Organizational culture and climate for patient safety in Intensive Care Units

Objective: To assess the perception of health professionals about patient safety climate and culture in different intensive care units (ICUs) and the relationship between scores obtained on the Hospital Survey on Patient Safety Culture (HSOPSC) and the Safety Attitudes Questionnaire (SAQ). Method: A cross-sectional study conducted at a teaching hospital in the state of São Paulo, Brazil, in March and April 2014. As data gathering instruments, the HSOPSC, SAQ and a questionnaire with sociodemographic and professional information about the staff working in an adult, pediatric and neonatal ICU were used. Data analysis was conducted with descriptive statistics. Results: The scales presented good reliability. Greater weaknesses in patient safety were observed in the Working conditions and Perceptions of management domains of the SAQ and in the Nonpunitive response to error domain of the HSOPSC. The strengths indicated by the SAQ were Teamwork climate and Job satisfaction and by the HSOPC, Supervisor/ manager expectations and actions promoting safety and Organizational learningcontinuous improvement. Job satisfaction was higher among neonatal ICU workers when compared with the other ICUs. The adult ICU presented lower scores for most of the SAQ and HSOPSC domains. The scales presented moderate correlation between them (r=0.66). Conclusion: There were differences in perception regarding patient safety among ICUs, which corroborates the existence of local microcultures. The study did not demonstrate equivalence between the SAQ and the HSOPSC.


INTRODUCTION
Patient safety results from actions which avoid, prevent and improve adverse events and injuries resulting from the process of health care delivery. Patient safety differs from general aspects related to medical care given its focus on the more negative side of quality of care, i.e., on care that is actually harmful, and not only on care that is less than good (1) .
Discussions about accountability and how to deal with errors within the health system are complex and controversial. In complex organizations, mistakes are rarely isolated -active errors and latent conditions contribute to bad outcomes. Active errors are norm violations and, latent conditions can be illustrated by managerial decisions and pressure for production, among other factors. The focus is on making the system safe by increasing the number of safeguards, making it difficult for the holes in the Swiss cheese to align (the Swiss cheese model proposed by Reason (2) to analyze system accidents), instead of focusing only on perfecting individual human training (2) .
There are organizational influences, consisting of decisions coming from high management, which impact the safety of the entire health care delivery system. Working conditions consisting of supervision, communication, equipment, knowledge and skill are relevant to safety and are affected by organizational and managerial decisions. Furthermore, there are also the personal characteristics of health professionals and of the patients themselves, which also determine the variability of clinical practice. Creating a safety culture that permeates all actions in health care delivery is a great challenge for patient safety in health care services.
Organizational phenomena are influenced by organizational culture -intrinsic values and beliefs, which are often not perceived but expressed through the climate manifested in a given period. It is influenced by leadership and processes of change, both internal and external.
Organizational climate and culture offer overlapping perspectives for understanding the experiences of people in organizational environments (3) . Even though they are constantly used interchangeably, organizational culture and organizational climate are different concepts (4) . Organizational climate refers to a set of perceptions shared by individuals about the organizational environment in which they work (5) , while organizational culture represents the basic premises about the world and the values that guide life in the organization (3) .
Using the image of an iceberg, culture refers to phenomena that are under the surface, but that exert a powerful influence on system safety. The external expression of these phenomena can be perceived in the form of the organizational climate (6) . Therefore, climate is the more readily perceptible and measurable part of organizational culture, while culture, due to its more core location, portrays a deeper aspect, difficult to measure.
Safety culture is defined as the product of individual and group values and standards, which determines the commitment to and the style of an organization's health and safety management. Organizations with a positive safety culture seem to have open communication and a structure based on trust established among individuals, mediated by the recognition of the importance of safety and the adoption of preventive measures in the organizational context. Measuring patient safety culture can serve as a snapshot of the organizational situation and also to help investigate the impact of interventions (7)(8) . Patient safety climate is defined as the measurable characteristics of organizational culture, via perceptions and attitudes of individuals at a given time (4) .
The Safety Attitudes Questionnaire (SAQ) and the Hospital Survey on Patient Safety Culture (HSOPSC) are currently among the most used patient safety climate and culture assessment instruments, respectively. Both have been validated in Brazil and are available for use in health care services (9)(10) .
