Patient safety climate in general public hospitals in China: differences associated with department and job type based on a cross-sectional survey

Objective This study analysed differences in the perceived patient safety climate among different working departments and job types in public general hospitals in China. Design Cross-sectional survey. Setting Eighteen tertiary hospitals and 36 secondary hospitals from 10 areas in Shanghai, Hubei Province and Gansu Province, China. Participants Overall, 4753 staff, including physicians, nurses, medical technicians and managers, were recruited from March to June 2015. Main outcome measure The Patient Safety Climate in Healthcare Organisations (PSCHO) tool and the percentages of ‘problematic responses’ (PPRs) were used as outcome measures. Multivariable two-level random intercept models were applied in the analysis. Results A total of 4121 valid questionnaires were collected. Perceptions regarding the patient safety climate varied among departments and job types. Physicians responded with relatively more negative evaluations of ‘organisational resources for safety’, ‘unit recognition and support for safety efforts’, ‘psychological safety’, ‘problem responsiveness’ and overall safety climate. Paediatrics departments, intensive care units, emergency departments and clinical auxiliary departments require more attention. The PPRs for ‘fear of blame and punishment’ were universally significantly high, and the PPRs for ‘fear of shame’ and ‘provision of safe care’ were remarkably high, especially in some departments. Departmental differences across all dimensions and the overall safety climate primarily depended on job type. Conclusions The differences suggest that strategies and measures for improving the patient safety climate should be tailored by working department and job type.


Strengths and Limitations of this study
This study was carried out in Shanghai, Hubei Province, and Gansu Province, which represented high, middle, and low socioeconomic status levels located in the eastern, central and western regions of China, with a large sample of 4,176 staff from public general hospitals.
This study was first to investigate the variation in the perception of patient safety climates among different departments and job types and the interaction of them in the public general hospitals of China.
We used hierarchical linear models (HLMs) to examine the relationships between work departments and job types with the patient safety climate survey responses.
To depict differences by job type within selected clinical departments, we added the interaction of the clinical department and job type variables with the model and graphically displayed the relevant predictions.
Although our analyses represented an important advance over prior studies because we adjusted for important known individual and hospital characteristics, other characteristics that were not measured could play some role in distinguishing personnel by working department and job type.
The results from 54 public general hospitals in three regions might not be generalizable to all hospitals in China, although our sample size was large and represented public hospitals in high, middle, and low socioeconomic level regions and the eastern, central, and western regions of China.

Introduction
Patient safety is a core issue in health care services. The Chinese government and hospitals have made great efforts to strengthen patient safety climate and to improve patient safety performance. [1][2][3][4] Because patient safety climate is associated with positive outcomes, such as greater error reporting, 5 less adverse events, 6,7 lower mortality rates, 8 and lower readmission rates. 9 Measuring patient safety climate and understanding its variations can be helpful in targeting efforts to improve patient safety. 10-12 However, the climate can vary within organizations in different ways. Previous studies have indicated that patient safety climate of particular departments varies both across and within institutions. 13-17 Therefore, neglecting the patient safety climate at the unit level will mask important local variations, and measuring the patient safety climate of departments can identify important opportunities for improvement.
Previous literature has suggested that variations in patient safety climate may be related to the pace and complexity of the work performed in different work areas. 15,16 Most existing studies investigating the unit-specific climate have focused on measuring the climate in one or several particular types of departments with higher levels of intrinsic risk, such as the operating room (OR), intensive care unit (ICU), and/or emergency department (ED). [13][14][15][16]18 However, research concerning patient safety climates of other units, such as the pediatric, internal medicine, surgery, and medical technical departments, is deficient.
Some studies have measured the perception of patient safety climate among personnel by job type. 14,[19][20][21][22] In some studies, physicians demonstrated more positive perceptions of patient safety climate than nurses and other clinical personnel. 16,18,21 However, in our previous study conducted in the hospitals of Pudong New Area, Shanghai (one of the municipalities in China), we found that the physicians responded with more negative perceptions of patient safety climate than nurses. 22 With the variation in the perception of patient safety climates among different departments and job types, the improvement efforts of patient safety climate should not be limited to the interventions at hospital level, but at department level and among different types of employees. Currently, limited articles have focused on patient safety climates at department level and among different types of employees in hospitals in China.
This study aimed to analyze the differences in the perception of patient safety climate among different working departments and job types in 54 public hospitals located in the eastern, central and western regions of China. We selected not only intrinsically hazardous departments but also others, including the internal medicine, surgery, obstetrics and gynecology, pediatric, and medical technical departments. We specifically explored: ① in which departments, the staff perceived patient safety climate more negatively, ② in which job types, the staff rated lower scores across safety climate dimensions, and ③ whether differences by job type persisted across departments.

