Low Concordance Between Pediatric Emergency Attendings and Pediatric Residents for Predictors of Serious Intracranial Injury

Abstract

Objectives: Minor head trauma is a common cause of pediatric emergency room visits. The Pediatric Emergency Care Applied Research Network head trauma clinical decision rules (PECARN-CDR) are designed to assist clinicians in determining which patients require imaging. However, only minimal data are available on the accuracy of residents' assessments using PECARN-CDR. Prior research suggests that trainees often come to erroneous conclusions about pediatric head trauma. The objective of the present study was to assess concordance between pediatric residents' and attending physicians' assessments of children with low-risk head trauma, with the ultimate goal of improving education in pediatric trauma assessment.

Methods: This is a retrospective cohort study analyzing concordance between pediatric residents and pediatric emergency attendings who provided PECARN-CDR-based evaluations of low-risk head injuries. It is a planned subanalysis based on a prospectively collected, multicenter data set tracking pediatric head trauma encounters from July 2014 to June 2019.

Results: Data were collected from 436 pediatric residents, who encountered 878 patients. In the case of patients younger than 2 years, low concordance between residents and attendings was observed for the following elements of the PECARN-CDR: severe mechanism (κ = 0.24), palpable skull fracture (κ = 0.23), Glasgow Coma Scale (GCS) score less than 15 (κ = 0.14), and altered mental status (AMS; κ = -0.03). There was moderate to high agreement between residents and attendings for loss of consciousness (κ = 0.71), nonfrontal hematoma (κ = 0.48), and not acting normally per parent (κ = 0.35). In the case of patients older than 2 years, there was low concordance for signs of basilar skull fracture (κ = 0.28) and GCS score less than 15 (κ = 0.10). Concordance was high to moderate for history of vomiting (κ = 0.88), loss of consciousness (κ = 0.67), severe headache (κ = 0.50), severe mechanism (κ = 0.44), and AMS (κ = 0.42). Residents were more conservative, that is, more likely to report a positive finding, in nearly all components of the PECARN-CDR.

Conclusions: Resident assessment of children presenting to the ED with minor head trauma is often poorly concordant with attending assessment on the major predictors of clinically important traumatic brain injury (abnormal GCS, AMS, signs of skull fracture) based on the PECARN-CDR. Future work may explore the reasons for low concordance and seek ways to improve pediatric resident education in the diagnosis and management of trauma.