Development and evaluation of a de-escalation training intervention in adult acute and forensic units: the EDITION systematic review and feasibility trial

Inclusion criteria

Studies were included that met the following criteria:

Population: healthcare staff (qualified and unqualified) working with adult populations (18–65 years) with mental health difficulties.

Intervention: training with a de-escalation techniques component.

Comparison: all controlled studies meeting eligibility criteria, irrespective of their control condition. Comparisons of two or more active interventions or of an active treatment with a ‘no treatment’ comparator will be included.

Outcomes (effectiveness): changes in rates of conflict (e.g. violence, verbal and physical aggression) and use of restrictive interventions (e.g. seclusion and physical restraint) post training. Secondary outcomes were cognitive (e.g. knowledge), affective (e.g. confidence) and skills-based (e.g. de-escalation performance) changes in trainee performance.

Outcomes (acceptability): Defined, for the purpose of the review, as intervention uptake, adherence and participant satisfaction and views.

Design: randomised controlled trials (RCTs), nRCTs, controlled before-and-after (CBA) and interrupted time series (ITS) designs. All studies that (1) asked trainees or trainers for their views of interventions using qualitative or quantitative methods, and/or (2) studies that quantitatively assessed non-participation, withdrawal, or adherence rates.

Rates of conflict

In total six studies of moderate quality provided outcome data related to rates of conflict. One study measured impact of training on conflict broadly (including aggression, self-harm, absconding, rule-breaking, medication refusal, drug and alcohol use) and found a significant effect (effect size (ES) 0.13) in reducing conflict. 10 This finding represented a 44% decrease in verbal aggression, a 53% decrease in physical aggression, a 72% decrease in self-harm and a 43% reduction in absconding attempts. 10 However, when this study was repeated using a controlled design, no effects on conflict outcomes were detected. 15

Another study reported reductions in patient assaults post training but did not test significance. 58 Two studies measured patient assault-related injuries. One found a reduction but did not test significance16 and the other found a statistically significant reduction in rate of injuries, although the effect size was not calculable. 17 Two studies found negative effects of training in relation to assaults on staff. One study measured impact of training on what they referred to as ‘disturbed behaviour’ (defined as any untoward incident involving patient behaviour that met the threshold for incident reporting) and found a non-significant negative effect of training between intervention and control wards. 35 The hazard ratio for staff assaults in the de-escalation training arm of the study was 48% higher than for staff in the usual care arm (control and restraint training). 35 One study found a significant, medium-sized effect (0.64) increasing rates of assaults post training, attributed to increased patient acuity in the follow-up period. 53

In total, 12 weak-quality studies provided outcome data related to rates of conflict. One study, again, measured conflict broadly and found no statistically significant difference in rates between trained and untrained wards. 15 One study found a significant overall reduction in conflict59 but the omission of SD prevented calculation of effect size. One study only referred to the measurement of ‘incidents’ but did not define what this meant or test the significance of a reported reduction in frequency post training. 26 Two studies found statistically significant reductions in assaults compared with control conditions, one study measuring rates at ward level47 and one following up individual trainees. 32 Neither study reported the required statistics to calculate effect sizes. Two studies found reductions in patient assaults compared with control conditions but neither reached statistical significance. 34,60 One study found a reduction in assaults comparing 12 months pre and post training but did not test significance. 25

Two studies measured aggression, incorporating verbal and physical aggression and, again, both found reductions compared with control conditions that did not reach statistical significance. 50,56 One study measured actual and threatened physical aggression separately and found a negligible reduction in actual aggression and an increase in threatened aggression although the statistical significance of neither finding was reported. 36 One study measured verbal aggression alone and reported a reduction without evaluating statistical significance. 41

Rates of containment

In total, four studies of moderate quality provided outcome data on rates of containment. One uncontrolled pre and post study measured impact of training on rates of ‘restraint’ (definition limited to ‘physical restraint’ interpreted as ‘manual restraint’ excluding seclusion or mechanical or chemical restraint). They found a significant reduction in frequency of restraint usage at 3 months (representing an 85% decrease) and 6 months (99% decrease) follow-up which was maintained at 9 and 12 months. 33 Two studies found no effect on containment use. One study measured containment broadly including manual restraint, seclusion, rapid tranquilisation, time out, PRN (extra psychotropic medicines), transfer to psychiatric intensive care, special observations (intermittent or continuous) and show-of-force and found no reduction associated with training. 10 The other moderate-quality study reporting no effect measured impact on PRN only. 53

In total, five studies of weak methodological quality provided outcome data on rates of containment. One uncontrolled pre and post study in two wards measured impact of training on frequency and duration of ‘restraint’ (definition included manual and mechanical restraint) and ‘seclusion’ (defined as confinement of a patient in a ward area from which they cannot freely leave). Both the number and duration of restraint and seclusion episodes in the training period (10 months) and in the post-training period (21 months) were significantly reduced compared with the pre-training period (21 months). 24 Because of the omission of SDs in their reporting, calculations of effect sizes were not possible. One study, which again used the broad definition of ‘containment’, measured rates pre and post training using a controlled design (three training wards and five control (no intervention) wards). 15 On the intervention wards there was a significant, medium-sized (0.33) pre and post reduction in containment. There was no overall difference in containment detected between intervention and control wards but there were two significant effects detected at individual containment item level (‘PRN’ and ‘Transfer to PICU’). No SDs were reported for the comparisons with control wards for these two items, so we were unable to calculate effect sizes.

An uncontrolled pre and post study in two wards measured the effect of training on ‘coercive measures’ but did not provide further definition. 47 They found a significant reduction in rates of coercive measures between the pre-training period (3 months) and the post-training period (3 months). However, because of the lack of reported means and SDs, no effect size calculation was possible. An additional, uncontrolled pre and post study measured rates of manual restraint and rapid tranquilisation associated with training. 34 They found a significant reduction in both manual restraint (0.4) and rapid tranquilisation (0.52) after training, once differences in patient demographics between pre and post periods (6 months pre, 6 months post training) were adjusted for. However, the confidence intervals (CIs) for these effects were wide (manual restraint CI 0.17 to 0.94; rapid tranquilisation CI 0.23 to 1.21). The final study, rated as methodologically weak, was an uncontrolled pre and post study in a single ward, which measured rates of ‘restraint’ and ‘seclusion’ (no more specific definition provided for either) associated with training. 62 They found a small (self-reported, the necessary statistics for calculation were omitted) non-significant effect of training in reduced rates of both seclusion and restraint.

Cognitive outcomes

Knowledge outcomes: In total, five studies of moderate methodological quality provided outcome data on changes in trainee knowledge following training. Four of these studies found significant improvements post training. 16,31,51,53 Two of these studies provided sufficient statistical information to calculate effect sizes: one reported a medium-sized effect (0.73)16 and one reported large effects (effect sizes ranged from 1.13 to 2.2 between different staff subgroups). 31 One study did not assess the significance of reported improvements in trainee knowledge. 28 Although these studies rated as moderate methodological strength according to the ROBINS-I, there were some specific problems with measurement that should be considered in the interpretation of training impact on this outcome. All five studies developed study-specific scales to measure knowledge outcomes and only one study provided any robust evidence of internal consistency and reliability [Ilkiw-Lavelle31 (Cronbach’s alpha 0.86)].

In total, six studies of weak methodological quality provided outcome data on changes in trainee knowledge. 5,19,21,36,37,41 Only two studies measured the significance of changes: one found a significant effect of training, but effect size calculation was not possible owing to the omission of a SD. 5 The other found a non-significant improvement in knowledge. 21 No weak study tested the internal consistency of their measure of de-escalation knowledge.

Affective outcomes: In total, 15 studies of moderate methodological quality and 11 studies of weak methodological quality reported on changes associated with trainee affective outcomes associated with training. These studies measured a range of relevant outcomes relating to emotion regulation, confidence and self-efficacy, fear and anxiety, attitudes, stress and burnout and empathy.

Emotion regulation: One moderate-quality study measured ‘emotion regulation’ which they defined as ‘awareness and control of feelings, especially fear and anger’ (p. 398)9 and found a significant, large effect (2.24) of training in enhancing this outcome (follow-up measurements were conducted between 3 and 6 months post training). Two weak studies measured emotion regulation outcomes. One weak study61 used a non-randomised controlled design and measured emotion regulation immediately pre and post training in role-play scenarios. They used an observer rated scale [including the following subscales: ‘calm’, ‘upset’, ‘angry’, ‘relaxed’, ‘defensive’ (p. 111)] and a self-report scale [including the following subscales: ‘anger’, ‘upset’, ‘anxiety’, ‘in control’ (p. 109)]. They found significant and large (0.76–5.24) pre-and-post effects of training on observer-rated items including ‘relaxed’, ‘defensive’, ‘upset’ and ‘calm’ but no significant change on ‘angry’. However, there were significant, medium to large (0.69–2.81) effects observed in the control group on ‘relaxed’ and ‘upset’, indicating a significant impact of practice effects on emotion regulation. No between-group analysis was conducted. In terms of self-report outcomes, they found significant, large effects in the trained group (0.79–1.5) on ‘anger’ and ‘in control’. There were no significant self-rated changes in the control group.

The other weak study measuring emotion regulation7 used an uncontrolled pre and post design and measured the following item related to emotion regulation (of a 20-item questionnaire measuring attitudes to aggression in student nurses): ‘When a patient gets aggressive, I get so nervous I can hardly think straight’ (p. 208). They found no effect of training on this item. Of the three studies measuring emotion regulation outcomes, only Bjorkdahl et al. 9 provided any evidence of internal consistency (Cronbach’s alpha 0.83).

Confidence and self-efficacy: Nine moderate-quality studies measured either trainee confidence16,23,28,43,46,49,51,53 or self-efficacy. 39 Five of these were non-randomised controlled studies23,39,46,49,53 and four uncontrolled pre and post studies. 16,28,43,51 Five studies measured these outcomes only immediately pre and post training. 16,28,46,51,53 Four studies included an additional longer-term follow-up, one at 2 weeks post training,43 one at 8 months,49 one at 6 months post training39 and one at 18 months. 23 All nine studies reported improvements in these outcomes associated with training at follow-up, eight found statistically significant improvements and one study53 failed to assess the significance of the improvement. Four studies provided the necessary statistical information to calculate effect sizes and, of these, three produced small effects (range 0.19–0.34)16,28,39 and one a large effect (0.87). 46 Improvements in confidence were retained in all studies that included a longer-term follow-up. 23,39,43,49

Nine studies of weak methodological quality provided outcome data related to confidence or self-efficacy. Eight measured confidence6,7,19,22,26,36,37,41 and one measured self-efficacy. 54 All nine studies used uncontrolled pre and post designs. Six studies measured these outcomes immediately pre and post training7,22,26,37,41,54 and three studies included longer-term follow-up time points. One study had an additional follow-up at 6 months,19 one study had four time points across 8 months (two pre, two post training, no more specific detail provided)6 and the final study measured confidence 12 months pre and post training. 36 All nine weak studies found improvements in confidence or self-efficacy but only five assessed significance6,7,22,37,54 and the statistical reporting in these five studies did not allow for effect size calculation. All three weak studies including longer-term post-training follow-ups found that improved confidence had been retained. 6,19,36 Of the 18 studies measuring confidence or self-efficacy outcomes, nine used validated measures,16,22,28,37,39,43,49,53,54 with Cronbach’s alphas ranging between 0.7153 and 0.97. 39

Fear and anxiety: Two moderate-quality studies measured outcomes related to staff fear or anxiety. 39,44 Both were non-randomised, controlled studies. One study measured these outcomes immediately pre and post training44 and the other measured pre and post and included an additional 6-month follow-up time point. 39 McIntosh39 measured ‘anxiety arousal’, which was defined as the level of concern about patient-perpetrated violence experienced by staff at work, and found no effect of training on this outcome post training or at 6 months follow-up. Nau et al. measured ‘fear’ (no further description provided) as a subscale of the observer-rated DABS. 44 They found significantly reduced fear from pre and post within trained subjects as well as between trained and untrained groups. The statistical reporting did not permit calculations of the sizes of these effects.

