Role of Transactive Memory in ICU team Resuscitaton

CHELSEA KRAMER

The Role of Transactive Memory in ICU Team Resuscitation

The intensive care unit (ICU) is branded as a complex environment. Evolving situations, rapid changes to patient condition and frequent interruptions challenge multidisciplinary teams' ability to work effectively. The interdependent, often ad-hoc nature of ICU teams makes them vulnerable to teamwork skill deficiencies. Particularly, issues in non-technical skills such as communication and teamwork during a crisis situation that can affect patient care. Team members are often juggling individual goals in conjunction with team goals (e.g., getting the task done effectively and efficiently). As such, sharing information does not equate to effective communication. Team communication is prone to misinterpretation leading to error, particularly in multidisciplinary teams.

In the past two decades, team cognition has emerged as an overarching perspective from which to develop theories of team performance. Mental model theory posits that shared knowledge among members facilitates teamwork, but overlooks the knowledge differences inherent to multidisciplinary teams. A more differentiated view of team cognition is captured in transactive memory theory, which is concerned with the prediction of group behaviours through an understanding of the manner in which individuals differentially process and structure information within the group. While it has been studied extensively in dyads and small work groups, and somewhat in virtual teams and organizations, the role of Transactive Memory in high-risk domains has been largely ignored.

Critical care teams (such as ICU and OR) tend have inconsistent membership, lack team training, and are prone to communication issues. These conditions are not ideal for the development TMS, yet the strong cultural norms of healthcare team roles offer an interesting juxtaposition. Despite the lack of team training, ICU teams still manage to perform successfully. Standardized team roles can allow people who have never met to function effectively in safety-critical domains. In aviation, it is often the case that the pilots, co-pilots and flight attendants meet each other for the first time at the start of each flight and proceed to seamlessly execute hundreds of flights a year. Nevertheless, planes still crash and medical errors still occur, so the need to study variation in team performance is still relevant. In the ICU, a team's structure can be a blend of assigned and assumed roles associated with professional titles, cultural hierarchies and the needs of the situation. An inconsistent team structure would seem problematic for effective teamwork in complex environments. Transactive memory can provide a framework to explain successful team performance. Teams can compensate for shifting roles if members are aware of each other's knowledge and expertise. This knowledge builds implicit coordinating mechanism that in turn facilitates team processes such as communication and coordination, and allow teams to deal with crisis situations.

The goal of this project is to answer four research questions: 1) What is the importance and applicability of transactive memory in the high-risk healthcare domain? 2) What is the relationship, if any, between transactive memory and non-technical skills (e.g. communication), and between transactive memory and technical skills (i.e. performance)? 3) What is the validity and reliability of the Transactive Memory Systems Scale and a behavioural markers checklist for measuring TMS in ICU teams during simulated crisis situations? and 4) What is the validity of the proposed conceptual framework describing the relationship between TMS, technical, non-technical skills in regards to team performance?

To answer the first question, an extensive literature review of the published literature covers the topics of teamwork, healthcare complexity, and team cognition. This review will be coupled with an ethnographic approach of interviews and questionnaires with subject matter experts in the ICU domain (Study 1). The triangulation of these efforts will be used to formulate behavioural markers within specific context of ICU teams. In Studies 2a and 2b, questions two and three will be answered through an empirical analysis of videos of high-fidelity ICU simulation, using the behavioural markers derived from study 1. The fourth question will be answered in a detailed cross-analysis of the results of the two empirical sub-studies in order to validate the conceptual framework of TMS and performance in ICU teams.

1.1 Medical Context: Resuscitation

Cardiac arrest is internationally defined as the "cessation of cardiac mechanical activity, confirmed by the absence of a detectable pulse, unresponsiveness, and apnea" (Zaritsky et al., 1995). When the heart ceases to provide sufficient blood flow and oxygen to the brain, the victim collapses, and will appear lifeless in the absence of vital signs (Vaillancourt & Stiell, 2004). In the absence of cardiopulmonary resuscitation (CPR) and/or electrical defibrillation, such electrical cardiac activity disappears (i.e., asystole), followed by death in a matter of minutes. When cardiac arrest occurs, the immediate and skilled action of first responders is critical. Once the resuscitation team arrives, a coordinated rapid and efficient exchange of information, along with continuous hands-on CPR measures, is essential. The 2010 North American Heart Association Guidelines dictate CPR as part of the "Chain of Survival" algorithm for cardiac arrest response: 1) immediate recognition of cardiac arrest and activation of the emergency response system; 2) Early CPR with an emphasis on chest compressions; 3) Rapid defibrillation 4) Effective advanced life support and 5) Integrated post-cardiac arrest care (American Heart Association, 2010).

