Design of a tailored and integrated violence prevention program in psychiatric wards and prisons

Abstract

BACKGROUND:

Violence and threats of violence against personnel at psychiatric wards as well as in the prison service is a major work environment problem. To date results from interventions to prevent violence and threats in these sectors have been inconclusive or of small effect. One of the reasons may be that violence and threats of violence occur as a consequence of a complex interaction between employee-level and management-level factors.

OBJECTIVE:

To design a tailored and theory-based intervention program directed at violence prevention in psychiatric wards and prisons that integrates the employee-level and management-level, and development of an evaluation design building on the Context, Process, and Outcome Evaluation Model.

METHODS:

The study follows a stepped-wedged design with 16 work units entering the intervention in four groups with differing start dates from September 2017 to January 2019. The context and process evaluation includes: calculating the implementation degree; mapping of contextual factors; interviews with unit-leaders and employees before and after the intervention. The outcome evaluation includes performing multi-level statistical analysis on data from a three-monthly questionnaire to employees at the participating workplaces.

RESULTS:

The first results will be available in 2020.

CONCLUSIONS:

The comprehensive evaluation of the intervention will give insight into the processes and effects of the intervention.

Keywords

ISRCTN86993466 Organizational intervention process evaluation stepped-wedged design

1 Background

Violence and threats of violence are a critical work environment issue in the healthcare sector and the prison and probation service, where employees are in close contact with patients or clients throughout their working hours.

In the following paragraphs, we will review the most recent research concerning the risk of violence, approaches to the prevention of violence, and theoretical considerations in relation to how interventions will work in practice.

1.1 Risk of violence and health consequences for staff in the healthcare sector and the prison service

Two of the most exposed occupations in the healthcare sector and the prison service are mental health staff and correctional officers [1 –3].

In a review on workplace violence in US mental health staff, Piquero and colleagues found prevalence rates of violence between 48% and 65% and prevalence rates of threats of violence between 55% and 68% [1, 3].

In Denmark, the prevalence of violence and threats of violence in mental health staff and correction officers is also high. In a study of 35 psychiatric worksites, 64% of the employees reported threats in a scolding manner and 34% reported that occasionally a hard object had been thrown towards them within the last year [4]. In the same study, all 83 worksites of the Danish Prison and Probation Service participated and in this sector 49% of the employees reported threats in a scolding manner and 8% reported having a hard object thrown towards them within the last year [4].

Experiencing violence and threats at work is associated with a number of health consequences such as symptoms of anxiety and depression, post-traumatic stress disorder, and reduced general health [5 –13]. Moreover, violence can lead to severe physical injuries that needs medical treatment [14, 15]. Also, studies have shown that violence and threats at work are related to negative outcomes such as dissatisfaction with work, low work engagement, risk of sick leave, intentions to leave and lowered productivity [6 , 16–19].

1.2 Prevention of violence and threats: Existing empirical and theoretical knowledge

In general, results from interventions to prevent violence and threats of violence have been inconclusive or of small effect. The majority of interventions to prevent violence at work has been directed at employees, in particular by the use of various training methods [20 –22]. These types of interventions often show improvements in knowledge and competencies, but only seldom measure effects on actual prevention of violence or threats of violence, so that the effect of these trainings on violence prevention is unclear [20, 23]. Systematic reviews show limited evidence for the effects of the hitherto tested interventions that comprises environmental modifications, training, and changes in policies and practices [20 –23]. This might be due to methodological shortcomings identified in the existing studies, such as small sample sizes, no control groups, use of non-validated measurements or failure to account for processual and contextual factors [20, 21]. The limited effects in previous studies might also be due to a narrow focus on only one specific intervention activity. Studies have shown that violence and threats of violence seem to be caused by the complex interaction between employee-level factors and management-level factors [24 –26]. A prospective multi-level study with 3011 employees from the human service industries found that psychosocial work environment factors, such as high emotional demands, low predictability, low level of influence over own work-situation, high degree of role conflict and low supervisor quality, were significantly related to higher risk of exposure to violence and threats at work after adjusting for relevant confounders [26].

This calls for violence prevention programs that include activities on several levels e.g., by integrating training of employees, modification of the environment, and organizational efforts [20 , 28]. It is necessary to tailor the intervention to the specific workplaces since the reasons that workplace violence occur are not only complex, but also vary strongly between occupations as well as between the individual workplaces [20, 21]. Yet, such comprehensive and tailored interventions are rare. To the best of our knowledge, the only studies that tested interventions on multiple levels (employee-level and group-level) in a rigorous evaluation design is The Safewards cluster randomized controlled trial [29] and a randomized control test (RCT) study of a Participatory Actions Research Approach to violence prevention in hospitals [30]. In the Safewards study, 10 small interventions directed at teamwork and the relation to the patients were conducted in 31 psychiatric wards; and showed a statistically significant decrease in rates of conflict, a known risk factor of violence or threats [12, 29]. However, no process data was collected making it difficult to understand how the intervention worked in the different units. Moreover, the Safewards study did not include interventions on management-level e.g., addressing the managements’ role in implementing and prioritizing the intervention activities nor did they allow for tailoring of the interventions to the specific workplaces. The RCT study by Arnetz and colleagues went a step further and tested tailored individual and organizational interventions that were developed by the units based on a risk assessment that gave information about rates of violence and threats of violence. The study involved 41 units from seven different hospitals with 21 units in the intervention group. They found a significant reduction in incident rates of violence and threats per 100 full time employees at six months follow-up, but not at 12 months. The limited effects of the existing intervention studies call for the development of more advanced intervention approaches.

From a theoretical point of view, the multi-level approach to workplace safety has been emphasized by Dejoy [31]. The core feature of this framework is to integrate interventions on management-level with interventions at employee-level and make the two groups jointly identify relevant safety security problems and possible solutions. The theoretical framework was originally developed for occupational safety, however violence prevention can be considered part of occupational safety and the framework has been supported empirically in two different studies, one of which examined violence prevention practices [24, 32]. Violence prevention practices can occur at all levels in the organization and comprises, for example, encouragement to register violence and threats of violence and prioritization of violence prevention in economic decisions. A prospective study with 3011 employees from the human service industries found that in units where top-managers, managers and employees all had higher levels of prevention practice, the employees were at significantly lower risk of being exposed to violent episodes. This implies that “ ... when managing workplace violence in high risk areas of human service work, there should be emphasis on the use of the violence prevention behaviors from top management, supervisor, and among coworkers” [24]. Dejoys’ framework has also been tested in a prospective controlled intervention study with 22 Danish small and large construction material companies [32, 33]. Kines and colleagues operationalized the integration of the management and employee-level through several activities. For example, by conducting steering group meetings with the participation of both managers and employees. Results showed that the intervention had a significant effect on the 8 small companies’ (10–19 employees) safety climate, a factor that has been shown to be related to actual accident rates [34].