Survey instrument
In the study, we applied the Patient Safety Climate in Healthcare Organizations (PSCHO) tool 23-25 to measure patient safety climate. We selected the PSCHO for measurement, because it demonstrated good reliability and validity 22,24,25 and captured particular underlying characteristics of Chinese culture (i.e., "fear of blame" and "fear of shame") .
The PSCHO contains 12 dimensions and three categories (based on the hospital, work unit, and interpersonal contributions to the safety climate). 23  In this study, we added two items to the PSCHO: "Staff can freely voice their opinions on patient safety" in the dimension "Psychological Safety" and "We will analyze the accidents or unexpected events in a timely manner" in the dimension "Problem Responsiveness". The survey also asked informants to provide demographic information, including gender, age, education, working years, monthly income, working department and job type. Then, we selected 3 prefecture-level cities/counties (areas) in Hubei Province and Gansu Province representing high, middle, and low socioeconomic status levels within each province/municipality. In each area, two tertiary public general hospitals and four secondary public general hospitals were selected.
Because the tertiary hospitals in Shanghai were not evenly distributed among districts, 6 tertiary public general hospitals were selected in Shanghai to represent the tertiary hospitals owned by universities, the Shanghai government, or district governments. Additionally, 12 secondary public general hospitals were selected from 4 districts (A-D). Because District A was the largest district in Shanghai and comprised both urban and rural areas covering approximately 20% of the total Shanghai population, 6 secondary public general hospitals were selected in this We recruited a sample of 18 tertiary hospitals and 36 secondary hospitals from 10 areas in 3 regions of China in this study.

Data sources
For each selected hospital, general data (i.e., hospital level, number of beds, number of physicians, and number of nurses) were collected, and an employee questionnaire survey was conducted.
In the employee survey, we randomly sampled 10% of the managers and administrative staff (at least 15), 10% of the frontline physicians (at least 15), nurses (at least 15) and health technicians and the staff working in the medical auxiliary departments (at least 5). The term manager is a hospital or department director (including clinical departments and administrative offices), and the administrative staff are employees working in the administrative offices related to patient safety and medical quality without managerial positions. Frontline workers are employees without management responsibilities who interact directly with patients. Internal medicine department, surgical department, obstetrics and gynecology, ICU, and ED employees were recruited in this survey.

Survey Data Analysis Psychometric Analysis
We conducted item analyses and a confirmatory factor analysis to test the reliability and validity of the PSCHO revised for the Chinese context. The study demonstrated that eleven dimensions had high internal consistency (Cronbach's α coefficients ranging from 0.77 to 0. 93  Bentler-Bonett Normed Fit Index and the Non-Normed Index values were all greater than 0.9. The adjusted goodness of fit (AGFI) was 0.83. However, the Goodness of Fit Index (GFI) was 0.84, which was slightly lower than the criterion for this index (GFI >0.85). Overall, the constructive validity of the PSCHO revised for the Chinese cultural context in this study was acceptable. 26,27

