2.1. Tool modification and decision support

Our team of investigators, which includes experts in pediatric emergency medicine and critical care, added 15 cardiac‐arrest specific fields, including Utstein variables, ^{14 , 15 , 16 , 17 , 18} to our previously validated PEDS tool. The original PEDS tool contained 36 fields and covered the entire spectrum of clinical cases, whereas the current tool is for OHCA, exclusively. The added variables were selected based on the Cardiac Arrest Registry to Enhance Survival (CARES) guidelines for OHCA data collection. Added variables included whether the arrest was witnessed, whether a bystander performed cardiopulmonary resuscitation (CPR) or defibrillation before EMS arrival, the etiology of the cardiac arrest, and if return of spontaneous circulation was achieved. To increase efficiency for clinical reviewers, we moved many existing data elements that could be abstracted by a non‐clinical data abstractor (eg, patient demographics, EMS scene time, interventions performed, and the timing of these interventions) into a separate section. Reliability between abstractors was achieved through training sessions with clinical experts (C.O.E., M.H., and G.M.) and evaluated through multiple rounds of comparison of agreement and discussion. The details of this process are noted elsewhere. ¹³ We created data fields to identify details of key interventions, including whether an intervention was indicated, delayed, or required multiple attempts to successfully complete. We made subsequent minor edits to the PEDS‐OHCA tool based on input from our team of 6 chart reviewers who are experts in pediatric emergency or critical care medicine.

To decrease subjectivity in the assessment of ASEs, we created a decision support matrix based on current PALS and NRP guidelines as well as the clinical expertise of our study team and chart reviewers who are pediatric emergency or critical care physicians or fellows (Table S1). This matrix is organized according to domains of care in the PEDS‐OHCA tool, and identifies the criteria, ASE subtype, severity, and preventability for common ASEs. We focused on ASEs identified in prior work examining pre‐hospital care for OHCA. ^{12 , 19}

The authors developed and tested a chart review tool to identify severe adverse safety events during the out‐of‐hospital care of children with cardiac arrest. Averaging 8 minutes per chart and with substantial inter‐rater agreement, this tool provides a systematic approach to evaluate the safety and quality of emergency medical services care for pediatric cardiac arrest.

Using REDCap electronic data capture tools hosted at Oregon Health and Science University, ^{20 , 21} we enhanced decision support for clinical reviewers by using key elements of data abstracted by non‐clinical reviewers and displayed the information in the corresponding domain of the tool. In the procedures and airway domains, this included patient age and weight, type of vascular access, number of attempts and success in establishing access, electrical therapies applied, airway procedures performed, size of equipment used, and the time between arrival on scene to when key interventions were performed (Figure 1). For the medications and fluids domains, this included medications administered, dose, concentration, route, and also the time between arrival on scene and first dose. Based on interventions, dosing, and times from the charts, notices appear in REDCap when there appears to be a deviation from the guidelines in the ASE detection matrix (e.g., medication dose given is too high or low for the child's weight, or time between arrival on scene and starting bag‐valve‐mask ventilation is too long).

Example of the PEDS‐OHCA tool and decision support in REDCap