Interventions to reduce bullying in health care organizations: A scoping review

Identification and development of the research questions

The three research questions are as follows: What types of interventions addressing workplace bullying have been tested and evaluated? What are the outcomes? How are the outcomes measured and the intervention evaluated?

Location, screening, and selection of relevant publications

An information scientist guided preliminary searches to determine tentative terms and to establish the project’s initial universe and size, identify relevant databases and appropriate search strategies/terms, and advise on data management tools. Our preliminary searches revealed terms such as harassment and mobbing were inappropriate for the purposes of a review of interventions and consequently were discarded. The term “harassment” can be a single act or event of mistreatment, whereas bullying implies repeated mistreatment over a period of time. As such, bullying and harassment require different institutional responses and interventions. “Mobbing” was found much more commonly in studies emanating from European countries and as such was not used to direct the focus to literature emanating from non-European countries, in accordance with the focus of the scoping review on the latter set of countries in general, and the country of the author’s program of research in particular, where the regulatory framework has been much slower to develop. However, these search terms would serve well for a systematic review of the literature, which often is done following a scoping review.

The following terms were used to search six electronic databases of published scientific literature: “health care and bullying,” “health professions and bullying,” and “workplace bullying and health care.”

Our search capitalized on the databases’ search engines, permitting free-text searching and controlled vocabulary subject-descriptor-identifier searching. Besides standard Boolean/logical operators, most of the databases support “proximity” or “adjacency” searching to increase the range of retrievals. Other common features included the ability to: (1) search by specific elements, such as document type, language, and year; (2) “limit” or “separate-out” by “types” of materials, e.g. “peer-reviewed” versus “non–peer-reviewed,” books, chapters-within-books, book reviews, conference proceedings; and (3) group search results by “time”—in some cases, with visual maps/charts—to peruse search results based upon the “evolution” of key search terms or concepts.

The review process (Figure 1) began with 508 sources, a combination of sources obtained from the databases, filtered for language (English), and other published literature found by bibliographic searches from key articles, or recommended by team members and other key informants. Refworks, an online research management, writing and collaboration tool, was used for storage of PDF copies of selected sources (see Appendix). A progressive series of three filters was then applied and we were left with 14 sources in total to review. OPEN IN VIEWER

Charting and organizing results

The remaining 14 sources were analyzed by three team members. To organize our analysis, a data chart was developed with the following dimensions: context of the study, purpose and characteristics of the intervention, characteristics of the evaluation research design, and outcomes of the intervention. The sources were discussed at a team meeting. Guided by the research questions and our data charts, six sources were excluded (two articles, three dissertations, and one book) as they did not fit the criteria of an evaluated intervention directed to, or including bullying, and one source was found to deal with patient, rather than coworker assaults.

Intervention outcomes

Two of the reviewed articles report unexpected outcomes (see Table 2). First, in Bourbonnais et al.’s ⁴⁵ quasi-experimental study, three findings were contrary to the hypotheses: psychological distress did not abate and both supervisor support and decision latitude decreased in the experimental group. The explanation offered by the authors for the unanticipated findings is the short time between the pre- and postintervention measurements. Most of the 56 interventions were not yet implemented by the postintervention data capture. Further, it was noted that the expectations of the interventions were especially high among caregivers and thus the surveys could be registering their dismay with their perceived limitations in decision latitude and supervisor support.

Second, Pate and Beaumont’s ⁴² longitudinal study with 200 public sector employees also produced an unanticipated result. Although the intervention realized a decrease in bullying, the level of trust in senior management also decreased over the three years. The percentage of employees who “agreed” or “strongly agreed” to the 5-point Likert scale survey item, “I feel bullying is a problem within the organization” dropped from 50% to 22%. Yet, contrary to expectations, those reporting “poor” or “very poor” levels of trust in senior management increased from 25% to 32% between the pre- and postintervention measures.

The authors advance several explanations for the deviant result. First, congruent with Bourbonnais et al.’s ⁴⁵ explanation for their unexpected findings, Pate and Beaumont speculate that three years is insufficient for the effects of the intervention to be noticed by employees. Second, the available measures and models to examine complex, multi-dimensional social phenomena such as trust and bullying are under-developed. While the measure used to assess bullying had the advantage of capturing perceptions of the pervasiveness of the problem rather than individual experiences of bullying, as a single survey item, it nonetheless obscures distinctive dimensions of the phenomenon. A similar limitation holds for the single item measuring trust in senior management. As the authors argue, employees might consider the bullying intervention to be management’s “one hit wonder”: successful yes, but would require concurrent and similarly successful initiatives to raise their level of trust in management.