Two decades ago, resuscitation researchers dismissed the in-hospital cardiac arrest population as unsuitable for resuscitation research because it was composed mostly of patients whose cardiac arrest was the terminal event of a fatal illness (Jastremski, 1993). A standardized solution, the Utstein style of reporting arose out of concern that resuscitation endeavors in different countries (and within countries) could not be compared meaningfully (Cummins et al., 1997). The "In-Hospital Utstein-Style Template" was developed to summarize the critical data elements essential for documenting in-hospital cardiac arrest (IHCA): time of event onset, time of cardio-pulmonary resuscitation (CPR) started/stopped, time to first defibrillation, time to advance airway management, time to administration of first resuscitation medications and time of sustained returned of spontaneous circulation (ROSC) (Cummins et al., 1997; Zaritsky et al., 1995). Since the creation of the Utstein style, a surge of epidemiologically consistent literature has significantly advanced our understanding of the factors involved with IHCA. Unfortunately, we have learned that survival rates of IHCA are low, and have been for some time. The overall survival rate (usually 10 to 20%), from cardiac arrest is calculated by the number of patients discharged from a hospital after successful CPR, divided by the total number of patients in whom CPR was attempted (Skogvoll, Isern, Sangolt, & Gisvold, 1999). In spite of efforts to prevent arrest or enhance resuscitation care before, during and after IHCA, survival rates have not improved in the past four decades (Abella et al., 2012). In this complex healthcare environment, despite training and standardization to improve team performance, errors and inefficiencies persist even among experienced teams.

Resuscitation from cardiac arrest is a frequent intensive care unit (ICU) emergency that requires both technical (e.g. medical knowledge) and non-technical team skills. The majority of Canadian and U.S. hospitals have organized teams (rather than individual healthcare workers) to respond to in-hospital cardiac arrests (Hunziker et al., 2011). Despite substantial efforts to make cardiopulmonary resuscitation (CPR) algorithms known to healthcare workers, resuscitation teams often deviate from CPR algorithms (Mattei, McKay, Lepper, & Soar, 2002; Peberdy, Silver, & Ornato, 2009). Resuscitation guidelines provide a logical, sequential algorithmic approach, but they have mainly emphasized technical tasks performed by individual rescuers and have not addressed issues of adapting to the complex nature of most actual resuscitations. There is mounting evidence that errors in the teamwork associated with patient care may be contributing to poor outcomes (Ornato, Peberdy, Reid, Feeser, & Dhindsa, 2012; Peberdy et al., 2012). Even with the high degree of technical knowledge required, errors that occur in high-stakes medical environments (such as during a resuscitation) often result from a problem with team functioning, rather than from a lack of clinical knowledge (Baker, Day, & Salas, 2006).

The challenging nature of healthcare teams necessitates the ongoing study and refinement of the factors that contribute to teamwork effectiveness, as it ultimately makes up each patient's level of care. Team skills are inherently cyclical; what serves as an outcome of one variable may serve as an input to another (Salas, 2005). Teams may actively modify and adjust their behaviours accordingly (Frese & Zapf, 1994) Empirical research on the underlying cognitive structures propelling teamwork cycles will improve our ability to identify and train the skills necessary to improve performance in complex, safety-critical domains. The execution of these processes help shape an overarching team cognition (Gorman, Cooke, & Kiekel, 2004), which will serve as an overarching theme of this prospectus. The following sections provide a review of the relevant team cognitive factors, including teams, teamwork and team performance in the complex healthcare environment.

2.1 Teams

Over the past 40 years, a "golden age" of interest of team research has yielded over 130 models of team, teamwork and team performance (Salas, Cooke, & Rosen, 2008, p. 541). The word "team" typically refers to a group of two or more people that work together towards a common goal in a coordinated and coherent fashion (Paris, Salas, & Cannon-Bowers, 2000). Athletic teams exemplify this classic definition: team members all have a common goal of winning and practice together regularly (even multiple times a day); individuals typically have one clearly designated role or position that is known by other members; team members often develop some type of shared-language to aid communication (e.g. plays, call-signs); and the team typically is assigned an official team leader or captain (Fiedor, Hunt, & Devita, 2011). Team sports are not played or won by individual members. To succeed, they must function effectively and efficiently.

Organizations in high risk domains, such as aviation, nuclear, military and healthcare, also rely on teams to deal with the complex nature of their environments. However, compared to athletic teams, the need to effective performance, and the potential consequences of poor performance are much more severe. This concept has been demonstrated in the famous quick decision making witness during the "Miracle on Hudson". On January 15, 2009, US Airways Flight 1549 struck a flock a geese following takeoff from LaGuardia Airport in New York. The crew noticed that both engines had lost power and informed air traffic control (ATC). As per standard protocol, the ATC sought the closest alternative runway, which happened to be in New Jersey airstrip (Atkins, 2010). The flight 'team', while having the common goal of a safe flight from point A to B, was distributed among the pilots in the cockpit, the flight crew in the cabin, and the ATC in the tower. Through communication via radio headsets, the team was able to coordinate their different perspectives on the immediate situation. The ATC thought Flight 1549 had sufficient time to make it to the alternate landing strip, and followed all of the appropriate communication protocols with the pilots. However, while the various ATC operators were discussing their options, Captain Sullenberger judged the immediate situation and decided to conduct an emergency landing in the Hudson River resulting in all 155 surviving passengers (Eisen & Savel, 2009).

