A total of 240 participants constituting 80 teams participated in the study. The teams consisted of two ambulance personnel and one dispatcher in the Bergen EMCC. Bergen is the second largest city in Norway, and Bergen EMCC covers approximately 450,000 people, with about 58,000 ambulance missions a year. Fifty-six percent of the operators reported less than one year of EMCC experience, 40% had between one and 12 years of experience, and 4% reported more than 12 years of experience.

SMMs, trust, performance satisfaction, situational awareness, and mission complexity were measured using a five-item, single-page, paper-based, self-report questionnaire. Each construct was measured using a single item 100 mm visual analog scale (VAS; McCormack et al., 1988). The VAS was constructed as a solid line with anchor points representing opposite descriptors (e.g., very low to very high).

Item details are as follows:

Closed-loop communication and all five team process variables were measured using frequency counts derived by study personnel from audio recordings of the interactions between EMCC operators and ambulance personnel.

According to Norwegian regulations, ethical approval for anonymous non-health-related data or quality improvement studies is not required. However, a formal letter of acceptance confirmed the approval of the study by the EMCC-administration, based on regulations in the Health Personnel Act (Helsepersonelloven) §26. Since a prerequisite for conducting the study was that data collection should not intervene with ongoing activity at the EMCC, several meetings were held with key personnel to develop the short questionnaire and entertain mutual understanding. Verbal and written information was presented to the EMCC-operators with a focus on anonymity and consent. Only written information was presented to the ambulance personnel.

As a standard procedure, all communication, internal in the EMCC and external toward the ambulances, are stored in a voice-recorded database. Thus, all voice recordings included in the present study were based on real-life ambulance missions. The authentic voice recordings consist of complete audio recordings from the emergency call to the end of the mission. Since the study was based on real life emergencies, team composition (e.g., EMCC operator and ambulance workers) were done using standard protocol for handling emergency calls in the EMCC. This includes random assignment to team based on availability of EMCC-operator and ambulance, distance to the emergency and urgency of need for care.

By retrieving the logged verbal communication, the behavioral markers of team processes and CLC could be scored. Based on the unique identification number of the mission, the anonymous data from the questionnaires were combined with an anonymized version of the stored communication (i.e., the names of the EMCC and ambulance personnel, and patient, were deleted). The variables were extracted by manually scoring pre-defined categories of verbal statements (see Table 1) from digital sound-files played on a computer. The scoring was done by three research assistants, trained in this specific methodology.

The questionnaire administered to the EMCC-operators was filled in immediately after completion of a mission.

Reliability and correlational analyses were conducted using SPSS 25. Two-tailed Pearson product-moment correlation coefficients were used in the bi-variate correlational analyses, and intra-class correlations were used to test the inter-rater reliability of the scoring of the coordinating mechanisms and the “Big five” team processes. The intra-class correlations were based on 25 voice recordings scored by all raters. Path analyses were conducted using AMOS 25, testing the a priori hypothesis model show in Figure 1. Separate analyses were performed for each of the three dependent variables of performance satisfaction, mission complexity and situational awareness. In the proposed model SMM, trust, and CLC acted as exogenous variables, while team processes and dependent variables acted as endogenous variables. The effects are based on generalized least square estimates. Ninety percent confidence intervals (90% CI) were determined using 1,000 bootstrapped samples. The proportion of explained variance in variables was calculated using multiple squared correlations. To compare the fit of the proposed model with the observed data, Comparative Fit Indexes (CFI) were computed for each dependent variable. A value close to 0.95 indicates a good fit between the hypothesized model and the observed data, whereas values in the range of 0.90–0.95 are considered acceptable (Hu and Bentler, 1999).

In the present study, 92% of the missions were coordinated by EMCC-operators with a background as ambulance paramedics, whereas 6% of the missions were coordinated by operators with a dual background as a registered nurse and ambulance paramedic (2% missing data).

The measure of “Big five” team processes showed an alpha-value of 0.593. This low alpha value indicated the multi-dimensionality of the measure. This is to be expected since the variable is a composite measure of five different team processes. Three raters scored the communications showing an intra-class correlation ranging from r = 0.726 to r = 0.967 on the different dependent measures. Means and standard deviations are presented in Table 2.

Positive correlations were found between SMM and the dependent measures of performance satisfaction and situational awareness (see Table 2). A significant negative correlation was revealed between SMM and complexity of the situation. The same pattern was found for the mechanism of trust. The only significant correlation involving CLC was a positive association to team processes, and the only inter-correlation between the mechanism was a relation between trust and SMM. These results partially support hypothesis 5.

