YJCRC-50869; No of Pages 7. Journal of Critical Care (2011) xx, xxx xxx

Transcription

1 YJCRC-50869; No of Pages 7 Journal of Critical Care (2011) xx, xxx xxx 1 2 Organizational and safety culture in Canadian intensive 3 care units: Relationship to size of intensive care unit and 4 physician management model, Q1 5 Peter M. Dodek MD, MHSc a,,1, Hubert Wong PhD b,1, Danny Jaswal MD a,1, Q3 6 Daren K. Heyland MD, MSc c,1, Deborah J. Cook MD, MSc d,1, 7 Graeme M. Rocker DM, MHSc e,1, Demetrios J. Kutsogiannis MD f,1, Craig Dale RN g,1, 8 Robert Fowler MD, MSc h,1, Najib T. Ayas MD, MPH a,1 Q2 a 9 Center for Health Evaluation and Outcome Sciences and Division of Critical Care Medicine, 10 Providence Health Care and University of British Columbia, Vancouver, British Columbia, Canada V6Z 1Y6 b 11 Center for Health Evaluation and Outcome Sciences and School of Population and Public Health, 12 University of British Columbia, Vancouver, British Columbia, Canada c 13 Department of Medicine, Queen's University, Kingston, Ontario, Canada d 14 Departments of Medicine, Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, Ontario, Canada e 15 QEII Health Sciences Centre, Dalhousie University, Halifax, Nova Scotia, Canada f 16 Division of Critical Care Medicine, University of Alberta, Edmonton, Alberta, Canada g 17 Trauma, Emergency and Critical Care Program, Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada h 18 Departments of Medicine and Critical Care Medicine, Sunnybrook Health Sciences Centre, 19 University of Toronto, Ontario, Canada Keywords: 29 Abstract 23 Organizational culture; 30 Purpose: The objectives of this study are to describe organizational and safety culture in Canadian 24 Safety management; 31 intensive care units (ICUs), to correlate culture with the number of beds and physician management 25 Critical care; 32 model in each ICU, and to correlate organizational culture and safety culture. 26 Intensive care unit 33 Materials and Methods: In this cross-sectional study, surveys of organizational and safety culture were administered to 2374 clinical staff in 23 Canadian tertiary care and community ICUs. For the completed surveys, scores were calculated for each of 34 domains. Average domain scores for each ICU 36 were correlated with number of ICU beds and with intensivist vs nonintensivist management model. 37 Domain scores for organizational culture were correlated with domain scores for safety culture. Institution where work was done: Center for Health Evaluation and Outcome Sciences, Providence Health Care and University of British Columbia, Vancouver, British Columbia, Canada. Research support: Canadian Institutes of Health Research, Michael Smith Foundation for Health Research, Canadian Researchers at the End of Life Network. Corresponding author. Tel.: ; fax: address: pedodek@interchange.ubc.ca (P.M. Dodek). 1 for the Canadian Researchers at the End of Life Network /$ see front matter 2011 Published by Elsevier Inc. doi: /j.jcrc

2 38 Results: Culture domain scores were generally favorable in all ICUs. There were moderately strong 39 positive correlations between number of ICU beds and perceived effectiveness at recruiting/retaining 40 physicians (r = 0.58; P b.01), relative technical quality of care (r = 0.66; P b.01), and medical director 41 budgeting authority (r = 0.46; P =.03), and moderately strong negative correlations with frequency of 42 events reported (r = 0.46; P =.03), and teamwork across hospital units (r = 0.51; P =.01). There 43 were similar patterns for relationships with intensivist management. For most pairs of domains, there 44 were weak correlations between organizational and safety culture. 45 Conclusion: Differences in perceptions between staff in larger and smaller ICUs highlight the 46 importance of teamwork among units in larger ICUs Published by Elsevier Inc P.M. Dodek et al Introduction 50 Organizational culture has been defined as a group's 51 shared assumptions, values, and norms [1]. Developed 52 through formal training or socialization, culture defines 53 valid ways to address problems and is passed on to new team 54 members as a resource governing internal and external 55 stakeholder interaction. Culture is the invisible force 56 behind the tangibles and observables in any organization, a 57 social energy that moves people to act. Culture is to an 58 organization what personality is to the individual a hidden, 59 yet unifying, theme that provides meaning, direction, and 60 mobilization [2]. Cultures help members deal with uncer- 61 tainty, on both an individual and collective basis, by defining 62 what is important in a given situation, providing guidance on 63 how individuals should perceive situations and interact with 64 each other, and providing members with accepted ways of 65 expressing and affirming beliefs, values, and norms [3]. That 66 is, organizational culture provides its members with 67 direction, purpose, and perspective. Safety culture is a 68 specific form of organizational culture, which addresses the 69 context related to achieving safe outcomes for patients [4]. 