Statistical Analysis
We used the percentage of "problematic responses" (PPRs) to measure the patient safety climate. A rating less than 3 for a positive statement or greater than 3 for a negative statement was identified as a problematic response. A lower PPR is indicative of a better perception of the safety climate. This scoring method identifies areas of non-uniformity in safety that are of potential concern and may benefit from interventions to improve the safety climate. [21][22][28][29][30] We computed the PPR for each safety climate dimension with each item in the dimension weighted equally. We also calculated the average PPR for all questions in the survey as a summary statistic, which we referred to as the "overall safety climate". These percentages were calculated as the averages of all responses received.
Comparisons among working departments and job types were calculated separately for the respondents who indicated that they worked in any of the 8 working departments (internal medicine, surgery, obstetrics and gynecology, pediatrics, ICU, ED, medical technical departments (MTD) and others) and then for employees by different job type (physicians, nurses, administrative staff, hospital or department directors and others).
We used a hierarchical linear model (HLM) to examine the relationships between work areas and worker characteristics with the patient safety climate survey responses. To account for the association between hospital membership and the safety climate, which was significant both overall and for each dimension (p<0.0001), we included the respondents' hospital as a random factor in the form of a 2-level random intercept analysis. We estimated HLMs in which each safety climate dimension and the overall safety climate were the dependent variables and the working department and job type measures were the independent variables. All models were controlled for variables describing other respondent characteristics (i.e., gender, age, education, length of working years, and monthly income) and hospital characteristics (i.e., region, hospital level, bed size, and doctor-nurse ratio).
To indicate differences in perceptions among relative groups of personnel, we used regression coefficients to predict the PPR for each dimension and overall by working department and job type, with variables representing other individual characteristics held constant at their means and setting variables representing hospital characteristics to the mean for the participating hospitals. Thus, the predictions indicate average respondents in average hospitals for each personnel category. To depict differences by job type (focusing on physicians with various professional titles and other employees) within selected clinical departments, we added the interaction of the clinical department and job type variables with the model and graphically displayed the relevant predictions. We focused on physicians in the interaction analysis because they were the most important staff in the healthcare service and we intended to investigate whether there were differences among chief physicians, attending physicians, residents or below and other personnel within a particular department.
The respondents were predominantly female (66.31%) and older than 45 years (52.43%). Nearly 52% of the respondents had worked within their hospitals for 10 years or more (Appendix 1).

Perceptions of the Safety Climate
In this multi-region study, the mean PPR of the overall safety climate among all 54 hospitals was 9.00%. The dimension with the highest PPR was "fear of blame and punishment" (64.81%), and the dimension with the lowest PPR was collective learning (2.10%) ( Table 1).

Variations among Working Departments
After controlling for the individual and hospital characteristics, the predicted PPR for the overall safety climate was the highest among the respondents working in EDs (9.63%), followed by the medical technical, pediatric and ICU departments (9.52%, 9.46% and 9.19%, respectively) and the lowest among the respondents in obstetrics and gynecology departments (6.82%) ( Table 2).
The PPRs among the respondents in the EDs, ICUs, medical technical departments, and pediatric departments were also higher across the dimensions. The PPRs in the EDs were significantly higher for the senior managers' engagement, unit recognition and support for safety efforts and collective learning. The PPRs in the ICUs were notably higher for the senior managers' engagement and psychological safety but relatively lower for fear of blame and punishment. The PPRs in the medical technical departments were considerably higher for 5 dimensions. The respondents in pediatrics responded more negatively on organizational resources for safety, collective learning and psychological safety.
Conversely, the respondents from anesthesiology and ORs and the surgery departments evaluated almost all dimensions more positively except fear of shame.
Additionally, the PPRs for the obstetrics and gynecology respondents were universally lower across dimensions, especially for fear of blame and punishment (Table 2).

Variations among Job Types
After adjusting for the personnel and hospital characteristics, the PPR reported by the physicians was the highest for the overall safety climate (10.19%) and the PPR among the managers was the lowest (7.45%) ( Table 3).
The PPRs of the physicians were consistently higher among the dimensions, especially for "organizational resources for safety", "unit recognition and support for safety efforts", "psychological safety", and "problem responsiveness". Additionally, the responses of the managers to the dimensions seemed to be more positive.
Although no significant differences were detected among the various job types for "fear of shame", "fear of blame and punishment", and "provision of safe care", the PPR among the nurses was slightly higher for "fear of shame" (Table 3).

