Article Text

Associations between safety culture and employee engagement over time: a retrospective analysis
  1. Elizabeth Lee Daugherty Biddison1,
  2. Lori Paine2,
  3. Peter Murakami3,
  4. Carrie Herzke1,
  5. Sallie J Weaver4
  1. 1Department of Medicine, Johns Hopkins School of Medicine, Baltimore, Maryland, USA
  2. 2Johns Hopkins University School of Medicine, Baltimore, Maryland, USA
  3. 3Department of Biostatistics, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA
  4. 4Department of Anesthesiology and Critical Care Medicine, Johns Hopkins School of Medicine, Baltimore, Maryland, USA
  1. Correspondence to Dr Elizabeth Lee Daugherty Biddison, Department of Medicine, Johns Hopkins School of Medicine, Baltimore, MD 21287, USA; edaughe2{at}


With the growth of the patient safety movement and development of methods to measure workforce health and success have come multiple modes of assessing healthcare worker opinions and attitudes about work and the workplace. Safety culture, a group-level measure of patient safety-related norms and behaviours, has been proposed to influence a variety of patient safety outcomes. Employee engagement, conceptualised as a positive, work-related mindset including feelings of vigour, dedication and absorption in one's work, has also demonstrated an association with a number of important worker outcomes in healthcare. To date, the relationship between responses to these two commonly used measures has been poorly characterised. Our study used secondary data analysis to assess the relationship between safety culture and employee engagement over time in a sample of >50 inpatient hospital units in a large US academic health system. With >2000 respondents in each of three time periods assessed, we found moderate to strong positive correlations (r=0.43–0.69) between employee engagement and four Safety Attitudes Questionnaire domains. Independent collection of these two assessments may have limited our analysis in that minimally different inclusion criteria resulted in some differences in the total respondents to the two instruments. Our findings, nevertheless, suggest a key area in which healthcare quality improvement efforts might be streamlined.

  • Patient safety
  • Safety culture
  • Attitudes

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

An environment that is deficient in joy and meaning, where the workforce is burdened by extreme production pressures, toxic sociocultural norms, and the risk that they could be physically or psychologically harmed, is an environment where both the workforce and patients suffer.1—Lucian Leape InstituteRoundtable on Joy and Meaning in Work and Workplace Safety, 2013


The patient safety movement has grown in parallel with increasing challenges to maintaining the well-being, engagement and resilience of the healthcare workforce. Efforts to measure, evaluate and improve processes and outcomes are increasingly complicated by narrow financial margins and exponential growth in the number of patients presenting with multiple comorbidities and complex care needs. These factors together have created a perfect storm in which the healthcare workforce struggles both to meet expectations and to flourish in doing the work to which they have dedicated their lives. For >14 years, studies have repeatedly found high levels of burnout, job-related stress, depression and turnover among members of the healthcare workforce.2–5

Theoretical models of job performance have underscored the effects of prolonged job-related stressors on individual behaviour, performance and retention, as well as effects on organisational outcomes, safety and reliability.6–8 In parallel, conceptual models of positive valence indicators of workforce well-being, such as employee engagement,9 and other constructs that can be collected under the broad and diverse umbrella of positive psychology (eg, participative management)10 and organisational citizenship behaviours,11 emphasise psychological involvement of employees in their work as a means to improve work outcomes.

Evidence supporting the relationship between indicators of workforce well-being, patient safety-related and other outcomes in healthcare continues to grow as well. For example, a study of 7076 registered nurses working in 161 hospitals found significant relationships between nurse burnout and rates of both hospital-acquired urinary tract infections (UTI) and surgical site infections (SSI).12 Their results indicated that each 10% increase in the proportion of nurses reporting high burnout was associated with an increase of nearly one UTI and two SSIs per 1000 patients. Extrapolating these findings, organisations could estimate annual cost savings of $28–69 million from prevented UTIs and SSIs were they to realise a 30% reduction in nursing burnout. Similarly, a study of 178 recently hospitalised patients and their main inpatient providers found correlations between aspects of physician burnout and patient satisfactory. Importantly, this study also found a significant increase in postdischarge recovery times for those patients whose physicians reported burnout symptoms. Such findings suggest that healthcare workforce well-being may also impact costs associated with illness and injury of their patients, including lost time from work.13

