A Qualitative Analysis of Observed Behavior of Pediatric Rapid Response Team Performance

This study was a qualitative, in situ, observational study of pediatric RRT activations. The study occurred at a 313-bed, freestanding, urban academic children’s hospital that serves as a teaching hospital and is the area’s level 1 trauma and burn center. This institution averaged 60 RRT activations per month during the study period. The RRT consists of providers from the PICU, including a critical care fellow or nurse practitioner, an ICU charge nurse, and an ICU respiratory therapist. The RRT members assemble at the patient’s bedside to join the patient’s primary acute care floor team, which consists of a bedside nurse, a floor charge nurse, a primary team physician or advanced practice provider, and a floor respiratory therapist. This work was part of a larger study of team leader training during which team performance was assessed.¹⁵ 

The research assistant attended a convenience sample of daytime RRT activations and collected field notes in situ during observations. This research assistant was familiar with basics of the RRT intent but was intentionally unfamiliar with other services and teams in the hospital, the hospital system at large, and specific roles of non-RRT personnel. This lack of familiarity was important to limit bias or presumptions about people’s roles or expectations. As part of a larger study, the research assistant was collecting data regarding team leader behavior and team performance and was asked to take field notes of what he observed even outside leader and team member behavior. Although the field notes could be about leader and team member behavior, we also asked him to record defining events during the RRT activation and answer “What made this RRT memorable?” Within 15 minutes of completion of the RRT activation, the research assistant organized notes taken during the event and wrote down initial observations and thoughts on the basis of these questions. The research assistant was supervised by a research coordinator with a direct role on our hospital’s RRT committee and experience as a qualitative researcher; the research coordinator and research assistant attended 10 RRT activations before the initiation of the formal data collection to test the data collection instrument and to reconcile observations and clarify procedural questions the assistant had. Thereafter, the research assistant was able to ask clarifying questions of the coordinator after data collection started.

Field notes of observed RRTs were analyzed by using conventional qualitative content analysis.¹⁶  Notes were organized and analyzed by using Dedoose software, version 6.1.18 (Los Angeles, CA). The text of the field notes was abstracted into excerpts or “units of analysis,” and codes were assigned. In some instances, the units of analysis included findings that received >1 code. Once all notes were coded, definitions were created for each code. Informational redundancy was reached, meaning no new codes could be induced from the units of analysis and each code had adequate findings to support their definitions.^17,18  One author (A.B.L.) analyzed all notes, identified units of analysis, and created the codes and definitions. This author is a clinician with previous training and experience in qualitative analysis. Using the coding library with definitions, the research assistant independently recoded 20% of observations. The percentage agreement between the 2 raters was 81%, an acceptable threshold for intercoder agreement.^19,20  Disagreement about codes applied to individual excerpts were resolved by consensus agreement. The research team then organized the codes into themes.

The research assistant observed 73 RRTs. For the 73 observed RRTs, 35 patients were transferred to a higher level of care and 38 patients remained on the floor. A total of 19 833 words of text from field notes were organized into 592 units of analysis. From the 592 units of analysis, 770 coded findings were identified and were organized into 21 codes.

After completion of the analysis, codes were categorized as enhancer codes, that is, they enhanced team performance, whereas threat codes were those that threatened RRT performance. Beyond these designations, the codes appeared to be naturally grouped; thus, the research group organized the codes into 3 larger themes: teamwork (335 units of analysis), leadership (286 units of analysis), and families and patients (149 units of analysis). The 3 themes are outlined in Tables 1–3 with related codes and definitions.

Teamwork (Table 1) represented more than half of all units of analysis. Enhancers of RRT performance included the codes “collegiality” and “order.” The observer noted examples of collegial conversation about management, orderly progression through the event, and openness to questions. Threats to RRT performance included observations such as providers not speaking up, indecision, unfamiliarity of providers, disruptive behavior, disorderly behavior, and lack of resources. Some teams struggled with difficulty reaching a plan, members being upset by the plan the team leader created or being seemingly hesitant to speak, unexpected interruptions or excessive noise, or emotion, such as anger or frustration, that heightened the RRT.

Leadership (Table 2) included 286 units of analysis. Leader behaviors that enhanced RRT performance included bedside manner, being composed, and being inclusive of staff. Specific behaviors included leaders introducing themselves to the patient and team members to establish rapport early, staying calm when the patient was unwell and when treatments were administered, and making efforts to seek out staff input. Threats to RRT performance included leaders who lacked presence, changing leadership, and leaders who were disrespectful to team members. There were examples of disrespectful statements, but also disconcerting were leaders who were unsure of what to do or instances when a leader had to leave abruptly.

Observations about patients and families are shown in Table 3. Enhancers of RRT performance included an engaged family, a well-appearing child, and a well-known patient. Family members who became participants in the team and engaged in conversation were observed when the PICU team members already knew the patient or family. Also, a child who was relatively well appearing and less acute set the team at ease. Threats to RRT performance included families and patients in distress, language barriers, and family exclusion. RRT activations occurred when the patient was in pain or distressed, when family members were openly upset or angry, or when families were attempting to participate but not engaged by the team. This was noteworthy, in particular, for family members who spoke limited English.

After conclusion of content analysis and organization of RRT themes, we observed that many of our enhancer codes and comparable threat codes, such as leader bedside manner versus lack of presence reflected opposite sides of an underlying construct. We observed that perceptions of the RRT as being orderly or chaotic could be explained by other team, leader, or family and patient codes. We propose a framework to understand these codes in Fig 1, with the goal to move from chaotic to orderly with identification of the related codes. Enhancer leadership codes, such as bedside manner and inclusive of staff, and family and patient codes, such as engaged family, all contribute to an orderly team. Likewise, threat codes, such as changing leadership, lack of resources, and family or patient in distress, all lead to a chaotic team.

