Organizational Role and Perception of Organizational Safety Culture: A Survey of Registered Nurses, Physicians, and Managers

Risk of error is inherent in healthcare practice. Decisions are complex, non-mechanized, sometimes urgent, and have potentially serious consequences. The term “safety culture” describes a combination of organizational factors that minimize the risk of error. Poor emphasis on safety culture in hospitals has been associated with numerous negative outcomes such as mortality (Berry et al., 2016), infection rates (Fan et al., 2016), surgical complications (Birkmeyer et al., 2016), nurse burnout (Vifladt, Simonsen, Lydersen, & Farup, 2016) and the patient experience (Abrahamson, Hass, Morgan, Fulton & Ramanujam, 2016). Researchers have previously demonstrated organizational role as influencing the perception of workplace safety culture (Banaszak-Holl et al., 2017). Nurses are particularly pertinent to organizations as they deliver the majority of direct patient care.

Nurses practice at the “sharp end” of error; they function at the point in the clinical process where latent factors such as misaligned organizational process, poor communication, and lack of resources converge to actualize an error event. Nurses practice at the “sharp end” of error; they function at the point in the clinical process where latent factors such as misaligned organizational process, poor communication, and lack of resources converge to actualize an error event (Center for Patient Safety, 2018). Recent studies have affirmed the relationship between nurse work environment and safety culture (Olds, Aiken, Cimiotti, & Lake, 2017) and organizational role and safety culture perception (Banaszak-Holl et al., 2017). The purpose of this study was to build upon previous research addressing the relationship between work role and safety culture by examining the impact of organizational role on perception of safety culture within a multi-state, multi-organization sample of healthcare professionals.

Background

Safety Culture
Safety is imperative in high-reliability organizations and settings such as nuclear aircraft carriers, air traffic control centers, and nuclear power plants (Barach & Small, 2000). These organizations rely on a systems approach to error management whereby every individual assumes an active role in the prevention and control of operational failures and hazardous events when they occur (Reason, 2000). Healthcare organizations in the U.S. have been working to address patient safety issues for decades. In 2000, the Institute of Medicine (IOM) published “To Err is Human: Building a Safer Health System,” to address growing concern about the prevalence of medical errors resulting in unnecessary illness, injury, and death. The report highlights significant lags in the healthcare industry to construct effective safety systems that can prevent and control error, similar to those found in other high-risk organizations (Donaldson, Corrigan, & Kohn, 2000; Kohn, Corrigan, & Donaldson, 2000).

Further identified is the significant risk of death from medical error in U.S. hospitals (Donaldson et al., 2000; Kohn et al., 2000). Despite awareness and efforts to curb medical error, nearly a quarter-million patient deaths result from medical errors every year, costing over $17 billion annually, and making medical errors the 3rd leading cause of death in the U.S. (James, 2013; Makary & Daniel, 2016; Van Den Bos et al., 2011). Nationally, multiple and varied organizations implore the healthcare industry to be vigilant in identifying root causes of medical error in order to reduce the alarming prevalence of preventable hospital-related injury, illness, and death (James, 2013; Makary & Daniel, 2016; Kohn et al., 2000).

Safety culture influences the management, control, and minimization of operational failures that can lead to medical errors within healthcare settings. Safety culture has been identified as an important component in determining how to prevent and control medical errors. Safety culture influences the management, control, and minimization of operational failures that can lead to medical errors within healthcare settings (Pronovost et al., 2006). Safety culture is a subset of organizational culture that pertains to the prevention, detection, and management of patient safety events in healthcare and informs the safety climate that captures the perceptions, practices, and procedures employees follow when executing tasks (Dodek et al, 2012, p. 1506; Hofmann & Mark, 2006; Schneider, 1990).

Medical settings that practice effective patient safety culture value open and honest communication; enforce safety measures including confidential reporting; and incorporate incident reporting systems as well as near-miss reporting systems (Barach & Small, 2000; Kohn et al., 2000; Nieva & Sorra, 2003). In “The Patient Safety Culture” model, organizational culture informs patient safety culture, which then determines the patient safety climate, thus leading to individual attitudes affecting overall levels of patient safety (Morello et al., 2013). Organizational commitment to safety improves organizational performance (Page, 2004).

The Role of the Nurse
Nursing practice functions at the “sharp end of error,” the place where latent organizational effects meet with human factors to be actualized as error (Center for Patient Safety, 2018). Nurses provide the majority of direct patient care in most healthcare settings and are responsible for patient surveillance (i.e., monitoring of patient condition to detect change and prevent negative outcomes). Nursing surveillance reduces medical error and patient mortality (Leape et al., 1995; Mitchell & Shortell, 1997).