Such findings underscore recommendations from the National Patient Safety Foundation, Institute of Medicine and numerous leading professional associations, articulating the need to address patient safety and care quality from a systems perspective that is both patient and workforce centred.14 Drawing from the theoretical lenses of positive psychology and concepts of psychologically healthy workplaces,15–17 this perspective is hypothesised to be critical for developing a workforce that has the bandwidth and motivation to be active, engaged organisational citizens, who feel supported by the system in which they work.1 Thus, creating the conditions that facilitate an engaged workforce may also offer pathways for improving patient safety and care quality. To this end, strategic goals dedicated to helping healthcare workers find joy and meaning in their work have been advocated by safety-focused national entities and professional associations alike.


Despite a history of study in other organisational contexts,8 ,18 ,19 relatively limited work in the healthcare context to date has explored hypothesised relationships between positive valence indicators of workforce well-being, such as employee engagement and organisational safety characteristics or attitudes.20 ,21 These are important relationships to explore, with potential to offer insight into unexamined barriers and facilitators of patient safety improvement and change efforts. Theories of high reliability and resilience22 ,23 suggest that organisations that are able to achieve high levels of reliably safe performance do so in part through shared norms of vigilance, willingness to speak up regarding concerns or creative ideas, a unified learning orientation and a shared motivation to continuously seek opportunities to improve. Arguably, engaged team members may have greater bandwidth and motivation to create or sustain these safety-oriented norms compared with team members who are actively disengaged or burned out. Thus, engagement may influence or moderate safety-related aspects of organisational culture and team member perceptions of that culture, as well as influencing safety and quality outcomes. Frameworks for patient safety improvement similarly highlight that improvement efforts, a form of organisational change, require both technical interventions (eg, checklists, changes to work processes or care algorithms) and adaptive interventions (eg, interventions aimed at engaging organisational members in the change process and developing or strengthening cultural norms).24 Much of the work in this area has focused on safety culture or workforce perceptions and attitudes regarding safety culture, known as safety climate.25 Employee engagement is another arguably important ‘adaptive’ component that has received comparatively little attention, however, particularly in the patient safety literature. Meta-analyses have demonstrated organisation-level effect sizes of −0.21 to −0.28 between indices of employee engagement and safety indices; however, only a very small number of studies were available to estimate these effects.26 Additionally, previous studies have not explored relationships between engagement and foundational aspects of reliably safe operations such as safety climate. To address this gap, we examined the relationship between employee engagement and healthcare worker perceptions of patient safety culture (often a foundational element that, when weak or poor, has been identified as a common root cause in both safety and quality breakdowns) across multiple years in a sample of >50 inpatient hospital units nested within a large academic health system in the northeastern US.

Employee engagement

Employee engagement, conceptualised as a positive, work-related mindset including feelings of vigour, dedication and absorption in one's work,9 ,26 has demonstrated an association with a number of important worker outcomes in healthcare, including staff turnover, productivity and patient safety events.26–28 When engaged, clinicians, staff and administrators feel ‘physically involved, cognitively vigilant and emotionally connected' to their work.26 ,29 Some theories define engagement as the direct opposite of burnout with the two constructs conceptualised as two ends of a single continuum. Engagement reflects relatively high reserves of emotional energy and capacity for mental resilience (high energy), a high degree of involvement in one's work fed by a sense of significance, enthusiasm, inspiration, pride and challenge, and a sense that one's efforts are meaningfully related to the feedback, evaluations and outcomes received in the context of one's work. Conversely, burnout reflects a combination of emotional exhaustion (low energy), cynicism (low involvement) and inefficiency (low sense of efficacy and/or learned helplessness).30 ,31 Thus, engaged healthcare workers view both their ‘daily process of care’ work and their more general professional work as meaningful, and they likely feel responsibility and a realistic degree of autonomy. They also likely perceive strong ties between their efforts, the evaluations of their work and other outcomes.32

