head injury - journal of neurology, neurosurgery, and ... · head injury is now a patient with a...

56ournalofNeurology, Neurosurgery, and Psychiatry 1995;58:526-539

NEUROLOGICAL MANAGEMENT

Head injury

Graham M Teasdale

The problems posed by head injuries are vast,varied, and vexed. One million patientsattend hospital in the United Kingdom everyyear; they present a wide range of types andseverities of injury and sequelae; and there ismuch controversy, particularly between dif-ferent specialties, about the appropriate pro-cedures for management from acute to latestages. Head injuries are a major health prob-lem because of their peak occurrence inyoung adult men; they account for manyyears of potential loss of life up to the age of65 years, and for many people with lifelongdisability. There are estimated to be as manypeople neurologically disabled due to headinjury as due to stroke. No single approach tomanagement can cater for the needs of allpatients and their families, but a reasonable,rational approach follows from a considera-tion of the nature of head injuries and theirconsequences, of the scale of the problem, ofthe methods of treatment, and of the person-nel and facilities available.

DefinitionA practical operational definition, used insurveys in Scotland, incorporates a definitehistory of a blow to the head, a laceration ofthe scalp or head, or altered consciousness no

matter how brief.' Unfortunately, in theInternational Classification of Diseases, thereis no single code for head injury, which iscovered by up to 10 rubrics that are notmutually exclusive and that are related topathological rather than clinical features. Thishas greatly limited the collection of reliablestatistics, except as part of special surveys,but the 10th edition does include an assess-

ment of the duration of unconsciousness.

13% in those attending hospital but accountfor a third of patients transferred to neuro-surgery and 58% of deaths. Less than oneadult in five and less than one child in 10 isadmitted to hospital, an overall rate inScotland of 313 per 100 000 per year.' For allages, the death rate from head injury in theUnited Kingdom is nine per 100 000 peryear; this accounts for 1% of all deaths butfor 15%-20% between the ages of 5 and 35.Death rates from head injury are alreadydeclining in road users as a reflection of exist-ing preventive measures and further reduc-tions should follow the increasing use of airbags.

Traumatic brain damageBrain damage after head injury can be classi-fied by pattern and by time course. The pat-terns of damage recognised by pathologistsand, increasingly by imaging in life, are essen-tially separated into focal and diffuse varieties(table 1). It must be accepted that in manypatients the most accurate description may beof a multifocal distribution-for example,multiple cortical contusions or multipleischaemic lesions. In the time course, the dif-ferentiation is between primary damage-developing at the moment of impact-andsecondary damage, due to the subsequentcomplications, which may be intracranial or

systemic insults (table 2). Classification canalso be based on mechanisms of injury-forexample, missile v non-missile-and onwhether or not there is a compound fracture,and an open or closed injury.Diffuse axonal injury-is the single mostimportant lesion in traumatic brain damage.4It is thought to be responsible for the extent

Department ofNeurosurgery,Institute ofNeurological Sciences,Southern GeneralHospital, Glasgow G514TF, ScotlandG M Teasdale

EpidemiologyThe best guide to the incidence of head injuryis the number of patients presenting to a hos-pital after injury; in Scotland 1976 per100 000 per year,2 in the United Kingdom a

total of nearly one million per year.3 Almosthalf of these are children less than 15 yearsold and males outnumber females by morethan two to one. Most injuries are due to a

fall (41%), followed by an assault (20%); theimportance of road traffic accidents increaseswith the severity of injury, they cause only

Table 1 Lesions causingfocal and diffuse patterns ofdamage after head injuryFocal Diffuse

Contusion Axonal injuryHaematoma: Hypoxia/ischaemia

ExtraduralSubduralIntracerebral

Swelling Diffuse vascularInfarct Fat embolismPressure necrosis SubarachnoidhaemorrhageAbscess Meningitis

526

on June 26, 2020 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.58.5.526 on 1 May 1995. D

ownloaded from

http://jnnp.bmj.com/

Head injury

Table 2 Complications after head injury that causesecondary results to the damaged brain

Intracranial Systemic

Haematoma HypoxiaSwelling HypercarbiaRaised intracranial pressure HypotensionVasospasm Severe hypocarbiaInfection FeverEpilepsy AnaemiaHydrocephalus Hyponatraemia

of the impairment of consciousness in theacute stage and to account for much of thedisability experienced by survivors in the laterstages after all types of injury.5 It consists ofscattered damage and division of axons

throughout the white matter of the brain.Injury to individual axons can be recognisedonly by microscopy on fatal cases-silverstains show "retraction balls", which repre-sent swollen blobs of axoplasm. These lesionsare distributed centripetally and with increas-ing injury extend from the subcortical whitematter into the centrum semiovale, internalcapsule, and brain stem. In more severe

cases, they are accompanied by haemorrhagefrom small macroscopic tissue tears. Theseare located typically in the parasagittal sub-cortical white matter-previously called a

gliding contusion-the corpus callosum, thesuperior cerebellar peduncle, and the dorso-lateral aspect of the brain stem. These lesionscan be recognised on the cut surface of thebrain in fatal cases, and are now beingdetected in many patients in life by CT orMRI.6 7

Ischaemic brain damage is by far the mostcommon secondary insult8 and is still foundin more than 80% of fatal cases, despite mod-ern intensive management.9 The frequency ofischaemic damage is contributed to byimpairment, as a consequence of injury, ofthe normal regulating mechanisms by whichcerebral vascular responses maintain an ade-quate supply of oxygen.' 11 The frequency ofsecondary ischaemic insults, particularly inpatients with other injuries, has been high-lighted by recent findings made with analysisof continuous monitoring.'2-'4 In a series ofpatients with varying severity of head injury,92% were found to have one or more insultslasting for at least five minutes, despite beingin a well equipped and staffed intensive care

unit.IPimanry and secondary traumatic brain

damage are becoming less easy to separate.Thus it is recognised that axonal injury, once

thought to occur at the moment of impactand be irreversible, may in fact evolve from a

partial injury, in continuity, to complete dis-ruption over some hours."5 The sequence

includes unfolding of the axolemma, loss ofmembrane properties, damage to the cyto-skeleton, and interruption of axoplasmic flowleading to local swelling and then disruption.Also, secondary damage from insults such as

hypoxia may occur within minutes, beforeeven paramedical roadside attention, andmerge with the damage resulting from the

biomechanical forces acting at the moment ofinjury. Nevertheless, the distinction is still auseful clinical concept and underlines theimportance of focusing management on theavoidance or reversal of secondary events.

Clinically the processes of primary and sec-ondary damage are reflected in three principlepatterns of evolution, each with implicationsfor management: (1) The patient loses con-sciousness or develops other neurological fea-tures at the time of injury, but improves astime passes; this correlates with damage thatis principally primary from which naturalrecovery is taking place. (2) The patient doesnot lose consciousness at the moment ofinjury but then deteriorates, or having lostconsciousness then begins to worsen; each ofthese signals the development of secondarydamage and demands immediate action. (3)Features of brain damage develop at themoment of or soon after injury and persistwithout change: such a patient may go on toshow natural recovery but also is at increasedrisk of secondary complications.Many of the issues in early treatment of

head injuries concern the appropriateapproach to investigation and management ofthese cases. The issues facing the clinician,therefore, are how severely is the patientalready injured, and what are the risks offuture deterioration and increased damage?

Classification: severityMuch of the confusion, scientific, clinical, aswell as medicolegal, that clouds discussionand fuels controversy about head injuries canbe traced to variations and discrepanciesbetween different approaches to classificationof severity of injuries. It is, therefore, impor-tant to discuss the purposes of classificationand the approaches that are used; clarificationof the confusion and the adoption generallyof a coherent consistent approach should leadto more fruitful discussion and agreement.The first purpose of classification of sever-

ity is in management in the acute stage: thecritical factors are the patient's condition onarrival at hospital, how this is evolving, andwhat complications can be expected. The sec-ond is the potential for recovery, after initialassessment and acute management have beencompleted-when the ongoing assessment ofthe depth and duration of neurologicalimpairment is of primary interest. The thirdconcerns the inter-relation between the injuryand late sequelae-here the total quantum ofinjury is important, both initial and due tosubsequent complications; and early severityis often assessed retrospectively-for exam-ple, by duration of amnesia; this is particu-larly relevant to medicolegal issues. Thedifference in perceptions between those whohave seen the patient at the acute stag- acci-dent and emergency consultants, general andorthopaedic surgeons, neurosurgeons-andthose who usually become involved only laterin the assessment of sequelae-neurologists,psychologists, and psychiatrists-reflect thesevarying standpoints.