One weak-quality study measured outcomes related to staff fear. This study adopted a non-randomised uncontrolled design36 and measured ‘Fear when managing aggressive or potentially aggressive patients’ (p. 212) 12 months pre and post training. They found reduced fear at 12 months but failed to assess the significance of this reported effect. Two out of three studies measuring fear or anxiety outcomes provided evidence of the internal consistency of scales used. Nau et al. 44 reported a Cronbach’s alpha of 0.87 for the DABS scale and McIntosh (2003) reported a Cronbach’s alpha or between 0.88 and 0.97 for the Self-efficacy scale. 39

Attitudes: Six moderate-quality studies measured outcomes related to trainee attitudes. 4,10,16,28,29,49 Two of these were non-RCTs29,49 and four adopted uncontrolled pre and post designs. 4,10,16,28 Five studies measured attitudes immediately pre and post training, with only one study including an additional 3-month follow-up. 49 Three studies measured trainee attitudes to patient aggression,28,29,49 two studies measured attitudes to using de-escalation4,16 and one study measured trainee attitudes to personality disorder. 10

Two moderate-quality studies reported significant effects of training in improving trainee attitudes to aggression. 28,49 Grenyer et al. 28 found significant effects of training on four of eight subscales of the ‘Attitudes to Aggressive Behaviour Questionnaire’, with effect sizes ranging between 0.33 and 0.97. Needham et al. 49 found a significant effect (sample size calculation was not possible due to statistical reporting) of training on a visual analogue scale which measured positive and negative attitudes towards patient aggression, but no effect of training on the ‘Perception of Aggression Scale’. One study found no effect of training on trainee attitudes. 29 Both studies measuring attitudes to using de-escalation found significant effects of training;4,16 a medium effect size of 0.39 was calculable for one of these studies. 16 The only study measuring attitudes to personality disorder found no effect of training. 10

Three weak-quality studies measured outcomes related to trainee attitudes. 19,41,47 All three studies used uncontrolled pre and post designs and all measured attitudes to patient aggression immediately pre and post training, except Collins,19 who included an additional 6-month follow-up time point. Only Needham et al. 47 found a significant effect of training on ‘Subjective perceptions of the severity of aggressive incidents’ (effect size calculation was not possible). Only three of nine studies measuring trainee attitudes provided a rigorous test of internal consistency. 16,29,49

Cronbach’s alphas for the Management of Aggression and Violence Scale (MAVAS) ranged from 0.25 (situational factors), 0.41 (external factors), 0.54 (internal factors) to 0.71 (management approach) on its four subscales. 29 Needham et al. 49 reported Cronbach’s alphas of 0.69 (factor 1) and 0.67 (factor 2) for the two subscales of the Perception of Aggression Scale (POAS). The final study, Calabro et al.,16 reported a Cronbach’s alpha of 0.68 of their study-specific measure of attitudes, knowledge, self-efficacy and behavioural intention in respect of training.

Stress and burnout: Two studies rated as moderate methodological quality measured stress and burnout as training outcomes. 10,51 Both studies used uncontrolled pre and post designs and both measured stress and burnout at a single time point immediately pre and post training. Bowers et al. 10 used the Maslach Burnout Inventory (MBI)13 (Cronbach’s alpha of 0.86) and Paterson et al. 51 used the General Health Questionnaire52 to measure stress and burnout (Cronbach’s alpha 0.82–0.86). Paterson et al. 51 found significantly reduced stress and burnout associated with training (effect size not calculable due to statistical reporting) and Bowers et al. 10 found no significant effect of training on burnout.

Two studies rated as weak in methodological quality measured burnout as a training outcome. 26,59 Both studies used uncontrolled pre and post designs and measured stress and burnout at a single time point immediately pre and post training. Goodykoontz and Herrick 26 used The Burnout Scale27 and Taylor and Sambrook59 used the MBI. 13 Both studies reported reduced burnout in trainees in the post-training period, but both failed to test the significance of these changes.

Skills-based outcomes: Seven moderate-quality studies provided outcome data related to trainee de-escalation skills. Two used a non-randomised controlled design29,44 and five used an uncontrolled pre and post design. 4,10,16,51,53 Six measured skills outcomes at a single time point immediately pre and post training4,10,16,29,44,51 and one study included an additional 15-month follow-up time point. 53 Six of the seven studies providing data on this outcome found a significant effect of training. 4,10,16,44,51,53 Effect sizes were calculable for two of these studies. In one study, medium effect sizes were reported for two skills outcomes relevant to de-escalation, measured with the support (staff supportiveness of patients) (0.6) and autonomy (degree of autonomy and independence granted to patients) (0.68) subscales of the Ward Atmosphere Scale (WAS). 10 The remaining study reported a small effect (0.16) of training on ‘behavioural intention’ (intention to use de-escalation). 16 Only three of seven moderate-quality studies measured skills outcomes using validated scales. These reported Cronbach’s alphas ranging between 0.69 and 0.79 (subscales of the WAS),10 0.87 (DABS)44 and between 0.25 and 0.71 (subscales of the MAVAS). 29

One weak study provided outcome data related to trainee de-escalation skills. They used a non-randomised controlled design61 and measured skills at a single time point immediately pre and post training. The authors found significant improvements associated with training on the following subscales of their observer-rated measure to assess skills changes: ‘ability to defuse the situation’, ‘ability to deal with the situation’, ‘ability to control the situation’, ‘ability to be supportive’, ‘ability to deal with criticism’ and ‘effective use of confrontation’ (p. 111). They found no significant effect of training on ‘eye contact’, ‘posture’ or ‘empathy’ (p. 111). Effect sizes for the subscales with significant effects ranged between 0.8 and 2.0. No Cronbach’s alphas were reported for these scales.

Participant satisfaction

A total of 12 studies provided some qualitative evaluation of participant’s views on and experiences of the interventions. In one study, many participants perceived the training to have no positive impacts but no further information on these views was discussed. 43 One study reported a mix of positive and negative perceptions of the training intervention. 19 The remaining 10 studies reported positive views of the training from the majority of, if not all, participants. Suggested improvements to the training by participants are described in the following four themes.

Duration, frequency and availability: Participants discussed the importance of offering regular refresher courses to increase the frequency of training and maintain learning. 26,31,41,58 Conflicting views on duration were reported; some felt training was too long and others requested that increased content and time to process learning be included. 26 Participants reported a need for more opportunities to complete de-escalation, and related, training courses in general. 20

Delivery methods: Participants felt training should be tailored to the clinical context in which they were employed31 and more specific clinical examples provided from these contexts. They felt it was important to include role plays and opportunities to practise real de-escalation interventions using a diverse array of case examples. 8,20,26,31 Live demonstrations were reported as preferable to video-recorded examples26 and there were requests for trainer observation and feedback on real staff–patient interactions. 31,58 Some participants requested smaller training groups,8 while others felt training ward teams together was preferable to support whole-team approaches. 25,41 It was felt that all members of the multidisciplinary team (MDT) should be trained in de-escalation techniques. 25,41 Access to a de-escalation manual on wards was also suggested. 58

Intervention content: In multiple studies participants felt it was important to include a more in-depth coverage of the components of de-escalation. 8,20 There was a perceived need for focus on early detection of aggression20 and consideration of ‘illness and non-illness related aggression’. 31 Participants felt it was important to cover theoretical models of de-escalation8 and also that training should be tailored to individuals and cultures. 20

Facilitator attributes: Participants felt trainers with current ward experience to be more credible facilitators and emphasised the importance of trainers linking training content with personal experiences working on the wards. 31

Formal knowledge

Unsurprisingly, knowledge of de-escalation, and alternatives to control and restraint, is important. 63,64 When combined with knowledge of the patient, being aware of a range of de-escalation strategies allows staff to select and adapt specific interventions, tailored to the individual, to avoid use of containment. 63,65–72 Conversely, a lack of knowledge can lead to staff ‘defaulting’ to an authoritarian approach. This may be particularly relevant to staff responses to absconding behaviour, where staff feel that there are few alternatives to disciplining approaches (warnings, deterrents) available. 73

Knowledge regarding psychopathology also appears to be important. An awareness of symptoms and psychiatric/behavioural indicators can assist staff in predicting the occurrence of an incident, its trajectory and the level of risk present. 64,65,74 Psychiatric knowledge may improve attitudes, beliefs and attributions regarding ‘problematic’ behaviour, all of which directly affect de-escalation behaviour. 68,71,75–79 Unfortunately, inaccurate beliefs and attributions are common; for example, aggression or agitation as ‘acting out’,71 staff ‘attention’ reinforces self-harm,80 engaging with suicidal patients is ‘inappropriate’,81 punitive techniques, such as denying access to therapy, lead to improvements in behavioural and emotional self-regulation. 79 Such beliefs can lead to authoritarian and emotional staff responses to conflict behaviours. 82

Knowledge of the patient

Patient knowledge allows staff to predict incidents,66,73,83,84 identify early warning signs of potential conflict behaviours,67 select and apply individually tailored de-escalation interventions63,68,70,72,83,85,86 and improves therapeutic relationships. 86,87 Diffusion of relevant information, to all stakeholders, to enhance this knowledge is important to de-escalation. 75 Data suggest that an awareness of patients’ typical presentation can be helpful in prediction of conflict. 66,67,73,83–86,88 Changes in presentation can serve as an ‘early warning’ for escalating behaviour, thus allowing staff to intervene early. 65,67,84 Staff should also be aware of individuals’ history, significant life events, and relationships with other patients, triggers and typical trajectory of distress. This knowledge facilitates the selection of appropriate de-escalation interventions. 67,69–71,86 Knowledge of the patient can also promote empathy among staff and provide a sense of predictability and safety during incidents. 71 Some evidence suggests that patients respond better to staff who know them. 88

Understanding the meaning of behaviour is also important. For example, conflict behaviour often has a communicative value;89 improved understanding of patient motivation improves staff attitudes towards patients,90 increases a desire to help and assists staff in moving from coercive to alliance-based responses to conflict behaviour. 77,82 Limited knowledge regarding the meaning or function of behaviour leads to a blanket response to escalation, without personalisation. 76

Empathic communication and interpersonal skills

Authentic engagement and empathic communication are perceived as sufficient, in and of themselves, to defuse unsafe situations without containment. Non-medicalised, authentic91,92 engagement with patients can create a sense of safety. 67,72 Rapport,73 active listening, and direct acknowledgement and validation of patient behaviour,85,90,93–96 experiences97 and concerns71,82,97 can be helpful to de-escalate a potentially dangerous incident. Calm, non-provocative language should be used,67,72,88 and any directions should be accompanied by an explanation and communicated with respect and care. 88 Limit-setting is generally acceptable to patients if delivered with empathy and an explanation. 88 Compassion and support should be maintained, even in the wake of a serious incident. 98

Data suggest that, while essential in managing all conflict behaviour, communication and interpersonal skills may be most important in the de-escalation of self-harming or suicidal behaviour. Engagement is described as ‘the difference between life and death’ (p. 309). 81 Both staff and patients identify the following qualities as useful when engaging in de-escalation of self-harm/suicide: empathy, respect, a willingness to help, respecting space, expertise and autonomy, negotiation, communicating hope and avoiding overreaction and judgement. 76–79,81,83,85,89,90,93–96,99–103

Therapeutic relationship

A strong, pre-established therapeutic relationship can optimise and enhance de-escalation,81 promote help-seeking before de-escalation is required,76,101,102 facilitate compliance82,97,98,104,105 and motivate patients to avoid dangerous behaviours. 73,75,90,94 The data indicate that developing the relationships for de-escalation can be difficult with involuntarily detained patients. 77 Patients may perceive trusting relationships with staff to be ‘risky’, citing concerns about reliability and consistency. 98 Staff must also be aware of the tension between casual, informal interactions, which serve to develop strong therapeutic relationships, and the need to maintain ‘professionalism’, something that is also valued by patients. 72 Use of containment interventions further damages patient trust in staff and results in a cycle of conflict followed by containment. 86

Assessment and flexible intervention

Assessments to inform individualised de-escalation approaches should be made at admission, including clinical history, current presentation, symptoms, suicidality and mental state. 73 Assessment, however, is considered an ongoing process, whereby staff are attentive to the full range of environmental, patient, milieu and relational factors66,67,72,106,107 that may precipitate conflict behaviours. Staff should be aware of person-specific triggers that may cause an escalation, and behavioural cues that indicate the likely onset of unsafe behaviour. 67,69,83,85,108 When conflict is active, staff must determine how and when to intervene,66,67 considering the success of past methods of intervention. 68 Even when containment is being used, stuff must repeatedly assess whether there is a necessity to continue. 104

Any assessment of conflict behaviour will inevitably incorporate, either implicitly or explicitly, an attribution of the cause of the behaviour. Data indicate that attribution directly impacts the nature of de-escalation intervention;71 for example, staff are less likely to use control and restraint if the behaviour is attributed to illness, rather than the person. 67

Intervention approaches are clearly articulated for aggression and self-harm/suicide; limited data are provided for the other conflict behaviours. Basic approaches, including emotional support, reassurance, comforting, focusing on the future, grounding, and distraction techniques,83,85,87,93,101 problem-solving, negotiation, collaboratively identifying solutions67,82,97,104,105 and limit-setting,86 can be used to effectively deescalate conflict situations. Passive intervention should be considered; giving an individual time and space, disengaging, or delaying intervention can allow patients to self-regulate. 63,66,67,69 Whatever intervention approach is taken, patient autonomy must be prioritised; punitive, or corrective, interventions generally lead to increased patient distress. 101

Data emphasise the need for individualised intervention, based on patient need. 76,89,91,95,107 Staff must be flexible in the selection and application of intervention, considering both the individual and the context. 68,71 The personalised adjustments are dependent on knowledge of the individual, highlighting earlier emphases on strong therapeutic relationships and assessment. Indeed, early discussions regarding personal triggers, behavioural indicators and preferences about how and when staff should engage can guide the selection and application of intervention strategies. 83,108

Awareness of antecedents

Good practice, in relation to de-escalation, appears to be contingent upon staff awareness of patient behaviours, interactions and milieu ‘flow’. 64–67,74 By maintaining an awareness of subtle changes in the environment, staff can identify the antecedents of conflict, informing decisions about intervention. As indicated previously, staff must have a good knowledge of the patients they are supporting to identify likely triggers, early warning signs, changes in emotional states and person-specific cues of aggression. 67,73,99 Similarly, characteristics of the milieu, such as noise levels, movement, pacing and agitation, can offer an early indication of the likelihood of conflict. Combined, this form of fluid assessment allows staff to effectively predict and neutralise conflict episodes without containment. However, some early warning signs are difficult to identify73 or are ignored by staff. 109 This may contribute to the advent or escalation of conflict behaviours or, importantly, avoid unnecessary intervention.