70 Although much has been written about the relationship 71 between organizational and safety culture and clinical 72 outcomes in health care, there are only a few empirical 73 studies that demonstrate this relationship [5,6]. Most of this 74 work has focused on the value of teamwork [7]. In critical 75 care medicine, knowledge about the relationship between 76 organizational and safety culture and outcomes is limited [ ] and may not be generalizable across a variety of health 78 care settings and different types and sizes of intensive care 79 units (ICUs). 80 Recently, there has been an interest in the association 81 between size of ICU and ICU outcomes [11,12] and the 82 association between physician management model and ICU 83 outcomes [13-15]. These associations may be related to 84 differences in organizational culture. Therefore, our hypoth- 85 esis was that there are differences in organizational and 86 safety culture among ICUs of different size and structure. If 87 organizational culture is related to safety culture, then it is 88 possible that changes in one of these measures will influence 89 the other. Therefore, our second hypothesis was that there are relationships between domains of organizational culture and safety culture. The objectives of this study were to describe organizational and safety culture in a sample of Canadian ICUs; to investigate the relationship between culture, the number of beds, and physician management model in each ICU; and to examine the relationship between domains of organizational culture and safety culture. 2. Methods 2.1. Participating ICUs Leaders of 23 geographically dispersed Canadian ICUs (each in a separate hospital) volunteered to participate in this study, which was part of a study of the relationship between organizational culture and family satisfaction in ICUs. This study was conducted between 2006 and 2008 in 17 ICUs in British Columbia, 5 ICUs in Ontario, and 1 ICU in Alberta Design and surveys This was a cross-sectional study. Based on a review of quantitative tools that have been used to measure organizational culture in ICUs, the Organization and Management of Intensive Care Units (short form) survey [16] was used. Dimensions of organizational culture measured in the 22 domains in this survey include physician-to-physician relationships, physician to other health professional relationships, general relationships and communication, nursing leadership, physician leadership, teamwork, perceived effectiveness, managing disagreements among physicians or among other health professionals, managing disagreements between physicians and other health professionals, authority, and general job satisfaction. This survey instrument has been found to have adequate reliability and validity [17]. Furthermore, higher scores on domains from this survey have been shown to correlate with lower risk-adjusted length of stay, lower nurse turnover, higher perceived technical quality of care, and higher perceived ability to meet family members' needs in ICUs [16]. Based on the original survey

3 Relationship to size of ICU and physician management model and with permission of the survey developers, 2 versions of 126 this survey were distributed 1 for physicians and 1 for all 127 other health professionals. There were no differences in the 128 content, only the perspective of the respondent. Domains that 129 include the phrases within group or between groups refer 130 to each of these 2 professional groups. 131 To measure safety culture, the Hospital Survey on Patient 132 Safety Culture prepared for the US Agency for Healthcare 133 Research and Quality [18] was used. The 12 domains of 134 safety culture measured in this survey were supervisor/ 135 manager expectations and actions promoting safety, organi- 136 zational learning/continuous improvement, teamwork in the 137 unit, teamwork across hospital units, communication open- 138 ness, feedback, and communication about error, nonpunitive 139 response to error, staffing adequacy, hospital management 140 support for patient safety, hospital handoffs and transitions, 141 overall perceptions of safety, and frequency of event 142 reporting. This survey has moderately strong validity and 143 internal consistency [19] and has been used in nearly hospitals [20]. We recognize that the surveys described 145 above measure organizational and safety climate, the 146 perceptions of culture. However, given the names of these 147 surveys and common usage, the terms organizational culture 148 and safety culture were used in this study. 