Occupational violence and aggression in urgent and critical care in rural health service settings: A systematic review of mixed studies

Abstract Rural/remote health services are vulnerable to occupational violence and aggression due to factors such as weapon accessibility, poor network coverage and distance to backup. This systematic review investigated (1) the nature of occupational violence and aggression perpetrated in rural/remote health service urgent care settings and (2) the availability and effectiveness of policies/interventions/recommendations that address occupational violence and aggression in this context. We searched Business Source Complete, CINAHL Complete, Health & Society, APAIS Health, Health Collection, PsycINFO, PubMed, Scopus, SocIndex and Web of Science. Included articles (peer‐reviewed, no grey literature and English language) addressed occupational violence and aggression in rural health service urgent care settings. Fifteen articles matched these criteria (total [rural/remote only, where specified] N ~ 2555) and were included in the final analysis. The Mixed Methods Appraisal Tool was applied to assess the risk of bias. A data extraction table and narrative synthesis are presented. The most common occupational violence and aggression type was verbal aggression. The primary perpetrator was patients. Risk factors reflected practitioner age, remoteness, sector, staffing, shift type and area of practice. Precipitating factors were alcohol/drugs, dissatisfaction and mental health conditions. Policy content and limitations and education/training programme effectiveness were not addressed. Community collaboration supported occupational violence and aggression prevention/management. Organisational culture should promote reporting, debriefing and post‐incident care for staff well‐being. Work environment and job/task design are priorities for safety, but with possible limitations for traumatised clients. Occupational violence and aggression policies/interventions in rural health settings must be systematically evaluated to inform best practices. Co‐funded by Swinburne Social Innovation Research Institute Interdisciplinary Seed Funding Scheme and SMART Rural Health Network.


| INTRODUC TI ON
Occupational violence and aggression describe 'incidents in which a person is abused, threatened or assaulted in circumstances relating to their work' (Victorian Department of Health and Human Services -DHHS, 2017). Occupational violence and aggression is a prominent issue across the healthcare sector worldwide (Auditor-General Victoria, 2015;Kerr et al., 2017;Morphet et al., 2018;Shea et al., 2017) and is estimated to affect 95% of Australian healthcare workers (Spelten et al., 2020). Occupational violence and aggression incidents place workers at risk of physical injury or even death, as well as increasing the risk of burnout and post-traumatic stress (Kerr et al., 2017;Rees et al., 2018;Spelten et al., 2020). There are also financial implications due to lost days of work, work incapacity claims and costs invested to improve work safety (Hassard et al., 2019;Maguire et al., 2018). Additionally, occupational violence and aggression may reduce the quality of clinical care for a variety of reasons, for example, burnout (Spelten et al., 2020), thus potentially compromising public health at a broader level.
Given the adverse nature of occupational violence and aggression and its implications within healthcare settings, there has been considerable investment in establishing guidelines and protocols to manage it. In Australia, the Victorian Departmental of Health and Human Services recently published the Code Grey Standards, which provide guidelines for 'an organisation-level response to actual or potential violent, aggressive, abusive or threatening behaviour, exhibited by patients or visitors, towards others or themselves, which creates a risk to health and safety' (Victorian DHHS, 2017).
However, the standards were described within the context of 'ideal' well-resourced health services, where there are teams of clinicians and security staff, and access to resources, equipment and contextspecific training. Rural health service sites take a range of forms and have varying team sizes, but many are small sites with restricted staffing levels and little or no access to security or legal services (Hills et al., 2019). In addition, rural health services may be vulnerable due to additional challenges including geographical distance to emergency backup/response (Grant & Hartanto, 2019), access of local residents to guns and knives (Lyneham, 2000), poor mobile phone connectivity (Grant & Hartanto, 2019;Thomas et al., 2020), built environments and grounds that are difficult to secure safely (Hills et al., 2019) and reduced privacy of healthcare workers in the community, resulting in a reluctance to pursue incidents (Hills et al., 2019). As such, when occupational violence and aggression incidents occur, it may be difficult for workers to know how best to respond, given these circumstances.
At the time of writing, there were no clear guidelines on how to design locally contextualised occupational violence and aggression practices for implementation within rural health services. Thus, we conducted a systematic review focussing on occupational violence and aggression in rural and remote health services in a high-risk setting: urgent care centres/emergency departments (Cabilan & Johnston, 2019;Gacki-Smith et al., 2009;Lau et al., 2004;Nikathil et al., 2017;Spelten et al., 2020). The objectives of the systematic review are to summarise and synthesise evidence relating to occupational violence and aggression perpetrated by patients/visitors in rural health service urgent care facilities, and to inform the management of incidents in rural health service urgent care settings, to support improve health and safety outcomes. Our research questions were as follows: What is known about this topic • Occupational violence and aggression are estimated to affect 95% of healthcare workers in Australia, placing workers at risk physically and psychologically, with financial implications for organisations, for example, absenteeism, work incapacity claims and costs invested to improve safety.
• Occupational violence and aggression may compromise public health at a broader level by reducing the quality of clinical care.
• Current guidelines for the prevention and management of occupational violence and aggression overlook risk factors in small, rural health services that often have limited staff and resources as well as additional challenges due to remoteness, for example, distance to emergency backup/response.

