Head injury accounts for approximately 700 000 emergency department (ED) attendances in England and Wales each year (National Institute for Health and Clinical Excellence (NICE), 2007), 90% of which are classified as minor (Glasgow Coma Score (GCS) 13–15). Diagnostic assessment can either use a clinical decision rule or unstructured assessment of individual clinical features to identify those who are at risk of intracranial injury and require computed tomography (CT) scans and/or hospital admission.
In moderate or severe head injury management, pathways are clearly structured with in-hospital care being essential to treatment (Joint Royal Colleges Ambulance Liaison Committee (JRCALC), 2006; NICE, 2007). However, with minor head injuries, approximately 8% will sustain intracranial injury and only 1% will require any specialist intervention (NICE, 2007). Management of this large proportion of patients involves a balance between under-investigation, which risks missed opportunities to provide early effective treatment for intracranial injury and over-investigation, which risks unnecessary radiation exposure and wasting of NHS resources.
The overall aim of this project was to use secondary research methods to determine the most appropriate diagnostic management strategy for adults and children with minor head injury (GCS 13–15) in the NHS. Two of the specific objectives were to:
With regard to paramedic practice, both the JRCALC and NICE guidelines for head injury/trauma are focused on the management of more serious injury and the prevention of secondary brain injury. Minor head injury patients will often present to the emergency services and this project looked at evidence that is relevant in the prehospital setting.
Methods
To describe current NHS practice, we mailed a questionnaire to the lead clinician of all major acute hospital emergency departments in the UK and analysed routine emergency department data from hospital episodes statistics (HES). Where possible, we correlated questionnaire responses with HES to determine whether service provision was associated with a difference in the proportion of patients admitted. Further methodological details are provided elsewhere (Goodacre et al, 2010).
For the review, several electronic databases (including MEDLINE, MEDLINE In-Process, CINAHL, EMBASE and the Cochrane Library) were searched from inception to April 2009 (updated searches to March 2010 were conducted on the MEDLINE databases only). Searches were supplemented by hand-searching relevant articles (including citation searching) and contacting experts in the field.
Inclusion criteria were cohort studies of patients with minor head injury in which a clinical decision rule or individual clinical characteristics (including biomarkers and skull radiography) were compared to a reference standard test for intracranial injury or need for neurosurgical intervention. Study quality was assessed using the quality assessment of diagnostic accuracy studies (QUADAS) tool. Where sufficient data existed in accuracy studies, we used meta-analysis to generate pooled estimates of sensitivity, specificity and likelihood ratios (Positive = PLR, Negative = NLR) (Box 1 and Table 1 for explanations).
Disease positive | Disease negative | ||
---|---|---|---|
Test positive | a | b | a+b |
Test negative | c a+c | d b+d | c+d a+b+c+d |
Positive predictive value (PPV) = a/(a+b) (The proportion of subjects with a positive test who actually have the disease) | |||
Negative predictive value (NPV) = d/(c+d) (The proportion of subjects with a negative test who do not have the disease) | |||
Sensitivity = a/(a+c) (The proportion of subjects with the disease that the test correctly identifies as positive) | |||
Specificity = d/(b+d) (The proportion of subjects without the disease that the test correctly identifies as negative) | |||
Positive likelihood ratio (PLR) = Sensitivity/(1-Specificity) (X - in definition below) (Test positive would be X times as likely to be seen in someone with the disease as someone without). |
Results
Survey of current practice
The survey of NHS emergency departments showed that nearly all had unrestricted access to CT scanning (adults 96%, children 94.5%). The median proportion of attendances admitted was higher for adults (18%) than for children (9%). Adults were usually admitted to an observation ward or clinical decision unit (61%), while children were usually admitted to an inpatient ward (87%). There was no evidence of an association between the proportion admitted and the admission team, location or requirement for senior or specialist approval (all P>0.1). The relationship to CT scan findings was not assessed with this survey.
Review
The literature searches identified 8003 citations. Of these, 93 full text papers were included for the assessment of diagnostic accuracy and 71 for the assessment of individual characteristics. The quality of studies and reporting was generally poor with inconsistent application of inclusion criteria, reference standards (i.e. CT scan or follow-up contact) and outcome criteria (i.e. the significance of different injuries). More methodological details can be found in the full project report and individual papers published elsewhere (Harnan et al, 2011; Pandor et al, 2011; Pickering et al, 2011).