149 After contact with the physician and nurse leaders of each 150 ICU, our research team invited all dedicated ICU staff 151 (physicians, nurses, respiratory therapists, pharmacists, 152 dieticians, pastoral care and social workers, and unit 153 coordinators) within each unit to complete these self- 154 administered surveys. Specifically, the research assistant 155 and the principal investigator (or coinvestigators for ICUs in 156 Edmonton, Hamilton, Toronto, Sudbury, and Kingston) 157 visited the ICUs to explain the purpose of the study, the 158 anonymity of the data collected, and the potential use of the 159 findings to the ICU leaders and staff members. A cover letter 160 that explained these issues and the survey instrument were 161 distributed to each staff member within each ICU. Complet- 162 ed surveys were collected centrally within each ICU over a 163 period of 4 to 6 weeks, then mailed to the principal 164 investigator by the local ICU leaders. There were no personal 165 identifiers on any survey. 166 During the period when we disseminated organizational/ 167 safety culture surveys to ICU staff, the medical and 168 nursing leaders of each ICU were asked to complete a 169 brief questionnaire, which described the organizational 170 structure of their ICU, including the number of open and 171 funded ICU beds and the physician management model. 172 If the respondents indicated that decisions regarding 173 admission and overall coordination of the care of ICU 174 patients were made by an ICU physician (a physician 175 who has been formally trained in critical care medicine 176 and who provides dedicated care to patients in the ICU), 177 then the physician management model in those ICUs was 178 called intensivist staffing. If any other model of care was 179 used, the physician management model was called 180 nonintensivist staffing Analysis The response rate for each ICU was expressed as the number of culture surveys returned divided by the number of surveys distributed in each ICU. Responses to each item were in the form of a Likert scale: strongly disagree, disagree, neutral, agree, and strongly agree. The frame of each question was aligned, and scores for each domain in each culture survey were calculated using the methods described by the respective survey developers [16,21] such that a score of 1 was the lowest (most negative perspective) and a score of 5 was the highest (most positive perspective). Median and interquartile ranges for each domain score for each ICU were plotted from smallest to largest ICU. To assess whether responses differed between physicians and other health professionals, the mean scores for each domain for each ICU separately for physician-respondents and for other health professionals were calculated. Then, for each domain, the physician mean scores were compared with the other health professional mean scores using a paired t test (n = 23, ICUs). Mean scores for each domain for each ICU were related to the number of ICU beds for that ICU using the Pearson correlation coefficient, and mean domain scores for ICUs that had intensivist staffing were compared with those for ICUs that had nonintensivist staffing using an unpaired t test. To assess the relationship between each domain score in the organizational culture survey and each domain score in the safety culture survey, for each pair of domains and within each ICU, the Pearson correlation coefficient was calculated based on the individual (respondent-level) domain scores. Then, box plots of these correlation coefficients by ICU ordered by the number of ICU beds in each site were constructed and visually inspected to assess: (1) the strength of the correlations (as depicted by the median in each box plot) and (2) whether the strength of the correlation might vary with the number of ICU beds. In adherence to convention, P b.05 is reported as statistically significant. However, because of the large number of comparisons explored in this work, the P values presented are not interpretable as probability statements in the hypothesis testing framework but serve as a convenient measure for ranking, which observed associations are more likely to be nonzero. This study was approved by the research ethics board of each participating institution. Formal written consent was not required because consent was implied by completion of the surveys. 3. Results 3.1. Description of ICUs and respondents A total of 2374 surveys were distributed at the 23 participating ICUs; 1285 surveys were returned for an overall response rate of 54%. These ICUs were in 9 tertiary care hospitals and 14 community hospitals. Eight of the ICUs

