Executive Summary

In adult and youth acute mental health care settings, aggression and violence are workplace hazards that are known to cause harm. These hazards are particularly prevalent in acute inpatient environments and arise within complex systems where clinical, environmental, organisational, and broader systemic factors interact. Understanding aggression and violence as outcomes that can arise from these interacting conditions informs the identification and implementation of proactive approaches to manage risk and prevent exposure. Systems thinking provides a robust framework for examining these complex safety issues, by identifying underlying contributors, understanding their interrelationships, and informing the development of systemic and sustainable strategies for reducing harm.

As part of their ongoing efforts to reduce the impact of Aggression and Violence, WorkSafe Victoria funded Opposite to undertake an exploratory research project applying systems thinking to the issue of Aggression and Violence in adult and youth acute mental health inpatient settings in Victoria. The aims were to apply systems thinking methods to enhance understanding of Aggression and Violence in these settings and to develop a series of recommendations and novel interventions designed to support enhanced prevention and management of the issues. The research involved six phases, including:

1. Literature Review. A review of relevant literature was conducted to identify and systematically organise the numerous factors previously found to contribute to Aggression and Violence in like settings.
2. Engagement in Project Design. This phase involved collaborating with WorkSafe Victoria and Safer Care Victoria to define priorities, refine the scope, and co-develop processes, ensuring relevance, feasibility, and shared ownership.
3. Actor Engagement. This phase gathered data through online surveys, one-on-one semi-structured interviews, and small group workshops to identify contributing factors and preventative measures for Aggression and Violence.
4. Systems Map Development. An AcciMap was created to show the risk factors which interact to create Aggression and Violence in Victorian adult and youth acute mental health inpatient settings. The PreventiMap outlined networks of potential interventions designed to optimise the prevention and management of Aggression and Violence in these settings.
5. Additional Groups (Lived and Living Experience (LLE) and Forensicare). To ensure the AcciMap and PreventiMap were comprehensive and did not omit perspectives of the Lived and Living Experience (LLE) (both carers and peer workers) and participants in Forensic acute mental health inpatient settings, additional data collection and analysis were conducted concurrently with validation sessions (Phase 6).
6. Systems Map Validation. A mixed-methods validation process was conducted to assess the accuracy, comprehensiveness, and relevance of two previously developed systems maps.

This final report presents the findings and recommendations.

Contents

1. Background
2. Project Summary
3. Literature Review
4. Engagement in Project Design
5. Actor Engagement
6. Systems Map Development
7. Additional Groups (Lived and Living Experience (LLE) and Forensicare)
8. Systems Map Validation
9. Results
10. Future Improvement Areas for Consideration
11. References

Aggression and Violence

It is clear from international and local research, as well as health service data, that Aggression and Violence is a significant, complex, and challenging problem that is especially prevalent within adult and youth acute mental health inpatient settings (Bekelepi & Martin, 2022; Jenkin et al., 2022; Odes et al., 2021; Whiting, Lichtenstein, & Fazel, 2021). The term Aggression and Violence refers to 'incidents in which a person is abused, threatened or assaulted in circumstances relating to their work' (WorkSafe Victoria, 2026). Such incidents can undermine the provision of a safe and therapeutic environment by placing consumers and mental health workers at risk of physical or psychological harm (Bowers et al., 2003).

Prevalence estimates across multiple studies indicate that 65% to 99% of workers have experienced general aggression in these settings, and rates of physical aggression range from 38% to 82% (Weltens et al., 2021). In a survey of Victorian mental health workers in Australia, 83% reported exposure to at least one form of violence in the previous 12 months (Tonso et al., 2016).

Impacts of Aggression and Violence Incidents

Incidents of Aggression and Violence in acute mental health inpatient settings can cause a range of harms including physical, psychological, social, and economic, which may be short and/or long-term (Jenkin et al., 2022; WorkSafe Victoria, 2025). The impacts of Aggression and Violence do not only affect the physical and mental health and wellbeing of the individual/s involved both directly and indirectly in the incident; Aggression and Violence can also have negative implications for quality of care, productivity, and intentions to remain in the workforce (Jang, Son & Lee, 2022; Jenkin et al., 2022; Søvold et al., 2021; Zhao et al., 2018). Tonso and colleagues (2016) found that amongst a survey of Victorian mental health workers who reported experiencing some form of violence in the past year, 33% rated themselves as being in psychological distress, and of this group over half (54%) reported being in severe psychological distress.

Aggression and Violence Management Approaches

Under the Mental Health and Wellbeing Act 2022, mental health and wellbeing service providers and decision-makers have a legislated obligation to reduce, and ultimately eliminate, the use of restrictive interventions — seclusion, bodily restraint and chemical restraint — within ten years. Restrictive interventions may only be authorised as a last resort, when all reasonable less restrictive options have been tried or considered and found unsuitable, and only where necessary to prevent serious and imminent harm to the person or another person. Restrictive interventions are recognised as contributing to escalation and trauma rather than reducing aggression (Royal Commission into Victoria's Mental Health System, 2021; National Mental Health Commission, 2020), reinforcing the need for approaches that address the system conditions in which Aggression and Violence emerges.

