‘What effect do safety culture interventions have on health care workers in hospital settings?’ A systematic review of the international literature

Introduction: Interventions designed to improve safety culture in hospitals foster organisational environments that prevent patient safety events and support organisational and staff learning when events do occur. A safety culture supports the required health workforce behaviours and norms that enable safe patient care, and the well-being of patients and staff. The impact of safety culture interventions on staff perceptions of safety culture and patient outcomes has been established. To-date, however, there is no common understanding of what staff outcomes are associated with interventions to improve safety culture and what staff outcomes should be measured. Objectives: The study seeks to examine the effect of safety culture interventions on staff in hospital settings, globally. Methods and Analysis: A mixed methods systematic review will be conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Searches will be conducted using the electronic databases of MEDLINE, EMBASE, CINAHL, Health Business Elite, and Scopus. Returns will be screened in Covidence according to inclusion and exclusion criteria. The mixed-methods appraisal tool (MMAT) will be used as a quality assessment tool. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials and non-randomised studies of interventions will be employed to verify bias. Synthesis will follow the Joanna Briggs Institute methodological guidance for mixed methods reviews, which recommends a convergent approach to synthesis and integration. Discussion: This systematic review will contribute to the international evidence on how interventions to improve safety culture may support staff outcomes and how such interventions may be appropriately designed and implemented.


Introduction
In 2000, the seminal report by the Institute of Medicine, To Err is Human, estimated over 90,000 patient deaths from medical error related to both system failure and poor organisational culture 1 . Since that time the incidence of patient safety events have remained high globally, due in part to the increasing complexity of healthcare and multi-morbidities 2,3 . To Err is Human recommended the concept of a patient safety culture to support the health workforce to provide safer care 1, 4 . A major focus is on the reduction and mitigation of patient safety events, building a culture of reporting and learning from events, and supporting those affected by harm. Patients are considered first victims of patient safety events, while members of the workforce may experience a second victim impact, including physical and psychological distress 5,6 . How organisations respond to and learn from safety events affecting patients, families and staff is a marker of their patient safety culture and impacts the future safety and wellbeing of all concerned.
A safety culture is about shared organisational beliefs, values, norms and procedures for patient safety 7,8 . Safety climate, on the other hand, relates to a group or team perception of safety culture in organisations 9 . Rather than a specific process or technology, interventions to promote a culture of safety are designed to improve organisational conditions, enhancing leadership and teamwork among health care providers to support safe patient care [10][11][12] . Safety culture interventions which target staff behaviour include executive walk-rounds or interdisciplinary rounds; multicomponent unit-based interventions; team training or communication initiatives 11 ; frameworks for assessing, reporting and improving patient safety concerns; and standalone courses 13, 14 . Patient safety educational interventions have also targeted the patient safety skills, attitudes and knowledge of healthcare workers to support safe patient care 15 , with safety culture likely moderating their impact 16 . While there are over 80 definitions of the terms "safety culture", "patient safety culture" and "safety climate" across the literature 17 , core distinctions are set out in Table 1.
The theoretical construct of "safety culture" is contested in the literature 18 . One perspective posits that safety culture exists as an empirical measurable entity, as a product of values, beliefs, attitudes and behaviours, to be manipulated by intervening in the healthcare system 4 . A contrasting perspective views the concept of safety culture more usefully deployed as a conceptual lens to inform analysis of the healthcare system 18 . In the latter, safety behaviours are a reflection of a generic organisational culture, emerging over time through other multiple internal and external system dynamics 19 . The dearth of robust, definitive explanations for changes in safety culture in the literature, particularly those that demonstrate little connection between safety culture interventions and reduced adverse events, is seen as supporting the conceptual approach 19 .
The empirical assessment of safety culture is also diverse, with multiple tools based on unique constructs of safety culture 17 . Furthermore, there is increased recognition of the need for triangulated data sources to assess the safety of healthcare organisations. Data on safety culture alone is not always considered sufficient to explain changes in safety behaviours and culture 19,20 . Additional data sources such as routine hospital data on adverse events, and multiple types of data such as quantitative and qualitative are needed to better inform analysis 19 . Recently, Churrucca et al. constructed common over-arching themes to synthesise domains of safety culture across different assessment tools 17 . These common themes enable comparison between various data collection instruments 17 .
Reflecting the above debates, evidence on the effectiveness of interventions to improve safety culture is weak, but suggests they contribute to clinical care processes and to improved clinician and staff perceptions of safety culture 11 . A safety culture can support staff to deliver effective and timely care to patients, increase staff competencies to collectively learn from patient safety events when they occur 21 , and can strengthen organisational support to healthcare teams to recover from patient safety events 22-26 . Despite a focus on patient safety within measures of safety culture, there is inconsistent Table 1. Definitions and terms.