Perceptions of patient safety culture

While conceptual clarity and theoretical grounding continues to emerge,33 ,34 patient safety culture has generally been defined as a group-level (eg, team, unit, department or organisation) construct reflecting patient safety-related norms, artefacts such as policies, procedures, communication and teamwork patterns, leadership and day-to-day behaviours. It has been proposed to influence a variety of patient safety outcomes, as well as staff satisfaction and performance.35–37 The related concept of patient safety climate reflects team members' perceptions or attitudes about the organisational safety culture in which they work. A great deal of peer-reviewed literature has examined interventions to build or improve patient safety climate,25 ,38 and climate has been found to moderate the effectiveness of other, technically oriented safety interventions, like checklists, training/education or procedural algorithms.39 Although theories of safety culture and climate have been developed in both the general organisational and management sciences that have articulated numerous factors affecting development and sustainment of both organisational and work-area-level climates,40 ,41 employee motivation and engagement are (1) commonly unaccounted for in these theories and (2) not robustly tested as potential influences on climate or as moderators of the relationships between climate and other safety-related outcomes. Arguably, theories and studies of engagement suggest that engaged employees likely have a higher probability of also displaying safety-related attitudes and behaviours that define safety climate, given that the high degrees of self-efficacy, motivation and sense of being invested in by their organisation may provide the bandwidth and attention necessary to develop safety-related attitudes and demonstrate safety-related behaviours. The inter-relationships between these constructs have not been extensively explored, however. In hopes of offering initial insight into the potential relationships between these two constructs, we examined them in a series of secondary analyses. Specifically, we asked: to what extent are group-level measures of engagement associated with group-level perceptions of patient safety culture and does that association persist over time? We hypothesised that work-area-level engagement and team member perceptions of safety culture in their work area would be positively related.


Since 2006, the Johns Hopkins Hospital (JHH) has administered the Safety Attitudes Questionnaire (SAQ) every 18–24 months. The SAQ has been shown to be psychometrically sound and reliable in a variety of healthcare settings.42 Additionally, JHH has assessed employee engagement across both inpatient and outpatient units annually since 2009 using an instrument developed by the Gallup organisation, known as the Gallup Q12.43 The Q12 was validated by the Gallup organisation using an iterative process that involved characterising a range of units from ones with high workplace turnover to, on the opposite end of the spectrum, ones with high-functioning, productive teams.27 ,43 Both surveys are distributed to all staff, using a mix of electronic and paper-based methods. All responses are anonymous. Results are reported to hospital, department and unit leadership in aggregate form. For the SAQ, all work areas with a survey response rate >60% receive a score on each of seven domains, as well as scores on individual survey items. All SAQ responses are scored on a five-point Likert scale from ‘disagree strongly’ to ‘agree strongly’, and scores are reported as the percentage of respondents in a given unit reporting that they ‘agree slightly’ or ‘agree strongly’ with a given statement. Thus, a unit's score for a given SAQ domain reflects the proportion of respondents in a given unit who rank their unit highly on the 3–7 items within that domain. Items for each domain are listed in online supplementary table A1. For the Q12, each work area manager receives a report containing scores on each of the 12 items, as well as Grand Mean score across all items.

In this analysis, time-matched SAQ and employee engagement data were analysed from 58 inpatient units in 2009, 61 in 2011 and 59 units in 2013. Based on institutional safety interest and needs, the Safety Culture Assessment was revised in 2013 such that four of the original seven SAQ domains had available data over all three time periods. In all analyses, a p value of <0.05 was considered statistically significant. Analyses were performed using R statistical software V. ,45


In 2009, of the 2722 total eligible staff assigned to the units in this analysis, 90.9% (n=2473) completed the SAQ. For 2011 and 2013, response rates for the SAQ were 73.8% (n=2646) and 73.2% (n=3020), respectively. Respondents included nurses, physicians and support personnel. Median scores across all units for the included SAQ domains are summarised in table 1 for all time periods.

