Teamwork and effective communication are considered to be important for the safe care of patients.1 This is especially poignant in the cloistered environment of the operating room (OR) where interprofessional teams work together in situations where an error can rapidly result in adverse consequences for patients.2 Because small errors can chain together in unexpected ways that slip past organizational, technological, and human defenses,3–5 individuals speaking up is often the final barrier to an adverse event in the making.6–10 A health care team member speaking up to raise concerns about risky or inappropriate actions of other team members can have a direct, immediate, and preventive effect on adverse outcomes.11 To strengthen this final barrier to an error chain reaching a patient, empowering health care team members to speak up has been part of widely disseminated teamwork training programs, such as the MedTeams project, the Veterans Health Administration Medical Team Training program, and TeamSTEPPS (Team Strategies and Tools to Enhance Performance and Patient Safety).12–14
Hurdles to speaking-up behaviors (those factors that are perceived to prevent a person from speaking up) have been examined in many fields outside of health care including psychology, business, and aviation. Cultural, professional, and organizational socialization predispose people to avoid speaking up in hierarchies where voicing concerns might be interpreted as disloyalty, disobedience, or disrespect.15–19 When employees in business confront speaking-up situations they weigh both the perceived safety versus the perceived cost of using their “voice” and the perceived efficacy versus the perceived futility of their speaking up. Both contextual factors (e.g., organizational culture, leadership style, workgroup size and structure) and individual factors (e.g., personality, experience and tenure, job attitudes) can affect the likelihood that employees will speak up in situations where they identify an error.19 Empirical studies, performed in the shadow of aviation accidents where the failure of aircrew members to speak up has been implicated as a cause, have identified factors associated with silence in that field. For example, relational issues (e.g., perceived hierarchy, fear of damaging relationships, avoiding conflict), content-related reasons (e.g., perceived futility of speaking up, efficiency versus safety), individual issues (e.g., fear of embarrassing someone, fear of being identified or labeled negatively, self-concept), and environmental issues (e.g., maintaining a positive team climate, fear of repercussions, cost versus safety) have been identified as leading hurdles to speaking up, and vary in importance for different role groups (e.g., captains, copilots, pursers, flight attendants).20–22
However, despite the existing literature in psychology, business, and aviation, there is little evidence about the social dynamics of speaking-up behaviors within health care. Within studies using self-report measures from questionnaires and interviews, factors such as not wanting to appear incompetent, bother or offend, and/or violate hierarchical norms are cited,23–29 but only one broad taxonomy of hurdles to speaking up, by one of us, has been offered.30 As far as we know, very few studies have looked beyond self-reported speaking-up intentions to observe actual speaking-up behaviors. In one such study, interns exposed to a 2.5-hour intervention on teamwork that included the skill of assertion (i.e., an intern would assert an opinion, through questions or statements, about the resuscitation process during critical times), which was added to the standard Neonatal Resuscitation Program, used assertion more often than control subjects in simulated resuscitation scenarios.31 In another study, engaging anesthesiology residents in a simulated scenario where opportunities to speak up to surgeons, supervisory anesthesiology faculty, and nurses were followed by a debriefing session resulted in an increased frequency of residents speaking up.32 On the other hand, O’Connor et al33 showed that a 90-minute training intervention that included instructions on speaking-up techniques did not lead to differences in speaking-up behaviors among interns. Kolbe et al34 found that speaking-up behaviors by nurses, but not residents, on a nurse and anesthesiology resident team were related to better technical performance of the team in simulated events during an anesthesia induction sequence. In a simulation-based study in anesthesia, Weiss et al35 showed that when confronted with at least one critical speaking-up situation, 41% of the participating nurses and residents did not speak up. In an interview study, Schwappach and Gehring36 showed that oncology health care professionals were particularly concerned with how to speak up and reported a lack of knowledge about speaking-up techniques. In sum, these few studies indicate that speaking up in health care (1) is beneficial but not common, (2) is difficult to train, and (3) could benefit from individuals feeling competent about how to speak up. However, as far as we know, educational means to encourage speaking up in ORs, where interprofessional teams work together intensely in time-critical, high-risk situations, have not yet been studied. All of the aforementioned studies focused their efforts on interns, residents, and nurses; it is unclear whether experienced, practicing physicians would exhibit similar behaviors, as role groups deemed to be of a higher hierarchical status have not been examined. Additionally, no study that we know of has examined enablers—those factors that are perceived to allow a clinician to speak up in spite of whatever hurdles may exist.
