Barriers to Adverse Event and Error Reporting in Anesthesia

Transcription

1 Barriers to Adverse Event and Error Reporting in Anesthesia Gaylene C. Heard, MBBS, FANZCA, MHumanFact,* Penelope M. Sanderson, PhD, FASSA, and Rowan D. Thomas, MBBS, FANZCA, MPH* BACKGROUND: Although anesthesiologists are leaders in patient safety, there has been little research on factors affecting their reporting of adverse events and errors. First, we explored the attitudinal/emotional factors influencing reporting of an unspecified adverse event caused by error. Second, we used a between-groups study design to ask whether there are different perceived barriers to reporting a case of anaphylaxis caused by an error compared with anaphylaxis not caused by error. Finally, we examined strategies that anesthesiologists believe would facilitate reporting. Where possible, we contrasted our results with published findings from other physician groups. METHODS: An anonymous, self-administered, mailed survey was conducted of 629 consultant anesthesiologists and 263 anesthesiology residents on the mailing list of the Australian and New Zealand College of Anaesthetists in Victoria, Australia. Participants were randomized into Error versus No Error groups for the specified anaphylaxis adverse event section of the survey. Data were analyzed using nonparametric descriptive and inferential tests. RESULTS: There were 433 usable returned surveys, a usable response rate of 49%. First, there was only 1 of 13 statements on attitudinal/emotional factors that influenced reporting of an unspecified adverse event caused by error with which more anesthesiologists agreed/strongly agreed than disagreed/strongly disagreed: Doctors who make errors are blamed by their colleagues. Second, when an error rather than no error had caused anaphylaxis, participants were more likely to agree/strongly agree that 6 statements about litigation, getting into trouble, disciplinary action, being blamed, unsupportive colleagues, and not wanting the case discussed in meetings, were perceived as reporting barriers. Finally, the most favored assistive strategies for reporting were generalized deidentified feedback about adverse event and error reports, role models such as senior colleagues who openly encourage reporting, and legislated protection of reports from legal discoverability. CONCLUSION: The majority of anesthesiologists in our study did not agree that the attitudinal/emotional barriers surveyed would influence reporting of an unspecified adverse event caused by error, with the exception of the barrier of being concerned about blame by colleagues. The probable influence of 6 perceived barriers to reporting a specified adverse event of anaphylaxis differed with the presence or absence of error. Anesthesiologists in our study supported assistive reporting strategies. There seem to be some differences between our results and previously published research for other physician groups. (Anesth Analg 2012;114: ) Anesthesiologists have been acknowledged as leaders in patient safety strategies, 1 yet there has been very little research on barriers they perceive to reporting adverse events and errors. Accurate reporting of errors is considered to be fundamental to improving patient safety 2,3 because you can t fix what you don t know about. 4 Underreporting in health care is evidenced by expert opinion and past research 2,3,5 8 and by the improved reporting behavior that occurs when reporting systems are introduced and supported. 9,10 From the *Department of Anaesthesia, St. Vincent s Hospital, Melbourne, Victoria; and Schools of Psychology, Information Technology and Electrical Engineering, and Medicine, The University of Queensland, Brisbane, Australia. Accepted for publication May 13, Funding: Self-funded by first author. The authors declare no conflicts of interest. Reprints will not be available from the authors. Address correspondence to Gaylene C. Heard, MBBS, FANZCA, MHumanFact, Department of Anaesthesia, St. Vincent s Hospital, PO Box 2900, Fitzroy, Melbourne, Victoria, Australia Address to gaylene.heard@svhm.org.au. Copyright 2012 International Anesthesia Research Society DOI: /ANE.0b013e e8 Reporting and learning are 2 critical features of an effective safety culture. 11 Counting adverse events and errors for the sake of it contributes little unless the data are analyzed and used to create safer systems. 2,3,12 When errors with no adverse outcomes are reported, free lessons are provided by analyzing what strategies were used to catch incipient errors or mitigate their effects before harm occurred. 13 Potential benefits from a culture of error disclosure and open discussion include improved professional learning and better patient outcomes from higher quality care, increased trust between patients and staff, less litigation, and a more realistic view by patients and staff of medicine s limitations. 14 The patient safety movement promotes reporting of errors but this contrasts with expert opinion that medicine has a traditional code of silence 14 around errors. Blame and shame contribute to underreporting. Higher reporting rates are positively correlated with independently defined measures of a safety culture, including lack of blame and punishment and fair treatment for staff involved in errors or incidents. 10 The 10 most frequently cited barriers to medical error disclosure are professional repercussions of disclosure, March 2012 Volume 114 Number 3

