Safety culture in the perception of public-hospital health professionals

ABSTRACT OBJECTIVE Evaluating safety culture in the perception of professionals working in public hospitals of the Unified Health System (SUS) of Distrito Federal, Brazil, three years after the implementation of the National Patient Safety Program (PNSP). METHODS Analytical cross-sectional study conducted in eleven public hospitals using the Safety Attitudes Questionnaire (SAQ) in electronic format. Stratified sampling was estimated according to the proportion of the total number of professionals in each hospital, as well as the representativeness of each professional group. The results of the total score and domains equal to or greater than 75 were considered positive. Descriptive and inferential analyses of professional groups and hospitals were carried out. RESULTS 909 professionals participated. The total score by professional group was negative (62.5 to 69.5) and the domains differed statistically in all cases. The eleven hospitals had a negative total score (61.5 to 68.6). The domains to attain positive performance were job satisfaction, stress recognition and teamwork climate. The lowest results were in working conditions and management perception domains, for which none of the hospitals had an average above 75. Differences were also found for domain means across hospitals, except in management perception. DISCUSSION Three years after the implementation of PNSP, the safety culture in eleven hospitals evaluated was weak, although the domains of job satisfaction, stress recognition and teamwork climate had positive results. The results can contribute to decision-making by managers, as safety culture is an essential element in the implementation of patient safety policy.


INTRODUCTION
Unsafe care and its harmful consequences to the patient have been reported since the 1980s. The World Health Organization (WHO) launched several initiatives focused on care safety, with greater emphasis since 2004 when it created the World Alliance for Patient Safety 1 . Brazil, part of this alliance, started the construction of a patient safety policy in 2001 with the creation of the Sentinela Network, a performance and safety observatory of health products 2 .
In 2013, Brazil intensified patient safety guidelines by implementing the National Patient Safety Program (PNSP), aiming to qualify health care. The safety culture is a transversal element that permeates the four axes of the program: encouragement of safe care practice; citizen involvement in their own safety; inclusion of the theme in teaching; and increased research on the topic 1 .
The term "safety culture" is used by organizations considered to be at high risk since the Chernobyl nuclear accident 3 . In healthcare, the safety culture is described as the product of individual and group values, attitudes, perceptions, skills and behavior patterns that determine a healthcare organization's commitment to patient safety management. Organizations with a positive safety culture are characterized by good communication between professionals, mutual trust and common perceptions about the importance of safety and the effectiveness of preventive actions 4 . The term "safety climate" is defined as the superficial and measurable characteristics of the safety culture based on the perceptions and attitudes of individuals at a given point in time [3][4][5] .
In line with international patient safety policies, PNSP follows the WHO definition of safety culture, which is based on five points: 1) all workers taking responsibility for their own safety, as well as for their colleagues', patients' and family members' safety; 2) prioritizing safety over financial and operational goals; 3) encouraging and rewarding identification, notification, and resolution of safety-related issues; 4) promoting organizational learning from the occurrence of incidents; and 5) providing resources, structure and accountability for the effective maintenance of safety 1,6 .
The Unified Health System (SUS) in Brazil provides health services to the population through various establishments, from basic units to hospitals. In this system, the aim is to promote a safety culture with emphasis on organizational learning and improvement, involvement of professionals and patients in the prevention of incidents, focusing on safe systems and avoiding individual accountability processes 6 . beds and 15,545 health professionals (Table 1). To map the installed capacity of hospitals and number of professionals, the National Register of Health Establishments (CNES) was used, according to data from May 2016. To preserve identification, each hospital was identified as HX, from H1 to H11, following the descending order of the number of beds.
The stratified sampling was calculated according to the proportion of the total number of professionals in each hospital and the representativeness of each professional group, with 869 professionals expected to answer the questionnaire ( Table 1). The inclusion criterion was to be a professional active relationship at the hospitals participating in the study. Professionals absent during the period of data collection were excluded.
SAQ is a self-applied instrument divided into two parts. The first one consists of 41 items that comprise six domains: teamwork climate (1 to 6), safety climate (7 to 13), job satisfaction (15 to 19), stress recognition (20 to 23), management perception (24 to 29) and working conditions (30 to 32). Items 14, 33 to 36 do not belong to any domain, but make up the total score, which is calculated with all statements. The second part collects data that characterize the professionals regarding gender, profession and years of experience in the field 8 . The responses to the items follow a five-point Likert scale, with the instrument's result varying from 0 to 100 for the total score and domains, where zero represents the worst and 100 the best perception of the safety climate. The safety climate is considered positive when the score is equal to or greater than 75 points 8 . For this research, we chose to transcribe SAQ into electronic format for mobile devices.