The SAQ is a questionnaire with 41 multiple-choice questions distributed among six domains: Teamwork Climate, Safety Climate, Perceptions of Management, Job Satisfaction, Working Conditions, and Stress Recognition. The SAQ -Short Form was validated in Brazilian Portuguese with a population of health professionals working in inpatient units at five Brazilian hospitals in three regions of the country. The author emphasizes that the SAQ may be adapted to any hospital area, for the items in each version of the instrument are the same (9) .
The HSOPSC consists of 42 items distributed among 12 dimensions: Teamwork within units, Supervisor/manager expectations and actions promoting safety, Organizational learning-continuous improvement, Executive management support for patient safety, Feedback and communication about error, Communication openness, Frequency of error reporting, Teamwork across units, Staffing, Handoffs and transitions between units and shifts and Nonpunitive response to error. In the validation study, the instrument was given to professionals working in inpatient units in two hospitals in the city of Rio de Janeiro, including staff from diagnostic and therapeutic support services (10) . The author recognizes the distinction between organizational climate and culture but chose to use climate and culture as synonyms in the Brazilian Portuguese version, as that is how they are used in the literature about patient safety in health organizations.
A study conducted in intensive care units (ICUs) in twelve hospitals in the United States assessed the reliability and predictive validity of the SAQ and the HSOPSC. The researchers concluded that the reliability of both instruments was similar, leaving the decision about which instrument to use up to researchers or the organization's head of patient safety (11) .
Most Brazilian institutions do not know what their patient safety climate or culture is and which are their greatest weaknesses and strengths in this regard. In order to investigate safety climate/culture, the present study used the SAQ and the HSOPSC.
The aim of this study was to assess the perception of health care professionals about patient safety climate and Santiago THR, Turrini RNT www.ee.usp.br/reeusp culture in intensive care units (ICU) and the relationship between the HSOPSC and SAQ surveys.

METHOD
This was an exploratory cross-sectional study developed in three ICUs at a public teaching hospital in the municipality of Sumaré, state of São Paulo, Brazil, in March and April 2014. There were 36 beds total in all three ICUs, 18 of which belonged to adult intensive care (AICU), 12 neonatal (NICU) and 6 pediatric (PICU). The hospital is accredited by the Brazilian National Accreditation Organization and by Accreditation Canada. The institution has a long history of improving the quality and safety of health care delivery and it is recognized both by accreditation programs and by other institutions in the region.
The study population comprised 197 professionals working in the ICUs both in health care delivery and administration capacities. The inclusion criteria for the professionals was to be part of the ICU work schedule and workers who had been at the unit for less than six months were excluded.
Based on a list of ICU staff members, research participants were given an envelope with an invitation to participate in the study, a free and informed consent form, the sociodemographic and professional data questionnaire, the SAQ and the HSOPSC. Health care professionals were instructed to return the envelope if they did not wish to participate in the study. After excluding the professionals with less than six months work experience in the unit and 20 questionnaires that were not completely filled out, the effective sample consisted of 88 participants.
The professional category variable for analysis was classified in three groups: Medical professional, nursing professional and other professionals. The nutritionist worked at both the AICU and PICU and, for purposes of analysis, her questionnaire was replicated for each one of the units.
The reliability of the SAQ and the HSOPC was determined using Cronbach's alpha. The survey items were analyzed by domain/dimension and as a complete scale using means, medians and standard deviations. For calculating SAQ scores, negatively worded items were reversed and then each item was submitted to the formula (m-1)x25 m, where m is the mean of the domain items in questions and range from 0 to 100 (8,11) . Scores above 75 indicated a strong area with regards to patient safety and scores under 50 represented areas of weakness (12) .
For the HSOPSC, after reversing negatively worded items, the percentage of positive responses for each dimension was calculated using the formula: % of positive answers for domain X=[number of positive answers for dimension X/total number of validated responses for dimension X items (positive, neutral and negative, excluding missing data)] x 100 (13) . Percentages higher than 75 represented satisfactory results regarding patient safety climate, while those ≤50 indicated weak dimensions.