Patient safety culture
One aspect of an organization's culture. It can be personified by shared values, beliefs, norms, and procedures related to patient safety among members of an organization, unit, or team. It influences clinician and staff behaviours, attitudes, and cognitions through cues about the relative priority of patient safety compared to other goals such as throughput or efficiency 11 .

Safety culture
The product of individual and group values, attitudes, perceptions, competencies, and patterns of behaviour that determine the commitment to, and the style and proficiency of, an organisation's health and safety management 4 .

Safety climate
Group or team perceptions of safety culture in organisations. These are measurable aspects of safety culture, in contrast to domains of culture such as behaviour and values 9 .

Amendments from Version 1
A short summary of our updates is: an enhanced review of the deployment of safety culture as a concept, and the differing perspectives therein; an incorporation of the new research question recommended; a clarification on the specific focus and limitation of the research; clarification on the treatment of varying measurement instruments in the review; and an adjustment to the reference in the Lancet, as observed by both reviewers.
Any further responses from the reviewers can be found at the end of the article REVISED evidence for the link between interventions seeking to enhance a safety culture and patient outcomes such as hospital readmission rates, length of stay, pressure ulcers and falls, or ventilator associated pneumonia 15,[27][28][29][30][31] . The knowledge base on safety culture and patient outcomes is mixed, and a safety culture can moderate the effects of patient safety interventions on patients.
Measures of safety culture focus on staff attitudes and behaviours related to the prioritisation of patient safety 4,32,33 . In hospitals, evaluations of safety culture typically measure workforce related domains such as teamwork, communication, or information exchange, safety culture outcomes such as reporting rates, and patient outcomes such as falls, length of stay, or readmission rates 11,27 . A minority of measures include domains related to staff outcomes such as staff safety behaviours, staff injury rates, or well-being following a patient safety event 34 . Interventions to enhance safety culture are generally targeted at enhancing the norms and behaviours of hospital staff, but without an associated focus or exploration of the staff outcomes associated with safety culture. Evidence suggests that a safety culture can alleviate stress and burnout for staff 35-37 , support recovery and learning from events, and safeguard against repeat events 38-41 by supporting staff to safely speak out about what is not working 34,42 . There is a now a need for a review of the impact of safety culture interventions on staff, to generate a common understanding of what staff outcomes are associated with interventions to improve safety culture and what staff outcomes should be measured. This research seeks to isolate staff outcomes, thereby examining the impact of safety culture interventions on staff where staff effects are claimed to take place.
Safety culture measurement tools focus on generating data to improve the safety of care provided to patients, and they do so by measuring staff perceptions of safety culture 43 , including staff attitudes, behaviours, and norms 17 . Depending on the instruments selected, staff outcomes and experiences may be captured through the deployment of safety culture measurement tools alone or in combination with other data sources. Prior research suggests a bi-directional relationship between safety culture and staff experience. The strongest evidence suggests that a safety culture influences certain staff outcomes such as staff behaviour to improve safety, event reporting, staff turnover, and injury rates among staff 34,44 . This literature reviews seeks to isolate staff outcomes as the point of interest including, but not limited to, burnout, stress, well-being, turnover or absenteeism 36,45 . In doing this, the review recognises studies that claim to manipulate safety culture.
Hospital staff experiences of and outcomes from interventions to improve safety culture are rarely examined in their own right. There is a knowledge gap on how safety culture affects staff in hospital settings. This systematic review seeks to address the knowledge gap and generate a conceptual understanding of associated staff outcomes based on the available published evidence, evaluating the state of evidence connecting safety culture to staff outcomes. Understanding these issues is important: as Shaw et al. have stated, 'there is no patient safety without health worker safety' 46 .

Systematic review aims and objectives
A systematic review of the literature will examine the state of evidence connecting interventions to improve safety culture to hospital staff outcomes. The research questions are: 1) How is safety culture defined in studies with interventions that aim to enhance it?
2) What effects do interventions to improve safety culture have on hospital staff?
3) What intervention features, safety culture domains or other factors explain these effects?

Design
This study is a mixed methods systematic review 47,48 of published literature guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines 49 .

