction in the number of incidents of restraint (used by 30% of interventions; n = 20) and (2) a reduction in the number of incidents of seclusion (used by 14% of interventions, n = 9). Eighteen evaluated interventions reported a significant reduction in incidents of restraint and six reported a significant reduction in incidents seclusion. The BCTs identified in > 20% of these interventions are shown in Figure 32.

FIGURE 32.

The BCTs identified in > 20% of interventions that successfully reduced incidents of restraint and seclusion.

As seen in Table 26, the two most frequently identified BCTs are again BCT 4.1 (instruction on how to perform the behavior) and BCT 12.2 (restructuring the social environment), with BCT 2.7 [feedback on outcome(s) of behavior] ranking third for both outcomes. BCT 13.2 (framing/reframing), BCT 13.2 [social support (practical)], BCT 1.4 (action-planning) and BCT 1.2 (problem-solving) were identified in higher percentages of interventions that reduced incidents of seclusion, as well as BCT 2.2 (feedback on behavior), BCT 2.3 (self-monitoring of behavior) and BCT 1.9 (commitment), which were absent from > 20% of the interventions that reduced incidents of restraint.

The list of identified BCTs does not account for the multiple procedures within these interventions, nor does it account for the small number of studies that reported significant reduction in incidents of seclusion (n = 6). When these are compared with all interventions, BCT 1.4 (action-planning) drops rank from third to fifth with both groups of interventions that reduced restraint and seclusion and BCT 2.7 [feedback on outcome(s) of behavior] and BCT 1.2 (problem-solving) are ranked third and fourth. The ranking of the two most frequent BCTs remains unchanged.

The intervention evaluations were implemented in different settings. Those that reported significant positive findings were separated by setting and the BCTs were examined (criminal justice was excluded from this analysis as there was only one case). Mental health and social care settings were both dominated by BCTs 4.1 (instruction on how to perform the behaviours) and 12.2 (restructuring the social environment).

The interventions designed for use in generic settings used the most BCTs. Notably, all used BCT 2.7 [feedback on outcome(s) of behavior]. Educational settings were the only setting where over half of the interventions used BCT 7.1 (prompts/cues) to change staff behaviour (Figure 33).

FIGURE 33.

The BCTs identified in > 50% of interventions reporting significant positive findings by setting.

Mechanisms of action

Mechanisms of action are theoretical constructs that explain how BCTs affect behaviour. Twenty-six mechanisms of action, derived from both the theoretical domains framework and a systematic review of 83 behaviour change theories, have been specified. Understanding how specific BCTs change behaviour can support intervention theory development and testing and inform increasingly effective and efficient interventions. The Theory and Technique tool215 was used to identify the mechanisms of action for which there were identified links with the BCTs most frequently identified in studies reporting positive findings. These are summarised in Table 27 in addition to the mechanisms of actions through which they are theorised to work.

The most common mechanisms of action related to BCTs used in studies with positive significant findings was environmental control and resources. This is defined as changes to ‘aspects of a person’s situation or environment that discourage or encourage the behaviour’. 216 BCTs linked with this were 3.2 [social support (practical)], 7.1 (prompts/cues), 12.1 (restructuring the physical environment) and 12.2 (restructuring the social environment).

The second most frequently occurring mechanism of action was behavioural cueing, defined as the process by which a behaviour is triggered from the external environment, the emergence of ideas or the performance of another behaviour. The BCTs that were related to this mechanism were 1.4 (action-planning), 7.1 (prompts/cues) and 12.1 (restructuring the physical environment).

The third most common mechanisms of action were knowledge, behavioural regulations and belief about capabilities. Knowledge was targeted by two BCTs: BCT 4.1 (instruction on how to perform the behavior) and BCT 4.2 (information about antecedents). Behavioural regulation was targeted by BCT 4.2 (information about antecedents) and BCT 1.2 (problem-solving). Belief about capabilities included BCT 1.2 (problem-solving) and BCT 4.1 (instruction on how to perform the behavior).

This more detailed look at where different BCTs were used has shown that where procedures had been used on their own interventions they showed little variation in the two most popular BCTs, 4.1 (instruction on how to perform the behavior) and 12.2 (restructuring of the social environment). After these two there was some variation by procedure. Exploration of outcomes was also hampered by the use of many different outcome measures used in interventions. When studies reporting positive findings in the two most frequently used outcomes (incidents of restraint and incidents of seclusion) were explored, these showed that the same BCTs were in most frequent use in interventions that successfully reduced incidents. The mechanisms of action related to the most frequently identified BCTs were those relating to changes to a person’s broad social and physical environment to encourage desired behaviours. These relationships are summarised in Table 27.