Alongside these regulatory imperatives, attention has shifted toward approaches that treat Aggression and Violence as a complex, multi-factorial phenomenon rather than a problem located in any single individual or cohort. The Systemic Model of Aggression and Violence in Mental Health Care (Cutcliffe & Riahi, 2013) is one such framework, integrating environmental-related, system-related, clinician-related and consumer-related contributing factors. Understanding the complex network of contributory factors is critical to intervening effectively, and is the foundation for the systems-thinking methodology applied in this project.

Applied Systems-Thinking

Given the range of contributing factors and the increased focus on causes beyond the individual, approaches that can take complexity into account when trying to understand the drivers of Aggression and Violence can be of particular use in furthering understanding. One such approach is systems thinking, which can be used to identify the factors underpinning complex safety-related issues and reveal how components of the system interact to create the issue (Meadows, 2008). This type of analysis also supports the identification of 'leverage points' in the system where targeted interventions can flow on to multiple points in the complex system to bring about improvements.

Systems thinking has mostly been applied in high-risk industries such as aviation, maritime, rail, public health, and mining (Hulme et al., 2019), with recent applications extending to aggression and violence (Salmon et al., 2022). For example, one recent WorkSafe study in Australia utilised Rasmussen's framework (Rasmussen, 1997) for understanding manual handling and psychological injuries in Victorian public hospitals (Aburumman & Morton, 2023).

Systems-Thinking in Acute Mental Health Inpatient Settings

Applying systems thinking to Aggression and Violence in mental health settings can be especially helpful because it allows for a holistic understanding of the complex interrelated factors such as staffing levels, consumer acuity, environmental design, and organisational policies that contribute to incidents, enabling the development of more targeted, sustainable interventions to improve both staff and consumer safety. To our knowledge, systems thinking has not been applied to Aggression and Violence in adult and youth acute mental health inpatient settings globally or nationally.

Recent WorkSafe initiatives are underway for the systemic understanding of Aggression and Violence for adult and youth acute inpatient settings in Victoria, aligned with the Mental Health Workforce Safety and Wellbeing Committee's 12-month action plan to better address psychological safety in the mental health sector.

WorkSafe Victoria engaged Opposite to undertake an exploratory research project applying systems thinking to the issue of Aggression and Violence in adult and youth acute inpatient settings in Victoria. The project goals were to:

• Apply systems thinking methods to enhance understanding of Aggression and Violence in adult and youth acute mental health inpatient settings in Victoria.
• Identify targeted intervention opportunities across system levels to support improved prevention and management of Aggression and Violence.

The outcomes are intended to help stakeholders better identify the contributing factors behind Aggression and Violence in these environments and to guide future decisions around implementing effective strategies to address the issue and reduce the frequency and associated impacts of Aggression and Violence.

Project Phases

Below is a brief overview of the methods used in this project. Further details about the specific approaches applied are provided throughout this report, as they arise in the relevant phases.

Project Scope

The scope of the project was determined in consultation and agreement with WorkSafe Victoria and Safer Care Victoria.

Introduction

This literature review was conducted to synthesise existing research on the factors contributing to Aggression and Violence in comparable settings. Applying Rasmussen's systems thinking approach to a risk management framework helped clarify the complexity of the issue and provided background and context for the project. The literature review was also one of the resources used to identify initial relationships between contributing factors.

Summary of Methods

Our approach involved compiling evidence from multiple existing reviews, following a method similar to that used in a previous SafeWork NSW study on Aggression and Violence in healthcare settings through a systems analysis. This approach enables a rapid synthesis of evidence and is well-suited for addressing broad research questions.

We mapped identified contributory factors onto systems framework categories proposed by Rasmussen (1997) and adapted by WorkSafe in a previous study for Aggression and Violence in Victorian public hospitals. The six categories were:

1. Government policy and budgeting
2. Regulatory bodies and external influences
3. Organisational governance and administration
4. Operational management
5. Frontline
6. Environment and equipment

Summary of Results

The Literature Review identified 71 factors contributing to Aggression and Violence in acute inpatient mental health settings. Below is a summary of key findings for each system level:

Engagement with stakeholders was integral to shaping the design, scope, and implementation of this project. WorkSafe Victoria and Safer Care Victoria provided valuable input throughout the study.

At the outset, we consulted with both stakeholders to identify areas of priority to be included within the project scope. Stakeholders were asked to review and refine the initial scope and aims so that the project reflected their priorities and was feasible within their operational contexts. This collaborative approach enabled the research team to incorporate diverse perspectives and ensure relevance. The specific aims of the project were co-developed by the research team in collaboration with both stakeholder groups, ensuring shared ownership of the process.

Additional engagement activities included email correspondence, regular meetings, and leadership consultations to establish a shared understanding of the project objectives and design elements. During these sessions, we outlined the rationale and benefits of the project, agreed on the careful development of processes and tools, established roles and responsibilities, and discussed key risks alongside mitigation strategies.

This initial engagement phase was critical to the project's success. It fostered transparency, trust, and consensus across all parties. We gratefully acknowledge and thank members of WorkSafe Victoria and Safer Care Victoria for their significant contributions, which informed the overall vision, scope, and intended outcomes of the study.

Introduction

The goal for Phase 3 of the research was to identify and document the factors that contribute to Aggression and Violence in adult and youth acute mental health inpatient settings, as well as to identify opportunities for preventative measures to reduce such incidents from occurring. This section presents an overview of the research methodology, and a summary of the key findings.

Methods — Procedure

To investigate contributing factors and preventative measures, a multi-method approach was employed. Data was systematically collected through a series of online surveys, one-on-one online semi-structured interviews, and small group online workshops.