How and when to intervene

Staff must be able to, firstly, identify situations that are becoming unsafe, and, secondly, predict the likely outcome; the outcome of each of these decision processes will dictate if, how and when they intervene. 66 Ultimately, staff must differentiate between behaviours that can be tolerated (benign), and those that require control. 67 If the decision is made to intervene, the nature of the intervention is often informed by ethical principles of respect, dignity, self-determination and safety, and an assessment of the patient’s historical responses to intervention. 72

Organisational culture

Ethos: The underlying beliefs, assumptions and values of an organisation directly influence the use of de-escalation procedures. The ethos of an organisation will manifest itself in the behaviours of its staff; principles such as respect for individuals and safety as a human right, and the expectation that staff will prioritise therapeutic engagement, remain calm when dealing with conflict, and focus on helping, rather than correcting, are linked to the use of de-escalation. 67,81,91 Unfortunately, coercive practices are often justified at the organisational level. For example, forced medication is typically justified by existing legal frameworks and the view that treatment is ‘necessary’. 104

Procedures: Formal clinical systems can prevent conflict behaviours and facilitate the use of de-escalation. Emergency systems and procedures must be reliable and effective if staff are to engage in de-escalation safely and confidently. 110 Additionally, risk assessment protocols and procedures for updating risk assessments should be clear, consistent and communicated among staff teams, to optimise intervention. 84 Co-produced care plans and ‘Positive Behaviour Support’ (PBS) plans also appear to enhance de-escalation; these documents are typically developed by psychologists and are designed to reinforce positive behaviours by enhancing staff and patient awareness of behavioural sequences. As such, they typically identify triggers, behaviour functions, early warning signs and patient preferences for reactive and proactive means of calming distress. 72,108

Staff support: Formal methods of staff support, such as debriefing and clinical supervision, have been found to promote active learning69 and can enhance de-escalation. 64,72,106 Staff express general dissatisfaction with a lack of post-incident debriefing and support. 73 Staff also value less-formal support processes to enhance de-escalation practices, such as emotional sharing and mutual support between colleagues. 80

Resources

Staff: Low staffing can be a barrier to the implementation of planned de-escalation strategies, such as PBS plans. 108 Low staff numbers also are implicated in de-escalation in medication refusal; when limited staff are available, or if there is a high load of acutely unwell patients, negotiation only occurs with patients who are deemed to be dangerous, or a positive outcome is predicted. 104 Some findings suggest, however, that staff mix is more important than absolute staff numbers. 71 Balanced staffing, ensuring a mix of skill and experience, can facilitate de-escalation, while a poor staff mix contributes to ineffective teamwork and poor communication. 69,71,72,106,109 Experienced staff are perceived better at deciding when to intervene, and are more likely to use de-escalation, in response to aggressive behaviour. 69

The type of staff also influences patient behaviour and the use of de-escalation. The presence of ‘authority figures’, such doctors and male nurses, can defuse conflictual situations, and having male nurses present makes staff more comfortable and confident engaging in de-escalation. 72 Well-trained nurses are seen to be most effective in managing self-harm, without containment. 80 Conversely, bank/agency staff may lack the skills and experience to identify escalating behaviours and can miss early opportunities to intervene. 99 Bank/agency staff are also often experienced by patients as aggressive in response to escalations. 109 Poorly or inadequately trained staff, and those who are unfamiliar with the ward and patients, find it difficult to predict escalating conflict behaviour. 71

Time: Often related to overall staffing levels, time, or lack thereof, represents a significant factor in the use of de-escalation on inpatient psychiatric units. The effectiveness of de-escalation is widely reported as being dependent on a strong, pre-existing therapeutic relationship; these relationships can only be developed over time. 97 Data indicate that a lack of resources, including time, has a direct impact on the quality of relationships, interventions such as sensory rooms,111 negotiation,104 observation93 and listening to patient concerns. 89

Activities: A lack of ward activities prevents these being used as de-escalation strategy. 72,112

Ward environment

Physical environment, social environment and rule application: Though difficult to manipulate, the physical environment of inpatient wards can contribute to the incidence of conflict and influence staff responses. The availability of a sensory room offers another de-escalation approach, and an alternative to containment. 113 Staff indicate that the availability of a sensory room allows agitated or aggressive patients a space to engage in self-soothing behaviours in private, thus regaining a sense of control. 111 If available, the use of sensory rooms should be incorporated into care plans. 113 When a ward has a high load of acutely unwell patients, staff may use de-escalation techniques, such as negotiation, less. 82 Patients value flexibility in the application of normal rules, where it can be facilitated safely, as a de-escalation strategy. 92,114,115

Staff peer support

Having a strong staff team, where staff feel they can rely on each other, enhances de-escalation. 70 Good communication and a sense of ‘community’ among staff allows for transfer of de-escalation knowledge. 63 When seeking support after an incident, staff prefer to receive this from peers who have shared similar experiences; this assists in processing emotions and may help staff to maintain therapeutic relationships with patients in advance of future de-escalation events. 75

Psychological skills

Six core psychological skill domains for de-escalation were identified from synthesis of stakeholder perspectives. These were divided into connecting domains, regulating domains and an additional domain, labelled as ‘pliability’. ‘Connecting domains’ related to staff members’ ability to engage in behaviours that confirm and validate patient distress and to their ability to reduce social distance, that is, to engage with patients in authentic, reciprocal human interactions on equal terms. ‘Regulating domains’ related, firstly, to staff’s ability to remain calm and assess and understand the threat level present, and the causes and function of the behaviour requiring de-escalation. They, secondly, related to staff’s ability to create the cognitive, emotional, and physical conditions needed for de-escalation, either through autonomy-confirming techniques (e.g. offering choices, providing time and space, collaborative problem-solving) or limit-setting techniques (e.g. instructions and deterrents).

The ‘pliability domain’ cut across connecting and regulating domains and referred to staff ability to mould their de-escalation behaviours according to individual patient preferences and the changing dynamics of critical incidents as they unfolded. For example, many staff participants pointed to wide variations in the levels of intimacy and informality they would adopt in de-escalation events depending on the patient involved. Female patients valued staff who were able to combine empathy and understanding with a firmness and discipline that enabled them to contain difficult emotions, for example Q2.

The core de-escalation skill domains were, broadly, presented as prerequisites for effective engagement in a series of ‘de-escalation processes’ which staff participants reported as occurring in a loosely linear pattern. These processes were noticing a relevant change in the patient’s behaviour then either engaging in a process of ‘watchful waiting’ or deciding to intervene. If intervention was deemed necessary, there was broad agreement that the first step should be to manipulate the environment, for example, by inviting the patient into a private space or encouraging other patients to leave. The second step was to ground and reassure, for example, by providing safety statements such as ‘you’re safe’, ‘we are nurses’, ‘you’re in hospital’, ‘we will not harm you’ (male ward manager 1, acute ward). The remaining steps related to a logical process of clarifying then helping to address the problem by distracting, resolving, or reframing. The importance of post-event support was widely emphasised irrespective of whether a process of ‘watchful waiting’ or intervention had been adopted.

The richest data in terms of specific de-escalation processes related to reframing techniques. Both staff and patients pointed to the importance of stimulating memory to de-escalation. For example, PICU staff described routinely asking the patient to recount events preceding admission and felt this helped to promote insight and re-contextualise the patient’s current circumstances. Patients often described staff who would point to their shared history with staff, or a time when they had done something well, as a means of stimulating positive memories and emotions that would reduce anger. Both staff and patients agreed that de-escalation was unlikely to occur where staff had failed to create positive short-term memory of their conduct through attentive engagement with patient emotions and demonstration of trustworthiness and care, for example Q3.

Physical skills

The physical de-escalation skills participants described, perhaps, counterintuitively, aimed to communicate vulnerability (not fear). This was to communicate a non-violent intention and to render violent actions less socially congruent. For example, many staff participants described initiating sitting down, even when this was unreciprocated by the patient, for this reason. Importantly, some patients described experiences where staff would remain seated when confronted from a standing position and experienced this as belittling. This demonstrated how, at a surface level, similar behaviours could communicate importantly different messages to patients. Initiating sitting communicates vulnerability by relinquishing an optimal physical stance for self-defence, whereas remaining seated communicates a lack of concern for the physical threat posed by the patient and is, thus, experienced as emasculating. Other physical behaviours staff described as useful in communicating vulnerability and trust included ensuring a minimal staff presence necessary to maintain safety and adopting open body language with arms placed by sides. There were exceptions. Some staff participants reported that keeping arms by sides interfered with their ability to conceal visible tremor in situations involving heightened aggression. There was broad agreement among staff that visible anxiety could cause feelings of stigma in patients during incidents, but patients did not always share in this agreement (e.g. Q4).

The main divergence between staff, patient and carer views related to the use of physical touch as a de-escalation skill. Female patients and carers identified a central role for use of touch. For example, carers discussed its importance when the level of distress was so high that the individual found it difficult to respond to verbal communication. The most frequently mentioned form of touch was holding patients’ hands, or even just for staff to be ‘within enough distance to put out their hand and comfort’ (female carer, Carer FG 1) (i.e. simply sharing the patient’s space in a more intimate way could be calming in and of itself). Carers talked a lot about the value of hugs as a way of responding to patient distress. Only ‘old-school’ (female carer, carer FG 1) staff were viewed as being willing to give hugs to patients. This indicated either that more experienced staff were more flexible in their implementation of interpersonal boundaries (i.e. able to consider individual and situational differences) or differing practice norms between generations of nurses. There was a consensus among carers that patients were so deprived of physical contact on inpatient wards that they would seek it through physical restraint (e.g. Q5).

Understanding people

Trauma and behavioural scripts: Both patients and staff accounts indicated the importance of staff understanding how aggressive behavioural scripts could be activated through behaviour that mimics patients’ past trauma. One patient drew comparisons between staff de-escalation approaches in different hospitals which underscored the importance of this understanding (Q6). Relatedly, many staff participants described de-escalation of anger that was, ostensibly, related to unmet needs but, they felt, was implicitly motivated by antagonism to authority. They described situations where attempts to meet needs and provide explanations were met with newly generated grievances, counterarguments, and escalating confrontation. They emphasised the importance of understanding the significance of authority representations in the context of both coercive psychiatry and childhood abuse and neglect, and to respond to the behaviour in unexpected and surprising ways that moved out of the anticipated dialogue, reframing authority representations and humanising staff. These responses typically involved humour, swearing or blunt acknowledgement of the dynamic, as it was unfolding, for example Q7.

Patient-specific knowledge: There was agreement across stakeholder groups on the importance of gathering and effectively utilising patient-centred information to inform de-escalation; for example, triggers of traumatic memories in staff behaviour or the environment, the social and cultural norms of the patient and known strategies for de-escalation that worked in the past. However, there were contrasts in carer and MDT professional perspectives on how this ‘patient knowledge’ was constructed. Staff tended to value one-directional processes such as MDT discussions, reading patient notes or handover sheets as useful sources of this information but also recognised that these processes could diffuse, through nursing teams, unhelpful and inaccurate preconceptions about patients. Carers, by contrast, felt that patient knowledge with utility to de-escalation could only be derived from authentic ‘knowing’ of patients, built up through consistently reciprocal interactions over time. In this sense, they prioritised knowing patients over knowledge of or about patients (e.g. Q8)

Moral formulations: Carers and patients felt that moral formulations constructed by staff as a way of understanding patient behaviour represented a key barrier to de-escalation. They provided numerous examples of evaluatively loaded expressions they had experienced in typical staff language, for example ‘bad’, ‘nasty’, ‘difficult’ (Female Carer 1). Carers felt that the moral judgements that were communicated through these expressions betrayed a superficial understanding of patients and that de-escalation required, instead, deep understanding and connection with the feelings and needs underlying patient behaviour. By contrast, staff tended to identify deficits in understanding in terms of specific illness knowledge. Most commonly, they reported a lack of knowledge of how to reduce distress in aggression they linked with auditory hallucinations (e.g. Q9).