What this paper adds
• This systematic review describes the nature of occupational violence and aggression in rural health services for nurses and healthcare workers generally, including incident characteristics (type, frequency and severity), perpetrator characteristics and risk/precipitating factors.
• Where possible, the review provides an overview of the effectiveness of policies/interventions/ recommendations for the prevention and management of occupational violence and aggression in rural health services.
• With all included studies but one conducted in Australia, the findings provide an insight into challenges and gaps in the prevention and management of occupational violence and aggression in rural health services in Australia.

| Search strategy
We searched the following electronic databases using the default set- However, after the initial search, we decided to narrow our review to literature that specifically addresses occupational violence and aggression in rural health service urgent care facilities due to time and budget restrictions.
We did not impose date and language restrictions as we wanted to capture a broad range of local and international literature. For literature in languages other than English, we attempted to find an English language version. The search was conducted on March 3, 2019, and March 4, 2019. Ethical approval was not required as we used published studies in which primary investigators obtained informed consent.

| Eligibility criteria
Articles were included if they were peer-reviewed/refereed literature (including refereed book chapters, qualitative and quantitative studies, mixed-method studies and review papers) that address occupational violence and aggression in rural health service urgent care settings. Exclusion criteria were as follows: (1)

| Study screening
The initial search identified 691 records after removing duplicates.
Articles were subsequently uploaded to Covidence (www.covid ence.org) for screening. First, co-authors 1 and 2 independently conducted title and abstract screening, whereby literature was sorted into 'yes', 'maybe' or 'no' categories based on the inclusion and exclusion criteria. This resulted in 573 articles being excluded.
Next, these co-authors independently conducted full-text screening on the 118 'yes' and 'maybe' records, recording the reason for exclusion. Disagreements were resolved through discussion or, if required, through consultation with an independent co-author (3).
We excluded 103 articles for not meeting the inclusion criteria and retained 15 articles for inclusion. Figure 1 summarises our process.

