Building effective critical care teams

Critical care is formulated and delivered by a team. Accordingly, behavioral scientific principles relevant to teams, namely psychological safety, transactive memory and leadership, apply to critical care teams. Two experts in behavioral sciences review the impact of psychological safety, transactive memory and leadership on medical team outcomes. A clinician then applies those principles to two routine critical care paradigms: daily rounds and resuscitations. Since critical care is a team endeavor, methods to maximize teamwork should be learned and mastered by critical care team members, and especially leaders.

Team and organizational research indicates that poor performance occurs because professionals in psycho logically unsafe teams are less likely to engage in behaviors that contribute to team eff ectiveness [12]. First, they do not speak up with concerns and suggestions, which is critical for coordinating eff orts, rectifying errors, and improving processes [13,14]. For example, Edmondson [4] found that nurses in psychologically unsafe units spoke up less about medication errors. Less knowledgesharing in psychologically unsafe places has been found in a variety of organizational settings, ranging from restaurant chains to web services to aircraft component manufacturing [15].
Second, individuals in psychologically unsafe teams engage less in team learning [1]. Learning is critical to team eff ectiveness, particularly in contexts such as health care where teams constantly implement new practices. A series of follow-up studies on the aforementioned 23 neonatal intensive care units showed that units with higher psychological safety used signifi cantly more learning activities (for example, dry runs). In turn, they also experienced more successful implementation of new practices and 18% lower patient mortality after 2 years [16,17].
Lastly, studies have also shown that psychological safety predicts problem-solving, another contributor to team eff ectiveness. A study of 37 (intensive care, medical/ surgical, and maternity) units in 14 hospitals showed that nurses in more psychologically safe units were more likely to participate in eff orts to help solve systemic problems [18]. Likewise, in such units, there were less workarounds [19].

Lesson 2: enable the development of transactive memory
Experienced teams often solve problems better than newly formed teams [20]. One explanation for why team performance improves over time is that experienced teams often develop a transactive memory system that enables them to make better use of each individual's knowledge, skills and abilities [21]. A transactive memory system is a group-level memory system that often develops in close relationships and work teams. It involves the division of knowledge responsibilities among team members with respect to the encoding, storage, retrieval and communication of informa tion from diff erent knowledge areas and a shared aware ness among team members of 'who knows what' and 'who will do what' [22]. A useful metaphor for describ ing transactive memory is a set of networked com puters each containing diff erent information but with a common directory [23]. Th e division of responsibilities among members of critical care teams are likely to refl ect formal roles such as 'physician' or 'nurse' but others may relate to team member's unique skills and experiences.
For example, one member of the team may be especially good at calming family members while another may be especially good at fi nding a good vein for an IV. Accurate and shared knowledge about each member's roles and responsibilities on the team often translates into more eff ective communication processes: members know who to ask, what to ask, when to ask, and how to ask for information from others in the system, and they can anticipate when and how to communicate information eff ectively to those who need it [24].
Team behaviors associated with a well-developed trans active memory include: specialization, credibility, and task coordination [25]. As team members gain an understanding of one another's skills and expertise, they develop knowledge that is diff erent but complementary (specialization). However, this occurs only if they trust in their teammates' expertise (credibility). Task coordination (ability of team members to coordinate their work effi ciently) occurs as members develop high levels of specialization and credibility leading to high team performance.
One necessary condition for transactive memory develop ment is cognitive interdependence: individuals must perceive that their outcomes are dependent on the knowledge of others, and those others' outcomes are dependent on their knowledge [26]. Th is provides the motivation for team members to attend to what others can potentially contribute to the team. Cognitive interdependence can be instilled in teams in a number of ways: through team simulations and team training, teambased rewards and the development of positive work relationships. Informal interactions and shared experiences through daily rounds provide opportunities for members to learn about the relative expertise of other members, to coordinate who does what, to observe members' skills in action, and to build positive relationships. Th ose systems set up by formal design (such as a listing of staff responsibilities or procedures in an ICU training manual) are either validated or modifi ed over time as the team discovers whether individuals assigned to specifi c roles are capable and willing to perform them [27].
It takes time to develop a well functioning transactive memory system. Physicians in newly formed medical teams may not always know what information is needed by other team members. In a simulation of emergency medical diagnosis, twenty teams of either two or three physicians who had never worked together performed a patient handoff using a human patient simulator. In most of these teams, the physician who held the patient's chart did not read the earlier diagnosis, and did not talk to the room. As a result, many teams made the incorrect diagnosis because the physician with the chart did not understand the information needs and roles of other team members [28].
A primary supposition of the increasing use of simulation training for complex medical tasks/teams, includ ing critical care resuscitation, asserts that transactive memory is an essential component of building highly performing teams. Although there are no systematic studies conducted in critical care that have compared the performance of new and experienced teams, Hunziker and colleagues [29] examined 50 teams that performed cardiopulmonary resuscitation simulation exercises. Teams that included members who had not worked together before provided less hands-on care and later initial shocks compared to teams that had trained together before exercises [29].
Transactive memory is not just important for team performance but it also boosts team morale and commitment. Michinov and colleagues [30] conducted a crosssectional study of 193 nurses and anesthetists from eight French hospitals. Th e survey contained measures of transactive memory, work-team eff ectiveness and work attitudes. Th e researchers found that transactive memory positively predicted members' perceptions of team eff ectiveness and also aff ective outcomes such as team identifi cation and job satisfaction.
Although studies of team transactive memory in organi zations have focused on the benefi ts of member specialization, building in a certain degree of overlapping knowledge so that members can substitute for one another as necessary may be especially important for critical care teams where team membership changes on a daily basis and non-routine, time-sensitive, life-threatening decisions are commonplace.