Variations within Working Departments among Physicians with Various Titles and Other Personnel
According to the models in which we evaluated both the working department and job type ((associate) chief physicians, attending physicians, residents or below and other personnel), working department differences across all the dimensions of safety Regarding the overall safety climate, (associate) chief physicians and residents or below in the pediatric departments responded with significantly higher PPRs (16.88% and 15.20%, respectively), attending physicians in the ICUs responded with remarkably higher PPRs (17.58%), and residents or below in the medical technical departments responded with dramatically higher PPRs than the other physicians and personnel (17.94%).
Additionally, attending physicians in the ICUs reported the highest PPRs among the compared staff regarding the senior managers' engagement, overall emphasis on patient safety, collective learning, and problem responsiveness (all above 10%), whereas (associate) chief physicians in the ICUs reported remarkably higher PPRs than the other physicians and staff for provision of safe care (68.27%). Residents or below in the EDs reported the most negative perceptions for unit safety norms and unit recognition and support for safety efforts among the respondents (12.52% and 25.88%, respectively). The PPRs of the chief and associate chief physicians in the pediatric departments were considerably higher than those of the other physicians and staff on organizational resources for safety (36.55%) and unit recognition and support for safety efforts (21.04%). The PPRs of the chief and associate chief physicians and residents or below in the pediatric departments were significantly higher than those of others for collective learning (both nearly 10%), psychological safety (both above 20%), and fear of shame (32.80% and 30.46%, respectively).

Physicians in the pediatric departments universally demonstrated prominently higher
PPRs for fear of blame and punishment. Attending physicians in the internal medicine departments perceived the highest PPR on organizational resources for safety. Residents or below in the medical technical departments responded most negatively on the overall emphasis on patient safety, unit safety norms, unit recognition and support for safety efforts and collective learning (13.20%, 14.21%, 24.52%, and 13.77%, respectively). Chief and associate chief physicians and residents or below responded significantly more negatively on psychological safety (26.88% and 25.51%, respectively) and problem responsiveness (15.16% and 22.80%, respectively). Additionally, the chief physicians responded with notably lower PPR for fear of blame and punishment (32.26%) and with significantly higher PPR for provision of safe care (40.37%) than the other physicians and staff ( Figure   1). Furthermore, the personnel in the obstetrics and gynecology departments universally perceived more positively than that in the other departments on fear of blame and punishment, especially chief and associate chief physicians did. However, higher PPRs among respondents in the surgery departments and anesthesiology and ORs for fear of shame did not depend on various job types.

Hospital Patient Safety Climate
The overall perception of the patient climate in this survey was relatively good (PPR=9%) But, substantial attention should be paid to the dimensions "fear of blame and punishment" (65%), "fear of shame" (20%),"provision of safe care" (16%) and "organizational resources for safety" (10%) according to the HRO theories. 31

General Safety Climate Variations by Department and Job Type
The results highlight differences in perceptions of patient safety climate among working departments and job types. We found that personnel in the pediatric, ICU, ED, and medical technical departments perceived substantially lower levels of safety climate than the staff in other departments. The PPRs given to the safety climate by physicians were systematically higher than those by other staff in this study, while it was opposite in Western studies. 13,16,19,32 Staff in the anesthesiology and OR departments and obstetrics and gynecology departments responded more positively across almost all dimensions of safety climate and overall. More resources may have been delivered and greater efforts made to overcome safety hazards in the anesthesiology department and ORs. 15,16 Furthermore, a notably higher percentage of personnel in the surgery departments problematically perceived the dimensions "fear of shame" and "fear of blame and punishment" more negatively than the other departments. Generally, surgeons have greater social identity and self-identity.
However, medical errors and/or accidents occurring in surgery are usually more severe than those in other medical departments. Therefore, staff in surgery will bear more pressure and face more serious consequences.

ICUs and EDs Should Receive More Attention
Working in ICUs and EDs is associated with a higher level of risk, complexity, difficulty, faster pace and lower predictability. 15,16 These work areas are intrinsically hazardous, and personnel working in these areas are prone to encounter a high workload and pressure that can result in burnout. [33][34][35][36] Previous studies have indicated that burnout in healthcare providers may lead to a reduction in patient safety. [37][38][39] Nahrgang at al. generally argued 40 that the mental and physical energy levels in burnout employees lessened safe work behaviors and thus increased the likelihood of errors and work-related injuries. The results of this survey suggest that more attention should continue to be paid to ICUs and EDs.
We also found that attending physicians in the ICUs responded more negatively, especially on the dimensions senior managers' engagement, overall emphasis on patient safety, collective learning, and problem responsiveness, whereas residents in the EDs responded negatively on the dimensions unit safety norms, unit recognition and support for safety efforts. These health care workers are important personnel at the front line in the corresponding departments who perhaps experience more safety problems. 41 Hospital managers and policy makers should carefully consider their opinions.