2 legal liability, blame, lack of confidentiality, negative patient or family reaction, humiliation, perfectionism, guilt, lack of anonymity, and the absence of a supportive forum for disclosure. 18 Uncertainty about definitions of errors and adverse events, lack of time to report, and knowing neither what nor how to report are further barriers. 2,19,20 Strategies to improve reporting include providing faster and improved feedback, 2 education in reportable event recognition, 2 user-friendly paper-based and computerized systems, 21,22 and payment for reports. 23 There has been little research specifically examining whether perceived barriers to reporting an adverse event differ based on whether or not that event was caused by an error. Unfortunately, the available previous survey research on physicians perceived barriers to reporting is further limited because results are reported for mixed groups of physicians and nurses 19,24 and fewer items specifically probe emotional barriers to reporting. 19,20,24 In a previous South Australian survey of physicians, 14% agreed or strongly agreed that My coworkers may be unsupportive and 21% agreed or strongly agreed that I am worried about litigation were perceived barriers to incident reporting. 20 In a United States survey of physicians perceived barriers to error reporting, Not knowing the usefulness of the report and Thinking that reporting has little contribution for improvement of quality of care were among the top 6 of 17 barriers. 19 In our study, the attitudes and perceived barriers of anesthesiologists to reporting adverse events and errors were explored. First, we asked whether anesthesiologists in our sample agreed or disagreed with the attitudinal/emotional factors we listed that may influence whether an adverse event caused by an error will be reported. Second, we asked whether there is a difference in perceived barriers to reporting a specified adverse event of anaphylaxis caused by an error compared with no error. Third, we asked which assistive strategies our participants thought could improve reporting. In our Discussion, we illustrate how our results differ from previously published surveys of perceived barriers to reporting in other physician groups. METHODS Participants Ethics approval was obtained from The University of Queensland. In accordance with its Ethics Guidelines, a waiver was granted for the need for individual signed participant consent because the study involved an anonymous mailed survey with minimal risk to participants. The Participant Information Sheet, mailed to potential participants in the survey, explained that participation was voluntary. The Australian and New Zealand College of Anaesthetists (ANZCA) Trials Group provided gatekeeper permission to send the survey to anesthesiologists and anesthesiologist trainees on its mailing list. We conducted an anonymous, self-administered survey of 629 consultant anesthesiologists and 263 anesthesiology residents on the ANZCA mailing list in the state of Victoria, Australia. The participant sample group is 84% of all consultant anesthesiologists and 100% of anesthesiology residents in Victoria. Materials Participants received a mailed survey package consisting of a covering letter, Participant Information Sheet, questionnaire, and a stamped, addressed reply envelope. A follow-up reminder was ed. The questionnaire had 5 sections: 3 sections listed statements to be rated using a 5-point Likert scale ranging from strongly agree to strongly disagree, a fourth section had questions on demographic characteristics, and a fifth section was for free text comments. The survey packages were identical except for section 2. Section 1 of the Questionnaire Section 1 specifically explored attitudinal and emotional factors that could influence whether an (unspecified) adverse event caused by an error would be reported. Because of the lack of published survey items on these attitudinal and emotional factors, the construction of the survey items for section 1 was informed by a literature review as part of a master s thesis, and was based on a published, empirically derived taxonomy of factors impeding the willingness of physicians to disclose medical errors. 18 Section 1 had 13 survey items, with 10 items based on the domain of attitudinal barriers in the taxonomy 18 and 3 based on the domain of fears/anxieties barriers. 18 Section 2 of the Questionnaire Section 2 of the survey had a between-groups design. It covered barriers to reporting an adverse event of anaphylaxis, with or without an error as its cause. This section contained a case scenario: You are anesthetizing a 32-yearold woman for an elective laparoscopic cholecystectomy. You give the patient an IV antibiotic after induction and the patient develops a very severe anaphylaxis, which you treat appropriately. The surgery is postponed and the patient, who remains intubated, is transferred to ICU for further monitoring and treatment. She makes a full recovery. There were 2 different endings to the scenario: For the group No Error: The patient had no history of drug allergy. For the group Error: You realize afterward you have given the patient that particular antibiotic in error, because the patient had told you preoperatively that she was allergic to it and you had written this on the anesthetic chart preoperatively. There were no features on the surveys to distinguish the group No Error from the group Error surveys, apart from the different endings to the scenario. There were 17 statements to rate as barriers to reporting the anaphylaxis, with 14 of these statements either the same as or similar to previously published survey statements. 20,24 For example, we changed statement wording such as on the ward 20,24 to at work to suit our anesthesiology participants, and preferred the word colleagues 24 rather than coworkers. 20 Three statements had been more substantially altered. Section 3 of the Questionnaire Section 3 covered strategies to improve reporting of adverse events and errors, with 17 statements on factors that March 2012 Volume 114 Number

