Trauma resuscitation: can team behaviours in the prearrival period predict resuscitation performance?

Abstract

Background

Optimising team performance is critical in paediatric trauma resuscitation. Previous studies in aviation and surgery link performance to behaviours in the prearrival period.

Objective

To determine if patterns of human behaviour in the prearrival period of a simulated trauma resuscitation is predictive of resuscitation performance.

Design

Twelve volunteer trauma teams performed in four simulation scenarios in a paediatric hospital. The scenarios were video recorded, transcribed and analysed in 10-second intervals. Variation in the amount of utterances per team member in the prearrival period was compared with team performance and implicit coordination during the resuscitation.

Key results

Coders analysed 18 962 s of video. They coded 5204 team member utterances into one of eight communication behaviour categories. Inter-rater reliability was excellent (an average of 83.1% across all four scenarios). The average number of communications occurring during the prearrival period was 18.84 utterances, with a range of 2–42 and a SD of 9.55. The average length of this period was almost 2 minutes (mean =117.30 s, SD=39.20). Lower variance in team member communication during the prearrival better was associated with better implicit coordination (p=0.011) but not team performance (p=0.054) during the resuscitation.

Conclusion

Patterns of communication in the prearrival trauma resuscitation period predicted implicit coordination and a trend towards significance for team performance which suggests further studies in such patterns are warranted.

Introduction

Teamwork is recognised as an essential part of optimal performance in healthcare settings,^{1 2} but is particularly challenging during paediatric trauma resuscitation. A successful paediatric trauma resuscitation requires the ad hoc formation of a team with members who may have limited previous interactions but who now need to coordinate their activities to diagnose and treat a trauma patent's most life-threatening injuries. Patients require a complete assessment within minutes, with interruptions necessary to implement life-saving treatments.³ Trauma teams must make rapid decisions that will often have immediate and significant consequences. They make these decisions without having detailed knowledge of the patient's injuries or medical history.⁴ These decisions can include prioritising chest tube insertion before abdominal exploration in a haemodynamically unstable patient with a pneumothorax and an undiagnosed aortic tear.

A successful trauma resuscitation requires extreme coordination to implement numerous and interrelated tasks simultaneously under significant time pressure. Outcomes may depend on communication and coordination in addition to the technical proficiency of team members. A team's failure to translate individual knowledge and skills into effective team activity constitutes a major obstacle to achieving improved patient outcomes.^{5 6} Many studies highlight the failure of providers to accomplish this.^7–10 Essential in optimising these types of teams is coordination.¹¹ Coordination can be either explicit or implicit. Implicit coordination occurs when team members anticipate the actions and needs of their colleagues and dynamically adjust their own behaviour, without directly communicating or planning the activity.¹² Although explicit and directive coordination is often cited as critical to good leadership and teamwork,^{2 13} research in other domains has linked higher implicit coordination with greater overall team effectiveness,¹⁴ and specifically found that teams who proactively communicated information had improved performance.¹⁵ Directive leader behaviour may improve team effectiveness initially, but teams characterised more by empowering leader behaviour outperform other teams as tasks unfold over time.¹⁶

This current work focuses on communication during the preparation period before task performance begins and is uniquely different that previous research which focuses on behavioural patterns during task performance.^{17 18} In trauma resuscitation, this preparation period is defined as the interval between the initial notification of the team and the patient's arrival in the trauma bay, which is a designated room in the emergency department for trauma patients. This period is used to prepare personnel and equipment for the subsequent resuscitation. An intervention with checklists during a comparable time period in surgery has been shown to improve mortality.¹⁹

Existing research in non-healthcare domains indicates that high levels of reciprocity during this preparatory period—that is, interaction characterised by balanced, two-way exchanges—is significantly related to higher subsequent team effectiveness during task performance.^{20 21} More generally, research on team dynamics indicates that ‘setting the tone’ for a new team is vitally important, as team norms and routines of interaction emerge very quickly.^{22 23} Should these emergent properties serve to inhibit reciprocity by squelching team member voice (ie, constructive challenges to the status quo with the intent of improving the situation,²⁴ or lowering psychological safety (ie, team member perceptions of feeling safe to share ideas, information and opinions without fearing blame),²⁵ subsequent team effectiveness may be hampered as well—and especially given a dynamic work environment requiring high levels of team information sharing and coordination.

