Problem/background

Approximately 6000 paediatric patients receive in-hospital cardiopulmonary resuscitation (CPR) each year, with 90% of these cardiac arrest events taking place in the paediatric intensive care unit (PICU).1–3 Patients who have cardiac arrests represent a significant proportion of paediatric mortality, with only 45% surviving to hospital discharge.1 High-quality CPR is associated with improved survival,3 however, CPR guidelines are often not met during resuscitation attempts.4

In an effort to improve compliance with American Heart Association (AHA) resuscitation guidelines, the AHA launched the ‘Get with the Guidelines-Resuscitation’ programme. Their goal was to create performance benchmarks and evidence-based feedback across all patient populations.5 As part of this effort, the AHA recommends specific parameters for CPR depth, rate, velocity and percentage of high-quality chest compressions. These are well studied modifiable factors that impact survival in patients who experience both in and out of hospital cardiac arrest.2 A retrospective cohort study from the AHA Get With The Guidelines-Resuscitation Registry showed that over 50% of adult cardiac arrests have deviations from recommended guidelines. Deviations from the AHA guidelines were associated with decreased rate of return of spontaneous circulation (ROSC).6 The use of resuscitation quality bundles like the CPR Coaching, Objective-Data Evaluation, Action-linked-phrases, Choreography, Ergonomics, Structured debriefing and Simulation (CODE ACES 2) have been associated with increased AHA CPR guideline compliance during in-hospital arrests.7 The use of multidisciplinary performance debriefing tools not only improves team performance, but improves patients short-term outcomes.8

There are a wide variety of approaches to improve the delivery of high-quality CPR, which include standard resuscitation courses, real time feedback devices and postarrest team debriefing tools. Standard resuscitation courses are the mainstay of CPR training and maintenance of certification, yet evidence is lacking that these courses improve outcomes.9 10 Courses of this nature have the issue of diminishing performance over time, with loss of skills as soon as 1 year after training.10 One approach to addressing this issue of skill loss over time is incorporating real time feedback via devices that are integrated into defibrillators and other devices, and these interventions have demonstrated moderate improvement with AHA guideline compliance. However, these systems focus strictly on improving technical aspects of basic life support performance.3

A more comprehensive approach to resuscitation education is integration of post-arrest team debriefing tools into reviews of resuscitation efforts. This approach has been demonstrated to improve CPR quality, short-term survival and neurological outcomes in both adult and paediatric patients.3 A multidisciplinary approach in postarrest debriefings of this nature further enhances compliance with paediatric AHA guidelines and improves CPR outcomes.4 Debriefings can also help facilitate discussion after an event to improve future communication among the resuscitation team members.11–13 Debriefings also allow healthcare workers educational opportunities that are relevant, goal-oriented, practical and collaborative.11 Current evidence suggests that paediatric post-cardiac arrest debriefing programmes are feasible and useful for providers, regardless of their participation in the resuscitation.8

Given the importance of high-quality CPR in improving patient outcomes, percentage of arrest resulting in ROSC and improved percentage of hospital discharges, we desired to study cardiac arrests in the Duke PICU to address the lack of a standardised system to track incidence, performance and outcomes of cardiac arrest through a comprehensive debriefing programme. Due to the lack of a tracking system or preintervention data we were unable to target specific areas for improvement during a cardiac arrest event nor were we able to track our cardiac arrest outcomes compared with published data. Therefore, we felt that it was critical to established a quality improvement programme around cardiac arrest events and CPR in our PICU.

Measurements

The Duke PICU is part of Duke Children’s Hospital (DCH), which is a quaternary care centre with a diverse population and wide range of patient diagnoses. The PICU has over one thousand patients admitted annually, with a wide spectrum of medical and surgical diagnoses, including both solid organ and bone marrow transplant recipients. Acute postoperative patients with congenital heart disease are cared for in a separate paediatric cardiac intensive care unit. The facility has full Extracorporeal Membrane Oxygenation (ECMO) capabilities including ECMO-CPR (ECPR).

At DCH, the leader for cardiopulmonary resuscitative efforts is either the PICU fellow, advanced practice provider (APP) or a paediatric intensivist in both the PICU or in-patient floors. The PICU is staffed 24 hours a day, 7 days a week by a paediatric intensive care physician who completed an Accreditation Council for Graduate Medical Education accredited paediatric critical care fellowship who is board eligible or certified. These physicians have a range of experience as paediatric critical care physicians from 2 to 10 years. Thus, a paediatric intensivist is present for all cardiac arrest events. All PICU physicians, APPs, residents, PICU fellows, nurses and respiratory therapists are trained in PALS.

