Implementation of a Pediatric Emergency Department Cardiopulmonary Resuscitation Quality Bundle Free

This QI initiative was conducted at an academic pediatric ED with an annual volume of 62 000 patients. The ED resuscitation bay has an annual volume of 4200 patients, and approximately 30 cardiac arrest patients each year. Volumes decreased in the setting of the COVID-19 pandemic; from June 2020 to June 2021, total ED volume was 50 000 and resuscitation bay volume was 2500. Teams in the ED resuscitation bay consist of at least 1 physician (1 attending, usually joined by a resident and fellow), 4 nurses (lead nurse, medication nurse, and 2 bedside nurses), 1 respiratory therapist, 2 patient care assistants (PCAs), and 1 paramedic. The PCAs are primarily responsible for chest compressions, and the paramedic is primarily responsible for defibrillator pad placement and defibrillator management. Our current defibrillators are the Zoll R-Series (Zoll Medical Corp; Pittsburgh, PA).

Each bed in the resuscitation bay has ceiling-mounted video- and audio-recording equipment installed, and each encounter is recorded. Video encounters are reviewed using a proprietary software program (Live Capture; B-Line Medical, Washington, DC). Patients and families consent to video recording through the general ED consent process. Our hospital’s institutional review board determined this project to be nonhuman subjects research.

All patients with OHCA and prearrival notification from emergency medical services (EMS) were included in this study. In our hospital’s catchment area, EMS organizations almost uniformly prenotify the pediatric ED dispatch of patients with OHCA. This allows the ED team time to prepare for the incoming cardiac arrest patient. Our previous cardiac arrest QI work has focused on patients with prearrival notification to standardize our process, with the goal of extrapolating our process to patients who arrive without notification or arrest in the ED. Exclusion criteria included traumatic cardiac arrest, use of an external compression device or signs of nonsurvivable arrest (rigor mortis, dependent lividity) on arrival where time of death was stated before the 2-minute mark of the video recording. Cases meeting exclusion criteria were reviewed by the cardiac arrest committee to ensure completeness and accuracy of data.

In August 2020, an interdisciplinary cardiac arrest committee was formed consisting of ED physicians, nurses, paramedics, and respiratory therapists. This committee’s work was conducted in collaboration with our ED’s Medical Resuscitation Committee and Center for Simulation and Research, whose QI work for critically ill children has been described previously.¹⁰  Additionally, our institution has considerable QI resources and expertise, with several investigators having completed advanced training in QI science.

The improvement team reviewed the results of the initial project and the relevant literature. In August 2020, we completed a detailed, video-based review of the care of patients with OHCA in our ED. Baseline data on discrete aspects of resuscitative care were collected during this review. Our ability to perform video review allowed us to precisely examine non-CPR interventions and adherence to AHA guidelines. We theorized that a standardized choreography and engagement of frontline providers would optimize performance of discrete aspects of resuscitation, including initial cardiac rhythm check, advanced airway placement, end-tidal capnography application, and approach to vascular access. We created a key driver diagram to illustrate the interactions between these key drivers and the subsequent interventions (Fig 1).

The improvement team selected a bundle of 5 elements of resuscitation to improve: (1) IV/IO placement, (2) first epinephrine administration, (3) advanced airway placement, (4) ETCO₂ monitoring, and (5) initial cardiac rhythm verbalization. After selecting bundle elements, we determined baseline bundle performance and explored both the reasons for delays in bundle completion and the relationships between elements. For example, time to first epinephrine administration is directly related to the placement of initial vascular access, and ETCO₂ monitoring may require advanced airway placement.