Search strategy
Searches will be conducted using the electronic databases of MEDLINE, EMBASE, CINAHL, Health Business Elite, and Scopus. The search strategy will include the following terms: patient safety culture, hospital workforce and management, hospital setting, and high-, middle-and low-income countries.
The primary search will begin in PUBMED with major keywords listed on the Medline Medical Subject Headings (MeSH) terms, such as "patient safety," "safety climate," and "hospital workforce". Search terms will be used in conjunction with Boolean Operators "AND" and "OR." Supplementary Files will provide details of the full search strategy. Reference lists of all articles included for full-text screening will also be searched. A sample of the search run through Medline is provided in Table 2.

Inclusion and exclusion criteria
Articles that report research involving both clinical and non-clinical members of the hospital workforce, and hospital management, will be included. Articles will be included if they are (a) quantitative, qualitative or mixed methods studies that evaluate an intervention in a hospital setting with an explicit aim to improve safety culture; (b) contain empirical data for analysis; (c) are available in any language; (d) are published in peer-reviewed scholarly journals; and (e) are published since 2000 (from the publication date of To Err is Human).
Articles will be excluded if they (a) describe interventions that do not have an explicit aim to improve safety culture, (b) do not measure the effect of the intervention on safety culture, and/or (c) have a primary purpose other than hospital staff or patient safety culture. From a theoretical perspective, the research approach may limit the review to studies that view safety culture as amenable to change within a healthcare system, and exclude conceptual studies. The approach adopted is an explicit focus limited to studies that seek to measure change in safety culture as a totality and not sub-elements of safety-culture only. Limitations of the study design, such as bias toward quantitative studies and interventions which target safety culture will be acknowledged in the findings. Inclusion and exclusion criteria are illustrated in Table 3. Article review process Search results will be entered to Endnote where duplicates will be removed. Titles and abstracts of remaining articles will be screened for inclusion by two reviewers in Covidence, a systematic review data management program. Disagreement will be resolved by discussion with a third reviewer. Those selected for full-text review will be assessed according to the inclusion and criteria for addition to the final sample, again by two reviewers and discussion with a third to resolve disagreements.
Data extraction and quality assessment Data will be extracted from each article and organized into a matrix. The articles will be examined to determine if and how well they measure effect on or capture outcome experience of health care workers and management following safety culture interventions. Data will be organised according to author, year, sample size, sample characteristics, place of publication, study setting, study design, intervention structure, intervention content, intervention duration, outcomes measured, measurement instrument, and principal findings. Data from included articles will be assessed independently by two authors, and disagreements will be resolved by discussion until a consensus is reached. The mixed-methods appraisal tool (MMAT) will be used as a quality assessment tool, with two authors discussing and cross-checking quality scores 49-51 . The Cochrane Collaboration's tool for assessing risk of bias in randomised trials 52 and non-randomised studies of interventions 53 will be employed to verify bias.

Data synthesis and analysis
Synthesis will follow the Joanna Briggs Institute methodological guidance for mixed methods reviews, which recommends a convergent approach to synthesis and integration 48  , and 'what staff outcomes and experiences are identified?' a thematic analysis of staff outcomes will be simultaneously combined with data on measures and effects available in the evidence.

Study status
Preliminary searches and study selection process piloted. We are currently formally screening search results against eligibility criteria.

Data availability
No data are associated with this article.
health © 2022 Adamson L. This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Laura Adamson
Department of Radiation Oncology, Sydney West Radiation Oncology Network, Crown Princess Mary Cancer Centre, Sydney, NSW, Australia Thank you for the opportunity to review this protocol to perform a mixed methods systematic review on healthcare worker staff outcomes of safety culture interventions.
The study protocol is clear. The rationale for investigating the gap in the literature is well stated and supported. I anticipate that there will be a variety of interventions, and safety culture assessments across the literature.