Thus, we addressed three questions in this study. First, would an educational intervention be effective in improving speaking-up behaviors of practicing nontrainee anesthesiologists when presented with realistic simulated clinical situations? Second, what would those speaking-up behaviors be when the issue emanated from a surgeon, a circulating nurse, or an anesthesiologist colleague? Third, what were the hurdles and enablers to speaking up in those situations?
The institutional review board of Partners HealthCare determined this study to be exempt.
We conducted a simulation-based randomized controlled experiment during a mandatory crisis management course for practicing nontrainee anesthesiologists from four academic medical centers and one community hospital in Boston, Massachusetts. From March 2008 to February 2011, we conducted the experiment during 71 regularly scheduled six-hour courses at the Center for Medical Simulation, which is also in Boston. We randomly designated each course as part of the intervention or control group in advance using a random number table. Each course hosted between 3 and 8 anesthesiologists with no more than 2 individuals from any one institution represented, with a total pool of 340 anesthesiologists attending the courses. At the beginning of a course, an independent technician randomly selected a single participant by the roll of a die to be the experimental “subject.” The remaining course participants were “observers” who participated in an unrelated scenario, but not in the one used for this experiment. Observers did actively participate in the debriefing session that followed the experimental scenario.
For both intervention and control groups, a one-hour standardized introduction to the course and a simulation scenario and debriefing session unrelated to this experiment preceded the experimental scenario. Following the unrelated scenario and debriefing session, we exposed participants in the intervention group to the intervention itself: a 50-minute workshop on speaking up. The workshop included an interactive didactic presentation that included a patient safety rationale, a rubric for speaking up modified from aviation (i.e., the two-challenge rule*), conversational skills (advocacy plus inquiry), and a role-play exercise. Then the experimental scenario was conducted. In the control group, we exposed participants to the 50-minute workshop on speaking up after the experimental scenario was conducted.
We designed the experimental scenario to elicit an opportunity to speak up, in turn, to a surgeon (designated event 1), a circulating nurse (designated event 2), and an anesthesiologist colleague (designated event 3) (all actors). The clinical case used for this scenario was that of a lightly sedated but awake patient undergoing a craniotomy with motor-mapping and tumor resection. The subject was first asked to relieve an anesthesiologist, who happened to be the anesthesia clinical director (ACD), at a midway point in the surgery. Sometime after taking responsibility for the anesthesia care, a circulating nurse took the subject aside and told him/her that the surgeon had been working for the prior 36 hours and was behaving in a sleepy manner (event 1). After the subject responded to this event, a phone call from the pathologist was answered by the circulating nurse, who, in due course, announced twice that he/she was going to put the call on speakerphone, an act that would allow the patient to hear her diagnosis and dismal prognosis (event 2). After the subject had responded to this event, the patient became agitated and the physiologic monitor indicated a sudden fall in the end-tidal carbon dioxide followed by profound hypotension indicative of a venous air embolism (VAE). The ACD returned to the room in response to a call for help from the circulating nurse, diagnosed the event as a VAE, and prescribed an incorrect treatment, changing the patient’s position by putting the head up (event 3). After responding to this event, the patient was successfully resuscitated by the subject, ACD, and other course actors.
Table 1 summarizes the three events in which the subject was expected to speak up and the action that was desired from the subject. Event 1 was predicated on the notion that a surgeon, who had been reported as sleep deprived by the circulating nurse and showed obvious signs of fatigue, might represent a patient safety risk. We expected the subject to speak up to the surgeon, ultimately getting help for him/her. Event 2 was predicated on the idea that the use of a speakerphone by a nurse in an OR with an awake patient is inappropriate. We highlighted and foreshadowed the inappropriateness of a call from the pathologist about an awake patient’s diagnosis with a prior speakerphone conversation about a different patient in the intensive care unit having a seizure, which caused the awake patient to become anxious and upset. Having had this experience, we expected the subject to verbally intervene and prohibit the circulating nurse from putting the pathologist’s telephone call about the awake patient on the speakerphone. Event 3 was predicated on the common knowledge that one of the treatment steps for a VAE is to put the patient in a head-down or flat position to prevent further air entry. Having the ACD make the mistake of calling for the head up was meant to add the weight of confronting a colleague of a higher hierarchical position to the decision to speak up. We expected the subject to challenge the ACD and ultimately order the head of the bed to be placed down or flat.