Lesson 3: leaders' behavior matters
Team leader behavior is a well-documented contributor to team eff ectiveness [31]. How team leaders structure the team, defi ne its purpose, remove organizational barriers, help individual members enhance their contribu tions to the team, and coach to help members use their collective resources to fulfi ll team goals all infl uence team success [32]. Additionally, team leaders are infl uential on an interpersonal level because their behavior is salient to team members. Members look to leaders' behaviors as an indicator of what is expected and accepted of their own behavior [33]. Th us, leader behavior shapes team climate and team capabilities. With respect to team climate, for example, research shows that leader inclusiveness -the extent to which leaders, such as physicians or nurse managers, sought and appreciated others' input, made themselves available, and displayed fallibility -predicts psychological safety. In fact, research has found that the diff erence in psychological safety between physicians and non-physicians is minimized in health care teams with empowering leadership [8]. In these units, psychological safety is high, as is team learning and performance [12].
Leaders who adopt an empowering leadership style share power and give more autonomy to subordinates [34]. Some of the behaviors associated with an em powering leadership style include: leading by example, informing, coaching, participative decision making and showing concern for the team [35]. In studies of management teams in a variety of organizations, an empowering leadership style has been positively associated with know ledge sharing and higher levels of team effi cacy, which in turn are positively associated with team performance [36]. Empowering leadership styles have also been associated with the job satisfaction of subordinates [15]. It is important to note that many of these studies were conducted in management teams where teams were not making life and death decisions. Research on trauma resuscitation teams suggests that there may be other important contingencies for critical care teams.
Yun and colleagues [37] observed admissions, shadowed teams, and interviewed specialists in an ethnographic study of trauma resuscitation teams over a six-month period. Th eir fi ndings demonstrated that the eff ectiveness of diff erent leadership styles depended on the severity of the patient's condition and the level of team experience. Directive leadership was more eff ective when a patient was severely injured, whereas empowering leadership was more eff ective when a patient was not severely injured. Directive leadership was also better when a trauma team was inexperienced, but empowering leadership was better when a trauma team had high experience working together.
Although it has not been addressed in previous studies, it is quite possible that empowering leadership during patient rounds, quality improvement meetings and other routine low-stakes situations may enable critical care teams to coordinate more eff ectively during traumas under directive leadership. Th e psychological safety that develops through empowering leadership may increase the likelihood that team members will notice and speak up if they see a physician making an error during a resuscitation [38].