3 Adverse Event and Error Reporting would help you to report adverse events and errors. These survey items were constructed based on literature review of strategies shown to increase reporting such as feedback, 2 education, 2 user-friendly systems, 21,22 payment, 23 and on the perceived barriers to reporting that physicians thought would be modifiable to increase reporting. 19 Section 4 of the Questionnaire Section 4 covered demographic characteristics, with questions on type of practice (public or private practice), years of anesthesia experience, age group, location (capital city or regional/rural), and gender. Section 5 of the Questionnaire This section was for participants free text comments about the survey or about reporting adverse events and errors. Procedure The survey was piloted on 20 anesthesiologists and 5 anesthesia residents to examine the usability of the survey, using contextual inquiry and modified cognitive walkthrough techniques. Minor changes were subsequently made to the wording of the survey instructions and to 3 survey items. The survey packages were prepared into 2 separate groups, group No Error and group Error, by the researchers and packed in random order for mailing using a randomization table. 25 The survey packages were addressed and mailed by ANZCA. The researchers did not have access to participants names and addresses. A follow-up reminder by was sent to participants by ANZCA 2 weeks after the initial mailing. Data Collation Each returned survey was given a unique identification number and an NE or an E depending whether it was in group No Error or group Error. Missing data were treated as null. Data were entered into an Excel spreadsheet (Microsoft, Redmond, WA). Statistical Analysis The data obtained from the Likert scales were treated as ordinal data, and nonparametric inferential tests were used. A 2 (groups) c (categories) 2 test was used to compare the demographic characteristics of group No Error and group Error. A series of 2 (groups) 5 (levels of agreement) Wilcoxon-Mann- Whitney (WMW) tests was used to compare the responses of participants in group No Error and group Error for section 2. A series of r (demographic categories) c (levels of agreement) Kruskal-Wallis tests was used to examine the effects of demographic characteristics on participants responses in sections 1 to 3. Post hoc analyses of statistically significant demographic influences were performed using 2 c WMW tests. To maintain the family-wise type I error rate at 0.05, all raw P values from the Kruskal-Wallis tests and WMW tests were adjusted for the effects of multiple comparisons using the Ryan-Holm step-down procedure, with raw P values arranged in ascending order and sequentially adjusted until all hypotheses were tested or until the adjusted P value was Data were analyzed using Stata version 10 (StataCorp, College Station, TX). A P value of 0.05 was considered statistically significant. Table 1. Demographic Characteristics of Participants a Characteristic Frequency (n) Percentage Age group (y) Gender Female Male Practice type Mostly private Mostly public Public/private Other Work location Capital city Regional/rural Mix capital/regional Overseas Years of experience 20 b b b Anesthesia resident Prevocational resident Retired a One participant did not answer any demographic questions and another participant answered all questions except years of experience. b Consultant anesthesiologist. RESULTS The usable survey response rate was 52% for consultant anesthesiologists and 39% for anesthesia residents, with an overall usable response rate of 49%. The usable response rate for women was 46% and for men 49%. Two participants did not complete section 2 but completed the other sections. Twenty other surveys had some missing data. Participants worked mainly in a capital city and in public hospital practice (Table 1). For section 2, there were 214 participants in group No Error and 217 participants in group Error, and as expected from the randomization process, 30 there were no statistically significant differences between their demographic characteristics. Section 1 The only statement with which more participants agreed or strongly agreed than disagreed or strongly disagreed was Doctors who make errors are blamed by their colleagues, with 46% (confidence interval [CI], 42% 51%) agreeing or strongly agreeing, 26% (CI, 22% 30%) disagreeing or strongly disagreeing, and 27% (CI, 23% 32%) neutral (Table 2). There were no statistically significant demographic influences on responses. Section 2 Participants in group Error were more likely to agree or strongly agree than those in group No Error that 6 of the 17 statements were barriers to reporting the anaphylaxis (Fig. 1, Tables 3 and 4). The 6 statements were: I am worried about litigation. I don t want to get into trouble. My colleagues may be unsupportive ANESTHESIA & ANALGESIA

4 Table 2. Section 1: Attitudinal and Emotional Factors Influencing Reporting an Unspecified Adverse Event Caused by an Error Statement agree Agree Neutral Disagree disagree I would protect my self-interests 0.7% ( ) 4.4% ( ) 7.7% ( ) 56.3% ( ) 30.9% ( ) ahead of the interests of the patient if I could, e.g., by hiding or denying an error. Competition with my peers would 0.2% ( ) 7.7% ( ) 10.9% ( ) 56.3% ( ) 24.9% ( ) prevent me from disclosing an error. If a doctor is careful enough he 1.4% ( ) 10.6% ( ) 9.4% ( ) 51.4% ( ) 27.2% ( ) or she will not make an error. It would affect my identity as a 0.5% ( ) 9.3% ( ) 13.5% ( ) 60.6% ( ) 16.1% ( ) doctor to admit to an error. Others don t need to know about 0.5% ( ) 11.7% ( ) 19.2% ( ) 55.3% ( ) 13.3% ( ) errors I have made. Disclosing an error, if you don t 1.2% ( ) 12.3% ( ) 19.1% ( ) 50.8% ( ) 16.6% ( ) have to, is an optional act of heroism. I would cover up an error I had 0.2% ( ) 9.3% ( ) 23.6% ( ) 52.2% ( ) 14.7% ( ) made if I could. If I admit to an error I will feel 1.9% ( ) 19.1% ( ) 16.0% ( ) 50.2% ( ) 12.8% ( ) like a failure. It would affect my self-esteem to 0.9% ( ) 28.2% ( ) 16.7% ( ) 42.3% ( ) 11.9% ( ) admit to an error. Doctors who make errors are 2.6% ( ) 15.2% ( ) 27.8% ( ) 46.0% ( ) 8.4% ( ) humiliated by their colleagues. Medicine has a culture of silence 6.1% ( ) 30.4% ( ) 23.0% ( ) 37.5% ( ) 3.0% ( ) where errors are not talked about. Doctors who make errors are 3.3% ( ) 43.1% ( ) 27.5% ( ) 25.4% ( ) 0.7% ( ) blamed by their colleagues. Doctors should not make errors. 