The present study will draw on this operationalization of Dejoy’s theory and the intervention activities will be described in detail in Section 2.2.

In this study, we will test a comprehensive intervention approach to violence prevention that:

Integrates employee-level and management-level activities

Is theory based

Is tailored to the specific needs of the workplace where the intervention is taking place

1.3 Evaluation approach

The evaluation of a multi-leveled tailored intervention requires a coherent and well-planned approach that takes into account both context, implementation, and effect [21].

1.3.1 Context

Evaluating a tailored intervention places special emphasis on the interplay between the context, the intervention and the implementation processes, and how this is affecting the working mechanisms of the intervention, and in turn the outcome. This approach requires an understanding of context, not only as a facilitating or hindering factor as seen in some process evaluation frameworks [35, 36], but as a reciprocal, transformational relationship where the context shapes the intervention and the intervention affects the context. This understanding is found in the Context, Process, and Outcome Evaluation Model for Organizational Health Interventions from Fridrich and colleagues [37] that we will use for evaluating the intervention. In this framework, a further distinction between omnibus and discrete context is made to facilitate analyses. The omnibus context is the wider context of the implementation process e.g., occupation, location and time. The discrete context refers to individual, group, leader, and organizational factors (IGLO – factors) relevant for the implementation process [37]. We will examine the impact of contextual factors at both levels.

1.3.2 Implementation process

To distinguish implementation failure from program failure it is important to closely monitor the implementation process [38]. To know whether an effect is attributable to the intervention, it is necessary to know whether the intervention was in fact implemented, and how it was implemented. Therefore, aspects like fidelity (correspondence with the originally intended idea of the program), program reach and dosage of the original program delivered by the interventionists are crucial for understanding the effect of the intervention [39] and will be assessed in this study.

1.3.3 Effect

The randomized controlled effect evaluation is the golden-standard of evaluation designs in terms of determining the effect of an intervention across contexts [40]. The effects of the intervention will be assessed by comparing intervention and control units in a cluster-randomized controlled trial with a stepped wedged design.

The aim of this article is twofold: Firstly, we describe a multi-level, theory-based and tailored intervention program to prevent violence and threats in psychiatric hospitals and in the prison service. Secondly, we present an evaluation design that through a thorough assessment of the context, implementation process and subsequent outcomes will enable the examination of the mechanisms that are at work in the multi-level violence prevention intervention.

2 Methods

We will first describe the recruitment process for the study and the workplace context, then the intervention content and its relation to Dejoy’s theory on integrated safety management. Finally, we present the design and methods of the evaluation of the intervention.

2.1 Recruitment

Recruitment of work units took place in the spring and summer of 2017. In total, we recruited 16 work units, eight from psychiatric hospitals and eight from the Prison and Probation Service. The recruitment process is illustrated in Fig. 1. To recruit eight work units from psychiatry we contacted either the top management or the work environment coordinator directly in 12 work units in two Danish regions; eight rejected due to competing change processes or disinterest in the intervention program and four accepted to participate. In addition, we contacted the person responsible for the working environment of psychiatric units in another region. He sent an open call for participation to all 13 psychiatric work units within the region; of which four units accepted to participate. To recruit eight work units within the Prison and Probation Service we contacted the centralized violence prevention coordinators. The participation in the study was then approved by the top management of the Prison and Probation Service. After approval the top management encouraged all prisons (in total 52 prisons or detentions) in Denmark to participate and 8 work units spread over three regions decided to participate in the study.

Fig.1

Flowchart of recruitment process.

Inclusion criteria were that work units a) had more than 10 employees; b) did not share a unit-leader; c) were not ambulant departments (for psychiatric units); and d) currently did not use a “Violence Prevention Package”. The “Violence Prevention Package” is a publicly sponsored intervention that shares some characteristics with the Integrated Violence Prevention intervention (IVP). All work units that chose to participate lived up to the inclusion criteria.

2.2 Intervention content

The intervention addresses the prevention of violence situations where the perpetrator is a client or patient who becomes violent while being served by the organization (type 2 violence) [41]. The intervention does not focus on other types of violence [41], for example where the perpetrator has other kinds of relations to the victim but still conducts the violent act in the workplace.

The core elements of the intervention are presented in Fig. 2 and can be divided into four phases: 1) Preparation, 2) Mapping of violence prevention practices, 3) Problem-solving process, and 4) Assimilation.

Fig.2

The context, process and outcomes of Integrated Violence Prevention.

2.2.1 Phase 1: Preparation

A telephone or face to face meeting is held with the management to introduce the forthcoming intervention activities and ask them to inform their employees about the project. At this meeting the unit-leader is asked to point out members for the steering group, which should consist of the unit-leader, a health and safety representative or a representative of the union and about 2–3 regular employees.

2.2.2 Phase 2: Mapping of the existing violence prevention practices

To guide a tailoring of the intervention focusing on areas of prevention relevant for the given work unit, the existing violence prevention practices are mapped. To this end, we conduct an interview with the unit-leader, a focus group interview with employees, and a survey among all employees in the work unit. The mapping takes place 2–4 weeks before the intervention activities begin.

2.2.2.1. Interviews with unit-leaders. Interviews are conducted with the unit-leader with personnel responsibilities. Information is collected on what works well and what are the challenges in the existing violence prevention practice specifically in relation to: training in violence prevention, culture of acceptance of violence and threats, daily structures for communication and coordination of violence prevention, and collaboration on violence prevention.

2.2.2.2. Focus group interview with employees. Focus group interviews with 3–6 employees are conducted. To maximize variation in viewpoints, we aim to recruit focus group interview participants who differ in: participation of the work environment organization, educational background and seniority. During the focus group interview, we introduce the participants to two tasks. The first one is a critical incident analysis where the participants describe a situation with a patient or inmate that did not escalate, and what they, their colleagues and the unit-leader did to handle the situation. The second task is to map what participants think is influencing good violence prevention. They are asked to list all points on a whiteboard or poster and to determine for each of the listed aspect whether this is working well in their work unit or whether it is something that could be improved. This list is used directly in the mapping presentation for the entire unit (see Phase 3: Problem-solving process).

2.2.2.3. Survey. For the mapping phase we developed a survey that contains 26 questions (including 7 batteries of questions) that cover areas that have been found to be important in violence prevention: a) policies and procedures of violence prevention, b) shared definition of violence and threats, c) introduction of new employees to violence prevention procedures, d) training and courses on violence prevention, e) techniques/means used for violence prevention, f) engagement in violence prevention and g) collaboration on violence prevention.

