Predicting TBI: Prediction rules or judgment?

Andrea Cruz (Texas Children’s) and Michelle Macy (Michigan) presented the best PEM-related articles of 2016 at AAP this past October. If you weren’t in the room that day, we’re going to be counting them down over here.

Comparison of Prediction Rules and Clinician Suspicion for Identifying Children with Clinically Important Brain Injuries After Blunt Head Trauma

Atabaki SM, et al. Academic Emergency Medicine 2016.

Unmet Needs Addressed

CT imaging has been overused to evaluate children with blunt head trauma despite associated risks to patients and costs. PECARN developed prediction rules to determine which children are at very low risk for clinically important traumatic brain injury (ciTBI). But how do these rules compare to clinical suspicion? In cases when the clinician had a low suspicion for ciTBI but ordered a CT anyway, this study determined reasoning for obtaining the CT.


  • Secondary analysis of data collected at 24 PECARN centers that derived and validated the original prediction rules.
  • The validation sample was used for analysis to minimize potential bias that would come with using the derivation sample.
  • The clinician suspicion of ciTBI was recorded on a scale. Less than 1%, 1-5%, 6-10%, 11-50%, and greater than 50%. Clinician suspicion did not have the benefit of the prediction rules since these were still being developed.
  • The study also asked for indications for CT when the suspicioin for ciTBI was less than 1%.

Main Results

  • For children under 2 years old, the sensitivity of the prediction rule for predicting ciTBI was 100% while the sensitivity of the clinician was 60%. The specificity of the rule was 54% while the specificity of clinical suspicion was 92%.
  • In children over 2 years old the sensitivity and specificity were similar to the under 2 group.
  • There were 7,688 children in which the clinical suspicion of ciTBI was less than 1%. Over a quarter of these patients (2,099) had a CT and these CTs identified 32 instances of ciTBI. The following is a table of the rationale for obtaining a CT in these children.

Why Did This Make Our Top 10?

  • Clinical decision rules were more sensitive than clinical suspicion
  • Clinical suspicion was more specific than clinical decision rules
  • Clinicians still ordered CTs when their suspicion for ciTBI was less than 1%
  • CT rates are likely higher in settings without PEM providers
  • Translating knowledge can address influences of clinician suspicion
Andrea Cruz MD MPH

Andrea Cruz MD MPH

Associate Professor of Pediatrics in Emergency Medicine and Infectious Diseases at Baylor College of Medicine
Andrea is the PECARN site PI for Texas Children’s Hospital and runs the Data Center for the Pediatric Emergency Medicine Collaborative Research Committee.
Andrea Cruz MD MPH