10.8% ( ) 31.4% ( ) 14.0% ( ) 28.1% ( ) 15.7% ( ) Data are presented as percentage with confidence interval. I am worried about disciplinary action. I may be blamed unfairly for the event. I do not want the case discussed in meetings. For the other 11 statements for which there were no statistically significant differences between groups Error and No Error (Table 3), at least 68% (lowest confidence limit) of participants disagreed or strongly disagreed with the following statements: Adverse event reporting makes little contribution to quality of care (93% disagree/strongly disagree; CI, 90% 95%). I don t know whose responsibility it is to make a report (86% disagree/strongly disagree; CI, 83% 90%). A good outcome of the case makes reporting unnecessary (86% disagree/strongly disagree; CI, 83% 90%). I do not know which adverse events should be reported (73% disagree/strongly disagree; CI, 68% 77%). Demographic Influences on Response to Section 2 Anesthesia residents in group No Error agreed or strongly agreed more with I am worried about disciplinary action than did consultant anesthesiologists (Table 5). Anesthesia residents in group Error agreed or strongly agreed more with I do not know which adverse events should be reported than did consultant anesthesiologists (Table 5). Section 3 Section 3 investigated strategies to improve reporting of adverse events and errors. More than 75% (lower confidence limit) of participants agreed or strongly agreed with 7 assistive strategies about feedback, role models, legislated protection, ability to report anonymously, and clear guidelines (Table 6). The only strategy with which the majority of participants disagreed or strongly disagreed was Payment for time taken to report. Demographic Influences on Responses to Section 3 Of note, anesthesia residents were more likely to agree or strongly agree than anesthesiologists in practice as consultants for 20 years, or those in practice as consultants for 10 to 20 years, with the strategy Payment for time taken to report (Table 7). Further analysis using age groups showed that 9% (CI, 0% 19%) of those aged 61 to 70 years compared with 41% (CI, 26% 57%) of those aged younger than 30 years agreed/strongly agreed with Payment for time taken to report. DISCUSSION Section 1: Attitudinal and Emotional Barriers The majority of the anesthesiologists surveyed did not agree with the attitudinal and emotional factors we examined. Direct comparisons with other physician groups for the emotional and attitudinal barriers explored in section 1 are difficult because of the lack of previously published survey research using these items. The traditional perfectibility model 31,32 of error prevention is a barrier to reporting. 18 The perfectibility model for error prevention is based on beliefs that March 2012 Volume 114 Number

5 Adverse Event and Error Reporting Figure 1. Frequency of responses to section 2 statements for which there was a statistically significant difference between responses for group No Error and group Error. physicians are capable of and should provide error-free practice, with anything less being unacceptable. 31,32 Traditionally, physicians have been socialized throughout their training to strive for error-free practice. 31,32 In our study, 79% (CI, 75% 83%) disagreed/strongly disagreed with the statement If a doctor is careful enough he or she will not make an error, suggesting that the anesthesiologists in the sample do not agree with this aspect of the perfectibility model. This may reflect education of anesthesiologists about human performance and the inevitability of error. However, at a more personal level, 21% (CI, 17% 25%) of our sample agreed/strongly agreed, If I admit to an error I will feel like a failure and 29% (CI, 25% 33%) agreed/strongly agreed, It would affect my self-esteem to admit to an error. Errors are a normal feature of human performance 33 and human ANESTHESIA & ANALGESIA

6 Table 3. Section 2: Perceived Barriers to Reporting Anaphylaxis, Group No Error Versus Group Error a Statement P value b Group agree Agree Neutral Disagree disagree I am worried about NE 4.7% ( ) 26.6% ( ) 9.8% ( ) 37.4% ( ) 21.5% ( ) litigation. E 12.9% ( ) 44.7% ( ) 17.5% ( ) 19.4% ( ) 5.5% ( ) I don t want to get NE 1.9% ( ) 16.8% ( ) 10.3% ( ) 44.8% ( ) 26.2% ( ) into trouble. E 3.7% ( ) 35.9% ( ) 17.5% ( ) 33.2% ( ) 9.7% ( ) My colleagues may be NE 0.9% ( ) 13.5% ( ) 16.4% ( ) 44.4% ( ) 24.8% ( ) unsupportive. E 3.7% ( ) 26.3% ( ) 16.1% ( ) 42.4% ( ) 11.5% ( ) I am worried about NE 0.5% ( ) 8.9% ( ) 11.2% ( ) 53.7% ( ) 25.7% ( ) disciplinary action. E 4.2% ( ) 18.4% ( ) 17.0% ( ) 49.3% ( ) 11.1% ( ) I may be blamed NE 0.5% ( ) 14.5% ( ) 12.6% ( ) 47.2% ( ) 25.2% ( ) unfairly for the event. E 1.4% ( ) 20.3% ( ) 18.0% ( ) 49.3% ( ) 11.0% ( ) I do not want the case discussed in meetings. Adverse event reporting makes little contribution to quality of care. I don t know whose responsibility it is to make a report. A good outcome of the case makes reporting unnecessary. I do not know which adverse events should be reported. Even if I don t give my details I