527


http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from


Teasdale

Table 3 Glasgow coma scale, coma score, and modifications for children underfive yearsold

In adults (score in normal adults is 15)Eye opening response:

SpontaneouslyTo speechTo painNone

Best motor response (in arms):Obeys commandsLocalisation to painful stimuliNormal flexion to painful stimuliSpastic flexion to painful stimuliExtension to painful stimuliNone

Best verbal response:OrientedConfusedInappropriate wordsIncomprehensible soundsNone

4321

654321

54321

Modifications of nortnal response in children under 5Best motor Best verbalresponse responseFlexion Smiles and criesLocalisation Smiles and criesLocalisation Sounds and wordsObeys commands Words and phrases

COMA, CONCUSSION, AMNESIAChanges in consciousness provide the basis ofmost approaches to the classification of sever-

ity'6; this reflects the importance of diffuseaxonal injury in the initial events and in caus-

ing later sequelae. The Glasgow coma scale(table 3) separately assesses eye, verbal, andmotor performance. 17 This separation, appro-

priate conceptually because each may changeindependently, and very convenient in prac-

tice, may have contributed to the wide accep-

tance of the Glasgow approach. Thetemptation, however, to summate the scores

of the different components into an overallcoma score ranging from 3-15 could not beresisted'8 '1 and the total "coma score" now

provides the most widely used basis for classi-fication (table 4).20 Nevertheless, its use needscritical review and some redefinition may benecessary, particularly in less severe injuries,21even if this is at the price of some initial con-

troversy.The most widely used definition for severe

head injury is now a patient with a Glasgowcoma score of 3-8. Originally, the definitionused in the international studies coordinatedfrom Glasgow,22 23 was that the patient was ina coma for six hours, coma being defined as

no eye opening, no comprehensible verbalresponse, and not obeying commands.24 Insome 80% of cases the notation for coma

translates into a coma score of 8 or less,hence the adoption of the score. The six hour

duration has become difficult to apply as aresult of severe head injuries now almost uni-formly being sedated, intubated, and venti-lated and are hence unassessable for manyhours, and initial severity is usually assessedby the findings on admission.

Moderate head injury was defined byRimmel et al as a patient with a coma score of9-12.25 This group until recently did notreceive as much attention as either the severeor lesser injuries. The group may be difficultto identify consistently and the definitionneeds scrutiny before much further work iscarried out.26The most unsatisfactory definition is of a

mild or even minor head injury as a patientwith a Glasgow coma score of 13-15.27 Theproblem is that patients with a coma score of15 make up by far the overwhelming numberof patients classified in this group.228 In prac-tice, a patient with a coma score of 15, com-pared with those with scores of 13 or 14, hasa much lower risk of complications at theacute stage2930 and fewer and less persistentsubsequent sequelae. The inclusion withinthe same category of all patients with a comascore of 13-15 underestimates the true sever-ity of the injury in patients with scores of 13or 14. It also gives an impression of undueseriousness to those with a coma score of 15.It is more appropriate to separate out patientswith a coma score of 15 and refer to these ashaving had a minor injury.

Description of severity in later stages isbased on the duration of alteration in con-sciousness-either of observed coma or ofamnesia."3 The duration of amnesia after theinjury-post-traumatic amnesia-is a widelyaccepted index. It may be difficult to estimateprecisely and is best regarded as a logarithmicscale: very mild, less than five minutes; mild,five to 60 minutes; moderate, one to 24hours; severe, one to seven days; very severe,one to four weeks; extremely severe, morethan four weeks. 16 32The classification of severity based solely

on changes of consciousness may sometimesoverlook the importance of focal injury.Computed tomography and MRI show thatcortical contusions can occur in the absenceof prolonged unconsciousness but lead toprolonged confusion and sequelae such asmemory impairment and epilepsy.33

PreventionPrevention is possible at three stages: fore-stalling the accident; minimising the degree of

Table 4 Classification ofhead injuries by the Glasgow coma score into severe, moderate, mild, and minor

GCS on Cases (%) Multiple Risk ofICHArival Attenders Admissions Injury (%/) No fracture Fracture Dead (%)

Minor 15 95 42 32 1 1:10 000 1:100 <1Mild 13/14 1 38 J 32 l 3-5Moderate 9-12 4 13 371:3 1:15 9Severe 3-8 1 7 63 1:50 1:8 35-40

ICH = Intracranial haematoma; GCS = Glasgow coma score.

Age

<6 months6-12 months1-2 years2-5 years

52.8


http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from


Head injury

injury occurring on impact; and reducing therisk of secondary complications-the focus ofmedical management in the acute stage.Accident prevention requires modification ofbehaviour by the public and is effective usu-ally only when enforced by legislation. Theintroduction of speed limits, the use of safetybelts by vehicle occupants, and the wearing ofhelmets by motor cyclists have all provedeffective. More stringent limits on the alcohollevel allowed in drivers, and the universal useof air bags, with rigorous enforcement couldfurther contribute to a reduction in injuriesdue to road accidents. This would leave alco-hol still a major contributor to injuries fromassaults and falls and in pedestrian victims ofroad accidents. There is increasing evidencethat the wearing of helmets by cyclists pre-vents injuries, but this remains to be backedup by legislation. The dangers of brain dam-age from boxing are well recognised.34 Whatdoctors should do is to emphasise the inade-quacy of current prefight medical examina-tions in minimising the risk and to point tothe long term dangers, highlighted by theincreasing evidence of a biological connectionbetween head injury and dementia.35

DiagnosisTwo questions need to be answered in everysuspected head injury: is it a head injury, andis it only a head injury? There is little doubtabout the occurrence of a head injury when aclear history is available from either the vic-tim or an onlooker. Difficulties arise in theperson presenting with impaired conscious-ness of unknown onset and duration, espe-cially when there is evidence of alcoholintake. There is compelling evidence thatwhen in doubt it is safer to regard the victimas having a head injury than to attributeimpaired consciousness to alcohol ingestionor, in the old person with focal signs, to theeffects of a stroke.36 Confirmation of an injuryto the head may come from careful examina-tion clinically or from the result of a skullradiograph. Conversely, when a head injuredpatient has impaired consciousness, there is atemptation to focus too much attention to thehead and to overlook important injuries else-where.37 The initial clinical examinationshould note carefully any abnormal neurolog-ical symptoms or signs as a reference pointfor comparison with subsequent examinationsand interviews. The niceties of the compre-hensive neurological examination are, how-ever, of considerably less relevance thanregular reliable assessments of consciousness.

ManagementThe essence of management of head injury isthe provision of optimum circumstances forrecovery from damage already sustained-principally primary damage-and the avoid-ance of the development or exacerbation ofdamage due to complications-principallysecondary damage. In the acute stage survivaland natural recovery can be expected in most

patients with minor, mild, or moderate injuryand hence the focus in these is on the identifi-cation of patients at risk of secondary compli-cations-principally a traumatic haematoma.The patient with a severe head injury who isin a coma has both evidence of already havingsustained a substantial amount of brain dam-age and also a much greater risk of bothintracranial and extracranial complications.

Assessment, diagnosis, investigation,observation, monitoring, treatment, and reha-bilitation each have a crucial part in the man-agement of head injury. The diversity ofinjuries and variations in resources availablemean that there is not a single approach thatis optimum for all victims. On the otherhand, attempts to tailor management to eachindividual patient, based on a process ofdeduction and deliberation, does not provideeffective care for head injuries. Instead, aswith all trauma, an approach based on aseries of recommendations, criteria, or guide-lines, is both more efficient and effective.These enable an approach that is consistentbetween cases and between centres; theyreduce confusion and enhance communica-tion and improve outcome.'839 Widelyaccepted approaches to head injury manage-ment have been recently reviewed.">4

ASSESSMENTThe approach to assessment varies with theperceived severity of injury. When the patienthas impaired consciousness, assessment andresuscitation must follow the principles of lifesupport, as it is taught in the AdvancedTrauma Life Support (ATLS) system." Theidentification and correction of an obstructedairway, of inadequate ventilation, or shockmust take priority over the detailed assess-ment of the patient's neurological state.43

Assessment can begin at the roadside andambulance staff can now score patients onthe Glasgow coma scale and report the levelof blood pressure and heart rate.Unfortunately, many patients still arrive atspecialised units either from the scene of anaccident or from another hospital withhypoxia, shock, or other factors worseningprognosis.45 On arrival at hospital, assessmentand resuscitation must be completed beforethe patient is moved for further investigationor treatment. The temptation to focus on thehead and carry out premature CT or otherinvestigations must be resisted in favour of aproper, thorough general survey and manage-ment.46

Assessment of the patient's neurologicalstate is quickly and effectively carried outwith the Glasgow coma scale for overall con-sciousness, noting any side to side differencesin limb movement to detect hemiparesis orother focal neurological deficit and examiningthe pupil size and response to light.18 The eyeopening, verbal, and motor responses of theGlasgow coma scale are well known and canbe modified for application to children underfive years old (table 3).4 Neurological assess-ment should be repeated and documentedoften-for example, at least every 10 minutes

529


http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from


Teasdale

Table S Indications for CT and referral ofpatients withTeasdale, et al30)

Indications for referral to neurosurgiWithout preliminary computed tomography:Coma persisting after resuscitationDeteriorating consciousness or progressive focal neurologiOpen injury: depressed fracture of vault or basal skull fractPatients fulfils criteria for CT in a general hospital when tI

reasonable time-for example, three to four hoursAfter computed tomography in general hospital:

Abnormal tomogram (after neurosurgical opinion on imagTomogram considered to be normal but patient's progress

Indications for computed tomography in geFull consciousness but with a skull fractureConfusion persisting after initial assessment and resuscitatUnstable systemic state precluding transfer to neurosurgerDiagnosis uncertain

during the first hcaccording to progrvalid guide to the ex

when the patient iand has a normal bloration is a signal t(as hypoxia, hypohaematoma.