A clinician's synthesis
All critical care is 'high stakes'; survival and its quality depend upon optimal implementation of high-quality decisions. Some decisions can be made more deliberatively, and if errant, there is opportunity for reversibility, whereas others may cause immediate irreversible endorgan dysfunction or death. Daily rounds and bedside resuscitations are common team activities that demonstrate how intra-team relationships can be activityspecifi c yet highly complementary. Th e impor tance of leadership, psychological safety and transactive memory in team eff ectiveness becomes self-evident if we dissect these everyday activities.

High-stakes, time-to-ponder: daily care and 'unit management'
A week after fi nishing fellowship, I walked into 'my' ICU at a 350-bed teaching hospital. I was the unit's fi rst intensivist, team cohesion was poor and mortality was 34% for patients whose expected mortality was 24% [39]. While I had exemplary role models, my critical care fellowship provided no formal instruction on how to lead, organize and motivate teams. Consequently, the fi rst decade was marked by trial and (frequent) error; but I believed that the excess mortality justifi ed 'trans for mation by brute force. ' Despite the fact that patient outcomes improved, the fi rst year was harrowing for my teammates.
Th e directive style that had served me well in sports antagonized team members. Nurses and residents cried. Everyone complained. Change is 'hard, ' even when implemented through gradual consensus-building, and there are easy ways and hard ways to achieve healthcare goals. I had to 'stumble onto' how to facilitate team performance.
Interdisciplinary rounds are the foundation of critical care. Stakes of the discussions and decisions are high, but there is ample time to build team skills and empowerment. I mistakenly treated other team members as 'reporters' , there to provide me with information so I could make most of the decisions. While this (immature) modus operandi may be expected from a just-graduated trainee, eager to assert his new skills and directive discretion, it simultaneously disrespects and fails to leverage the talents and energies of teammates. Th e intensivist can not be at every bedside, every minute. She depends highly on others (especially nurses and physician extenders) to keep her apprised of substantial developments as they unfold. It is in the ICU leader's and patients' interests to have teammates who: are well trained, to distinguish between substantive versus irrelevant signal; have learned how to administer the multitude of daily, evidence-based therapies (automatically) without constant oversight and reteaching; and feel safe, empowered and prepared to speak up -during rounds or subsequently -when they see something that has not been reported to the team that they think may impact (the team's collective synthesis and thus) individual care plans or systems safety.
Strong leaders can, but they do not need to, dominate team discourse. Consensual leadership cultivates psychological safety, which in turn promotes positive transactive memories. Team rounds take a bit longer, but time is well invested because both routine and emergency care are positively impacted (see below). A number of concrete 'leadership practices, ' exercised during rounds, help achieve these goals.
First, leaders should actively illicit the observations and viewpoints of others on the team (even when he thinks he knows the answer). Historic hierarchies discourage non-doctors and/or junior team members from speaking up. If a patient's bedside nurse has failed to share his/her observations or suggestions, simply ask: 'Have we missed anything?' Th is technique simultaneously provides psychological safety by transmitting to the individual, and all team members, that their observations/opinions are welcome . . . or even expected. Transactive memory is formed that 'my perspective' matters -and so in future rounds (and emergencies; see below), non-physician team members feel more comfortable contributing their -often critical -observations.
Teaching is another important technique of promoting team spirit and psychological safety. Healthcare professionals share a heritage of self-improvement as a means to reduce human suff ering. By promoting a team culture of shared inquiry and intellectual discovery, the leader demonstrates benefi cence; genuine interest in the growth of teammates. Teaching must not be unilateral didacticism. All team members can be encouraged to bring their latest learning to the team for consideration. Meritorious ideas should be implemented, thereby demonstrating that good ideas (not just those of the team leader) will receive consideration and can be actualized. By cultivating rounds as a shared time/space for group learning, the democratic leader builds psychological safety and transactive memory. Both the team's interpersonal modus operandi and its ever-increasing library of shared learning and practices eventually become 'traits' -hard-wired, unconsciously, into daily activities.
Another powerful technique is to regularly name team victories and failures. Open, non-accusatory discussion of failures and errors permits us to share the negative emotions that can otherwise erode spirit and eff ectiveness. A particularly powerful tool is for the team-leader to choose his own mistakes -acknowledging fallibility, providing the 'safety' for others to do the same. By acknowledging that errors are part of human being, and that the only inexcusable sin is to not learn collectively from our errors, guilt can be neutralized or even 'spun' into positive experiences that cultivate willingness to selfreport for systems-improvement in the future (that is, positive transactive memory). Similarly, team victories can be named and shared. When a patient, not expected to live, leaves the hospital on a path to recovery, it is not the intensivist's victory but rather the result of the team's collective eff ort and energy. In medical ICUs, one in fi ve patients do not leave hospital alive, and a sizeable proportion of the remainder experience suboptimal survivor ship. So it is particularly important for the team not to get discouraged, to collectively celebrate the team's promotion of 'good saves' and 'good deaths' .
Team science concepts apply equally to monthly quality assurance meetings. Representatives from each group of professionals bring concerns or new practices for consideration by the group. If consensus builds that an item requires action, processes are deconstructed and reconstructed to address the problem or install a solution. Each team member brings a diff erent point of view to inform process engineering, champion implementation and monitor solutions.