Introduction

In March 2020, the UK press217 reported that the Ministry of Justice was being challenged by Article 39, a children’s charity, to release details of child restraint in prisons. The Ministry of Justice was asked to release records giving the reasons for the use of pain-inducing restraint [i.e. ‘managing and minimising physical restraint’ (MMPR)] had been used, on 260 occasions during 2017–18, in young offender institutions and secure training centres. 217

The use of restrictive practices with children remains a pertinent, ongoing issue. Children who are already vulnerable may be at higher risk of being restrained. Wisdom et al. 114 reported that, according to the New York Office of Mental Health, children were five times more likely than adults to be put in restraint or seclusion. Restraints seem to be performed more frequently on elementary/primary students (aged 5–11 years) than their older peers. 113,218

Overall, the literature on interventions to reduce restrictive practices in CYP’s settings presents a complex picture. There is a lack of consistency and comprehensiveness in reporting. This results in an evidence base that is of limited value for informing decisions about the development, implementation and evaluation of interventions. There is a strong tendency for interventions to have multiple components, and this, combined with the poor reporting, further limits the potential of the evidence base to inform practice. The study results therefore suggest an urgent need to simplify and streamline interventions and intervention reporting.

Analysis of the robustness of the results

As discussed (see Effectiveness), the review was unable to identify robust evidence of effectiveness and could only draw tentative conclusions regarding four promising BCTs. Robust conclusions may be drawn concerning the frequency of use of specific intervention components, notably staff training and programmes that restructure the social or physical environment. Furthermore, the BCT analysis found that the most frequently occurring BCTs tended to occur in the same interventions that showed promising outcomes. Interventions are listed by study design in Appendix 8.

Strengths

To our knowledge, no previous reviews have looked at both the outcomes and components of interventions that aim to reduce restrictive practices in CYP’s institutional settings. This study aimed to identify the most promising intervention components and recommend them to be tested within a trial setting, with a view to seeking future funding to develop and test an intervention based on the results of the review. The recommendations generated by this review are transferable, and this is one of its strengths: institutions that have children in their care could benefit from interventions that are better defined, more acceptable to children and staff, more completely and accurately implemented, and more cost-effective.

Table 28 compares and contrasts key findings from the current review of the use of restrictive practices in children’s institutions settings, with key findings from the companion review of the use of restrictive practices in adult mental health settings. 46

Limitations

The search strategy combined traditional search techniques for retrieving research and grey literature, with a scanning approach to identify potential alternative sources of relevant material. This had the advantage of enabling the retrieval of diverse records that reported intervention content and was useful for mapping the number and range of interventions; however, the diverse quality of reporting in some records retrieved in this way presented a challenge for the meaningful assimilation of findings. For example, a lack of detailed description of interventions may have masked the presence of BCTs, that consequently were not detected.

The adoption of a broad approach to searching and inclusion criteria led to the inclusion of a wide range of interventions in diverse formats. No criteria for exclusion on the basis of quality were developed. This is substantially different from normal systematic reviews of evidence. However, current practice in this area is of adopted interventions without a clear evidence base. There may remain some institutional settings/interventions which were not captured in our search as a result of databases searched.

The literature search was restricted to English-language records and there was limited evidence from countries outside the USA, so the findings may have limited international transferability. As noted in one of the included studies, there was a:

[C]lear divide between the numbers of published studies coming out of America compared to the rest of the world [which was] likely to change in the coming years.

Wilson et al. 230

Reproduced with permission under the CC BY-NC-ND 4.0 licence

The scope of the study did not allow for detailed analysis of effectiveness by setting, population group, culture, national context, or institutional ethos. Contextual factors, for example criminal justice settings or service users with intellectual disability, are likely to be highly relevant to decision-making regarding interventions. In addition, the finding that the evidence was weak restricted the scope of the study to examine the effectiveness of BCTs used in interventions. In these terms, transferability of the tentative conclusions about effectiveness has yet to be demonstrated. These are relevant areas for future research.

Implications for policy and practice and future research

The need to reduce restrictive practices in CYP’s institutional settings is ongoing: this should be considered a priority for policy-makers. However, without clarity about current use of restrictive practices in CYP’s institutional settings, evaluation of interventions could remain problematic.

It is clear that some groups may experience more severe restrictive practices than others; therefore, better understanding of the influence of gender, ethnicity, disability, and institutional setting and its governance is required. Exploration of how interventions can be adapted to reduce restrictive practice for different contexts is urgently needed.