m worried they ll track me down. The forms take too long to fill in and I just don t have time. When I am busy at work I forget to make a report. I don t feel confident the information I provide will be kept confidential. I never get any feedback after I report an adverse event. I wonder about who else will have access to information I disclose. As long as the staff involved learn from incidents it is unnecessary to discuss them further NE 0.5% ( ) 3.3% ( ) 6.6% ( ) 52.4% ( ) 37.2% ( ) E 0.9% ( ) 13.8% ( ) 9.2% ( ) 58.6% ( ) 17.5% ( ) NS NE 0.9% ( ) 1.4% ( ) 5.2% ( ) 49.6% ( ) 42.9% ( ) E 0.0% ( ) 1.9% ( ) 5.1% ( ) 47.2% ( ) 45.8% ( ) NS NE 0.5% ( ) 5.1% ( ) 10.8% ( ) 55.6% ( ) 28.0% ( ) E 0.0% ( ) 2.8% ( ) 8.3% ( ) 59.4% ( ) 29.5% ( ) NS NE 0.5% ( ) 5.6% ( ) 9.3% ( ) 57.0% ( ) 27.6% ( ) E 0.5% ( ) 2.3% ( ) 9.2% ( ) 59.0% ( ) 29.0% ( ) NS NE 0.5% ( ) 21.9% ( ) 8.9% ( ) 48.6% ( ) 20.1% ( ) E 1.4% ( ) 12.4% ( ) 8.8% ( ) 57.6% ( ) 19.8% ( ) NS NE 1.9% ( ) 7.9% ( ) 22.4% ( ) 47.2% ( ) 20.6% ( ) E 2.8% ( ) 12.0% ( ) 22.7% ( ) 46.3% ( ) 16.2% ( ) NS NE 1.4% ( ) 26.2% ( ) 22.9% ( ) 39.2% ( ) 10.3% ( ) E 1.4% ( ) 23.0% ( ) 19.8% ( ) 46.1% ( ) 9.7% ( ) NS NE 2.3% ( ) 37.4% ( ) 10.8% ( ) 37.4% ( ) 12.1% ( ) E 2.8% ( ) 28.2% ( ) 13.9% ( ) 44.0% ( ) 11.1% ( ) NS NE 4.7% ( ) 19.2% ( ) 23.8% ( ) 37.8% ( ) 14.5% ( ) E 5.5% ( ) 25.0% ( ) 24.1% ( ) 35.7% ( ) 9.7% ( ) NS NE 2.3% ( ) 21.0% ( ) 25.7% ( ) 40.2% ( ) 10.8% ( ) E 6.0% ( ) 18.5% ( ) 29.6% ( ) 35.7% ( ) 10.2% ( ) NS NE 7.1% ( ) 33.8% ( ) 19.7% ( ) 30.5% ( ) 8.9% ( ) E 7.4% ( ) 34.3% ( ) 25.9% ( ) 24.1% ( ) 8.3% ( ) NS NE 0.0% (0 0) 1.9% ( ) 9.8% ( ) 61.7% ( ) 26.6% ( ) E 0.5% ( ) 5.1% ( ) 9.2% ( ) 66.3% ( ) 18.9% ( ) Data are presented as percentage with confidence interval. NE group No Error; E group Error; NS not significant. a Wilcoxon-Mann-Whitney test. b Corrected P values are shown. Raw P values have been corrected for effect of multiple comparisons using Ryan-Holm step-down Bonferroni correction. fallibility has been described, using an aviation context, as like gravity, weather, and terrain, just another foreseeable hazard 11 that requires robust preventative measures at both individual and system levels. The placing of self-interest before patient interests 18 and allowing competition with peers to inhibit disclosure 18 were not agreed with by the majority of our anesthesiologists. This contrasts with expert opinion that the competitive environment of March 2012 Volume 114 Number

7 Adverse Event and Error Reporting Table 4. Magnitude and Direction of Difference Between Section 3 Responses Where There Were Statistically Significant Differences Between Group No Error and Group Error a Statement Group agree, % Agree, % Neutral, % Disagree, % disagree, % I am worried about litigation. NE 18 (6.2) 16 (7.3) E 8.2 (0.9) 18.1 (5.4) 7.7 (0) I don t want to get into trouble. NE 11.6 (0) 16.5 (6.6) E 1.8 (0) 19.1 (7.6) 7.2 (0) My colleagues may be NE 0.3 (0) 2.0 (0) 13.3 (3.1) unsupportive. E 2.8 (0) 12.8 (2.2) I am worried about disciplinary NE 4.4 (0) 14.9 (4.5) action. E 3.7 (0.2) 9.5 (0.5) 5.8 (0) I may be blamed unfairly for NE 14.2 (4.1) the event. E 0.9 (0) 5.8 (0) 5.4 (0) 2.1 (0) I do not want the case discussed at meetings. NE 19.7 (8.1) E 0.4 (0) 10.5 (3.4) 2.6 (0) 6.2 (0) Items are placed in the rows corresponding to the preferred response. NE group No Error; E group Error. a Expressed as difference between raw percentages. Items in parentheses are the percentage difference between the upper 95% confidence limit of the lower group and the lower 95% confidence limit of the higher group. Zero denotes confidence limits overlap. Table 5. Section 2: Statistically Significant Demographic Influences on Responses of Group No Error and Group Error Group NE E Statement I am worried about disciplinary action. I do not know which adverse events should be reported. Demographic characteristic agree Agree Neutral Disagree disagree 20 y consultant a 0.0% ( ) 5.0% ( ) 15.0% ( ) 50.0% ( ) 30.0% ( ) y b 0.0% ( ) 0.0% ( ) 2.7% ( ) 66.7% ( ) 30.6% ( ) 10 y c 0.0% ( ) 3.4% ( ) 15.3% ( ) 52.5% ( ) 28.8% ( ) Resident 1.8% ( ) 23.2% ( ) 8.9% ( ) 51.8% ( ) 14.3% ( ) 20 y consultant d 0.0% ( ) 9.4% ( ) 5.7% ( ) 49.1% ( ) 35.8% ( ) y e 4.2% ( ) 4.2% ( ) 10.4% ( ) 56.2% ( ) 25.0% ( ) 10 y 0.0% ( ) 12.8% ( ) 8.6% ( ) 64.3% ( ) 14.3% ( ) Resident 2.3% ( ) 25.0% ( ) 9.1% ( ) 59.1% ( ) 4.5% ( ) Data are presented as percentage with confidence interval. NE group No Error; E group Error. a More than 20 years as consultant versus residents, P b Ten to 20 years as consultant versus residents, P c Fewer than 10 years as consultant versus residents, P d More than 20 years as consultant versus residents, P e Ten to 20 years as consultant versus residents, P Raw P values obtained from Wilcoxon-Mann-Whitney post hoc analysis after Kruskal-Wallis testing and corrected P values shown, corrected for multiple comparisons using Ryan-Holm step-down Bonferroni correction. medical training and practice frequently inhibits error disclosure 34 and that hiding, denying, or covering up errors occurs to protect oneself. 31,34,35 However, in our study, only 5% (CI, 3% 7%) of anesthesiologists agreed/strongly agreed that they would protect their self-interests ahead of a patient s, for example, by hiding or denying an error, and only 10% (CI, 7% 12%) agreed/strongly agreed that they would cover up an error if they could. These results may be affected by social desirability bias, whereby survey respondents answer questions as a good person should rather than what they actually believe. 