We then designed interventions to improve the efficient performance of each bundle element. To address vascular access, we tasked the nurse on the patient’s right side (or the ED fellow, if present) to place a right tibial IO immediately after patient transfer to the ED stretcher and back defibrillator pad placement. To improve the timing of the first epinephrine dose, we asked the medication nurse to prepare epinephrine as soon as possible, ideally before patient arrival, and the nurse on the patient’s right side to administer the first dose as soon as vascular access placement was confirmed. Throughout didactic and in situ training, we emphasized also immediately assessing prehospital IV/IO access sites on arrival to the trauma bay. For advanced airway placement, we instructed the respiratory therapist (or ED fellow, if present) to place a SGA immediately after patient arrival (if not already placed by EMS). There is currently no strong evidence for the superiority of any initial airway management strategy for pediatric cardiac arrest,^1,11,12  but initial placement of an SGA provides advantages (“may improve ventilation, reduce the risk of aspiration, and enable uninterrupted compression delivery”¹ ) while avoiding the disadvantage of ETT placement (takes longer and may interrupt compressions). If return of spontaneous circulation is achieved in our ED, the SGA is subsequently exchanged for an ETT before pediatric intensive care unit admission. The respiratory therapist was also tasked with placement of an ETCO₂ monitor on the supraglottic device. Finally, we tasked the ED paramedic with prompting the physician team lead to perform an initial cardiac rhythm assessment.

Although our improvement work began in summer 2020, implementation was delayed until December 2020 because of policy limiting in-person education and simulation resulting from COVID-19 precautions. Initial cardiac rhythm assessment education was disseminated to the paramedics and PCAs in December 2020 via mandatory annual skills checkoff sessions. Education of the core nurses was completed in December 2020 via a virtual presentation on the following interventions: multiple doses of epinephrine drawn up prearrival, immediate assessment of prehospital access patency, and immediate placement of a tibial IO on patient arrival and epinephrine administration after access confirmation. Respiratory therapist team meetings were used to present education on advanced airway placement, and didactic education was also conducted at physician division meetings to address all bundle elements.

We used in situ simulation to further disseminate and reinforce all interventions with the entire resuscitation team. In January 2021, a cardiac arrest in situ simulation with an initial rhythm of asystole was performed weekly in the ED. In February 2021, a ventricular fibrillation in situ simulation incorporating bundle elements was developed, and in March 2021, weekly in situ simulations (4 total) were performed, with emphasis to the whole ED team on the cardiac arrest bundle elements. The March 2021 simulation debriefing script was modified based on feedback from the January in situ simulations.

Finally, beginning in March 2021, we started performing hot debriefs in the ED within 4 hours of each CPR event. The hot debrief is a structured, team-based discussion of the cardiac arrest, led by the nurse manager and physician team leader. The discussion offered an opportunity for all care team members to provide feedback on case management and highlighted CPR bundle elements to encourage standardized patient care. At the same time, we also began e-mailing specific and actionable video review-based feedback to all team members involved in the case using a standardized CPR report card (Supplemental Fig 4) in addition to narrative feedback, which is a more detailed description of successes and suggestions for improvement. An advantage of the dual debrief system is that even when hot debriefs were unable to be performed because of staffing issues, our QI team was always able to send the cold debrief feedback e-mails. In addition, regularly scheduled refresher in situ simulations were unable to be performed in the second half of 2021 because of staffing shortages during the COVID-19 pandemic as well as a hospital-wide moratorium on simulations resulting from a move to a new building.

We used the same approach to patient identification and data collection as in our original project.⁸  The same study team member (M.C.) collected data using a standard electronic form. The video review software records continuously, which allows the exact time of patient arrival and interventions to be noted. We have previously demonstrated the superiority of video review for collection for time-based resuscitation data.^8,10  For each cardiac arrest case, M.C. collected the time of patient arrival and the time each element of the cardiac arrest bundle was performed. These data were stored using a Research Electronic Data Capture electronic database.¹³ 

For IV/IO placement, we documented time of successful placement of a vascular access device by successful blood return or flush. If the patient arrived with prehospital access, IV/IO placement was documented at the time the access point was successfully flushed. For epinephrine administration, we documented the time epinephrine was flushed through the vascular access device. For advanced airway placement (both for airway devices placed in the ED and prehospital), we documented the time of confirmation of correct placement of either an SGA or ETT via either verbal confirmation, visualization of chest rise, or visualization of ETCO₂ waveform. For ETCO₂ application and monitoring, we documented the time of the first measured ETCO₂ waveform on the monitor, including if an ETCO₂ monitor was attached to a bag-valve mask. Finally, for initial cardiac rhythm verbalization, we documented the time of the first pause in compressions where either a rhythm is verbalized to the team or a “shock” or “no shock” decision is verbalized.