INVESTIGATION: RADComputed tomogr;value rapidly in thecomplications afteradvantages in diagiment in the outcomonly when the availtpolicies aimed at en

earlier stage, preferaof neurological deteonance imagingparenchymal abnorravailability and prastill the mainstayhead injury, where icome is most clear.The issue that C]

and that continuestriage of patients; w:

fq~ ~ ~

-1: 1 0f^)

0.2

'~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~......

*i10000

CsciutJsnrefs rlv pri re- or-'"F raCt'L] N, N(J N o IN!t

Figure 1 Effect of age, skullfracture, and impaired consthe risk ofa traumatic intracranial haematoma (data fro;

recent head injury (from between the number of CT scans that it isfeasible to carry out and the likelihood that

ial unit the investigation will contribute to manage-ment?

.cal deficits Guidelines for selection of patients for CTture were first promulgated a decade ago, whenhis cannot be performed within a scanners were largely restricted to neuro-

surgical units.5051 More recently the increas-es transferred electronically) ing availability of CT scanners in general;is unsatisfactory

hospitals has led to a reappraisal and widen-meral hospitals ing of the criteria for scanning (table 5).3°ion Unfortunately, the opportunity provided byy the availability of CT52 in many hospitals is

often not turned to the advantage of patientswith head injury because access to the scan-ner is limited to the normal working day,whereas most injuries occur at nights and atweekends. When this is the case, neurologists

iur and then afterwards and neurosurgeons should press for the estab-ess. The findings are a lishment of an "out of hours service" with, if,tent of brain damage only necessary, image transfer for consultationis adequately oxygenated with the neurosurgical unit. Without anlood pressure. Any deteri- emergency service, the restricted availability) seek complications such of the CT scanner in a general hospital cantension, or intracranial lead to inappropriate delay before the patient

is investigated and a delay in the diagnosis ofremediable intracranial complications.53

XIOLOGY It is neither feasible nor desirable that allaphy proved its clinical head injuries should undergo CT. Instead,diagnosis of intracranial there is now evidence from several studieshead injury.48 Despite its that the factors that identify the likelihood ofnosis, however, improve- a patient having either abnormal CT or aLe of head injury occurred remediable intracranial lesion can be deducedability of CT was allied to from clinical features.'03'54 The key factorsisuring investigation at an are the depth and duration of alteration ofLbly before the occurrence consciousness, the result of a skull radiographrioration.38 Magnetic res- (fig 1) and, in a few cases, the presence ofis more sensitive to focal neurological signs. Whereas in the past,malities,749 but the greater when scanning required neurosurgical trans-cticability of CT make it fer, this could be advocated only in patientsof acute investigation of with both impaired consciousness and theLts value in improving out- skull fracture-in whom the risk of

haematoma was as high as one in four,5" it isrbrought into clear focus, now more reasonable that all patients withto be controversial, is the persisting impairment of consciousness afterhat is the optimum match arrival at hospital are considered for CT (fig

2).3° In the patient in a coma, this should fol-low transfer to the regional neurosurgicalunit; in the patient with a coma scorebetween 9 and 14 whose condition does notreturn to normal within one to two hours ofinjury, scanning should be carried out locally.

Skull radiographs can be omitted from ini-tial assessment if CT is to be carried out. Askull radiograph, however, retains an impor-tant place in the investigation of patients whoare fully conscious, in whom the finding of askull fracture raises the risk of intracranialcomplications by more than 200-fold.Computed tomography should be performedif a skull fracture is present. The use of CT inall cases is unjustified, because of the greaterradiation exposure (some twofold) and thegreater cost (twofold to fourfold) comparedwith a skull radiograph.

Skull radiographs still have value in theivsD7. " 'YeftCA,.*s 's,-detection of fluid levels in the sphenoid sinus,

intracranial air, or depressed skull fracture,sciousness and coma on increasing each of which signals an open injury and them Teasdale et al 0). risk of intracranial infection. Indications for

530


http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from


Head injury

Figure 2 Flow chartformanagement ofpatientswith head injury accordingto level of consciousness,skull radiograph, and CTin district general hospitalsand regional neurosurgicalunits 6(rom Teasdaleet aP0).

Patient fullyconscious

No indicatiorfor radiograph

Observationat home

Patient with impaired consciousness orneurological signs

I Radiologyiy of skull

Negative FPositive

Urgent CT

Negative Positive

Observe in hospital untilfully recovered

Patient in coma or withdeteriorating consciousness

Resuscitate

Refer toneurosurgical unit

skull radiographs include high or mediumvelocity impact with a broad hard surface55and association with other features such aspost-traumatic amnesia, leakage of CSF, or

bleeding from nose or ears, a large scalphaematoma, or laceration. In the unconsciouspatient, radiographs should also include filmsof the cervical spine, chest, and of any areas

suspected of associated fractures.

CLINICAL OBSERVATION

The twin purposes of observation and moni-toring are the determination of the pattern ofevolution of the patient's neurological stateand the detection of complications. Theintensity of observation and monitoring isdetermined by the extent of any existing braindamage and the perceived risk of deterioration.

For the patient who is conscious and whodoes not have a skull fracture, the decision iswhether or not to admit to hospital for obser-vation. At least four out of five patients are

discharged for observation at home and theproportion of patients admitted can bereduced further without adverse effects.5657Every accident and emergency department

Table 6 Indications for admission for observation ofhead injuries

The folowing list of indications for admission ofpatients with head injuries should be displayed in acci-dent and emergency departments

Confusion or any other depression of consciousness at the time of the examinationSkull fractureNeurological symptoms, or signs, or bothDifficulty in assessing the patient-for example, because of ingestion of alcohol, epilepsy, or

other medical conditions that cloud consciousness. Children are also difficult to assessLack of a responsible adult to supervise the patient and other social problemsBrief amnesia after trauma with full recovery is not necessarily an indication for admissionIf the patient is to be observed outside hospital he or she should be discharged with a head

injury "warning card" into the care of a responsible person

dealing with head injuries should have anestablished policy, clearly displayed andwidely known. Table 6 shows the acceptedindications for admission. Patients who areconscious without any of these features canbe discharged for observation under the careof a relative or responsible adult who shouldbe given a list of instructions in the form of a"warning card" about what to observe andwhat to do. In such low risk cases, admissionto hospital confers no advantage when ahaematoma does develop.58

In the patient who has impaired conscious-ness but is still talking, assessments of con-sciousness should be repeated every one totwo hours. Cardiovascular respiratory andother "vital functions" can be assessed bycustomary clinical methods with recordingsmade intermittently. The more severe theimpairment of consciousness, however, themore likely it is that these traditional methodswill fail to disclose important deviations fromvalues that ensure satisfactory cerebralperfusion and oxygenation and the greaterthe indication for continuous and invasivemethods.

Table 7 Monitoring techniques and investigations used inpatients with severe head injury (from Della Carte59)Routine Selective

Glasgow coma score Intracranial pressurePupils Cerebral perfusion pressureTemperature Central venous/pulmonary arteryArterial 02 saturation pressureEnd tidal CO2 ElectroencephalogramArterial blood pressure Jugular bulb 02 saturationElectrolytes/haematology Transcranial Doppler

CT: on admission; postoperative(24h-72h);if intracranialpressure is raised; if clinicaldeterioration occurs

531


http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from


532

MONITORINGDetailed continuous monitoring is needed inall patients who are not obeying commandsor who have a head injury and other seriousinjuries (table 7). Arterial pressure should bemonitored with indwelling arterial cannula toprovide a continuous record. This will pro-vide access for intermittent measurements ofblood gases as indicated by the results of con-tinuous monitoring of arterial oxygen satura-tion using pulse oximetry. An ECG should bemonitored continuously and monitoring ofcentral venous pressure or pulmonary arterypressure may be indicated when shock hasoccurred or is suspected. In a ventilatedpatient, the measurement of end tidal CO,with a capnograph is a useful check to theadequacy of ventilation, supplemented by thecontrols inbuilt into modem ventilator systems.The role of monitoring intracranial pres-

sure in head injuries is still controversial.Although it is not employed in all units deal-ing with severe head injuries, it has at leasttwo clear benefits.42 The first is in providingan indication of severity of space occupyingeffects from a focal intracranial lesion. Thesecond, when coupled with arterial pressuremeasurements, is in calculating cerebral per-fusion pressure, the critical determinant ofoverall cerebral blood flow. All techniques areto a degree invasive, however, and have a riskof intracranial infection of 2%-8% and ofcausing intracranial haemorrhage (< 1%) andepilepsy (1%).The value of monitoring intracranial pres-

sure in the management of a patient with anintracranial clot became established when itbecame apparent that CT was showing manymore focal lesions than could be expected torequire evacuation.60 Deciding as soon as pos-sible whether to evacuate a clot in a stablepatient minimises the risk of neurologicaldeterioration and improves outcome. CertainCT features show strong correlation with arise in intracranial pressure and by themselvesare indications for evacuation.4041 Such fea-tures are pronounced midline shift (1 cm),loss of visualisation of the third ventricle andperimesencephalic cisterns, and dilatation ofthe ventricle contralateral to the lesion, par-ticularly the temporal horn. When CT fea-tures leave doubt about the need foroperation, continuous monitoring of intracra-nial pressure should be instituted and thelesion evacuated if pressure is sustained above20-25 mm Hg. If the pressure remains belowthis level for 24 hours, then the likelihood ofdeterioration is very small. Intracranial pres-sure should be monitored after evacuation ofan intracranial clot to provide an early warn-ing of the development of a recurrent or newhaematoma and because after evacuation of asubdural or intracerebral clot brain swellingand raised pressure are frequent.