2.2.3 Phase 3: Problem-solving process

Phase 3 is initiated by a seminar for the unit-leader and the top manager, followed by a seminar for all employees in the specific unit and their unit-leader. At the one-hour management seminar, the project team presents results from phase 2 (mapping) together with research results about management’s role in preventing violence and threats. Next, a three-hour seminar is held with the unit leader and as many employees as possible. During the first hour the results from phase 2 are presented; during the second hour the project team facilitates a process through which employees identify the most relevant areas of violence prevention; and during the third hour the employees work in groups to develop possible solutions and then prioritize which solutions they find most important. The resulting suggestions for prevention activities are handed over to the steering group (i.e. the unit-leader, the work environment representative or union representative and 2–3 regular employees) as inspiration for their further work.

Following the seminars members of the project team conduct 3 monthly coaching sessions with the unit-leader to support his/her own work with management-specific action plans on how to support the violence prevention work of the employees. The focus of the coaching is specifically but not exclusively on how to act supportive during the employee seminar and the steering group meetings. The first session is held in prolongation of the management seminar.

The project team also facilitates three monthly steering group meetings where the appointed members work with action plans in an iterative process trying out solutions and adjusting goals for the violence prevention effort. The action plans are qualified by using a dialogue tool developed by the project team: “the good suggested solution”. In using this tool, the participants are guided to reflect upon the consequences of the suggested solution. The consequences are considered for their colleagues working other shifts, for the management, for the patients/inmates, for their core task, from a practical point of view, and from a resource point of view in order to plan steps towards making it a good solution for all parties. Special emphasis is made on how to involve as many employees of the work unit as possible in the activity. The actions identified to qualify the suggested solution are finally noted in the action plans.

2.2.4 Phase 4: Assimilation

In the assimilation phase the intervention activities are no longer facilitated by the project team, and units have responsibility for conducting these activities themselves. The idea behind this phase is that the units thereby will start to adopt the intervention concept and make it part of their own approach to systematically work with violence prevention.

The assimilation phase lasts 3 months and during this period the steering group continues to have monthly meetings that are unassisted, i.e. the steering group participants facilitate the meetings themselves without assistance from the project team. In the 6th month a telephone meeting is offered to the unit-leader to evaluate the process, and talk about plans for future violence prevention work. Prior to this meeting a brief summary of the progress regarding reported violence prevention practices is developed by the project team and send to the unit-leader. The summary is based on the implementation degree (see Section 2.4.2), follow-up interviews (see Section 2.5.1), and the short questionnaire (see Section 2.4.1).

2.3 Design of the evaluation

The evaluation of the intervention consists of a coherent context, process and effect evaluation.

The effect evaluation is conducted in a multicenter stepped-wedged interventional design with start dates stratified and randomly assigned to work units. Results will be reported in accordance with the CONSORT-statement [42].

The context and process evaluation serves both a formative and summative purpose. For the formative purpose information from focus group interviews will be used to adjust and improve the implementation based on how context, implementation and intervention content fit together. For the summative purpose primarily qualitative data are gathered before, under, and after the intervention from each of the participating work units. The reporting of the context and process evaluation will follow the Consolidated criteria for reporting qualitative research (COREQ) [43]. In Table 1 we list all the hypotheses of the study.

Table 1
Hypotheses of the study

Hypotheses on effect and implementation

1. Main effect a) In Psychiatry ... ... The intervention will lead to a significant increase in the degree to which unit-leaders and employees continuously work on violence prevention practices based on their registrations and experiences.

b) In the Prison and Probation Service ...

2. Implementation and main outcome a) In Psychiatry ... ... Work units with a higher degree of implementation will have a greater increase in the degree to which unit-leaders and employees continuously work on violence prevention practices based on their registration and experiences than work units with a lower degree of implementation.

b) In the Prison and Probation Service ...

3. Implementation and specific outcomes a) In Psychiatry ... ... At work units with a high degree of implementation there will be a greater increase in a) cooperation between unit-leaders and employees; b) attention to violence prevention; c) the number of actions taken to prevent violence and threats; d) the violence prevention climate; e) the employees’ self-efficacy in violence prevention; and f) the employees’ sense of safety at work.

b) In the Prison and Probation Service ...

4. Ceiling effect Work units with a higher baseline score of outcome measures will experience smaller improvements post intervention due to a ceiling effect, even when the implementation degree is high.

General hypotheses on context and process

5. Contextual factors shape implementation The implementation of the intervention will be shaped by contextual factors such as changes in management, activities of the working environment organization and the organization’s readiness for change.

6. The role of auxiliary processes We will explore the role of the auxiliary processes (trust between management and employees, affective commitment, balanced attributions of responsibility of prevention work, and reciprocity between unit-leaders and employees in violence prevention efforts) on the implementation process (see also Fig. 1 and the explanation of auxiliary processes in the text).

Apart from the hypotheses in Table 1, we have several specific assumptions about the context and its relation to the implementation. These are represented as factors in Fig. 2 in the boxes of the discrete context and the omnibus context, and will be presented in paragraph 2.5.

Figure 2 illustrates how the context, process and effect evaluation are integrated in the Context, Process and Outcome Evaluation model [37]. In line with the authors of this model, we argue that integrating process and effect evaluation of complex organizational interventions is crucial to fully understand, interpret and use the results.

Figure 2 show that the intervention is considered part of a change process that starts already before the actual intervention activities and extends to after these activities. The auxiliary processes are theoretically based factors that are thought to facilitate this change process. The four auxiliary processes derive from the framework of Dejoy [31] and consists of a) Trust between management and employees, b) Affective commitment of employees to the work unit; c) The process of more balanced attributions of responsibility for preventing violence; and d) Increased reciprocity between management and employees. The auxiliary process c) and d) are similar, but in c) both parts acknowledge that there are aspects concerning violence prevention that the other part can influence and aspects that are out of their control, while d) emphasizes the psychological contract between management and employees, where both parts will do their best to optimize safety. Since not much is known about the working mechanisms of these factors, we will use this study to explore these mechanisms. This might help us to better understand how and if these auxiliary processes contribute to the change process.

2.3.1 The cluster-randomized controlled trial

We conduct a cluster-randomized controlled trial with a stepped wedged design to estimate the effect of the intervention. The 16 participating work units were randomized to four different starting dates (October 2017, January 2018, October 2018, and January 2019). At each starting date two psychiatric units and two units from the Prison and Probation Service enter the intervention. The work units in each sector were stratified by geography meaning that work units that are geographically close to each other will receive the intervention at the same time or shortly after each other, to diminish the spill-over effect in work units that are located in the same city. While not receiving the intervention the work units are considered controls as illustrated in Appendix 1.