Conceptualisations of de-escalation

Despite the breadth and extent of skills described by participants, there was a substantial cohort of participants, represented across all stakeholder groups, who rejected the concept of de-escalation as consisting of a discrete set of skills. Rather, these participants conceptualised de-escalation as a ‘process’ that occurred within the context of a mutually trusting relationship. The centrality of relationship quality to de-escalation was, most clearly, illuminated in the context of medication refusal. Multiple staff provided accounts of patients who would only accept medication from specific staff. They described how these preferences could be accommodated by waiting until the preferred staff member was next on shift to avoid the need to enforce medication. In one extreme example, the patient’s carer would come into the inpatient environment each evening to administer the patient’s medication. This relationship-focused, flexible conceptualisation of de-escalation contrasted sharply with other commonly held understandings. For example, some participants appeared to conceptualise de-escalation as efforts to enforce patient compliance without using restrictive interventions. These participants found it difficult to engage with the potential for meaningful negotiation or conceding ground to meet patient needs, and often viewed de-escalation as part of a sequence of events that inevitably ended in restraint. So enmeshed, at times, were participant conceptualisations of de-escalation and restraint that they did not always meaningfully distinguish between the two interventions. For example, participants often observed that multidisciplinary professionals did not get involved in ‘de-escalation’ because ‘de-escalation’ would damage their relationships with patients, for example Q10.

Ward team culture

Regime: Patient, ward staff and carer accounts focused extensively on ward team cultures that they felt worked in opposition to de-escalation practice. These accounts reflected a cycle of violence, in which staff would use maladaptive coping strategies (e.g. excessive rules, stringent boundary maintenance) to manage their fear of patient violence. Patients would then respond to these coping strategies with violence, thereby reinforcing staff fear and furthering the perceived need for more boundaries and rules. In addition to aggravating rules, patients also described, as a feature of dysfunctional teams, attempts to control patient narratives of their experiences as inpatients, as a way of maintaining safety. For example, patients observed a marked intolerance among staff of dissent and criticism of practice. During the limited occasions when patient views on practice were consulted, for example, in ward community meetings, they described feeling steered towards accepting explanations for poor practice or shortfalls in provision that, they felt, were unduly apologetic to staff and services. An additional manifestation of fear cultures was demands for unconditional patient respect of staff, irrespective of whether this was reciprocated in staff treatment of patients. The sense of injustice resulting from these practices was compounded by the presence of what they called ‘propaganda materials’ (female patient 4, acute ward) in the environment (e.g. Q13). The propaganda materials that this patient was referring to were posters in ward environments demanding patient respect of staff (without concomitant demands for staff to respect patients) or expressing zero tolerance of patient violence. Patients reported that these cultural practices had such an aggravating impact on patient’s general affect and level of arousal that attempts at de-escalation, where attempted, were unlikely to be successful.

Vulnerability: The consequences of fear cultures within ward teams were not limited to staff treatment of patients, but also affected their treatment of each other. Descriptive accounts of these teams indicated that they displayed a marked contempt for vulnerable colleagues (e.g. new starters, newly qualified nurses, students and non-regular staff) and a suspicious disdain for outsiders (e.g. MDT professionals, senior nurses). Both groups were perceived as representing a threat to established risk-management practices (the former group through unfamiliarity, and the latter group through unwelcome change initiatives). Participants described a culture in which the practices of new starters would be closely monitored, by established staff, to assess and, if necessary, modify their inclination towards assimilation to the existing culture. This generated pressure in vulnerable staff to demonstrate ‘competence’ through practices that conflicted with de-escalation (e.g. premature use of restrictive interventions and inflexible application of ward rules). These social dynamics therefore encouraged sadomasochistic ways of relating to others, where avoidable conflict was initiated to gain social approval, inviting violence from patients, for example Q14.

Ideology

Deserving and undeserving patients: A universal feature of teams, described by all stakeholder groups, was shared ideologies related to patient deservingness of care. Staff often used the term ‘behavioural’ to signal to each other whether an observed behaviour or patient was worthy or unworthy of de-escalating responses. This was based on a subjective clinical judgement as to whether the person’s behaviour could be linked to mental illness. There was, rarely, reflection on how, given the complexity of psychopathologies and trauma characterising inpatient populations, these judgements could ever be made with any degree of accuracy. Carers linked these ideologies with a desire to make patients ‘take responsibility’ (Carer FG) and that this desire manifested itself at the point of de-escalation through unhelpful behaviours, for example Q15.

Beliefs about behaviour change: The perceived need to encourage patients in distress to ‘take responsibility’ at the point of de-escalation was, again, linked by carers with what were labelled as ‘behaviourist beliefs’ within the analysis. Commonly shared in ward teams, these included beliefs that empathetic responses to distress ‘reward’ or ‘reinforce’ negative behaviours and coping strategies, and the belief that making hospital ‘too nice’ (MDT FG 1) for patients lengthens inpatient stays. They also related to more fundamental staff beliefs on the value of punishment and reward in eliciting desired behaviours from patients. Patient and carer views often sharply contrasted with ward staff views in this respect. The former tended towards the view that undesirable behaviours could be reduced through addressing underlying sources of distress and were sceptical about the value of punishment and reward in long-term behaviour change, for example Q16.

Multidisciplinary team participants emphasised that behaviourist beliefs functioned, in practice, in contradiction to their intended aims, in that patients would ‘progress the pathway in a different way’ (MDT FG 1) (i.e. become so unwell or so aggressive that they required transfer to PICU or forensic services). Although not exclusively, behaviourist beliefs and assessments of deservingness appeared more commonly in discussions of patients labelled with personality disorder and directly informed differential treatment, for example Q17.

Attitudes to personality disorder: There was evidence that National Institute for Health and Care Excellence (NICE) guidelines recommending alternatives to inpatient admissions for people with borderline personality disorder had exacerbated already pernicious attitudes to patients within this group. Many staff expressed nihilistic views of working with these ‘untreatable’ (Male Staff Nurse 2, PICU) people, often held the view that hospital is not the right place for them and bemoaned the lack of available specialist services. It was unclear whether this viewpoint reflected a genuine need for an alternative setting or a desire among staff to limit the time they had to interact with a patient group they found difficult. This uncertainty was reflected in the diversity of explanations provided for why patients with diagnoses of personality disorder should not be on acute wards. For example, the same participant expressed the contradictory positions that it was not helpful for patients with this diagnosis to be in hospital at all, and that it was simply that acute staff lacked the skills to manage this patient group, necessitating a specialist service, for example Q18. Overall, it was clear that these attitudes prevented staff from connecting with the distress underlying behaviour in this group, perceived as so fundamental to de-escalation by carer participants.

Organisational structure and culture

Hierarchical barriers to information availability and utility: There were extensive data from all stakeholder groups describing how hierarchical professional structures and group interests limit the availability of accurate information about patients to enhance de-escalation. Participants described marginalisation of perspectives at every layer of inpatient organisational structures: between medical and nursing staff, between senior nursing management and ward teams, between multidisciplinary teams and unqualified nurses, between nursing teams and patients, and between inpatient services, broadly, and carers, for example Q19. There was a strong perception among many participants, cutting across stakeholder groups, that these structural problems ensured that those who knew least about the patient had most influence on de-escalation strategies.

Senior nurses consistently expressed the view that psychiatrists tended to be invested in traditional medical hierarchies and disregarded the views of nurses. They drew on strategies for de-escalation of medication refusal as examples of this, where, they felt, nurses were often required to enforce medication when de-escalation strategies were available. For example, these nurses reported that nursing proposals to delay forced treatment, to identify and resolve psychosocial stressors were often overruled by psychiatrists. They made the broader observation that the paternalism inherent to the medical model undermined the confidence of nurses to make common-sense decisions in relation to de-escalation. For example, MDT professionals reported a common belief among nursing staff that depot dates determined categorically when the medication should be administered, and that this resulted in the avoidable use of force to administer medication. This was perceived as partially driven by anxiety about asking doctors to rewrite depot prescriptions and partially by mistaken beliefs about the authority of these systems (professionals noted a lack of specific policy guidelines on the issue). In contrast, MDT professionals felt that a ‘three-day grace period’ in which to negotiate with the patient was unlikely to result in any significant deterioration in mental state or increases in other risks. Carers broadly and some staff, for example Q20, felt there was a lack of information provided to patients about antipsychotic medication and side effects, which they felt was rooted in the, they felt mistaken, belief that increased transparency would create more, rather than fewer, problems with adherence.

Other structural barriers precluding relevant staff input on de-escalation included the exclusion of unqualified staff from ward rounds. This had the effect of marginalising, from key de-escalation decisions, the staff group who spent most time with the patient, and prevented unqualified staff from developing the knowledge required to de-escalate typical patient concerns, for example Q21.

Power and alienation: Senior nursing leadership was perceived as failing to foster the quality of relationship with ward staff that would make accurate information about sources of conflict and de-escalation strategies available. A culture of distance and alienation between front-line ward staff and nursing management was attributed to a lack of meaningful clinical input and blame-based contact. This had evidently engendered a siege mentality among staff, and a sense that those outside the direct care team lacked a credible understanding of the context in which decisions and expectations were implemented. Participants indicated this created a culture of back-covering and secrecy that rendered critical events involving de-escalation invisible to senior leaders. Thus, there was an identified need for closer working relationships that fostered open dialogue around de-escalation practice and supported learning rather than blame, for example Q22.

Exclusion of patient and carer perspectives: Power imbalances between different groups also affected the quality of information gathered after de-escalation had occurred. The exclusion of patient perspectives from written communications was perceived as having a significant impact on the availability of accurate information for de-escalation. For example, the lack of direct patient contribution to nursing notes was felt to limit reliable communication of what had worked in terms of de-escalation. Nursing notes were, more broadly, identified as a major source of patient distrust in qualified nurses, who were perceived as documenting incidents from a biased and self-exculpatory perspective that was then taken as authoritative by others reading their accounts. Conversely, increased attempts to embed patient perspectives in systems (e.g. through co-produced de-escalation plans) were reported positively across stakeholder groups, for example Q23.

Overall, carers described similar experience to patients in terms of marginalisation from discussions and decisions relevant to de-escalation. Where their perspectives were sought, they were given the impression that staff ‘didn’t actually believe anything [they] said’ (Male Carer 1). MDT professional discussion of how they involved carers in decision-making confirmed these perspectives, as they exclusively framed carer involvement in terms of how to get carers to ‘see the benefits’ (MDT FG 2) of a plan they might be resistant to, to get them ‘on board’ (MDT FG 2). Conflicting perspectives with carers were perceived by MDT professionals as stemming fundamentally from a lack of knowledge on the part of carers.

Positive risk-taking: Staff were very much aware of the potential value in adopting less restrictive approaches to self-harm but felt that broader organisational support for positive risk-taking was currently lacking. Consequently, nursing staff were perceived as feeling the need to intervene aggressively at every incident of self-harm and that clinical management was almost exclusively based upon removal of means. MDT professionals felt that this only escalated patient desire to self-harm and tended to increase severity when the opportunity eventually arose, for example Q24.

Environmental stressors

Staff and patients described sources of environmental stress that had a persistently aggravating impact on patients’ level of arousal and affect. They emphasised that any attempt to enhance de-escalation must first address these sources of distress. Environmental stressors were subsumed under four categories: environmental signifiers of coercion and disrespect; ward procedures; patient community conflict; sensory input.

Environmental signifiers of coercion and disrespect: Patients described a range of visual cues and design features of inpatient wards that were either fear-inducing or alienating, or both. They identified as particularly unhelpful posters within the environment that instructed or threatened patients about their behaviour. Patients also found visual evidence of coercion within the environment (e.g. seclusion-room doors and pinpoint alarms) distressing and felt that staff used these as visual cues to intimidate patients into compliant behaviour. Staff accounts, to some extent, corroborated these perceptions, for example Q25.

The door to the nursing office remaining closed was a consistent and significant source of alienation for patients. They described feeling intimidated approaching this ‘staff-only’ space and difficulty in rousing staff from within to meet their needs. This resulted in many examples of escalating frustration and distress occurring immediately outside the staff office, for example Q26. Notably, multiple patients identified being invited into staff-only areas (e.g. nursing offices or medication clinics) as an effective de-escalation strategy. This highlighted the significance of segregated staff spaces both as a potent source of conflict and as a symbol of disrespect for patients.