Intracranial pressure can be monitored byinsertion of a fluid filled catheter into the ven-tricle or, increasingly, by insertion of a solidstate fibre optic system into the CSF path-ways or direct into brain parenchyma. Thefirst usually requires a burr hole performed in

the operating theatre, whereas the second canbe carried out with a twist drill in the inten-sive care unit. Although the techniques formonitoring intracranial pressure are not nec-essarily restricted to neurosurgical intensivecare, just as an overconcern about the headinjury in a comatose patient with multipleinjuries can be dangerous, so too can be anovernarrow focus on observing and treatingintracranial pressure. Unless monitoring isaccompanied by the knowledge and experi-ence necessary to appreciate the importanceof first considering the cause of raised pres-sure before any treatment, and of the need foreasy access to and use of repeat CT and otherinvestigations needed to interpret intracranialpressure findings, benefits from monitoring itmay be outweighed by adverse effects. Thusit is inappropriate and harmful to attempt tolower intracranial pressure by medical mea-sures when its primary cause is an expandingintracranial lesion requiring evacuation. It isalso crucial to be aware that an apparentlysatisfactory reduction of intracranial pres-sure for instance, by hyperventilation orbarbiturate treatment - may be at the price ofactually inducing cerebral ischaemia as aresult of vasoconstriction,36' or net loweringof cerebral perfusion pressure by concomitanthypotension62 and result in a worse out-come.63

Continuous monitoring of jugular venousoxygen saturation can provide valuable infor-mation when treating raised intracranial pres-sure.64 A decrease in venous bloodoxygenation indicates increased extractioneither due to reduction in blood flow or dueto an increase in metabolism. Conversely,very high levels of venous oxygenation indi-cate cerebral hyperaemia. Jugular oxygen sat-uration can be measured continuously withindwelling fibre optic catheters but, at thepresent stage of development, readings haveto be interpreted with considerable cautionbecause abnormal values can result fromtechnical factors. The catheter should berecalibrated at least every 12 hours againstthe findings of a co-oximeter on withdrawnblood samples and these should be repeatedto check any apparently abnormal values. Inthe presence of arterial hypoxaemia andanaemia, it is probably preferable to deter-mine the absolute content of blood and hencethe cerebral oxygen extraction rather thanrely on saturation values. The precise valuesof venous saturation that are optimum arestill being debated; values below 50%-60%indicate excessive extraction and potential forischaemia, at least regional, and values of85% indicate hyperaemia.

Cerebral blood flow can be determinedintermittently from a variety of techniques,but none of these has as yet found a place inroutine management. An index of cerebralblood flow is provided by the velocity ofblood flow in the intracranial arteries and thiscan be measured intermittently or continu-ously, typically in the middle cerebral artery,using transcranial Doppler sonography.65This non-invasive technique also finds

Teasdale


http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from


Head injury

application in the care of patients with sub-arachnoid haemorrhage and is becomingincreasingly available in neurosurgical units.Changes in mean velocity or in indices of pul-satility-the difference between systolic anddiastolic flow velocities-aid the interpreta-tion of information from intracranial pressuremeasurements and cerebral oxygen extractionvalues. Velocity falls and pulsatility increaseswith reducing cerebral perfusion pressure andblood flow. A high velocity can be due tohyperaemia or narrowing of the vessels due totraumatic cerebral vasospasm.

Studies of cerebral electrical activity canprovide information about brain function inpatients who are unconscious either due tothe head injury or because of pharmacologicaltreatment. The potentials evoked bysomatosensory stimulation provide a usefulindex of integrity and are prognostically use-ful but have not established a value in practi-cal care. When neuromuscular paralysis isemployed to permit ventilatory treatment,however, there are advantages in continuousmonitoring of cerebral activity using a simpli-fied device such as a cerebral function monitor.

Monitoring with which staff are not famil-iar or which produces technically capriciousresults is useless and even dangerous. Theincreasing complexity of monitoring used inthe management of patients with serious headinjury is a strong argument for concentratingsuch cases, and hence experience, in regionalneurosurgical centres, where the expertise tocarry out the measurements and interpret andact on their findings can be developed andsustained.

Management oftraumatic brain damageThe early scepticism, if not pessimism, aboutthe prospects for recovery of a patient whoremained in a coma, despite effective earlyresuscitation, has been dispelled by the evi-dence from many sources that half or moresuch victims can recover and make an inde-pendent recovery. The prospects for improv-ing the outcome of such injuries have beenconsiderably heightened by evidence frommodem methods of monitoring of the occur-rence of secondary insults, likely to exacer-bate brain damage, that were not detected bymethods available a decade or more ago.'2Such insults worsen outcome" opening the

Table 8 Indications for intubation and ventilation ofpatients with recent head injury(from Gentleman et al43)Immediately:Coma (not obeying, not speaking, not eye opening)-that is, Glasgow coma score < 8Loss of protective laryngeal reflexesVentilatory insufficiency (as judged by blood gases):

Hypoxaemia (Pao2 < 9 kPa on air or < 13 kPa on oxygen)Hypercarbia (Paco2 > 6 kPa)

Spontaneous hyperventilation causing Paco2 < 3-5 kPaRespiratory arrhythmia

Before transport within or between hospitals:Significantly deteriorating conscious level, even if not in a comaBilaterally fractured mandibleCopious bleeding into mouth (for example, from skull base fracture)Seizures

An intubated patient must also be ventilated; aim for Pao2 > 15 kPa; Paco2 4 0-45 kPa

Pao2 = arterial oxygen tension; Paco2 = arterial co2 tension.

way to improving an outcome by preventionof secondary insults or minimisation of theirconsequences. Also, the greater understand-ing of the pathophysiology of injury that hascome from animal experiments and clinicalobservation in the past decade has enabledthe development of more rational policies formanagement and more appropriate targetedtreatment in individual cases.66 This isencompassed within the concept of neuro-intensive care, distinguished in its conceptsand techniques from general intensive care.No longer is it satisfactory for patients withhead injury to be sedated, paralysed, and ven-tilated in the absence of appropriate neuro-monitoring and investigation facilities.

NON-OPERATIVE MANAGEMENT OF SEVEREHEAD INJURYThis covers a range of techniques employedto prevent and treat complications consideredto be liable to produce secondary damage.Many of these complications are systemic-for example, hypotension, hypoxia, hyper-capnia, hypothermia, electrolyte imbalance-rather than intracranial. The need for meticu-lous standards of management of severelyinjured or ill patients, and the methodsinvolved apply just as much to a serious headinjury. As a general principle, such systemicdisturbances are both more common andmore serious in their effects than intracranialdisturbances,67 even after the initial resuscita-tion and emergency measures have been car-ried out. 14 By contrast with the unanimityabout the importance of these factors, there isstill considerable variation of opinion aboutthe employment of methods primarily aimedat treating raised intracranial pressure andbrain swelling after head injury. This is partlybecause such methods have not been shownby a randomised control trial to substantiallyimprove outcome, and it is difficult to con-ceive of the feasibility of carrying out such atrial. Also, it reflects the fact that brainswelling and raised intracranial pressure maybe a consequence of brain damage, ratherthan a primary factor in producing damage.There is, nevertheless, an acceptance that thesupervention of raised intracranial pressure,reduction in cerebral perfusion pressure, andischaemia on an already damaged brain, withheightened vulnerability as a result of injury68cannot be other than undesirable.69

Ventilation of unconscious patients iswidely used (table 8). The aim should be tokeep arterial oxygen saturation as close aspossible to 100%, using increases in inspiredoxygen and positive end expired pressurewhen necessary. Ventilation should beadjusted to maintain arterial CO, tensionnormal or slightly subnormal. The practice ofequating ventilation of head injury withhyperventilation must be abandoned in viewof the evidence Qf the resulting cerebralhypoxia and impaired outcome that result."363Hyperventilation should be employed onlybriefly, not least because its effects are onlytemporary. Hypotension, whether as a conse-quence of hypovolaemia due to inadequate

533


http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from


Teasdale

fluid replacement, or the use of sedativedepressant drugs must be avoided.When raised intracranial pressure occurs

and simple causes such as neck position, air-way obstruction, abnormal breathing pat-terns, fever, and seizures, have been excludedand a surgically remediable intracranial lesionoccupying space has been ruled out, twoprincipal approaches to treatment areemployed. The first is the use of osmoticdiuretics such as mannitol with a view towithdrawing fluid from either the normalbrain or areas of brain oedema. The usualstarting dose is 0-5 g/kg body weight withadjustment as determined by the effects onintracranial pressure and cerebral perfusionpressure. Additional measures includefrusemide to sustain the osmotic gradient,infusion of colloid to maintain circulating vol-ume, and the avoidance of hyperosmolarity(serum osmolarity more than 320 mmol/l).The alternative is to use sedative or hypnoticdrugs such as propofol or thiopentone toreduce cerebral metabolism and hence inducea fall in blood flow and blood volume.