It is argued that virtual teams (i.e., dispersed team members) operate more autonomously, requiring a higher quality of intra-team teamwork (Johnson et al., 2011). The SMM approach suggests a conceptual model of teamwork characterized by a causal flow from coordinating mechanisms through team processes to performance (Salas et al., 2005). For instance, Mathieu et al. (2000) found that SMM was positively related to team processes and, subsequently through a complete indirect effect, performance in co-located flight simulator teams. However, the prediction of team processes as mediators between mechanisms and performance was not supported in the present study. The results from the path analysis revealed that the EMS-teams’ SMM enhances the evaluation of performance satisfaction and situational awareness, while reducing perceived mission complexity. However, this occurred without the mediating role of the “Big five” team processes. One reason for the findings could be that the transition from face-to-face communication to teamwork in virtual teams actually alters the underlying mechanism, causing successful execution of the team’s mission by increasing the importance of shared cognitions representing vital aspects of the mission. The “Big five” approach describes various processes with behavioral markers that are important for the face-to-face interaction among a co-located team. For a physically dispersed team, there is a need for additional processes to coordinate the team and to establish a shared understanding of task, team and occurring challenges. When team members are separated, the team competencies are reduced, relying more on mechanisms that aims at distributing information equally among team members in order to create sharedness within the team. This need for shared understanding could be fueled by a requirement within the team in order to adapt their behavior to higher order systems. Pre-hospital missions often involve standard medical procedures and a need for integrating these procedures in the medical evacuation chain (i.e., higher order systems). This creates requirements of a shared cognitive representation of the status at the site of the emergency as well as projection in the near future and the requirements are met by the coordinating mechanism of shared mental model. Thus, the need for a shared understanding in order to integrate and coordinate the team’s activity with present and future intentions of the team, guides team interaction based on SMM while verbal markers of team processes are reduced. Our claim is that the increased need for coordinating mechanisms, and especially SMM, could be generalized to other virtual teams. Geographically dispersed team members would perform their professional procedures, but in order to coordinate their activities at present and toward future states, mechanisms that creates sharedness of understanding is crucial. In the absence of visual cues, verbal communication becomes crucial to establish shared mental models.

The main finding of shared mental model being the only predictor of performance lends some support also from studies using a related research approach. Shared mental model is characterized by a shared understanding in the team of tasks, procedures, team members and equipment (Jonker et al., 2011; Sinval et al., 2020). The shared understanding approach shows similar characteristics and refers to the degree to which people concur on the interpretation of significant concepts, when sharing a perspective (mutual agreement) or can act in a coordinated manner (Agredo-Delgado et al., 2020). Agredo-Delgado et al. (2020) investigated the development of shared understanding in teams using computer-supported collaborative work. They found a marked effect on both sharedness within the group as well as performance toward the objectives when focusing on three phases of teamwork. The pre-process phase was composed of activity related to design and specification of objectives and possible disagreements followed by the execution of collaborative activity (process phase). In the post-process phase, a review of the process related to specific objectives was conducted. The design of the Agredo-Delgado et al. (2020) study indicated a mediating role of team processes, Thus, an abundance of empirical data from different research approaches supports the importance of shared cognitive representation of objectives, procedures and team interaction. However, a discrepancy exists of the mediating role of team processes. One reason for the discrepancy could be found in the type of tasks studied. The present study differs from most of the research using the “Big five” approach by studying teams conducting real life critical missions.

Significant aspects when evaluating team performance is the team members’ subjective experiences of mission success. Important factors when rating mission success are the team’s ability to fulfill their objectives when all aspects are considered (i.e., performance satisfaction), gain a correct understanding of the situation (i.e., situational awareness), and obtain clarity regarding the situation (i.e., complexity). SMM being positively related to the evaluation of team situational awareness is also in line with previous research (Macintosh et al., 2009; Saus et al., 2010; Johnsen et al., 2017). Macintosh et al. (2009) found team SA to be related to decision making in face-to-face teamwork in United Kingdom delivery suits. In the present study, this was shown to also be the case for team SA in dispersed medical teams. In a review of teamwork in primary care, Fiscella and McDaniel (2018) emphasized enabling factors for frequent face-to face communication (e.g., space configurations) in order to generate and maintain shared mental models, shared goals, and shared decision making in primary care teams. The negative association between SMM and complexity shows that increased levels of SMM results in a decreased evaluation of the complexity of the situation. This is a notable finding since a perception of reduced situation complexity (e.g., simplicity) may increase teams’ recognition-based decisions (Drury et al., 2012) and prompt the use of standard operating procedures. In a study in the medical domain using the SMM approach, Schraagen (2011) reported complexity as a major risk factor in non-routine pediatric cases. Complexity was related to longer operations as well as more negative patient outcome. However, the study reported that one of the teams studied encountered a more severe case a few weeks after the first, resulting in improved teamwork, higher degrees of attending and employing procedures in a heedful way (Schraagen, 2011). Taken together, the present study expands previous knowledge by showing that SMM is the only team behavior predicting EMS team effectiveness in virtual teams conducting real life missions. Coordinating mechanisms and “The Big-five” team processes.

The results show that the mechanisms of SMM and CLC predict the “Big-five” team processes. This is in line with the predictions of previous models (Salas et al., 2005), where the purpose of the coordinating mechanisms was to support an even distribution of information, fueling the “Big five” team processes resulting in enhanced performance (Uitdewilligen et al., 2018). The effect of team processes on team effectiveness has previously been reported in the medical domain. Studies of team processes within the nursing role (Kalisch et al., 2009), as well as studies of Norwegian trauma teams (Westli et al., 2010; Johnsen et al., 2017), have reported the same predicted effect. However, a common factor in these studies was the co-location of team members involving face-to-face contact. This influence of team process on performance was not supported in our study involving dispersed EMS teams. The direct effect of SMM outperforming an effect mediated by team processes could be caused by an increased need for shared knowledge structures when the team members are located separately.