Substantial Concerns in the Pediatric Departments
Based on this investigation, the personnel in pediatrics responded with relatively higher PPRs in many dimensions and regarding the overall safety climate, especially for organizational resources, which was an alarming finding. This result was consistent with the increasingly severe shortage of pediatricians in China reported in some articles and new sources. [42][43][44] Statistical analyses showed that the number of pediatricians in China increased by only 5,000 from 1995-2010, 45 and the current number of pediatric doctors per 1,000 kids was only 0.5, which was one-third of the ratio in the US. 46 Moreover, the issue will become more problematic after implementation of the Two-Child policy in China. Pediatricians are constantly leaving medical practice especially in the primary healthcare institutions, and a lot of new medical graduates are not willing to join. 47 In this study, chief and associate chief pediatricians were most worried about the patient climate. The healthcare human resource shortage would be very likely to impact health care quality and patient safety negatively.

Medical Technical Departments Need Attention
Another meaningful finding of our study was that the perceptions of patient safety climate among respondents in the medical technical departments were relatively worse than those in other departments regarding several dimensions of safety climate and overall, particularly among the residents in the medical technical departments.
Those residents responded with significantly higher PPRs for overall emphasis on patient safety, unit safety norms, unit recognition and support for safety efforts, collective learning, psychological safety, problem responsiveness (6 of 12 dimensions), and overall safety climate than the other staff. Hospital managers might not pay sufficient attention to the medical technical departments compared with the clinical departments, especially the residents' training and motivation in these departments.

Limitations
First, the data were based on self-reporting, which might involve recall/report bias.
Second, because this study was a cross-sectional study, we could not rule out the potential for omitted variables. Although our analyses represented an important advance over prior studies because we adjusted for important known individual and hospital characteristics, other characteristics that were not measured could play some role in distinguishing personnel by working department and job type. Third, the results from 54 public general hospitals in three regions might not be generalizable to all hospitals in China, although our sample size was reasonably large and represented public hospitals in high, middle, and low socioeconomic level regions and the eastern, central, and western regions of China.

Notes:
The first type of models include independent variables: working departments and job types (frontline physicians, frontline nurses, managers, medical technicians, and others), and, in the second type of models only, interactions of different titles of physicians and various working departments. In addition, models control for    This study was first to investigate the variation in the perceived level of patient 7 safety climates among different departments and job types and the interaction of 8 them in the public general hospitals of China. 9 We used 2-level random intercept models to examine the relationships between 10 work departments/job types and the patient safety climate survey responses and  other characteristics that were not measured could play some role in 18 distinguishing personnel by working department and job type.

19
The results from 54 public general hospitals in three regions might not be 20 generalizable to all hospitals in China, although our sample size was large and 21 represented public hospitals in high, middle, and low socioeconomic level 22 regions and the eastern, central, and western regions of China. In this study, we added two items to the PSCHO: "Staff can freely voice their 6 opinions on patient safety" in the dimension "Psychological Safety" and "We 7 analyze accidents or unexpected events in a timely manner" in the dimension 8 "Problem Responsiveness". These items were added because they reflect a more 9 general psychological safety climate (not specific to certain concerns) and timely 10 responses to adverse events in hospitals, respectively. The survey also asked 11 informants to provide demographic information, including gender, age, education, 12 working years, monthly income, working department and job type.   We recruited a sample of 18 tertiary hospitals and 36 secondary hospitals from 10 11 areas in 3 regions of China in this study.

13
Data sources 14 For each selected hospital, general data (i.e., hospital level, number of beds, number 15 of physicians, and number of nurses) were collected, and anonymous, paper-based, 16 self-administered employee questionnaires were distributed and collected by trained 17 coordinators in the surveyed hospitals or regions according to our study design.