In this paper, we examine whether communication patterns among team members during the prearrival period are predictive of implicit coordination and team performance during subsequent trauma resuscitation. We hypothesised that a more ‘balanced’ pattern of communication among team members in the prearrival period will correlate with better team performance and implicit coordination. In this context, ‘balance’ refers to more people ‘sharing the floor’ versus one or two team members doing most or all of the speaking.²⁶ Previous research indicates that when team members participate in a more open information exchange before task initiation,²⁷ and particularly in face-to-face settings,²⁸ the more shared their mental representations of roles and tasks become. We expected this shared cognition would translate into (1) greater implicit coordination during trauma resuscitation and (2) improved team performance.

Methods

Participants

Twelve emergency room trauma teams working at a Mid-Atlantic children's hospital volunteered to participate in a simulation study designed to evaluate the impact of a checklist on trauma resuscitation task performance. This current study is a secondary analysis of the simulations performed for this study.²⁹ The IRB approved the previous study and secondary analyses such as this one were also included in that IRB. Participants consented to video recording and analysis. Each team participated in four scenarios, resulting in a total of 48 videos. One video recording was damaged, resulting in a total of 47 usable videos.

Each nine-member team was composed of a lead physician (the team leader who was either an attending emergency room physician or a paediatric surgeon), a surgical resident, an airway physician (who was either an anaesthesiologist, or a critical care or emergency room fellow), a bedside nurse and a recording nurse and three confederates (a second bedside nurse, a respiratory therapist and a medication nurse). The confederates served in the roles of team members that are a standard part of the trauma resuscitation team and followed instructions given by other team members. Confederates in the respiratory therapy and the nurse right roles provided the trauma teams cues that were necessary for the progression of the scenarios such as stating asymmetry in chest rise which is not easily replicated on a simulation manikin. They did not provide suggestions or guidance to the teams. Demographic data on the participants, including years of experience, age and gender were not collected.

Simulations

The simulations were conducted in situ in the trauma bay. The 12 teams participated in four scenarios.²⁹ In each scenario, the teams had to identify and manage two potentially life-threatening injuries, including hypotension, pneumothorax, seizure and profound hypothermia. High-fidelity manikins were moulaged and programmed to represent the features of each scenario. Scenarios began with a scripted text page notification of the patient. Teams had time to prepare the room and themselves for the patient. A confederate emergency medical services (EMS) provider then brought in the manikin on a stretcher and gave a scripted report about the patient.

Outcome measurements

Following other work examining teams in high-reliability contexts,³⁰ implicit coordination was measured using the ‘information anticipation ratio.’¹⁴ This ratio represents the number of information transfers divided by information requests. Higher ratios show that individuals are anticipating the information that other team members need instead of explicitly having to be asked to provide this information. The information anticipation ratio in this study was derived from coding of team members' communications and represented the ratio of provided information regarding an action or patient divided by information requests. The sentence, ‘Blood pressure is 60 over 36’ is an example of provided information, while the question ‘Do you hear breath sounds?’ is an example of a request for information (table 1). To reflect the importance of addressing both life-threatening conditions, we measured team performance using the time from patient arrival to completion of the second life-saving intervention corresponding to the second life-threatening injury.

Table 1.