Mannitol is considered to be most effectivein raised pressure due to focal space occupy-ing lesions whereas sedatives are more appro-priate in patients with raised intracranialpressure due to vascular dilatation-typicallychildren with preserved cerebrovascular CO2activity and cerebral electrical activity.65 In allcircumstances, care must be taken to avoidhypotension and it is sometimes more appro-priate to maintain cerebral perfusion pressureby raising blood pressure pharmacologicallythan to strive to reduce intracranial pressure.

DRUG TREATMENT OF HEAD INJURY:NEUROPROTECTIONVarious agents have been used or are beingconsidered that aim to interfere with the mol-ecular, biochemical, cellular, and microvascu-lar processes involved in traumatic braininjury.70 None has yet been shown clearly tobe of benefit. This is particularly true ofsteroids. After many years of debate, severaltrials at various doses have failed to showbeneficial effects and even adverse conse-quences have been noted.71 This contrastswith the benefit of steroids in brain swellingdue to tumour and highlights the mechanisticdifferences in the processes.The increased understanding of traumatic

and ischaemic brain damage that has comefrom intense research in recent years haspointed to the importance of mechanismssuch as increased intracellular calcium, exci-totoxicity from excessive glutamate and otherexcitatory amino acids, and lipid peroxida-tion.72 There are promising indications fromexperimental studies of benefit from a rangeof neuroprotective drugs-calcium ion chan-nel antagonists, glutamate receptor blockingagents and antioxidants-and also hypother-mia. Clinical studies are underway orplanned.Most neuroprotective agents have side

effects on cardiovascular and CNS functionand the achievement of an acceptable efficacy

and safety margin is likely to depend on thepatient receiving effective neurocritical care.Certainly, this will be necessary in the trialsneeded to determine efficacy and the concen-tration of severely injured patients in appro-priate facilities should be encouraged.

ANTICONVULSANTSThe frequency of seizures (5%), both in theacute and late phase after head injury73 hasprompted the use of anticonvulsant drugs as aprophylactic measure rather than as aresponse to a declared epileptic event. It washoped that prophylactic treatment and thesuppression of seizure events would lead to areduced occurrence of late continuingseizures. Trials have shown that this is not thecase for such late epilepsy.74 75 One study,however, has shown a clear reduction inseizures but only in the first week afterinjury. The precise level of risk of seizuresthat merits treatment remains debated; likemost British neurosurgeons, I prefer to with-hold treatment until such time that a seizureoccurs.

ANTIBIOTICSThere is controversy also about the use ofantibiotics in the prophylaxis of infection inpatients with an open injury, particularly dueto a fracture at the base of the skull resultingin CSF rhinorrhea, or otorrhea, or intracra-nial air. One school of thought argues that theuse of broad spectrum antibiotics does notreduce infection and simply promotes theoccurrence of antibiotic resistant bacteria. Onthe other hand, I believe that there is reason-able recent evidence77 to support the long sus-pected value of prophylaxis with penicillin orallied agents, against pneumococcus, which isthe most common organism and which cancause explosive and irreversible deterioration.

OPERATIONS IN HEAD INJURYMost intracranial haematomas after a headinjury are intradural-subdural, intracerebral,or both. Effective operative management ofthese lesions demands more access than canbe achieved by a burr hole and requiresneurosurgical and neuroanaesthetic expertiseand facilities. With early referral for CT,there now should be very few, if any, occa-sions that a non-neurosurgeon needs to con-template intracranial surgery-even for thesimpler but much more rare solitaryextradural haematoma. When a haematoma isdetected or strongly suspected, a large bolusof mannitol-for example, 1 g/kg-can "buytime" for transfer to the neurosurgical operat-ing theatre.

Intracerebral contusions and smallerhaematomas can pose difficult decisions.Some surgeons favour a conservativeapproach: intracranial pressure is monitoredand if raised, especially if cerebral perfusionpressure is reduced, medical methods areused and operation performed only if thesefail. Although a conservative approach maybe reasonable initially, if intracranial pressureis raised and the CT shows a focal space

534


http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from


Head injury

occupying, mass effect, operation should bepreferred. Evacuation does not risk injury tosurrounding recoverable brain tissue-areasof low CT density adjacent to a contusionidentify cytotoxic oedema in irreversibly dam-aged brain; evacuation of a mass lesion willalways be a more secure method of improvingintracranial pressure-volume relations, andfatal herniation can occur from local shiftwhile intracranial pressure is being "con-trolled" medically.Open injuries are less urgent indications for

operation. Debridement and repair of a com-pound depressed skull fracture within 12-24hours of injury avoids or minimises infection.Repair of a basal skull fracture is usually post-poned until any associated leakage of CSF haspersisted for several days, at which time thefrontal lobe swelling, due to the contusionscustomary in such cases, has subsided. Thereis a trend for early (< 12 hours) operation incraniofacial injury, the optimum time for cor-recting deformity, but this should be avoidedif an associated head injury is any more thanminor or mild severity.

Where should head injuries be treated?Provision of care for head injuries needs toincorporate the assessment and arrangementfor home observation of the minor injuriesthat make up 80% of hospital attenders;arrangements for admission and observationof mild and moderate injuries; CT, eitherbefore or after neurosurgical referral andtransfer; and for the resuscitation, continuingcare and definitive management of patientswith a severe head injury or a head injurycombined with serious injuries elsewhere.78

Facilities needed for initial assessmentinclude staff trained and able to assess con-sciousness, to apply guidelines for skull radi-ography, to interpret its findings, and to applyguidelines for discharge or hospital admis-sion. These must be available at all timeswithin a district general hospital that dealswith trauma cases. The arrangements forcases needing admission to hospital willdepend on whether the hospital also containsa neurosurgical unit. If this is not so, it ispreferable for head injuries admitted forobservation to be grouped together in a spe-cific short stay or observation unit whereexpertise and experience can be maintainedin assessment and, where necessary, referralonwards; it is less satisfactory for patients tobe admitted for observation on acute surgicalwards, whether general surgical ororthopaedic.

Transfer of a patient both within a hospitaland from a district hospital to a neurosurgicalunit is fraught with hazard.1279 The distancetravelled is less important than the risk ofchanges in homeostasis developing insidi-ously, unnoticed and untreated, because ofinadequate monitoring, equipment, or inex-perienced escort.45 8>82 Dedicated transferteams, travelling from major centres to collectseverely ill patients have been shown to be ofvalue83 but most often the responsibility will

lie with the staff of the referring hospital.Before transfer, the patient must be renderedstable; life threatening extracranial injuriesdemand priority.84 If there is any concernabout airway or oxygenation, intubation andventilation should be established. The staffaccompanying the patient should be experi-enced in the care of the unconscious injuredpatient. The minimum is a doctor, preferablyan anaesthetist or a doctor with anaesthetictraining and experience, and a trained nurseor paramedic. They must be familiar with thepatient's condition before the journey, withwhat can go wrong during transport, and withthe procedures and equipment needed.During transport there must be reliable intra-venous access, the ECG should be monitoredcontinuously as should the blood pressureand arterial oxygen saturation by pulseoximetry. The monitoring equipment shouldpreferably have the ability to display trends,to store data, and to print hard copy for lateranalysis. Facilities must be available to re-position or replace an endotracheal tube tocontinue ventilation if the oxygen supply fails,to continue or replace intravenous treatment,and to deal with cardiac arrest. With anappropriately equipped ambulance and ade-quately experienced staff, road transport willbe appropriate for most transfers, air transferbeing needed only for inaccessible locationsor for very long distances.85The preferred location for the definitive

and continued management of the severelyinjured patient is the regional neurosurgicalcentre, and neurosurgical intervention ishighly cost effective.86 Although district gen-eral hospitals can increasingly undertake CT,and provide general intensive care, the temp-tation to retain severely injured patients inwhom CT seems not to show a surgical com-plication should be resisted. As stressedabove, the expertise needed to interpret CTscans, to decide on the need for repeat scan-ning, to carry out and interpret comprehen-sive "neuromonitoring", and to apply itsresults, requires the build up and continua-tion of expertise; experience depends Cn acase load of sufficient numbers that is likelyto be found only in a regional centre. Actingagainst this are the factors, professional pres-tige and market forces, that lead to pressuresnot to transfer such patients for neurosurgicalcare but to manage them in general intensivecare units in a general hospital. This limitsthe employment of comprehensive neurocriti-cal care and also divorces the victim's familyfrom the informed advice, counselling, sup-port, and follow up that should be availablein a specialised unit.The merits of "trauma centres" are cur-

rently under debate and investigation.Proponents emphasise the benefits of earlymultidisciplinary intervention in patients withserious multiple injuries. Others point to thevery small proportion of accident victimsrequiring such intervention and the difficul-ties in selecting out such patients, in arrang-ing for their transport to a dedicated singlecentre, and in the centre being adequately