Nurses act as key communicators, gathering information at the bedside and translating assessments and patient concerns to physicians and family members. Nurse staffing levels, availability of nursing resources, nurse-responsive management, and overall nursing work environment have been identified as factors which impact patient safety (Aiken, Clarke, Slone, & International Hospital Outcomes Research Consortium, 2002; Needleman et al, 2002; Olds et al., 2017; Page, 2004). Research demonstrates patient safety culture is compromised in settings where nurses work more than 40 hours per week (Wu et al., 2013).

Organizational Role and Safety Culture
Managers’ opinions of patient safety culture vary significantly from direct care providers’ impressions and opinions in the same healthcare setting. Several studies have documented the disconnect between managers who assumed that a smaller number of operational failures occurred among their nurses, and direct care nurses who reported a greater number of observed operational failures, compared to the managers (Singer et al., 2009; Tucker & Edmonson, 2003; Tucker & Spear, 2006). This disconnect between expected versus observed failures indicates that perceptions of safety culture in healthcare settings vary across professional disciplines and roles. The disconnect contributes to the difficulty that direct care nurses may experience when trying to convince managers why organizational operation failures exist and persist (Tucker, 2004). It also provides rationale for large-scale and in-depth examination into factors that impact the relationship between organizational role and safety culture.

Theoretical Framework and Hypotheses

Factors related to high-risk work within error-prone environments, which contribute to the understanding of error risk in healthcare settings, have been identified by Reason (2003) and Page (2004). First, healthcare is complex and decisions are often heterogeneous between patient situations. The urgency of these decisions adds to their complexity. Second, the risk of serious outcome from an error is high, and patients are often vulnerable, unable to and/or uncomfortable with questioning care practices. Third, in most settings, healthcare is provided to many people at one time, often without adequate resources. Fourth, healthcare delivery remains an inexact science, adding to the complexity of decision making. And lastly, there remains a level of negative sanction for reporting or discussing error, which limits organizational learning. Among these factors, two primary risks emerge for individual healthcare workers: the amount of direct patient care and the critical nature of consequences if an error is made (Page, 2004). In this study, it is presumed that nurses, being at the ‘sharp end’ of an error, are more aware of error and threats to organizational safety culture than physicians or managers.

Methods

The purpose of this cross-sectional secondary analysis was to contribute to knowledge and understanding surrounding role differences and perception of safety culture guided by the following hypotheses:

1. Based upon previous literature, nurses will perceive a lower level of safety culture within their organizations than physicians or managers.
2. Nurses, at the sharp end, will report more adverse safety events than physicians or managers, and that number of events reported will be a significant negative predictor of safety culture perception.
3. Organizational experience (measured through length of hospital employment and hours worked per week) will have a significantly negative relationship with perception of safety culture.
4. Respondents who have direct patient contact will report negative perception of safety culture more often than those who do not provide direct patient care.

In the following section, we describe our study methods, analysis, and results.

Sample and Instrument
AHRQ released the “Hospital Survey on Patient Safety Culture” in 2004, and since that time it has become the criterion standard of patient safety culture tools (Sorra & Nieva, 2004; Sorra, Khanna, Dyer, Mardon, & Famolaro, 2012). The AHRQ survey is administered annually to all hospital staff (Sorra & Nieva, 2004).

Data for this study were derived from a national sample of hospitals that administered the AHRQ safety culture survey through a contractual agreement with Press Ganey Associates. Data for the study were collected in 2011 and became available in analyzable form in 2014. The dataset had an initial sample size of 43,968 respondents. Over 3.5% of data (1,633 cases) included survey responses that were missing in a systematic pattern and were excluded from analysis. The remaining sample included 13,753 nurses, 2,133 physicians, 3,041 managers, and 23,408 respondents taking other staff positions. Only nurses, physicians and managers were included in the current analyses. Only registered nurses were included in the nurse sample. The final analytical sample included 17,858 respondents: 13,072 nurses, 2,043 physicians, and 2,743 managers working within a multi-state sample of 66 hospitals.

Survey items are listed in Table 1. For more information and a fully copy of the survey please refer to www.ahrq.gov.

Table 1. Items in the Hospital Survey on Patient Safety Culture

Analysis
The outcome variable for the analyses was the respondent overall mean score on the AHRQ Hospital Survey of Patient Safety Culture, minus the single outcome measure asking respondents to grade the hospital unit patient safety culture which was excluded for reasons of normality and distribution. Predictor variables included in the model corresponded to the proposed hypotheses and included occupational status (compared to nurse), length of time (years) working in the hospital, number of adverse safety events reported in the previous year, number of hours worked per week, and whether the role included direct patient care. Correlation between variables and data distribution was determined prior to specification of statistical models. All correlations were below .2, with the exception of occupational status and patient direct care, which had a correlation of .58. Data distribution of the outcome variable was acceptable.