Some minor comments and recommendations:
With multiple safety culture measurement tools within in the literature, the authors should ensure they note the disparity between the tools. Consideration will be needed when comparing the different assessment tools and the authors should avoid direct comparison. I wonder if any additional questions around what measurement tools of safety culture are currently used.
interventions and the impact on staff outcomes. I applaud the authors for taking on such a review, as I can imagine that actually working through the papers, with their likely vast array of different types of interventions and definitions of safety culture, will be a significant undertaking. I do have a few comments and suggestions for the authors to consider in their revisions.
Is the rationale for, and objectives of, the study clearly described?
The paper clearly describes a set of research questions. However, I do have a couple of queries for the authors on the scope of the questions. The term safety culture is contested within academia and practice. The authors describe some of this variation in the term in their rationale, but I think they could, and perhaps should reflect this a little more in the body of their introduction. This is particularly important given that there is an implicit assumption embedded in this type of review (exploring the impact of an intervention on an outcome) that the construct that is foundational to the intervention can be manipulated, therefore leading to some change in the outcome. In the context of safety culture, there are some authors that would argue that it cannot be manipulated, but rather is an emergent property of a system -an outcome of other processes and structures -or at the very least, bidirectional (as the authors themselves mention). Of course, we can't be sure what position is 'right' but I do think the authors need to be clear that the manipulation of culture is in itself contested, and that by doing the review, they are taking an explicit position on this discussion. A good commentator on this issue is Stian Antonsen -I would urge the authors to engage in his work in this paper and the subsequent final review.
In terms of the research questions, because of this disagreement within the field I wonder if the review would benefit from an explicit research question about the nature of the definition of safety culture contained within studies looking to manipulate it? ○ Is the study design appropriate for the research question?
Largely, I think the design is appropriate. However, I do have a slight concern about the exclusion of research that does not present an explicit measure of safety culture as part of the intervention. This would exclude papers reporting research that is based broadly on the position that improving safety culture or facets of safety culture (including all of the varied ways that culture is conceived -for example, team communications and team working) might lead to improvements in an outcome of interest, but that don't undertake measurement of 'safety culture'. This in and of itself is not necessarily a problem, but I think the authors need to be clear the impact this will have on the review and the findings. It is also likely to exclude most qualitative work, which from my reading of the methods is included in the review.

○
Are sufficient details of the methods provided to allow replication by others? Yes.

○
Are the datasets clearly presented in a useable and accessible format?
The databases to be reviewed are stated clearly.

General comments:
Just a small, perhaps pedantic, point. At the end of the rationale, the authors state: "Understanding these issues is important: as the Lancet has stated, 'there is no patient safety without health worker safety' 43 ." The Lancet did not state this; rather, the authors who were published in The Lancet said that within that particular paper.

○
There are a couple of typographical errors in the manuscript -perhaps editorial could attend to these? ○ Is the rationale for, and objectives of, the study clearly described? Partly

Are sufficient details of the methods provided to allow replication by others? Yes
Are the datasets clearly presented in a useable and accessible format? Yes Competing Interests: No competing interests were disclosed.
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
Author Response 10 Nov 2022

Mairéad Finn
Thank you to both reviewers for the time and reflection taken to appraise this work. Their feedback greatly enhances the piece and we are grateful for the opportunity to add clarity and further investigate the literature. A short summary of our updates is: an enhanced review of the deployment of safety culture as a concept, and the differing perspectives therein; an incorporation of the new research question recommended; a clarification on the specific focus and limitation of the research; clarification on the treatment of varying measurement instruments in the review; and an adjustment to the reference in the Lancet, as observed by both reviewers.
Jane O'Hara: Thank you for these welcome observations. Drawing on the work of Stian Antonsen, in our introduction we have further developed varying theoretical approaches to the understanding and investigation of safety culture. We draw attention to differing approaches and acknowledge the specific focus of our research question relates to studies that view safety culture as manipulable (see also inclusion and exclusion criteria). We will clarify that no one position is right and be clear that the manipulation of culture is in itself contested, and that by doing the review, we are focusing on studies that take an explicit position on this discussion. Our position aligns to studies that seek to measure changes to culture in a system, though we do not argue this is 'right'.
Furthermore, we have since conducted the search and know now the variations in scale and time frame of included studies. For example, one study conducted a 3-year long intervention to influence culture. Several interventions were multi-pronged across long time frames. We will discuss these findings in the final review.
We have added a research question on the definitions of safety culture contained within the included studies: "How is safety culture defined in studies with interventions that aim to enhance it." We will record these and any theory of change documented in the included studies.
We have made both the explicit focus and limitation of our study more clear to readers. In the section on inclusion and exclusion criteria, we include the statement: "From a theoretical perspective, the research approach may limit the review to studies that view safety culture as amenable to change within a healthcare system, and exclude conceptual studies. The approach adopted is an explicit focus limited to studies that seek to measure change in safety culture as a totality and not sub-elements of safety-culture only. Limitations of the study design, such as bias toward quantitative studies and interventions which target safety culture will be acknowledged in the findings." Thank you for identifying the correct authorship within the Lancet publication, we will amend.

Competing Interests:
There are no competing interests.