On completion of the experimental scenario (25 ± 4 minutes in duration), a trained facilitator (D.B.R., J.W.R.) led a structured debriefing session on the scenario (51 ± 13 minutes in duration). The discussion topics included each of the three speaking-up opportunities (events 1–3); anesthetic management of an awake craniotomy patient; signs, symptoms, and treatment of a VAE; and teamwork during resuscitation. The experimental scenario and debriefing session were video recorded.
At the end of each course, all participants were asked to complete a course eval uation survey asking them to rate the overall quality of the course, the quality of the instruction, and the quality of all the scenarios in the course on a five-point scale (excellent, above average, average, below average, poor). Participants were not explicitly told which scenario was part of the study. A total of 337 participants completed the course evaluation survey (3 course evaluation surveys were inadvertently not distributed).
Experimental scenario video analysis
Two investigators (M.C.M.P.-S., R.D.M.), blinded to the course’s assignment to the intervention or control group, were trained using the first three videos from each group (intervention [n = 35] and control [n = 36]) to transcribe the conversations involving the subject during the experimental scenario and then to code the conversations and the actions taken for each of the three events. Both investigators coded all of the transcribed conversations and actions, and a third investigator (D.B.R.) resolved any conflicts. For event 1, the actions noted included whether the subject asked the surgeon if he/she wanted help and/or asked the circulating nurse to get help for the surgeon. For event 2, the action noted was whether the subject tried to stop the circulating nurse from putting the pathologist on speakerphone either before or after the patient heard her diagnosis. For event 3, the action noted was whether the subject verbally responded to the command to “put the head up” after a VAE was diagnosed, allowed the bed to be put in a head-up position, and/or asked for an explanation (because the correct treatment for a VAE is to position the head down or flat).
Experimental scenario debriefing session video analysis
A single investigator (M.K.), with a background in organizational psychology, analyzed a balanced random sample of 40 experimental scenario debriefing session videos (20 from the control group and 20 from the intervention group). The investigator was blinded to the study hypotheses, the courses’ assignment to the intervention or control group, and subjects’ performance during the scenario. The investigator coded hurdles (factors participants reported as keeping them from speaking up) and enablers (factors participants reported as helping them to speak up) by means of event sampling based on verbal statements. For the purposes of this study, a coding unit was defined as a participant statement that could be assigned to at least one code.
Hurdles were coded on the basis of Raemer’s30 taxonomy, and the investigator supplemented new codes when necessary. To validate new codes, we consulted the organizational behavior literature on speaking up and information sharing for conceptual validation of other reasons for not speaking up.15,16,19,20,30,37–48 There were six hurdle codes (gender issue; stereotypes of others on the team; did not realize speaking-up opportunity was happening; impaired situational awareness: crisis or complexity; not considered a speaking-up situation; and futile to speak up) for which we could not find a respective equivalent in the existing frameworks. Because these six codes represented a considerable number of participants’ statements, they were integrated inductively in the final code list.
Similarly, enablers were coded based on a list of deductively and inductively developed codes. We used Raemer’s30 taxonomy of hurdles to speaking up as an initial framework. For example, whereas familiarity with the individual was a frequently mentioned hurdle to speaking up, many participants mentioned it as an enabler as well (e.g., “We’ve known each other for a long time, this makes it easy to say something”). Thus, some of the hurdle codes could be adopted as enabler codes. When participants mentioned an enabler not yet represented by the initial code list, the investigator created a new code. Again, to validate new codes, we consulted the organizational behavior literature on speaking up and information sharing and looked for conceptual validation wherever possible.15,16,19,20,30,37–48 Otherwise, codes were added inductively in an attempt to extend recent conceptualizations of speaking up.
For codes, definitions, examples, and related references for hurdles and enablers to speaking up, see Supplemental Digital Tables 1 and 2 at http://links.lww.com/ACADMED/A322.
We used HyperRESEARCH, version 3.5.2 (ResearchWare, Inc., Randolph, Massachusetts) to manage codes.