Clinical decision rules
The Canadian CT head rule (CCHR) (Stiell et al, 2001) was the most widely validated adult rule with a sensitivity of 99–100% and a specificity of 48–77% for neurosurgical injury (NSI) using the high-risk criteria. Using the high or medium-risk criteria gave a sensitivity of 99–100% for NSI and 80–100% for any intracranial injury (ICI), with corresponding specificities of 37–48% and 39–51%, respectively.
Other rules assessed included the New Orleans CT rule (NOC) (Haydel et al, 2000); Scottish Intercollegiate Guidelines Network (SIGN) (Scottish Intercollegiate Guidelines, 2009); NEXUS II (Mower et al, 2005); and the European Federation of Neurological Sciences (EFNS) (Vos et al, 2002) guidelines. None were as accurate in the prediction of NSI or ICI than the CCHR. Relevant to the UK, the current NICE guidelines are based on this CCHR (Table 2).
Minor head injury defined as:
|
High risk criteria (for neurological intervention)
|
Medium risk criteria (for brain injury on CT)
|
Rules for children were less well validated. Several had high sensitivity and acceptable specificity in derivation cohorts but the limited validation data suggested specificity was poor. The UK NICE guidelines are based on the derivation cohort data from the children's head injury algorithm for the prediction of important clinical events (CHALICE) study (Dunning et al, 2006) which remains the best option for UK practice at present. The recently published pediatric emergency care applied research network (PECARN) rule (Kuppermann et al, 2009) demonstrated high sensitivity and specificity but would result in a considerable number of normal scans being performed to identify one ICI or NSI. This is not acceptable in the paediatric population because of the high risks of radiation exposure at this young age.
Individual characteristics
The results are summarized separately for adults, children and infants (Table 3).
Age Group | PLR >10 | PLR 5–10 | PLR 2–5 | PLR <2 |
---|---|---|---|---|
Substantial significance | Marked significance | Moderate significance | Unlikely significance | |
Adults |
|
|
|
|
Children |
|
|
|
|
Infants |
|
|
||
Predicting intracranial injury in infants
Aged 0–2 years in most studies, despite the limited number of varied studies:
Predicting intracranial injury in adults
Predicting intracranial injury in children
Aged 2–16 in most studies, the most useful clinical characteristics were:
Discussion
This study has identified the current best evidence for the management of minor head injuries in the UK. In adults, the most accurate and validated decision rule is the CCHR (on which the NICE guidelines are based) with a high sensitivity and acceptable specificity for identifying both intracranial and neurosurgical injury. In children, the evidence is less clear with poor validation of the currently existing rules. In both age groups, direct comparison of decision rules was hampered by inconsistent patient selection, rates of CT use, follow-up methods and diagnostic criteria.
This systematic review and meta-analysis found that most of the clinical characteristics that we analyzed had some value (PLR >2) for diagnosing intracranial injury (Table 2). Most notably, skull fracture (whether clinical or radiological), post-traumatic seizure, focal neurological deficit, decrease in GCS or persistent vomiting all indicated a markedly increased risk of intracranial injury in adults (PLR > 5).
Similarly, a depressed skull fracture, basal skull fracture, focal neurological deficit, coagulopathy or post-traumatic seizure, markedly increased risk of intracranial injury in children and depressed skull fracture or focal neurological deficit in infants. Despite the limitations of the data, clinical characteristics of limited diagnostic value included loss of consciousness and headache in adults and scalp haematoma and scalp laceration in children.
The failure to demonstrate diagnostic value of many characteristics for identifying neurosurgical injury probably refects the limited data available for this outcome and should not be interpreted as showing that individual characteristics are of limited value. There are good theoretical reasons to anticipate that characteristics that are useful for diagnosing any intracranial injury will also be valuable for diagnosing specifically neurosurgical injury.
Clinical decision rules for minor head injury are based upon individual clinical characteristics, with the presence of a criterion indicating the need for CT scanning (or hospital admission prior to the widespread use of CT). There is substantial variation in the criteria used by each rule and it is interesting to examine the diagnostic value of each item, as estimated in our meta-analysis.