Our primary outcome measure was time to completion of the 5 bundle elements (in seconds). Our SMART aim was to reduce completion of all 5 bundle elements from a baseline of 300 seconds to less than 120 seconds within 1 year of implementation. We also measured time to completion of each element and whether an element was ever completed.

Our balancing measures were time to defibrillator pad placement and time off the chest in the first 120 seconds. We chose these elements because they are key metrics for quality CPR, and we wanted to make sure that emphasizing the bundle elements did not lead to increased time off the chest or delayed time to defibrillator pad placement.

We used statistical process control charts to determine if our interventions were associated with improvement in the time to performance of the cardiac arrest bundle as a whole and each bundle element individually. Because the incidence of patients presenting in cardiac arrest is relatively rare, we used I charts to track our measures over time. In this type of control chart, each data point represents an individual patient case. We used standard rules for interpretation of a Shewhart chart to determine special cause variation.¹⁴ 

Our video review data for baseline data collection started in May 2019, and we began our cardiac arrest bundle interventions in December 2020. Between May 2019 and June 2021, 56 OHCA cases managed in our resuscitation bay met study inclusion criteria. The median patient age was 0.41 years (IQR 0.16-2.49) and 54% of the patients were male.

Figure 2 demonstrates an improvement in time to completion of bundle elements from a mean of 302 seconds at baseline to 147 seconds. This shift occurred in February 2021 after the initiation our cardiac arrest bundle interventions and has been sustained through June 2021.

We detected special cause variation for 4 of 5 bundle elements. Mean time to initial cardiac rhythm verbalization decreased from 215 to 115 seconds (Fig 3). This shift occurred after paramedic training in December 2020 emphasizing the paramedic role in prompting for timely rhythm check and the paramedic’s role as a CPR coach. Time to the first dose of epinephrine also met criteria for special cause variation, with a decrease in mean time to completion from 204 to 123 seconds (Supplemental Fig 5). We noted this change after the beginning of the second phase of in situ simulations in March 2021. Time to advanced airway placement and time to ETCO₂ measurement met criteria for special cause variation (Supplemental Figs 6 and 7). After the first and second phase of the cardiac arrest bundle in situ simulation, special cause variation occurred for both measures. These simulations emphasized routine supraglottic airway placement and confirmation with ETCO₂ application. Mean time to advanced airway placement decreased from 174 to 64 seconds, and mean time to ETCO₂ measurement decreased from 158 to 56 seconds. There was an observed trend toward improvement in time to vascular access (IV or IO), but this did not meet criteria for special cause variation (Supplemental Fig 8).

Our balancing measures, time to defibrillator pad placement and time off the chest during the first 2 minutes of CPR, did not change during the study period (mean time to defibrillator pad placement was 90 seconds before December 2020 and 74 seconds after, mean time off the chest was 13 seconds before and 11 seconds after).

OHCA is a high-acuity, low-frequency event in the pediatric ED. These resuscitations are complex, with multiple critical interventions in the first minutes of arrival to the ED. Based on previous work in our ED on CPR handoff choreography,⁸  we theorized that a standardized choreography and engagement of frontline clinicians would improve the performance of a bundle of 5 discrete elements of resuscitative care. After implementing our interventions, time to bundle completion, as well as time to completion of 4 of the 5 bundle elements, improved. Although the time to bundle completion remains above our goal, most bundle elements now have a mean completion time of less than 120 seconds.