In, complex, fast-paced, dynamic domains, errors can lead to severe consequences when the task complexity exceeds the capacity of an individual; when the task environment is ill-defined, ambiguous, and stressful; when multiple and quick decisions are needed; and when the lives of others depend on the collective insight of individual members (Salas, Cooke, & Rosen, 2008).

Distributed responsibilities allow teams to process massive amounts information, thus reducing the cognitive load on individuals (Hazlehurst, 2004). The distribution of cognitive workload can allow teams to function more effectively than individual members (Fiore et al., 2010). For instance, working in teams (versus working alone) has been shown to mitigate the negative effects of task interruption as demonstrated by reduced individual workload, and faster resumption time to a resource monitoring task (Tremblay, Vachon, Lafond, & Kramer, 2011). The next section discusses how the beneficial impact or potential pitfalls of teamwork are important for the healthcare domain.

2.1.1 Healthcare Teams

Healthcare teams operate in an environment characterized by stress, workload, complexity and high stakes for decision and action errors (Salas, Rosen, & King, 2012). Healthcare complexity stems from the individual nature of patients; the frequency, public exposure, and applied investigation methods with respect to adverse incidents (e.g., Taneva, Plattner, Byer, Higgins, & Easty, 2010) and error (e.g., Wu, Folkman, McPhee, & Lo, 2003); training and evaluation of professional skills (e.g., Nurok, Lipsitz, Satwicz, Kelly, & Frankel, 2010); and the constant coordination and re-coordination of behaviours as each members' actions can immediately require another team member to react appropriately (Helmreich, 1996).

The dynamic, interdependent, and adaptive interaction of team members toward a common and valued goal (Salas, Dickinson, Converse, & Tannenbaum, 1992) facilitates performance of complex, critical tasks (Cooke, Kiekel, & Helm, 2000). In fact, many medical procedures such as surgery, emergency medicine, and anaesthesia can only be performed by teams. In these group procedures, clinical performance and patient safety are key functions of group coordination (Kolbe, Burtscher, Manser, Kunzle, & Grote, 2011). Multidisciplinary teams play a vital role in healthcare because some level of collaboration between healthcare providers is always necessary as no single discipline or specialty can meet all of a patient's needs (Ellingson, 2002). A hospitalized patient, for example, may need a physician to provide a diagnosis and treatment plan, a nurse to administer medications, a dietitian to monitor food intake, a physical therapist to aid in muscle strengthening and flexibility, and a social worker to coordinate home care following release (Ellingson, 2002). This is particularly true in dynamic domains of surgery, intensive care, and trauma where effective multidisciplinary teamwork has been shown to be important for safe patient care (Burtscher et al., 2011).

Teams composed of multiple disciplines must consider the coordinated needed among crews of the same profession, as well as coordination between professional groups. Anesthesia teams, for example, are constantly engaged in planning (e.g. planning the correct amount of a certain medication), problem solving (e.g. finding the correct reason for suddenly rising blood pressure), decision making (e.g. deciding on the right time for intubation), and psycho-motor performance (e.g. laryngoscopy) (Burtscher, Kolbe, Wacker, & Manser, 2011; Schulz et al., 2011). These crew-specific demands are further complicated by the need to coordinated with the other professions throughout "mixed-motive" medical procedures (e.g. choosing between the surgeon's request to proceed with the operation and the need to further stabilize the patient) (Kolbe et al., 2011).

The increasing complexity in healthcare and the demanding nature of healthcare services has made multidisciplinary healthcare teams an integral part in the delivery of patient care (Sutton, Liao, Jimmieson, & Restubog, 2005). However, in reality different professionals do not always understand each other and cooperation does not always work (Schoop, 1999). Perceptions of teamwork have been shown to vary substantially by caregiver type, with physicians often rating aspects of teamwork higher than nurses (Makary et al., 2006). Team members often must juggle individual and team goals, interacting with multiple team goals (Keyton & Beck, 2010). Physicians, nurses, pharmacists, technicians, and other health professionals must coordinate their activities to deliver safe and efficient patient care (Baker et al., 2006). According to their different interests and tasks, healthcare professionals may notice different things about the same patient and, therefore, may prioritize and react to different aspects of patient care (Reddy et al., 2001). Healthcare workers perform interdependent tasks (e.g., a surgeon cannot operate until a patient is anesthetized) while functioning in specific roles (e.g., staff surgeon, surgical resident, anesthesiologist) and sharing the common goal of safe care.