From prior studies32,34 and our experience with behavioral changes after receiving simulation-based education, we estimated that our subjects would speak up in one-third of the independent speaking-up opportunities they encountered in the experiment without an educational intervention. Additionally, as our experience indicated substantial immediate improvement in performance of learned skills from simulation-based education, we expected our subjects to speak up in two-thirds of independent speaking-up opportunities immediately following an educational intervention. To detect this improvement in speaking up with 95% probability and a power of 80% to reject the null hypothesis were it false, we would require 34 subjects in each group.
We compared the intervention and control group subjects’ institutional distribution and experience levels using chi-square and Mann–Whitney U tests, respectively. We compared the course evaluations from all intervention and control group participants using chi-square tests. We counted and compared the actions for each event for the intervention and control group as a proportion using a two-tailed Fisher exact test. For all of these comparisons, we considered P < .05 significant.
For the analysis of the debriefing session videos, we exported the raw coded data from the HyperRESEARCH software into SPSS Statistics, version 20 (IBM Corp., Armonk, New York), which we used for further data analysis. A matrix was produced in which each row included the participant’s statement, event number, and all codes that we assigned to the statement.
For a method summary, see Figure 1, a CONSORT (Consolidated Standards of Reporting Trials) experimental flow diagram showing subject recruitment and the reasons for which videos were excluded from the analysis.
There were no statistically significant differences in institutional membership (P = .80) or experience (14.8 ± 11.1 years since completing anesthesia training) (P = .42) between the intervention and control group participants. There were no statistically significant differences between intervention and control group course evaluation surveys for the overall quality of the course (P = .97), the quality of the instruction (P = .76), or the quality of all the scenarios in the course (P = .56). Of the combined participants in the simulation-based crisis management courses (n = 337), for the overall quality of the course, the quality of instruction, and the quality of all the scenarios in the course, 324 to 325 (96%) participants rated these items as excellent or above average, 4 to 5 (1%) rated them as average, 0 (0%) rated them as below average or poor, and 8 (2%) provided no rating.
Subjects’ actions when presented with events 1–3
Table 2 shows the incidence of speaking up for the three events. For event 1, 19/32 (59%) and 15/32 (47%) subjects in the intervention and control groups, respectively, asked the sleepy surgeon if he/she wanted help and asked the circulating nurse to call for help. For event 2, 4/32 (13%) and 5/31 (16%) subjects in the intervention and control groups, respectively, tried to stop the speakerphone conversation prior to the patient hearing her diagnosis. For event 3, 9/31 (29%) and 6/32 (19%) subjects in the intervention and control groups, respectively, verbally disagreed with the head-up treatment for the VAE and ordered the position to be changed to head down or flat. No statistically significant difference between the intervention and control group subjects was seen for any of the observed speaking-up actions for any of the three events (for P values, see Table 2).
Hurdles and enablers to speaking up
There were 40 subjects and 109 observ ers (149 participants total) in the 40 debriefing session videos analyzed. Because subjects and observers equally contributed to the debriefing sessions (i.e., all of them reported personal hurdles and enablers to speaking up), we pooled codes from the comments made by both subjects and observers for analysis. The number of instances of each code were counted and compared as a percentage of the total number of codes between the intervention and control groups. Finding no differences between the codes used by the groups, all of the codes were pooled and presented with respect to their category (i.e., climate, content, relational, and self) (see Figures 2 and 3). Of the 911 comments coded about hurdles, the five most frequently mentioned hurdles to speaking up were uncertainty about the issue (119 [13%]), stereotypes of others on the team (i.e., simplified conceptions of members of a certain group such as surgeons or circulating nurses) (68 [7%]), familiarity with the individual (62 [7%]), respect for experience (61 [7%]), and the repercussion expected (56 [6%]). Not having an environmental norm (36 [4%]) was mentioned as a further hurdle to speaking up. Of the 764 comments coded about enablers, the five most frequently mentioned enablers to speaking up were realizing the speaking-up problem (148 [19%]), having a speaking-up rubric (110 [14%]), certainty about the consequences of speaking up (63 [8%]), familiarity with the individual (60 [8%]), and having a second opinion or getting help (58 [8%]). Some participants mentioned that they should have spoken up (29 [4%]), and some reported that additional enablers to speaking up were maintaining professional respect (39 [5%]) and having an environmental (44 [6%]) as well as an internal speaking-up norm (23 [3%]).