535


http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from


Teasdale

staffed at all times, when appropriate casesmay be sporadic and unpredictable. Whateverthe cause of multiple injuries, an injury to thehead has the greatest influence on outcome.87When a trauma centre exists or is being

developed in a hospital in which there is alsoa neurosurgical unit, neurosurgical adviceand intervention should be readily availablebut should not take priority over the diagnosisand management of serious systemic injuries,hypotension in particular being an adversefactor to the recently injured brain. In prac-tice, injuries requiring the intervention ofanother surgical specialty occur in only a fewpatients with head injury, even those with asevere injury. Miller (personal communica-tion) found that of 440 admitted headinjuries, only 14 needed an orthopaedic oper-ation, four a maxillofacial operation, andthree a general surgical operation but 80needed a neurosurgical procedure. It is,therefore, more appropriate for a head injury,after initial resuscitation, to be transferredwith appropriate safeguards for neurosurgicalassessment and management in a fullyequipped unit with round the clock neuro-surgical cover. The Society of BritishNeurological Surgeons has strongly recom-mended in In Safe Neurosurgery that traumacentres should be developed only in hospitalswith a neurosurgical service. It is doubtfulthat without this any centre can regard itselfas offering comprehensive trauma care.

Outcome, prognosis, and predictionThe outcome that can be expected after ahead injury is of much concern to a victim'srelatives and carers.88 The traditionalapproach of doctors has been to stress theuncertainty of the situation-borrowing theHippocratic aphorism "no head injury is tootrivial to ignore nor too serious to despairof'-coupled with an emphasis on pessimism("hanging crepe") to prepare the relatives fordeath or disability and to protect staff fromcriticism when this ensues. Reasonablyreliable predictions of expected outcomecan, however, be made on the basis of thewealth of data described over the past twodecades. 1689 90 When considering prognosis,certain safeguards must be applied; an esti-mate of prognosis made too soon after injuryis fallible when the patient's condition ispartly a reflection of a high alcohol or lowoxygen level, correction of which may lead torapid recovery, or when delayed complica-tions supervene and lead to subsequent deter-ioration. A balance needs to be struckbetween the accuracy of later predictions andtheir lesser usefulness in practice.

In the individual patient one of the mostimportant prognostic factors is age, outcomeworsening progressively with the increasingage. By incorporating the additional informa-tion about the severity and type of brain dam-age gained from clinical observations(consciousness, motor patterns, pupil reac-tions, eye movements)9' and investigationssuch as CT92 an increasingly clear and reliable

estimate of the probable outcome can begiven for individual patients. Murray et al,89used a simple computer program to calculateoutcome probabilities for a large series ofpatients with head injury and studied theeffect of providing clinicians with predictionsat the time that a patient was under acutemanagement. Although this did not altermanagement overall, intensive methods wereused less among patients with a very poorprognosis and were redirected towardspatients with an intermediate, uncertain prog-nosis, for whom outcome was presumablyless predetermined and who had more poten-tial to respond.Outcome after head injury depends pre-

dominantly on the degree of mental sequelae,in particular changes in personality and infor-mation processing and, in only a few patients,on the degree of any persisting physical limi-tations.93 For this reason, scales developed forpatients with stroke or other types of neuro-logical damage are inappropriate in the headinjury population. The Glasgow outcomescale described by Jennett and Bond94 distin-guished three classes of conscious survival, interms of consequent handicap: the severelydisabled patient, who is unable to live inde-pendently and unable to shop or travel onpublic transport; the moderately disabledpatient, who is independent but does notresume previous employment or sociallifestyle; and the patient who makes a goodrecovery, but who is not necessarily free fromneurological and neuropsychological limita-tions.

After severe head injury, 10%-20% ofpatients remain severely disabled for sixmonths or longer. At this stage only 1%-3%are categorised as being vegetative; hardly anyof these subsequently improve to be evenseverely disabled and none make an indepen-dent recovery. On either side of these out-comes, the distributions of deaths andindependent recovery have an inverse rela-tion, depending on the population consid-ered. For severe injuries, defined as notobeying commands, the proportion of deathsis 30%,95 when in addition the patient is in acoma, with no eye opening and no compre-hensible verbal response, mortality is 35%-40%96 and when this state has been present forsix hours or more, mortality approaches 50%.'

After moderate, and in particular, mildhead injury, most patients are at worst onlymoderately disabled and the original Glasgowcoma scale94 has been criticised as too crudefor such populations.97 The device of definingupper and lower levels in each category ofconscious survival only partially removes thisdifficulty.93 A wide range of neuropsychologi-cal tests98 and inventories of emotional,behavioural, and social status have beendescribed but no single one has yet beenwidely adopted. In assessing outcome, it iscrucial that information is not obtained onlyfrom the patient or even from the familypractitioner.99 A true picture of the patient'sstate, and in particular of its impact upon thefamily as a whole can be obtained only when

536


http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from


Head injury

psychologists are involved in the assessmentand where the family and carers are inter-viewed. This is particularly important if a

patient is being assessed for the purposes of a

claim for damages, when an underestimate ofthe consequences of injury may have adverseeffects upon the settlement for the patient andperhaps expose the assessor to subsequentcriticism.

Postacute care

After discharge from acute care, a range ofproblems arise. On the one hand, there is theneed, in those with persisting limitations, foreffective rehabilitation, support, and reinte-gration to minimise handicap. On the other,there is the problem of the management ofpatients without obvious physical or mentalsequelae but with persisting symptoms andrestrictions in activity.

Although there has been considerable con-

troversy over the efficacy of rehabilitation forhead injury, there is evidence that it is benefi-cial particularly when begun soon after injuryand provided intensively by specialists in neu-

rological problems.'00 101 Unfortunately, manypatients fail to receive the "seamless" conti-nuity of management that is optimum.102 Thisdeficiency reflects both the shortage in theUnited Kingdom of appropriately trained andexperienced specialist services and the chasmthat often lies between acute medical servicesand community services. Moreover, mostrehabilitation services and therapists aredirected towards specific physical limitationsand their consequent disability, whereas men-

tal limitations are much more important afterhead injury in causing disability and handicapand limiting return to previous social andworking lifestyles.

Macauley's aphorism "The business ofeverybody is the business of nobody",'03 can

be applied to head injuries. A wide range ofprofessional disciplines, of support services,and of charitable and voluntary organisationscan contribute to the recovery of a headinjury victim. They need to be coordinatedeffectively and specific interventions usedappropriately and selectively. Unfortunately,all too often coordination and communica-tion is lacking and, given the range of prob-lems after head injury, it can be tempting toregard the patient as "somebody else's prob-lem". In some parts of Britain no effectiveorganisation exists, in others the organisationis appropriate but facilities needed for benefitare inadequate; in a few centres, comprehen-sive progressive rehabilitation and communityreintegration are available. All concernedshould press for the institution in every regionof an integrated policy for postacute care ofhead injury. Someone should be given theresponsibility of establishing the service andfor arranging for its implementation; in effectthis person's job should be to make sure thateveryone else does their job!'04At the time of discharge from hospital the

patient and the family should be counselledabout the nature of the head injury, its likely

sequelae in the short term, and the likelihoodof progressive recovery. Sufficient informa-tion should also be conveyed to the generalpractitioner about the nature and severity ofthe head injury and the initial approach tosubsequent care. In the patient with a minoror mild injury all that may be needed is reas-surance, relief of symptoms-for example,headache-and arrangements for review andreturn to activities after an interval that isdetermined by the severity of injury and sub-sequent symptoms. In the uncomplicatedmild or minor injury, review by the generalpractitioner after a week may show a satisfac-tory progress, enabling return to work. Thisshould be discouraged, however, if symptomspresent at discharge-for example headache,dizziness, or mental limitations-persist orbecome exacerbated. At around a month afterinjury, patients who continue to have a signif-icant problem should be referred or recalledfor assessment and for planning of rehabilita-tion and reintegration socially, professionally,etc. This service would be facilitated if aregional register of head injuries was estab-lished, with close liaison with neurosurgicalunit staff. Subsequently, further remedial ser-vices may be required for a year or longer,depending on the patient's progress and thefindings of subsequent assessments.