In many ways, multidisciplinary healthcare teams represent the antithesis to the well-trained athletic team because of the tendency to be formed ad-hoc. The ad-hoc nature of many types of healthcare teams makes it difficult to practicenecessary skills such as communication, organization, group problem solving. For instance, hospital trauma teams (e.g. code-blue) serve to prevent death in suddenly critically ill patients (Sarcevic, Marsic, Lesk, & Burd, 2008). Such a critical function should be the target of frequent and effective training programs, as is the case with daily practice sessions of athletic teams. However, different professions are brought together quickly assembled (e.g. trauma resuscitation) or for a very short period of time (e.g. OR team). Once the crisis or procedure is over, they may never work together (in whole or in part) again. In many cases, several members of the team (or even the whole team) may have never worked together before. In sport, the connection between ineffective behaviours of a newly formed team and the outcome are clearer; delays in action, miscommunication, and errors can result in a fumbled play or a lost game. In healthcare, the lack of team familiarity could increase the likelihood of a fatal patient outcome. The consequences of poor performance in teams working in high-risk domains can be life altering, meaning teams must function at the highest level (Boies & Howell, 2011; Wilson, Salas, Priest, & Andrews, 2007).The characteristics of effective teams must be able to account for organizations that are complex, tightly coupled, hierarchical; time compressed and relies upon coordinated actions. The proponents of teamwork necessary for effective teams are discussed next.

2.2 Teamwork

Teamwork is not a fixed state; it is a multidimensional, dynamic construct that refers to a set of interrelated cognitions, behaviors, and attitudes that occur as team members perform a task that results in a coordinated and synchronized collective action (Salas, 2005; Salas et al., 2008). Teamwork, particularly in safety-critical, HRO domains, is not a natural product of working together (Undre, Sevdalis, Healey, Darzi, & Vincent, 2007). Teamwork must be 'learned and practiced' (Wallin, Meurling, Hedman, Hedegård, & Felländer-Tsai, 2007). At the same time that teamwork is deemed necessary, it is also an often underappreciated and misunderstood skill (Paris et al., 2000; West & Field, 1995). Training interventions designed to improve team processes in healthcare are often based on common sense notions, positive team climate, or tradition rather than on empirical validation of their effects on performance (Salas et al., 2008). In these cases, teamwork is often used interchangeably with the notion of 'effective team coordination'-but the connection between team processes and performance is blurred. To differentiate these two descriptor, this section deals with a number of encompassing models of teamwork behaviours, followed by their coordinating mechanisms.

2.2.1 Models of Teamwork Behaviour

Teamwork is an essential component of safety in most work environments, especially in healthcare (Salas et al., 2012). Theories of teamwork provide an understanding of what is important to team effectiveness and hence what is important to measure, however multiple models exist. In the absence of an "ideal" model tailored for a specific domain (and has received empirical validation) general models provide an overview of the critical features of teamwork (Rosen et al., 2008). A number of desirable team characteristics (i.e., team knowledge, skills and attitudes) are identified in the literature as required for effective teams, regardless of membership consistency (Baker et al., 2006; Salas, 2005; Salas, Sims, & Klein, 2004). A frequently cited theoretical model, the "Big Five", describes teamwork (see Table 1) in terms of five core dimensions (i.e., team leadership, mutual performance monitoring, backup behaviour, adaptability, and team orientation) and three coordinating mechanisms (i.e., mutual trust, shared mental models, and closed- loop communication; Salas, 2005). Adopting this model of teamwork as the basis for team performance measurement system would mean that measures would be developed to capture each of the five core teamwork dimensions and the three coordinating mechanisms.

The expectations associated with the role of team leadership are daunting because failure of the team's leader to guide and structure team experience is posited by a wealth of research to have such an important impact on performance (e.g., Hunziker et al., 2010). Through a synthesis of the related research, the Big five identifies three overarching leader functions: 1) The team leader has a role in the creation, maintenance and accurate of the team's shared mental model, that is, a shared understanding of the team objectives, the team constraints, the roles of each team member, and the resources available to them. 2) The team leader facilitates team effectiveness by monitoring the internal and external environment of the team to facilitate team adaptability and the ensure teams are not caught off guard when the changes occur. 3) The team leader establishes behavioural and performance expectations, while tracking the abilities and skill deficiencies of each member. Salas (2005) argues that team leaders ultimately facilitate team effectiveness not only by synchronizing and combining the individual contributions of each of the team members but also by insuring individuals on the team understand their interdependence and the benefits of working together.

Mutual performance monitoring is defined in the big five model as the team's ability to keep track of fellow member's work while carrying out their own. The model proposes that the information gathered through mutual performance monitoring that affects team performance by identifying errors or lapses. This information, expressed through feedback and backup behavior, boosts the team from the sum of individual performance to the synergy of teamwork and ultimately to team effectiveness. Similar to leadership, the prerequisites for effective mutual performance monitoring are shared mental models of the task and team responsibilities. This mental model provides teammates with a common understanding of what other team members are supposed to be doing at any given moment, and what they should be doing next (Peterson, Mitchell, Thompson, & Burr, 2000).

As a compliment to mutual performance monitoring, the big five model proposes that backup behaviour affects team performance directly by ensuring that all aspects of the team tasks are completed. If a team member's workload threshold is surpassed, the team can engage in backup behaviors by shifting work responsibilities to other underutilized team members as it becomes necessary (this is a form of team adaptation, described next). Should the tasks of the overloaded team member not be facilitated or taken over, it is expected that team performance will drastically degrade. Depending on the type of task, the compensatory behaviour may manifest differently (e.g., physically taking of the task, ensuring that the error is corrected, or providing support). Again, back up behaviour would require the existence of some type of shared mental model and mutual performance monitoring as they would form the foundation for team members' decisions of when and how to provide the necessary backup.