Recruitment

Recruitment focused on stakeholders who currently share responsibility for managing Aggression and Violence in adult and youth acute inpatient settings. WorkSafe Victoria, Safer Care Victoria and Opposite identified relevant subject matter experts through their industry networks and invited them to participate via email. A snowball sampling method was also employed.

• Surveys. A total of 64 participants took part in the survey. Of these, 33 responses were incomplete, and only 2 of the incomplete responses contained analysable data. Consequently, data from 33 participants were included in the analysis.
• Interviews. 26 participants took part in interviews.
• Workshops. A total of 47 people took part across eight workshops.

Phase 3 Participant Summary

Workshops accounted for the largest share of data collection (44%), followed by surveys (31%), with interviews contributing the smallest portion (25%). Survey participants were predominantly from adult inpatient settings (79%), with smaller proportions working across youth settings (13%) and combined settings (8%). Interview participants were mostly working with adults (59%), with some working across both youth and adult populations (35%), and one from the youth-only sector.

In terms of roles, OHS Professionals made up the largest group (28%), followed by Frontline Staff (25%), with Directors and Senior Managers/Associate Directors each representing 10%. Among survey participants, frontline staff made up the largest group (52%), whereas OHS professionals were most common among interview participants (62%).

The broader cohort representation reflected these roles, with Clinical Staff comprising the largest group (30%), followed by Senior Managers (26%), and WorkSafe Representatives (15%). Among survey participants, Clinical Staff made up the largest group (64%), whilst WorkSafe representatives were most frequent among interview participants (54%). Workshop participants included a large representation of Senior Managers (45%) as well as Union/Other Representatives (29%).

Methods — Data Analysis

Data analysis followed a three-stage process:

1. Data preparation, involving extraction, selection, and de-identification of survey, interview, and workshop data.
2. Generating initial codes for each data source to identify contributing factors and preventative measures.
3. Organising codes into themes through an iterative thematic analysis.

Data Analysis — Stage 1 (Data Preparation)

The data preparation stage involved a multi-step process to extract, select, and de-identify data from three primary sources: online surveys, one-on-one online semi-structured interviews, and small group online workshops.

• Data Extraction and Demographic Categorisation. Raw data was first extracted from all survey responses, interview transcripts, and workshop outputs. Demographic and contextual data points were categorised according to specific rules to ensure consistency across the diverse data sources.
• Data Selection and Consolidation. Only those data points most relevant to the specific project research were selected from each source.
• Ethical Considerations and De-identification. Prior to analysis, all data underwent a thorough de-identification process to protect participant anonymity and confidentiality.

Data Analysis — Stage 2 (Generating Initial Codes for Each Data Source)

The primary objective of this analysis was to systematically identify, categorise, and analyse all stated contributing factors and preventative measures related to Aggression and Violence. Consistent with Rasmussen's systemic framework, the analysis sought to find factors at each level of the system. The final levels were: (1) Environment, (2) Consumers & Their Carers, (3) Workforce (Staff), (4) Operational Management, (5) Organisation / Company, and (6) Government, Regulatory & External Bodies.

Key analytical steps included:

• Analytical Framework and Process. A structured, AI-assisted methodology was employed using Google's Gemini Pro, guided by precise instructional user-created prompts. All analysis and interpretation were conducted by a human researcher.
• Foundational Prompt Development. A detailed analytical prompt was established prior to the analysis, developed and refined through an iterative testing process using dummy data.
• Iterative Batch Processing. Data for each resource was processed iteratively in small batches of participants.
• Data Deconstruction and Thematic Coding. Each qualitative response was broken down into individual, distinct ideas. Each idea was classified as either a "Contributing Factor" or a "Preventative Measure/Control", categorised into one of the six predefined Main Categories, and assigned a Sub-Category.
• Consolidation, Verification and Synthesis. Following the AI-driven coding, a manual human review was conducted.
• Application Across Multiple Data Sources. This core methodology was consistently applied to analyse qualitative data from each of the data sources.

Data Analysis — Stage 3 (Organising Codes into Themes)

This study employed a systematic, iterative thematic analysis approach to synthesise qualitative data gathered from multiple sources. The methodology was applied independently to two distinct datasets: one examining the contributing factors to Aggression and Violence, and a second analysing proposed preventative measures and controls.

The methodology consisted of two distinct phases: Framework Development and Final Application.

• Iterative Framework Development. The core of each analysis was the development of a unique thematic coding framework, referred to as 'Second Level Subcategories'.
• Pilot Analysis. An initial dataset was subjected to a preliminary thematic analysis, generating an initial set of broad thematic categories forming a pilot codebook (Version 1.0).
• Iterative Refinement. Subsequent datasets were analysed sequentially, with the existing codebook used as a foundation. The process was repeated until a definitive Final Project Codebook was established.
• Application and Data Dictionary. Once each thematic framework was finalised, the original raw data files were systematically re-analysed using the Final Project Codebook.

Methods

A draft AcciMap and PreventiMap were created by the research team using the results of the initial analysis. These maps were developed on an online, digital tool called a 'MiroBoard'.

Results

The AcciMap and PreventiMap analyses were validated through future workshops and stakeholder engagement (see Phase 6). Initial analysis also highlighted cohort gaps, such as the perspectives of individuals with LLE and Forensicare personnel, which was addressed in subsequent data collection (see Phase 5).

