Management of patients with mild head injury (MHI) is open to debate.¹ In the last few years, there has been a trend towards earlier diagnosis, implying an extensive use of computed tomography (CT), rather than admission and observation. The National Institute for Clinical Excellence (NICE) has recently proposed new evidence based recommendations on all steps of the management of patients with MHI.² In the diagnostic algorithm, coagulopathy (history of bleeding, clotting disorder, or current treatment with warfarin) is not considered a predictor variable necessitating early CT in subjects without loss of consciousness (LOC) or amnesia since injury. This statement conflicts with previous guidelines, where history of coagulopathy, independently of symptoms, indicated CT.³

Since 1999, all cases with MHI attending the Emergency Department of our district hospital have been treated and registered in a comprehensive database according to predefined procedures.³ Our criteria for CT and/or hospital admission are wider than the NICE criteria; in particular, there is routine detailing of NICE variables, but in addition, all subjects with coagulopathy have an early CT, independently of symptoms and signs after injury. This provides the opportunity to determine the risk related to coagulopathy and the accuracy of the NICE recommendations.

We analysed the data of 7955 consecutive patients within 24 hours from trauma, who had been triaged for an acute MHI. MHI was defined as an injury of the head, other than any superficial injury to the face, Glasgow Coma Score (GCS) definitely 14 or 15, in subjects aged ⩾10 years. We excluded 1258 more patients because of unclear history of the trauma as primary event, major trauma with unstable vital signs, GCS <14, penetrating injuries, pregnancy, or voluntary discharge. All patients re-attending for complaints after discharge (282 cases) underwent a CT scan and in this study were considered only once. All patients received written recommendations at discharge for home observation and complaints that would require referral back to hospital for further evaluation. Observers were instructed to check for symptoms and signs, and for any change in patients’ clinical status for 7 days.

According to NICE, CT scan is recommended in the presence of: (a) GCS <13 at any point and/or equal to 13 or 14 at 2 hours after injury, (b) any sign of basal skull fracture, (c) any focal neurological deficit, (d) post-traumatic seizure, (e) vomiting (>one episode), and (f) amnesia of events before impact >30 minutes, (g) risk factors (coagulopathy, age ⩾65 years, dangerous mechanism of injury), provided that patients have experienced some LOC or amnesia since injury. In our protocol,⁴ CT is mandatory for subjects with risk factors, in particular amnesia and/or LOC (but excluding old age), independently of signs and symptoms.

Following our protocol, 4081 out of 4547 (89.8%) eligible patients had an early CT scan. In 3580 early CT was also indicated according to the NICE protocol; in 501, CT scans were performed in subjects outside the NICE protocol. These patients had CT because of coagulopathy (warfarin therapy) in 66 cases (13.2%), diffuse headache in 178 cases (35.5%), previous neurosurgical intervention in 26 cases (5.2%), history of seizures in 22 cases (4.4%), dangerous mechanism of injury in 172 cases (34.3%), and recent alcohol and/or drug misuse in 58 cases (11.6%).

Clinically important intracranial lesions were demonstrated in 477/3580 (13.3%) patients of the NICE group. Neurosurgical intervention was required within 7 days in 97 patients (2.7%) for haematoma evacuation or for elevation of depressed skull fracture. At follow up (6 months), 36 patients (0.1%) had an unfavourable outcome (death, persisting vegetative state, or severe disability by the Glasgow Outcome Scale), rated by an expert physician on the basis of a structured telephone call.

In the 501 NICE negative cases, 40 patients (8.0%) had an intracranial haemorrhagic lesion: intracerebral haematoma (20 cases); intracerebral haematoma plus subarachnoid haemorrhage (2); intracerebral haematoma plus subdural haematoma (3); subarachnoid haemorrhage (2); subarachnoid haemorrhage plus subdural haematoma (1), subdural haematoma (11); and epidural haematoma (1). This prevalence is lower compared with NICE positive cases (Fisher’s exact test, p  = 0.0006), but nevertheless NICE recommendations would not have led to early detection of these 40 lesions, for which neurosurgical intervention was required in five (12.5%): intracerebral haematoma evacuation (1 case), subdural haematoma (3), subarachnoid haemorrhage plus subdural haematoma (1). At follow up, only one patient died after 9 days for causes related to intracerebral haematoma, the remaining having a favourable outcome. In these 40 NICE negative cases with haemorrhagic lesions, coagulopathy was the main factor leading to CT scan in 16 cases (40%), and was associated with a fivefold increase in the risk of intracranial lesions (table 1). With logistic analysis, coagulopathy was the only predictor variable associated with CT lesions in asymptomatic patients not fulfilling NICE criteria for early CT. Six patients, re-evaluated for complaints after a median (interquartile range) time of 144 hours (66 to 168), had an intracranial lesion detected by a second CT; four belonged to the NICE positive group, two were in the NICE negative. None had coagulopathy.

The post hoc analysis of our prospective database demonstrates that NICE recommendations for CT scanning identify the majority of patients with intracranial lesions in subjects attending the ED for MHI. However, the exclusion of coagulopathy as a factor always indicating CT impairs the diagnostic accuracy of NICE guidance. Routine use of CT scanning is not cost effective; more than 90% of CT scanning are negative in subjects with MHI, and at least 98% are negative for epidural haematoma, the event requiring immediate intervention. A more liberal policy for CT use, making CT mandatory in patients with coagulopathy, independently of head trauma severity, would indicate only 66 additional CT in our total cohort of 3581 (less than 2.0%), with a 1:4 probability of identifying an intracranial lesion.

The indications for CT use in MHI are subject to a continuous debate.⁵ Our data strongly suggest that the restrictive use of CT proposed by NICE in the presence of risk factors may be generally accepted. However, in the light of our data we suggest that CT should also be considered for all subjects with coagulopathy.