The code team for the PICU includes a paediatric intensivist, a paediatric intensive care fellow, a paediatric intensive care nurse practitioner, a respiratory therapist and 1–3 PICU nurses. There are three code carts in the PICU. Two code carts are the Paediatric Code Carts (Blue) that are made by ARMSTRONG MEDICAL (A-SMART CART SYSTEMS). Each Paediatric Blue Code Cart is checked each shift by the charge nurse to ensure the seal is intact. If the seal is broken, then a full cart check is done. The carts are checked monthly for expired products and medications. There is also one Universal Code Cart (Red) (MMP Medcart), which is standardised hospital-wide. This code cart is checked and restocked once a month. The Hospital Code Committee oversees contents and processes related to the Universal Code Cart (Red) while the Paediatric Code Blue Committee, which is a subcommittee of the Hospital Code Committee, oversees the Paediatric Code Carts (Blue). There are two defibrillators in the PICU (LIFEPAK 20e) and we use Physio Control Quick-Combo pads (paediatric and adult) for defibrillation,

A key driver diagram was created to inform our project and is included in online supplemental documents. From January 2017 to December 2018, all cardiac arrests in the PICU were tracked and reviewed in detail. A paediatric critical care fellow was notified when a patient received chest compressions of any duration in the PICU. Within 48 hours, the PICU fellow and faculty code champions performed a comprehensive review of the available data.

Each individual cardiac arrest was analysed and presented at a monthly multidisciplinary Morbidity and Mortality (M&M) conference. Year-to-date data were aggregated and presented at the beginning of each calendar year. PICU faculty and staff members, graduate medical education trainees and consulting services attended each meeting regardless of their involvement with specific patients or resuscitation efforts.

At each session, a templated review of each arrest was presented. This template is included in online supplementary material. Each presentation started with the ground rules for participation with emphasis on the peer-protected nature of the discussion and importance of a professional and blame-free environment. Next, we addressed basic event information including time and date of event, duration of the arrest, time to CPR, quality of CPR including rate and depth, medications administered, rate of ROSC, need for ECPR and patient outcome. Patient information included: medical history, admission diagnosis, procedures, labs and prior pertinent imaging. Vital signs were tracked for the prior 12 hours to the event and throughout the entire resuscitation effort. We presented graphics of arrest data including vital signs, ECG with rate and rhythm, pulse oximetry tracing and arterial line waveform, non-invasive cardiac output monitor data if on invasive ventilatory support and near-infrared spectroscopy trends. Code-specific information included prearrest rhythm, CPR interruption and rationale (pulse check, intubation attempt, ECMO cannula placement), defibrillation attempts with preshock and postshock rhythm and chest compression quality including rate and depth. We assessed compliance with AHA guidelines. Cardiac arrest events were considered compliant if they followed all the recommended AHA guidelines for CPR including: no pauses in CPR lasting longer than 10 s, assessment of high quality compressions by analysis of the arterial wave form or pulse oximeter waveform, code medications given at appropriate paediatric dosing and intervals, shockable rhythms were treated at appropriate Joule dosing, and there was no greater than a 2 min delay in CPR once a patient met cardiac arrest criteria. We targeted areas for clinical care and CPR improvement including assessment of prearrest warning signs, delayed time to initiation of CPR, communication to and from the code leader and code documentation. Examples of high-quality CPR and resuscitative efforts and exceptional teamwork and communication were also highlighted during the reviews. Peer-to-Peer discussions and feedback were preformed after each cardiac arrest event with nursing, provider and respiratory therapist. These discussions were standardised and included a set of questions addressing: events leading up to the arrest, individual concerns about the patient prior to the arrest, were concerns addressed prior to arrest, what went well during the arrest, any equipment or medication issues, were code roles defined, was there closed loop communication, what did not go well and what could have been done differently in the same situation and finally additional constructive thoughts (table 1). Examples of templates for providers, respiratory therapists and nursing peer-to-peer discussions are also included in online supplemental materials.

