Objects The aim of the authors is to derive a safe, effective and feasible symptom-driven CT rule in fully conscious children ≥3 years with symptoms after head trauma, based on time-framed clinical course, radiological findings, outcome measures and prognosis of patients.

Methods Fully conscious but symptomatic children ≥3 years after head injury (1997–2010) with CT performance ≤2 h since injury were included in the study. Additional exclusion criteria were set for patient selection. Evolution of clinical symptoms of patients in 24 h since injury was the focus in current study. Clinical data were extracted from standardised medical records on admission and observation charts.

Results Data of 1897 eligible cases were retrospectively reviewed. Traumatic brain injury (TBI) was revealed radiologically in 73 cases (3.8%). Eight cases underwent surgery. Recursive partitioning analysis identified the following factors in the CT rule: any delayed headache commenced between 4 and 10 h since injury; significantly worsening headaches present between 2 and 12 h since injury; vomiting between 6 and 12 h since injury; and headache without significant changes persisted ≥12 h since injury. It has a sensitivity of 100% (95% CI 95.0% to 100.0%) and specificity of 72.1% (95% CI 70.0% to 74.1%) to predict cases with TBI.

Conclusions A symptom-driven CT rule has been derived to identify cases at high risk of having TBI in fully conscious, but symptomatic children with mild closed head injury. To be practical, an additional observation rule is added.