36 Our results also reflect what participants think, and this may be different than what they would actually do when faced with a real situation. Blame and a code of silence 14 inhibit error reporting. After a serious error, physicians report high levels of emotional distress, shame, guilt, self-reproach, self-perceptions as failures, fear of blame and criticism, feeling isolated from colleagues, fear that others will find out about the error, and concern about their professional reputation. 15,37 39 These factors may contribute to the code of silence, together with past advice from insurance companies 34 and lawyers 34 to not disclose errors, as well as the influence of the perfectibility model. 31,32 For our participants, the statement Doctors who make errors are blamed by their colleagues was the only statement with which more anesthesiologists agreed/strongly agreed (46%; CI, 42% 51%) than disagreed/strongly disagreed (26%; CI, 22% 30%). It is a very entrenched aspect of human psychology to blame others. 40 Blame is one of the core elements of vulnerable system syndrome 40 in organizations, whereby organizations become more liable to adverse events because blame inhibits thorough and far-reaching investigation of adverse events and also impedes the development of robust and multilayered safety systems. 40 To create a safety culture, fair treatment of those who make errors and less punishment and blame are required ANESTHESIA & ANALGESIA

8 Table 6. Section 3: Strategies to Improve Reporting Statement agree Agree Neutral Disagree disagree Generalized deidentified feedback 35.2% ( ) 59.0% ( ) 4.6% ( ) 0.5% ( ) 0.7% ( ) about reports received from the anesthetic community. Role models, e.g., senior colleagues, 35.0% ( ) 56.0% ( ) 6.5% ( ) 1.8% ( ) 0.7% ( ) departmental directors who openly encourage reporting. Legislated protection of information 45.2% ( ) 44.6% ( ) 7.7% ( ) 1.6% ( ) 0.9% ( ) you provide from use in litigation. Ability to report anonymously. 34.8% ( ) 49.2% ( ) 10.2% ( ) 4.4% ( ) 1.4% ( ) Clear guidelines about what adverse 19.7% ( ) 63.1% ( ) 11.6% ( ) 5.4% ( ) 0.2% ( ) events and errors to report. Information on how confidentiality will 25.3% ( ) 54.2% ( ) 12.8% ( ) 5.6% ( ) 2.1% ( ) be maintained if you supply your name. Individualized feedback to you about 20.4% ( ) 58.6% ( ) 14.8% ( ) 5.5% ( ) 0.7% ( ) reports you submit. Paper forms for reporting provided in 20.6% ( ) 52.1% ( ) 17.3% ( ) 9.3% ( ) 0.7% ( ) each theatre. More support from colleagues. 15.8% ( ) 53.8% ( ) 23.7% ( ) 5.3% ( ) 1.4% ( ) Less blame attached to those who 16.0% ( ) 50.8% ( ) 19.7% ( ) 10.9% ( ) 2.6% ( ) report errors. ANZCA Continuing Professional 12.5% ( ) 53.7% ( ) 18.3% ( ) 12.0% ( ) 3.5% ( ) Development points for reports. Access to computer-based reporting 18.7% ( ) 46.3% ( ) 16.9% ( ) 13.9% ( ) 4.2% ( ) systems from home. Education about the purpose of 9.7% ( ) 49.0% ( ) 28.8% ( ) 10.9% ( ) 1.6% ( ) reporting. Computer-based reporting systems. 13.4% ( ) 38.9% ( ) 27.1% ( ) 15.7% ( ) 4.9% ( ) Training on how to use computerbased 9.9% ( ) 41.7% ( ) 31.5% ( ) 12.3% ( ) 4.6% ( ) system. Training on how to fill in paper forms 9.9% ( ) 35.3% ( ) 33.9% ( ) 18.3% ( ) 2.6% ( ) for reporting. Payment for time taken to report. 2.3% ( ) 17.2% ( ) 24.6% ( ) 39.9% ( ) 16.0% ( ) Data are presented as percentage with confidence interval. ANZCA Australian and New Zealand College of Anaesthetists. Culture has been defined as the way we do things around here, and can vary among and within organizations. 41 Although most of our sample participants disagreed/strongly disagreed or were neutral that medicine has a culture of silence where errors are not talked about, we still had 37% (CI, 32% 41%) of our sample participants who agreed/strongly agreed with this statement, suggesting that silence may still influence our participants. Section 2: Adverse Event With or Without Error Anesthesiologists in group Error were more likely to agree with barriers concerning litigation, disciplinary action, trouble, blame, lack of support from colleagues, and not wanting the case discussed in meetings, than those in group No Error. Traditionally, physicians have been discouraged by medical indemnity insurers, hospital lawyers, and risk managers from disclosing errors. 34 Reporting is inhibited by concern that adverse event reports disclosing error could be discoverable material by a plaintiff s lawyers. 2,42 Opinion varies as to whether error disclosure affects risk of litigation, 43 but 58% (CI, 51% 64%) of our group Error anesthesiologists perceived the risk of litigation as a barrier to reporting an adverse event caused by error, compared with 31% (CI, 25% 37%) of group No Error, suggesting that litigation remains a barrier to error reporting. In previous survey research, 21% of a mixed sample of physicians from medical units, surgical units, emergency departments, and intensive care units in the state of South Australia agreed (strongly agreed/agreed) I am worried about litigation is a barrier to reporting incidents. 20 Victoria and South Australia have similar tortbased adversarial legal systems and the same national specialist and general practitioner training programs. Our study s very specific and personalized case scenario may have resulted in more concern about litigation. A sense of isolation from colleagues and fear of criticism after making an error often have a major impact on physicians who have made a medical error, 15,39 which may account for anesthesiologists in group Error being twice as likely to agree/strongly agree than those in group No Error with the statement, My colleagues may be unsupportive. In previous South Australian research, 20 the barrier of My coworkers may be unsupportive was agreed (agreed/strongly agreed) with by 14% of surveyed physicians, similar to the percentage agree/strongly agree in our group No Error for My colleagues may be unsupportive but less than half the percentage agree/strongly agree for our group Error for this statement. Perceived support from colleagues may differ with whether or not one has made an error. The anesthesiologists in our sample seem to be aware of the importance of reporting adverse events for improved March 2012 Volume 114 Number