Table 1

Median SAQ scores (and IQR) by domain across included units and time periods by year

For employee engagement in 2009, of the 3222 assigned staff on the included units, 63.3% completed the Q12 (n=2041). For 2011 and 2013, response rates for employee engagement were 64.2% (n=2024) and 68.3% (n=2382). The median engagement survey response rates by unit within the three time periods were 64.0%, 66.0% and 72.0%, respectively. For these time periods, the median unit-level Grand Mean engagement scores were 3.84 (IQR 0.34), 3.91 (IQR 0.39) and 3.84 (IQR 0.31). In order to account for potential non-linear associations and non-normal distribution of scores and to provide robustness to outliers, Spearman's rank correlation coefficients were chosen to examine the association between unit-level engagement and safety attitude scores. Spearman's correlations between each of the four SAQ domain scores and Grand Mean engagement scores are shown in table 2.

Table 2

Spearman's correlations between SAQ domains and employee engagement grand mean by year

We then explored the relationships between each of the 12 items that comprise the employee engagement Grand Mean score and the four SAQ domains (see table 3 for safety and teamwork climate domains and online supplementary table A2 for complete data). Only correlations of 0.40 or higher, listed in bold in table 3, were considered robust. The 0.40 criteria were chosen to reflect a moderate effect size based upon common effect size conventions.46 ,47 Results suggested consistent relationships between safety climate, which focuses on perceived norms related to communication, learning, and error management and engagement items reflecting (1) clarity of goals and expectations, (2) opportunities for employee development and constructive feedback and (3) supervisory support. Teamwork climate, which focuses on team relationships and resolution of conflict, was also consistently associated with perceptions of voice (ie, that one's opinions and input is meaningfully counted).

Table 3

Spearman's correlations for individual Q12 employee engagement items and SAQ domains


Our results suggest moderate to strong positive correlations (r=0.43–0.69) between employee engagement and unit-level indices of safety attitudes. All four safety attitude domains demonstrated similar, moderate to positive correlations with engagement across three time points, suggesting a relatively consistent relationship between the two measures. Importantly, this relationship remained relatively stable despite increases over time in the number of eligible respondents. Although changes in unit structure over time (eg, moves and reconfigurations occurring largely in 2012) prevented us from conducting true longitudinal analysis, the results suggest a potentially meaningful association between employee engagement and healthcare worker perceptions of safety culture that may warrant more robust future investigation. Clearly numerous factors may mediate, moderate or otherwise help to characterise and explain this simple observed association. Future research can provide valuable insight in these areas.

Others have evaluated the association between either safety culture or engagement on a variety of outcomes important to healthcare delivery and patient safety, and many methods proposed to improve scores on one or the other instrument.25 ,48–51 In practice, many hospitals across diverse international settings have chosen to invest significant resources in measuring healthcare worker perceptions of patient safety culture and/or engagement on a regular basis in an effort to improve organisational culture, staff satisfaction, commitment and patient safety. For example, the UK's National Health Service (NHS) has invested in measuring employee engagement across its nearly 1.3 million staff since 2009.52 The NHS has also supported evaluations of patient safety culture metrics and periodic efforts to measure perceptions of culture across multiple trusts since 2006.53 ,54 Additionally, accreditation standards, including those of the Joint Commission and Joint Commission International, call for regular assessment of safety culture.55 In the USA alone, >650 hospitals voluntarily contributed patient safety culture survey data to a national database maintained by the Agency for Healthcare Research and Quality in 2014.56 Despite widespread implementation of these metrics, the relationship between safety culture and employee engagement had remained relatively unclear.57–59 Also unclear has been the degree to which the two tools reflect attitudes or behaviours similar enough to warrant focusing time and effort on use and improvement of a single tool, or, alternatively, whether interventions may be developed to impact both arenas more efficiently. Additionally, our results suggest value in understanding the reciprocal impact of different interventions designed to improve engagement, reduce burnout or improve perceptions of patient safety culture.

