This study assessed coordination behaviours during trauma teamwork IRL and during simulation. It showed a similar distribution of utterances related to task and information management in both domains, with confirmations being the most common coordination behaviour. Detailed knowledge about how these domains compare adds information that can be used to present a validity argument for trauma communication research in simulation. This is valuable since conclusions about behaviour in the IRL are drawn upon behaviour in the simulated environment. Further, coordinating behaviours and communication types have been classified in teamwork from different domains of healthcare and other organizations. These have been confirmed and validated in emergency teams [10, 25] and trauma teams [2, 11, 14]. The present study inquired a more detailed description about the relevance in this specific domain. This might be informative to simulation based training, as specific patterns of communication can be linked to the specificities of the trauma environment.

The trauma team is often described as performing task-oriented teamwork, and the pace is high. There are working manuals and a specific order of patient examination that structure the work performed. Although task-focused in character, these teams’ communication is largely represented by information processing (65%). The “push of information” is considered positive for team performance [14, 22, 32], and this type of coordination behaviour is called “PIWR” in the CoMeT-E system. It functions to update team members about particular information that is, by the sender, believed important for the receiver without carrying out specific guidance, instructions, or questions.

We found, however, that explicit information management through “request information” and “confirmation” constituted larger proportions of the information management activity, each comprising 20% of the information related communication, whereas the implicit information related behaviour “PIWR” amounted to just below 10%.

One important difference between the IRL and simulation domains was that “give information after request” was significantly more prevalent during simulation. This is most reasonably the result of instructor delivery of information upon request from the team members. Although in theory, this might reduce the need for “PIWR” in simulation (as team members get information updates through such common information exchange), we did not find any significant reduction in “PIWR”.

Overall, the proportions of task management coordination behaviours were similar between the two domains. The two most important task management coordination types were “task distribution”, accounting for as much as 43%, and “confirmation”. Explicit task distribution is generally considered positive for accomplishing tasks in emergency teams [5, 7, 19]. In the present study, earlier described online commentaries, which are comments about own activities and task executions, were probably comprised of both “PIWR” and “initiate action”, depending on the exact spell-out. Such comments are often more implicit and aim to make others aware of steps taken or planned in a process that could possibly induce back-up behaviours [10, 16]. Task-related communication in the form of task distribution and instruction is more beneficial to performance in heterogeneous teams, which, to a greater extent, need explicit instructions. An experienced and homogeneous team might benefit from reducing task-related communication, as overt communication slows down task execution and makes less room for information processing and decision-making processes [16, 25]. Technical skills training with a honing of tasks is a pillar of trauma team training [6], and even though such training is technical in nature, it could supposedly affect communication in a positive way by reducing explicit task-related communication.

The main purpose of TTR communication is to deliver important information updates with the aim of aligning mental models within the team. In earlier studies, this has been conceptualized as either TTR, “push of information”, or PIWR [14, 15, 32]. We considered all utterances directed to the team in this study and found that overall in both domains, around 20% of the information-related TTR communication was “PIWR”, thus representing the first level of information sharing to the team (fact sharing) [30]. Such communication has been linked to improved performance in clinical diagnostic processes [29]. However, such an outcome could not be demonstrated in emergency resuscitation scenarios [25, 29].

“Review process” and “general situation assessment”, representing the provision of interpretation of facts and possibly projection sharing, were less frequent in this study, both IRL and during simulation. These types of communication have been associated with the alignment of mental models. Fact sharing might result in the oversharing of basic facts that are already known to the other team members, and thereby carries the risk of contributing to information overload (Sohrab et al. 2015). This tendency might, to a great extent, result from the trauma team handling algorithm-like processes in which background assumptions and representations are alike. Therefore, the sharing of basic facts is not merely a sharing of facts but a triggering of similar mental processes across individuals. An example of this is the subtle expression of concern by repeating a fact about a patient, such as “the systolic blood pressure is at 80”, with the implicit intention of pointing at the need to address this concern in action [3]. Such commentaries serve to decrease potential attention narrowing in other team members, and are fundamental to safe teamwork [4]. With a similar frame of reference, the need to share interpretation and mental projections represented by the “review process” and “general situation assessment” is reduced. Future studies could test the hypothesis that team mental models can be updated and corrected through the use of PIWR in algorithm-like processes, whereas “review process” and “general situation assessment” are required in knowledge-related processes.

Information related TTR were overall more explicit than expected, with “confirmation” being an important coordination behaviour in both domains. However, “request of information” was one of the most important information related TTR in simulation, but not used to the same extent IRL. Overall, the findings imply that besides providing a push of information and assessments, a central communicator also addresses information gaps and acknowledges others. Both of these coordination behaviours represent TTR that is more explicit in nature than earlier conceptualizations of information-related TTR.

Task management TTR in both domains mainly regarded “task distribution” and the “initiation of actions”. The initiation of actions delivered to the team probably functions as the previously described concept of online commentaries—that is, to update the team on steps taken, which stimulates implicit actions and back-up behaviour [10]. However, regular task distributions dominated the “initiation of actions” regarding TTR, which underlines the need for a central speaker to explicitly instruct, delegate, and coordinate task executions in a trauma team and underlines team member interdependence.