4 4 P.M. Dodek et al. t1:1 Table 1 Respondents to organizational/safety culture surveys t1:2 t1:3 Position n % t1:4 Registered nurse t1:5 Respiratory therapist t1:6 Physician t1:7 Unit assistant/clerk/secretary t1:8 Physician assistant/nurse practitioner 38 3 t1:9 Other t1:10 Missing t1:11 Total were in teaching hospitals, and 15 were in nonteaching 232 hospitals. The number of ICU beds across sites ranged from to 29 with a mean of 13. In the 16 intensivist-staffed ICUs, 234 the mean number of beds was 15, whereas in the nonintensivist-staffed ICUs, the mean number of beds was Most of the respondents were nurses, but respondents 237 included all ICU professionals (Table 1) Relationship between culture scores and 239 respondent or type of ICU 240 Organizational and safety culture domain scores for each 241 ICU were generally favorable (Appendix E1), but scores 242 were generally higher for communication and perceived 243 effectiveness and generally lower for problem solving, 244 teamwork or relations between units, and for staffing 245 adequacy. The lowest scores were for hospital management 246 support for patient safety. Physicians' scores for the domains 247 of between-group communication accuracy, medical director 248 budgeting authority, and staffing adequacy were lower than 249 those of other health professionals, but physicians' scores for 250 the domains of between-group communication openness, 251 communication timeliness, and physician leadership were 252 higher than those of other health professionals (Table 2). 253 Correlations between domain scores and number of ICU 254 beds showed positive relationships between perceived 255 relative technical quality of care, effectiveness at recruitment 256 and retention of staff, and medical director budgeting 257 authority and size of ICU higher scores in larger ICUs 258 (Table 2). These same relationships were found for ICUs that 259 had intensivist staffing (Table 2). Conversely, scores for 260 frequency of adverse events reported and teamwork across 261 hospital units were inversely related to the size of the ICU 262 and intensivist staffing lower scores in larger ICUs or those 263 that had intensivist staffing (Table 2). There were no other 264 statistically significant relationships between any culture 265 domain score and ICU size or staffing model Correlation between organizational and 267 safety culture 268 Correlations between all pairs of domain scores in the 269 organizational and safety culture surveys showed that there are weak but consistently positive relationships between domains that are specific to patient safety and all domains of organizational culture (Appendix E2). There were stronger correlations between related constructs from the 2 surveys, for example, teamwork across hospital units. None of the plots of the correlation coefficients against the number of ICU beds exhibited any pattern suggesting that the strength of the correlation varied with the number of ICU beds. 4. Discussion We found that organizational and safety culture scores reported by the staff from a cross section of community and teaching ICUs in Canada were generally favorable. Furthermore, physicians differed from other health professionals in their perceptions of some of these culture domains, especially related to communication and leadership characteristics of physicians. For example, physicians rated between-group communication openness and physician leadership higher than did nonphysicians. Staff from larger ICUs and ICUs with intensivist staffing perceived their ICUs to have a higher quality of care, greater ability to recruit and retain staff, and more budgeting authority of their medical director. However, staff in larger ICUs and those with intensivist staffing also reported fewer adverse events and perceived less teamwork across hospital units than did the staff from smaller ICUs and those that did not have intensive staffing. This finding highlights the potential importance of a focus on teamwork as a quality improvement initiative for larger ICUs. The similar findings for large ICUs and those that have intensivist staffing are likely related to the fact that most of the ICUs that have intensivist staffing are large. The weak correlations between most of the domain