The new SIGN guidelines is designed to help doctors provide early and more effective treatment for patients with head injury, as Mr Douglas Gentleman explains

A million people attend hospital each year in the UK after a head injury, 10% of whom have an actual or potential brain injury. Although outcomes have improved over the past 30 years, avoidable disability and death still occur because of failure to implement what is already known.

The Scottish Intercollegiate Guidelines Network (SIGN) head injury guideline development group was motivated by a wish to see this situation change.

Development of SIGN guidelines

The production of a SIGN guideline follows a clear process. The first step is to identify an important clinical problem where practice varies.

A broadly based expert group is then formed, and agrees on the key questions to be asked and a literature search strategy to answer them. Time spent at this stage on defining these questions and setting the limits of the topic is time well spent.

The quality of the evidence found in the literature search is appraised in a structured way before being used as the basis for preliminary conclusions in a draft guideline. At this point the group consults widely (including expert external review), and must be prepared to revise and do further work in the light of this feedback.

The final recommendations in the guideline must be explicitly based on robust scientific evidence, but must also be pragmatic, clinically realistic, and capable of being implemented across a wide range of geography and practice.

Distributing a booklet from a royal college in Edinburgh does not achieve change in inner-city Glasgow or the rural Highlands: local clinicians need to 'buy in' if the exercise is not to be an academic one.

Finally, the process should identify areas where more research is needed, include arrangements for auditing local implementation, and build in a review of the guideline 2–4 years hence.

The head injury guideline development group

The group was chaired by Graham Teasdale, Professor of Neurosurgery at Glasgow University, and included 20 clinicians (mostly doctors) from surgical specialties, A&E, anaesthetics, and radiology.

There was only one GP and no patient representative. However, since this group was set up SIGN has recognised the key role of primary care in the new NHS and the need to hear the voice of service users when interpreting published evidence to make policy choices.

We began our task in May 1997, and consulted on a draft guideline in little over a year. A surprising time was then needed to amend it in response to the consultation exercise, to show more clearly the links from each recommendation to the evidence found in the literature, and to get formal approval from the SIGN parent committee.

The guideline was eventually published in October 2000. Perhaps there is scope for speeding up the process.

We were conscious of the risk that the guideline might spin out of control and become an unfocused textbook of head injury. Some aspects are uncontroversial, such as the need for vigorous early resuscitation. Others interest only a few specialists, such as neuro-intensive care. Some really need their own guideline, such as brain injury rehabilitation or how to manage trauma in rural practice.

We decided to leave these aspects largely alone and selected our questions with care to address the issues that most concern the many doctors and other clinicians who treat head-injured patients each year in Scotland:

  • Which patients should be referred to hospital?
  • What are the indications for skull films, and for a computerised tomographic (CT) scan?
  • Which patients can safely be discharged from A&E, and which need admission to hospital?
  • How should the latter be monitored, and for how long?
  • Which patients should be referred for a neurosurgical opinion?

We also looked (belatedly) at the special points to be considered in young children with a head injury.

Finally we considered the service implications of our recommendations.

Searching the literature

To answer the above questions, we used a search strategy devised by SIGN's Information Officer, working with us. We searched the Cochrane Library, Embase, Healthstar, and Medline from 1985, adding material identified by group members.

As expected, meta-analyses, systematic reviews, and randomised controlled trials (RCTs) on head injury proved to be rare. This is not a reproach to researchers in that field, but reflects the difficulty in designing RCTs where there are so many clinical variables, adverse events happen to relatively few patients, and treatments cannot be blinded.

Where other evidence was weak or absent we sought observational studies instead. Overall we found some answers but also many gaps, indicating the need for more research.

Grading the evidence

SIGN uses a standardised system to grade the quality of published evidence and the strength of the recommendations based on it:

  • Level I evidence is most robust, and comes from meta-analysis of RCTs (Ia) or RCTs themselves (Ib).
  • Level II evidence comes from controlled studies without randomisation (IIa) or other quasi-experimental studies (IIb).
  • Level III evidence comes from non-experimental descriptive studies (comparisons, correlations, and case studies).
  • Level IV evidence – the weakest but most abundant – comes from expert reports or authoritative clinical opinions.

Any type of study must be well-designed for a recommendation to rely on it. Grade A recommendations require at least one RCT within a body of literature that is of good quality and consistency and has a direct bearing on the recommendation. We were unable to make any Grade A recommendations.

Grade B recommendations are based on well-conducted clinical studies without RCTs, and Grade C recommendations on evidence from expert committee reports and/or the clinical experiences and opinions of respected authorities. The evidence in the literature allowed us to make Grade B and C recommendations only.

Disseminating the guidelines

All SIGN guidelines aim to improve patient care, by finding the best available evidence to support clinical practice and disseminating it throughout Scotland in the guidelines.

No longer does every consultant and GP routinely get every guideline, which risked devaluing the process by deluging clinicians with material in areas irrelevant to their clinical practice. Instead, each trust in Scotland now has a guideline coordinator who can target SIGN guidelines to relevant clinical teams.

Guidelines can be bought from the SIGN office at the Royal College of Physicians of Edinburgh, and are available on the SIGN website at Each guideline includes a 'quick reference guide' – a double-sided A4 laminated sheet summarising the key points (see Figures 1 & 2,below). This easy-to-hand summary has proved popular in GP surgeries and ward offices.

Figure 1: Front of the A4 laminated Quick Reference Guide containing key points of the guideline
head injury guideline p1
Figure 2: Reverse of the A4 laminated quick Refernce guide containing key points of the guideline
head injury guideline p2

A recently issued CD-ROM brings together for the first time most existing SIGN guidelines and quick reference guides in a new user-friendly format.