| Quality assessment
Co-authors 1 and 2 independently conducted risk of bias (quality) assessments using the Mixed Methods Appraisal Tool (Hong et al., 2018). This tool includes five methodological criteria questions (yes, no and unclear) for each study type (see Mixed Methods Appraisal Tool 2018 user guide; Hong et al., 2018) and is considered an effective and efficient quality appraisal tool for empirical research, with good inter-rater reliability (Pace et al., 2012). Disagreements were resolved by discussion, with the involvement of an independent co-author (3) as necessary.
A quality score for each study was calculated based on the percentage of 'yes' responses divided by five, with scores ranging from 0% (no criteria met) to 100% (all five criteria met). Quality scores are presented in Table 1. Most studies were of adequate quality: seven studies (46.7%) were high quality, four studies (26.7%) were moderate quality and three studies (20.0%) were low quality. One included article is a literature review and thus was not assessed using the tool. No studies were excluded from the review based on quality scores. Inter-rater reliability for quality scores was calculated using a two-way mixed-model intra-class correlation coefficient (Koo & Li, 2016). The intra-class correlation coefficient was 0.57, indicating moderate-to-good agreement between assessors.
Co-authors 1 and 2 independently assessed the certainty of the body of evidence, as related to the objectives of the review and the research questions (see Table 1). We assessed the certainty of evidence as high (4), moderate (3), low (2), very low (1) or uncertain/not applicable (0) with regard to the following criteria for upgrading the certainty of evidence: (a) Relevance of the sample: (1) data are aggregated but include rural health services, healthcare workers working in rural health services or healthcare workers working in emergency/urgent care in rural health services (very low); (2) data are specific to various healthcare workers in rural health services (low); (3) data are specific to a particular subgroup of healthcare workers in rural health services (moderate) or (4) data are specific to emergency/urgent and critical care healthcare workers in rural health services (high); (b) (i) Relevance to RQ1: (1) describes one of the following criteria: occupational violence and aggression type, frequency, severity, source/perpetrator characteristics and risk factors (very low); (2) describes two of the previous criteria (low); (3) describes three of the previous criteria (moderate) or (4) describes at least four of the previous criteria (high) and/or (ii) Relevance to RQ2: (1) broadly discusses policy or interventions (very low); (2) makes broad recommendations for policy or interventions, based on evidence (low); (3) discusses the effectiveness of specific policies or interventions (moderate) or (4) discusses the effectiveness of specific policies and interventions (high). We averaged scores across criteria (a) and (b) to generate a total score (0-4). Scores ranged from 0.67 to 3.67. Inter-rater reliability (intra-class correlation coefficient) for certainty was 0.98 indicating excellent agreement.

| Data extraction
For each article, we extracted the objectives, methods and findings (see Table 1).
Articles were divided randomly among co-authors 1 and 2 and samples of each co-authors' extraction (approximately 30%) were checked for consistency in the early stages.
A narrative synthesis, structured around the nature and prevalence of occupational violence and aggression incidents, policies/ interventions/recommendations to manage incidents and the effectiveness of these policies/interventions/recommendations, is presented below. Quantitative synthesis was not possible as the studies were not sufficiently homogeneous due to varying samples, research methods and data analytic approaches. In addition, some studies provided disaggregated data for the population of in-  Participants were asked to report on their experiences of patient and visitor assault in the past 4 working weeks. They were then asked to (a) rate the importance of 26 'protective' factors to help prevent and manage assault and (b) indicate the presence of factors in their work setting. Statistical tests: Descriptive statistics (percentages and frequencies) and binary regression analysis (presence of protective factors as a predictor of patient and visitor assault) with control of potential covariates: nursing division, work location, work type, employment, duty roster and clinical setting 36% of participants reported patient visitor assault in the past 4 working weeks and 46% reported three or more incidents. Verbal abuse (90%), physical abuse (45%) and threat of harm (27%) Patients were approximately 2.5 more times assaultive than visitors Men aged over 50 years accounted for 54% of patient visitor assault incidents Protective factors were high standard facilities, personal protective equipment (mobile phones and personal duress alarms), sufficient staffing levels and effective enforcement of policies by management (policy details not provided) Risk factors were working in public settings; accident and emergency, aged care and mental health settings; working on a rotating roster and working on night duty Aggression management training was neither a protective factor nor a risk factor for patient visitor assault; the authors speculated that training could be 'inadequate'

Main findings
MMAT score (%) Certainty score (0-4) Fisher et al. (1996)  Remote area nurses aged 20-29 were more likely to experience sexual harassment on duty and on call More violence occurred in smaller communities, both on duty and on call; respondents on call 24 h were more likely to experience all types of violence Respondents without a security escort were more likely to experience verbal aggression and obscene behaviour on-call, physical violence on duty and on call and sexual harassment on and off duty Those living in independent accommodation were the least likely to experience violence Perpetrators tended to be male, Aboriginal and either a client or his family Violent incidents tended to occur mostly at night The location of violence was mostly frequently the clinic/hospital Incidents were associated with a range of negative emotions yet there were significant numbers who perceived the frequency and severity of incidents as low, except with physical violence Broader social context, for example, alcohol (all participants), dissatisfaction with the service provided, unreasonable/illegal requests, mental illness, grieving/sadness and rioting, were identified as contributing factor The majority of participants did not feel adequately prepared for their work (46.9% had no orientation prior to commencing their current position and 30.8% had no orientation or employer-based education since commencement). Only 53.6% received crosscultural information; 24.3% received personal safety information There was little or no support post-incident reporting apart from temporary leave; 32.8% no longer felt confident reporting future incidents. Only 52.8% always reported incidents GRANT et al.