The team of researchers responsible for data collection attended training, awareness-raising meetings with hospital managers (directors, managers and professionals from the Patient Safety Center) and a presentation of the research project and researcher team. Subsequently, researchers visited hospitals and invited professionals to participate. Mobile devices were made available so that the Informed Consent Term (TCLE) and SAQ could be accessed.
For the analysis, professionals were divided into the following groups: 1) managers; 2) doctors; 3) nurses; 4) other graduate assistance professionals; 5) nursing technicians; 6) other assistance technical professionals; 7) non-assistance professionals from the support team.
The Kolmogorov-Smirnov test was used to assess data normality. According to their distribution, quantitative data were expressed as mean ± standard deviation (SD) or as median and interquartile range 25-75% (25-75% IQ). Categorical variables were expressed as number and percentage (%). For quantitative variables, Student's t-test or Mann-Whitney test were used when we had two groups and the Analysis of Variance (ANOVA) or Kruskal-Wallis  Table 3).
Among professional groups, managers had the highest total SAQ scores (69.5; SD = 14.5), although groups did not differ statistically (p = 0.067). For all groups, the mean total SAQ score was below 75. Regarding SAQ domains, all professional groups differed. Job satisfaction was the only one in which all groups had an average above 75. The domains with the most unfavorable results were safety climate, management perception and working conditions, with a mean lower than 75 in all groups. In teamwork climate, physicians, nursing technicians and other technical-level assistance professionals had an average above 75.
In the stress recognition, all groups had an average above 75, except for nursing technicians and non-assistance professionals from the support team (Table 3).
In teamwork climate, the post-hoc analysis showed that physicians had higher scores when compared to nurses  Table 3].
In job satisfaction, managers differed from physicians     Table 3].
In all domains, SAQ scores showed statistically significant differences among the hospitals participating in the study (Table 4). Thus, the total SAQ score ranged from 61.5 (SD = 12.  (Table 4).

DISCUSSION
The perception of safety culture among professionals was negative, with a mean total SAQ score below 75. Like other studies done in Brazil 9-12 , the result of job satisfaction was the only one evaluated positively by all professional groups between the domains. The domains evaluated negatively by all professional groups were working conditions, management perception and safety climate, the former with the worst performance. In a study carried out in three Brazilian public hospitals, the perception of safety culture among professionals was negative, with mean scores ranging between 65 and 69, and job satisfaction was also the domain with the best evaluation 9 .
Negative evaluations of the safety culture by professionals have also been observed in other countries 13,14 . A study done in Sweden with surgical teams showed that the perception of safety attitudes was negative, except for job satisfaction, which had an average score above 75 in all professional groups -a result again similar to ours 13 . In intensive care units of ten Australian hospitals, the perception of safety culture was negative in most services, with less than half of the professionals identifying it as positive 14 . These findings suggest the need for initiatives aiming to improve the safety culture of professionals in health institutions. For example, a study in the United States found significant increases in the half-yearly follow-up of the SAQ after the implementation of programs aimed at improving quality and safety associated with a significant reduction in preventable harm, serious adverse events and adjusted hospital mortality 15 .
In the assessment by hospitals, job satisfaction also had the highest scores -a fact also observed in other Brazilian studies [9][10][11][12] . In other countries, job satisfaction is one of the domains with best evaluation 13,[16][17][18][19] . Although there was a positive evaluation in all professional groups, managers had significantly higher scores than the other groups. This aspect can be explained because, in general, managers tend to have a more positive perception of the safety culture in their institutions when compared to other professionals 20 .
The critical performance of working conditions is like that found in other studies in Brazil and other countries, being always one of the worst-evaluated domains [9][10][11][12][16][17][18][19] . Studies done in hospitals in Sweden 13,21 and Australia 22 also showed a negative perception of working conditions, but with better scores than those observed in this study. However, physicians in these countries had a more positive perception than other professionals, differing from our results.
The health system is made up of high-risk services that are still considered to be of low reliability due to the countless adverse events that continue to happen daily around the world 1,6 . In this sense, making this system more secure requires resources, structure and responsibility for the effective maintenance of safety. The participants' perception of the precarious working conditions reflects the need for improvements in the assessed hospitals 1,23 . Furthermore, confronting working conditions results with good evaluations of job satisfaction can signal the preservation of the altruistic dimension of the health professional, which is reflected by the feeling of the social usefulness of what is produced. Interpersonal relationships, bonds of camaraderie, ways of coordination and cooperation, tacit rules of mutual help and coexistence among workers can increase job satisfaction, even in situations that are precarious for performance 24 .