Quality of care: auditAnalysis of the management and the outcomefrom head injuries has already resulted inchanges in practice that have been rewardedby better results. In the early 1970s somestudies showed that avoidable mortality inpatients with head injury occurred often dueto delayed recognition of the presence of anintracranial haematoma.36 105 106 The recogni-tion that CT could detect haematomas beforethey caused deterioration and the identifica-tion of clinical features correlating with thelikelihood of a haematoma, enabled the pro-duction of guidelines aimed at early referralto neurosurgery. These were first used infor-mally within the neurosurgical unit inGlasgow49 and then published and widelyadopted nationally.5' Continuing audit of out-come has shown that these changes in proce-dure were accompanied by reductions inmortality in operated cases (table 9).49 57 89

Insults during transfer from district hospi-tal to neurosurgery were identified as anothersource of brain damage and death and

Table 9 Introduction ofguidelines for management ofhead injuries and associated improvements in outcome ofpatients with a traumatic intracranial haematoma (INSGlasgow)

Total (% dead)

1974-7Referred to neurosurgery whenclinically deteriorating 305 (38)

1978-84Local criteria for earlier acceptance 659 (30)

1985-8Published national guidelines forearlier referral 222 (23)

537


http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from


Teasdale

disability that might be avoided by a betterstandard of care during transfer.49Dissemination of this knowledge led, over aperiod of 1 1 years, to a rise from 1 1% to 82%in the use of endotracheal intubation andventilation in transfer. This was associatedwith a fall in hypoxaemia on arrival from 22%to 8% and a reduction in the number ofunidentified or undertreated major extracra-nial injuries from 31% to 1 1% and inhypotension from 11% to 6%. Mortality fellfrom 45% to 32% and independent recoveryrose from 40% to 58%.79

It will be important in the future to ensurethat the gains resulting from the foregoingneurosurgically and neuroanaesthetically ledimprovements in practice are maintainedagainst the background of alterations in theorganisation of medical care (for example,trust hospitals, internal markets) and the dis-semination of previously centralised facilities(for example, CT). Each health purchasingauthority should identify standards of prac-tice, based on the guidelines -for the processof care. Contracts with providers should con-tain these specific recommendations andshould require that audit is carried out toensure that they are being used properly.

Questions to be asked include what pro-portion of head injury attenders undergo skullradiography, hospital admission, and neuro-surgical transfer and with what outcome?Samples should be surveyed in detail to dis-cover if practice in regard to individualpatients conforms with the agreed standards.The quality of interpretation of investigationsshould be checked. Where skull radiographsare interpreted by accident and emergencystaff, all radiographs should be reviewed by asenior radiologist. Likewise, CT performedoutside the neurosurgical department shouldbe reviewed, either at the time through imagetransfer and consultation with an experiencedclinician in the neurosurgical centre, or as amatter of routine periodic quality assessment.

Every death of a patient who arrives at hos-pital with a head injury, or at the very least allcases in which the patient was known to havetalked after injury, should be subject todetailed review at regional level. Withoutoverview at this level, the dispersal of headinjuries to different hospitals and even differ-ent departments means that audit at unit levelwill not be satisfactory.

Improved outcomes for head injuries havebeen achieved by the standards in managementdescribd in this review, that have developedin the past decade. It is essential to safeguardthese standards as the foundation for buildingsystems for even better care in future.

1 Jennett WB, Murray A, MacMillan R, et al. Headinjuries in Scottish hospitals. Scottish head injury man-agement study. Lancet 1977;ii:696-8.

2 Strang I, MacMillan R, Jennett B. Head injuries in acci-dent and emergency departments at Scottish hospitals.Injury 1987;10:154-9.

3 Jennett WB, MacMillan R. Epidemiology of head injury.BM3 1981;282:101-7.

4 Adams JH, Doyle D, Ford I, et al. Diffuse axonal injuryin head injury: definition, diagnosis and grading.Histopathology 1 989;15:49-59.

5 McLellan DR, Adams JH, Graham DI, Kerr AR,Teasdale GM. The structural law of the vegetative stateand prolonged coma after non-missile head injury. In:

Papo I, Cohadon F, Massarotti I, eds. Le coma trauma-tique. Padova: Liviana Editrice, 1986:165-85.

6 Zimmerman RA, Bilaniuk LT, Harkney DB, et al. Headinjury: early results of comparing CT and high fieldMR. Am7Radiol 1986;147:1215-22.

7 Jenkins A, Teasdale GM, Hadley D, et al. Brain lesionsdetected by magnetic resonance imaging in mild andsevere head injury. Lancet 1986;ii:445-6.

8 Graham DI, Adams JH, Doyle D. Ischaemic brain dam-age in fatal non-missile head injuries. _7 Neurol Sci 1978;39:213-34.

9 Graham DI, Ford I, Adams JH, et al. Ischaemic braindamage is still common in fatal non-missile head injury..7 Neurol Neurosurg Psychiatry 1989;52:346-50.

10 Lewelt W, Jenkins LW, Miller JD. Autoregulation ofcerebral blood flow after experimental fluid percussioninjury of the brain.7Neurosurg 1980;53:500-11.

11 Lewelt W, Jenkins LW, Miller JD. Effects of experimen-tal fluid percussion injury of the brain on cerebrovascu-lar reactivity to hypoxia and hypercapnia. .7 Neurosurg1982;56:332-8.

12 Andrews PJD, Piper IR, Dearden NM, Miller JD.Secondary insults during intrahospital transport of headinjured patients. Lancet 1990;335:327-30.

13 Gopinath SP, Robertson CS, Contant CF, et al. Jugularvenous desaturation and outcome after head injury. -7Neurol Neurosurg Psychiatry 1994;57:717-23.

14 Jones PA, Andrews PJD, Midgley S, et al. Measuring theburden of secondary insults in head injured patientsduring intensive care. _7 Neurosurg Anesthesiol 1994;6:4-14.

15 Povlishock JT. Traumatically induced axonal injury:pathogenesis and pathobiological implications. BrainPathology 1992;2:1-12.

16 Jennett WB, Teasdale GM, Braakman R, et al.Predicting outcome in individual patients after severehead injury. Lancet 1976;ii: 1031-4.

17 Teasdale GM, Jennett B. Assessment of coma andimpaired consciousness. Lancet 1974;ii:81-4.

18 Teasdale GM, Jennett WB. Assessment and prognosis ofcoma after severe head injury. Acta Neurochir 1976;34:45-55.

19 Teasdale GM, Murray G, Parker L, Jennett WB. Addingup the Glasgow coma score. Acta Neurochir (Paris)1979;1(suppl 28):13-16.

20 Williams DH, Levine HS, Eisenberg HM. Mild headinjury classification. Neurosurgery 1990;27:663-4.

21 Gomez PA, Lobato RD, Ortega SM, et al. Mild headinjury: differences in prognosis among patients with aGlasgow coma score of 13-15 and analysis of factorswith an abnormal CT scan. Neurosurgery 1995 (inpress).

22 Jennett WB, Teasdale GM, Galbraith S, et al. Severehead injuries in three countries. .7 Neurol NeurosurgPsychiatry 1977;40:291-8.

23 Jennett WB, Teasdale G, Braakman R, et al. Prognosis ofpatients with severe head injury. Neurosurgery 1979;4:283-9.

24 Teasdale GM. Assessment of head injuries. Br_7 Anaesth1976;48:761-6.

25 Rimmel RW, Giordani B, Barth JT, Jane JA. Moderatehead injury: completing the clinical spectrum of braintrauma. Neurosurgery 1982;11:344-51.

26 Johnstone AJ, Lohlun JC, Miller JD, et al. A comparisonof the Glasgow coma scale and the Swedish reactionlevel scale. Brain Injury 1993;7:501-6.

27 Rimmel RW, Giordani B, Barth JT, et al. Disabilitycaused by minor head injury. Neurosurgery 1981;9:221-8.

28 Swann IJ, MacMillan R, Strang I. Head injuries at aninner city accident and emergency department. Injury1981;12:274-8.

29 Mendelow AD, Teasdale GM, Jennett B, et al. Risks ofintracranial haematoma in head injured adults. BM.71983;287:1173-6.

30 Teasdale GM, Murray G, Anderson E, et al. Risks oftraumatic intracranial haematoma in children andadults: implications for managing head injuries. BM.1990;300:363-7.

31 Russell WR. The traumatic amnesias. Oxford: OxfordUniversity Press, 1961.

32 Teasdale GM, Brooks N. Traumatic amnesia. In:Frederiks JAM, ed. Handbook of clinical neurology. Vol. 1.Clinical neuropsychology. Amsterdam: Elsevier, 1985:185-91.

33 Wilson JTL, Teasdale GM, Hadley DM, et al. Post trau-matic amnesia: still a valuable yardstick. _7 NeurolNeurosurg Psychiatry 1994;57: 198-201.

34 Roberts AJ. Brain damage in boxers. London: PitmanMedical, 1969.

35 Roberts GW, Gentleman SM, Lynch A, et al. ,B Amyloidprotein disposition in the brain after severe head injury:implications for the pathogenesis of Alzheimer's disease._7 Neurol Neurosurg Psychiatry 1994;57:419-25.

36 Galbraith S. Misdiagnosis and delayed diagnosis in trau-matic intracranial haematoma. BM.7 1976;1:1438-9.

37 McLaren C, Robertson C, Little K. Missed orthopaedicinjuries in the resuscitation room. .7 R Coll Surg Edinb1983;28:399-401.

38 Teasdale GM, Galbraith S, Murray L, et al.Management of traumatic intracranial haematoma.BM.7 1982;285:1695-7.

39 Miller JD. Neurological emergency. Head injury._7 Neurol Neurosurg Psychiatry 1993;56:440-7.

40 Bullock R, Teasdale GM. Head injuries-surgical man-agement: traumatic intracranial haematomas. In:

538


http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from


Head injury

Braakman R, ed. Vinken and Bruyn's handbook of clinicalneurology: head injury. Amsterdam: Elsevier SciencePublishers, 1990;24:249-98.