Main findings
MMAT score (%) The proportion of female nurses and the age of the sample was slightly higher than in the population of nurses in New South Wales at the time of the study. Measures: demographic questions, a single item related to the frequency of participation in aggression management training or updates in the previous 5 years; items type of patient aggression experienced in the previous 3 months and the resulting number of days taken off work; and the 5-item efficacy in Dealing with Challenging Behaviours Scale (Hastings & Brown, 2002) Statistical tests: Factor and reliability analysis of Dealing with Challenging Behaviours Scale and correlational analysis of ordinal and interval variables The median number of times participants had received aggression management training in the previous 5 years was once; 39.5% had not participated in training, 36% had received training once, 17.2% twice and 6.9% three or more times Excluding managers and specialists/consultants, 76% of participants had experienced patient aggression in the previous 3 months. Registered nurses and midwives were most affected (83.4%) The most common form of aggression was verbal abuse, followed by verbal threats and intimidation, physical threats and intimidation and physical violence 96.7% had taken no days off work, 2.4% one or 2 days and two participants had taken either 10 days or 60 days off work 70% of participants reported self-efficacy in the low-to-medium range There was a moderate association between verbal threats or intimidation and physical threats or intimidation, suggesting co-occurrence. Participation in aggression management training showed a weak, negative association with verbal abuse experienced, and a weak, positive association with perceived self-efficacy GRANT et al.

Main findings
MMAT score (%) Certainty score (0-4) Mayhew and Chappell (2003)  Clients were the perpetrators in over three-quarters of all violent events, while relatives or visitors were the perpetrators in nearly one-third of threat and verbal abuse events Client-/patient-initiated violence was reported to be most common among those who suffer from injury, illness, brain injury, dementia or a semi-comatose state, mental health problems, those affected by drugs and/or alcohol and those recovering from anaesthesia Relative-/visitor-initiated violence was more commonly perpetrated by people from lower socio-economic backgrounds, although some reported perpetrators worked in skilled jobs Perpetrators were identified as male in 49% of the events, female in 23% and both male and female in 5% (gender was not reported in 23% of the events) The authors estimated that only between 8% and 12% of all occupational violence and aggression events are formally reported and recorded on official databases

| RE SULTS
All included studies but one (Jackson & Ashley, 2005) were conducted in Australia. Seven focused on nurses and midwives (Farrell et al., 2014;Fisher et al., 1996;Hegney et al., 2003;Hills, 2008; Lyneham In Lyneham's (2000) study of Australian rural and remote emergency nurses, verbal abuse and threats were the most prevalent form of occupational violence and aggression, followed by verbal abuse by phone and physical intimidation or assault respectively.
There were more incidents involving weapons in rural health services than remote ones.
With regard to severity, Ross-Adjie et al. (2007) found that 'violence against staff' was ranked as the top stressor among Australian emergency department nurses. However, the type and frequency of occupational violence and aggression were not measured in this study, and data were aggregated for rural and metropolitan respondents. Hills (2008) found that the most common form of occupational violence and aggression for Australian rural nurses was verbal abuse, followed by verbal threats and intimidation, physical threats and intimidation and physical violence. Drill-down data for areas of practice were not provided. Excluding managers and specialists/consultants, more than three-quarters of participants had experienced patient aggression in the previous 3 months, with registered nurses and midwives most impacted. There was a moderate association between 'verbal threats or intimidation' and 'physical threats or intimidation', suggesting co-occurrence.

Farrell et al.'s (2014) study of Australian nurses and midwives
found that the main types of occupational violence and aggression experienced were verbal abuse, physical abuse and the threat of harm in that order; almost 40% of participants had experienced occupational violence and aggression in the past 4 weeks, with half reporting three or more incidents. Rural and regional nurses and midwives were overrepresented in this sample and data were aggregated across all geographical locations.