The most promising outcomes were from the participatory interventions, as discussed in subsequent paragraphs, illustrated in Figure 2, and detailed in Table 2. Intervention teams consisting of representatives from all units and levels of the workplace express the epistemological commitment of bringing multiple perspectives to the identified problem and subsequent intervention development and implementation. In Bourbonnais et al.’s ⁴⁵ study, an intervention team was established through voluntary recruitment of direct care providers (nurses, orderlies, and auxiliary nurses) and included representatives from the human resources department, unions, and nursing management. The team’s procedures were drawn from German health circles, which have otherwise reported stress-prevention consequences. The intervention team was given latitude to arrive at specific interventions; members were granted release time for team meetings and follow-up sessions in each of their respective work units to disseminate and gather comments from their coworkers. The measurable outcomes from Bourbonnais et al., ⁴⁵ were (1) a significant drop in psychological demands and in the effort-reward ratio in the experimental and not the control group and (b) more favorable postintervention scores on all of the psychosocial factors in the experimental, compared to the control group. OPEN IN VIEWER

Collette’s ⁴⁶ study with registered nurses (RNs) in 17 Australian hospitals was another participatory intervention. The “best practices team” consisted of 14 nominated RNs and management representatives. Bullying was not an explicit focus of the intervention, so the outcomes are expressed in terms of recruitment and workplace culture. RNs at “high” risk of leaving the workplace and the profession dropped from 40% to 28% with similar rates of increase in workplace culture indicated by high levels of trust in management, effective communication, and a desirable place to work. Further, the initiative of the best practices team to implement two “bullying advocates” was favorably reported on.

Jury et al.’s ⁴⁷ reported outcomes are equally impressive. Their large-scale, multi-site intervention granted autonomy to each district and division within the large 68,000-employee organization of Queensland Health to develop and implement their own action plans in response to the workplace culture survey results. Some of the resulting action plans included educating staff about harmful behaviors, increasing staff in decision-making through focus groups and forums, and improving communication processes. At the end of the three-year intervention, all 11 measures of organizational climate, six measures of supervisor support, and three individual psychological health measures showed improvement. Along with developing the commitment amongst the organization’s senior leadership, the authors credit the success of the intervention to involving the staff. “Ensuring staff are provided the opportunity to hear their results and contribute to strategies designed to improve them is a key to success when making a commitment to improving the workplace culture for 68,000 employees” (p. 376).

Meloni and Austin’s ⁴³ intervention, the fourth intervention that we class as participatory, was launched by forming a “Working Group” of employees nominated by their coworkers to “guide the rollout of the chosen strategies” (p. 93), although the details of how those strategies were chosen and by whom is not provided. Following the three-year intervention, all measures of bullying showed improvement: more employees were aware of the available support mechanisms, knew how to report bullying, and trusted that bullying would be appropriately handled if reported. The informal complaints of bullying increased as the authors’ expected because of the greater levels of general awareness of bullying among the employee groups. Similar to Collette ⁴⁶ and Bourbonnais et al., ⁴⁵ bullying was not the sole focus of the intervention or the evaluation. The results of the employee engagement dimension of the evaluation are in line with the above evidence of the intervention’s success: the percentage of employees who are openly positive, optimistic, and engaged about the organization’s future, ready for change, and to trust management rose from 28% to 37%. ⁴³

In contrast to the above-discussed four intervention studies, the nonparticipatory interventions had less striking outcomes. First is Stevens’ case study, initiated by nursing managers in a large Australian hospital. The strongest outcome of the top-down intervention was the marginal decrease in turnover of nursing staff from 28% to 22% within the first year and stayed constant in the three subsequent years. Second is Pate and Beaumont’s intervention with UK’s public sector employees, which focused on employees’ trust in senior management. The bullying intervention was primarily undertaken to restore employees’ confidence in managerial initiatives. Its two-prong strategy consisted of the dismissal of a number of “worst offending” senior employees following a top-down identification of the “offenders,” and implementation of a zero-tolerance bullying policy with compulsory training for all employees. As reported above, over the three-year period, the trust in senior management decreased, rather than to increase.

We resist unilaterally asserting the conclusion of superior outcomes from the participatory interventions due to the brevity and vague reporting in some of the articles. For instance, the only reported outcome from Crawford’s intervention, in which bullied employees were given the opportunity for confidential individualized consultations with the author, himself a psychotherapist, is that all who attended exhibited obvious signs of distress. Likewise, the limited reporting of Chima’s study furnishes no evidence substantiating the claim that the champion role was “positively evaluated by individuals, management, and organizational levels” and the champions themselves were “empowered, confident, and valued by staff across the organization” (p. 65).