19
In the employee survey, we randomly sampled 10% of the managers and 20 administrative staff (at least 15), 10% of the frontline physicians (at least 15), nurses 21 (at least 15) and health technicians and the staff working in the medical auxiliary    areas of non-uniformity in safety that are of potential concern and may benefit from 2 interventions to improve the safety climate. 21 We computed the PPR for each safety climate dimension with each item in the 5 dimension weighted equally. We also calculated the average PPR for all questions in 6 the survey as a summary statistic, which we referred to as the "overall safety 7 climate". These percentages were calculated as the averages of all responses received.

8
Comparisons among working departments and job types were calculated separately 9 for the respondents who indicated that they worked in any of the 9 types of working 10 departments (internal medicine, surgery, obstetrics and gynecology, pediatrics, ICU,

11
ED, anesthesiology and operating room (OR), clinical auxiliary departments (CAD) 12 and others) and then for employees by different job type (frontline physicians, 13 frontline nurses, medical technicians, managers and others). To test the appropriateness of using a two-level model to account for the nesting of

5
In this multi-region study, the mean PPR of the overall safety climate among all 54 6 hospitals was 9.00%. The dimension with the highest PPR was "fear of blame and 7 punishment" (64.81%), and the dimension with the lowest PPR was collective 8 learning (2.10%) ( Table 1).    The PPRs of the physicians were consistently higher among the dimensions, 3 especially for "organizational resources for safety", "unit recognition and support for 4 safety efforts", "psychological safety", and "problem responsiveness". Additionally, 5 the responses of the managers to the dimensions seemed to be more positive. ( Figure   6 2, Appendix C-D).  In this survey, the result revealed that the overall perception of the patient climate 1 was relatively good (PPR=9%). But, substantial attention should be paid to the 2 dimensions "fear of blame and punishment" (65%), "fear of shame" 3 (20%),"provision of safe care" (16%) and "organizational resources for safety" (10%) 4 according to the HRO theories. 39 The high prevalence of "fear of blame and 5 punishment" and "fear of shame" may be attributed to inappropriate systems of 6 performance assessments and rewards, hierarchical management style, bad 7 doctor-patient relationships as well as quintessential Chinese notion of "face". Since 8 hierarchical management style and quintessential notion of "face" are often seen in 9 Chinese culture, our results of high PPRs in "fear of blame and punishment" and in 10 "fear of shame" may exist in other hospitals with dominant Chinese culture. the surgery departments problematically perceived the dimensions "fear of shame" 22 and "fear of blame and punishment" more negatively than the other departments.

23
Generally, surgeons have greater social identity and self-identity. However, medical 24 errors and/or accidents occurring in surgery are usually more severe than those in 25 other medical departments. Therefore, staff in surgery will bear more pressure and 26 face more serious consequences.  The results showed that the PPRs by physicians were systematically higher than This finding enlightened that hospital managers and unit directors should set up and 12 continuously improve internal managerial system to stimulate frontline physicians to 13 report or share their experiences and information on patient safety and healthcare 14 quality, and to facilitate them to participate in related improving projects proactively.

15
The measures to appoint frontline physicians to serve as unit quality controllers, to Besides, the results revealed that the perceptions by managers were relatively 22 consistent with frontline workers on many dimensions, which was different with 23 many previous researches. 22 with our previous study in Shanghai 2013. 22 The clinical departments' directors 2 in China are physicians who also provide healthcare for patients in the front line.  and optimize systems in such a way as to protect ICU and ED patients from We also found that attending physicians in the ICUs responded more negatively, 14 almost all above 10% PPRs, whereas attending physicians and residents in the EDs  First, the data were based on self-reporting, which might involve recall/report bias.

5
Second, because this study was a cross-sectional study, we could not rule out the 14 15

16
Our study highlighted differences in perceptions of patient safety climate among and 17 within working departments and job types that were not previously documented in 18 China.     punishment" were universally significantly high, and PPRs of "fear of shame" and 21 "provision of safe care" were remarkably high, especially in some departments. 22 Department differences across all dimensions and the overall safety climate 23 primarily depended on job type. China's public general hospitals. 9 We used 2-level random intercept models to examine the relationships between 10 working departments/job types and the patient safety climate survey responses.

11
To depict differences by job type within selected working departments, we     "strongly disagree" to "strongly agree," with a neutral midpoint.