Coded team behaviours

Behaviour	Definition	Example
Information request	Request for information or verification	‘Do you hear breath sounds?’
Information request response	Response to a request for information	‘No breath sounds’
Opinion/advice request	Request for opinion or advice (ie, the individual needs information on what to do and/or how to do it)	‘Should I give more fluids?’
Opinion/advice request response	Response to request for opinion or advice (solicited or unsolicited)	‘Yes, give another 800 cc bolus’
Permission request	Request for permission to complete an action (unlike opinion/advice request, individual knows how to complete the action but needs or desires authorisation to do so)	‘May I start the secondary survey?
Provide action/patient information	Provision of unsolicited information about an action or patient information (including summaries and updates)	‘Blood pressure is 60 over 36’
Direct command	Command that is given when an actor clearly conveys what needs to be done and who needs to do it	‘Reassess breath sounds for us, (name of team member)’
Indirect command	Command given when there is ambiguity in what needs to be done and/or who needs to do it	‘We need to put the c-collar back on’

Balance in communication

Balance in communication among team members was measured using the variance in the total number of communications (ie, utterances) made by team members during the prearrival period. Lower variances represent greater balance and higher variances represent less balance in communication. As an example, if the head physician made multiple utterances and none of the other eight team members spoke, the variance would be high. Conversely, if all nine team members each made an equal number of utterances, the variance would be low.

Classification of team member communication

The simulation scenarios were recorded using a digital video camera. A physician, familiar with the scenarios and medical terminology, created verbatim transcripts of the videos. Two members of the research team who were blinded to our hypotheses coded team member utterances (or statements that represent complete thought units) in 10-second intervals while referring to the written transcripts for clarity. The coding scheme was adapted from a protocol that had been used in previous research on communication patterns in a high-reliability context.²⁰ Previous studies of dyadic or team interaction have used coding intervals of 10 s.^{31 32} Coders watched or listened to recordings in predefined 10-second increments, noting either the binary occurrence or frequency of target behaviours. Although, this technique increased the granularity of the coding, it also increased the ease and speed at which coders completed and compared their work. This interval is long enough to capture one complete instance of any of our target behaviours but not so long as to likely capture more than one instance.³³ Coders recorded the utterance and which team member spoke. Coders did not include the same code for a team member more than once in a given interval and assigned only one code per transcripted utterance. In addition, when an utterance began in one 10-second interval and ended after the start of the next interval, the utterance was only recorded for the interval in which it began.

Inter-rater reliability

The two coders randomly selected and coded three of the 12 videos in each of the four simulation scenarios to assess inter-rater agreement. Initial coding disagreements were discussed, and videos were recoded until an acceptable level of inter-rater agreement (an average of 83.1% across all four scenarios) was achieved on these videos.³⁴ Each coder then independently coded half of the remaining videos.

Statistical analysis

Multilevel regression (ie, hierarchical linear modelling) was used to account for statistical non-independence given the repeated measures nature of the data (ie, teams responding to four scenarios). Multilevel regression can control for clustering and heteroscedasticity and provides efficient estimates of parameters. We first conducted a series of log-likelihood ratio tests to examine whether any of the following variables related to either of two outcome measures: the scenario to which the team was responding, the order of the scenarios, and the checklist study condition (ie, checklist or control). As longer prearrival periods could be related to more total communication, we controlled for the length of each period. We then proceeded with the tests of the hypotheses using multilevel regression to examine the relationship between variance in team member communication before patient arrival and the two outcomes of interest, namely: (1) the information anticipation ratio, and (2) the time until the team completed the second intervention. When testing each hypothesis, we examined the same model for each of the two measures. In these models, intercepts were allowed to vary, but slopes were not allowed to vary for model parsimony. A series of tests of model fit revealed that including random effects for the slopes did not significantly improve model fit. Thus, we retained the more parsimonious model only including the fixed effects for the slope. The primary predictor of interest was communication variance. Since the hypotheses were directional, we used one-tailed significance tests.

Results

The two coders analysed 18 962 s grouped into 10 s intervals. They coded 5204 team member utterances into one of the eight categories (table 1). Given the scarcity of studies examining communication occurring before patient arrival, we first present some relevant descriptive findings (table 2). An average of 18.84±9.5 utterances, (range of 2–42) occurred during the prearrival period. The average length of the prearrival phase period was 117.3 s (range: 4–200) (SD=39.20 s). On average, there were 9.55 (SD=3.62) communication utterances made per minute during this period. In comparison, there were 17.19 (SD=4.83) communication utterances made per minute after the patient arrived. The lead physician spoke more than half of all coded communication utterances during the prearrival period (mean =10.11, SD=5.57).