There are three interrelated findings about speaking up from this study involving practicing nontrainee anesthesiologists from five different institutions (four academic medical centers and one community hospital). First, the educational intervention—which included a patient safety rationale for speaking up, a rubric for speaking up (the two-challenge rule), conversational skills, and a role-play exercise—was not effective in getting subjects to speak up more frequently to a surgeon, a circulating nurse, or an anesthesiologist colleague in the three events we tested. Importantly, this is powerful evidence that speaking-up behaviors are deeply rooted and difficult to change, at least with education alone. The emerging curricula for teamwork training in health care, virtually all of which contain educational content to enhance speaking up by providing rubrics, code words, encouragement, rationale, analogies, and slogans, may be insufficient on their own to get clinicians to speak up more readily.
The second finding was the limited frequency and quality of speaking up by the practicing nontrainee anesthesiologists in the presented events. It is reassuring that almost all, around 90%, of our subjects did something constructive to get help for the sleepy surgeon. The circulating nurse who alerted the subject to the surgeon’s lack of sleep and expressed his/her concern for the patient directly to the subject may have facilitated this. About 15% of our subjects in both the intervention and control group made what we considered to be a weak attempt to do something to protect the patient’s safety by asking the surgeon if he/she wanted help but not following up when the surgeon denied needing help. Add to this that about 10% of subjects from both groups did nothing to speak up in this situation and a full quarter of our subjects did not effectively speak up for the patient’s safety. Only half of our subjects spoke up in a manner that we considered to be the most effective—speaking up to the surgeon and then asking the circulating nurse to get help from outside the OR.
In the case of event 2, the majority of subjects (approximately 60%) did nothing to try to stop the inappropriate speakerphone conversation, and about another 25% of subjects only tried to stop the conversation after the patient heard her diagnosis. As we believed we had made the situation quite obvious and egregious, it is surprising that so few of our subjects spoke up. Our impression from the debriefing sessions was that in this event the attention of the subject was often so intensely focused on his/her immediate care of the patient that the larger environment of the OR was mostly ignored. Interestingly, among those who did speak up, the concern was often addressed to the surgeon, not the circulating nurse. As the nurse was the initiator of the error and was the only person with access to the speakerphone, it seemed to us that speaking up directly to the nurse would be the most appropriate action. The impression we had was that the subject was often reticent to supervise the circulating nurse and preferred that the surgeon convey the order to stop.
Most surprising of all was that only about 25% of subjects spoke up to their anesthesiologist colleague about the positioning of the head after the diagnosis of a VAE. In the debriefing sessions, there was no instance where someone argued that putting the head up was the correct treatment, suggesting that there was not a lack of knowledge about the correct treatment for a VAE. There were discussions about its priority, the relative effectiveness of putting the head down or flat, or the effectiveness of some of the other recommended maneuvers, such as trying to withdraw gas via a central line. The most common reason stated by subjects for not speaking up in this event was that there were too many issues to think about during this crisis situation.
The final finding was the extensive list of hurdles and enablers coded from the sample of debriefing session videos. Uncertainty about the issue was mentioned twice as often as any other hurdle. Similarly, realizing the speaking-up problem, certainty about the consequences of speaking up, and certainty about the issue were frequently mentioned enablers. These findings point to uncertainty as a particular hurdle to speaking up in health care. Notably, our subjects had on average about 15 years of experience since completing anesthesia training, indicating that “not knowing for sure” is a hurdle that is not limited to trainees. This hurdle is not yet explicitly included in recent models of voice behavior16,19,20 but has important implications for designing educational interventions, which should include how to speak up even when uncertainty about how to manage a patient care issue is present.
The next most common hurdles that participants reported had to do with social relationships, such as stereotypes of others on the team, respect for experience, and familiarity with the individual. The clinician’s view of teamwork and interpersonal processes at work are seemingly deeply embedded; this may partially explain the negative result of this study. Despite being experienced anesthesiologists, our subjects did not appear to self-identify as leaders with reasons legitimate enough to tackle the socially awkward task of speaking up for patient safety. Interestingly, familiarity with the individual was not only mentioned as a hurdle but also as an enabler. Further research on how clinicians decide whether a colleague they are familiar with is someone they can or cannot speak up to would be valuable. Similarly, maintaining professional respect (i.e., assuming the person who would be spoken up to is competent or ambitious enough to understand the concern) was mentioned repeatedly as an enabler of speaking up. Participants also often mentioned having a second opinion or getting help as an enabler. Overall, this set of findings extend current knowledge of speaking up by demonstrating that speaking up is not only an individual decision-making process but also a dynamic, social process. This suggests that including the social dynamics of speaking up in educational interventions (e.g., how to invite speaking up and how to deal with being spoken up to) may help improve speaking-up behaviors.