Writing guidelines is not enough. It is also important to see them put into practice, and each SIGN guideline includes sections on service delivery and recommendations for audit and research.

The process of monitoring how effectively SIGN guidelines are put into practice is still in its infancy. There is a need for robust local systems to look at each SIGN guideline and decide how it can be translated into practice in a way that is both locally plausible and consistent with a national guideline.

Content of the head injury guideline

Our recommendations on who to send to hospital (Table 1, below) will be widely agreed, and indeed are already in use. Some rural GPs have access to community hospitals, and admission there may be a reasonable alternative to a lengthy transfer to a general hospital if the perceived risk of complications is low, e.g. if supervision at home is impractical but there is no clinical concern.

Table 1: When to refer a patient with a recent head injury

A head-injured patient should be referred to hospital if any of the following is present:

  • Impaired consciousness (Glasgow Coma Score <15/15) at any time since injury
  • Amnesia for the incident or subsequent events
  • Neurological symptoms, e.g. severe persistent headache, nausea and vomiting, irritability or altered behaviour, or a seizure
  • Clinical evidence of a skull fracture, e.g. CSF leak, peri-orbital haematoma
  • Significant extracranial injuries
  • A mechanism suggesting a high energy injury (e.g. road traffic accident), possible penetrating brain injury, or possible non-accidental injury in a child
  • Continuing uncertainty about the diagnosis after first assessment
  • Medical co-morbidity, e.g. anticoagulant use, alcohol abuse
  • Adverse social factors, e.g. no-one able to supervise the patient at home

The risks associated with various combinations of neurological and skull film findings are well known (Table 2), and now form the basis of head injury triage in the UK. However, as CT scanners have become more widely available worldwide, evidence has emerged that even an apparently mild head injury can be followed by clinically silent but potentially important CT abnormalities.

Table 2: Risk of an operable intracranial haematoma in head-injured patients

Glasgow Coma Score Risk Other features Risk
15 1 in 3615 None 1 in 31300
    Post-traumatic amnesia (PTA) 1 in 6700
    Skull fracture 1 in 81
    Skull fracture and PTA 1 in 29
9–14 1 in 51 No fracture 1 in 180
    Skull fracture 1 in 5
3–8 1 in 7 No fracture 1 in 27
    Skull fracture 1 in 4

Despite the predictive power of a skull fracture, the use of plain films to select patients for scanning now seems old-fashioned, even perhaps dangerous. We therefore recommend the more liberal use of scanning to get a definitive answer to the key question: is there intracranial damage?

There are important implications for professional time if more CT scans are to be done (especially out of hours), and issues about the risk/ benefit analysis of transferring patients from places where there are no scanners to hospitals with scanners. There are few data on which to draw conclusions.

Deciding whether someone needs to be admitted or discharged requires clinical skill, and again depends on the perceived risk of intracranial complications. Clinical monitoring is performed with variable skill in different places, and we have made recommendations on how (and how often) to make observations, what to check before sending the patient home, what advice to give the family, and, above all, how to recognise neurological deterioration and react to it quickly.

The particular contribution of neurosurgeons is the management of haematoma and other intracranial complications. We have recommended when to seek neurosurgical advice with a view to transferring the patient.

Over the past 20 years the threshold for seeking advice has progressively fallen, and neurosurgeons should be willing to help even when the clinical and radiological findings clearly do not merit transferring the patient to the neurosurgical unit.

Children with a head injury

We came rather late to the problem of head-injured children. There is remarkably little in the literature about this, but the similarities to head-injured adults are greater than the differences.

When a child is discharged from A&E, clear written guidelines should be given to the parents. The threshold for admitting children for observation should be low, and reasons include the lack of a responsible adult, unsatisfactory social circumstances, and suspected non-accidental injury. In the last case, a doctor experienced in this condition and a social worker should be consulted.

Children should be nursed in a children's ward under the care of an appropriately experienced consultant. Using the Glasgow Coma Scale in children under 5 years of age requires great care and paediatric experience, and clinical features such as a tense fontanelle are as important as skull fracture in deciding on the need for a CT scan.

Children under 3 years of age are especially hard to evaluate, and there should be a low threshold for consulting a specialist paediatric surgery unit.

Transfer of a child to a neurosurgical unit should be done by a team experienced in the transfer of critically ill children. Children suffering significant head injury should be followed up by a specialist multidisciplinary team. More research in this area is needed.


Like every other SIGN guideline, this one is intended to help individual clinicians manage individual patients. However, it would be foolish to ignore the rapidly changing landscape around it.

For example, a report from the Royal College of Surgeons of England in 1999 recommended that in future, head injury care should be delivered by A&E and neurosurgery, with general and orthopaedic surgery involved only in cases of multisystem trauma.

Such a policy would, of course, require a substantial shift in resources (and in reality increased resources) over the next 10 years, as well as expanded services for follow-up and rehabilitation after serious head injury.

The future shape of head injury services in the UK will be decided in the boardrooms of health departments and royal colleges, not in the resuscitation room or the operating theatre, but at least the new SIGN guideline may ensure that some of the decisions that are taken have some grounding in evidence.

  • Copies of Early Management of Patients with a Head Injury are available from the SIGN Secretariat, Royal College of Physicians, 9 Queen Street, Edinburgh EH2 1JQ (tel. 0131 225 7324). They are free of charge within the NHS in Scotland; elsewhere the cost is £7.50 per copy plus £1.00 postage. The guidelines can also be downloaded free of charge from the SIGN website at

Guidelines in Practice, February 2001, Volume 4(2)
© 2001 MGP Ltd
further information
| subscribe