Project personnel entered data into Research Electronic Data Capture (REDCap) database hosted by Duke University. REDCap is a secure, web-based application designed exclusively to support data capture for research studies. Members of the research team completed all required trainings in the protection of human subjects. In this way, access to any patient identifiers was limited to the minimum necessary to carry out this project. All data and records generated were kept confidential in accordance with institutional policies on subject privacy and HIPPA regulations. This project was reviewed by the Duke IRB and granted IRB exemption. It was not appropriate or possible to involve patients or the public in the design, conduct, reporting or dissemination plans in this quality improvement initiative.

Design/strategy

With our first Plan-Do-Study-Act (PDSA) cycle, we set out to establish the creation of a multidisciplinary paediatric cardiac arrest review process. Information about the review programme was sent to faculty and staff by email and posters were placed in the PICU provider work areas. The development of a ‘code corner’ was created to house code sheets and debriefing documents in one easily visible centralised location. Two PICU fellows were trained on the code review process, including use of the central telemetry system. We found that despite education, awareness and a centralised location for arrest-related forms, there were multiple missed code event notifications early in the process. After eliciting feedback during the cardiac arrest review meetings from nursing, PICU fellows and respiratory therapists, we noted many did not know how to notify the committee members of an arrest and were unaware of some code related documentation processes. During this time, we also held cardiac arrest reviews as an independent educational conference. However, subjectively minimal attendance with often less than ten people in attendance and little discussion about arrest events. Given our initial challenges, we sought to improve notification of cardiac arrests and to improve attendance at the cardiac arrest review meetings. Despite low attendance at review meetings, we were encouraged by the overall enthusiasm and buy-in from all invested participants to develop a standardised approach to the cardiac review process. We successfully completed nine cardiac arrest reviews during our first PDSA cycle that occurred between January 2017 and April 2017. However, although we did complete nine cardiac reviews, many of the reviews had missing information due to lack of timely notification.

To help improve notification and education around the cardiac arrest review process, our second PDSA cycle involved developing a multidisciplinary committee made up of PICU nurses, respiratory therapists, PICU fellows, paediatric intensivists, nursing leadership and respiratory leadership. This second PDSA cycle spanned from April 2017 to July 2017. This committee was responsible for notifying the PICU fellow of recent arrests and helping staff to fill out appropriate documents. Other responsibilities included facilitating peer to peer discussion and feedback following a code event and attending two review meetings each year. Committee members were encouraged to use this educational work for tiered advancement in their respective fields. Feedback on cardiac arrest areas for improvement was critical in creating a more streamlined approach to cardiac arrest care as well as to improve AHA compliance. Creation of the multidisciplinary cardiac arrest committee allowed for more recruitment of more interested individuals across multiple PICU roles. Given the nature of the project dependent on feedback and education across disciplines, attendance at the cardiac arrest meetings was critical. Given robust multidisciplinary participation at monthly PICU M&M conferences, cardiac arrest review meeting was added to the M&M PICU conference agenda. This improved more attendance and participation in event discussions and feedback. Subjectively, we went from five to six participants at cardiac arrest review meetings to a full room with standing room only during M&M conferences.

Despite development of a cardiac arrest committee, we continued to have difficulty with timely notification of PICU cardiac arrests. In order to review all relevant telemetry data during an arrest event, the cardiac arrest review committee needed to be notified within 24 hours of the arrest. After 48 hours, the telemetry data are removed from the bedside Philips monitors and cardiac rhythms, arterial line and pulse oximetry waveform data may not be retrievable. Cardiac arrest notifications were often greater than 48 hours after the event occurred, leading to increased number of arrest reviews with incomplete waveform and cardiac rhythm analysis. Therefore, our third PDSA cycle spanning from July 2017 to January 2018, involved a daily automated text message to the PICU charge nurse phone asking about cardiac arrests that occurred in the PICU over the past 24 hours. Based on a yes or no reply, a follow-up phone call was made to obtain patient information. This allowed us to capture all cardiac arrests occurring in the PICU within 24 hours of an event. In order to provide a more secure data collection platform, a REDCap database was also created to streamline the review process and allow for a secure centralised location for all data.