AcciMap

The AcciMap for Aggression and Violence in adult and youth acute inpatient settings in Victoria was created. The AcciMap identified 30 factors and 155 subfactors organised across six system levels. Upstream pressures at the government and organisational levels, including sustained funding constraints, legislative and regulatory conditions, and patterns in leadership engagement, create conditions that flow through to operational and workforce levels, manifesting as constrained staffing, cumulative workforce strain including burnout, and capability development gaps. At the consumer and environment levels, high clinical acuity, the impact of involuntary status, limitations in physical design, and reduced therapeutic quality of ward environments completed the picture of a system in which Aggression and Violence is a predictable outcome of accumulated pressures rather than isolated incidents.

PreventiMap

The PreventiMap for Aggression and Violence in adult and youth acute inpatient settings in Victoria was produced. The PreventiMap outlined 27 measures and 118 submeasures across the same six levels. Effective prevention was identified as requiring coordinated action across governance, workforce capability, environmental design, and relational care practices. Safety-focused design, appropriate staffing models, comprehensive training, and dignity-based consumer engagement were identified as central pillars.

As described above, both maps were further refined and validated through Phases 5 and 6. The final versions of the AcciMap and PreventiMap are presented in the interactive maps below.

Introduction

To ensure the AcciMap and PreventiMap were comprehensive and did not omit perspectives of the Lived and Living Experience (LLE) (both carers and peer workers) and participants in Forensic acute mental health inpatient settings, additional data collection and analysis were conducted concurrently with validation sessions (see Phase 6).

Methods — Procedure

A total of 13 semi-structured interviews and 4 group workshops were conducted with participants from the LLE (Carer), LLE (Peer Worker), and Forensicare groups, yielding 17 participants in total. The same interview and workshop structures used in the previous data collection phase were applied, ensuring consistency in how data were gathered across the project.

Interviews provided an opportunity for in-depth, individual exploration of participants' experiences and perspectives on Aggression and Violence contributing factors and preventive measures. Workshop sessions complemented this by allowing for group discussion, enabling participants to build on and respond to each other's contributions.

An overview of the participants' demographic information is presented in Table 2.

Phase 5 Participant Summary

Workshops comprised the larger portion of the additional groups sample (52%) across four sessions, with interviews accounting for the remaining 48%. Workshop 1, Workshop 2, and Workshop 3 each contributed 15%, and Workshop 4 contributed 7%. All workshop participants worked in adult settings (100%). Interview participants were predominantly from adult settings (69%), with smaller proportions working in youth settings (23%) and across both youth and adult populations (8%).

In terms of roles, Carers made up the largest group overall (41%), followed by Peer Workforce (37%) and OHS Professionals (11%). Among interview participants, Carers were the most common role (54%), followed by OHS Professionals (23%) and Frontline Supervisors/Managers (15%). Workshop 1 participants were exclusively Carers, whilst Workshop 2, Workshop 3, and Workshop 4 participants were exclusively Peer Workforce.

The broader cohort representation reflected these roles, with LLE (Carer) comprising the largest group (41%) and LLE (Peer Workforce) 37% of the total sample.

Methods — Data Analysis

Prior to analysis, all data were de-identified. Data from the 13 interviews and 4 workshops were analysed iteratively and simultaneously alongside the ongoing validation sessions.

Analytical Approach and Map Updates

Analysis was conducted using thematic analysis, a qualitative method for systematically identifying and interpreting patterns of meaning across data sources. Themes were developed inductively from participant data and then reviewed deductively against the existing map content to identify gaps, areas of insufficient coverage, and items warranting revision.

The analytical focus was applied at two levels for each map. For the AcciMap, analysis examined content at both the factor level and the sub-factor level. For the PreventiMap, the equivalent examination was conducted at the measure and sub-measure level. Across both maps, three types of findings were distinguished: content that was absent and warranted addition; content that was present but insufficiently prominent or described; and content that participants had reframed or challenged.

Results

17 data sources were analysed across 13 interviews and 4 workshops. All six levels of both maps were assessed for comprehensiveness across contributing factors and preventive measures. Both maps were confirmed as broadly comprehensive. The findings extended and enriched existing content, with the overall structure remaining intact. Key themes of LLE inclusion, therapeutic environment, and policy-to-practice gaps ran consistently across both maps and all source types.

Key Findings (Contributing Factors)

Detailed findings are reflected in the final AcciMap presented below and further described in Appendix L.

Key Findings (Preventative Measures)

Detailed findings are reflected in the final PreventiMap presented below and further described in Appendix M.

Introduction

Validation sessions engaged stakeholder groups in reviewing the maps. A mixed-methods validation process was conducted to assess the accuracy, comprehensiveness, and relevance of two previously developed systems maps: an AcciMap mapping contributing factors to Aggression and Violence in adult and youth acute mental health inpatient settings, and a PreventiMap mapping corresponding preventative measures.

Methods — Procedure

The validation drew on two complementary data sources: a series of stakeholder validation sessions and an online validation survey. Both maps were presented and assessed within the same sessions and survey.

Validation Sessions

Eight validation sessions were conducted between December 2025 and February 2026. Sessions comprised five individual interviews and three group workshops, all held online using the Miro collaborative whiteboard platform. Each session presented the full AcciMap and PreventiMap and invited participants to engage with both maps through annotations directly on the Miro board.