Ultra-high stakes, no-time-to-ponder; cardiopulmonary resuscitation
On Th anksgiving Day, a young man with variceal hemorrhage was banded shortly after admission, but soon a river of blood poured from his intubated mouth. In the ensuing 2 hours, three nurses, a respiratory therapist, gastroenterologist, three residents, one fellow and I moved refl exively and simultaneously. Participating nurses' assignments were cross-covered, saline was squeezed by residents with the largest triceps, another central line was placed, another quick endoscopy failed, blood and body warmers were fetched and applied. Blood products were fetched and administered, and fi nally a Blakemore tube was placed. 'Th inking aloud' throughout, we shared ideas and modifi ed the approach several times as the situation evolved. Roughly 10 units of packed cells, 5 fresh frozen plasma, 3 platelets, 10 L normal saline and 24 hours later, he was stable and not one organ system had failed. Th e Blakemore was defl ated, repeat banding was successful, he was extubated and went home shortly thereafter. Th at Th anksgiving Day, without congas or fanfare, our team played 'the beautiful game' .
Each member's individual training and excellence, coupled with our team's transactive memory and psychological safety (to off er opinions that diff ered from mine), resulted in an optimal outcome, where he could have died that same day. Transactive memory from daily rounds and previous resuscitations were the predicate for both the coordination of eff orts (that is, knowing what to do, feeling comfortable acting independently even before an order was issued) and the real-time clinical problem solving (that is, everyone in the room felt safe to speak up and off er opinions about what to do next based on accumulated, pooled experience of the team). Note that my role necessarily became more directive than during rounds -teams expect a resuscitation leader, and deliberate democratic processes are temporarily/partly suspended. Th is highlights that 'contingent leadership, ' that is, determined by the venue or situation, is most germane to critical care. But even during emergencies and my more directive approach, nearly everyone off ered suggestions that helped us synthesize and strategize. Th inking aloud in our resuscitations is a quality/safety check (in case reasoning/logic is wrong) and broadcasts to the team 'what is next' , so team members can participate, prepare and better coordinate their activities.

Conclusion
Th ere are insuffi cient data to prove that behavioral science can be applied to improve ICU outcomes [40,41]. Nonetheless, ICU management is, fundamentally, teambuilding that is teachable and learnable. Educational programs might include formal training in behavioral sciences to complement young clinicians' developing repertoire of medical science. World Cups are not won by individuals, and since teams are the sharp end of the critical care stick, it is reasonable to believe that team performance impacts patient outcomes as much or more than brilliance of any one individual team member. Trainees in critical care need not struggle for a decade after training to acquire these skills by trial and error at the expense of nurses, therapists, patients and the team.