5
In this study, we added two items to the PSCHO: "Staff can freely voice their 6 opinions on patient safety" in the dimension "Psychological Safety" and "We 7 analyze accidents or unexpected events in a timely manner" in the dimension 8 "Problem Responsiveness." These items were added because they reflect a more 9 general psychological safety climate (not specific to certain concerns) and timely 10 responses to adverse events in hospitals, respectively. The survey also asked 11 informants to provide demographic information, including gender, age, education, 12 working years, monthly income, working department and job type.   We recruited a sample of 18 tertiary hospitals and 36 secondary hospitals from 10 11 areas in 3 regions of China.

13
Data sources 14 For each selected hospital, general data (i.e., hospital level, number of beds, number 15 of physicians, and number of nurses) were collected, and anonymous, paper-based, 16 self-administered employee questionnaires were distributed and collected by trained 17 coordinators in the surveyed hospitals or regions according to our study design.

19
In the employee survey, we randomly sampled 10% of the managers and  Sampled employees who were willing to participate in our study filled the 10 questionnaires and returned them to the coordinators. Psychometric analysis 14 We conducted item analyses and confirmatory factor analysis to test the reliability  Statistical analysis 4 We used the percentage of "problematic responses" (PPRs) to measure the patient 5 safety climate. A rating of less than 3 for a positive statement or greater than 3 for a 6 negative statement was identified as a problematic response. A lower PPR is 7 indicative of a better perception of the safety climate. This scoring method identifies 8 areas of non-uniformity in safety that are of potential concern and may benefit from 9 interventions to improve the safety climate. [21,22,[33][34][35] 10 11 We computed the PPR for each safety climate dimension with each item in the 12 dimension weighted equally. We also calculated the average PPR for all questions in 13 the survey as a summary statistic, which we referred to as the "overall safety 14 climate." These percentages were calculated as the averages of all responses received.

15
Comparisons among working departments and job types were calculated separately 16 for respondents who indicated that they worked in any of the 9 types of working 17 departments (internal medicine, surgery, obstetrics and gynecology, pediatrics, ICU, 18 ED, anesthesiology and OR, clinical auxiliary departments (CADs) and others) and   To test the appropriateness of using a two-level model to account for the nesting of   The respondents were predominantly female (66.31%) and older than 45 years 8 (52.43%). Nearly 52% of the respondents had worked within their hospitals for 10 9 years or more (Appendix A).

12
In this multi-region study, the mean PPR of the overall safety climate among all 54 13 hospitals was 9.00%. The dimensions with the three highest PPRs were "fear of 14 blame and punishment" (64.81%), "fear of shame" (20.42%) and "provisions of safe 15 care" (16.31%), and the dimension with the lowest PPR was "collective learning"

13
After adjusting for other personnel and hospital characteristics, the PPR reported by 14 physicians was highest for the overall safety climate (10.19%, 95%CL: 9.52%, 15 10.86%), whereas the PPR among managers was the lowest (7.45%, 95%CL: 5.47%, 16 9.43%). The PPRs for "fear of shame" and "fear of blame and punishment" were 17 generally high among staff with various job types (Table 3).

21
The survey results revealed that the perception of overall patient climate was 22 relatively good (PPR=9%). However, substantial attention should be paid to the    more severe than those in other working departments. Therefore, staff in these 2 departments will experience more pressure and face more serious consequences.  Besides, the results also revealed that managers' perceptions were relatively 26 consistent with those of frontline workers on many dimensions but differed from 27 those reported in many previous studies. [22,24,34,42]  China are physicians who also provide healthcare for patients on the front lines.   We also found that attending ICU physicians responded more negatively; almost all 20 PPRs were above 10%. These health care workers are important personnel on the 21 front lines in their departments and perhaps experience more safety problems.
[43] 22 Hospital managers and policy makers should carefully consider their opinions.

11
To depict differences by job type within selected working departments, we    "strongly disagree" to "strongly agree," with a neutral midpoint.

20
After controlling for the hospital and other individual characteristics, the top four 21 predicted PPRs for the overall safety climate were in the ED (9.63%), CADs 22 (9.52%), pediatrics (9.46%) and the ICU (9.19%); the two lowest predicted PPRs 23 were in the obstetrics and gynecology department (6.82%) and anesthesiology and 24 the OR (7.44%).