Validation Survey

An online validation survey was administered in parallel with the sessions. For the AcciMap, participants rated each of the contributing factors and their associated sub-factors. For the PreventiMap, participants rated each of the preventative measures and their associated sub-measures. Ratings were made using a four-point ordinal scale: Not Relevant, Somewhat Relevant, Relevant, and Highly Relevant. The survey was completed by 33 respondents. Five rows were entirely empty and were excluded, yielding a usable sample of n=28.

Phase 6 Participant Summary

Surveys accounted for the largest portion of the validation sample (47%), followed by workshops combined (44%), and interviews contributing the smallest share (8%). Among workshop sessions, Workshop 2 was the largest (25%), followed by Workshop 3 (12%) and Workshop 1 (7%).

All workshop participants worked in adult settings. The majority of interview participants worked across both youth and adult populations (80%), with the remainder working in adult settings only (20%). Among survey participants with valid population data, most worked in adult settings (94%), with one participant from an adolescent acute mental health setting.

In terms of roles, Rep. (Union/Other) was the most common across the full sample (39%). OHS Professionals represented 19% of the total sample. Among survey participants, Workforce (Staff) made up the largest group (36%), followed by Directors (21%) and OHS Professionals (14%).

Methods — Data Analysis

Prior to analysis, all validation data were de-identified. A mixed-methods approach was adopted, combining quantitative analysis of survey relevance ratings with qualitative analysis of session annotations, verbal discussion notes, and open-ended survey responses.

• Validation Session Analysis. Session data were analysed systematically across all eight sessions. For each annotated factor, measure, sub-factor, or sub-measure, the content of written sticky notes, the number and location of importance markers, and any general notes associated with a system level were extracted and recorded.
• Survey Analysis. Survey data were analysed in two stages. In the first stage, quantitative ratings were analysed separately for the two rater groups. In the second stage, open-ended responses were extracted and thematic analysis was applied.
• Integrated Analysis and Map Updates. Findings from both data sources were integrated into consolidated implications analyses. Each factor and measure was assessed against four categories: Strengthen, Expand, Revise, and Validate.

Results — Summary of Findings

Analysis across the 13 interviews and 4 workshops surfaced a substantial range of new and refined content across both maps.

• AcciMap. Findings identified contributing conditions at every level of the map, from government and legislative conditions, including the structural separation of AOD and mental health systems and the limited accountability mechanisms for inquiry recommendations, through to workforce-level factors such as gaps in neurodiversity-informed training and patterns in peer worker deployment, and environment-level factors including the therapeutic impact of ward design and the role of sensory stressors in escalation.
• PreventiMap. Findings highlighted a broad range of preventive measures not previously captured, spanning government-directed policy reform, organisational cultural transformation, operational practices such as pre-shift huddles and structured admission orientation, workforce training in relational and neurodiversity-informed skills, consumer-facing practices centred on autonomy and humanised de-escalation, and specific environmental provisions from sensory comfort items to outdoor recovery spaces.

Across both maps, themes related to LLE inclusion, the therapeutic quality of the care environment, and the gap between policy intent and ward-level reality were particularly prominent and consistent across sources.

Key Findings — Contributing Factors

Broad Endorsement. A large majority of factors and sub-factors in the AcciMap were endorsed by participants as relevant and accurate representations of contributing factors to Aggression and Violence.

Five key findings emerged with the strongest and most consistent evidence:

Detailed findings are reflected in the AcciMap presented below and further described in Appendix N.

Key Findings — Preventative Measures

Broad Endorsement. Findings indicate broad validation of the PreventiMap. Endorsement levels were generally stronger than those observed in the AcciMap validation.

Five findings with the strongest and most consistent evidence:

Detailed findings are reflected in the PreventiMap presented below and further described in Appendix O.

AcciMap & PreventiMap Findings — Contributing Factors

The AcciMap analysis demonstrated that Aggression and Violence in acute inpatient mental health settings emerges from the interaction of pressures distributed across all levels of the system rather than from isolated incidents or individual behaviours. Participants consistently described Aggression and Violence as a predictable outcome of accumulated systemic pressures that shape the experiences, behaviours, and decision-making of consumers, staff, managers, and organisations alike. Across all phases, the findings highlighted how upstream policy, funding, and regulatory conditions flow through organisational and operational systems before manifesting at the unit level as heightened acuity, workforce strain, reduced therapeutic engagement, and increased likelihood of escalation.

System-Level Drivers

The findings highlighted how upstream policy, funding, and regulatory conditions flow through organisational and operational systems before manifesting at the unit level as heightened acuity, workforce strain, reduced therapeutic engagement, and increased likelihood of escalation.

Participants consistently identified government, regulatory, and broader system-level pressures as foundational contributors to Aggression and Violence risk. Sustained funding constraints (CF-1.3), variable cross-sector coordination (CF-1.2), and legislative tensions relating to restrictive interventions and occupational safety (CF-1.4) were described as creating persistent instability across the sector.

Many participants described the mental health system as operating in a largely reactive mode, where demand pressures, limited inpatient capacity, and insufficient community supports constrained the ability of services to provide optimal therapeutic care. Bed shortages and constrained system capacity (CF-1.3.2) were repeatedly identified as creating throughput pressures that flowed directly into ward-level operations, contributing to unsafe admissions, inappropriate placements, and reduced therapeutic time with consumers.