Adaptability, as described by the big five model, is the team's capacity to recognize deviations from the expected action and readjust their actions accordingly (Salas, 2005). Adaptive behaviour helps teams respond to unexpected demands and assign meaning to that change, and to carry out a new plan of action. From the Big Five's perspective, adaptability includes not only the ability to changes, but that quality and effectiveness of that change to deal with the change in the environment. Like the connection among the aforementioned Big Five characteristics, effective adaptability is bound by shared mental models, effective engagement in mutual performance monitoring and backup behaviour.

The final Big Five dimension of team orientation is posited more as an attitude, versus a behavioural characteristic, as the previous four. Team orientation is viewed as both a preference for working with others but also a tendency to enhance individual performance through the coordination, evaluation and utilization of tasks inputs from other members while performing group tasks. Team orientation differs from team cohesion (e.g., Undre, Sevdalis, Healey, Darzi, & Vincent, 2006b), which is an attraction to work with a particular team rather than a general preference to work in team settings. This would have particular value to assessing healthcare teams which often work in ad-hoc conditions and evolving team membership.

Table 1.

Characteristics of Effective Teams; adapted from Salas (2005).     

Note: * denotes measures that could be used within the scope of this thesis.

Recognizing the frequent overlap of team performance models, Rousseau, Aubé and Savoie (2006) conducted a comprehensive literature review on the behaviors most likely to influence teamwork effectiveness. The authors defined a hierarchy of team behaviors (see Figure 1) organized around four functions based on action regulation theory (Frese & Zapf, 1994) whereby individuals steer their own activities in correspondence with some goal. During the cycle of action in regulation theory, people monitor their environments, gathering information in order to plan a course of action. While executing the plan, one can actively influence the environment, and the results are fed back regarding one's actions (Frese & Zapf, 1994).

In Rousseau and colleague's (2006) hierarchical model of team behaviours, goals, information collection, planning, execution, and feedback correspond to (a) preparation of work accomplishment, (b) task-related collaborative behaviors, (c) work assessment behaviors, and (d) team adjustment behaviors. Preparation for accomplishment refers to the analysis of the work to be done and planning the tasks. This includes the identification of the main tasks and analysis of the environment and available resources. Similar to the benefits of having good team leaderships as defined in the Big 5, the team is then able to define a common goal and develop a plan, meaning a series of actions to achieve this goal. Collaborative task-related behaviour refers to the implementation of the action plan established during the preparation stage. Similar to the Big 5's communication and adaptability, this function includes behavior coordination, cooperation and information exchange of information. During collaboration, team members can share all of the necessary information and work together to perform the task. Work assessment behaviours refer to monitoring the overall performance of the team, and the environment in which the work is performed. Similar to mutual performance monitoring, this ensures that the team is going in the right direction and nothing is compromising team performance. Team adjustment behaviours refer to the way in which the team responds to changes or observed deviations from the initial plans. Like the Big 5, these include back up behaviours, intra-team coaching, collaborative problem solving and innovation provide support to adjustment behaviours. Lafond, Jobidon, Aube and Tremblay (2011) empirically tested aspects of the Big Five and Rousseau's hierarchical model in a functional simulation of command and control in a complex and dynamic firefighting task. In their study of 24 three-person teams, the two most important predictors of team performance were mutual performance monitoring and team coordination. Specifically, more frequent communication, and when communication was goal-oriented, was positive predictors of team performance. The following section examines coordination mechanisms of teams.

Figure 1. Rousseau's (2006) schematic Representation of the Hierarchical Conceptual Structure of Teamwork Behaviours

2.2.2 Coordinating Mechanisms of Teamwork

As has been shown in the previous sections, teamwork is rarely, if ever, defined as a singular unit. It is generally broken down into a number of observable and non-observable behaviours such as communication, coordination and collaboration (Wilson et al., 2007). Although the specific behavioural aspects of various teamwork models can differ (Lafond et al., 2011), many team performance models are bound by similar 'coordinating mechanisms'. Effective team behaviours are believed to be facilitated by these internal mechanisms such as mental models, trust and attitudes, which cannot be observed directly. To understand the effect of certain teamwork behaviours on performance within a dynamic context, it is equally important to understand the related cognitive underpinnings.

Teams manage interdependent work by ensuring that the right things happen at the right time through coordination. Coordination can be viewed as "the management of interdependencies among activities in terms of actors, goals, time, space, quality of products" (Bardam, 2000, p. 158). Teams can coordinate explicitly (i.e., purposefully) using external artifacts (e.g., schedules, plans, procedures, etc.) or by communicating (e.g., orally, in writing, formally, informally, interpersonally, in groups) (Espinosa, Lerch, & Kraut, 2002). Communication as a coordinating mechanism in healthcare refers to the quality and the quantity of the information exchanged among members of the team (Undre, Sevdalis, Healey, Darzi, & Vincent, 2006b). Communication covers a wide range of interactions, including interpersonal communication, communication technology, medical education, health policy, and mass communication. It takes many forms, from a brief informal talk between colleagues to formalized written documents between professionals.