In line with current research on speaking-up climate, our results show that not having an environmental speaking-up norm is considered a hurdle to speaking up, whereas the existence of a speaking-up norm is considered an enabler. This suggests that promoting speaking up, promoting inviting speaking up, and promoting friendly or benign responses (and avoiding hostile responses) to speaking up as desirable or normative in an OR culture would encourage the practice of speaking up.
In addition, participants spoke fre quently of having a speaking-up rubric, in this instance a two-challenge rule, as an enabler to speaking up in the events presented. Apparently, the participants liked the concept of a two-challenge rule being disseminated within their OR environment even though in the experimental scenario those subjects in the intervention group who had learned this rubric prior to the scenario did not speak up more frequently than their colleagues who had not learned about the two-challenge rule beforehand. It could be that the subjects needed more practice or learning about the speaking-up rubric to feel comfortable using it themselves. Alternatively, the participants may have believed that it would need to be deeply embedded in the OR environment, in which surgeons and circulating nurses are included, before it would be effective, but this was beyond the scope of our analysis. Participants also quite frequently stated that they simply should have spoken up. This suggests that practicing speaking up in simulations might enhance clinicians’ confidence that they could do so in a real-world situation even if they did not do so in the simulation.
There are several limitations to our study that should be noted. First, the study was conducted in a simulation environment, and the actions taken in simulated events might be different than the actions that would occur spontaneously for an analogous problem in the real-world environment of an OR. Although we went to great lengths to provide a simulation with the highest fidelity possible, our subjects had all attended similar simulations in the past, and our course participants rated the quality of all the scenarios in the course very highly, there always remains the notion in the minds of the simulation participants that “this is not real.” How behavior is affected by each individual’s perception of this realism is not known. Additionally, factors such as unfamiliarity with the particular clinical case used in the scenario, the compression of time often noted in simulations, and deviation from usual practice patterns might have affected speaking-up behaviors. Similarly, another difference from the real world is that in these simulation courses, a group of colleagues are observing, which may affect the actions of the subjects. Although there appears to be a strong tendency for subjects in simulations to try to “do their best” because of the observation, the precise effect on their behavior is unknown. Third, we were able to present only three events to the subjects, one each for the surgeon, the circulating nurse, and the anesthesiologist colleague. The particular events had attributes that certainly affected the subjects’ actions. For example, some of our events (e.g., the sleepy surgeon) occurred in a context where “not much [was] happening” at the anesthesia station, while another of the events happened during an evolving anesthesia crisis (e.g., a VAE). Although it can be argued that the impetus to speak up is crucially important in each event, it is also important to note that the cognitive load on the subject was vastly different depending on the event. We must, therefore, be careful in extrapolating our results beyond the particular events we tested. Finally, we were able to present only a single educational intervention to the participants using a particular team of instructors. That the participants (most of whom were from academic medical centers) rated the quality of the instruction and the overall course very highly (excellent or above average) suggests that this is an exemplar intervention. Although we tried to present a compelling rationale and a practical rubric and to provide practice for speaking up over a reasonable time frame within this intervention, other educational interventions having different content, presentations, and/or durations could be more effective than ours.
An educational intervention—including a patient safety rationale, a speaking-up rubric, conversational skills, and a role-play exercise—alone was ineffective in improving the speaking-up behaviors of practicing nontrainee anesthesiologists; this was probably due to the substantial social- and uncertainty-related hurdles that inhibit speaking up. Other measures to change speaking-up behaviors such as an institutional emphasis on speaking up, including the adoption of a standardized rubric for speaking up, and educational interventions on how to invite speaking up, how to speak up, how to deal with being spoken up to, and how to speak up even when uncertainty is present, could be implemented and might improve patient safety. Expectations for education-only approaches to changing speaking-up behaviors should be tempered.
Acknowledgments: The authors would like to acknowledge the significant contributions of the anesthesia faculty of Harvard-affiliated hospitals; the faculty and staff at the Center for Medical Simulation; Brinda Kamdar, MD; and Mark Hoeft, MD, as well as the support of the Anesthesia Patient Safety Foundation, which made this work possible.