The present study showed a high correlation between trust and SMM, which is in line with previous findings (Hanna and Richards, 2018). While both SMM and CLC were related to team processes, trust was not. McComb et al. (2017) compared nurses’ and physicians’ level of SMM and trust. They reported a low level of SMM regarding the perceived role responsibilities between nurses and physicians. The two professions showed an equal level of trust toward physicians, while the physicians rated their trust toward nurses lower compared nurses’ evaluation of trust toward their own profession. A seemingly surprising discovery in the present study was the finding of no relation between trust and team processes. Trust could be defined as a belief that the team members will perform expected actions and recognize and protect the interests of their colleagues, as well as a willingness to allow for risk-exposure among members working interdependently (Salas et al., 2005). A possible explanation could be that both CLC and SMM are mechanisms directly involved in generating, maintaining, and altering knowledge structures in the teams, while the impact of trust flows through an indirect emotional component resulting in team members acceptance of being monitored. In addition, for teams consisting of EMCC-operators and paramedics, the function of SMM and CLC are more easily communicated verbally, while signals of trust or lack of trust are harder to communicate and perceive when not being in visual contact. Thus, separating team members geographically alters the mechanisms underlying teamwork and team effectiveness by extinguishing the role of trust. The model fit of the tested model varied from poor (performance satisfaction), to acceptable (mission complexity), to good (situational awareness). This indicates that the model needs to be improved to show more consistency over different indices of team effectiveness.

The result from the present study shows the importance of the coordinating mechanism of SMM. Four types of SMMs have been proposed (Cannon-Bowers et al., 1995; Cannon-Bowers and Salas, 1998). This includes shared cognition of equipment, operational task, interaction as well as of the team members. Based on the SMM approach, cross-training is suggested to present an important method to increase shared team knowledge structures (McCann et al., 2000; Marks et al., 2002). Cross training refers to a rotation of team roles following training in one’s own role requirements. This provides an opportunity to experience new learning for all team members and facilitate information sharing that will extend and maintain SMMs. Superior training effects have been found in teams characterized by high levels of SMMs. Espevik et al. (2011b) reported significantly higher levels of performance after only one training session in naval cadet teams exposed to cross-training. Implications of the present study point not only toward the type of training strategy (e.g., cross training), but also toward the training objectives (types of SMM). Thus, in targeting future training it will be paramount to define the type of SMM in need of training (“What”) and the training strategy (“How”) to increase shared cognitions in EMS teams.

Some limitations should be noted. The present study relies on single item measures on several of the variables. From a historical standpoint, reliability and validity issues have discouraged the use of single item measures of psychological constructs. Thus, multiple-item scales are favored (Nunnaly, 1967). However, this view has been challenged and there are several reports supporting the use of single-item variables (Wanous and Reichers, 1996; Wanous et al., 1997). This is founded on empirical data showing high test -retest reliability (Gardner et al., 1998), as well as well as high correlations with multiple item scales (Littman et al., 2006). The validity is also revealed by single item measures effectively predicting outcomes (Littman et al., 2006), including job satisfaction (Wanous et al., 1997; Nagy, 2002). Although there are limitations to single-item measures, potential advantages should also be noted. Advantages like cost-efficiency, greater face validity, and a possible increased willingness of respondents to take time to complete the questionnaire instigated by a less intrusive method compared to the use of multi-item scales.

The item measuring situation awareness is not totally in line with Endsley’s model of situational awareness (Endsley, 1995). The model describes situational awareness as being constituted by three hierarchical organized levels. These levels include detection of critical signals (level 1), understanding of the situation (level 2) and prediction into the near future (level 3). However, the question is meant to capture the core element of the concept by tapping into the individual’s cognitive representation and understanding of the emergency at hand. It could be argued that that this includes level 1 and level 2 in Endsley’s (1995) model.

Both the dependent variables and two of the coordinating mechanisms were evaluated by the EMCC-operators only. This was caused by the study’s intention to not intrude on the ongoing activities. Logistical, methodological, and ethical constraints prevented recordings of data from the ambulance personnel or regarding outcome for the patients. However, the EMCC-operator is a significant team member. By acting as a communication hub, initiating and coordinating the mission, the operators could be viewed as team leaders, and leaders’ evaluation of team effectiveness is imperative.

Another limitation is the lack of control concerning the team members’ identification with the unit studied. The results could be influenced by the participants defining their team membership differently. The ambulance crews could view themselves as part of one team and the EMCC-operators as another team, resulting in measuring inter-team coordination rather than teamwork. Despite this, the units that were studied match the formal definition of teams, justifying the approach of the current study. Finally, nations vary in the organization of their pre-hospital services. Education and training levels, the size of organizations, and the amount and type of operations may differ, limiting the generalizability of the results.