Analysis of Pediatric Cervical Spine Injury Evaluation Workflow in the Emergency Department

The decision to image children for cervical spine injuries (CSI), an uncommon injury with high associated morbidity and mortality, is complex. X-rays and computed tomography are often used to screen for CSI; however, radiation exposure increases lifetime cancer risk. We recently developed a CSI prediction rule to inform pediatric imaging decisions. To guide implementation, we conducted a workflow analysis of CSI screening in children after blunt trauma.

We interviewed emergency departments (ED) and trauma clinicians at 21 hospitals following the applied cognitive task analysis task diagram, knowledge audit, and simulation interview approaches. Interviews were coded using a combined deductive-inductive approach to construct a workflow diagram and identify critical decisions with associated workflow junctures, decision makers, physical locations, and cognitive demands. We noted areas of high, medium, and low variability.

We interviewed 48 participants (emergency medicine physicians and advanced practice providers [n = 22], nurses [n = 14], and surgeons [n = 12]) across 21 hospitals located in the Western (n = 15), Northeastern (n = 5), and Midwestern (n = 1) USA. Critical decisions within the pediatric CSI imaging decision workflow included trauma triaging, spinal motion restriction application, clinical clearance, decision to image, and imaging type. There was moderate-to-high variability between hospitals for most workflow junctures and decision makers. However, more consistent patterns emerged, such as a high level of ED and trauma attending involvement throughout the process. The knowledge audit revealed a heavy cognitive load associated with decision making. "Big picture," "past and future," "noticing", "job smarts," and "equipment" were highly relevant cognitive demands.

Moderate-to-high between-hospital variability in workflow for decision making around CSI evaluation for pediatric trauma patients could potentially complicate CSI prediction rule implementation.