Study	Reason for exclusion
Barrett 2009	Did not include outcome measures as specified in our PICOS. This study examined the effect of a targeted team‐building intervention (organisation/employer level) that was aimed at improving group cohesion, turnover and nurse satisfaction in an acute care teaching hospital in the United States of America (US). It was a quasi‐experimental pre‐post intervention design without a control group. There was no matching of participants pre‐ and post‐test and each unit in which participants were located had its own individual dynamics and issues that needed to be addressed.The study outcomes did not include a change in the number of reported cases of bullying or level of absenteeism.
Beirne 2013	Study design not as specified in our PICOS. A qualitative case study to compare two anti‐bullying initiatives (organisation/employer level); one in the public and one in the private sector in the United Kingdom (UK).They highlighted the complexity of bullying in the workplace and called for a more grounded approach to engage with the specific workforce. Not a control study.
Bortoluzzi 2014	Study design not as specified in our PICOS. This study examined the predictors of bullying (individual/job interface level) in an observational study among nurses in public hospital corporations in northern Italy. It showed that leadership style explained 33.5% of the variance in the onset of bullying: this is useful, but no intervention was tested.
Bourbonnais 2006a	Did not include an intervention as specified in our PICOS. This study tested a participative intervention (job/task level; see Bourbonnais 2006b for full details of intervention) to prevent workplace‐related mental health problems among 'care providing personnel' in two hospitals in Quebec, Canada. Whilst it was effective in that regard, their focus did not extend to prevention of bullying per se.This is a psychosocial intervention, not focused on bullying.
Brunges 2014	Did not include an intervention as specified in our PICOS. This study from the US takes a long‐term approach consisting of several interventions (organisation/employer level) and although some interesting effects were seen on workplace engagement and job satisfaction, their study lacked precision and did not focus on bullying prevention.The improvements/interventions are spread over long periods and the 'results' are diffuse, and due to the prolonged timeframe, it was not possible to control a number of variables.
Ceravolo 2012	Did not include an intervention as specified in our PICOS. This was a pre‐ and post‐intervention survey of registered nurses’ perception of lateral violence and turnover in the workplace (organisation or employer level). Improvements were noted following workshops designed to enhance assertive communication skills, raise awareness of the impact of lateral violence behaviour, and develop healthy conflict resolution skills. No control group was used.
Chipps 2012	This was a pilot study described as a 'quasi‐experimental pre‐test and post‐test comparison' of an educational programme (individual/job interface level), with 16 participants.The group acted as their own control.
Collette 2004	Study design not specified in our PICOS. This was a case study, examining a team‐based approach to the retention of nursing staff (organisation/employer level) in a hospital in East Melbourne, Australia. This study only had an indirect impact on bullying and there was no control group.
Cooper‐Thomas 2013	Study design not specified in our PICOS. This was a survey of a convenience sample of 727 employees from nine healthcare organisations in New Zealand, which focused on the potential buffering effects of perceived organisational support, and organisational anti‐bullying initiatives (organisation/employer level).
Crawford 1999	Did not include outcome measures specified in our PICOS. Reports on two organisational interventions in two organisations in the UK aimed at preventing bullying in the workplace. The first intervention was the implementation of the Dignity at Work Policy and procedures in an organisation where bullying had been identified as an issue (society/policy level). The outcomes from the policy implementation were not clear. The second organisational intervention briefly described was the response of an organisation to the systematic bullying of staff by a manager (individual/job interface). It was reported that the bully left the organisation but the reason was not stated. There was insufficient detail about the intervention and lack of data from which evidence of effectiveness of either intervention could be determined.The study outcomes did not include a change in the number of reported cases of bullying or level of absenteeism.
Egues 2014	Did not include an intervention as specified in our PICOS. This study from the US provides weak evidence that education workshops have an effect on knowledge of student nurses. However, it is not prevention in a workplace setting (unclassified level of intervention).
Feda 2010	Did not include an intervention as specified in our PICOS. A case control design was used, in educational workplace settings in the US, to analyse nine different written violence policies and their impact on work‐related physical assault (unclassified level of intervention).
Gedro 2013	Study design not specified in our PICOS. This is a case study which was focused on workplace incivility from the US. It mainly includes a description of the workshops and feedback from participants (organisation/employer level).
Gilbert 2013	Study design not specified in our PICOS. A survey of 238 students from a business school in the US, which sought to understand the complexities of workplace bullying by exploring the use of a bullying policy as a means of mitigation, particularly in relation to gender norms (society/policy level).
Grenyer 2004	Did not include outcome measures as specified in our PICOS. Reports on a pilot of an aggression minimisation programme for all public health staff who were at risk in New South Wales (Individual/job interface level). It involved twenty‐two hours of training divided into four modules. Two pilot samples were evaluated and the outcomes focused on the perceived confidence of staff in dealing with incidents of aggression and not on the outcomes of relevance to this review.
Griffin 2004	Did not include an intervention as specified in our PICOS. An exploratory design from the US with an applied intervention of ‘cognitive rehearsal techniques', which staff were encouraged to use as a shield against incidences of lateral violence (Individual/job interface level). There was no control nor any pre‐ or post‐test measures. The intervention was focused on 'how to respond' if bullied. Hence, it was considered to be a management of bullying intervention rather than prevention of bullying.
Holme 2006	Study design not specified in our PICOS. This paper reports on a consultancy project from the UK where managers in a company of 900 staff were trained to implement a new harassment and bullying policy (society/policy level), through involvement in work‐based projects. This was a case study with no control.