Clinical processes: Both patients and staff provided extensive accounts of ward processes that, they felt, in their current format were incompatible with de-escalation. These processes generated feelings of dependence, exclusion or isolation, humiliation and loss of control, in patients. These were, consistently: ward rounds, nursing shift handovers, prescribing consultations, waiting times, medication rounds, mealtimes and admission. In respect of admissions, specifically, patient participants emphasised that experiencing trauma (use-of-force and/or social isolation) at admission can make it impossible to recover the trust in staff necessary for de-escalation to work. Staff and patients identified the need for improvements to provide a greater sense of power and control over these experiences. These included: greater involvement of patients in handover and prescribing; smaller ward rounds with fewer attending staff; measures to avoid social isolation or use of force on admission; more efficient management of patient requests; elimination of institutionalised practices around medication rounds (inflexible timing, queuing, use of the ‘stable-door’); and creating a less dehumanising experience of mealtimes (i.e. being observed, plastic cutlery, being required to remain in the dining area until patients have finished eating), for example Q27.

Patient community conflict: Patients prioritised patient community conflict as a source of environmental stress. There were widespread reports of insufficient and, often, no staff present in communal areas to detect and intervene in patient-to-patient conflict. Patients felt that rules mandating patient assembly and blocking patient access to bedrooms impeded their ability to escape victimising behaviour and wanted these practices eliminated or minimised, for example Q28. The patient community dynamic that was most focused on by staff participants was what they perceived as ‘prison cultures’ on male wards. This referred to perceived patterns of behaviour in which patients (typically young, frightened, with prior experience of the criminal justice system) would attempt to establish a hierarchy within the patient community by initiating conflict with patients and/or staff. This required careful attention in relationships: for example, by providing subtle reassurance (that does not undermine masculine self-confidence), orienting the patient to the differences between hospital and prison and avoiding escalating confrontation in the presence of other patients (furthering the perceived need to secure dominance). There was some limited support among patients and staff for training patients in de-escalation, for example Q29.

Sensory input: Patients acknowledged the contribution environmental noise made to escalations, by impairing their ability to monitor the source of auditory stimulation and increasing fear of other patients. However, many felt this problem was currently managed through the joyless suppression of all noise without reference to meaning and context. Patients were broadly critical of the absence of calming spaces equipped with sensory equipment that would enable their escape from distress caused by other patients and environmental noise, for example Q30. Several patients and nurses made specific reference to the value of having objects to safely kick and punch, for example, beanbags, mattresses and pillows, as a de-escalation strategy. This activity was perceived as useful in providing distraction from urges to self-harm or be violent, reducing anger and coping with distressing auditory hallucinations, for example Q31.

Professional boundaries

Staff conceptualisation of appropriate professional boundaries has an important motivating impact on de-escalation engagement. Descriptions of boundary maintenance ranged from staff who identified self-disclosure and reciprocity as their primary protection against assault, to staff who concealed their name badge from patients to maintain anonymity. The maintenance of narrow boundaries was felt to be underpinned by beliefs relating to how personal safety is maintained, that is, by displaying no emotional vulnerability, disclosing no personal information, and controlling narratives by accepting no patient criticism. Beliefs about appropriate professional boundaries also restricted staff ability to use physical touch as a de-escalation strategy, identified as important by both patients and carers. Broadly, participants felt that narrow professional boundaries precluded authentic connections with patients and thereby de-escalation capability, for example Q32.

Role perception

Correctional, health-educator and task-oriented role perceptions were identified as problematic in terms of de-escalation engagement. Helpful role perceptions had the following features: a conceptualisation of working with ‘people not patients’, an appreciation of reciprocal and authentic relationships (characterised as ‘professional friendships’) and humility, for example Q33.

Social identity

There was evidence gender, seniority and age have an important influence on the range of strategies available to de-escalating staff. Staff recognised that female patients who had experienced sexual trauma would often respond better to a female than to a male at times of distress (due to perpetrators of sexual abuse being overwhelmingly male). However, staff failed to recognise that the same was likely to be true of male patients under their care and did not explicitly recognise or discuss how they would manage a situation where male patients were triggered by the presence of male staff if they were distressed or dissociating. This was exacerbated by staff beliefs that ‘self-harm is more on the female side than the male side’ (female occupational therapist 1, acute ward), meaning that staff on male wards didn’t feel there was a value to attending training on self-harm and suicide, making it unlikely that staff would be able to recognise and respond appropriately to male self-harming behaviours. Overall, staff seemed better applying knowledge of trauma to female patients, understanding the impact of this on how they might need to be de-escalated, and being able to recognise and respond to self-harming behaviour, and had difficulties seeing male patients as vulnerable in a similar way.

Counterproductive beliefs about gender also extended to the staff team, with staff reporting that male staff members felt they needed to take the lead in conflict situations, pushing female staff out of the way to be at the front, since they felt they ‘should be the protector’ (MDT FG 1) of female staff, to ‘not let a woman get hit’ (MDT FG 1). This was seen as unhelpful for de-escalation for numerous reasons. As discussed above, it neglected the impact of male staff re-triggering a patient who had experienced trauma from a male perpetrator. The presence of men (particularly in a group) was also felt to increase male aggressiveness as it could be threatening to their masculinity, as well as not making use of the fact that similar ideologies within the patient group meant they were less likely to hit a female member of staff or even ‘present as aggressive’ (MDT FG 1) in front of them. This set of beliefs about men needing to take the leading role in de-escalation and restraint was not endorsed by MDT participants and was felt by them to increase the number of patient assaults on staff and aggravate conflict situations.

Male patients had a strong view that women possessed feminine qualities that were naturally more suited to de-escalation. It was also felt easier for female nurses to stimulate empathy and remorse in male patients because of the commonality of moral commitments against violence to women. Conversely, female gender could be a disadvantage in the case of men from cultures in which accepting guidance/instruction from females is uncustomary. There was broadly an indication that de-escalation between opposite genders was uncomplicated by gender competition. A minority felt more senior staff could employ authoritative approaches that would not be accepted from junior staff. There was agreement across participants that young staff de-escalating older patients created an abnormal and potentially undignified social dynamic, which required skill and resilience to navigate, for example Q34.

Intervention component descriptions (aims, materials, procedures, delivery methods, persons responsible for delivery, relevant locations, dose and frequency)

Psychology-delivered components

1. 2. ‘Conflict formulation’

The conflict formulation model aims to:

1. deconstruct moral judgements about conflict by:

   1. providing a model of formulation that does not pathologise the patient or see conflict originating only from them

   2. encouraging staff to imaginatively engage with patient perspectives (needs and feelings) and to be open and honest about how their own feelings and needs contribute to conflict situations

   3. think critically about the social, institutional and historical context in which these interactions occur

2. provide a forum for mutual support in staff teams.

Each session requires bespoke materials developed for the EDITION study. These included, firstly, a feelings card pack (n = 90) and a needs card pack (n = 72). These were developed in partnership with the study’s PPI advisory panel. Secondly, they included a laminated PowerPoint slide titled ‘Understanding feelings and needs in context’ detailing relevant contextual factors to conflict situations at the level of individuals, the environment/institution and social injustice (Figure 4). These were derived from our development work and supplementary evidence review.

FIGURE 4.

Understanding feelings and needs in context.

Each session followed the stepwise process outlined in Figure 5. The first step is to identify, from an individual volunteer or a staff team, a conflict with a patient that is provoking negative emotion and/or causing difficulties that need resolving. This will be the focus of the session. Then step 2 is to enable the team to describe the conflict, ventilate emotions and express intuitive understandings of patient behaviour. The facilitator should not attempt to modify the group’s understandings, biases or self-protective mechanisms. Step 3 is a group exercise to identify feelings and needs of staff and patient in relation to the conflict. Identifying feelings is used for the purpose of identifying and articulating unmet needs. Needs-based analysis is useful to progressing beyond moral formulations of behaviour. 138 While it is possible to formulate moral judgements about the strategies people use to get their needs met, it is difficult to morally evaluate the presence or absence of a need. Identifying needs also provides a useful basis for resolutions to reduce conflict. Step 4 aims to enhance group members’ understanding of, and empathy for, factors outside staff and patient control that might explain why feelings and needs sequences are expressed the way they are in this unique care context/working environment. The purpose of this is to re-locate individual, trait-based explanations for conflict to richer, historically, environmentally and situationally informed explanations. Again, analysis of care or work context factors can provide further practical indications of how the context can be changed to reduce conflict. In stage 5, the group is asked to summarise and integrate the discussions and perspectives of the group to develop a shared functional understanding (feelings, needs, context) of staff and patient inputs into the relevant conflict; then, to plan around meeting needs and changing the work or care context to reduce negative emotion and conflict.

FIGURE 5.

Conflict formulation model.

1. 3. ‘Negotiated boundaries’

The ‘Negotiated boundaries’ groups aimed to:

1. minimise unnecessary boundary setting and enhance de-escalation skills

2. create team cultures that are protective of vulnerability and receptive to new/marginal ideas and perspectives on clinical practice

3. enhance the quality of staff–patient relationships through critical reflection on professional boundaries and role perceptions

4. provide a forum for mutual support in staff teams.

Each session requires bespoke materials developed for the EDITION study. Three separate magnet packs were developed based on three session topics. The first pack (n = 42) had a single patient behaviour labelled on it, for example ‘spitting on the floor’. These were derived from literature review of boundary and limit-setting evidence in mental health inpatient settings. Every patient behaviour reported as preceding a boundary-setting response from staff was extracted from the literature. The magnets consisted of behaviours falling under the categories of sexual expression; structure, rules and routine; aggression and exploitation; relationships and trust; emotional expression; self-harm and suicide. The second pack (n = 88) had a single interpersonal boundary labelled on it, for example ‘hiding your surname from patients or concealing or modifying ID badges’. These were derived from review of evidence related to interpersonal boundaries in mental health inpatient settings. Interpersonal boundaries fell under the following categories: equity and expediency; intimacy; distance; exploitation and abuse; disclosure; generosity and kindness; values; dress; emotional expression. The third pack (n = 23) had a staff behaviour, attitude or attribute labelled on it, for example ‘staff displaying visible self-harm scars’. The magnets fell under the following categories: attitudes to vulnerability; attitudes to divergent opinions, practices and approaches to care; social pressure and courage; victimisation and bullying; team responses to stress and trauma; secrecy. Facilitators were provided with a magnetic whiteboard to facilitate the session, which was installed in a private room in the clinical settings.

Each of the three, distinct, sessions followed the same process:

1. Participants will then be asked to review each boundary and organise the magnets into ‘negotiable’ and ‘non-negotiable’ groups.

2. For each boundary, participants will be asked to:

   1. identify whose need the boundary serves (i.e. staff’s, patients’ or both)

   2. identify the specific need the boundary protects (e.g. safety, respect, equality, cohesion, convenience)

   3. explore ways of negotiating this boundary, including:

      1. alternative ways the same need can be met without setting blanket boundaries on patient behaviour

      2. skills in implementing the boundary safely.

Clinical psychology staff (clinical psychologists and assistant psychologists) were trained in both ‘Conflict formulation’ and ‘Negotiated boundaries’ in one training session (duration 2–3 hours). Invitations were sent to all members of clinical psychology at each of the study sites. Training took place in a booked room, away from the ward areas at each participating clinical site, and was undertaken in the first week of the 8-week intervention embedding phase on each of the 10 participating wards. The training consisted of presenting the rationale for both groups, then practising both groups as they would be delivered in practice. All psychologists trained in the intervention were provided with a detailed facilitator manual. Piloting of the groups indicated that both groups had a duration of approximately 1 hour. Feedback from the expert stakeholders and PPI advisory panel indicated that the conflict formulation model should be used at a frequency of twice per month and the boundaries sessions once per month. The sessions with ward staff were held in a private room away from ward area.

Senior nurse-delivered components

1. 4. Debriefing

The debriefing model aims to:

1. create a culture of accountability where ward staff can learn from critical events without fear of recrimination;

2. create closer working relationships between nursing leadership and ward staff that foster shared responsibility for safety;

3. improve staff’s ability to develop a functional understanding (feelings, needs, context) of incidents where seclusion and restraint have been used, including both staff and patient inputs into conflict.

The debriefing model (Figure 6) was similar to the ‘Conflict formulation’ model but with additional emphasis on data-gathering, diffusion of learning, and shorter-term changes to staff behaviour, the routine or the environment, that could rapidly avoid recurrence of the need for seclusion or restraint. After each episode of restraint and seclusion staff ward staff were asked to complete a proforma with the following sections: incident facts (description of relevant behaviours without causal explanations or subjective descriptions, e.g. ‘aggressive’); your feelings prior to initiation of seclusion or restraint; your needs prior to initiation of seclusion or restraint; context questions (resourcing, location, time, impact of others present, environment) and actions needed to prevent recurrence.