9 Adverse Event and Error Reporting Table 7. Section 3: Statistically Significant Demographic Influences of Years Qualified, Gender, and Practice Type on Strategies to Improve Reporting Statement Payment for time taken to report. Less blame attached to those who report errors. More support from colleagues. Role models, e.g., senior colleagues, departmental directors who openly encourage reporting. Demographic characteristic agree, % Agree, % Neutral, % Disagree, % disagree, % 20 y consultant a 1.8 ( ) 8.8 ( ) 26.3 ( ) 37.7 ( ) 25.4 ( ) y consultant b 0.0 ( ) 11.9 ( ) 25.0 ( ) 42.9 ( ) 20.2 ( ) 10 y consultant 2.3 ( ) 18.0 ( ) 25.8 ( ) 39.8 ( ) 14.1 ( ) Resident 5.0 ( ) 28.0 ( ) 21.0 ( ) 42.0 ( ) 4.0 ( ) 20 y consultant c 9.7 ( ) 39.0 ( ) 30.1 ( ) 16.8 ( ) 4.4 ( ) y consultant d 16.7 ( ) 45.2 ( ) 25.0 ( ) 9.5 ( ) 3.6 ( ) 10 y consultant 15.5 ( ) 58.1 ( ) 16.3 ( ) 7.8 ( ) 2.3 ( ) Resident 23.0 ( ) 60.0 ( ) 8.0 ( ) 9.0 ( ) 0.0 ( ) 20 y consultant e 11.5 ( ) 39.8 ( ) 39.0 ( ) 6.2 ( ) 3.5 ( ) y consultant f 11.9 ( ) 50.0 ( ) 29.8 ( ) 7.1 ( ) 1.2 ( ) 10 y consultant g 13.2 ( ) 64.3 ( ) 17.0 ( ) 4.7 ( ) 0.8 ( ) Resident 27.0 ( ) 59.0 ( ) 11.0 ( ) 3.0 ( ) 0.0 ( ) 20 y consultant h 25.4 ( ) 58.8 ( ) 14.0 ( ) 1.8 ( ) 0.0 ( ) y consultant i 25.0 ( ) 61.9 ( ) 7.1 ( ) 2.4 ( ) 3.6 ( ) 10 y consultant j 31.8 ( ) 61.2 ( ) 3.9 ( ) 3.1 ( ) 0.0 ( ) Resident 58.0 ( ) 41.0 ( ) 1.0 ( ) 0.0 ( ) 0.0 ( ) Ability to report anonymously. Female k 47.3 ( ) 41.2 ( ) 8.4 ( ) 2.3 ( ) 0.8 ( ) Male 29.4 ( ) 52.9 ( ) 11.0 ( ) 5.0 ( ) 1.7 ( ) Role models who openly encourage reporting. Private practice l 23.1 ( ) 61.2 ( ) 11.6 ( ) 3.3 ( ) 0.8 ( ) Public practice m 46.0 ( ) 48.4 ( ) 3.6 ( ) 1.6 ( ) 0.4 ( ) Mixed private/public 11.8 ( ) 78.0 ( ) 8.5 ( ) 0.0 ( ) 1.7 ( ) Data are presented as percentage with confidence interval. a More than 20 years as consultant versus residents, P b Ten to 20 years versus residents, P c More than 20 years versus residents, P , and 20 years versus 10 years, P d Ten to 20 years versus residents, P e More than 20 years versus residents, P , and 20 years versus 10 years, P f Ten to 20 years versus residents, P g Fewer than 10 years versus residents, P h More than 20 years versus residents, P i Ten to 20 years versus residents, P j Fewer than 10 years versus residents, P k Female versus male, P l Private versus public practice, P m Public versus mixed private/public, P Raw P values obtained from Wilcoxon-Mann-Whitney post hoc analysis after Kruskal-Wallis testing. P values shown have been corrected for multiple comparisons using Ryan-Holm step-down Bonferroni correction. quality of care. Only 2% (CI, 0.1% 3.5%) of participants in our study agreed/strongly agreed that Adverse event reporting makes little contribution to quality of care, compared with the 29% of the South Australian physicians who agreed that Adverse incident reporting is unlikely to lead to system changes. 20 Furthermore, in a survey of American physicians working in departments of internal medicine and surgery in a large Midwest academic medical center in the United States, the perceived barriers to error reporting of Not knowing the usefulness of the report and Thinking that reporting has little contribution for improvement of quality of care were among the top 6 of 17 barriers, 19 whereas in our study, Adverse event reporting makes little contribution to quality of care was the statement least agreed with of our 17 statements. Our participants were also less likely to agree with, I don t know whose responsibility it is to make a report, with only 4% (CI, 2% 6%) agreeing/strongly agreeing with this, compared with 38% of physicians in another study. 20 Similarly, only 26% (CI, 22% 30%) of our participants agreed/strongly agreed that The forms take too long and I just don t have time, compared with 54% of physicians in another study. 20 Section 3: Strategies to Improve Reporting The importance of obtaining generalized feedback is highlighted by the fact that Generalized deidentified feedback about reports received from the anesthetic community was the most popular strategy in our study. The need for role models who encourage reporting was similarly popular. Our participants commented very enthusiastically in our survey s free-text section about supportive and nonjudgmental department heads and senior anesthesiologists who discuss their own errors with junior staff, and actively lead the way in reporting. Leadership that actively encourages and supports reporting of incidents and errors is recognized in other safety-critical industries as essential to create a safety culture. 44 The need for legislated protection of adverse event and error reports from legal discoverability was also supported by participants. Successful reporting systems should be confidential, nonpunitive, systems oriented (recommending changes in systems, processes, or products), involve expert analysis by those who understand clinical settings, and be timely and responsive. 45 The anesthesia residents and less experienced anesthesiologists agreed or strongly agreed more than the more experienced anesthesiologists with the assistive strategies ANESTHESIA & ANALGESIA