To date, there have been few studies in which RRTs were directly observed, with few observations per study.^6,21–23  Our study, with 73 total observations, adds a more extensive and comprehensive investigation to the literature. We confirmed some known enhancers and threats to RRT performance gleaned from surveys and focus groups, but we also revealed phenomena not currently described in the literature. Factors that enhanced RRT performance most frequently were leader composure and team collegiality. Although collegiality is not novel, these items came up frequently and perhaps reflect the relative importance of this for team function. Novel factors that enhanced order were leader decisiveness and composure as well as engaged family members and a well-appearing patient. These features foster open dialogue regarding medical management and ability to reach a consensus in a timely manner. In contrast, chaotic RRT activations suffered from threats such as a lack of leader presence, expression of disruptive behavior from providers, exclusion of family, and the family or patient being in distress. The presence of excessive personnel and interruptions, such as code blue activations or entry of consultant teams, was perceived as more chaotic.

The teamwork concepts of collegiality and morale, not speaking up, negative attitudes of participants, and indecision have previously been described.^24–28  Morale and teamwork was described by Benin et al⁷  as a source of empowerment for nurses. Lack of resources, such as ICU bed availability, was also previously identified by Shearer et al²²  and Thrasher et al²⁴  as a threat to RRT performance.

Good leadership is essential to the success of ICUs, operating rooms, and trauma teams.^29–32  Training RRTs can improve team performance, but few studies address leadership directly.^5,15,33  We found that leaders who remained composed (calm and decisive) enhanced the team function, whereas those who lacked presence, were distracted, or appeared indecisive were threatening to the team. To improve RRT performance, team leader training could focus not only on avoiding disrespectful behavior but on maintaining bedside manner and decisiveness even in the face of diagnostic and treatment uncertainty.

The number of observed disruptive behaviors during the RRT activation was striking. Previous studies have highlighted poor nurse-to-physician interactions, feelings of intimidation, belittlement, or expectations of poor interactions.^{26–28,33,34}  In additional work, researchers have commented on the emotional intelligence of providers and the ability to monitor one’s own emotions and those of others.^35,36  To these we add specific examples of chaos, including comments that expressed anger, sarcasm, or passive aggression; conflicts about how to proceed with patient triage; comments about how the patient was adding to the burden of work; and conversations over the phone about unrelated patient care. Medical team training literature distinguishes technical skills and nontechnical skills.^15,37  Our results suggest that providers may benefit from honing nontechnical skills to avoid behaviors such as thoughtless comments and conversations or phone calls about other patients.

The role that a family and patient play in team performance is unique. Much of the focus on RRT improvement regarding families is in empowering them to activate RRTs. However, we observed that high levels of family or patient stress or comfort threatened the team. Patients who were perceived as well and families who engaged with the team were enhancers to the team because teams appeared to be almost relieved to find a patient to be okay. In contrast, some RRTs had family members present who were never spoken to or updated. Team members cannot dictate a family’s emotional state, but they can take the opportunity to improve responses to families and patients in distress and to address language barriers quickly. Although RRT research and training is focused greatly on the medical team, our findings emphasize a need to focus on patient and family engagement not just for the benefit of the patient and family but to enhance the team, reduce chaos, and enhance order and collegiality. In previous literature, authors have commented on family activation of RRTs, and most recently, the impact of relationships with families and family empowerment during escalation of care is gaining notice.^38–41  In preparing for future interventions, considering formal training on how to address families, respond to their emotions, and maintain composure in the setting of family or patient distress would be an important goal.

We propose a framework relating these codes and themes that may be used to inform interventions to improve RRT performance. Although the framework appears imbalanced, with more evident chaotic codes, it is likely that some contrasting positive behaviors were not as evident to the observer and thus not recorded and coded. For example, whereas one RRT might have suffered from a lack of resources, an RRT with adequate resources might not have been notable. The utility of the framework to direct training, education, research, or qualitive improvement work may focus on mitigating features of chaotic RRTs and thus would lead to orderly RRTs. The entire framework or parts of it may be used. For example, an intervention to train RRT leaders may be focused on composure and bedside manner, on recognizing the impact of a distressed patient and family member, or on being prepared to address a lack of resources. A quality improvement initiative toward minimizing noise, interruptions, and excessive personnel during the event may help to improve the sense of order. Training providers to acknowledge and separate their own emotions and stressors from other elements of patient care may also have a role in improving RRT performance.

Our study had several limitations. We observed RRTs only during the day for the convenience of our independent observer; however, this time restriction may not accurately capture all elements of RRT function. Additionally, having an observer at any hospital event may change behavior. Nevertheless, we believe that the presence of many personnel at these events helped reduce the Hawthorne effect by making the presence of the observer less obvious. Using only 1 observer is limiting as well (1 person may miss things, and although we had a training period with a research coordinator, we do not have an assessment of interrater reliability of observations themselves). The assessment of whether someone is in distress or well appearing might have been inaccurate from a nonclinician, and the experiences of barriers and facilitators by the medical team members might not have been picked up on without the benefit of a direct interview. Finally, the ability to generalize is limited because this is a single-site study.

The use of qualitative analysis to review in situ observations of a large number RRT activations has enabled us to gain a more comprehensive understanding of enhancers and threats to team functioning. We have identified interruptions, provider stress levels, and disruptive behaviors as targets of opportunity for improvement. Acknowledging family and patient input during RRT activations may also aid team performance. We propose a new framework for understanding threats and enhancers of RRT performance that can help in developing training modalities for RRTs.