FIGURE 6.

Debriefing model.

An interview proforma for patients to complete, at their preference, with or without either a senior nurse (external to the ward team, a modern matron, operational manager or nurse consultant) or a member of the clinical psychology team, was also developed. This should be completed, voluntarily, and as soon as the patient was well enough to engage with the process. The patient proforma included the following sections: incident description; your feelings before restraint/seclusion; your needs before restraint/seclusion; context questions [behaviour of others, memories, bad news, waiting, staff communication (asked to stop, asked to do, request denial), environment, physical health, time]. Senior nurses and/or psychologists were provided with the EDITION study’s feelings and needs cards packs to facilitate this activity with patients.

The patient proforma was developed with the study’s PPI advisory panel to increase acceptability. This co-production exercise raised the important question of whether the engagement of patients in post-incident debriefing, and the sensitive questions implicated in this (e.g. memories triggering the event), might require patients to re-live traumatic experiences and cause harm. The advisory group were emphatic that the primary trauma in this context is the seclusion and restraint itself, and that gathering of pertinent information from patients is essential to limiting patient exposure to further trauma. The group did recommend that there should be emphasis on the voluntariness of patient involvement in the process.

Once the data-gathering process was complete, senior nurses were asked to populate charts with the feelings, needs, context and resolution data they had gathered from staff and patient proformas. An example populated chart is provided in Report Supplementary Material 3, SM3.12 (not a real scenario). They were then asked to organise a meeting (1 hour duration) with ward staff involved in the incident. During these meetings, staff teams were asked to inspect the chart for its validity and refine, as needed, based on group discussion, and/or populate cells where no data had been collected, then agree on courses of action to prevent recurrence. Once these meetings were concluded, the senior nurse would e-mail ward staff to ensure diffusion of learning to all relevant persons. The debriefing sessions were planned to take place, ideally, after every incident in which seclusion or restraint was used. In environments where restraints and seclusion happened at such high frequency this would be unfeasible, senior nurses were asked to select the event occurring that week that they felt represented the most valuable learning opportunity for teams.

1. 5. ‘Symmetrical feedback’

The symmetrical feedback intervention aims to:

1. enhance the impact of staff and patient feedback on clinical practice and improve safety

2. provide staff and patients with opportunity to feedback anonymously and without recrimination

3. strengthen lines of accountability between clinical leadership and ward staff

4. improve de-escalation practice by creating a culture of open and constructive criticism of practice.

Two metal feedback boxes per ward were installed in patient and staff areas. Ward managers were advised to install them in discreet areas so that anonymous feedback could be provided. This was easily facilitated for staff but presented more of a problem for patient areas owing to potential ligature risk. The feedback boxes were advertised with posters (developed with the PPI advisory panel) displayed on the walls, explaining the purpose, emphasising the option to give feedback anonymously and encouraging feedback. Two-sided feedback proformas were provided alongside the boxes. Proformas were identical for staff and patients and asked respondents to provide: a description of the issue (positive or negative); impact of the issue on staff and/or patients; suggestions to resolve the issue raised; tick boxes to indicate if the respondent would like public, private or no feedback to their concern.

Feedback boxes were emptied once weekly by senior nurses (operational managers, modern matrons or nurse consultants) and checked for urgent safeguarding information. They then collated feedback for once-monthly feedback sessions with ward teams. The structure of these sessions was developed from a review of evidence relating to the impact of feedback on performance. If a shame response is triggered by negative feedback, learning is inhibited, and performance deteriorates rather than improves. 139 A process was, therefore, designed explicitly to avoid triggering shame responses in staff attending the monthly feedback sessions. Accordingly, the structure of each session was designed to frame the process of responding to critical feedback as an intellectual, skills-based (rather than emotional) exercise to make improvements to practice.

To achieve these objectives, the hour-long feedback sessions (held in private rooms away from the ward) were divided into three 20-minute sections. In the first section, attendees were asked to review with the senior nurse and ward manager examples of genuine patient feedback unrelated to their ward, hospital or trust (example feedback forms were developed using genuine patient feedback extracted from publicly available Care Quality Commission reports) and were asked to respond to them as if they were a complaint that had been made about their ward. In the second 20-minute section, the senior nurse and ward manager reviewed feedback from the staff box, inviting suggestions, discussion and reflection from staff on how leadership could make appropriate changes. A key function of this was to role-model open and reflective responses to critical feedback. In the final 20 minutes of the session, the group reviewed the actual patient feedback from their ward to identify actions needed to improve practice and the environment for patients. Sessions were, therefore, structured to ensure that attendees were not exposed to the content most likely to be threatening to self-perceptions until they had been habituated to the process of responding to critical feedback.

Senior nurses were trained in both ‘Debriefing’ and ‘Symmetrical feedback’ models in one training session (duration 3 hours). The training involved presentation of evidence and the rationale underpinning the two models and opportunities to practise facilitation. Invitations were sent to senior nurses (modern matrons, nurse consultants and operational managers) at each hospital. Training took place in private rooms at the clinical sites. All senior nurses trained in the interventions were provided with detailed facilitator manuals.

Methods for measuring adherence to the EDITION intervention

A 30-item checklist for assessing staff adherence to the 11 components of the EDITION intervention was developed and piloted in the WP3 feasibility evaluation.

Staff-reported clinical outcomes

Staff-reported clinical outcomes were collected at four time points at weeks 1 and 8 (start and end of the pre-training period) and weeks 16 and 24 (start and end of the post-training period). At each time point, staff completed the following three questionnaires:

1. (1) Capabilities, opportunities and motivation to use de-escalation (COM-B questionnaire).

The COM-B is a six-item questionnaire that assesses physical capability, psychological capability, physical opportunity, social opportunity, conscious motivation and automatic motivation to engage in a given task. Capabilities, opportunities and motivations are measured on 11-point Likert-type scales. The scale demonstrates reliability (ICCs 0.554 to 0.833) and validity (pairwise correlations rs < 1.0).

1. (2) Attitudes to Containment Measures Questionnaire (ACMQ)

The ACMQ (staff version) is an 11-item questionnaire that measures staff perceptions of the acceptability of restrictive practices on 5-point Likert scales (strongly agree to strongly disagree). The scale demonstrates good construct validity and high internal consistency (α = 0.97). 155

1. (3) Attitudes to Personality Disorder Questionnaire (APDQ)

The APDQ is a 37-item, valid and reliable (Cronbach’s alpha 0.9412) measure of mental health staff attitudes to people diagnosed with a ‘personality disorder’. Items are rated on 6-point Likert scales between ‘never’ and ‘always’.

1. (4) Violence Prevention Climate (VPC)

The VPC is a 14-item questionnaire that measures the VPC on inpatient mental health wards. It has two subscales, ‘staff actions’ and ‘patient actions’, which demonstrate good internal consistency (Cronbach’s alpha 0.89 and 0.76156). Items are rated on 5-point Likert scales (strongly agree to strongly disagree).

Patient-reported clinical outcomes

Patient-reported outcomes were collected at seven time points: three in the pre-training phase, one in the embedding phase and three in the post-training phase. The increased time points for patients were planned to enable the detection of potentially important impacts of training for patients who would already be discharged/transferred at later time points. At each time point, patients completed the following three questionnaires.

1. (1) Perceived Expressed Emotion in Staff Scale (PEESS)

The PEESS is a 20-item measure with three subscales (criticism, intrusiveness and supportiveness). It has demonstrated validity and reliability (subscale alphas 0.72 for criticism, 0.68 for intrusiveness, 0.87 for supportiveness157).

1. (2) Coercion Experience Scale (CES)

The CES is a 44-item valid and reliable (Cronbach’s alpha 0.67–0.93158) measure of the psychological impact of restrictive interventions on patients. It has six sub-scales: ‘humiliation’, ‘physical adverse effects’, ‘interpersonal separation’, ‘negative environmental influences’, ‘fear’ and ‘coercion’. Items are rated on a 5-point Likert scale from ‘not stressful’ to ‘extreme stress’.

1. (3) VPC

As above. The VPC can be administered to staff and patients.

Knowledge and attitudes

On multiple wards staff were observed to lack a trauma-informed approach to care. This presented in the form of dismissive responses to patient distress, callous humour when discussing a patient’s traumatic experiences with colleagues and patients indicating that they did not feel comfortable approaching members of staff for support. There was potential evidence of compassion fatigue evidenced by trivialising responses to patient expressions of anxiety, for example ‘You’re definitely not dying’, avoidance of distressed service users and conversations between colleagues mocking patients. In contrast, there was evidence of staff awareness of the negative effects of stigmatising language, particularly about personality disorders. Gaps in staff knowledge in relation to the ward environment and risk were inferred from failures to monitor trigger points and make efforts to modify what appeared to be an environment that was too chaotic for anxious patients to tolerate.

Skills

Deficits in interpersonal skills in staff were observed, for example, in social and self-awareness (staff were observed whispering and laughing to each other in communal areas in a manner that appeared intended to exclude patients). Researchers noticed a style of nursing communication which they attributed to the desire, among staff, to avoid reinforcing behaviour that they found difficult to manage (e.g. patient criticism of the ward regime, other concern or reassurance-seeking). This style of communication involved withholding facial or verbal empathy when cues of distress were expressed and in situations where it appeared socially congruent to respond with empathy. This behaviour was perceived as intentional and underpinned by an understanding of psychological behaviourism that informed them that these responses would be most likely to make the problem (and the person) ‘go away’. Because of these interactions, patients were often left without interventions to reduce distress.

Emotional expression and attitudes to vulnerability

Two wards in particular presented evidence of staff valuing emotional authenticity in interactions with patients and interactions between colleagues. This was evident in displays of empathy and sincerity when interacting with patients and discussion of the emotional impact of their work with colleagues. Staff openness with patients in terms of their emotions was observed in younger staff or new starters. On the other wards observed, there was no evidence of discussion of emotions between colleagues and no expressions of vulnerability in staff interactions with patients, indicating that an outward appearance of emotional resilience was a valued aspect of maintaining safety. Evidence of attitudes towards vulnerability was also observed in interactions between ward staff and the observing researchers. On three of the participating wards, researchers described feeling ‘taken care of’, ‘supported’ and ‘included’. This might have suggested a more protective attitude to more vulnerable colleagues, but researchers did note that these actions did not appear to correspond with treatment of bank and agency staff.

Social distance and the social environment

Researchers observed marked social distance between ward staff and patients, exemplified at the most fundamental level by ward staff’s apparent preference for congregating in the ward office rather than sharing the social spaces of patients. The ward office door was continuously noted as a source of division and conflict between staff and patients. Nursing offices doors were, typically, kept shut. Service users were often observed knocking on the office door in attempts to seek staff support or make requests, which were commonly ignored by staff in the office. Some patients would make attempts to enter the office, or knock profusely, or become irate due to being forced to wait for a response from staff. In these instances, staff would typically leave the office to provide support, to reprimand or re-direct patients away from the office.

On most wards, conscious attempts to reduce social distance between staff and patients were observed, at least superficially. In some instances, researchers characterised these attempts as ‘performative’: for example, engaging warmly with patients when staff needed information or a bureaucratic task completing, and routinely sitting with patients at mealtimes but only interacting with other staff. In one of the female forensic wards, there was signage displayed to enforce social distance between staff and patients. For example, extending beyond the ubiquitous ‘please knock’ signs, this ward had a poster that instructed patients not to come within 6 feet of the nursing office door. This had been installed as a solution to the disturbances patients were causing attempting to rouse staff from inside. Other signage included ‘Zero Tolerance’ and ‘Respect Us’ posters. These posters typically presented kindly and receptive-looking people dressed in nurses’ uniforms, and appeared to communicate a message proximal to ‘please don’t hit me, I’m only trying to do my job’. The image these posters conveyed of professionals often seemed at odds with the attitudes and behaviours staff displayed towards and around patients.

Notably, social distance was much less marked between patients and non-regular staff, indicating either that unfamiliarity with the staff team increases motivation to engage therapeutically with patients or that the desire for social distance in permanent staff increases due to burnout and/or exposure to toxic culture. Despite these negative observations, researchers noted ‘exceptions to the rule’ in staff teams, where individual staff members would engage with patients in a warm, approachable and authentic manner. Again, they did observe that these individuals tended to be either non-regular staff or staff who were relatively new to the team. Researchers noted that patients approached these staff to get their needs met, repeatedly, and when alternative staff appeared to be available.

Organisational context

Insufficient staff numbers were commonly observed across all wards. There was an important difference in organisational context observed. On the two adult forensic female wards, researchers observed a near-total absence of more senior clinical managers and clinicians. Importantly, these two wards were observed as having the most hierarchical behaviours and relationships between staff groups and between staff and patients. For example, they observed that ward managers had a separate office, off the ward, and that qualified nurses often spoke to support workers and housekeepers in a condescending manner. These observations might indicate that the modelling of social relationships at more senior layers of the hierarchical structure permeates to ward-level relationships, including those between staff and patients.