10 of Less blame attached to those who report errors and More support from colleagues. It is unclear whether this is because the residents and less experienced anesthesiologists perceive themselves as being blamed and unsupported, or whether they are more educated about the effects of blame and lack of support on reporting, and are thus more able to identify and name what happens in their departments when they do report. The least popular strategy was Payment taken for time to report, although interestingly, our young participants were approximately 4 times more likely to agree with this than our older participants. This may reflect generational differences in expectations of payment for professional activities. Limitations of the Study Our sample participants were anesthesiologists and anesthesiology residents in Victoria, Australia, and thus the results may not be transferable to anesthesiologists in other countries and possibly, although less likely, to other Australian states. The medicolegal context in Victoria and other Australian states is an adversarial, tort-based system in which patients can sue for medical negligence. Different medicolegal environments in other countries may influence perceived reporting barriers. An advantage of our Australian setting is that a recently published survey research on reporting barriers is also Australian, 20 although this research was conducted in the state of South Australia rather than Victoria. However, medical practitioners in Australia undergo their specialist or general practice training, examinations, and accreditation under the auspices of national Australian, not state-based, organizations, which may minimize differences between participants in different states. Responder bias may also have influenced the results. For example, those who responded to our survey may be more interested in adverse event and error reporting than the nonrespondents. Other potential confounding factors include social desirability bias, 36 which may have particularly affected the results for sections 1 and 2. Furthermore, the order in which statements are presented in surveys, the wording used, and the context in which they are considered can also affect results. 46,47 For example, the survey items in section 2 were answered for a specific case scenario and patient outcome. Different cases and different patient outcomes may elicit different perceived barriers to reporting. Our study used previously published survey items where possible, but the items had not, to our knowledge, undergone psychometric analysis. A validated measure of health professionals attitudes to clinical adverse event reporting has been developed using a mixed group of physicians and nurses. 48 However, our study has a more specific focus on emotional barriers and error reporting. CONCLUSION The majority of anesthesiologists in our study did not agree that attitudinal/emotional barriers would operate in the context of reporting an unspecified adverse event caused by an error, with the exception of concern about being blamed by colleagues. When a specified anaphylaxis event had been caused by an error, 6 perceived barriers to reporting the specified event differed with the presence or absence of error. Anesthesiologists in our study strongly supported assistive reporting strategies. Finally, there seems to be differences between our sample and other physician groups for some perceived barriers to reporting, such as the contribution of adverse event reporting to quality of care, time required to complete forms, and whose responsibility it is to report. DISCLOSURES Name: Gaylene C. Heard, MBBS, FANZCA, MHumanFact. Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript. Attestation: Gaylene C. Heard has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files. Name: Penelope M. Sanderson, PhD, FASSA. Contribution: This author helped design the study, write the manuscript, review the data analysis, and give statistics advice. Attestation: Penelope M. Sanderson has seen the original study data, reviewed the analysis of the data, and approved the final manuscript. Name: Rowan D. Thomas, MBBS, FANZCA, MPH. Contribution: This author helped conduct the study, analyze the data, and review the manuscript. Attestation: Rowan D. Thomas has seen the original study data, reviewed the analysis of the data, and approved the final manuscript. ACKNOWLEDGMENTS The authors thank Professor John Ludbrook, University of Melbourne, for statistics review and advice. The authors also thank Associate Professor David Story and the ANZCA Trials Group for permission to mail the survey to ANZCA fellows and trainees, and the staff at ANZCA who addressed and mailed the survey packages. REFERENCES 1. Gaba DM. Anesthesiology as a model for patient safety in health care. BMJ 2000;320: Kohn LT, Corrigan JM, Donaldson MS, eds. To Err Is Human: Building a Safer Health System. Washington, DC: National Academy Press, Department of Health. An Organisation with a Memory: Report of an Expert Group on Learning from Adverse Events in the NHS. Norwich, UK: The Stationery Office, Bagian JP, Lee C, Gosbee J, DeRosier J, Stalhandske E, Eldridge N, Williams R, Burkhardt M. Developing and deploying a patient safety program in a large health care delivery system: you can t fix what you don t know about. Jt Comm J Qual Improv 2001;27: Classen DC, Pestotnik SL, Evans S, Burke JP. Computerized surveillance of adverse drug events in hospital patients. JAMA 1991;266: Cullen DJ, Bates DW, Small SD, Cooper JB, Nemeskal AR, Leape LL. The incident reporting system does not detect adverse drug events: a problem for quality improvement. Jt Comm J Qual Improv 1995;21: Stanhope N, Crowley-Murphy M, Vincent C, O Connor AM, Taylor-Adams SE. An evaluation of adverse incident reporting. J Eval Clin Pract 1999;5: Vincent C, Stanhope N, Crowley-Murphy M. Reasons for not reporting adverse incidents: an empirical study. J Eval Clin Pract 1999;5: Catchpole K, Bell MD, Johnson S. Safety in anaesthesia: a study of 12,606 reported incidents from the UK National Reporting and Learning System. Anaesthesia 2008;63:340 6 March 2012 Volume 114 Number