2.3.2 Main and secondary outcomes

Both the main outcome (degree to which unit-leaders and employees continuously work on violence prevention practices based on their registration and experiences) and all secondary outcomes (represented in Fig. 2 in the outcome box) are measured by a three-monthly questionnaire.

We deliberately chose not to include self-reported frequencies of violence and threats as primary, but only as secondary outcomes. First, we do not expect it to be possible to find a statistically significant effect on frequencies of violence and threats during the follow-up period of the project because the incidence of violence and threats is multi-causal, and although the intervention is tailored to each work-unit we do not expect work-units to be able to address enough of the possible causes within the project period. Second, previous studies have suggested that violence prevention interventions can cause an increase in registration of violence and threats, because of the raised awareness an intervention brings [44, 45]. Third, official registers of violence and threats are not available and barriers for registering in the official systems are numerous and include for example a culture of acceptance of violence and threats as part of the job, cumbersome systems, perceptions of uselessness of registrations, fear of being known by the perpetrator in case of trial where the victim must witness etc. This makes incidence of violence and threats an unreliable outcome measure for the IVP intervention. We therefore chose a primary outcome measure that we believed would be possible to influence with the intervention activities within the project period, and that would have a relation to the incidence of violence and threats. There was no validated instrument available that suited this purpose. The Violence Prevention Climate Scale [46] was the only validated instrument that we evaluated to some extent lived up to these criteria as it has shown associations with levels of violence and threats and we therefore included it as a secondary measure [46 –48]. All other outcome measures are self-constructed or adjusted items.

2.4 Data for the effect evaluation

In the following all data-sources of the evaluation will be described one by one. For an overview of data-sources see Table 2.

Table 2
Overview of data-sources

Measurement tool Time for collection: Primarily used for effect, process or context evaluation

Throughout the study Only during the intervention period

Short questionnaire Every three months Effect evaluation

Implementation degree registration form After each intervention activity Effect and process evaluation

Field notes After each intervention activity Process and context evaluation

Registration of action plans In phase 3 and 4 problem solving and assimilation Effect, process and context evaluation

Before interviews with unit-leaders and with employees In phase 2 mapping Context evaluation

After interviews with unit-leaders and employees After phase 4 assimilation Process evaluation

Registration of context information Every three months Context evaluation

Measurement tool	Time for collection:	Primarily used for effect, process or context evaluation
Short questionnaire	Every three months		Effect evaluation
Implementation degree registration form		After each intervention activity	Effect and process evaluation
Field notes		After each intervention activity	Process and context evaluation
Registration of action plans		In phase 3 and 4 problem solving and assimilation	Effect, process and context evaluation
Before interviews with unit-leaders and with employees		In phase 2 mapping	Context evaluation
After interviews with unit-leaders and employees		After phase 4 assimilation	Process evaluation
Registration of context information	Every three months		Context evaluation

2.4.1 Short questionnaire

All main and secondary outcomes are measured by a 13-item web-based questionnaire send to all employees in the 16 participating work units every three months during the entire intervention period (see Appendix 2: Short questionnaire for the prison and probation service). This means that the first group of work units that enter the intervention will have measures every three months from 1 month prior to the intervention until 21 months after the intervention. The last group of work units that enter the intervention will have measures every three months from 16 months prior to the intervention until 6 months after the intervention.

2.4.2 Implementation registrations

Implementation degree is measured by registration of fidelity, reach and dose received after each intervention activity. Reach of the intervention is operationalized by documenting who (for example unit-leader or employee) and how many participants showed up as planned to the intervention activities. Fidelity to the intervention concept concerns to which degree the facilitator was loyal to the planned program and purpose of the intervention activities. It is operationalized by documenting to which degree the facilitator reached the specific sub-goals of each activity e.g. On the 3rd steering group meeting to which degree the facilitator gives the steering group sufficient tools for driving steering groups meetings by themselves in the assimilation phase. The dose of the intervention received by participants concerns the number of intervention activities held (for example steering group meetings) and the quality of participation in these activities (for example active participation in the discussions). Each activity is weighted in terms of its estimated importance for the overall effect of the intervention, we chose e.g. to weight the three coaching sessions and the three steering group meetings equally because we assume that both levels are of equal importance to the outcome. In the same way we weighted the aspects of reach, fidelity and dose received for each activity in terms of its estimated importance for the overall effect of the activity. An example of this weighing is that we generally chose to weigh the aspect of “dose received” a slightly higher since we assumed that things like the quality of the discussions and action plans would be crucial for the effect of the intervention. Each element and each aspect is scored by two members of the project group facilitating the intervention activities immediately after each intervention activity. When all activities have been completed a total score on a scale from 0–100% is calculated (See Appendix 3: Implementation degree registration forms).

A separate registration is made of reach and some aspects of fidelity in the assimilation phase where members of the project group are not present at the work units. Based on information from the contact person at the specific work unit the three unassisted steering group meetings are scored on a scale from 0–100% if the agenda and format of the project is used and a scale from 0–75% is used if meetings are held about violence preventive actions, but in another format. The mean of the score for the unassisted steering group meetings and the before mentioned registrations of the intervention activities are afterwards calculated as the final implementation degree.

2.4.3 Statistical analysis

We will perform parallel analysis on the two participating sectors, i.e. psychiatric hospitals and the prison service. Thus, all analyses described below will be performed on each of the two sectors separately.

The statistical analysis for testing the intervention’s effect on continuous adjustment of violence prevention practices (hypothesis 1a and 1b) is based on an intention-to-treat approach, using a linear mixed model. In this analysis we assume that the main outcome will be normally distributed. We will investigate this assumption graphically based on residual plots and if we detect ceiling or flooring tendencies, we will approach the analysis differently. Workplace, work unit, and employee will be entered into the model as random effects to account for correlation between observations. Due to the stepped-wedged design of the trial, we will enter the time dependent factors, calendar time and intervention-stage, as systematic effects.

In addition to the intention to treat analysis, we will perform an analysis that assesses whether the effect of the intervention is mediated by implementation degree (hypothesis 2). In this analysis, we will enter implementation degree as a time dependent variable into the model described above. Through further sensitivity analyses, we will assess whether pre-existing conditions (such as readiness for change or a high level of cooperation between management and employees) a) predict implementation degree and b) either fully or partly explain the effect of the intervention. Through this analysis we will be able to assess whether the implementation of the intervention or the pre-existing conditions at the work unit account for the effects.

Finally, we will perform secondary analyses to assess the effect of the intervention on secondary outcomes, i.e. cooperation between unit-leaders and employees, attention to violence prevention, the number of actions taken to prevent violence and threats, the violence prevention climate, the employees’ self-efficacy in violence prevention, and the employees’ sense of safety at work (hypothesis 3).