* In aviation, the two-challenge rule is that when observing an action that makes one uncomfortable, one should speak up by challenging the action from a position of curiosity. If the recipient of the challenge is unresponsive, the observer should speak up a second time by challenging from a position of concern. If the recipient is unresponsive to that challenge, the observer is authorized and expected to take over the controls. For health care, we have changed the last step to calling for help from someone in a position of authority to intervene.
1. World Health Organization. Multi-professional Patient Safety Curriculum Guide. http://www.who.int/patientsafety/education/curriculum/tools-download/en/
. Accessed October 8, 2015
2. Edmondson A.. Speaking up in the operating room: How team leaders promote learning in interdisciplinary action teams. J Manage Stud. 2003;40:1419–1452
3. Perrow C.. Accidents in high-risk systems. Technol Stud. 1994;1:1–19
4. Rudolph J, Repenning N.. Disaster dynamics: Understanding the role of quantity in organizational collapse. Adm Sci Q. 2002;47:1–30
5. Shrivastava P. Bhopal: Anatomy of a Crisis. 1987 Cambridge, Mass Ballinger Publishing
6. Edmondson A.. Disrupted routines: Team learning and new technology implementation in hospitals. Adm Sci Q. 2002;46:685–716
7. Hofmann D, Stetzer A.. The role of safety climate and communication in accident interpretation: Implications for learning from negative events. Acad Manage J. 1984;41:644–657
8. Reason J. Managing the Risks of Organizational Accidents. 1997 Brookfield, Vt Ashgate
9. Weick K, Roberts K.. Collective mind in organizations: Heedful interrelating on flight decks. Adm Sci Q. 1993;38:357–381
10. Weick K, Sutcliffe K, Obstfeld D.Sutton R, Staw B. Organizaing for high reliability: Processes of collective mindfulness. In: Research in Organizational Behavior: An Annual Series of Analytical Essays and Critical Reviews. 1999;Vol 21 Greenwich, Conn JAI Press:81–123
11. Okuyama A, Wagner C, Bijnen B.. Speaking up for patient safety by hospital-based health care professionals: A literature review. BMC Health Serv Res. 2014;14:61
12. Morey JC, Simon R, Jay G, et al. Error reduction and performance improvement in the emergency department through formal teamwork training: Evaluation results of the MedTeams project. Health Serv Res. 2002;37:1553–1581
13. Young-Xu Y, Neily J, Mills P, et al. Association between implementation of a medical team training program and surgical morbidity. Arch Surg. 2011;146:1368–1373
14. King H, Battles J, Baker D, et al.Henriksen K, Battles JB, Keyes MA, et al. TeamSTEPPS: Team strategies and tools to enhance performance and patient safety. In: Advances in Patient Safety: New Directions and Alternative Approaches, Performance and Tools. 2008;Vol 3 Rockville, Md Agency for Healthcare Research and Quality
15. Milgram S.. Behavioral study of obedience. J Abnorm Psychol. 1963;67:371–378
16. Detert J, Edmondson A.. Implicit voice theories: Taken for-granted rules of self-censorship at work. Acad Manage J. 2011;54:461–488
17. Detert J, Burris E.. Leadership behavior and employee voice: Is the door really open? Acad Manage J. 2007;50:869–884
18. Hirschman A. Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States. 1970 Cambridge, Mass Harvard University Press
19. Morrison E.. Employee voice behaviors: Integration and directions for future research. Academy Manage Ann. 2011;5:373–412
20. Bienefeld-Seall N, Grote G.. Silence that may kill: When aircrew members don’t speak up and why. Aviat Psychol Appl Hum Factors. 2012;2:1–10
21. Morrison EW, Wheeler-Smith SL, Kamdar D.. Speaking up in groups: A cross-level study of group voice climate and voice. J Appl Psychol. 2011;96:183–191
22. Bienefeld N, Grote G.. Speaking up in ad hoc multiteam systems: Individual-level effects of psychological safety, status, and leadership within and across teams. Eur J Work Organ Psychol. 2014;23:930–945