Conclusion and recommendations for future research

In this review, we found many authors remarking on the paucity of workplace bullying interventions with published results. Finding only eight articles that meet the scoping criteria of tested, evaluated interventions in health care workplaces, the results of our scoping review substantiate their claims. We offer several plausible explanations for the scarcity. First, organizations are initiating their own research rather than turning to experts and academics to conduct analyses, precipitated by the problems of bullying, dwindling retention rates, and low employee morale coming to the attention of senior management. Another motivation for organizations to take up their own research is the burgeoning legislative changes related to workplace bullying. In addition to establishing complaint-based resolution mechanisms, many of these legislative instruments place responsibility on employers to maintain, as can be reasonably expected, a bullying-free work environment. As a result, human resource departments, unions, and professional associations have been engaged in creating and delivering staff training programs, expanding the scope of existing resolution structures, such as joint management-labor Workplace Occupational Health and Safety Committees, and implementing organizational level anti-bullying policies. Although not legislatively required, some workplaces then take note of the effects of these initiatives with “soft” measures and somewhat perfunctory assessment procedures. Significantly, the only Canadian intervention in our reviewed set of eight articles was set in Quebec, which pioneered the legislative protections for targets of workplace bullying by amending the Labour Standards Act in 2004. Legislative changes in other jurisdictions are similarly recent with the exception of Sweden, the first EU country to enact anti-bullying legislation in 1993. The Australian Model Work Health and Safety Act was finalized in 2011 and is yet to be implemented in all jurisdictions. In the UK, the Protection from Harassment Act was passed in 1997, which offers legal redress for workplace bullying. In the United States, the Healthy Workplace Bill has been introduced in some state legislatures but not yet enacted. In light of how recent legislative changes are, it is not surprising that all eight articles were published since 2000.

A second explanation for the scarce literature is the limited valid and reliable measures capturing the effects of interventions. Relatedly, organizational research is problematic. Workplaces are complex, dynamic environments not well suited to traditional experimental research designs, arguably the gold standard for social scientific research.^50–52 Abstract knowledge, in its creation and application in organizational contexts, has been found to be made “sticky” by workplace climate, imperatives, and structures.^53–55 “Intervention science” takes place within the constraints of organizational life and rigorous, academic evaluations are not workplace priorities. ⁵⁶ Indeed, they can interfere with the day-to-day operations directed to the organizational mandates such as profit generation or service provision. Notably, only one of the eight reviewed articles used a quasi-experimental design. Perhaps the incompatibility of the workplace setting and the rigors of experimental designs accounts for the short time frame of the Bourbonnais et al.’s study. Indeed, the one year of elapsed time between pre- and postmeasures was the shortest of the reported timeframes: others were 15 months, ⁴⁶ three years,^42,43 and four years. ⁴⁸ The authors in their companion article ⁴⁵ reveal that the battery of interventions was prematurely assessed after a year because some of the 56 specific interventions developed by the intervention team took longer to implement.

Additional hurdles to using experimental designs in organizational research are the thorny ethical questions, which arise when an intervention is thought to improve the workplace for employees. As Flannery ⁵⁷ argues, it is difficult to justify using control groups, who by definition are denied the interventions, when, in fact they could mitigate acute organizational dysfunction, individuals’ negative health outcomes, and all their attendant repercussions.

Although the above-delineated limitations to organizational research are appreciable, they are no reason for a moratorium on intervention evaluation research. ⁵⁶ The impediments to conducting organizational research, however, are reason to reassess our standards for evaluating the outcomes of the research. Perhaps the gold standard of experimental designs should not be unwaveringly applied—a logic that guided the authors in the semantic filtering for this review. The standards regarding reporting of interventions and their results, on the other hand, do need to be firmed. Incremental advances in interventions and their evaluations necessitate detailed disclosure of intervention contexts, timeframes (duration, intensity, and frequency), type of data collected, analytical methods, and changes in organizational context coinciding with intervention (e.g. expansion of services, hospital closure). As noted above, details of the interventions and their evaluations in the reviewed studies were sparse.

Bullying in the health care sector, if left unchecked, will inevitably continue to exacerbate the existing recruitment and retention problems, which in turn have long-term economic consequences for health care systems. The published studies in this scoping review, albeit limited in number and details, are evidence of the value of the research to direct both future interventions and their evaluation procedures. The results of this review make a compelling case for interventions based on participatory principles and including employees from all levels of the organization in the co-creation of intervention goals and implementation as well as their evaluation strategies. Increasingly, legislative instruments, collective agreements, and other regulatory mechanisms are being used to address bullying in health care workplaces. To be fully operative, the decision-makers involved in the associated policy development, training programs, and intervention implementation need to be supported with relevant information. The scoping review is one step toward meeting these knowledge needs.