Participants also identified variable coordination between mental health services, alcohol and other drug services, housing systems, emergency services, and community providers (CF-1.2.1; CF-1.2.6) as contributing to delayed discharge, repeat admissions, and consumer distress. The broader mental health system was frequently described as absorbing the consequences of conditions occurring elsewhere in the service system.

Legislative and policy settings were also identified as contributing to operational uncertainty and escalation risk. Participants described tensions between consumer rights frameworks, restrictive intervention requirements, and occupational safety obligations (CF-1.4.1), alongside variation in regulatory interpretation and practical uncertainty regarding implementation at the ward level.

Phase 5 workshops surfaced a broader pattern: well-intentioned policies set at government or system level are applied within inpatient environments without sufficient weight given to operational impact and without the corresponding risk control package needed to make them workable. The blanket smoking prohibition for involuntary consumers (CF-1.4.11) was the most-cited example. Participants described its impact as predictable and consistently underestimated — a recurring source of frustration was that the operational consequences were foreseeable to those working on the ward, yet the policy was still adopted. Nicotine withdrawal, restricted movement, and limited or ineffective substitutes (e.g. NRT) combine to create a highly predictable escalation pathway. The smoking example is significant in its own right, and also points to a broader pattern of decisions made at a distance from the work setting.

Regional and rural inequities also emerged strongly during validation activities. Participants described services outside metropolitan areas as disproportionately impacted by workforce shortages, limited specialist support, delayed emergency responses, limited community pathways, and reliance on temporary workforce arrangements.

Two further system-level factors were identified through the project's own engagement and consultation activities. Ongoing tension between worker-safety and therapeutic-care framings (CF-1.6) shaped stakeholder discussions and at times contributed to difficulty establishing shared priorities and coordinated approaches to Aggression and Violence risk. Related to this, the absence of a shared cross-sector understanding of Aggression and Violence (CF-1.7), including differing conceptual perspectives, terminology, and evidence interpretation, occasionally acted as a barrier to collaborative problem-solving and slowed progression toward sector consensus.

Organisational and Governance Factors

Participants consistently described organisational leadership, governance, and culture as shaping the conditions under which unit staff attempted to manage risk. Leadership disconnect from ward-level realities (CF-2.3.1), limited accountability for safety (CF-2.3.2), and organisational priorities perceived to favour financial or operational pressures over safety (CF-2.3.6) were frequently identified throughout the analysis.

Many participants described consultation processes as tokenistic or predetermined, particularly in relation to infrastructure design, operational reform, and workforce change processes. Staff and LLE participants reported limited influence over decisions that directly affected safety and care delivery.

A recurring theme across workshops and interviews was the perception that incident reporting rarely resulted in meaningful action. Participants described reporting systems and investigation processes as burdensome, overly focused on compliance, and limited in orientation toward prevention and organisational learning. Investigations were often perceived as focused on individual behaviours rather than systemic drivers, and unlikely to result in sustained change.

Participants also described broader cultural patterns that contributed to normalisation of violence within inpatient settings. Across multiple services, Aggression and Violence was described as having become increasingly viewed as "part of the job," contributing to reduced reporting, diminished expectations for change, and ongoing psychological harm to staff.

A further set of governance factors was identified through the project's own consultation and oversight activities. Limited psychological safety within governance and consultation forums (CF-2.6) was observed in uneven participation patterns, premature closure of lines of discussion, and hesitancy to share dissenting perspectives. Closely related, unbalanced oversight and decision-making practices (CF-2.7), characterised by disproportionate focus on terminology and methodology, occasionally slowed progression of broader systems discussions. Limited collaborative evidence and knowledge development (CF-2.8) reflected differing assumptions across stakeholder groups about acceptable forms of evidence and the appropriate balance between empirical, operational and LLE perspectives. Finally, highly risk-sensitive lived and living experience engagement models (CF-2.9), while well intentioned, at times introduced procedural complexity that reduced the agility and authenticity of consultation activities.

Operational Pressures and Service Delivery

Operational management factors were identified as intensifying many of the systemic and organisational pressures experienced on the unit. Participants described environments characterised by high workloads, competing demands, staffing instability, and sustained pressure to maintain patient flow.

Bed management pressures were repeatedly identified as overriding clinical considerations, resulting in unsafe admissions, inappropriate consumer placements, and reduced opportunities for preventative engagement. Unsafe or inappropriate admission and placement describes situations where the consumer's acuity, gender, cultural needs or care requirements are not matched to the unit's capability or current consumer mix. These placements are often driven by bed-flow pressure, demand peaks or limited alternative options rather than by clinical suitability, and contribute to risk for the consumer, other consumers on the unit, and the workforce. Managers were often described as constrained by administrative burden, limited authority, and competing organisational expectations.

Participants also identified several operational conditions that contributed to escalation risk, including:

• inconsistent communication and handover processes
• limited or static risk assessments
• variable continuity of care
• inconsistent implementation of procedures
• limited capacity for proactive planning

Phase 6 returned often to procedural non-adherence. Participants described staff non-adherence to procedures (CF-4.5) as an outcome of workload intensity, procedural impracticality, staffing shortages, and competing operational demands rather than as an individual behavioural problem. This factor was the most frequently revisited across validation sessions.

Workforce Factors

Participants described a workforce operating under sustained psychological and operational strain. Burnout, cumulative trauma from exposure to repeated incidents over time, fatigue, and emotional exhaustion (CF-4.4) were identified as pervasive across many services and were described as directly affecting staff wellbeing, reflective capacity, and therapeutic engagement.