scores in the organizational culture and safety culture surveys suggest that these surveys each measure unique domains and are not interchangeable. However, stronger correlations between similar domains (eg, teamwork across hospital units) suggest that for some domains, these surveys are measuring similar constructs. These findings also indicate that organizational culture may influence safety culture or vice versa. Measures of organizational culture are important because they have been linked to measures of organizational performance outside the critical care setting. For example, cultures that convey beliefs, values, and norms suggesting that members of the group are respected and fairly treated have been known to impact specific attitudes and behaviors of those members, including increasing helping behaviors, performance, and lowering absenteeism [22,23]. Such cultures can affect individuals by providing an opportunity for social comparisons. Lind et al [24] argued that when evaluating whether someone is treated fairly, individuals often incorporate others' treatment into their judgments, such that when others are treated fairly, individuals are more likely to deem their own treatment to be fair, whereas when others

5 Relationship to size of ICU and physician management model 5 t2:1 Table 2 Relationships between organizational/safety culture scores and respondent, number of ICU beds, or intensivist staffing t2:2 t2:3 Organizational and safety culture domains Mean for physicians (SD) Mean for nonphysicians (SD) Correlation with ICU beds Mean for nonintensivist staffing hospitals (SD) Mean for intensivist staffing hospitals (SD) t2:4 Organizational culture t2:5 Within-group communication openness 4.27 (0.40) 4.13 (0.54) (0.19) 4.0 (0.17) t2:6 Between-group communication openness 4.27 (0.34) 3.86 (0.61) a (0.35) 3.8 (0.23) t2:7 Within-group communication accuracy 3.81 (0.37) 3.59 (0.54) (0.25) 3.5 (0.20) t2:8 Between-group communication accuracy 3.41 (0.60) 3.81 (0.53) a (0.22) 3.7 (0.23) t2:9 Communication timeliness 4.09 (0.37) 3.83 (0.48) b (0.21) 3.8 (0.13) t2:10 Nursing leadership 3.64 (0.48) 3.58 (0.47) (0.25) 3.4 (0.23) t2:11 Physician leadership 3.88 (0.36) 3.41 (0.50) a (0.37) 3.4 (0.23) t2:12 Unit relations with other units 3.51 (0.57) 3.34 (0.59) (0.30) 3.2 (0.24) t2:13 Perceived effectiveness at recruiting and retaining nurses 3.05 (0.79) 3.04 (0.62) (0.37) 3.0 (0.34) b t2:14 Perceived effectiveness at recruiting and retaining physicians 3.27 (0.84) 3.30 (0.55) 0.58 a 2.8 (0.27) 3.3 (0.25) a t2:15 Absolute technical quality of care 3.94 (0.40) 3.84 (0.39) (0.26) 3.8 (0.17) t2:16 Relative technical quality of care 3.46 (0.52) 3.44 (0.56) 0.66 a 3.0 (0.31) 3.5 (0.29) a t2:17 Perceived effectiveness at meeting family member needs 3.82 (0.51) 3.66 (0.48) (0.30) 3.7 (0.26) t2:18 Within-group problem-solving conflict strategy 3.43 (0.45) 3.25 (0.54) (0.22) 3.1 (0.20) t2:19 Between-group problem-solving conflict strategy 3.28 (0.46) 3.11 (0.59) (0.28) 3.1 (0.20) t2:20 Within-group avoiding conflict strategy 3.73 (0.43) 3.59 (0.49) (0.23) 3.5 (0.17) t2:21 Between-group avoiding conflict strategy 3.80 (0.24) 3.58 (0.50) (0.21) 3.6 (0.18) t2:22 Nursing director budgeting authority 3.49 (0.54) 3.54 (0.59) (0.34) 3.5 (0.22) t2:23 Medical director budgeting authority 2.81 (0.71) 3.23 (0.48) a 0.46 b 3.0 (0.29) 3.2 (0.20) b t2:24 Nursing director patient care authority 3.61 (0.46) 3.49 (0.58) (0.29) 3.3 (0.30) t2:25 Medical director patient care authority 3.87 (0.42) 3.81 (0.57) (0.32) 3.9 (0.26) t2:26 Overall job satisfaction 3.99 (0.77) 3.92 (0.61) (0.24) 3.8 (0.21) t2:27 Safety culture t2:28 Overall perceptions of safety 3.21 (0.41) 3.24 (0.56) (0.33) 3.1 (0.31) t2:29 Frequency of events reported 3.18 (0.41) 3.19 (0.66) 0.46 b 3.3 (0.14) 3.1 (0.13) b t2:30 Supervisor expectations promoting patient safety 3.31 (0.29) 3.34 (0.37) (0.14) 3.3 (0.18) t2:31 Organizational learning 3.59 (0.36) 3.51 (0.43) (0.23) 3.4 (0.19) t2:32 Teamwork within units 4.13 (0.32) 3.94 (0.41) (0.22) 3.8 (0.31) t2:33 Communication openness 3.77 (0.30) 3.66 (0.43) (0.18) 3.6 (0.22) t2:34 Feedback and communications 3.33 (0.42) 3.30 (0.57) (0.19) 3.1 (0.24) t2:35 Nonpunitive response to error 3.20 (0.31) 3.23 (0.49) (0.13) 3.1 (0.22) t2:36 Staffing 2.70 (0.47) 2.99 (0.48) b (0.30) 3.1 (0.35) t2:37 Hospital management support for safety 3.00 (0.65) 3.20 (0.65) (0.38) 3.0 (0.22) t2:38 Teamwork across hospital units 3.11 (0.54) 3.22 (0.53) 0.51 b 3.3 (0.21) 3.0 (0.19) a t2:39 Hospital handoffs and transitions 2.80 (0.42) 3.01 (0.50) (0.15) 3.1 (0.14) t2:40 a P b.01 for comparison between groups or correlation. t2:41 b P value between.01 and.05 for comparison between groups or correlation. 