To derive a reliable clinical decision rule for the use of computed tomography (CT) in children who sustain minor head injury, investigators from the Pediatric Emergency Research Canada (PERC) Head Injury Study Group prospectively enrolled consecutive children aged ≤16 years from 2001 to 2005. Children were eligible if they had sustained non-trivial blunt head trauma within the preceding 24 hours and presented with an initial Glasgow Coma Score (GCS) of ≥13 to one of 10 Canadian pediatric emergency departments. Witnessed disorientation, loss of consciousness, definite amnesia, or persistent vomiting or irritability was required for inclusion in the study. Children with acute focal findings on neurologic examination, obvious depressed skull fracture, history of developmental delay, or head injury due to suspected non-accidental trauma were excluded. Examining physician assessors recorded 26 standardized clinical items prior to CT scan. Recursive partitioning was used to determine the best set of predictors among those items found to have high inter-observer agreement between physician assessors and strong association with outcome variables.
The need for neurologic intervention served as the primary outcome (defined as death within seven days, need for a neurosurgical procedure, or endotracheal intubation for treatment of head injury). Brain injury on CT was the secondary outcome measure (defined as any acute intracranial finding on CT attributable to the injury, excluding non-depressed or basilar skull fractures). Patients who did not undergo CT scanning were contacted for a structured telephone interview at 14 days after the injury, and those children with persistent neurologic symptoms returned for clinical reassessment and CT.
Of the 3,866 patients enrolled, 2,043 (52.8%) underwent CT. Mean age of subjects was 9.2 years; 95 (2.5%) had a score of 13 on the GCS, 282 (7.3%) had a score of 14, and 3,489 (90.2%) had a score of 15. Brain injury was detected in 4.1% of all enrolled subjects. The rate of neurologic intervention was 0.6%. Data from 245 eligible subjects were excluded because of incomplete telephone follow-up. Four high-risk predictors (those associated with the need for neurologic intervention) and three medium-risk predictors (those associated with brain injury on CT scan) were identified.
The four high-risk predictors included: 1) GCS <15 at two hours after injury; 2) suspected open or depressed fracture of the skull; 3) worsening headache; and 4) irritability. When any one of these four high-risk predictors was present, the sensitivity for predicting the primary outcome (need for neurologic intervention) was 100.0% (95% CI, 86.2–100.0%) and would require that 30.2% of patients with minor head injury undergo CT scan. The medium-risk predictors included: 1) any sign of basilar skull fracture (eg, cerebrospinal fluid otorrhea/rhinorrhea, Battle’s sign, hemotympanum, or “raccoon” eyes); 2) large, boggy scalp hematoma; or 3) dangerous mechanism of injury (eg, fall from ≥3 feet, 5 stairs, or...