Accessible guidelines for a core outcomes set that is feasible for researchers and practitioners to use in real-world settings could be a valuable step towards improving practice in CYP’s settings.

With regard to both the more widely used interventions and the stand-alone interventions, there appears to be little appetite for simplicity. Simple interventions would facilitate like-for-like comparison so that it would be more possible to identify key ingredients and understand what works and what does not work. However, practice experts may be aware of factors that suggest complex interventions are suitable; therefore, interventions developers could consider all relevant factors (i.e. theoretical, practical, contextual) to ensure optimum conditions for delivery and evaluation of effective interventions. Alternatively, if complex interventions are inevitable, recent guidance suggests better reporting of complexity considerations,234 and better reporting of intervention development generally. 235

Those interventions that are developed to reduce staff use of restrictive practices need to be better defined, with clear links to theory, and contain more robust and rigorous approaches to evaluation. Specifically, it may be worthwhile to address the question of why so much training is directed at staff, many of whom are likely to have been previously trained to work in the setting. It may also be worthwhile to address data-monitoring, which is a component of many of the interventions; it has the potential to activate powerful psychological mechanisms such as shame and social norms when combined with feedback, and could be a quick and straightforward means of generating useful data. Key practice, policy and research recommendations are presented in Table 29.

Conclusion

Despite numerous enquiries, policy initiatives and recommendations, there remains ongoing concern about the use of restrictive practices in CYP’s institutional settings. The impact of restrictive practices on the psychological and physical welfare of both CYP and staff should not be underestimated. The care of CYP will remain suboptimal unless there is a sustained focus on reducing these practices. Without a sustained effort, these practices will continue to occur in institutional settings worldwide.

This study has generated, to our knowledge, the first known synthesis of the evidence on the content and effectiveness of interventions to reduce restrictive practices in CYP’s institutional settings. This synthesis provides a useful resource for practitioners, policy-makers and researchers aiming to implement or develop a restraint reduction intervention.

The new information generated adds to the research evidence base in the form of a comprehensive description of interventions, their components, context and outcomes, as far as can be ascertained from the limited evidence. The limitations are important because they suggest a need for caution in the use of interventions assumed to be effective.

Taken as a whole, these suggestions from our research for practice, policy and research can inform the development and testing of different models to reduce restrictive practices across a range of CYP’s institutional settings. They have the potential to affect the everyday practice of professionals working with children by supporting management decision-making in children’s services with regard to staff training and other interventions. The new insights generated from this study could lead to improved therapeutic outcomes, organisational efficiencies arising from reduced staff sickness and litigation costs, and better subjective experiences of children and staff. These could contribute towards improving the health and safety of vulnerable children and the staff who work with them in institutional care by protecting them from trauma, injury and death, and could thus benefit wider society.

Contributions of authors

John Baker (https://orcid.org/0000-0001-9985-9875) was responsible for leading the study and the protocol development, and contributed to literature searches, data extraction and analysis, and writing and editing the report.

Kathryn Berzins (https://orcid.org/0000-0001-5002-5212) was responsible for co-ordinating the study and contributed to literature searches, data extraction and analysis, BCT mapping, and writing and editing the report.

Krysia Canvin (https://orcid.org/0000-0001-6571-6411) contributed to literature searches, data extraction and analysis, BCT mapping, and writing and editing the report.

Sarah Kendal (https://orcid.org/0000-0001-8557-5716) contributed to literature searches, data extraction and analysis, BCT mapping, and writing and editing the report.

Stella Branthonne-Foster (https://orcid.org/0000-0002-5545-9199) was our PPI representative throughout the project, and contributed to the public involvement, the Plain English summary and the dissemination strategy.

Judy Wright (https://orcid.org/0000-0002-5239-0173) developed and advised on the search strategies and contributed to the final report.

Tim McDougall (https://orcid.org/0000-0002-9843-9315) provided expertise and contributed to writing and editing the final report.

Barry Goldson (https://orcid.org/0000-0002-9714-868X) developed and advised on the search strategies and contributed to the final report.

Ian Kellar (https://orcid.org/0000-0003-1608-5216) provided expertise in regards to BCT mapping, writing and editing the report.

Joy Duxbury (https://orcid.org/0000-0002-1772-6874) developed and advised on the search strategies and contributed to the final report.

Data-sharing statement

All available data can be obtained from the corresponding author.

Disclaimers

This report presents independent research funded by the National Institute for Health and Care Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, the HSDR programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, the HSDR programme or the Department of Health and Social Care.

Academic database searches

Search terms