322 are treated unfairly, individuals may incorporate these 323 negative experiences into their own judgments. This 324 phenomenon has been termed the common knowledge effect. 325 In health care, there are emerging data to show that positive 326 aspects of organizational culture, especially teamwork, are 327 linked to improved clinical outcomes, especially related to 328 patient safety [6,7]. 329 Organizational culture has also been linked to perfor- 330 mance of ICUs. For example, a before-after study of 331 introducing a culture of improvement in 1 ICU was 332 associated with improvements in processes, outcomes, and 333 costs of critical care [25]. A study of introducing a senior 334 executive adopt-a-work unit program in another ICU [26] was associated with an improvement in patient safety and in staff perceptions about the culture of safety. In a study of 13 ICUs, differences in standardized mortality rates among these ICUs were more strongly related to interaction and coordination among the staff of each ICU than to administrative structure, amount of specialized treatment, or teaching status of the hospital [27]. Another crosssectional survey of 42 ICUs showed that caregiver interaction (a term that includes culture, leadership, coordination, communication, and conflict management ability) is related to lower risk-adjusted length of ICU stay, lower nurse turnover, and a greater perceived ability to meet family members' needs [16]. A recent multicenter study found that a

6 6 P.M. Dodek et al. 348 higher perception of management by ICU staff correlates 349 with lower hospital mortality [10]. A cross-sectional survey 350 in 3 hospitals documented that ICU nursing' reports of 351 collaboration were correlated positively with improved 352 patient outcomes [28]. Interestingly, nurse ratings of 353 collaboration and teamwork are lower than those of 354 physicians working in the same ICUs [29]. Although we 355 did not find any differences between physicians and other 356 professionals in their perceptions related to teamwork, we 357 did find differences in their perceptions of communication, 358 leadership, and adequacy of staffing. 359 Our study adds to the body of knowledge about 360 organizational and safety culture by demonstrating the 361 relative consistency of domain scores across a variety of 362 ICUs. In addition, we showed that certain scores are related 363 to size of the ICU or to the type of physician management 364 model size and physician management model are likely 365 correlated to each other and to other features of large ICUs 366 (eg, complexity of patients). Interestingly, the same (larger 367 and intensivist physician staffing) ICUs that had high scores 368 for perception of technical quality of care had low scores for 369 frequency of reporting adverse events and for teamwork 370 between units. It is not known whether the lower score for 371 frequency of reporting adverse events reflects a lower 372 incidence of adverse events or simply a lower frequency of 373 reporting these events. It is also not known if there is any 374 relationship between the findings of low perceived 375 teamwork between units and low frequency of reporting 376 adverse events. Although the relationship between volume 377 of admissions and quality of care has been reported 378 previously [12], we are not aware of any previous reports 379 regarding the relationship between ICU size and reporting 380 of adverse events or measures of organizational culture. Our 381 finding that there are differences in frequency of reporting 382 events among ICUs of different size contrasts with the 383 observation that there are no differences in this domain by 384 size of hospital when the unit of analysis is the entire 385 hospital [20]. Perhaps smaller ICUs could improve their 386 performance by addressing quality-of-care issues, and larger 387 ICUs could improve their performance by facilitating the 388 reporting of adverse events and by developing teamwork 389 with other units in the hospital [6,7]. Furthermore, our 390 finding that there are correlations between some domains of 391 organizational culture and safety culture builds on the 392 finding that specific types of organizational culture correlate 393 with safety climate in hospitals [30]. 