An association was found between the scales and professional category, time working at the unit, and participant's age. For the qualitative variables, Person's chi-square test or Fischer's exact test were used; for quantitative variables, ANOVA, or the analysis of variation was used. Person's correlation was used to investigate the relationship between the SAQ and HSOPSC surveys. Level of significance was set at 5%.

RESULTS
The participants were distributed among the three ICUs as follows: 27 (30.7%) in the AICU, 26 (29.5%) in the PICU and 35 (39.8%) in the NICU. Most were female (94.3%) and age ranged from 21 to 58 years old, with a mean of 37.3 years (SD±9) and median of 36.5 years.
All three ICUs presented homogeneous distribution regarding professional category and gender, with a predominance of female nursing professionals. In these units, with the exception of the PICU medical coordination, unit coordinators carried out leadership roles concomitantly with direct health care delivery to patients.
The majority had a high school or university degree, with the exception of one (3.7%) participant from the PI-CU who only had completed elementary school. The AICU presented a greater percentage of professionals with university degrees (74.1%) with a statistically significant difference (p=0.029). Of the respondents with university degrees, 32 (36.4%) had completed graduate-level courses, 27 graduate specialization programs and five had complete masters or doctoral programs and worked at the AICU and PICU.
The largest group consisted of nursing professionals (71.6%), with the highest concentration in the NICU (82.9%) and the lowest in the AICU (55.6%), while the number of respondents in the "other professionals" category was higher in the AICU.
The mean time at the job was 6.7 years (SD±4.4), with a median of three years. Mean time working at the unit was 8.5 years (SD±4.3). Values for time since graduation were higher, with a mean of 10.8 years (SD±7.5) and a median of 10 years. The average weekly workload held by most nursing professionals was 36 hours.
No statistically significant difference among ICUs was observed regarding gender, age, professional category, time at the job in the hospital, time since graduation and workload. However, there was a statistically significant difference for time at the job in the unit (p=0.027), with the greatest mean in the NICU.
Cronbach's alpha resulted in good reliability for the SAQ and HSOPSC scales, with 0.85 and 0.90 respectively.
Regarding the SAQ scale, only the PICU and NICU obtained scores higher than 75, with the greatest score presented by the NICU. Analysis by domain demonstrated that the overall score for Job satisfaction was 81.2 for the three ICUs; the scores for the PICU and NICU were higher than 75 in the Teamwork climate domain. The lowest score, around 55, was observed in all the ICUs regarding the Perception of management -hospital administration domain and the Stress Analysis of HSOPSC scores in the ICUs demonstrated that only the Supervisor/manager expectations and actions promoting safety dimension obtained an overall score above 75%. The weakest dimension of safety culture was observed in the Nonpunitive response to error dimension, with 29.6%, followed by Communication openness, with 50.4% (Table 2). In general, the PICU presented the highest scores when compared to the other ICUs and its strongest dimensions were Organizational learning-continued improvement (78.9%) and Teamwork within units (82.5%). The difference regarding the last dimension was very significant when compared to the other ICUs. Hospital-related dimensions presented lower scores than unit-related ones. There was no statistically significant difference among types of ICUs.
Most professionals considered patient safety very good (69.3%), excellent (14.8%) and average (15.9%) In both the HSOPSC and SAQ, less than half of the dimensions/domains presented values considered "strong" for patient safety (over 75% of positive answers on the HSOPSC or over 75 points on the SAQ).
In the three ICUs, most participants declared not having reported any safety event in the last 12 months. Only the AICU declared notifying 21 or more events, and 14 participants did not answer the question (Figure 1).
There was a statistically significant difference between SAQ domains and gender. Recognition of stress was higher among men than women (p=0.004). There was also a correlation between age and Recognition of stress; the older the respondent, the higher the recognition of stress (p=0.049).
On comparing the HSOPSC scale and professional categories, nursing professionals presented a better perception toward Organizational learning-continuous improvement and Staffing than those in the "other professionals" category (p=0.034).
There was a positive correlation between some HSOP-SC domains and quantitative variables, such as Level of patient safety and weekly workload (p=0.020), in which the greater the weekly workload, the better the perception toward the level of patient safety. A positive correlation was also observed between the Staffing dimension and weekly workload (p=0.011).
Frequency of error reporting and time at the job in the hospital were negatively correlated. In other words, the longer the respondent had worked at the hospital, the higher frequency of error reporting (p=0.049).