23. Belyansky I, Martin T, Prabhu A, et al. Poor resident–attending intraoperative communication may compromise patient safety. J Surg Res. 2011;171:386–394
24. Coats RD, Burd RS.. Intraoperative communication of residents with faculty: Perception versus reality. J Surg Res. 2002;104:40–45
25. Lyndon A, Sexton JB, Simpson KR, Rosenstein A, Lee KA, Wachter RM.. Predictors of likelihood of speaking up about safety concerns in labour and delivery. BMJ Qual Saf. 2012;21:791–799
26. Shreves JG, Moss AH.. Residents’ ethical disagreements with attending physicians: An unrecognized problem. Acad Med. 1996;71:1103–1105
27. Sutcliffe KM, Lewton E, Rosenthal MM.. Communication failures: An insidious contributor to medical mishaps. Acad Med. 2004;79:186–194
28. O’Connor P, Ryan S, Keogh I.. A comparison of the teamwork attitudes and knowledge of Irish surgeons and U.S Naval aviators. Surgeon. 2012;10:278–282
29. Reader T, Flin R, Lauche K, Cuthbertson BH.. Non-technical skills in the intensive care unit. Br J Anaesth. 2006;96:551–559
30. Raemer D.Yano E, Kawachi I, Nakao M. The clinician’s response to challenging cases. In: The Healthy Hospital: Maximizing the Satisfaction of Patients, Health Workers, and Community. 2010 Tokyo, Japan Shinohara Shinsha:27–32
31. Thomas E, Taggart B, Crandell S, et al. Teaching teamwork during the Neonatal Resuscitation Program: A randomized trial. J Perinatol. 2007;27:409–414
32. Pian-Smith M, Simon R, Minehart R, et al. Teaching residents the two-challenge rule: A simulation-based approach to improve education and patient safety. Simul Healthc. 2009;4:84–91
33. O’Connor P, Byrne D, O’Dea A, McVeigh TP, Kerin MJ.. “Excuse me:” Teaching interns to speak up. Jt Comm J Qual Patient Saf. 2013;39:426–431
34. Kolbe M, Burtscher M, Wacker J, et al. Speaking up is related to better team performance in simulated anesthesia inductions: An observational study. Anesth Analg. 2012;115:1099–1108
35. Weiss M, Kolbe M, Grote G, et al. Agency and communion predict speaking up in acute care teams. Small Group Res. 2014;45:290–313
36. Schwappach DL, Gehring K.. “Saying it without words”: A qualitative study of oncology staff’s experiences with speaking up about safety concerns. BMJ Open. 2014;4:e004740
37. Kobayashi H, Pian-Smith M, Sato M, Sawa R, Takeshita T, Raemer D.. A cross-cultural survey of residents’ perceived barriers in questioning/challenging authority. Qual Saf Health Care. 2006;15:277–283
38. Kruglanski AW, Webster DM.. Group members’ reactions to opinion deviates and conformists at varying degrees of proximity to decision deadline and of environmental noise. J Pers Soc Psychol. 1991;61:212–225
39. Rudolph JW, Simon R, Raemer DB, Eppich WJ.. Debriefing as formative assessment: Closing performance gaps in medical education. Acad Emerg Med. 2008;15:1010–1016
40. Edmondson A, McLain Smith D.. Too hot to handle? How to manage relationship conflict at work. Calif Manage Rev. 2006;49:6–31
41. Duhigg C. The Power of Habit. Why We Do What We Do in Life and Business. 2012 New York, NY Random House
42. Kish-Gephart J, Detert J, Treviño L, Edmondson A.. Silenced by fear: The nature, sources, and consequences of fear at work. Res Organ Behav. 2009;29:163–193
43. Liang J, Farh C, Farh J.. Psychological antecedents of promotive and prohibitive voice: A two-wave examinination. Acad Manage J. 2012;55:71–92
44. Edmondson A.. Psychological safety and learning behavior in work teams. Adm Sci Q. 1999;44:350–383
45. Milliken F, Morrison E, Hewlin P.. An exploratory study of employee silence: Issues that employees don’t communicate upward and why. J Manage Stud. 2003;40:1453–1476
46. Fast N, Burris E, Bartel C.. Managing to stay in the dark: Managerial self-efficacy, ego defensiveness, and the aversion to employee voice. Acad Manage J. 2014;57:1013–1034
47. Grant A.. Rocking the boat but keeping it steady: The role of emotion regulation in employee voice. Acad Manage J. 2013;56:1703–1723
48. Liu W, Tangirala S, Ramanujam R.. The relational antecedents of voice targeted at different leaders. J Appl Psychol. 2014;98:841–851