CLC is considered an important communication tool that is used in fast-paced team practice to acknowledge correct information transfer and the execution of an instruction. In our conceptualization of CLC, we excluded question–answer communication as well as possible call-outs in which no acknowledgement happened or the reception part consisted of an acknowledgement (confirmation) without a check-back of the call-out information. We included check-backs and acknowledgements occurring subsequent to the execution of tasks as they fit the CLC structure. We looked for the use of CLC and detected it as any part of CLC that ultimately required at least two steps of the original description. Surprisingly, we found a small proportion of the utterances was associated with CLC, and significantly less used IRL. The proportional use of CLC is not easily compared to previous studies, since in most cases, CLC has been counted and reported as a number per team, or the frequency of completed check-backs to observed call-outs, without relating it to the bulk of communication.

In this study, the CoMeT-E system was modified according to the material and the research questions. One of the two added codes for coordination behaviour was “confirmation”, which proved to be one of the most prevalent ones. Sometimes, the confirmation was part of a CLC, such as a “check-back”, but in most cases, it was provided as “yes” or “OK” and was not associated with CLC. This leads us to the conclusion that information and task management are in many instances acknowledged in ways other than through check-backs, perhaps at the expense of repetition and thereby precision but with the possibility of reducing communication load.

In a sociolinguistic study of obstetric teams, the best-performing teams oriented towards CLC but with shorter linguistic structures compared to the textbook variant [1]. Some scholars have argued that the ritualization of speech ignores the dynamics of interaction and the multiple ways in which a message is best delivered [1], and the findings from our study support the suspicion that real-life CLC might differ from textbook CLC and that firm ritualization might increase communicative load at the expense of efficiency [25].

Further, we expected the use of CLC would be coupled to task management, but in this study, CLC was a greater part of information management communication in the IRL domain, and task management in the simulated domain. It is reasonable to believe that more tasks and administrations were executed in the simulation domain, since the simulations were created to induce rapid decision-making and resuscitation task coordination. In a study assessing CLC use in IRL trauma teams, the use of CLC was more frequent in the most serious cases [2]. The use of check-backs also seems to be more frequent in response to medication orders than intravenous fluid administration orders or task delegations [2, 9]. In our study, besides the task-related CLC, the overall task-related communication was similar to IRL, but there was also significantly higher CLC use regarding information management in simulation. This suggests the presence of at least some level of a Hawthorne effect, i.e. the knowledge of being watched and probably assessed on behalf of communication increases the use of “desired communication” [20]. The high frequency of confirmations and clarifications unrelated to CLC suggests that for many call-outs other ways to confirm the sender message are used.

This study is strengthened by the comprehensiveness of the description of the communication used, and the use of both the IRL and the in-situ simulated environment.

The weaknesses include the use of a single center material, which might limit the generalizability, and although rich in verbal communicative events, a limited number of studied cases. Although all teams were composed from the daily schedules, we have not assessed possible interpersonal relationships in the different teams, or the possible impact of different contextual conditions. Treating qualitative data, such as communication, quantitatively will not be completely explanatory, as communication is dynamic, and situationally coupled. Although the quantifications provide some information, they do not reveal other important aspects that are relevant for the understanding of communication, and which can be addressed through pure qualitative inquiries that carry the possibility to penetrate deep into a smaller sub-set of situations. Examples of this are particularities of the case, the training and experiences of the team members, and other situational factors applied to the timing and setting of each trauma resuscitation. Further, this study did not address the quality of clinical practice in the trauma teams including the relationship between communication and medical action undertaken. The quantifications and comparisons of verbal utterances collected from different trauma cases, handled by different teams might bias the results, and a comparison simply based on the IRL or simulation domains might obscure other important situational factors. However, e.g. Gundrosen et al. [10], qualitatively studied team talk in simulation, and opened for the possibility that the patterns seen in simulation might be specifically shaped by the use of a dummy and an instructor providing the “missing cues”. This might affect team talk patterns, based on the environment regardless of particularities of the case. From our results we have shed further light on this important educational issue.

A main methodological consideration regards our use of the CoMeT-E system for the classification of coordination behaviours. The acquisition of a comprehensive coding system was important for a truthful analysis of our material, as we aimed to quantitatively describe and compare the use of speech in different domains. The CoMeT-E system was modified to fit the material and the research aim, and TTR was operationalized as all verbal communication directed to the team. Thus, the coding system was complemented with a specific code and category relating to relational talk (e.g. greetings and salutations). Further, confirmations were added as a separate coordination behaviour to achieve an exhaustive coding scheme. CLC was originally considered a coordination behaviour distinct from the categories of information and task management. A strength is our separate coding of CLC to explain its association with task and information management and its association with specific coordination behaviours. As we did not include call-outs that were never responded to, no data about “missed” confirmations or check-backs was collected.

Substantial work was executed to frame the codes according to their meaning, thereby increasing strict and coherent coding. In this study, the coordination behaviours “request team member information”, “give team member information”, and “evaluate decision” were used at a minimum, which disqualified any conclusions drawn from those observations. The original CoMeT-E system would benefit from reducing alternative coding of behaviours associated with decision-making from three to one to achieve a useful instrument. For many purposes, it would also be serviceable to coalesce codes in which retrieval and giving of information and team information are separated into “information retrieval” and “team member information retrieval”.

We coded all utterances except for patient utterances, which were left out for ethical reasons. This might have somewhat skewed the IRL utterances. It is not typically for patients to talk a lot during resuscitation, but a less injured patient can be interviewed and provide information to the team.