There was great interest in the creation of a nurse lead education committee around cardiac arrest events based on feedback from nursing at M&M conference. We felt that expanding the cardiac arrest review committee was critical in providing education around AHA guidelines and other interventions suggested during review committee meetings. During our fourth PDSA cycle, from January 2018 to July 2018, we created a nurse led Cardiac Arrest Education Committee consisting of six to eight nurses with at least 2 years of critical care experience who worked a variety of day and night shifts. Objectives of this committee included: leading peer-to-peer postarrest debriefing sessions with bedside nursing staff to review specific resuscitative elements, attending at least two code committee reviews per year, and spearheading peer education around various cardiac arrest topics. Five cardiac arrest education topics were chosen during the first year including: (1) getting to know your code cart, (2) how to draw up code medications, (3) CPR basics including rate, depth and hand placement, (4) CPR during ECMO deployment and (5) rhythm recognition and interventions. These topics were chosen by the nursing committee members based on feedback from the prior year’s cardiac arrest review meetings. An education bulletin board was placed in the break room of the PICU and each staff member was required to sign a sheet stating they read through the education materials. In addition, hands on simulations were led by the PICU fellows/APPs during both day and night shifts for relevant topic areas.

We continue to review all cardiac arrests occurring in the PICU and present data at our monthly multidisciplinary M&M conference. Through sustained quality improvement efforts, we hope to improve our cardiac arrest outcomes over the next few years. Eventually, we hope to reduce the number of paediatric cardiac arrests in the PICU through early recognition of at-risk patients and timely intervention.

Conclusions/lessons/limitations

The primary aim of this project was to create a multidisciplinary process to review cardiac arrests and implement a programme to improve the quality of CPR. The multidisciplinary approach was critical to our success. Involvement of attending physicians, fellows, nurses, respiratory therapists and leadership created an engaged and invested working group. Key lessons learnt included the importance of incorporating a new process into existing workflow and having early involvement of all stakeholders.

We also sought to increase compliance with AHA guidelines during CPR events. Our AHA compliance increased from 17% to 76% from 2017 to 2018. It is difficult to draw conclusions from this as 73% of cardiac arrest had incomplete reviews in 2017. However, we feel that the creation of a comprehensive cardiac arrest review process with multi-disciplinary feedback could have led to increased AHA guideline compliance.

Despite the initial enthusiasm for the project, we identified several issues with reporting and reviewing cardiac arrests in a timely manner. Early on, several arrests were not reported, and the bedside telemetry was unable to be reviewed. In retrospect, implementing an automated daily text messaging system to the charge nurse sooner would have enabled timely reporting for this novel programme in our unit. Identifying both nursing leadership and respiratory therapy leadership earlier in the process would have also helped with staff investment and continued staff education.

This project highlighted several barriers to optimal team performance during resuscitations which could have contributed to the increase in AHA compliance during CPR between 2017 and 2018. Items as simple as changing the notification system for RT supervisor and PICU attendings of an in-unit code allowed for automatic notification of key team members. Other improvements attributed to our cardiac arrest initiative include development of an identification system for high-risk airways and consistent identification of the code leaders during resuscitation attempts.

While this project clearly improved our processes, there are several limitations. First, the lack of baseline data prior to our interventions limits our ability to fully demonstrate the impact of this initiative. Variability of fellow/APP staffing along with Registered Nurse (RN) and Respiratory Therapist (RT) staff turnover made it challenging to keep team members abreast of initiatives coming out of the code committee. Some of the early data collected were incomplete or not collected at all due to lack of communication. Missing 73% of our 2017 cardiac arrests prevented any meaningful clinical inferences to be drawn. Better notification and data capture would have enabled a more robust comparison between the two study years.

Cardiac arrest reviews are now an integral part of our PICU’s M&M conference ensuring that the reviews process is sustainable and will continue to occur on a monthly basis. Each academic year, a new fellow representative will join the cardiac arrest review committee, so that a minimum of 3 fellows (one per year) will be part of the review process, ensuring sustainable fellow presence. A paediatric intensivist faculty mentor will continue to oversee the cardiac reviews and provides mentorship to fellows. Next steps include continued data collection and tracking, understanding risk factors for cardiac arrest in a medical-surgical PICU, identification of pre-arrest trends which could hopefully prevent cardiac arrest events from occurring.

In conclusion, our cardiac arrest review and education programme allows for comprehensive multidisciplinary review of individual events, informs quality improvement initiatives, ensures consistent high-quality CPR and will hopefully lead to improved patient outcomes.