Patient safety culture in a university hospital

ABSTRACT Objective: to assess patient safety culture in a university hospital. Method: cross-sectional study with data collection through the Hospital Survey on Patient Safety Culture applied in electronic device. A total of 381 employees were interviewed, corresponding to 46% of the sum of eligible professionals. Data were analyzed descriptively. the Cronbach’s alpha was used to calculate the frequency and reliability. Results: most were women (73%) from the nursing area (50%) and with direct contact with patients (82%). The composites related to “teamwork within units” (58%, α=0.68), “organizational learning - continuous improvement” (58%, α=0.63), “supervisor/manager expectations and actions promoting patient safety” (56%, α=0.73) had higher positive responses. Nine composites had low positive responses, with emphasis on “nonpunitive response to error” (18%, α=0.40). Only the item “in this unit, people treat each other with respect” had positive response above 70%. The patient safety assessment in the work unit was positive for 36% of employees, however only 22% reported events in past year. Conclusion: the findings revealed weaknesses in the safety culture at the hospital, with emphasis on culpability.


Introduction
Patient safety culture corresponds to values and behaviors of members in an institution and collectively represents the degree of institutional commitment with the safety of its processes (1) . This construct reflects intangible aspects of health care, influenced exceedingly by the leadership, supervision and feedback to professionals (2) . Caregivers recognize to be inserted into an institution in which to follow the procedures is important. Therefore, they mark out their actions by performing the good practices of the area and providing information for its continuous improvement (3) .
Institutions with patient safety potentially provide safe care of better quality to their patients. The best scores on dimensions regarding safety culture were related to the lower incidence of surgical site infection in hospital (4) , reduction of injuries, critical adverse events and risk-adjusted mortality (5) . In risk-adjusted morbidity analyses of the patients and characteristics of the hospital, however, the positive responses of safety culture were not related to mortality in patients with acute myocardial infarction (6) , nor was affected after reduction of catheter-associated infections (7) .
The safety culture in healthcare environments is typically assessed by quantitative surveys based on individual items and combination of composites (1) . In The evaluation of patient safety culture is the first step to find the aspects that require improvement in this process.
In the Brazilian context, some initiatives to measure and evaluate safety culture in institutions have been registered (8)(9)(10)(11) , revealing weaknesses in different aspects. There still prevails the perception that failures in patient safety point to individual responsibilities and, consequently, punitive actions for the professional. This posture prevents the establishment of the improvements required. In the Northern Region of Brazil, which is historically less developed and with lower supply of health professionals and services (12) , this scenario is possibly more prevalent. This region of the country lacks investigations on safety culture. The objective of this research was to assess the patient safety culture in a university hospital from Manaus, Amazonas. Healthcare and administrative employees (including public servants, temporary employees or professionals of the multi-professional and medical residency program) working at least for three months in the institution were elected. Employees that were separated, on leave, or worked outside the main building of the hospital were ineligible.

Cross
Participants were selected by convenience sampling. The HSOPS was translated, transculturally adapted and validated for use in the Brazilian context (13)(14) . never, rarely, sometimes, most of the time, always).
The results were evaluated based on the performance of each item and composite. The items and composites with 75% of positive responses were considered strong and the ones less than 50% were considered weak (15) . In these rounds, the need to improve the writing of three questions of the HSOPS was observed, as stated in a previous analysis (16) . The term "event reports" in questions C1 and G1 was replaced by "notifications", term consolidated in Brazilian health services. Question A5 was written as "sometimes, the best patient care is not provided due to the excessive workload" instead of "staff (regardless of employment relationship) in this unit work longer hours than is best for patient care" (16) .   Table 1 demonstrate that most of the respondents were women with mean age of 39±11 years. More than 80% had direct contact with patients and 50% had graduate  and "nonpunitive response to error" (0.40). The majority of items (31/42) had negative responses, and only the item A4 -"in this unit, people treat each other with respect" had more than 70% of positive responses (data not presented).
Patient safety culture assessment in the work unit was positive for 36% of employees, according Table 3.
Of these, the majority filled out no reports in the last 12 months (78%) and 2% filled out six reports or more.  The instrument used had good reliability using the Cronbach's alpha in two thirds of the composites.
The strategy used to improve the understanding of some questions, as pointed by other researchers (16) , increased the reliability of the composites in relation to validation (14) . Another strategy would be the exclusion of low-performance questions (14) , however the instrument would have less items than the HSOPS originally developed. A new version of the HSOPS was validated for the Brazilian context and developed in an interface of electronic application (17) . The reliability of the instrument was high (α=0.92), possibly avoiding the interpretation limitations of the version applied in this investigation (14) .
The composite with lowest proportion of positive responses was the "nonpunitive response to errors", which also had the lowest reliability. In addition to this composite having a problematic aspect in institutions -the culpability culture -, it consisted of only negative questions, which required higher attention on interpretation and had less reliability in questionnaires (18) . Analyses of psychometric properties of HSOPS point to possible weaknesses in measuring the patient safety culture (19) . Composites with lower scores may reflect the writing of items and not necessarily the weaknesses in safety culture.
The result found in the composite "nonpunitive response to error" resembles studies carried out in intensive care in Brazil, in which this composite had the lowest proportion between composites of patient safety culture (14% to 29%) (8,(20)(21) . These lower positive responses were also observed in a systematic review with meta-analysis, in which seven of 11 studies included showed the lowest frequencies in the composite (22) .
Another factor that limits the results is the selection process by convenience of respondents, which decreases the representativeness of the hospital staff. The HSOPS ignores the recommendations on the sampling processthus, the questionnaire can be forwarded by e-mail and only the respondents are analyzed (15) . We know that recruitment of participants influences the results, especially in internet surveys (23) . On the other hand, almost half of all employees eligible to the survey were interviewed and included in this study.
Our findings proportionally had more positive responses than a study carried out in Southern Region of Brazil in 2016 with 59 participants of the health team of an intensive care unit, whose variation was from 14% to 47% of positive responses (21) . On the other hand, we had less positive responses than study carried out in 2014 in a teaching hospital of São Paulo with 88 health professionals, in which the safety culture reached proportions between 29% to 75% (nonpunitive response to error and supervisor/manager expectations and actions promoting patient safety, respectively) (8) .
Composites with better scores (organizational learning -continuous improvement, teamwork within units and supervisor/manager expectations and actions promoting patient safety) were similar to the strengths observed in Saudi studies, but had modest positive responses given other international studies (20,(24)(25)(26) .
Most respondents reported no adverse events in the past year. If on the one hand there is recognition of error and the importance of communicating it, on the other hand there is omission of it due to absence of communication (27) .
Previous studies had better results, with proportions of reports between 22% to 53% (8,(20)(21)(22)25) . National estimates indicate incidence of 5% of preventable adverse events during hospitalization (28) . The systemic approach to error, as opposed to the culpability, is strategic to improve the healthcare processes, covering the human nature involved in the processes and the complexity of health activities (29) .
Unsafe procedures must be redesigned and monitored to avoid the occurrence of the error, which results from latent and active faults in the system and not from an isolated individual.
Our findings result from the interviews with almost

Conclusion
The patient safety culture in the university hospital was evaluated as still fragile. To invest in systematic approach to errors, professional team and management is a priority to strengthen the patient safety at hospital.
The implementation and assessment of improvements in care, associated with the systematic measurement of the safety culture are strategies to increase the patient safety in hospital.