Table 2.

Descriptive statistics for pertinent study variables

Variable	N	Total	Mean	SD	Range
Prearrival period
Time (seconds)	46	5396	117.3	39.20	4–200
Total utterances	44	829	18.84	9.55	2–42
Lead physician	44	445	10.11	5.57	0–23
Surgical resident	44	36	0.82	1.88	0–11
Recording Nurse	44	20	0.45	1.02	0–4
Anaesthesiologist	44	65	1.48	1.78	0–6
Other*	44	180	4.09	2.50	0–14
Resuscitation period
Time (seconds)	45	14 915	331.44	118.21	142–657
Utterances	47	4375	93.09	33.00	36–178

*Other category included as airway physician, bedside nurse and confederates.

Prior to testing the hypotheses, we first conducted initial analyses to examine whether any of the following variables related to either of two outcome measures: the scenario to which the team was responding, the order of the scenarios, and the checklist study condition (ie, checklist or control). Log-likelihood ratio tests revealed that the scenario significantly predicted time until the team completed the second intervention, χ²Δ(3)=19.28, p<0.001, but did not predict information anticipation ratio, χ²Δ(3)=2.25, p>0.500. Thus, we controlled for scenario in the primary analyses. To be consistent, we included it as a control variable for both outcomes. Order of scenarios did not predict time until the second intervention, χ²Δ(1)=0.07, p>0.500, or the anticipation ratio, χ²Δ (1)=2.33, p=0.233. Similarly, the checklist condition did not predict time until the second intervention, χ²Δ(1)=0.43, p>0.500, or information anticipation ratio, χ²Δ(1)=0.43, p>0.500. Therefore, we did not include these two variables as covariates in the final models. Lower variance in team member communication (greater balance) was associated with a higher anticipation ratio (p=0.011) (table 3), our indicator of more implicit coordination during the resuscitation.

Table 3.

Anticipation ratio as a function of variance in team communication

Fixed Effect	Coefficient	SE	t-Ratio
Intercept, γ00	1.87	0.57	3.30*
Length of prearrival period, γ01	0.00	0.00	0.02
Variance in team communication, γ02	−0.02	0.01	−2.71*
Random effect	Variance	χ2	df
Intercept, U0	0.02	15.077	11
Level-1 residual	0.67
Deviance		108.39
Number of estimated parameters

Three effect-coded variables were included in the model to control for the scenarios to which the team was responding. For clarity of presentation, and to emphasise the results of primary interest, the results for those variables are not presented in the table.

*p<0.05.

Lower variance (ie, more balance) in team member communication was not significantly associated to team performance (p=0.054) (table 4).

Table 4.

Time until second intervention as a function of variance in team communication

Fixed effect	Coefficient	SE	t-Ratio
Intercept, γ00	390.09	56.67	6.88**
Length of prearrival period, γ01	−0.49	0.46	−1.05
Variance in team communication, γ02	1.25	0 0.76	1.64*
Random effect	Variance	χ2	df
Intercept, U0	441.46	16.165	11
Level-1 residual	6329.72
Deviance		499.24
Number of estimated parameters		8

Three effect-coded variables were included in the model to control for the scenarios to which the team was responding. For clarity of presentation, and to emphasise the results of primary interest, the results for those variables are not presented in the table.

*p=0.054, **p<0.001.

Discussion

This study indicates that teams with more balanced communication during the prearrival period have more efficient communication (ie, greater implicit coordination) after patient arrival. To the best of our knowledge, this study represents the first investigation examining prearrival communication as a predictor of subsequent coordination in the healthcare domain. These results also raise several other questions for future studies. Was the prearrival communication reflective of teams who simply knew each other better or at least had an optimal level of psychological safety? What was the actual reason communication was more balanced?

We used the anticipation ratio as a surrogate for implicit coordination. The anticipation ratio has been previously described¹⁴ and found to be reflective of implicit coordination. The more team members who are proactively providing information negating the need for information requests could be reflective of a team that already knows what information is needed to move the resuscitation forward and therefore offer it. In contrast, teams that are attempting to coordinate more explicitly may try to accomplish this goal by repeated or unanswered questions.

