Design: retrospective descriptive study of a national trauma registry, 2017-2023
Population: 12,030 US children with isolated mild traumatic brain injury
Primary outcome: head CT use by hospital type and patient characteristics
Effect: overall 43.5-51.5%; adult hospitals 59.5%, mixed 44.0%, paediatric 38.8%

Twelve thousand and thirty children with isolated mild traumatic brain injury treated at hospitals contributing to the US National Trauma Data Bank between 2017 and 2023 were examined for head CT use, with hospitals classified as adult, paediatric or mixed.

Overall CT use ranged from 43.5% to 51.5% and barely moved across seven years. Where the child arrived mattered more than when: 59.5% at adult hospitals, 44.0% at mixed, and 38.8% at paediatric hospitals. Rates were higher with lower Glasgow coma scale scores and impaired consciousness, which is appropriate, and also higher at non-teaching and smaller hospitals, which is not. Among scanned children, most were male, white, privately insured and at level I or II trauma centres.

A twenty-point gap between adult and paediatric hospitals in the same country, for the same injury, under the same national guidance, is a statement about who is doing the assessing rather than about the children. Validated decision rules exist precisely to allow a period of observation instead of a scan, and observation requires someone confident enough in paediatric assessment to wait. That is the resource that differs. The same logic applies wherever a child with a minor head injury first meets a clinician whose usual patient is an adult - which in India is most of the time.

Apply a validated paediatric head injury decision rule rather than scanning on clinical impression.
Observation for four to six hours is an accepted alternative to immediate CT in intermediate-risk children.
Rates were highest where adults are the usual patient, not where injuries were most severe.
Document the decision rule variables you assessed, including the negative ones.
Discuss radiation risk explicitly with parents; it is often the missing half of the conversation.

Why it matters

It locates the variation in who assesses the child rather than in how injured the child is, which is the part a department can change.

The statistics, in plain English

This is descriptive trauma registry data, so the comparison between hospital types is not adjusted for how ill the children were on arrival, and some of the gap will reflect case mix rather than practice. But a difference of roughly twenty percentage points, stable across seven years and across thousands of children, is too large and too consistent to be case mix alone.