The essence explicit coordination is the active sharing of information. Blakar (1985) proposed five pre-conditions for effective team communication: a) Team members must have shared social reality within which the exchange of messages can take place, including a shared language base and perception; b) Team members must be able to take the perspective of others into account in relation to both their affective and cognitive position; c) Team members must be motivated to communicate; d) There must be "negotiated and endorsed contracts of behaviour" (i.e., agreement among team members about how interactions take place); and finally e) the team must attribute communication difficulties appropriately, so if one of the other preconditions is not being met, the team is able to correctly identify the problem and develop a solution.

From a macrocognitive (Cacciabue & Hollnagel, 1995) perspective, shared meaning among team members is implicitly, rather than explicitly recorded in their messages (Keyton & Beck, 2010). Macrocognition in teams is used to capture team mental processes in collaborative contexts (Fiore et al., 2010; Letsky, Warner, Fiore, Rosen, & Salas, 2007). The term describes how teams move between internalization and externalization of cognition during team interaction (Letsky et al., 2007). Team members are often juggling individual and team goals in conjunction with multiple team goals (e.g., getting the task done effectively and efficiently). Thus, sharing information does not equate to effective communication (Keyton & Beck, 2010). Team communication is prone to misinterpretation leading to error, particularly in multidisciplinary teams. For example, the intended meaning stemming from an individual's internalized processes (e.g. mental model) may not be sufficiently translated to their external behaviour (e.g., verbal communication).

Mental models are one of the most commonly cited forms of implicit coordinating mechanisms (Jones, Ross, Lynam, Perez, & Leitch, 2011). Mental models are organized knowledge structures that allow individuals to predict and explain the behavior of the world around them, to recognize and remember relationships among components of the environment, and to construct expectations for what is likely to occur next (Rasmussen, 1979; Rouse & Morris, 1986). Though mental models are somewhat of a blanketed term, they are in general based on some degree of shared team and/or task specific/related knowledge (Gorman et al., 2004; Salas et al., 2008). This shared knowledge helps team members understand what is going on with the task, and also helps them anticipate what is going to happen next, and which actions team members are likely to take, thus helping them become more coordinated (Mathieu, Heffner, Goodwin, Salas, & Cannon-Bowers, 2000).

2.2.3 Interim Summary

Thus far the review has emphasized the importance of effective teamwork when coordinated, interdependent behavior is required, particularly in safety-critical medical emergency teams. Understanding how teams perform successfully in high-risk settings can provide insight into the process by which effective teamwork is achieved (Xiao, Moss, Mackenzie, Seagull, & Faraj, 2002), potentially leading to better systems designs and training methodologies. However measuring multidisciplinary teamwork in the healthcare environment is a challenge. The interaction mechanisms that stem from the multidisciplinary nature of healthcare teams are complex. The factors that facilitate or constrain teamwork processes, such as information exchange, decision making, and action are influenced not only by the group's composition, but also by knowledge that team members have about one another through previous relationships and interactions. To work effectively together, team members need to monitor each other's actions while engaged in one's own task, know their own and teammate's role responsibilities, and anticipate each other's needs (Baker et al., 2006; Sarcevic, Marsic, Lesk, & Burd, 2008). The internalized cognitive requirements, or coordinating mechanisms discussed earlier, must be transformed into external team behaviours in order to influence team performance. The concepts of shared mental models, team coordination, and team communication are often described as support for effective team behaviours. This family of implicit team functions is also well-represented as a multitude of theories specific to team cognition that are posited to moderate (not just support) team performance.

In the past two decades, team cognition has emerged as an overarching perspective from which to develop theories of team performance. Team cognition (Cooke et al., 2003) takes the view that knowledge representations and transformations on those representations (i.e., cognitive processing) occur not only within an individual mind but also between individuals and between individuals and the environment (Fiore et al., 2010). In the early 1990s, a number a theories pertaining to shared mental models (e.g., Stout, Cannon-Bowers, Salas, & Milanovich, 1999) and shared situation awareness (Endsley & Bolstad, 1994) began to explore the view that people think differently together than they do alone - as captured in the Gestalt motto, "whole is greater than the sum of parts". The main idea being that cognition of teams yields an emergent property that should be measured on a holistic, not summative level. This thesis plans to examine behavioural markers of effective (and ineffective) teams using the coordinating mechanisms of shared mental models and transactive memory as possible underlying theoretical frameworks.

3.1 Mental Models

Mental models are formed when people, through interaction with their environment, develop a subjective internal representation of its systems and of their role in the environment in which they operate (Rouse & Morris, 1985; Sinreich, Gopher, & Ben-Barak, 2005). People can have unique mental models about information that is specific to them, or shared mental models of common situations that include others. A shared mental model is naturally constructed through team cognitive processes, and can be understood as individual mental models amalgamated through team interactions (Gutwin & Greenberg, 1996). Shared mental models allow individuals to work together (Stout et al., 1999). The shared mental model in teams is thought to provide team members with valuable information during team processes that further allows for performance predictability and adaptability (Wiley, Cannon-bowers, & Salas, 2012).