We know that three of the participating work units are planning to shift from three separate unit-leaders to one shared management during the intervention period. This means that in these work units the part of the intervention directed at management will be less likely to be delivered and received as intended. Moreover, we expect that the three work units with shared management will be more similar once they share management. We will therefore perform sensitivity analyses for the primary analysis. In this sensitivity analysis, we will enter change in management as a new time dependent categorical variable with three possible stages: no change in the organization, being informed about the change, and the actual change in management. Work units that do not experience change in management during the intervention period will be assigned the “no change in the organization” value of the variable. Based on experiences from other intervention projects [49] we know that changes will most likely also occur in other participating work units. In these cases we will likewise perform sensitivity analyses that address these changes.

2.4.4 Power calculations

We expect to recruit approximately 400 participants distributed among the 16 work units, 8 at psychiatric hospitals and 8 in the Prison Service. If we assume a dropout rate of 25%, we would have 300 employees in total, 150 employees in each trade, to include in the analysis. Traditional formulas for power calculations assume that all observations are independent. In this study the measurements of outcome are not independent as the employees are clustered in work units, which in turn are clustered in workplaces. We do not know, however, the size of the correlations at any of these levels (across time for the same employee, between employees in the same work unit, and between different units within the same workplace). Also, to perform a power calculation we need information on the residual variance of the main outcome. However, as the main outcome of the intervention is a newly developed measure, we do not have this information. We are therefore not able to report an actual power calculation for this study.

2.5 Context and process evaluation

For the context and process evaluation we use the model of Fridrich and colleagues as a guiding framework, see Fig. 2 [37]. We first placed the intervention activities in the model and identified possible and plausible outcomes. We then chose the contextual factors based on literature on general contextual factors for organizational interventions [39 , 50–54] and our specific assumptions about which factors would influence the implementation of the IVP intervention. Based on the identified factors we developed 10 guiding assumptions. We chose not to make assumptions on the relation between contextual factors and the outcomes, since we believe that many more factors than the ones we address with our intervention could potentially influence the outcomes. We will control for some of these contextual factors in the statistical analyses (e.g. turnover) but otherwise we consider the investigation of these mechanisms to be out of the scope of this study. We aim to investigate the mechanisms that we try to improve with the intervention as illustrated in Fig. 3.

Fig.3

Overview of types of assumptions made about the intervention.

This means that we will not look at the relation between e.g. change in management (contextual factor) and feeling safe at work (outcome) but only on e.g. change in management (contextual factor) and the extent to which the intervention was implemented (implementation). The assumptions about the influence of context on implementation are presented in Table 3.

Table 3

Specific context assumptions

General context
1. Sector	Implementation degree will generally be higher in the psychiatric units than in the Prison Service.
Context related to changes in personnel
2. Staff cuts^*	At work units where staff cuts occur before or under the intervention there will be a lower degree of implementation.
3. Staff turnover^*	At work units with high staff turnover there will be a lower degree of implementation.
4. Change in management^*	At work units where there is a change in management just before or under the intervention the implementation will be negatively or positively affected.
Intervention experiences
5. Potentially competing change processes	At work units where additional, potentially competing change processes occur before or during the intervention the degree of implementation will be negatively or positively affected.
6. Previous good experiences with similar interventions (in terms of process or content)^*	Work units that are familiar with conducting similar interventions will have higher levels of implementation.
Culture, attitudes and existing structure
7. Existing culture of participation	Work units with an existing culture with a high extent of employee participation will have higher levels of implementation.
8. Motivation for participation in the steering group	At work units where the motivation for participation in the steering group is high there will be a higher degree of implementation.
9. Resources^*	At work units with low levels of resources caused by e.g. a heavy inmate/patient group, intervention fatigue, competing change processes etc. there will be lower levels of implementation.
10. Existing structure of the working environment organization	At work units where the working environment organization has a high degree of meeting activity and a systematical approach to improving the working environment the degree of implementation will be higher.

^*Potentially explored statistically.

To explore the context specific assumptions, we will use five types of data: field notes, registration of action plans, focus group interviews, individual interviews, and registration of contextual factors through email with a contact person. The marked assumptions could potentially be quantified and explored in relation to the statistical analysis. The rest of the assumptions on context will be explored qualitatively in relation to the implementation degree.

2.5.1 Data-collection methods

Interviews with unit-leaders prior to intervention: Interviews with unit-leaders are conducted in each of the 16 participating work units. To interview the management is especially important in terms of assessing the pre-existing context of the intervention since they are responsible for shaping the existing structures of the work units, and play a pivotal role in the existing culture. We use a semi-structured interview guide that is designed to ensure data on all our context assumptions.

Focus group interviews with employees prior to intervention: Focus group interviews are conducted in each of the 16 participating work units. See Section 2.2 for information on selection of participants. We use a semi-structured interview guide that is designed to capture data on all our context assumptions.

Focus group interviews and individual interview with employees after the intervention: Two employee interviews will be conducted after the intervention. The first one will be conducted with 2–4 employees that did not take part in the steering group and aim to gain knowledge about the dissemination of the action plans initiated by the steering group and unit-leader. We will conduct the interview without the presence of a steering group member to get richer information on evaluation and assimilation of the intervention itself and to ensure that the interview persons will not be restricted in reporting their attitudes towards the intervention because of the presence of a colleague that participated in the steering group. The semi-structured interview guide will be informed by the actual action plans that the participants initiated during the intervention to be able to discuss the tangible actions and changes. The aim is to gain information on the experienced effect of the intervention in their work unit.

The second interview will be conducted with the working environment organization representative if the person has been part of the steering group, otherwise the interview will be conducted with an employee from the steering group. The aim of this interview is to get richer information on what actual action has been taken to prevent violence and threats and how the steering group meetings as a platform for these actions worked. The semi-structured interview guide will also here be informed by the actual action plans that the participants initiated during the intervention to be able to discuss the tangible actions and changes.

Interview with the unit-leader after the intervention: For this interview we use a semi-structured interview guide for a summative evaluation with a focus on the unit-leaders’ role in the intervention.

Registration of contextual information: The contact person of each work unit is contacted by e-mail or phone every three months throughout the 2-year project period. In this email or phone interview we ask for information on changes in management, staff cuts or other changes in the unit. If they report any of the abovementioned changes and have not yet received the intervention, the subject is followed up in the relevant upcoming interviews to get further details on the impact on the implementation process. All email correspondence conducted with the purpose of planning intervention activities that contains relevant information on the context assumptions will also be registered as data.

Field notes: Members of the project team facilitating the intervention activities write field notes after each completed intervention activity. The notes are structured around the context and process assumptions so that data on all relevant hypotheses are covered with each field note.

Action plans: All action plans from assisted and unassisted steering group meetings and from the coaching sessions are either photographed, scanned, or hand copied.