Health services
A study of occupational violence and aggression in Jamaican urban and rural health services (Jackson & Ashley, 2005) found that psychological violence was considerably more common than physical violence over a 1-year period; however, urban and rural data were aggregated. Mayhew and Chappell (2003)  Overall, incidents were more prevalent among nurses than allied health professionals and general practitioners, with the latter two groups reporting similar levels. Nurses more commonly reported all forms of violence, except telephone threats, which were reported just as commonly by general practitioners. Urgent care was among the fields of practice in this study, but drill-down data were not available.
A study (Brock et al., 2009) focusing on a secure room in an Australian rural health service reported average use was about once per month, or at a rate of 1.1 uses/1000 emergency department visits.
Concern for staff safety was cited in 10% of the admissions.
Beattie et al. (Beattie, Griffiths, et al., 2018a;Beattie, Innes, et al., 2018b) conducted a study in Australian metropolitan and regional (remote) health services focussing on healthcare workers who were responsible for occupational violence and aggression prevention and management or who had experienced incidents. They found that there was a normalisation of occupational violence and aggression due to its frequency, especially in emergency departments, aged care and mental health. Metropolitan and regional data were reported in aggregate.
Together, these studies suggest that verbal aggression (abuse, threats and intimidation) is the most common type of occupational violence and aggression. Other types include obscene behaviour, property damage, verbal abuse by phone and physical aggression (abuse, assaults and intimidation). Given that all but one study discussed in this section focuses on the Australian context, we elaborate the discussion of our findings accordingly. are also needed to reinforce a zero-tolerance culture in rural health services and their communities.

| Perpetrator characteristics
For Australian remote area nurses, perpetrators tended to be male, Aboriginal clients or their family members (Fisher et al., 1996). For Australian rural and remote emergency department nurses, patients and their friends/relatives were the most common perpetrators (Lyneham, 2000). In Mayhew and Chappell's (2003) study (Australian rural and urban data aggregated), clients perpetrated over three-quarters of violent events, while relatives/visitors perpetrated nearly one-third of threat and verbal abuse events. Relative/ visitor-initiated violence was more commonly perpetrated by people of lower socio-economic backgrounds, although some perpetrators worked in skilled jobs. Around half of the perpetrators were male.
In Alexander et al.' (2004) study of Australian rural health services, respondents (various areas of practice) were unanimous that patients were the most distressing source of occupational violence and aggression. Farrell et al. (2014) found that patients were 2.5 times more assaultive than visitors, with men over 50 years accounting for more than half of the occupational violence and aggression incidents (data were aggregated across geographical locations in this Australian study). In the only international study included (Jackson & Ashley, 2005), patients were reported to be the primary perpe- Similarly, Lyneham (2000) found that alcohol, waiting time and drugs were the most three common precipitating factors of violence in Australian rural emergency departments. In contrast, alcohol, drugs and socioeconomic factors were the most three common precipitating factors in remote emergency departments in this study, suggesting that risk factors vary by local context. In Mayhew and Chappell's (2003) study (Australian rural and urban data aggregated), patient-initiated violence was most common among those who suffer from injury, illness, brain injury, dementia, mental health problems, substance abuse and anaesthesia aftereffects. Qualitative responses of several interviewees from rural health services in Mayhew and Chappell's study suggested not having a medical officer on site at the emergency department, and limited staff after hours were additional risks. Emergency departments, intensive care units, urgent and critical care, rural healthcare settings and remote/rural sites at night were among the settings identified as a 'high risk' for violence based on the collated evidence from 400 health workers.
In a Jamaican study (Jackson & Ashley, 2005), staff aged 34 years and below reported higher levels of verbal abuse compared to older staff, consistent with Australian findings that younger healthcare workers face a higher risk. Physical violence was slightly more common among staff aged 44 years and below, male staff and within general medicine, psychiatric wards, administration units and emergency units, respectively. The risk of physical violence was reduced among staff who were older, worked at night suggesting possible differences in staffing levels compared with the Australian context (Magin et al., 2010) or worked mostly with mentally disabled, geriatric or human immunodeficiency virus/acquired immune deficiency syndrome patients. Physicians and nurses were at an increased risk of experiencing physical violence overall, and staff who worked mostly with psychiatric patients were at increased risk of physical violence compared with those who worked with other patients (urban and rural data were aggregated). Similarly, the most frequent precipitating factor for occupational violence and aggression among nurses in Alexander et al.'s (2004) study of Australian rural health services was patient psychiatric issues. Service issues (allied health professionals) and drug and alcohol issues (general practitioners) were also reported as precipitating factors. A study in Australian metropolitan and regional (remote) health services (Beattie, Griffiths, et al., 2018a;Beattie, Innes, et al., 2018b) noted the challenges of engaging clients with altered cognitive states: stress and frustration and previous trauma can compromise clients' assessment of risk and safety, resulting in inappropriate reactivity to workers providing care.
Metropolitan and remote data were aggregated in this study. Magin et al.' (2010) study of Australian rural general practitioner registrars found that perceived risk was higher after hours than during business hours due to isolation, lower staffing levels and reduced police presence. Occupational violence and aggression, including perceived risks, was a significant source of stress particularly for female, young or less experienced registrars.
In Farrell et al.' (2014) study of Australian nurses and midwives, risk factors for occupational violence and aggression were working in public healthcare settings; accident and emergency, aged care and mental health settings; and a rotating roster and night duty. Rural and regional nurses and midwives were overrepresented in this sample, and data were aggregated across all geographical locations included.
The above studies indicate that risk factors for occupational violence and aggression include age of practitioner (younger), rurality/ remoteness, sector (public), staffing levels (low), type of shift (e.g., on-call and night shift-Australian studies and rotating roster) and area of practice (e.g., emergency department, intensive care unit and urgent and critical care). Common precipitating factors for occupational violence and aggression in the Australian context are alcohol and drugs, dissatisfaction with service/waiting times and mental health conditions. These findings suggest that triage processes for mental health and alcohol and other drugs are needed to reduce the risk of occupational violence and aggression and may be particularly important when individuals with a history of behaviours of concern present to health services.