Although the time to first vascular access and first epinephrine were lower, mean time to completion of these bundle elements remain higher than 120 seconds. Moreover, delays in these 2 elements were responsible for many of the delays in bundle completion. Interestingly, when patients arrived in the resuscitation bay without prehospital vascular access, the standardized choreography for immediate IO placement and epinephrine administration was successful. However, delays often occurred when patients arrived with an EMS IO in place; these prehospital IOs were frequently unable to be flushed, and there was a delay while placement was verified before placing a new IO in the ED. This is consistent with previous cadaver studies showing that pediatric IO placement is often inaccurate, especially in infants.^15,16  Based on these findings, our next education efforts have focused on simultaneous verification of prehospital access and placement of an IO as a second point of access to prevent delays in verification of prehospital IO placement.

Engaging respiratory therapists was essential to improving SGA placement and ETCO₂ measurement. This observation is consistent with other studies demonstrating the feasibility of supraglottic device placement instead of intubation in cardiac arrest.^11,12,17  There were several notable exceptions to timely SGA placement, which were caused by extenuating circumstances including new COVID-19 precautions early in the pandemic and an infant with an anatomically difficult airway.

The decrease in time to initial rhythm verbalization was significant. This is a crucial component to the resuscitation; although shockable rhythms are rare in pediatric patients, survival decreases significantly the longer these rhythms persist without recognition. We believe this observed improvement was due largely to our original improvement initiative, in which the ED paramedic improved the time to placement of the defibrillator pads. By getting the defibrillator pads on quickly after patient arrival, the paramedic is primed to think about the rhythm and have the mental space to prompt rhythm verbalization. Reinforcing this choreography with our didactic training and in situ simulations appears to have helped sustain this improvement.

Most previous studies on improving care of patients with OHCA have focused on improving CPR. For example, studies have shown that interventions focusing on CPR coaching can improve the quality of CPR, including chest compression quality and compliance with AHA recommendations.^18–21  Although these studies are generally consistent with our findings, our study had a notably different focus (ie, discrete, non-CPR aspects of resuscitative care that follow AHA recommendations and have the potential, especially early rhythm recognition and early epinephrine administration, to affect patient outcomes). By standardizing the choreography, our interventions decreased both the time to completion of these recommended elements as well as the variability in time to completion, leading to increased consistency in team performance.

The strengths of this study include a multidisciplinary team approach and the ability to review the exact time of each intervention by video review, rather than relying on code documentation.²²  Additionally, standardizing the choreography for each bundle element empowered our nurses, respiratory therapists, and paramedics to quickly complete crucial tasks while decreasing the cognitive burden on the physician code team leader.

The limitations of this study include that it was performed at a large, tertiary care pediatric ED, which may limit generalizability because other institutions may not have as many resources or staff available for cardiac arrests. In addition, lack of video review capabilities may limit implementation of these interventions in other institutions because it would be difficult to have an impartial observer at all OHCA resuscitations to evaluate compliance with cardiac arrest bundle elements. Finally, although our interventions are based on AHA recommendations and physiologic principles, we do not include patient outcome data in this report.

In the future, we hope to report on the influence of our CPR quality initiative on survival to hospital discharge. Our next phase of cardiac arrest improvement care will focus on CPR coaching and CPR quality metrics using defibrillator and ETCO₂ data. Other next steps include sustaining our progress through regularly scheduled in situ simulations, as well as standardizing our processes for after return of spontaneous circulation care and declaration of death.

Our implementation of a cardiac arrest QI initiative has significantly decreased the time to bundle completion, as well as crucial individual resuscitative elements. Our interventions were developed based on multiple key drivers, allowing us to develop a reproducible and sustainable choreography for the first critical moments of each OHCA resuscitation.

AHA

American Heart Association

CPR

cardiopulmonary resuscitation

ED

emergency department

EMS

emergency medical services

ETCO₂

end-tidal capnography

ETT

endotracheal tube

IO

intraosseous

IV

intravenous

OHCA

out-of-hospital cardiac arrest

PCA

patient care assistant

QI

quality improvement

SGA

supraglottic airway