Several interacting workforce pressures were repeatedly identified:

• high turnover and attrition of experienced staff
• increasing reliance on agency and temporary workforce arrangements
• growing proportions of junior or inexperienced staff
• inconsistent skill mix across shifts
• limited access to reflective practice and supervision

Participants also identified gaps in workforce capability development. Variable training quality, limited onboarding, constrained de-escalation capability development, and limited preparation for managing complex presentations were all identified as contributing factors.

Phase 5 findings strengthened this area significantly through the identification of neurodiversity-related capability gaps. Participants reported that autistic distress and "meltdowns" are frequently misinterpreted as deliberate aggression due to limited workforce understanding and adaptive communication strategies that have not yet been embedded.

Interdisciplinary tensions and inconsistent approaches to care were also identified as contributing to fragmentation during escalation events. Participants described conflict between professional groups, inconsistent behavioural expectations, and differing philosophies regarding restrictive interventions as reducing cohesion and increasing risk.

Consumer Experience and Escalation

Participants consistently emphasised that many aggressive incidents occur within contexts of acute distress, fear, trauma, confusion, and perceived loss of control. High clinical acuity, severe mental illness, substance use, trauma histories, cognitive impairment, and complex comorbidities (CF-5.1) were all identified as increasing vulnerability to escalation.

Importantly, participants strongly cautioned against framing aggression solely as an individual consumer characteristic. LLE engagement significantly reframed the understanding of Aggression and Violence by highlighting the extent to which aggression may emerge as a reactive response to coercion, restrictive environments, involuntary treatment processes, and experiences of being unheard or disempowered.

The revision of the factor relating to involuntary status and loss of rights was particularly significant in this regard. Participants described aggression in these contexts not simply as initiating violence, but frequently as a defensive or reactive response to system-generated distress.

Participants also identified several relational contributors to escalation, including:

• inconsistent communication
• limited explanation of rules and processes
• lack of transparency regarding care decisions
• limited consumer involvement in planning
• perceived unfairness
• experiences of disrespect or loss of dignity

Environmental and Infrastructure Conditions

Participants described many inpatient environments as outdated, overcrowded, noisy, sterile, institutional, and poorly designed for therapeutic care.

Layouts with blind spots, limited lines of sight, restricted exits, weaponisable objects, and constrained duress systems were all identified as increasing physical safety risks for both consumers and staff.

At the same time, participants emphasised that the environment contributes psychologically to escalation through overstimulation and sensory overload, boredom, limited privacy, and reinforcement of institutional control. Many participants described environments that felt punitive or custodial rather than therapeutic.

Limited access to meaningful activities, quiet spaces, sensory supports, and therapeutic environments was repeatedly identified as increasing agitation and reducing opportunities for early de-escalation.

Validation activities identified physical design and infrastructure conditions as one of the most strongly endorsed contributing factor domains across the entire dataset, resulting in the Environment and Equipment level receiving high-priority designation during Phase 6 validation.

PreventiMap Findings — Preventative Measures

The PreventiMap analysis demonstrated that preventing Aggression and Violence requires coordinated action across all levels of the system rather than reliance on isolated interventions or reactive incident management. Participants consistently described effective prevention as dependent on the interaction of system-level reform, organisational leadership, operational capability, workforce support, therapeutic care practices, and safety-focused environmental design. Across all phases, the findings reinforced that sustainable reductions in Aggression and Violence require proactive, preventative, and relational approaches that address underlying system drivers rather than focusing solely on crisis response.

Participants did not describe prevention as a single initiative or program, but rather as the cumulative effect of multiple protective conditions operating together. These included psychologically safe leadership, stable staffing models, therapeutic environments, proactive risk assessment, consumer dignity, strong team culture, and meaningful engagement with LLE perspectives. Psychologically safe leadership refers to leadership that creates conditions where staff, consumers and LLE contributors can raise safety concerns, disagree, or share emerging risks without fear of negative consequences, and where those concerns are visibly heard and acted on.

System-Level Prevention and Reform

Participants consistently identified system-level reform as foundational to sustainable Aggression and Violence prevention. Improved funding models (PM-1.2), stronger inter-agency coordination (PM-1.5), and clearer legislative and regulatory frameworks (PM-1.4) were described as essential for reducing the downstream pressures contributing to escalation within inpatient units.

Many participants argued that prevention efforts within hospitals would remain limited unless broader community and external support systems were strengthened. This included improved housing pathways, increased alcohol and other drug support services, expanded community mental health programs, and better discharge coordination processes. Participants described how limited community capacity often resulted in avoidable admissions, prolonged inpatient stays, and repeated presentations, all of which increased pressure on acute settings.

Participants also identified the need for greater consistency across the mental health system through standardised training, clinical frameworks, reporting approaches, and operational expectations (PM-1.6). Variation between services was frequently described as contributing to inconsistent safety practices and uneven workforce capability.

Several participants also advocated for legislative and regulatory reform that more effectively balanced consumer rights, therapeutic care, and workforce safety. This included clearer guidance regarding the use of restrictive interventions, stronger minimum safety standards, and more practical implementation support for legislative reforms.

Regional and rural participants particularly emphasised the need for more equitable system planning and resource distribution to address persistent workforce and infrastructure disadvantages.