Within the span of knowledge held by individuals within a team, two primary shared mental models are deemed crucial to team performance: the team-related shared mental model (Team-SMM) and the task related shared mental model (Task-SMM; Mathieu et al., 2000). The Team-SMM includes an understanding of team interactions, and teammates' knowledge skills, abilities, beliefs, preferences, and tendencies; the Task-SMM includes knowledge of typical task strategies, procedures and team-environment (Lee, 2007). The shared mental model in teams is thought to provide team members with valuable information during team processes that further allows for performance predictability and adaptability (Paris et al., 2000). Shared mental models help team members coordinate multiple perspectives for the effective solutions to given problems. Teams that have a shared mental model can work more smoothly to respond and adapt to the needs of the team (Gorman, Cooke, & Amazeen, 2010).

Within the contexts of this thesis, a gap in the explanatory power of shared mental models the lack of clarity surrounding the "degree of sharedness". Some authors posit that while shared mental models are desired, not all knowledge possessed by each team member needs to be shared with the other members in order to improve team performance (Cannon-Bowers, Tannenbaum, Salas & Volpe, 1995). A synthesis of the literature (e.g. Cannon-Bowers & Salas, 2001; Cannon-Bowers et al, 1993; Langan-Fox, Anglim & Wilson, 2004; Lim & Klein, 2006) suggests that there is probably not a single mental model that must be shared among team members, but that there are most likely multiple mental models co-existing among team members at a given point in time, especially when carrying out complex tasks. Most theories of mental models focus on the degree of knowledge overlap between team members only as a positive predictor of team performance. They discount the inherent and expected knowledge differences that come with multidisciplinary teams that may serve to enhance the breadth of team knowledge.

Healthcare workers have historically operated in distinct silos and have been trained in separate professions and possess distinct expertise, yet somehow these individual must coordinate to deliver safe care. Physicians must accurately communicate treatment information to the nurse based, in part, on information the nurse presents to the physician regarding the patient's condition. Orders are written on the basis of this discussion and the physician's examination of the patient. These orders are distributed to the pharmacy, X-ray, labs, physical therapy, etc., so that other health care professionals can collect additional information to provide insight regarding the patients or initiate treatment. Each member relies on knowledge of each other's activities, yet the cognition of the team of patient care is clearly distributed among other people and artifacts. Healthcare teams work in complex sociotechnical environments based on the interdependent interactions of humans, technology and cultural norms (Hazlehurst, Gorman, & McMullen, 2008). Individuals know only part of what the team as a whole knows, and the knowledge in teams is distributed unequally among the members, characterizing teams as socially shared distributed cognitive systems (Yoo & Kanawattanachai, 2001). The methods by which teams manage to operate within differentiated and distributed cognitive systems is captured within transactive memory theory, which will be one of the main drivers of this work. Transactive memory was selected as the main theoretical driver, as it encompasses aspects of mental models, coordination mechanisms, and knowledge differentiation which is suitable (but not yet investigated) for the empirical investigation of multidisciplinary teams in complex healthcare environments.

3.2 Transactive Memory

The study of transactive memory is concerned with the prediction of group (and individual) behavior through an understanding of the manner in which groups process and structure information (Wegner, 1987). Transactive memory has two components: 1) an organized store of knowledge contained in the individual memory of group members, and 2) a set of knowledge-relevant transactive processes that occur among group members (Wegner, Erber, & Raymond, 1991). In other words, TM is a combination of individual minds and the communication among them. Individual memories are stored in individual minds, but it possible to know what the other one knows (at least to some degree).

Transactive memory theory views the memory component as an evolving construct, changing as new information is added and old memories are reinterpreted, and as areas of expertise are created or reassigned (Hollingshead & Brandon, 2003). What makes the memory system transactive are the "transactions" or communications among group members used to encode, store, and retrieve in- formation from the individual memory systems (Wegner, 1987, 1995). A transactive memory system (TMS) is a set of individual memory systems in- combination with the communication that takes place between individuals' (Wegner, Giuliano, & Hertel, 1985). Transactive memories stem from team member interactions, a natural osmosis of one another's knowledge through direct (e.g., communication) and indirect (e.g., observation) means. This process simultaneously creates the transactive memories while individuals also develop an implicit TMS of dividing the information processing responsibility in different domains of knowledge among them based on their shared agreement on the knowledge distribution in the group (Michinov, Olivier-Chiron, Rusch, & Chiron, 2008)..

The mere storage of disparate pieces of information by related individuals does not constitute a transactive memory system; there must also be knowledge sharing among individuals. These knowledge-sharing transactions can change the memory structures in the system (Wegner, 1987). Thus, communication is an essential component in the encoding and retrieval of information in transactive memory systems. Direct interpersonal communication is a typical but not required means by which members find out about one another's areas of expertise and develop a transactive memory system. In TMS, communication is the medium for transferring information from one individual to another, determining or assigning responsibility for that type of info based on expertise or first to encounter (Hollingshead, 1998). Each person in the group becomes a specialist in some areas but not others and all come to expect each member to be able to process and access information in specific domains - this reduces the cognitive load of each person, while providing the group access to a larger pool of information across domains (Hollingshead, 1998).