Adults
Most adult rules use GCS less than 15, focal neurological deficit, loss of consciousness, vomiting and amnesia (Table 4). Our meta-analysis of these individual characteristics suggested that loss of consciousness has little diagnostic value, although this may refect its use as a selection criterion in many studies. The other four criteria were supported by our meta-analysis, although vomiting was only useful if it was persistent. Most rules did not specify that vomiting had to be persistent.
GCS less than 13 on initial assessment in the emergency department |
GCS less than 15 at 2 hours after the injury on assessment in the emergency department |
Suspected open or depressed skull fracture |
Any sign of basal skull fracture (haemotympanum, ‘panda’ eyes, cerebrospinal fluid leakage from the ear or nose, Battle's sign) |
Post-traumatic seizure |
Focal neurological deficit |
More than one episode of vomiting |
Amnesia for events more than 30 minutes before impact. |
Age 65 years or older |
Coagulopathy (history of bleeding, clotting disorder, current treatment with warfarin) |
Dangerous mechanism of injury (a pedestrian or cyclist struck by a motor vehicle, an occupant ejected from a motor vehicle or a fall from a height of greater than 1 m or five stairs). |
Approximately half the rules specified suspected basal or depressed skull fracture, age, seizure, decreasing GCS, mechanism of injury or coagulopathy as criteria. Our meta-analysis suggested that these were useful criteria (or at least fall from a height and bicycle or pedestrian RTC were useful with regards to mechanism of injury). Conversely, several rules used headache as a criterion, whereas our meta-analysis suggested this was of limited diagnostic value. Interestingly, this criterion also seems to have been added to NICE guidelines (NICE, 2003; 2007) by some NHS Trusts (Goodacre et al, 2010).
Overall it appeared that NICE guidelines matched the findings of our meta-analysis very well (perhaps better than any other prediction rule) in terms of including criteria that are diagnostically useful and excluding those that are not. We found little evidence to support the application of additional criteria to the NICE guidelines.
Children and infants
Most rules for children use loss of consciousness, GCS less than 15, skull fracture, vomiting, headache and visible injury as criteria. Our meta-analysis of the individual characteristics supported the use of loss of consciousness, GCS less than 15, skull fracture, vomiting and headache (if severe or persistent), but suggested that scalp laceration/ haematoma or an undefined headache were of little diagnostic value. Less than half the rules used focal neurological deficit, amnesia or seizures as criteria, few used mechanism of injury and only one used coagulopathy as criteria (Table 6).
Loss of consciousness lasting more than 5 minutes (witnessed) |
Amnesia (antegrade or retrograde) lasting more than 5 minutes |
Abnormal drowsiness |
Three or more discrete episodes of vomiting |
Clinical suspicion of non-accidental injury |
Post-traumatic seizure but no history of epilepsy |
GCS less than 14, or for a baby under 1 year GCS (paediatric) less than 15, on assessment in the emergency department |
Suspicion of open or depressed skull injury or tense fontanelle |
Any sign of basal skull fracture (haemotympanum, ‘panda’ eyes, cerebrospinal fluid leakage from the ear or nose, Battle's sign) |
Focal neurological deficit |
If under 1 year, presence of bruise, swelling or laceration of more than 5 cm on the head |
Dangerous mechanism of injury (high-speed road traffic accident either as pedestrian, cyclist or vehicle occupant, fall from a height of greater than 3 m, high-speed injury from a projectile or an object). |
However, our meta-analysis suggested that these criteria were all potentially diagnostically useful. Overall the CHALICE (Dunning et al, 2006) and NEXUS II (Mower et al, 2005) rules appeared to be most consistent with the findings of our meta-analysis, in terms of including criteria that are diagnostically useful and excluding those that are not.
At present, patients with minor head injury require transfer to the emergency department and clinical review prior to a definitive decision being made. The diagnostic accuracy of the identified rules is high for adults, and could potentially be developed in to a tool for the prehospital setting. Having a sensitivity of 99–100% for neurosurgical injury with an associated specificity of 48–77% means that, using the CCHR high-risk criteria, almost no patients with significant injuries will be missed. It also means that over half of those with no concerning injury will be correctly identified which could, potentially, lead to a reduction in transport rate.
For children, more research is required to validate the existing decision rules but adoption of the CHALICE-based NICE guidelines could, again, reduce the number of transports required.
Conclusion
Evidence around factors that can predict significant brain injury in seemingly minor head trauma is now well established. Application of the current knowledge to prehospital practice is an area for exploration and could have a significant impact on service use.