Linear mixed effect models were fitted to the data using STATA 14.2 (StataCorp, 2015). Mixed effect models were utilized to account for the clustering or non-independence of errors between respondents who worked in the same hospital. These models were performed in the total sample (n = 17,858), and then specified for each occupational status (13,072 nurses, 2,043 physicians, and 2,743 managers). In the total sample model, the outcome variable of overall mean survey score modeled with the predictor variables of status (nurses as the reference group), length of time working at hospital, adverse events reported, hours worked per week, and direct interaction with patients. In the subsample models by staff position, the overall mean scores were predicted by length of time working at hospital, events reported, hours worked per week, and direct interaction with patients.

Results

Managers had the highest overall mean (potential range 0-100) Safety Culture survey score (72.7) followed by physicians (65.8) and nurses (64.2). Approximately a third of managers (32%), physicians (35%) and nurses (37%) were employed five years or fewer by their hospital. As predicted, nurses reported safety events more frequently than managers or physicians. Fifty-one percent of managers had not reported an event in the past year, and an additional 18% had recorded one or two events.

Physicians reported events even less frequently, with 61% reporting no safety events, and an additional 25% reporting one or two. Among nurses, 29% reported no events in the past year, an additional 41% reported one or two, and 19% reported 3-5 safety events in the past year. More than half of nurses reported working 40 hours or less per week; 56% worked 20-39 hours and 34% worked 40-59 hours per week. Nineteen percent of physicians and 9% of managers worked 20-39 hours per week. Most mangers (79%) and most physicians (38%) worked 40-59 hours a week, with a notable 20% of physicians working 60-79 hours weekly. Ninety-five percent of nurses, 94% of physicians, and 36% of managers reported having direct contact with patients. Predictor variables frequencies by occupational status are displayed in Table 2.

Table 2. Predictor Variable Frequencies by Occupational Status

The mixed-effects model demonstrated that, compared to nurses, managers reported significantly higher mean scores on the Safety Culture survey. Longer tenure at the hospital predicted higher overall mean scores, and number of events reported predicted lower overall mean scores. Compared to those who reported direct interaction with patients, those who reported no interaction with patients reported significantly higher mean Safety Culture scores. Number of hours worked per week was not a significant predictor variable.

Nurses reported more adverse events than those in other roles, and number of events reported predicted a lower overall safety culture perception. Three of the proposed hypotheses were confirmed. Nurses had significantly lower mean Safety Culture scores than physicians and managers, and those with direct patient contact had significantly lower perceptions of their hospital’s safety culture. Nurses reported more adverse events than those in other roles, and number of events reported predicted a lower overall safety culture perception. Number of hours worked per week was not a significant predictor, and years worked at the hospital had a positive influence on safety culture, opposite of what had been predicted. Approximately 9% of the total variance in the mean Safety Culture scores is represented at the hospital level. Mixed effect model results for the full sample are displayed in Table 3.

Table 3. Results of Full Sample Model: Predictor Variables on Overall Mean Safety Culture Score (N=17,858)

Subsample models for each occupational status demonstrated similar results. In the mixed effect model for nurses only (n =13,072), years of hospital employment predicted higher mean Safety Culture scores. More events reported and having direct patient interaction predicted lower scores. Number of hours worked per week was not a significant predictor. In the subsample model for physicians (n =2,043), number of events reported and direct interaction with patients predicted lower Safety Culture Scores. Years of hospital employment and hours worked weekly were not significant predictors. The manager subscale model (n = 2,743) demonstrated that years of hospital employment had a significant positive effect on Safety Culture scores, as did number of events reported, a finding that differs from the other subsample models. Work hours and direct patient interaction were not significant predictors for this group.

Discussion

Consistent with proposed hypotheses, the role of nurse, number of reported adverse safety events, and direct patient contact predicted a lower perception of organizational safety culture. Years of hospital employment and hours worked per week did not consistently demonstrate the predicted effect on safety culture perception. It is possible that there is a cohort effect; more recently educated individuals may have more acute awareness of safety implications. Expectations of work place may also vary based upon age and years of training. Individual level variables were not available for this current analysis, and future research would benefit from adding individual education and age to the analysis. Understanding the specific nature of a cohort effect is important as it can lead to a more focused and impactful distribution of costly training resources.

