Over the last year the problem of radiological investigation of the acutely head injured child has been much featured in this journal. We were not so much concerned with what to do with the severely injured—such children are, invariably, intubated and mechanically ventilated, and the most pressing question is to identify whether they need immediate surgery. Head computed tomography (CT) is the obvious investigation. Rather, the real problem arises in the uncooperative 1 to 8 year old who just won’t lie still long enough for CT—is it really worth getting one even if it means endotracheal intubation and a brief anaesthetic, or will a skull x ray suffice, or should the child be admitted for observation until we are sure all is well?

Dunning et al reviewed 16 previously published papers and provided us with a meta-analysis of variables that predict significant intracranial injury in minor head trauma.¹ Kupperman commented on this report and presented the broader context of what should or shouldn’t be done in the emergency department with this information.² Later in the year the CHALICE (children’s head injury algorithm for the identification of significant clinical events) study group presented the implications of applying United Kingdom National Institute of Clinical Excellence (NICE) guidelines to the emergency department management of exactly such children.³ The authors reported the theoretical impact of advocating early CT in place of skull x ray and admission. Data from 10 965 children attending three hospitals between February 2000 and August 2002 were used in the model. Twenty five per cent of the patients had a skull x ray, 0.9% had head CT, and 3.7% were admitted. The authors calculated that adherence to the NICE guidelines would have resulted in a 0.3% skull x ray rate, an 8.7% CT rate, and a 1.4% admission rate. They concluded that implementing the new guidelines should not increase the workload caused by patients attending the emergency department with head injury, but will “move their management from the observation ward to the radiology department.” In this issue of the journal, Reed et al take this particular discussion one step further by describing what has actually happened in their emergency department.⁴ But before we discuss the detail and implications of these paediatric data, we should briefly review the broader perspective of emergency neuroradiology for minor head injury in adults.

In 2001 Stiell et al developed a highly sensitive clinical decision rule for use of CT in adults with minor head injuries as an alternative to risk stratification based on skull x ray.⁵ Ten Canadian centres conducted a prospective cohort study of consecutive adults who presented with a Glasgow coma scale (GCS) score of 13–15 after head injury. The investigators were interested in two main outcomes: the need for neurological intervention, and clinically important brain injury on CT. In 3121 patients, 8% had clinically important brain injury, and 1% required neurological intervention. The Canadian CT Head Rule was derived from these data and consisted of five high risk factors (failure to reach GCS of 15 within 2 hours, suspected open skull fracture, any sign of basal skull fracture, vomiting more than twice, or age older than 65 years) and two medium risk factors (amnesia before impact longer than 30 minutes, and dangerous mechanism of injury). The high risk factors were 100% sensitive for predicting the need for neurological intervention, and the investigators calculated that application would require only 32% of patients to undergo CT. Later, in 2003, Stiell et al validated this rule in a second cohort of a similar number.⁶ Application of the Canadian CT Head Rule into emergency department practice in the UK has been discussed,^7,8 and the department in Cambridge has evaluated its impact on clinical practice.^9,10 A “before and after” study—seven months in 2001 and nine months in 2002—of minor head injury (GCS 13–15) in individuals older than 15 years showed that CT rates increased from 47/330 (14%) to 58/267 (20%), admissions for observation increased from 111/330 (34%) to 119/267 (45%), and skull x ray rates fell from 33% to 1.6%. The suggestion from these data in adults is that, while the “rule” does reduce skull x ray rate, lowering the threshold for CT does not obviate the need for admission—presumably, in Cambridge, this is still being used as a diagnostic test of potential neurosurgical intervention.

In this issue of the journal, Reed et al report a “before and after” study in children aged between 1 and 14 years attending one hospital emergency department with head injury.⁴ Data from 1867 patients attending between August 2002 and July 2003 were compared with data from 1535 patients attending between august 1998 and July 1999. The hospital had decided that routine skull x rays had no place in the paediatric emergency department for children aged 1 year and older, and they had rules to guide admission for observation or head CT. Between the two periods there was no change in admission rate (10.1% v 10.9%), no change in proportion of patients with abnormal CT scans (25% v 26%), and no difference in identification of intracranial injury (0.2% v 0.4%) or in rates of neurosurgical intervention (0% v 0.1%). However, twice as many individuals were exposed to CT (1% v 2.1%). These data are important because of two clinical questions that arise from them.

First, why do we admit children with minor head injury? Is it because of the mechanism of injury, or the history, or a reduction in GCS score, irrespective of the radiological findings? Minor head injury in children is not the same as in adults. The mechanism of head injury is different: children are not the drivers of automobiles or motorcycles; rather, they are usually the passengers, pedestrians, or cyclists. We have no data to indicate that a GCS score of 13–15 has the same meaning in children and adults,¹¹ and amnesia is impossible to assess in the young child. It now seems likely that, like the adult experience,^9,10 introducing a CT rule in children will have no impact on reducing admission rates. In contrast to the suggestion from the CHALICE study group,³ management is not moved “from the observation ward to the radiology department.” In fact, in the report by Reed et al the radiology department became much less involved (before and after: 20.6% v 1.9%). Clearly, the end points we are most interested in are intracranial injury and need for neurosurgical intervention. We know that admission is a poor diagnostic test for these outcomes—that is, it is not very specific: the vast majority of children (97%) admitted for observation are discharged the following day without any further treatment being given.⁸ So we don’t really have an answer to our question – why do we admit children? – but it is clear that we will need to learn more about our behaviour and the decision making process, and how we can make it more efficient in children.

Second, what are the implications on public health of carrying out more CTs for minor head injury? In the report by Reed et al there is the suggestion that the radiation dosage per case of head injury was equivalent across the two series (0.045 v 0.042 mSv), and was therefore not influenced by the introduction of the CT rule. However, the key issue is that the CT rate was doubled. In the first series 16 children had CT: 13 patients were exposed to 2 mSv from CT (11/13 being normal), three were exposed to 2.11 mSv from skull x ray and CT (1/3 normal scans), and 337 were exposed to 0.11 mSv from skull x rays alone. In the second series 39 patients were exposed to 2 mSv from CT (29/39 normal scans). Kupperman recently reviewed for the journal the risk of radiation from CT²: the lifetime attributable risk for fatal cancer from one CT ranges from 1 per 2000 scans for young infants to 1 per 5000 for those 10 years old.^12,13 Taken together with the point above, if admissions and interventions are unaltered by the introduction of a CT rule we need to consider who has most to lose from the reduction in skull x rays at the expense of increasing the CT rate, and give thought to broadening the debate by also giving parents a choice.

The study by Reed et al is a step in developing better head injury protocols for children. We now know what actually happens to children in one emergency department. In the ensuing years we should hear more from various large groups: the CHALICE study, the CATCH study (Canadian assessment of tomography for childhood head injury), NEXUS (the national emergency x radiography utilisation study), and ISHIP (the international study of head injury project).