Management perception was the second domain with the lowest mean among the groups of professionals and in most hospitals studied. Similar results have been reported by other studies 25,26 . In the study that evaluated the perception of nurses working with acute care in six Australian hospitals, this domain had the worst evaluation 25 . A similar situation was found in a survey carried out in Taiwan 26 . The perception of management was negative for all professional groups, and again the management group had the best score. Low scores in this domain suggest the need to improve management processes. It is essential to bring front-line professionals closer to decision makers to avoid generating a scenario in which management is not seen as a strength, but as a weakness for the safety culture, as reported in this Taiwan study 26 . Another study, also carried out in Taiwan, referred to management perception as a causal domain, as well as teamwork climate and stress recognition. Initiatives directed towards causal domains not only directly improve the domain itself, but also the performance of other domains. In management perception, other affected domains were teamwork climate, safety climate, job satisfaction and working conditions, which reinforces the importance of actions aimed at improving the management capacity of health services 27 .
Regarding patient safety, the safety climate situates the moment in which health services meet, guiding actions, promoting comparative assessment between services and monitoring results after the implementation of policies over time 9,10,23 . Although this domain had negative mean scores in all groups, there was a more positive perception by professionals with a technical level, especially when compared to nursing technicians and nurses. In this regard, physicians have responded less positively than nurses and other assistance professionals in other studies 23,28 -a result that deserves to be explored in further research.
Teamwork climate had a positive performance, with physicians having a significantly higher perception compared to other professional groups -which was also observed in a study carried out in two hospitals in Australia 22 . In fact, this domain has been considered a strong point by professionals involved in direct patient care 20 . As observed in another study carried out with nurses in university hospitals in Sweden 21 , in a study carried out in Slovenia, teamwork climate had the highest scores among SAQ domains 28 .
Although all domains are equally important for a safety culture, studies have shown that favorable results in teamwork climate and in safety climate are associated with lower rates of infections related to healthcare 29 . An association was also found with reduced rates of adverse event notification with teamwork climate, safety climate, working conditions and management perception. This suggests that efforts aimed at improving the perception of these domains can improve the quality of care 29 .
Stress recognition signals the professional's ability to recognize that their performance can be influenced by stressing factors 27 . Although this domain had a positive evaluation, it was still negative in four hospitals and was worse evaluated by non-assistance professionals from the support team when compared to professionals from the front line. A previous study showed that licensed practical nurses (professionals with a secondary level of education, who circulate in the operating room) had lower mean scores than perioperative physicians and nurses. That means they were less able to recognize that their performance may be influenced by stressors when compared to other professionals 13 . An Australian study also showed that other health professionals scored lower than doctors and nurses 22 . In studies that compare different professional groups, those with less education or not directly involved in care had a more negative perception, a point that can be explored in future studies. Understanding the differences between groups of professionals is essential to direct assertive initiatives, as this domain provides a view of the professionals' own understanding of their limitations under physical, psychological and emotional stress 8 .
One of the limitations of the study was the impossibility of randomizing participants due to the weakness of information systems about professionals in each hospital. Another limitation was not having compared the domains between hospital units, as the literature points to the existence of subcultures within the same organization 15 . The comparison between groups of professionals in each hospital was also left out, as they go beyond the objectives of this study. Furthermore, although most studies that assessed the culture of safety have focused on the assessment of health professionals directly involved in care 13,21,23,28 , the inclusion of the manager group is important because they play a key role in promoting patient safety 30 , which also allows assessing the dissociation between the managers' self-assessment and the health professionals' management perception.
Safety culture assessments have taken place more frequently in recent years, and its applications are diverse, such as genuine safety climate evaluation 9,23 , assessment before and after interventions 21 and combined measurements that seek to associate results 27,29 . Three years after the implementation of PNSP, safety culture in eleven hospitals evaluated was weak, however the domains of job satisfaction, stress recognition and teamwork climate had positive results.
We advise managers to invest in improvement initiatives, especially in areas with greater weaknesses, as they are important elements for patient safety and quality of care. The results point to fundamental issues, however, they do not cover the subject thoroughly, which requires additional studies that address the differences in the safety climate between the units that make up each hospital, as well as qualitative studies to deepen the understanding of the findings of this study.