Limitations and future directions

Although this analysis of routinely collected organisational data does not support inferences regarding causality, it does advance understanding of the directionality and degree of association between healthcare worker perceptions of safety culture and engagement at work. Independent collection of these two assessments did limit our analysis in that minimally different inclusion criteria, turnover and unit staffing allocation methods resulted in differences in the total number of respondents surveyed by the two instruments. Additionally, although all engagement data were collected via online survey for the units included in this study, 39% of hospital units collected SAQ data in 2009–2010 via paper, while only 14% used only paper in 2013, which may have introduced non-response bias. We were not able to control for survey modality, but we note that even though overall response rates decreased across the time period of our study, actual survey scores did not demonstrate a uniform pattern of change corresponding to the decreased use of paper-based methodology. Arguably, moreover, conducting the surveys independently helped to mitigate some forms of mono-method bias. Our findings must also be considered in the context of multiple comparisons. Adjusting for multiple comparisons (eg, using a Bonferroni-style correction method) can reduce type I error (ie, the error of rejecting the null hypothesis when it is true) but it will be at the cost of inflating type II error (ie, incorrectly accepting the null hypothesis to be true when instead the alternative hypothesis is correct). Given the exploratory nature of this work, we opted to not apply such a correction when interpreting statistical significance, though we recognise its importance. In light of this fact, although results with a p value of <0.05 are considered significant and worth mentioning, we note that all of our findings related to engagement Grand Mean scores (table 2) demonstrated p values at or below 0.001, and, on the item, level we focused on findings with associations of 0.40 and above (table 3). Finally, our findings may also be limited by the fact that we were unable to control for staffing levels and severity of illness in the study units, given the retrospective nature of the work. Our findings do, however, suggest an area in which healthcare quality improvement efforts might be streamlined.


Overall, our results suggest an association between unit-level perceptions of safety culture and employee engagement over time in a large urban academic medical centre. Future research is needed in order to more fully understand temporal patterns and causal pathways between engagement and the broader construct of patient safety culture; however, our results support conceptual arguments that patient safety and working environments that motivate and engage employees are meaningfully associated. Practically, these findings support the development of approaches to improving patient safety that pair human resource, development and leadership interventions with traditional safety and quality improvement approaches.


We would like to acknowledge the important contributions of the members of the Johns Hopkins Provider Behavior Group including Paula Kent, Myles Leslie, MarieSarah Pillari, Melinda Sawyer, Carol Woodward, and Jiangxia Wang. We would also like to thank Drs Cynthia Rand and Michelle Eakin for their helpful feedback in the development of this work.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Twitter Follow Sallie Weaver at @SallieJWeaver

  • Contributors ELDB and SJW contributed to study design, analysis and manuscript writing. LP contributed to the study design and manuscript editing. PM contributed to analysis and manuscript writing. CH contributed to study design and manuscript editing.

  • Funding This study was supported by a grant from the NIH NHLBI (K23HL098452 to the Johns Hopkins University) and a grant from the Johns Hopkins Institute for Clinical and Translational Research (ICTR), which is funded in part by Grant Number 1KL2TR001077-01 from the National Center for Advancing Translational Sciences (NCATS), a component of the National Institutes of Health (NIH), and NIH Roadmap for Medical Research. Its contents are solely the responsibility of the authors and do not necessarily reflect the official views of Johns Hopkins ICTR, NCATS or NIH.

  • Competing interests None declared.

  • Ethics approval Johns Hopkins IRB-X.

  • Provenance and peer review Not commissioned; externally peer reviewed.