Karakas 2015	Study design not specified in our PICOS This study was a non‐controlled before and after study from Turkey, which focused on assertiveness training for nurses who had scored 204 points or more on a mobbing instrument which 'demonstrated that they had experienced mobbing'.There was no control (Individual and/ job interface level).
Lasater 2015	Did not include outcome measures as specified in our PICOS. This was an interrupted time series study from the US, which focused on a three‐part educational intervention (organisation/employer level), addressing incivility in the workplace.
Latham 2008	Did not include an intervention as specified in our PICOS. This study from the US was based on a description of the impact of a mentor and advocacy programme on the broader context of a healthcare workforce environment (organisation/employer level). The outcomes were measured through a survey, with the focus on perceptions of the impact of the programme on the environment in which the registered nurses worked and not specifically on bullying.The intervention was not focused on bullying at work.
Longo 2011	Study design not specified in our PICOS. This was a programme evaluation of a healthcare workforce partnership community collaboration from the US, aimed at nursing retention (society/policy level). It involved a range of initiatives which culminated in a train the trainer conference. There was no control group.
Léon‐Pérez 2012	Did not include an intervention as specified in our PICOS. This was a two‐wave prospective intervention study in a Spanish manufacturing corporation, which focused on conflict management training of 42 employees, not on prevention (organisation/employer level). It did not employ a control group.
Mallette 2011	Did not include outcome measures as specified in our PICOS. An experimental educational intervention using a pre/post design with a control group from Ontario, Canada. The intervention was computer‐based learning, using avatars in scenarios to address horizontal violence (individual/job interface level). The study outcomes did not include change in the number of reported cases of bullying or level of absenteeism.
Meloni 2011	Did not include outcome measures as specified in our PICOS. A case study approach to the implementation and evaluation of a zero tolerance of bullying and harassment programme (organisation/employer level) in one hospital in Australia. There was no control, and outcomes were based on employee satisfaction surveys.
Melwani 2011	Did not include an intervention as specified in our PICOS. This study focused on three experiments that tested the outcomes of being a recipient of contempt in the work domain (individual/job interface level) at a university in the US. Contempt is a possible component of bullying, but the study did not focus on prevention.
Mikkelsen 2011	Did not include outcome measures as specified in our PICOS. This Danish study used a quasi‐experimental approach to evaluate interventions in two organisations (organisation/employer level). The Interventions were largely educational in nature, including directed teaching sessions, meetings, and paper‐based information. The results were broadly qualitative and there were no control groups. Source of funding: Danish Work Environment Research Fund and The National Research Centre for the Working Environment.
Nikstatis 2014	Did not include outcome measures as specified in our PICOS. This was a before‐and‐after design from the US, with 38 participants, testing an educational intervention on the causes and effects of incivility, using case studies and discussion of team building skills and ways to prevent incivility (job/task level). The study did not employ a control group.
Oostrom 2008	Did not include an intervention or outcome measures as specified in our PICOS. This was an evaluation of an aggression management training programme from The Netherlands (Individual/job interface level). Using an alternative approach to a control group, the authors of the study referred to as an internal referencing strategy, which they considered 'ruled out some major threats to internal validity without the need for a control group'. The intervention dealt with the management of aggression rather than prevention of bullying at work. The study outcomes did not include change in the number of reported cases of bullying or level of absenteeism. The intervention was not focused on bullying at work.
Pate 2010	Did not include outcome measures as specified in our PICOS. This was a longitudinal study, which produced limited data on perceptions of bullying in a single organisation in the UK, following the implementation of bullying and harassment policies (organisation/employer level). It clearly indicated how leadership by a CEO can effect a perception of positive change in an organisation, but pointed to the difficulty of measuring the success of workplace bullying policies. The study did not employ a control group.
Probst 2008	Did not include outcome measures as specified in our PICOS. The authors reported on initial outcomes that appeared to improve employees' knowledge and understanding of the interrelated job associated problems (society/policy level). The International Labour Organisation multilevel longitudinal intervention (SOLVE) focused on the reduction of psychosocial problems in the workplace; stress, tobacco, alcohol, and drugs, HIV/AIDS and violence. However, the data did not allow for a comprehensive evaluation of SOLVE, but were limited to giving an indication of how employees had gained knowledge. The intervention was not focused on bullying at work.
Stagg 2011	Did not include an intervention as specified in our PICOS. This study utilised an intervention designed by Griffin 2004. While this study from the US was aimed at determining whether cognitively rehearsed responses to common bullying behaviours decreased bullying, we judged that it did not focus on prevention but rather on how to increase staff nurses' knowledge of workplace bullying management (Individual/job interface level).
Stevens 2002	Study design not specified in our PICOS. This was a case study within a broad review of the workplace, conducted in a large Australian teaching hospital (organisation/employer level). No research was involved.
Strandmark 2014	Study design not specified in our PICOS. This was a Swedish study, which employed a community‐based, participatory research approach (society/policy level), which aimed to achieve zero tolerance for bullying.
Wagner 2012	Study design and outcome measures not as specified in our PICOS. This was a post‐hoc analysis, with 339 participants in the US, who undertook training in new norms of workplace culture to prevent and resolve incidents of workplace violence (organisation/employer level). The study did not include measures of effectiveness or outcome measures; it was not a before‐after design, nor did it have a control group.
Woodrow 2014	Study design not specified in our PICOS. This was a case study from the UK, designed to explore the policies and procedures in place to prevent bullying, and to examine the extent and quality of local implementation of bullying policies (organisation/employer level). No comparative research was involved.