3 Discussion

This study aims to contribute to the violence prevention research in several ways. To the best of our knowledge this is the first study of a tailored, theory-based intervention integrating both employee-level and management-level actions in violence prevention. In the evaluation of this intervention, we apply a coherent approach of a context, process and effect evaluation, something that is strongly recommended by authors of systematic reviews of psychosocial work environment interventions and safety interventions [21, 55]. We expect that the evaluation of this intervention will bring about both generalizable knowledge of the effects as well as knowledge on the mechanisms that are at work in creating those effects. The thorough investigation of contextual factors will enhance the possibility of transferring experiences to other sectors in gaining knowledge on the prerequisites of successful implementation. The stepped-wedged design has the advantage that all participating work units receive the intervention at some point during the study which may enhance the motivation to participate. To use this design we measure the dependent variable 10 times throughout the study period, with a number of measurements before and after the intervention period, which improves the power in the statistical analyses [56].

The intervention is both participatory and open with regard to the types of changes to improve violence prevention. This approach makes it possible to tailor the intervention to the specific needs of the work unit. Both the participatory and tailored aspect have been recommended by other organizational health and safety intervention researchers [57]. This makes the intervention more complex and could lead to complexity bias, i.e. that the outcome might depend on several contingencies outside the control of the investigator [58]. As a recommended in a review on how to make organizational interventions work [59] we respond to this problem by including a thorough context and process evaluation. Through this comprehensive evaluation we aim to gain insight into these contingencies that might to some extent be outside the control of the investigator but still possible to describe and analyze in order to understand their influence on the implementation of the intervention.

The study faces several challenges. The stepped wedged design introduces a risk of contamination between clusters because of the long study period [56]. We address this issue by stratifying clusters by geography, meaning that work units that are geographically close to each other will receive the intervention at the same time or shortly after each other, to avoid that a work unit adapts parts of the intervention, before actually starting the intervention. The long study period also introduces a risk of learning curve bias [58] where the project team become more skilled during the study period and therefore the effect of the intervention will be better for the last cluster of work units. We intend to assess this bias by gathering extensive field notes from all intervention activities with detailed information on adaptions made for the specific activity.

Another critique of the stepped-wedged design is that the intervention is rolled out to all groups before the effectiveness is established [56]. Using a cluster RCT with sequential implementation would allow for estimation of the effect before the intervention is rolled out to the control group. As our intervention is preventive in nature and not a treatment, we believe that it is ethically sound to introduce the intervention to all participants before estimation of the effect. The very open nature of the intervention also allows the work units to focus on meaningful aspect that they want to work with. This opens up for the possibility to tailor the intervention to work units’ current needs, which heightens the probability that work units will experience positive side-effects, even in the absence of effects on the pre-defined outcomes.

Another challenge is that we use a method to calculate implementation degree that has not been tested before. We do so out of the conviction that it is crucial to include the implementation degree in the effect evaluation, and we found no satisfying existing way of calculating implementation degree.

Lastly, we have chosen a primary outcome measured with one self-constructed item. We did so for several reasons. Firstly, it is difficult to use long questionnaires in a stepped-wedged design, since the analysis is dependent upon high response rates throughout the entire project period (in this case every three months throughout the 2-year project period). With long questionnaires we would risk a very low response rate and statistical analysis would be impossible. Secondly, we looked for a primary outcome measure that was close enough to the intervention to realistically be affected by the planned intervention activities and at the same time could be a plausible indicator of future prevalence of violence and threats.

4 Concluding remarks

If the intervention proves successful it could be a step in the direction of a more reflective practice of violence prevention. As a general model for violence prevention the IVP intervention has the potential to be used in other trades such as the service trade, teaching, or other healthcare areas than psychiatry. The intervention addresses the management level, and information on the role of managers in violence prevention could in itself add to other existing violence prevention models such as e.g. Safewards [60]. The results of the study will be generalizable to other Scandinavian countries that have similar participative work cultures, work environment structures and approaches to psychiatric patients and prisoners. The first results are expected to be published in 2020.

4.1 Ethical considerations

We will collect data by use of questionnaires, focus groups, interviews, action plans, implementation documentation and background documents (for example organizational structure) from the participating workplaces. As no biological data is collected, ethics approval by the Danish National Committee on Biomedical Research Ethics is not requested nor required according to the Danish regulations. Before the initiation of data collection the project has been approved by The National Research Center for the Working Environment that has been authorized by the Danish Data Protection Agency to assess their internal projects (Journal no. 2015-57-0074). The facilitators of the intervention will have access to both external and internal supervision on the facilitation process.

Conflict of interest

None to report.

Footnotes

Appendix 1: Stepped-wedged design

Cluster	September 2017	October–December 2017	January–March 2018	April–June 2018	July–September 2018	October–December 2018	January–March 2019	April–June 2019	July–September 2019	October–December 2019
1	Mapping/Baseline	Interven-tion	Interven-tion	Follow-up	Follow-up	Follow-up	Follow-up	Follow-up	Follow-up	Follow-up
2	Baseline	Mapping/Control	Interven-tion	Interven-tion	Follow-up	Follow-up	Follow-up	Follow-up	Follow-up	Follow-up
3	Baseline	Control	Control	Control	Mapping/Control	Interven-tion	Interven-tion	Follow-up	Follow-up	Follow-up
4	Baseline	Control	Control	Control	Control	Mapping/Control	Interven-tion	Interven-tion	Follow-up	Follow-up

Appendix 2: Short questionnaire for the prison and probation service

The following questionnaire is tailored to the prison and probation service. Another equally tailored exists for psychiatry and can be acquired upon request.

Appendix 3: Implementation degree registration forms

The following registration forms are summary schemes. Separate detailed registration forms exist for each of the activities: the mapping survey, the management seminar, the employee seminar, for each of the three management coaching sessions, for each of the three assisted steering group meetings, and for each of the three unassisted steering group meetings. The registration forms for each activity are available upon request.

Based on information from the contact person at the specific work unit the three unassisted steering group meetings are scored on a scale from 0–100% if the agenda and format of the project is used and a scale from 0–75% is used if meetings are held about violence preventive actions, but in another format.

The mean of the score for the unassisted steering group meetings and the assisted intervention activities is calculated as the final implementation degree.

Acknowledgements

The study is funded by The Danish Working Environment Research Fund (Grant number: AMFF-24-2016-03-2016510163). Trial registration: ISRCTN86993466. We would like to thank our Intern Signe Marie Lundsted Hurup for contributions to the development of the intervention tools.

References

Piquero

, Piquero

, Craig

, Clipper

. Assessing research on workplace violence, 2000-2012. Agression and Violent Behavior. 2013;18:383–94.