Knowledge and attitudes

Evidence of staff practice changes because of increased trauma awareness was observed on one of the wards in the post-intervention period. Ward staff displayed an increased awareness of patient triggers and the importance of not undermining patients. The negative attitudes to patients expressed in conversations between staff in the pre-intervention period persisted. There was evidence of changes in knowledge and behaviour in relation to sensory modulation.

Skills

Researchers observed changes in staff skills after training. Field notes reflect adjustments of tone and body language to being calm, open and less threatening. On one ward, multiple instances of staff giving patients space and time (e.g. being willing to tolerate escalations without use of containment) were observed, which was emphasised in the de-escalation training. Honesty and clarity about the timeliness of completing patient requests were observed across most wards, with staff providing clear explanations of when requests would be met, or why they couldn’t be. Although some instances of staff not keeping promises and taking excessive time to follow through with requests were noted, these were in the minority. Staff communication and social skills were generally observed as having improved. The researchers observed evidence of patients appearing increasingly comfortable approaching staff because of these changes.

Emotional expression and attitudes to vulnerability

Researchers observed changes in the respect in the post-intervention period: for example, greater active inclusion of student nurses in conversations, more welcoming receptions to visitors to the ward and increased mutual support in the staff teams.

Social distance and the social environment

Significant physical and social segregation between staff and patients remained present on wards but active attempts by staff to reduce this were observed. For example, staff were seen actively joining groups of patients in activities, sitting with patients. Although the office door often remained a physical barrier between staff and patients and a source of conflict, staff congregation appeared less commonly. Researchers noted increases in staff emphasis on patient choices and autonomy. On one of the female adult forensic wards, a psychiatrist was observed extensively engaging patients with the intervention and encouraging patients to engage with the environmental audit tool and attend community meetings to discuss outcomes.

Systems and organisational context

There were no changes in the staff resourcing problems observed in the pre intervention but, as alluded to in previous themes, there was some evidence observed of senior clinicians on the wards engaging patients in the intervention.

Psychiatrists (ward rounds and collaborative prescribing)

Affective attitude Most psychiatrists in PICU and forensic settings felt that the collaborative prescribing intervention was not feasible in these settings. PICU patients were perceived as lacking capacity when prescribing decisions were made (i.e. close to admission). By the time they were well enough to engage they were ready to be transferred to a less-restrictive setting or be discharged. Psychiatrists in forensic settings pointed to a lack of opportunity to use the intervention, within the study time frames, as patients were often on stable medication and changes were needed infrequently (Q1).

Burden Psychiatrists found some of the ward round standards burdensome. The requirements to escort patients to and from the ward round room and facilitate patient access to ward round records generated additional workload and increased the duration of ward round (Q2).

Confidence There was evidence of a lack of confidence among some psychiatrists in giving patients meaningful choices. Two PICU psychiatrists gave examples of choices they would be willing to grant, such as between oral or intramuscular medicines, but did not feel confident in allowing patients to choose a specific medicine. This was grounded in scepticism of patient skills and knowledge to make informed choices and concern over accountability if unsafe events occurred after a patient had chosen their own medicine (Q3). Confidence issues also affected psychiatrist engagement with the ward round standards. Some felt insufficiently confident escorting patients from ward rounds because they are not trained in physical restraint techniques.

Opportunity costs A minority felt that existing collaboration in prescribing (this was limited to information provision on side effects) was adequate. From this perspective, any novel intervention targeting collaborative prescribing represented opportunity costs without benefit. Others were concerned that increasing patient involvement would result in changes to patient expectations that could not be met. They were concerned that patients might learn to demand changes to their medicines every time they encountered a distressing side effect (Q4). They felt that increasing patient sensitivity to side effects and involvement in their minimisation might reduce adherence and increase medication-related conflict with staff. They emphasised that none of the currently available antipsychotics are without distressing side effects. Psychiatrists also pointed to opportunity costs associated with ward round standards. For example, PICU psychiatrists stated that while they understood the rationale for not asking patients about symptoms in ward rounds, they observed that they did not always have time to see patients in any other context except ward round. As such, the ward round was the only setting in which vital clinical and risk information could be gleaned from patients.

Ethicality Some PICU psychiatrists, while recognising, in principle, the value of increasing patient choice in prescribing decisions, felt that the need to manage risk outweighed the benefits of increased autonomy (Q5). They provided examples of patients choosing medicines that they were happy with but would not adequately reduce their risk to others (this perspective missed that the intervention allowed prescribers to limit choices to medicines they were willing to prescribe). Ideals in relation to doctor–patient relationships also influenced how acceptable they found the intervention. Two psychiatrists appeared to characterise the ideal doctor–patient relationship as the sharing of knowledge and advice by a responsible expert with a naïve and passive recipient of medical intervention. They felt, fundamentally, that patients relied on them to make important decisions on their behalf.

Intervention coherence Important issues related to understanding were raised but these often related to broader concepts (e.g. shared decision-making and collaborative working) rather than the specific interventions. Psychiatrists often claimed to work collaboratively with patients but rarely provided examples that extended beyond information provision (Q6).

Perceived effectiveness The consensus of PICU and forensic psychiatrists was that the intervention may be effective in reducing medication-related conflict in general adult acute mental health settings, that is, where patients have the capacity and present lower risk. Multiple psychiatrists felt some of the ward round standards may cause paradoxical increases in conflict: for example, that offering patients the option of using first names might reduce patient respect for psychiatrists or generate alienation by failing to recognise the power imbalances between patients and psychiatrists (Q7).

Psychologists (conflict formulation and negotiated boundaries)

Affective attitude Psychologists uniformly reported staff enjoyment of the sessions, that the formats of the groups were stimulating, novel, safe and, occasionally, humorous. They often commented that the groups stimulated alternative perspectives on patients and boundaries that would not have been elicited without the groups. They did report that the conflict formulation group placed demands on staff to be vulnerable in that it required them to label and articulate feelings in in the presence of colleagues. The format required participation from all present, so there was no opportunity to avoid this entirely. Psychologists commonly observed initial discomfort and reticence in staff and there was conflicting evidence on whether this problem was resolved through repeated exposure. Some felt that staff’s willingness to be vulnerable increased through repeat attendance and others felt less emotionally open staff did not increase engagement over multiple sessions.

Some psychologists reported anxiety that the conflict formulation group would lead to disclosures of historical trauma by staff participants that they would be unable to contain. Although they observed no evidence of psychological distress, they did highlight that the groups tended to be ran with groups of staff that they knew quite well. There was a view that the psychological competency of facilitators and their relationship dynamic with the participant group need careful consideration to reduce the risk of adverse effects. A minority of psychologists reported unwanted emotions during the groups. One reported that the groups revealed punitive attitudes among attending staff that made her feel angry and that they found it difficult to suppress this emotion when providing alternative perspectives (Q8).

Burden Burden was reported in terms of time required to prepare for and facilitate sessions. There was uniform agreement that the time required was worth the investment given the value they attached to the groups. Participants reported that preparation time reduced with familiarity (Q9).

Confidence Two psychologists reported their confidence was affected by awareness that the groups were part of a research trial and that, because of this, they wanted to deliver them well. Another described feeling a lack of confidence managing discussions where there was marked imbalance of opinions in the group. They expressed a lack of confidence in reducing distress in staff holding marginal perspectives (Q10).

Ethicality Psychologists felt that the values of the groups aligned with the values of the participating wards and their own. They did discuss value clashes between group participants but felt these were positive and an intended function of the groups (Q11).

Intervention coherence All psychologists demonstrated understanding of the intended functions of the groups and indicated that attending staff shared this understanding. However, some psychologists reported that the card exercises involved in the Conflict Formulation model were not immediately intuitive to attending staff, especially when it came to identifying their own feelings (Q12). There was evidence that the distinction between EDITION’s novel reflective practice groups and existing reflective practice had not always been made clear in the promotion of the groups to staff and that, in a minority of wards, this might have reduced attendance.

Perceived effectiveness Psychologists felt that the groups were effective in increasing staff’s awareness and flexibility in boundary application with patients. A secondary outcome perceived was greater team cohesion and enhanced empathy between colleagues. There were caveats to this, in that these were perceived from observations of groups rather than actual observations of changes in practice on the wards. Some were uncertain of how long these changes would be sustained (Q13). They reported changes in staff’s functional understanding of patient behaviour, especially in relation to perceived motives.

Occupational therapists (sensory modulation and support plans)

Affective attitude Occupational therapists were positive about the sensory modulation and support-planning intervention but were concerned about safety (Q14). They felt they lacked the in-depth knowledge of sensory assessment to deliver the intervention safely. They were concerned about adverse reactions to sensory intervention. These concerns were found in one participating trust but not in the other.

Burden Burden associated with the intervention included identifying COVID secure spaces for the room and disinfecting the equipment before and after use. Some of the equipment was quite expensive and there were concerns that opportunity costs would mount as equipment would be damaged by patients and require replacement (Q15).

Opportunity costs Participants identified the potential for damaged relationships if a patient with sensory impairment has an adverse reaction to a piece of equipment (this did not happen during the study). They observed that time spent engaging patients in sensory interventions comes at the expense of alternative interventions. Because the intervention was occupational therapist (OT)-led, and OTs were external to the ward team, it was felt that that it could cause alienation in nursing staff who might perceive the intervention as another ‘top-down’ innovation imposed without adequate support (Q16). OTs recommended that the training and implementation approach should be more joined-up between OTs and nursing staff in future evaluations.

Ethicality OTs broadly endorsed the values underpinning the intervention, which they felt were consistent with OT ‘models of practice’ and that sensory modulation was an important gap in current provision (Q17). However, in one of the trusts, OTs felt that the infrastructure to upskill the workforce and deliver the intervention was lacking. Consequently, they felt they had a duty of care to patients not to implement an intervention that may cause harm. OTs at the other trust, who held the view that a simple assessment of sensory items patients found alerting and calming was sufficient to maintain safety, raised no concerns.

Intervention coherence OTs demonstrated an accurate understanding of the aims of the intervention, that is, to increase patient access to sensory intervention through the upskilling of nursing staff, to manage patient distress proactively rather than reactively (Q18) and to modify adverse environmental impacts on patient arousal and affect.

Perceived effectiveness OTs felt the intervention had the potential to be effective (Q19), with caveats. They were concerned that the intervention could be overstimulating for acutely ill patients. Simplifying and making the plans more accessible to patients was felt to enhance effectiveness.

Senior nurses (debriefing and symmetrical feedback)

Affective attitude Senior nurses presented positive attitudes towards the two senior nurse-led components and the EDITION intervention overall but described a perception that some ward staff members were overwhelmed with the breadth of changes demanded by the project (Q20). One senior nurse expressed frustration at the lack of engagement from ward staff and patients on their allocated ward. Again, this was the low secure male ward reporting low levels of conflict and containment, suggesting that the EDITION intervention may be more suited to acute settings.

Burden Senior nurses described substantial time commitment invested in the two intervention components and identified engaging staff in completing the debriefing and feedback forms as the most time-consuming elements (Q21). They reported ward busyness as a major constraint to nursing staff’s engagement. There was an overall view that the EDITION interventions had real value but would be optimised by implementation support provided by a seconded band 6 nurse (it was felt 0.4FTE would be sufficient time).

Confidence Some limitations were identified in the training provided to senior nurses. One participant felt that the training was too brief and that providing training inclusively of ward staff would have enhanced implementation and confidence. Confidence was further reduced by COVID-19 delays, which lengthened the time from training to implementation (Q22). Confidence was perceived as increasing through the process of delivering the interventions and engagement with the research team. The senior nurses recommended refresher training and regular supervision with the research team and intervention leads at other study sites, to enhance confidence.

Opportunity costs Senior nurses reported that they prioritised the EDITION over existing interventions. As such, this resulted in already embedded interventions not being utilised (Q23).

Ethicality Senior nurses recognised the value in the debriefing intervention and the importance of providing opportunities for ward staff and patients to raise issues with management directly. Most felt the intervention aligned with ward staff values and participating trusts (Q24). They did observe different attitudes towards the interventions between established and junior/inexperienced staff. They felt more experienced individuals often held core restrictive beliefs that were resistant to change and misaligned with the philosophical orientation of the interventions.

Intervention coherence The feedback in terms of understanding of the intervention, again, related to the large number of intervention components and the resultant confusion this caused among ward staff. They reported that this confusion resolved through staff engagement with the interventions over time. Some described patient written content on debriefing and feedback forms that indicated they had not understood the purposes of the interventions.