| RQ 2. What is the availability and effectiveness of policies / interventions / recommendations that address occupational violence and aggression perpetrated by patients/ visitors in rural health service urgent care facilities?
4.2.1 | Policies Hegney et al. (2003) investigated the availability and perceived adequacy of occupational violence and aggression policy among private, public and aged care sector nurses in Australia. Policy content was not examined. In the public sector only, occupational violence and aggression policies were significantly less prevalent in remote areas than in capital cities and large rural centres. Men perceived policies as less effective than did women, and the level of experience was inversely related to perceived effectiveness. Citing differences in occupational violence and aggression across the context of practice, the authors argued that a one-size-fits-all approach is unlikely to be effective. In Lyneham's (2000) study, around 50% of Australian rural and remote nurses reported that their emergency department had a policy relating to occupational violence and aggression. However, (global) satisfaction with administrative responses to occupational violence and aggression varied.
Existing studies fail to elaborate on the content or shortcomings of available occupational violence and aggression policies. Studies of policy effectiveness in rural health settings are needed to inform occupational violence and aggression policy development. These findings are consistent with a recent call for targeted, effectively operationalised legislation, policies and penalties for occupational violence and aggression in non-metropolitan settings (Hills et al., 2021).
Expert consultancy and peer-review mechanisms for occupational violence and aggression policies, including consultation with best practice forums in other contexts (e.g. mental health and forensics), may advance work in this area.

| Interventions and recommendations
Pushing the training agenda nurses servicing a multi-cultural, rural/remote community identified five broad practices that nurses used to avert or reduce occupational violence and aggression: being safe, available, respectful, supportive and responsive (see Table 1).
With regard to intervention effectiveness, Hills (2008) found that participation in aggression management training in the previous 5 years showed a weak, negative association with the proportion of verbal abuse and a weak, positive association with self-efficacy.
Not surprisingly, 70% of participants reported self-efficacy for dealing with occupational violence and aggression in the low-tomedium range. In the Beattie, Griffiths, et al. (2018a) and Beattie, Innes, et al. (2018b) studies, Australian healthcare workers (urban and regional/remote aggregated) recommended interventions for occupational violence and aggression such as trauma-informed care and identified emotional intelligence (resilience and mindfulness) and de-escalation as important training content areas, as opposed to the current focus on managing extreme physical violence, for example, restraint. However, Beattie, Griffiths, et al. (2018a) and Beattie, Innes, et al. (2018b) subsequent studies found that trauma-informed care practices were not necessarily embedded in education, policies and procedures. Furthermore, in health services that were implementing these practices, data on effectiveness were not available (metropolitan and regional/remote data were aggregated). These