Determining the antecedents of these different patterns of prearrival communication has important practical training implications. More balanced communication during this period may reflect interpersonal comfort and reciprocity among team members, perhaps resulting from greater familiarity working with each other. Consistent with this idea, healthcare teams composed of members who are more familiar with each other tend to outperform newly formed teams.³⁵ Other factors that may impact these prearrival patterns include the distribution of expertise among team members, individual personalities, and the members' experiences immediately before the trauma resuscitation (eg, impacting their mood and cognitive focus during the prearrival period) and would be worthy of future studies.

Additional exploratory analyses

Further analyses also revealed that a greater number of utterances by the head physician during the perarrival period was associated with a lower anticipation ratio and more time until the second intervention (controlling for the length of the prearrival period and the total amount of communication; both p<0.05). An additional analysis indicated that the number of utterances made by the head physician was strongly correlated with the variance in the total utterances (again controlling for the length of the prearrival period and the total amount of communication; p<0.05), suggesting that these two variables both represent ways to conceptualise ‘balance’ in team communication.

These results may explain the primary findings regarding variance and warrant further investigation. The findings could be interpreted in several ways. One possibility is that the team leader intimidated the team, resulting in less psychological safety and team members feeling uncomfortable asking questions or offering information or suggestions. Another possibility is that the team leader sensed a need to take on a more vocal and dominant role—perhaps inferring that the team was inexperienced or that team members would be hesitant to offer information and their perspectives. The role of silence in team communication during surgery is complicated, ranging in interpretation from acquiescence to anger.³⁶ Teams with leaders who encouraged members to speak up were more successful at implementing new cardiac surgery technology.³⁷

Limitations

Although the anticipation ratio has precedent as a valuable indicator of team communication effectiveness, the use of additional team process metrics in subsequent investigations is needed. For instance, the effects of the prearrival communication may impact, and operate through variables such as psychological safety, team cohesion and team member familiarity or liking which were not measured here. We chose time to second intervention for our performance measure given that the clinical conditions during trauma resuscitation often are difficult to diagnose and because the team frequently must treat multiple life-threatening injuries in parallel instead of sequentially. Using the time until the final intervention as our measure of team performance was intended to capture the importance of time. This performance measure is controversial with some teamwork literature advocating that performance measures should include the assessment of non-technical skills such as number of leadership utterances and numbers of completed closed loop communications. Since we were studying the impact on these nontechnical skills on performance, we did not include these skills in the performance measure. In addition, the original study of which this is a secondary analysis actually used adherence to the ATLS protocol since the intention of the checklist was to improve ATLS adherence. Knowing that the checklist did impact ATLS adherence, we purposely avoided that as a performance measure. The simulation scenarios were designed for the ‘patient manikin’ to have two life-threatening injuries that the team needed to diagnose and treat. Since time is an objective measure to assess a team's ability to do that, we used it as our primary performance measure. This type of performance measure limits the usefulness of our results. Certainly, teams who accomplished diagnosis and treatment could be just lucky and fast instead of reliably expert teams.

This is a secondary analysis of a simulation study. Additional limitations imposed include lack of certain demographic data of the participant group that could have impacted our finding such as experience level of the group. In addition, we weren't able to discern why a more balanced form of communication led to a higher ratio of responses to requests for information. Content of the communication could not be analysed for these purposes since content would seem to matter. In addition, since this was a simulation study, no impact on real patient outcomes could be assessed. The association of communication patterns during the prebriefing and team performance only approached significance and further studies need to be performed to examine whether there is any association at all.

As medicine has incorporated team training concepts, appreciation for the importance of concepts like coordination has grown. Our current study highlights that advocating for explicit coordination may be inappropriate for expert teams who do (or should do) much of their coordination implicitly. Learning how to promote implicit coordination—and, in turn, realise the improvements in patient outcomes that may follow seems essential. Findings from the current study suggests that altering the interpersonal dynamics occurring before the patient even arrives may be one way to achieve this.