Previous research has suggested that number of hours worked per week compromise safety outcomes (Rogers, Hwang, Scott, Aiken, & Dinges, 2004). The current findings did not indicate a significant link between working many hours weekly and perception of safety culture. However, more events reported and direct interactions with patients were impactful predictors. Respondents to this survey were queried about patient interaction in a binary “yes/no” manner, which provides limited understanding of the relationships among hours worked per week, reported adverse events, and patient interaction. Future research should examine the proportion of time spent in direct interactions with patients as well as number of years of traditional full or part-time hours per week working with patients. A plausible link to safety culture may not be hours worked per week, but rather, the proportion of the total work hours spent in direct patient care throughout an individual’s career. Understanding these relationships may help hospitals identify individuals at risk of developing a misperception of safety culture and interventions for ameliorating this risk, perhaps through safety-related educational policies.

The available data was not amenable to investigation surrounding the impact of previous safety levels on safety culture perception by hospital employees. For example, if a hospital with a poor record of patient safety had experienced recent improvements, that hospital may be perceived by employees as having a fairly good safety culture despite persistent poor overall performance. Conversely, it is unclear whether respondents with experience in multiple hospital settings have different (or more accurate) perspectives on safety culture. Further research in this area could contribute to knowledge and understanding surrounding basis of comparison when perceiving safety culture. Understanding this information may help hospitals identify individual-related factors for developing a misperception of safety culture.

There are a number of limitations to this study. First, a convenience sample of Press Ganey client hospitals was drawn for this analysis. As noted previously, the data set provided limited data regarding respondent characteristics and work responsibilities. Finally, data were collected in 2011; follow up analyses on more recent data would provide comparison data for the results of this study.

Conclusion

A primary implication of this study is the importance of nurse input into safety decisions, and for management and/or those with little patient interaction to seek input from persons working at the ‘sharp end’ of healthcare delivery. A primary implication of this study is the importance of nurse input into safety decisions, and for management and/or those with little patient interaction to seek input from persons working at the ‘sharp end’ of healthcare delivery. Quality initiatives such as Magnet designation and Page (2004) recommendations have highlighted the importance of including direct-care nurses in decision making processes. Organizational interventions should be directed toward promoting safety culture among nurses, physicians, and managers who have direct interactions with patients. This study provides the basis for future research to investigate organizational safety issues and interventions through the lens of the interaction between direct care providers and patients in the hospital setting.

Authors

Kathleen Abrahamson, PhD, RN
Email: kaabraha@purdue.edu

Dr Abrahamson is an Associate Professor in the Purdue University School of Nursing. Dr. Abrahamson’s research is focused on social and organizational factors that impact health services delivery. Her research addresses nursing home quality, patient safety, and the work environment of nurses. Her methods include qualitative analysis, survey development, structural equation modeling, and multi-level modeling.

Rong Fu, PhD, MS
Email: Rfu@siena.edu

Dr. Fu is an Assistant Professor of Sociology at Siena College. Her main fields of study include medical sociology, health, aging, and research methods.

Karis Pressler, PhD
Email: kpressler@purdue.edu

Dr. Pressler is the project manager of the Indiana Safer Medication Administration Regimens and Treatment (SMART) Campaign, that works to reduce unnecessary polypharmacy in nursing homes throughout the state. Her research areas of interest include the sociology of health and disability, and evaluating programs that aim to maintain and improve quality of life among older adults.

Bradley R Fulton, PhD
Email: bfulton@pressganey.com

Dr. Fulton has been a Senior Analytics and Research Scientist at Press Ganey Associates for 10 years. In this role, he is responsible for maintaining and defending Press Ganey’s methodological and statistical processes and standards, and conducting research pertaining to patient satisfaction across a number of health care settings.

Kristopher H. Morgan, PhD
Email: kmorgan@pressganey.com

Dr. Morgan is the Director of Workforce Analytics at Press Ganey Associates and has worked as a senior scientist at Press Ganey for 7 years. In this role, he is responsible for new metric development, instrument validation, maintaining Press Ganey’s methodological and statistical processes and standards, and conducting research pertaining to employee and patient experience across a number of health care settings.

James G. Anderson, PhD
Email: andersonj@purdue.edu

Dr. Anderson is Professor of Medical Sociology and of Health Communication at Purdue University. He earned a B.E.S. in Chemical Engineering, M.S.E. in Operations Research and Industrial Engineering, M.A.T. in Chemistry and Mathematics, and a Ph.D. in Education and Sociology from the Johns Hopkins University. He is the former Director of the Division of Engineering of the Evening College at Johns Hopkins University.