Immediate and ongoing staff support and safety
Australian remote area nurses in Fisher et al.'s (1996) study reported that there was little or no post-incident support for occupational violence and aggression apart from temporary leave. This may reduce confidence in reporting future incidents; only half of the respondents always officially reported incidents. In Lyneham's (2000) study, incident reporting was even lower, and this was attributed to beliefs that violent incidents were 'part of the job' and 'nothing would be done' by the health service administration. Indeed, more than half of all Australian respondents (rural, remote and metropolitan) in this study reported that they had never received support following serious incidents, with several respondents describing the administration as punitive and blaming staff. Mayhew and Chappell (2003)  design (e.g., staffing levels) are identified as priorities to ensure staff safety. The review also highlighted the importance of an organisational culture that acknowledges and supports staff affected by occupational violence and aggression and a community culture that facilitates collaboration between rural healthcare workers and public sector organisations such as the police.

| S TRENG TH S AND LIMITATI ON S
This systematic review includes several strengths: pre-registration, adherence to PRISMA guidelines, clear objectives and narrowly focused research questions and comprehensive database searches with no date restrictions, to allow for the potential inclusion of a broad range of studies. In addition, we conducted a rigorous quality assessment, with a reliable tool, and achieved moderate-to-good inter-rater agreement for quality scores. Finally, we developed and implemented a new index to assess the certainty of the body of evidence, with excellent inter-rater reliability, thus supporting an objective summary of the literature.
While our review aimed to cover international literature on occupational violence and aggression in rural health service urgent care facilities, we limited our search to literature in the English language due to a lack of resources for translation. Consequently, we might have missed relevant international literature published in other languages. All studies but one was conducted in the Australian context.
Additionally, quantitative analysis of the results was not possible due to the heterogeneity of the samples, research methods and data analytic approaches in the included studies. A further limitation was that only some studies reported disaggregated data for the population of interest, which reduces the generalisability of the findings to the target population of rural healthcare workers in urgent and critical care.
The systematic review is also limited by the quality of the included studies, with fewer than 50% of the studies being of high quality. The most common study limitations included sample representativeness relative to the target population (quantitative); absence of clear links among data sources, collection and analysis and interpretation (qualitative) and insufficient explanation of divergences/inconsistencies between quantitative and qualitative results and overall quality, based on adherence to each tradition (mixed methods studies).
With regard to certainty, most studies performed well on relevance to the research questions, but the relevance of the sample was often compromised (see above

| CON CLUS ION
Based on the findings of the review, research recommendations include systematic identification and evaluation of occupational violence and aggression policies and education and training interventions in rural health services settings. Such information is not currently available in the literature.
Clinical recommendations concern balancing security upgrades with trauma-informed care for patients in distress. Consistent changes to the physical work environment and job/task design in rural health services should be made as a priority to ensure staff safety.
Policy recommendations include organisational policies that support an anti-violence culture by encouraging incident reporting, debriefing (by independent and suitably qualified individuals) and ongoing post-incident support to promote staff well-being. The literature reinforces the importance of collaboration between rural health services and police and community organisations to support healthcare workers' responses to occupational violence and aggression.

AUTH O R CO NTR I B UTI O N S
Sharon Grant: 45% set the parameters for the review, completed article screening, completed data extraction, completed quality appraisal and certainty ratings and manuscript writing (lead). Stephanie Hartanto: 45% set the parameters for the review, completed PROSPERO registration, completed database searches and upload to Covidence, completed article screening, completed quality appraisal and certainty ratings, manuscript writing and formatting to template.
Diane Sivasubramaniam: 5% Assisted with screening, that is, resolving disagreements with article inclusion/exclusion and quality appraisal ratings and manuscript editing. Kaye Heritage: 5% Assisted with setting the parameters for the review and provided feedback on data extraction and thematic analysis and manuscript editing.

ACK N OWLED G EM ENT
There are no acknowledgements. Open access publishing facili- Network.

CO N FLI C T O F I NTE R E S T
On behalf of all authors, the corresponding author states that there is no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sets not applicable to this article as no data sets were generated or analysed during the current study.

PROS PE RO LI N K
h t t p: //w w w. c r d .yo r k . a c . u k /P RO S P ERO/d i s p l ay_ r e c o r d .