Spector

, Zhou

, Che

. Nurse exposure to physical and nonphysical violence, bullying, and sexual harassment: A quantitative review. International Journal of Nursing Studies. 2014;51(1):72–84.

Konda

, Tiesman

, Reichard

, Hartley

. U.S. Correctional Officers Killed or Injured on the Job. Correct Today. 2013;75(5):122–3.

Rasmussen

, Hogh

, Andersen

. Threats and Physical Violence in the Workplace: A Comparative Study of Four Areas of Human Service Work. J Interpers Violence. 2013;28(13):2749–69.

Lanctôt

, Guay

. The aftermath of workplace violence among healthcare workers: A systematic literature review of the consequences. Aggression and Violent Behavior. 2014;19(5):492–501.

Barling

, Rogers

, Kelloway

. Behind Closed Doors: In-Home Workers’ Experience of Sexual Harassment and Workplace Violence. Journal of Occupational Health Psychology. 2001;6(3):255–69.

Rogers

, Kelloway

. Violence at work: Personal and organizational outcomes. Journal of Occupational Health Psychology. 1997;2:63–71.

Richter

, Berger

. Post-traumatic stress disorder following patient assaults among staff members of mental health hospitals: A prospective longitudinal study. BMC Psychiatry. 2006;6:15.

Wykes

, Whittington

. Prevalence and predictors of early traumatic stress reactions in assaulted psychiatric nurses. The Journal of Forensic Psychiatry. 1998;9(3):643–58.

10.

Dillon

. Workplace violence: Impact, causes, and prevention. Work: Journal of Prevention, Assessment & Rehabilitation. 2012;42(1):15–20.

11.

Vaez

, Josephson

, Vingård

, Voss

. Work-related violence and its association with self-rated general health among public sector employees in Sweden. Work. 2014;49(1):163–71. Ref Type: Abstract

12.

Hogh

, Viitasara

. A systematic review of longitudinal studies of nonfatal workplace violence. European Journal of Work and Organizational Psychology. 2005;14(3):291–313.

13.

Madsen

IEH

, Burr

, Diderichsen

, Pejtersen

, Borritz

, Bjorner

, et al. Work-related Violence and Incident Use of Psychotropics. American Journal of Epidemiology. 2011;174(12):1354–62.

14.

Erkol

, Gökdoe∂an

, Erkol

, Boz

. Agression and violence towards health care providers – A problem in Turkey? Journal of Forensic and Legal Medicine. 2007;14:423–8.

15.

Langsrud

, Linaker

, Morken

. Staff injuries after patient-staff incidences in psychiatric acute wards. Nordic Journal of Psychiatry. 2007;61(2):121–5.

16.

Leather

, Lawrence

, Beale

, Cox

, Dickson

. Exposure to occupational violence and the buffering effects of intra-organizational support. Work & Stress. 1998;12(2):161–78.

17.

Clausen

, Hogh

, Borg

. Acts of offensive behaviour and risk of long-term sickness absence in the Danish elder-care services: a prospective analysis of register-based outcomes. International Archives of Occupational and Environmental Health. 2012;85(4):381–7.

18.

Rugulies

, Christensen

, Borritz

, Villadsen

, Bültmann

, Kristensen

. The contribution of the psychosocial work environment to sickness absence in human service workers: Results of a 3-year follow-up study. Work & Stress. 2007;21(4):293–311.

19.

Svedberg

, Alexanderson

. Associations between sickness absence and harassment, threats, violence, or discrimination: A cross-sectional study of the Swedish Police. Work. 2012;42(1):83–92.

20.

Anderson

, FitzGerald

, Luck

. An integrative literature review of interventions to reduce violence against emergency department nurses. Journal of Clinical Nursing. 2010;19(17-18):2520–30.

21.

Runyan

, Zakocs

, Zwerling

. Administrative and behavioral interventions for workplace violence prevention. American Journal of Preventive Medicine. 2000;18(4):116–27.

22.

Wassell

. Workplace violence intervention effectiveness: A systematic literature review. Safety Science. 2009;47(8):1049–55.

23.

Price

, Baker

, Bee

, Lovell

. Learning and performance outcomes of mental health staff training in de-escalation techniques for the management of violence and aggression. British Journal of Psychiatry. 2015;206(6):447–55.

24.

Gadegaard

, Andersen

, Hogh

. Effects of Violence Prevention Behavior on Exposure to Workplace Violence and Threats: A Follow-Up Study. J Interpers Violence. 2015;33(7):1096–117.

25.

Andersen

, Gadegaard

, Hogh

. Risikofaktorer for vold og trusler pa arbejdspladsen. Arbejdsmedicinsk Klinik Herning, Universitetsklinik, Institut for Psykologi, Kobenhavns Universitet.; 2015.

26.

Andersen

, Hogh

, Biering

, Gadegaard

. Work-related threats and violence in human service sectors: The importance of the psycho-social work environment examined in a multilevel prospective study. Work. 2018;59(1):141–54.

27.

Brooks

, Staniford

, Dollard

, Wiseman

. Risk Factors, Consequences, and Management of Aggression in Healthcare Environments. Contemporary Occupational Health Psychology. Wiley-Blackwell; 2010. pp. 329–53.

28.

Gillespie

, Gates

, Fisher

. Individual, relationship, workplace, and societal recommendations for addressing healthcare workplace violence. Work. 2015;51(1):67–71.

29.

Bowers

, James

, Quirk

, Simpson

, SUGAR (Service User and carer Group for Research led by Professor Simpson hosted by City University London), Stewart D, et al. Reducing conflict and containment rates on acute psychiatric wards: The Safewards cluster randomised controlled trial. International Journal of Nursing Studies. 2015;52:1412–22.

30.

Arnetz

, Hamblin

, Russell

, Upfal

, Luborsky

, Janisse

, et al. Preventing patient-to-worker violence in hospitals: outcome of a randomized controlled intervention. J Occup Environ Med. 2017;59(1):18–27.

31.

DeJoy

. Behavior change versus culture change: Divergent approaches to managing workplace safety. Safety Science. 2005;43(2):105–29.

32.

Kines

, Andersen

, Nielsen

, Pedersen

. Improving safety in small enterprises through an integrated safety management intervention. Journal of Safety Research. 2013;44:87–95.

33.

Nielsen

, Pedersen

, Andersen

, Kines

, Andersen

, Rasmussen

, et al. Integreret ulykkesforebyggelse på tværs af brancher og virksomhedsstørrelser. Arbejdsmedicinsk klinik, Regionshospitalet Herning & Det Nationale Forskningscenter for Arbejdsmiljø; 2012.

34.

Zohar

. A group-level model of safety climate: Testing the effect of group climate on microaccidents in manufacturing jobs. Journal of Applied Psychology. 2000;85(4):587–96.

35.