394 Strengths of this study include the use of validated survey 395 tools, the number and variety of participating ICUs, the 396 absolute number of health care professionals who responded, 397 and the consistency of responses for similar domains in the surveys. Limitations include the nonrandom sample of ICUs 399 because participation in this project was voluntary and the % response rate. Because of the anonymous nature of the 401 survey, we did not capture any descriptors about non- 402 responders, so we were unable to compare responders to 403 nonresponders. The survey design of this study precluded capturing details about safety programs and actual error or 404 complications rates within hospitals; therefore, we could not 405 validate opinions of the respondents about safety. In 406 addition, we did not capture information about tacit 407 knowledge of how work is done. The cross-sectional design 408 also precluded observations of changes in culture scores over 409 time. Finally, as in any observational study, confounding 410 may be a cause of the associations that we observed. For 411 instance, larger ICUs are usually located in larger hospitals 412 and may be systematically different from smaller ICUs (eg, 413 care for more complex patients and provide technical and 414 other services such as hemodialysis not available in smaller 415 ICUs). Given the likely strong correlations between size of 416 ICU and these additional descriptors, it may be difficult to 417 tease out the independent effect of ICU size. 418 In summary, we found that staff from 23 tertiary and 419 community ICUs generally report a favorable organiza- 420 tional and safety culture. However, differences in percep- 421 tions between staff in larger and smaller ICUs highlight 422 the importance of improving teamwork among units in 423 larger ICUs Supplementary materials related to this article can be 426 found online at doi: /j.jcrc Acknowledgments 428 This study was sponsored by the Canadian Institutes of [1] Schein EH. Organizational culture and leadership. 3rd ed. San 451 Francisco: Jossey-Bass; [2] Kilman RH. Gaining control of the corporate culture. San Francisco: 453 Jossey-Bass; Health Research, the Michael Smith Foundation for Health Research, and the Canadian Researchers at the End of Life Network (CARENET). The CARENET is supported by a grant from the Canadian Institutes of Health Research and Heart and Stroke Foundation of Canada. None of the funding agencies had any role in the study design, collection of data, analysis and interpretation of data, writing of the manuscript, or the decision to submit the manuscript for publication. The authors thank all members of CARENET who have contributed to the evolution and development of this work. The authors also thank Drs Sandra Robinson and John Millar for their input into the design of this project, Carol Honeyman and Mahi Etminan for coordinating the dissemination and collection of surveys, and Monica Norena and Hong Wang for their statistical work. The authors also thank the medical directors and research coordinators at each of the participating sites. Dr Fowler is a clinician-scientist of the Heart and Stroke Foundation (Ontario). Dr Deborah Cook holds a Canada Research Chair with the Canadian Institutes of Health Research. References

7 Relationship to size of ICU and physician management model [3] O'Reilly C, Chatman J. Organizational commitment and psychological 456 attachment: the effects of compliance, identification and internaliza- 457 tion on prosocial behavior. J Appl Psychol 1986;71: [4] Ruchlin HS, Dubbs NL, Callahan MA. The role of leadership in 459 instilling a culture of safety: lessons from the literature. J Healthc 460 Manag 2004;49: [5] Scott T, Mannion R, Marshall M, Davies H. Does organizational 462 culture influence health care performance? A review of the evidence. 463 J Health Serv Res Policy 2003;8: [6] Hoff T, Jameson L, Hannan E, Flink E. A review of the literature 465 examining linkages between organizational factors, medical errors, 466 and patient safety. Med Care Res Rev 2004;61: [7] Musson D, Helmreich R. Team training and resource management in 468 health care: current issues and future directions. Harvard Health Policy 469 Rev 2004;5: [8] Kutsogiannis DJ, Hague C, Triska OH, et al. The organizational 471 structure of intensive care units and its influence on patient outcomes. 472 Healthcare Management Forum 2001;14: [9] Pronovost PJ, Berenholtz SM, Goeschel C, et al. Improving patient 474 safety in intensive care units in Michigan. J Crit Care 2008;23: [10] Huang DT, Clermont G, Kong L, et al. Intensive care unit safety 477 culture and outcomes: a US multicenter study. Int J Qual Health Care ;22: [11] Peelen L, de Keizer NF, Peek N, et al. The influence of volume and 480 intensive care unit organization on hospital mortality in patients 481 admitted with severe sepsis: a retrospective multicentre cohort study. 