Perceived effectiveness Senior nurses deemed the recommended debriefing process was effective but that running debriefing sessions with staff from busy acute wards was not always feasible. One senior nurse felt that, while recognising the need for closer working relationships between senior clinicians and ward staff, the existing hierarchical disconnection may have reduced staff’s willingness to be open and honest during debriefing sessions within the trial period. Senior nurses in forensic settings reported much fewer problems with contextual feasibility and felt that both interventions were highly relevant to their clinical environments (Q25). These nurses planned to continue implementation after the study had finished, including wider roll-out to other forensic settings in the trust.

RRPIs (de-escalation training)

Affective attitude Positive views of the EDITION de-escalation training were expressed by RRPIs at all study sites. They valued the content on trauma, the group activities and the service user and carer researcher facilitation (Q26). All RRPIs were extremely positive and enthused by the wider project and happy to have been involved in its implementation.

Burden Because a lot of the training content was perceived as ‘out of the comfort zone’ for RRPIs there was a significant burden identified in terms of reading and re-reading the facilitator manuals outside of the 1-day ‘train-the-trainer’ training (Q27). There were practical burdens associated with delivering the training in the context of the COVID-19 pandemic, in terms of new restrictions on room capacity. One of the video case studies involved a patient spitting at a member of staff and one of the RRPIs had experienced this in their clinical practice. They reported that this resulted in the re-living of this experience and mild distress. However, they had used this experience as an education tool to explain, to trainees, the neurobiological processes involved in re-experiencing traumatic events in response to interpersonal/environmental stimuli.

Confidence RRPIs commonly reported a lack of confidence, especially in the delivery of the neurobiological aspects of the trauma module (Q27). They expressed anxiety about being asked questions that they could not answer while delivering the training. Confidence improved through delivery of the training but, like the senior nurses, they wanted more supervision meeting with members of the research team and suggested a forum for RRPIs to discuss the material and problem-solve in any future evaluation of the training.

Opportunity costs RRPI involvement in the delivery of the de-escalation training required delegation of normal training to colleagues. Other than mild guilt associated with this RRPIs reported no opportunity costs associated with the delivery of training.

Ethicality Broadly, trainers felt the values and ethics behind the training package aligned with their own personal and professional beliefs in terms of increasing the ‘patient-centredness’ of inpatient mental healthcare. There was less certainty in views as to whether the training content aligned with ward staff’s views. Like the clinical psychologist who reported experiencing anger at the punitive attitudes the Negotiated Boundaries exercise elicited, one RRPI described feeling ‘angry and ashamed’ by the cynical attitudes to increasing patient control over their experiences of inpatient care. The discussions this RRPI referred to related to the Creating Trauma-Sensitive Clinical Environments (module 2), which introduces a range of measures (Patient Handover, Patient-Reported Environmental Audit Tool, Collaborative Prescribing) designed to reduce patient experiences of powerlessness. Encouragingly, the RRPIs reported that many of the participating wards are continuing to use the EDITION interventions and that the RRPI teams continue to use the de-escalation training content.

Intervention coherence Prevention and management of violence and aggression (PMVA) trainers presented a coherent understanding of the training package and the project. Their main observation in response to interviewer questions about ward’s staff’s understanding was that they felt some staff had the ‘misconception’ that the training attempted to excuse intolerable patient behaviours. However, this may have represented values differences between the RRPIs and the research team, on one hand, and ward staff, on the other, rather than differences in understanding.

Perceived effectiveness RRPIs felt that the training had improved attitudes in staff who had attended. There were different perspectives on how the impact on service user and carer facilitation could be maximised. The research team consciously avoided the use of lived experience ‘testimonies’ on the advice of our PPI advisory panel. Rather we wanted the service users and carers to be educators and facilitators of the training content. Some RRPIs felt that this resulted in the servicer user researcher facilitator’s (Grundy) lived experience not being properly contextualised until the final module of the day (focused on nursing interventions to reduce isolation and shame associated with voice-hearing). Other RRPIs disagreed with this perspective and felt that having the voice-hearing content towards the end of the training allowed relationships between the ward staff and the service user and carer researchers to develop throughout the day without preconceptions about diagnoses influencing this process. On the contrary, they felt this enhanced rather than diminished the emotional and educational impact of the content on voice hearing.

Ward staff (Patient Handover, Patient-Reported Environmental Audit tool, Insiders’ Guide and Welcoming Committee)

Affective attitude In general, ward staff and ward managers identified value in the interventions and benefit from attending de-escalation training. Most participants described the training as novel and informative and appreciated the group activities and the service user and carer facilitation. There was an exception. On one of the wards, a ward manager reported that her staff had found some of the content condescending and more in keeping with environments that had not already done significant work to reduce the use of restrictive interventions (Q28). This participant drew attention to a video case study in which PRN was used without consideration of alternatives when, in fact, none of EDITION’s video case studies included a scenario of this nature (there was one video case study in which a patient is refused PRN due to there being no ‘objective signs of distress’). The observation that this feedback came only from a ward with a particularly strong identity in terms of reducing use of containment interventions raised some interesting possibilities. Either the ward was already functioning at a level too advanced for the EDITION training or receptiveness to evidence-based practice is reduced in wards where a strong sense of positive identity has been developed through engagement with practice-based evidence and innovation.

The non-training-based interventions were particularly valued by ward staff participants. They were evidently pleased with the reflective practice models, which were experienced as useful and interesting. There was one important exception. One member of ward staff felt the Conflict Formulation model had the potential to raise difficult memories for staff and cause distress. Ward staff felt generally that the interventions to involve patients more in care and the maintenance of safety (Patient Handover, Patient-Reported Environmental Audit tool, Insiders’ Guide and Welcoming Committee) were needed and, perhaps, most encouragingly, that they were valued by patients who were keen to engage with them. One ward manager held the view that ward staff were not sufficiently qualified to engage patients with sensory modulation equipment and that this should be the preserve of OTs alone. Overall, ward staff expressed some of the most positive views of the EDITION interventions among all professional groups. This is, perhaps, surprising given that they assumed much of the burden of intervention. It is possible that attendance at de-escalation training with the whole team protected nursing staff from the ‘silo effect’ that other professionals identified as a limitation of the project, or that it was a consequence of selection bias in the ward staff who were interviewed.

Burden Substantial burden was associated with implementing the interventions. Some of the recommended sensory equipment was expensive and all of it required a financial investment by the wards. There was additional burden associated with the sensory equipment in the context of the COVID-19 pandemic, as a raft of new requirements for infection control were introduced. Burdens associated with the ward interventions were exacerbated by staff reports of heavy workloads (Q29), high acuity and the introduction of other new initiatives by the trust that coincided with the EDITION project. One staff member drew particular attention to burden associated with the patient handover intervention and felt that the time this intervention demanded was likely to outweigh any potential benefit.

Confidence The interventions delivered by ward staff were generally felt to be within their remit and abilities. With regard to the training, one member of nursing staff reported that the de-escalation training had significantly increased their confidence as it had validated their existing practices (potentially indicating that their existing practice contrasted with other team members’).

Opportunity costs The main opportunity costs of the interventions, from ward staff’s perspective, was that they had to be completed either in addition to essential routine activities or instead of additional activities that could have been completed.

Ethicality Nursing staff and ward managers, broadly, felt the training and interventions were aligned to their organisation’s values. These participants emphasised the importance of ‘collaborative working’, ‘least restrictive practice’ and ‘patient-centred interventions’. One staff member highlighted the importance of ensuring those staff assigned to lead intervention components shared the values of the project (implying they did not always feel that had been the case during the study).

Intervention coherence Ward staff demonstrated a clear understanding of the aims of the project in terms of enhancing de-escalation and reducing conflict and containment. However, they reported that patients’ understanding of the interventions was not always apparent (Q30). There were multiple reports of patients misunderstanding the purpose of interventions, attributed partially to a lack of capacity on the part of patients and partially to lack of adequate explanation by the ward and the research teams.

Perceived effectiveness Most commonly, staff felt the training was valuable and interventions provided a simple but effective way with which to improve care. Some staff felt the overall project had created a safer working environment, offered valuable opportunities for reflection and important dialogue, and offered a fresh perspective. Others felt the interventions had made little-to-no difference to the overall ward environment and staff–patient interactions. Again, participants expressing this view worked on the ward that had a strong, pre-existing identity centred on the reduction of containment interventions.

The sensory modulation and support-planning intervention was perceived as effective in enhancing staff’s understanding of patients and providing a safe space for patients to escape distressing sensory input/interpersonal conflict (Q31). Ward staff felt the Insiders’ guides helped in both managing patient expectations and providing reassurance on admission. Participants reported that the guides remained in use after the study, which provided an important indicator of acceptability and perceived clinical value. The Patient-Reported Environmental Audit tool was reported as providing a ‘helpful, fresh perspective’ on environmental safety by involving patients in decision-making.

Patients (Patient Handover, Patient-Reported Environmental Audit tool, Insiders’ Guide and Welcoming Committee)

Affective attitude Patients from forensic inpatient settings presented overwhelmingly positive and some negative sentiments to the interventions they experienced. They felt the Symmetrical Feedback and Debriefing interventions were good initiatives and provided positive feedback on the layout of My Handover, debriefing and feedback forms. Multiple patients felt that the interventions addressed important shortfalls in existing practice. One patient felt strongly that the involvement of patients in completing debriefing forms was an important positive change in practice as it allowed patients an opportunity to express their version of events following use of restraint and or seclusion.

However, some patients stated they were happy with current practice and so felt the suggested changes were unnecessary. Some patients expressed discomfort with the idea of addressing psychiatrists by their given names and others stated they didn’t feel escorting patients back from ward round to the ward was an appropriate job for consultant psychiatrists. Patients also expressed negative feelings related to the tedium of completing paperwork as a part of the interventions. Some reported a perception that interventions were not completed properly, or with care, by staff, specifically the patient handover, debriefing and feedback forms.

Patients provided varied feedback on sensory and support plans. Some felt they were a good initiative and could increase staff understanding and were impressed by the idea of introducing sensory activities and personalised plans (Q32). Others felt that sensory interventions were childish and condescending for use with adult patients. Collaborative prescribing was felt to be a positive intervention, but patients were sceptical as to whether doctors would adopt the intervention. The Insider’s guide also brought mixed feelings from different patients in forensic settings. Some reported feeling proud of their involvement in the guides and others that they were inappropriate for these setting as they suggest the ward was like a hotel not a hospital.

Burden Few burdens were identified by patients. Interestingly, they presented conflicting perspectives of the effect of the interventions on staff workload. Some felt that interventions increased workload and strain on staff, with particular emphasis on the patient handover, which patients felt was completed too regularly, and added significant unnecessary work as only important information needed to be handed over. In contrast, other patients felt the patient handover increased efficiency by making staff more proactive in addressing patient needs. Patients recognised that staffing resource was lacking and as a result engaging in additional activities increased burden for an already strained workforce (Q33).

Confidence Of the few comments provided by patients with regard to their confidence in engaging in interventions, only the feedback intervention was discussed. Multiple patients felt concerned about giving feedback due to fears of negative consequences, or due to generally feeling too timid to provide feedback on the ward (Q34). They noted how giving feedback directly through their primary nurse might increase confidence to engage in the intervention. These findings rather undermined the assumption implicit in the intervention that providing opportunities to give feedback anonymously would provide patients with the confidence to provide honest feedback on practice.

Ethicality Some patients perceived that staff were not taking some interventions seriously, specifically the debriefing and feedback interventions. One felt that the values of staff should be questioned if they did not want to engage with interventions that they felt were clearly designed to protect patients and improve quality of care (Q35).

Intervention coherence Some patients reported inadequate explanation of the interventions, and some felt that the language used in the sensory and support-planning documents was too complex and needed to be simplified (Q36).

Perceived effectiveness Patients perceived a potential overall benefit to the project and, in the environments where it was implemented, an observable positive effect of the sensory modulation and support-planning intervention. Staff support for and buy-in to interventions was felt to be pivotal for their success. On some of the wards, patients reported than the patient handover and debriefing interventions had reduced staff–patient conflict and improved staff response to patient expressions of distress. Again, patients observed that the patient handover had continued to be implemented after the end of the trial phase of the EDITION project, indicating perceived value and acceptability.

Patients felt that adjustments to the time frames of certain interventions would have improved effectiveness. Many patients highlighted that debrief forms needed to be provided to patients with temporal proximity to when the incident had happened after seclusion had been terminated, but with a delay of no more than 2 weeks after this. They suggested forms were often provided after a significant delay following the incident, which they felt limited effectiveness.

Differing feedback was provided about when was best to provide the Insiders’ guides to new admissions. Multiple patients felt the guide should be provided early in the admission process, with as early as the first 24 hours being suggested by one patient, while another argued they should only be provided after the first week to allow time to process the admission experience. A further suggestion was made to give patients time to consider all the information in the guide before being introduced to the welcoming committee.