Biron

, Karanika-Murray

. Process evaluation for organizational stress and well-being interventions: Implications for theory, method, and practice. International Journal of Stress Management. 2014;21(1):85–111.

36.

Nielsen

, Randall

. Opening the black box: Presenting a model for evaluating organizational-level interventions. European Journal of Work and Organizational Psychology. 2013;22(5):601–17.

37.

Fridrich

, Jenny

, Bauer

. The context, process, and outcome evaluation model for organisational health interventions. BioMed research international. 2015;2015.

38.

Stame

. What Doesnt Work? Three Failures, Many Answers. Evaluation. 2010;16(4):371–87.

39.

Durlak

, DuPre

. Implementation matters: A review of research on the influence of implementation on program outcomes and the factors affecting implementation. American Journal of Community Psychology. 2008;41(3-4):327–50.

40.

GRADE Working Group. Grading quality of evidence and strength of recommendations. BMJ. 2004;328(7454):1490.

41.

Loveless

. Workplace Violence: A Report to the Nation. University of Iowa Injury Prevention Research Centre; 2001.

42.

Schulz

, Altman

, Moher

. CONSORT 2010 Statement: Updated guidelines for reporting parallel group randomised trials. BMJ. 2010;340.

43.

Tong

, Sainsbury

, Craig

. Consolidated criteria for reporting qualitative research (COREQ): A 32-item checklist for interviews and focus groups. International Journal for Quality in Health Care. 2007;19(6):349–57.

44.

Arnetz

, Arnetz

. Implementation and evaluation of a practical intervention programme for dealing with violence towards health care workers. Journal of Advanced Nursing. 2000;31(3):668–80.

45.

Whittington

, Wykes

. An evaluation of staff training in psychological techniques for the management of patient aggression. Journal of Clinical Nursing. 1996;5(4):257–61.

46.

Kessler

, Spector

, Chang

, Parr

. Organizational violence and aggression: Development of the three-factor Violence Climate Survey. Work & Stress. 2008;22(2):108–24.

47.

Spector

, Yang

, Zhou

. A longitudinal investigation of the role of violence prevention climate in exposure to workplace physical violence and verbal abuse. Work & Stress. 2015;29(4):325–40.

48.

Spector

, Coulter

, Stockwell

, Matz

. Perceived violence climate: A new construct and its relationship to workplace physical violence and verbal aggression, and their potential consequences. Work & Stress. 2007;21(2):117–30.

49.

Olsen

, Albertsen

, Nielsen

, Poulsen

KBr

, Gron

SMF

, Brunnberg

. Workplace restructurings in intervention studies - a challenge for design, analysis and interpretation. BMC Medical Research Methodology. 2008;8(1):39.

50.

Nielsen

, Abildgaard

. Organizational interventions: A research-based framework for the evaluation of both process and effects. Work & Stress. 2013;27(3):278–97.

51.

Damschroder

, Aron

, Keith

, Kirsh

, Alexander

, Lowery

. Fostering implementation of health services research findings into practice: A consolidated framework for advancing implementation science. Implementation Science. 2009;4(1):50.

52.

Wierenga

, Engbers

, Van Empelen

, Duijts

, Hildebrandt

, Van Mechelen

. What is actually measured in process evaluations for worksite health promotion programs: A systematic review. BMC Public Health. 2013;13(1):1–16.

53.

Havermans

, Schlevis

, Boot

CCR

, Brouwers

, Anema

, van der Beek

. Process variables in organizational stress management intervention evaluation research: A systematic review. Scandinavian Journal of Work, Environment & Health. 2016;42(5):371–81.

54.

Randall

, Cox

, Griffiths

. Participants’ accounts of a stress management intervention. Human Relations. 2007;60(8):1181–209.

55.

Egan

, Bambra

, Petticrew

, Whitehead

. Reviewing evidence on complex social interventions: Appraising implementation in systematic reviews of the health effects of organisational-level workplace interventions. Journal of Epidemiology and Community Health. 2009;63(1):4–11.

56.

Kotz

, Spigt

, Arts

, Crutzen

, Viechtbauer

. Use of the stepped wedge design cannot be recommended: A critical appraisal and comparison with the classic cluster randomized controlled trial design. Journal of Clinical Epidemiology. 2012;65(12):1249–52.

57.

LaMontagne

, Noblet

, Landbergis

. Intervention development and implementation: Understanding and addressing barriers to organizational-level interventions. In: Biron

, Karanika-Murray

, Cooper

, editors. Improving Organizational Interventions for Stress and Well-Being: Addressing Process and Context. 1st ed. New York: Routledge; 2012. pp. 21–38.

58.

Jadad

, Enkin

. Bias in Randomized Controlled Trials. In: Jadad

, Enkin

, editors. Randomized Controlled Trials: Questions, Answers, and Musings. Second Edition ed. Wiley Online Library; 2008. pp. 29–47.

59.

Nielsen

. Review Article: How can we make organizational interventions work? Employees and line managers as actively crafting interventions. Human Relations. 2013;66(8):1029–50.

60.

Bowers

. Safewards: A new model of conflict and containment on psychiatric wards. Journal of Psychiatric and Mental Health Nursing. 2014;21(6):499–508.

Hypotheses on effect and implementation
1. Main effect	a) In Psychiatry ...	... The intervention will lead to a significant increase in the degree to which unit-leaders and employees continuously work on violence prevention practices based on their registrations and experiences.
	b) In the Prison and Probation Service ...
2. Implementation and main outcome	a) In Psychiatry ...	... Work units with a higher degree of implementation will have a greater increase in the degree to which unit-leaders and employees continuously work on violence prevention practices based on their registration and experiences than work units with a lower degree of implementation.
	b) In the Prison and Probation Service ...
3. Implementation and specific outcomes	a) In Psychiatry ...	... At work units with a high degree of implementation there will be a greater increase in a) cooperation between unit-leaders and employees; b) attention to violence prevention; c) the number of actions taken to prevent violence and threats; d) the violence prevention climate; e) the employees’ self-efficacy in violence prevention; and f) the employees’ sense of safety at work.
	b) In the Prison and Probation Service ...
4. Ceiling effect	Work units with a higher baseline score of outcome measures will experience smaller improvements post intervention due to a ceiling effect, even when the implementation degree is high.
General hypotheses on context and process
5. Contextual factors shape implementation	The implementation of the intervention will be shaped by contextual factors such as changes in management, activities of the working environment organization and the organization’s readiness for change.
6. The role of auxiliary processes	We will explore the role of the auxiliary processes (trust between management and employees, affective commitment, balanced attributions of responsibility of prevention work, and reciprocity between unit-leaders and employees in violence prevention efforts) on the implementation process (see also Fig. 1 and the explanation of auxiliary processes in the text).