Organisational Leadership and Culture

Participants consistently described leadership and organisational culture as critical enablers of prevention. Accountable governance structures (PM-2.1), psychologically safe organisational cultures (PM-2.7), and integrated safety planning systems (PM-2.6) were all viewed as foundational protective factors.

Many participants described prevention-oriented organisations as those in which leadership maintained visible engagement with 'on the ground' realities, actively listened to staff and consumers, responded transparently to safety concerns, and demonstrated sustained commitment to implementing change. Participants repeatedly contrasted this with reactive or compliance-focused approaches that prioritised incident management over long-term prevention.

A strong reporting culture (PM-2.5) was identified as another important protective factor. Participants described the importance of staff feeling psychologically safe to report incidents, near misses, and emerging risks without fear of blame or negative consequences. Effective organisations were described as those that closed the loop following reporting by communicating findings, implementing recommendations, and demonstrating visible learning.

Participants also strongly emphasised the value of co-design and stakeholder engagement (PM-2.3). Across workshops and interviews, staff and LLE contributors highlighted the importance of involving consumers, peer workers, unit staff, unions, and health and safety representatives in policy development, service design, and environmental planning.

Importantly, several participants described organisational culture change as both achievable and highly impactful. One LLE contributor described a service that reduced seclusion admissions from approximately 300 per year to zero over a three-year period through deliberate cultural transformation focused on relational care, leadership commitment, and reduction of coercive practices.

Summary of Project Findings

This project applied a six-phase, iterative systems thinking methodology to investigate Aggression and Violence in adult and youth acute inpatient mental health settings in Victoria.

Insights from Initial Data Collection (Phase 4)

Phase 4 produced an AcciMap and PreventiMap. The AcciMap identified 30 factors and 155 subfactors organised across six system levels. The PreventiMap outlined 27 measures and 118 submeasures across the same six levels.

Findings from Lived and Living Experience (LLE) and Forensicare Groups (Phase 5)

Phase 5 involved four workshops and 13 interviews (n=17). Analysis produced a total of 26 changes to the AcciMap (25 new subfactors, one revised) and 44 new submeasures added to the PreventiMap.

Key Phase 5 additions included: blanket smoking prohibition as a predictable escalation trigger and as an example of the broader pattern of policy set at a distance from operations; bed-flow pressure leading to admission or placement decisions where consumer acuity, gender, cultural needs or care requirements are not matched to the unit's capability or current consumer mix; neurodiversity training gaps; peer worker deployment patterns that limit their distinctive contribution; rostering-driven desensitisation; security staff preparedness gaps; and reframing of involuntary status aggression as reactive or defensive in nature.

Phase 5 preventative measure additions included: supervised smoking areas; pre-shift huddles; staged consumer orientation; proactive peer worker deployment; relational and connection training; dignity provisions; and cultural transformation.

Insights from Systems Map Validation (Phase 6)

Phase 6 involved eight workshops and a survey, producing a total of 33 changes to the AcciMap (12 revised, 2 removed, 10 new, 9 high-priority designations) and 20 changes to the PreventiMap (1 revised measure, 4 revised submeasures, 8 new submeasures, 7 high-priority designations).

Phase 6 confirmed strong validation of both maps. The most strongly endorsed contributing factor domains were physical design and infrastructure, system funding and resourcing conditions, cumulative workforce strain (including burnout and fatigue), and high consumer acuity. Staff non-adherence was reframed by participants as a systemic factor — described in the AcciMap as "Barriers to Safety Procedure Adherence". Regional and rural inequity emerged as a new theme.

Most strongly supported preventative measures: safety-in-design, appropriate staffing levels and skill mix, comprehensive staff training, and post-incident debriefing. Zero tolerance approaches and consumer accountability language were consistently challenged and revised.

Summary & Limitations

Across all phases, the findings consistently demonstrate that Aggression and Violence in acute inpatient mental health settings is a systemic issue arising from the interaction of factors across all levels of the system. The AcciMap and PreventiMap collectively show that Aggression and Violence is a predictable outcome of system design, shaped by policy settings, resource constraints, environmental conditions, workforce capability, and consumer experience.

The research highlights that upstream drivers, particularly systemic underfunding, policy constraints, and service demand, create conditions that flow through to frontline care, where they manifest as high acuity, staff fatigue, reduced therapeutic engagement, and increased likelihood of escalation. The inclusion of Lived and Living Experience (LLE) perspectives in Phase 5 added significant depth to this picture, surfacing drivers and prevention opportunities that had not been captured through earlier data collection, and reframing aggression as frequently reactive to system-generated coercion.

The validation process in Phase 6 confirmed that both the problem space and the solution space are well understood and strongly supported by stakeholders. Reducing Aggression and Violence is achievable, but requires a shift from reactive, incident-focused responses to proactive, system-wide strategies that address underlying drivers and enable safer, more therapeutic environments.

Limitations

• The literature review included studies that were heterogeneous in design, perspectives, and types of Aggression and Violence, which may limit consistency and generalisability.
• The project provides limited insights into all aspects of the system, as not every perspective or process could be captured.
• The data relied on input from key actors, which may introduce bias based on their roles and experiences within the system.
• The relationships identified in the systems maps were derived qualitatively rather than statistically, which may affect the precision of their interpretation.
• The systems maps represent an imposed hierarchical structure to organise information for clarity, which may not fully capture the complexity or true nature of the system.