Several SAQ and HSOPSC domains presented statistically significant correlations, albeit at moderate levels. The highest values of Pearson's correlation occurred between: Overall HSOPSC and SAQ Safety climate (r=0.69 and p<0.000), Overall HSOPSC and Overall SAQ (r=0.66 and p<0.000), HSOPSC Feedback and communication about error and Overall SAQ (r=0.58 and p<0.000), the HSOPSC

DISCUSSION
Both instruments, the SAQ and HSOPSC, are widely used in other countries in open units (wards) and closed units, like ICUs and surgical centers (14)(15)(16) .
The target population of this study consisted of multiprofessional teams working in ICUs, and the purpose was to understand the perceptions about the units held by different professionals and possible differences among ICUs categorized by age group. The results showed that the highest scores were obtained by the pediatric and neonatal ICUs.
The Recognition of stress domain of the SAQ addresses to what extent professionals recognize stress as a factor leading to situations of risk for patient safety. The NICU presented the highest recognition of stress compared to the other units.
There are studies that report that older and more experienced professionals tend to handle stress better than younger and less experienced ones (17) . In the present study, however, there was no difference regarding the age of professionals working in the three ICUs; nonetheless, the NICU contained professionals who had worked for the most time in intensive care.
The Recognition of stress domain of the SAQ differs from the other domains (18) . In contrast with the other five, which address respondents' perceptions toward their work environment and their relationship with the team and superiors, Recognition of stress is aimed at self-assessment, touching on how respondents perceive themselves when in stressful situations and how this relates to situations of vulnerability regarding patient safety in their work. According to the authors of the scale, recognition of stress is distinct from safety climate as a measurement and more studies should be conducted to assess the separation of this domain from the rest of the scale.
A study that applied the SAQ in 44 NICUs in the United States revealed that professionals who suffered from burnout tended to have worse perception toward safety culture, which indicates a relationship between stress and safety culture (19) .
None of the three ICUs scored lower than 50 on the SAQ domains, although only Teamwork climate and Job satisfaction proved to be strong areas in patient safety. The lowest scoring domain in all three ICUs was Perception of management, both at the unit and hospital level.
Of the ten ICUs analyzed in a study conducted in Australia and New Zealand, nine also presented weak scores in Perception of management; the highest score was in Teamwork climate, demonstrating that respondents were more positive with regards to teamwork collaboration than the other domains (20) . Similarly to other studies (20)(21) medical professionals scored higher on Teamwork climate than nursing professionals. It is not unusual for operational pro-fessionals to hold different perceptions than leadership regarding safety issues (22) .
A study comparing HSOPSC scores in different countries, namely Japan, Taiwan, and the United States, found low percentages of positive answers for Communication Openness in Taiwan (23) . According to the authors, of the three countries, Chinese culture is more strongly characterized by authoritarian leadership, which means that collaborators are potentially afraid to express their concerns and problems with regards to patient safety. Despite being an Asian country, Japan has a national policy of reporting errors and patient safety problems since 2002, differentiating it from other Asian countries.
Teamwork climate involves collaboration among professionals, but also the quality of communication and in this aspect, the HSOPSC identified weakness in the Communication openness dimension. In other words, there was not much room for discussing mistakes. This finding was corroborated by the low scores obtained in other dimensions that deal with communication.
The AICU presented weak scores for patient safety in the Handoffs and transitions between units and shifts domain, which means that the transfer of important information regarding patient safety can falter in the communication process. Another communication-related aspect is the HSOPSC domain Feedback and communication about error, which also presented weak scores in the safety security process.
The greatest weakness in all the ICUs was the Nonpunitive response dimension of the HSOPSC. This aspect touches on the theme of the culture of fairness and nonculpability. A just culture is a characteristic of safety culture in which people recognize the difference between what is acceptable and unacceptable (24) . The scores obtained on communication-related and error-reporting dimensions suggested the need for greater investments in these aspects of the units.
The NICU concentrated professionals with the greatest amount of time working in the unit. This fact may reflect greater job satisfaction that naturally leads to less professional turnover, given that the score on the SAQ domain Job satisfaction was also higher in the NICU. The Staffing dimension of the HSOPSC also scored higher in the NICU and PICU.