Transactive memory can be viewed as a subset of team mental models, but is not the same construct (Kitaygorodskaya, 2006). Team mental models require that team members hold common or overlapping cognitive representations of task requirements, procedures, and role responsibilities (Smith, 1999). The content of team mental models is abstract theories of situations and actions focused on the task behaviours and the environment (Jones, Ross, Lynam, Perez, & Leitch, 2011; Rouse & Morris, 1985) Transactive memory is more narrowly restricted to know knows what, who is good at what and who does what. Transactive memory concerns itself with knowledge of information held within the group and is not focused primarily outside of the group. Since team mental models focus on certain tasks and how they interact with the environment, it would seem logical that an effective Transactive memory would be critical to an effective team mental model. Transactive memory would store the knowledge which would allow the team to interact with the task environment.

3.2.1 Development of Transactive Memory Systems

TMS development involves three underlying processes: encoding, storage and retrieval (Wegner, 1995). During the encoding stage, group members obtain information about one another and create directories of meta-memory indicating who knows what in the group. New knowledge coming into the group is labeled accordingly and then stored with the group's expert in the domain. Group members can retrieve this knowledge by identifying the label of the required knowledge, associating it with the relevant group expert, and locating this expert in the group. TMSs are said to exist when three conditions are met: specialization, task coordination, and task credibility (Hollingshead & Fraidin, 2003a). Specialization represents the extent to which group members' knowledge is differentiated, coordination is the extent to which the group works well together with few misunderstandings, and credibility refers to the extent to which group members trust others' expertise and are comfortable taking their advice.

The strategy in which the team distributes the responsibility for acquiring and encoding information is part of the TMS. For that system to function effectively, each person needs an accurate representation of its structure to know who to ask. The extent to which people know who has or is expected to have ownership of various information can affect their communication process and the quality of their decisions. It is not just who stores the info but how it is stored, this is determined by group member's specific roles and responsibilities within their TMS. People that are a part of some kind of relationship develop such differentiated TMS via implicit or explicit processes - based on personal expertise, negotiated agreements, or circumstantial knowledge allocation (e.g. first person who encounters it), (Wegner, 1987).

Wegner states that there are two sources of information people use to decide who is to be the acknowledged location of a set of labeled knowledge in the group. Individuals are seen as linked to knowledge on the basis of their personal expertise, or through the circumstantial knowledge responsibility that accrues as a result of how the knowledge has been encountered by the group. Known experts of a domain tend to be held responsible for the encoding, storage, and retrieval of any new information encountered in that domain. This may be the case for professional disciplines in a healthcare team. But what happens when multiple experts are present and likely have overlapping expertise? If a clear candidate is not available, Wegner states that formal groups will make the assignment of responsibility for information domains to individuals on other bases. For example, a work manager asking an employee to become trained in a certain software program; to which all future questions of new employees would be directed at the trainee. In less formal groups, more subtle rules are used to direct knowledge responsibility. For example, the person who initially reports a new piece of knowledge to the group may be held responsible (e.g. who ever received the phone call about the party is charged with coordinating the details). Effective TM should not leave the responsibility for information to chance. If a clear expert does not exist, an explicit or implicit channel for processing that information should be clearly established (Wegner, 1987).

3.2.2 Evolution of Transactive Memory Theory

Wegner's original proposal of the concept of Transactive Memory came from his desire to revive and re-define the relevant literature of social sciences pertaining to the 'group mind' (Wegner, Giuliano, & Hertel, 1985). The group mind was criticized for being superfluous to the study of overlapping individual minds - making the group mind no different from group members. To differentiate himself from the tainted concept of "group mind' Wegner coined the term "transactive memory" in hopes of establishing a more verifiable (and falsifiable) analysis of group mental processes (Wegner, 1987). Since the mid-1980's, Transactive memory theory has evolved from a sociological framework to explain implicit memory structures in dyads and romantic couples (Hollingshead & Fraidin, 2003; Wegner et al., 1991), to projections of large multi-teamed organizations (Nevo, Benbasat, & Wand, 2012). Transactive memory provides a means of understanding how people think together. It describes a "social network of individual minds that surpasses mere uniform agreement because is connects disparate minds" (Wegner, 1987).

3.2.3 TMS in Couples vs. Strangers

The earliest work on transactive memory equates interdependence with intimacy, implying that we think about things in ways together that we would not alone (Wegner et al., 1985). In this respect, transactive memory can be viewed as a means of conceptualizing how people in close relationships may depend on each other for acquiring, remembering, and generating knowledge. Wegner and colleagues first examined transactive memory by introducing new structures for memory organization in intimate couples to impose disorganization to the unique way in which couple holds information (Wegner, Erber, & Raymond, 1991). Individuals working in pairs either with their partner or someone new were exposed to items from various categories of knowledge. Some pairs were instructed to share the memory task through explicit assignment of categories, where others were not given any structure (a