482 Crit Care (London, England) 2007;11:R [12] Kahn JM, Ten Have TR, Iwashyna TJ. The relationship between 484 hospital volume and mortality in mechanical ventilation: an instru- 485 mental variable analysis. Health Serv Res 2009;44: [13] Pronovost PJ, Angus DC, Dorman T, et al. Physician staffing patterns 487 and clinical outcomes in critically ill patients: a systematic review. 488 JAMA 2002;288: [14] Treggiari MM, Martin DP, Yanez ND, et al. Effect of intensive 490 care unit organizational model and structure on outcomes in 491 patients with acute lung injury. Am J Resp Crit Care Med 2007; : [15] Levy MM, Rapoport J, Lemeshow S, et al. Association between 494 critical care physician management and patient mortality in the 495 intensive care unit. Ann Intern Med 2008;148: [16] Shortell SM, Zimmerman JE, Rousseau DM, et al. The performance of 497 intensive care units: does good management make a difference? Med 498 Care 1994;32: [17] Shortell SM, Rousseau DM, Gillies RR, et al. Organizational assessment in intensive care units (ICUs): construct development, reliability, and validity of the ICU nurse-physician questionnaire. Med Care 1991;29: [18] Sorra JS, Nieva VF. Hospital survey on patient safety culture. (Prepared by Westat, under Contract No ). AHRQ Publication No Rockville, MD: Agency for Healthcare Research and Quality; [19] Blegen MA, Gearhart S, O'Brien R, et al. AHRQ's hospital survey on patient safety culture: psychometric analyses. J Patient Safety 2009;5: [20] Hospital Survey on Patient Safety Culture Comparative database report. AHRQ Publication No , March Rockville, MDAgency for Healthcare Research and Quality; [21] Kho ME, Carbone JM, Lucas J, Cook DJ. Safety climate survey: reliability of results from a multicenter ICU survey. Qual Saf Health Care 2005;14: [22] Colquitt JA, Noe RA, Jackson CL. Justice in teams: antecedents and consequences of procedural justice climate. Pers Psychol 2002;55: [23] Naumann SE, Bennett N. A case for procedural justice climate: development and test of a multilevel model. Acad Manage J 2000;43: [24] Lind EA, Kray L, Thompson L. The social construction of injustice: fairness judgments in response to own and others' unfair treatment by authorities. Organ Behav Hum Dec 1998;75:1-22. [25] Clemmer TP, Spuhler VJ, Oniki TA, Horn SD. Results of a collaborative quality improvement program on outcomes and costs in a tertiary critical care unit. Crit Care Med 1999;27: [26] Pronovost PJ, Weast B, Bishop K, et al. Senior executive adopt-a-work unit: a model for safety improvement. Crit Care Med 2004;30: [27] Knaus WA, Draper EA, Wagner DP, Zimmerman JE. An evaluation of outcome from intensive care in major medical centers. Ann Intern Med 1986;104: [28] Baggs JG, Schmitt MH, Mushlin AI, et al. Association between nursephysician collaboration and patient outcomes in three intensive care units. Crit Care Med 1999;27: [29] Thomas EJ, Sexton JB, Helmreich RL. Discrepant attitudes about teamwork among critical care nurses and physicians. Crit Care Med 2003;31: [30] Singer SJ, Falwell A, Gaba DM, et al. Identifying organizational cultures that promote patient safety. Health Care Manage R 2009;34:

8 Appendix E1. Domain scores for organizational and safety culture (median and interquartile range) by hospital, according to number of ICU beds. Appendix E2. Correlations between domain scores for safety culture and domain scores for organizational culture (median and interquartile range for correlation coefficients).

9 Our reference: YJCRC P-authorquery-v8 AUTHOR QUERY FORM Journal: YJCRC Article Number: Please or fax your responses and any corrections to: Shayna Bayard Tel: Fax: Dear Author, Any queries or remarks that have arisen during the processing of your manuscript are listed below and highlighted by flags in the proof. Please check your proof carefully and mark all corrections at the appropriate place in the proof (e.g., by using onscreen annotation in the PDF file) or compile them in a separate list. For correction or revision of any artwork, please consult Any queries or remarks that have arisen during the processing of your manuscript are listed below and highlighted by flags in the proof. Click on the Q link to go to the location in the proof. Location in article Q1 Q2 Q3 Query / Remark: click on the Q link to go Please insert your reply or correction at the corresponding line in the proof Please confirm that given names and surnames have been identified correctly. Please provide the postal code of the affiliations. Please verify if appropriate that all authors are linked to the provided collaboration. Thank you for your assistance. Page 1 of 1