Regarding the HSOPSC, some dimensions obtained scores lower than 50% of positive answers, which, according to the authors of the instrument, indicates areas of weakness (13) . However, the overall assessment also demonstrated a predominance of positive answers in the average range, between 50% and 75%.
In the present study, the Organizational learning-continuous improvement dimension was best assessed by nursing professionals than by the other professionals category, probably due to continued education programs developed for nursing professionals that are commonly part of the nursing organizational dynamics. Leadership plays an important role in team development (6) .
The percentage of positive answers for the Teamwork within units dimension were significantly higher in the PICU when compared with the AICU and NICU. The PICU is part of the group of pediatric units in the hospital, which also consists of the Referral Pediatric Emergency Unit and the Pediatric Ward. Medical and nursing leaderships are common in the three areas. This probably contributes to the higher positive perception toward teamwork in this unit.
Concern about the quality and safety of health care delivery was recognized by professionals when scoring patient safety, a dimension of the HSOPSC, with a prevalence of excellent or very good evaluations for both ICUs, even though most of the dimensions scored between 50% and 70%.
Despite being accredited by two entities due to the quality of its service, the instruments demonstrated areas for potential improvement in the hospital. Being an accredited hospital can explain the scores close to 75% obtained on the Supervisor/manager expectations and actions promoting safety and Organizational learning-continuous improvement dimensions of the HSOPSC, but which was not observed in the other dimensions of the scale.
In Brazil, discussions about error in health services are relatively recent and the imputation of guilt on the professionals make mistakes is a strongly rooted practice, present since professional training. Relationships of power within health services are also observed between doctors and nurses and within the hierarchical structure of the organization between leadership and subordinates.
Furthermore, more bureaucratic organizational cultures are less inclined to carry out quality development than organizations with more equality regarding teamwork (16) .
Even though there were statistically significant results in the correlation between the HSOPSC and the SAQ, the highest correlation values found were of moderate strength, a fact that was also observed in another study that compared the two instruments (11) . This result may be related to the fact that most of the items in the HSOPSC domains address error directly, differently from the SAQ scale. The highest correlations were obtained between the total scores of the scale with the Safety climate domain of the SAQ.

CONCLUSION
Even though they were located at the same hospital, professionals working in the neonatal, pediatric and adult ICUs presented distinct recognitions of some aspects of patient safety culture and climate. In general, the PICU and NICU presented higher scores than the AICU, which suggests the need for investigating whether the affective relationship between professionals and children contributes to improving patient safety performance. Furthermore, the results corroborated those of other studies that report the presence of local microcultures within organizations.
In consonance with the global literature on patient safety, the results of the assessment of both the SAQ and HSOPSC in the present study pointed to some communication related aspects that need to be developed and present opportunities for improvement in this regard. Effort in this domain would probably make it easier for professionals to report important information for improving system safety. Strategies that favor the safe transfer of information in handoffs and transitions between units and shifts must also be developed.
This study did not aim to demonstrate whether the instruments are equivalent, despite the fact that both assess similar phenomena and possess moderate correlation between the scales as a whole and in some dimensions and domains.
reSUMen Objetivo: Evaluar la percepción de los profesionales de la salud sobre el clima y la cultura de seguridad del paciente en las unidades de cuidados intensivos (UCI) y la relación entre los instrumentos Hospital Survey on Patient Safety Culture (HSOPSC) y el Safety Attitudes Questionnaire (SAQ). Método: Estudio transversal realizado en un hospital universitario en el estado de São Paulo, Brasil, en marzo/abril de 2014. Se aplicaron los instrumentos HSOPSC y el SAQ, y un instrumento para la información socio-demográfica y profesional de los profesionales de la UCI de adultos, pediátrica y neonatal. Se utilizó la estadística descriptiva para el análisis de los datos. Resultados: las escalas mostraron buena fiabilidad. Se observaron importantes debilidades para la seguridad del paciente en la dimensión "condiciones de trabajo" y "percepciones de gestión" del SAQ y en la "respuesta no punitiva a los errores" del HSOPSC. Fortalezas en la SAQ fueron "el clima en el equipo de trabajo" y la "satisfacción en el trabajo" y en el HSOPSC las "expectativas y