41 Bullock R, Teasdale GM. Head Injuries I and II. In:D Skinner, P Driscoll, R Earlam, eds. ABC of majortrauma. BMJ 1990;300:1576-9.

42 MillerJD, Dearden NM. Measurements, analysis andthe management of raised intracranial pressure. In:Teasdale GM, Miller JD, eds. Current neurosurgery.Edinburgh: Churchill Livingstone, 1992:119-56.

43 Gentleman D, Dearden NM, MidgleyS, Maclean D.Guidelines for resuscitation and transfer of patients withserious head injury. BMJ 1993;307:547-52.

44 American College of Surgeons Committee on Trauma.ATLS course manual. Chicago: American College ofSurgeons, 1993.

45 Gentleman D, Jennett B. Hazards of inter-hospital trans-fer of comatose head injured patients. Lancet 1981;ii:853-5.

46 Henderson A, Coyne T, Wall D, Miller B. A survey ofinterhospital transfer of head injured patients with inad-equately treated life-threatening extracranial injuries.Aust NZJ7 Surg 1992;62:759-62.

47 Reilly PL, Simpson DA, Sprod I,Thomas L. Assessingthe conscious level in infants and young children. A pae-diatric version of the Glasgow coma scale. Childs NervSyst 1988;4:30-3.

48 Ambrose J, Gooding M, Uttley D. EMI scan in the man-agement of head injuries. Lancet 1976;i:847-8.

49 Teasdale GM. Management of head injuries. ThePractitioner 1982;226: 1667-73.

50 Teasdale GM, Teasdale E, Hadley D. Computedtomography and magnetic resonance imaging classifica-tion of head injury.Jf Neurotrauma 1992;9:S249-57.

51 Briggs M, Clarke P, Crockard A, et al. Guidelines for ini-tial management after head injury. Suggestions from agroup of neurosurgeons. BMJ 1984;288:983-5.

52 Jennett WB, Macpherson P. Implications of scanningrecently head injured patients in general hospitals. ClinRadiol 1990;42:88-90.

53 Marsh H, Maurice-Williams RS, Hatfield R. Closedhead injuries: where does the delay occur in the processof transfer to neurosurgical care? BrJ_ Neurosurg 1989;3:3-13.

54 Servadei F, Piazza G, Seracchioli A, et al. Extraduralhaematomas: an analysis of the changing characteristicsof patients admittedfrom 1980 to 1986. Diagnostic andtherapeutic implications in 158 cases. Brain Injury 1988;2:87-100.

55 Nee PA, Hadfield JM, Yates DW. Biomechanical factorsin patient selection for radiography after head injury.Injury 1993;24:471-5.

56 Miller JD, Jones PA. The work of a regional head injuryservice. Lancet 1985;i: 1141-4.

57 Miller JD, Jones PA, Dearden NM, TocherJL. Progressin the management of head injury. Br 7 Surg 1992;79:60-4.

58 Miller JD, Murray LS, Teasdale GM. Development of atraumatic intracranial haematoma after a "minor" headinjury. Neurosurgery 1990;27:669-73.

59 Della Corte F. Guidelines for monitoring during the first72 hours in severe brain injury. Second international sym-posium on severe head injury. Parma, 1994. Paper 55.

60 Galbraith S, Teasdale GM. Predicting the need for oper-ation in the patient with an occult traumatic intracranialhaematoma.JT Neurosurg 1981;55:75-81.

61 Cruz J. Combined continuous monitoring of systemicand cerebral oxygenation in acute brain injury:preliminary observations. Crit Care Med 1993;21:1225-32.

62 Schwartz ML, Tator CH, Rowed DW, et al. TheUniversity of Toronto injury treatment study: a prospec-tive randomised comparison of pentobarbital and man-nitol. Can JNeurolSci 1984;11:434-40.

63 Muizelaar JP, Marmarou A, Ward JD, et al. Adverseeffects of prolonged hyperventilation in patientswith severe head injury: a randomised clinical trial.JNeurosurg 1991;75:731-9.

64 Dearden NM. Jugular bulb venous oxygen saturation inthe management of severe head injury. Current Opinionin Anaesthesiolog 1991;4:279-86.

65 Chan KH, Dearden NM, Miller JD, et al. Multimodalitymonitoring as a guide to treatment of intracranial hyper-tension after severe brain injury. Neurosurgery 1993;356:547-53.

66 Bannan P, Teasdale G. Neuroprotection in head injury.In: Reilly P, Bullock R, eds. Head Injuries 1995 (inpress).

67 Piek J, Chesnut RM, Marshall LF, et al. Extracranialcomplications of severe head injury. J Neurosurg 1992;77:901-7.

68 Jenkins LW, Moszynski K, Lyeth BG, et al. Increasedvulnerability of the mildly traumatised rat brain to cere-bral ischaemia: the use of controlled secondaryischaemia as a research tool to identify common or dif-ferent mechanisms contributing to mechanical andischaemic brain injury. Brain Res 1989;477:211-24.

69 Marmarou A, Anderson RL, Ward JD, et al. Impact ofICP instability and hypotension in outcome in patientswith severe head trauma. Jf Neurosurg 199 1;75:S59-66.

70 Faden AI, Salzman C. Pharmacological strategies inCNS trauma. Trends in Pharmacological Sciences1992;13:29-35.

71 Todd N, Teasdale GM. Steroids in human head injury:clinical studies. In: R Capildeo, ed. Steroids in diseases of

the central nervous system. John Wiley and Sons Ltd,1989:151-61.

72 McIntosh TK. Novel pharmacological therapies in thetreatment of experimental traumatic brain injury: areview. JNeurotrauma 1993;10:215-61.

73 Jennett WB. Epilepsy after non-missile head injuries.London: Heinemann, 1975.

74 Young B, Rapp RP, Norton JA, et al. Failure of prophy-lactically administered phenytoin to prevent late posttraumatic seizures. J7 Neurosurg 1983;58:236-41.

75 Penry JK, et al. A controlled prospective study of thepharmacological prophylaxis of post traumatic epilepsy.Neurology 1979;29:600-1.

76 Temkin NR, Kilmea SS, Wilensky AJ, et al. A ran-domised, double blind study of phenytoin for the pre-vention of post-traumatic seizures. N EnglJf Med 1990;323:497-502.

77 El Jamal MS. Antibiotic prophylaxis in unrepaired CSFfistulae. BrJNeurosurg 1993;7:501-6.

78 BrydenJS, JennettWB. Neurosurgical resources andtransfer policies for head injuries. BMJ 1983;286:1791-3.

79 Gentleman D. Causes and effects of systemic complica-tions among severely head injured patients transferredto a neurosurgical unit. Int Surg 1992;77:297-302.

80 Miller JD, Sweet RC, Narayan R, Becker DP.Early insults to the injured brain.JAMA 1978;240:4319-442.

81 Kohi YM, Mendelow AD, Teasdale GM, et al. Extra-cranial insults and outcome in patients with acute headinjury-relationship to the Glasgow outcome scale.Injury 1984;16:25-9.

82 Gentleman D, Jennett B. Audit of transfer of uncon-scious head injured patients to a neurosurgical unit.Lancet 1990;335:330-4.

83 Bion IF, Wilson IH, Taylor PA. Transporting crucially illpatients by ambulance: audit by sickness scoring. BMJ1988;296: 170.

84 Munro HM, LaycockJRD. Interhospital transfer stan-dards for ventilated neurosurgical emergencies. BritishJournal of Intensive Care 1993;3:210-14.

85 Stanforth P. Head injuries-to be transferred or not [edi-torial]. Injury 1994;25:491-2.

86 PickardJD, BaileyS, Sanderson H, et al. Steps towardscost benefit, analysis of regional neurosurgical care.BMJ 1990;301:629-35.

87 Gennarelli TA, Champion HR, Sacco WJ, et al.Mortality of patients with head injury and extracranialinjury treated in trauma centers. J Trauma 1989;29:1193-201.

88 Barlow P, Teasdale G. Prediction of outcome and themanagement of severe head injuries: The attitudes ofneurosurgeons. Neurosurgery 1986;19:989-91.

89 MurrayLS, Teasdale GM, Murray GD, et al. Does pre-diction of outcome alter patient management? Lancet1993;341: 1487-91.

90 Murray LS, Teasdale GM, Murray GD, Jennett WB.Guidelines for decision making improves outcome afterhead injury. JNeurotrauma 1993;10:S101.

91 Jennett WB. Assessment of the severity of head injury.J NeurolNeurosurg Psychiatry 1976;39:647-55.

92 Marshall LF, Marshall SB, Klauber MR, et al. A newclassification of head injury based on computerisedtomography. J7 Neurosurg 199 1;75:S 14-20.

93 Jennett WB, Snoek J, Bond MR, et al. Disability aftersevere head injury: observations on the use of theGlasgow outcome scale. J Neurol Neurosurg Psychiatry1981;44:285-93.

94 Jennett WB, Bond MR. Assessment of outcome aftersevere brain damage. Lancet 1975;i:480-4.

95 Bailey I, Bell A, Gray J, et al. A trial of the effect ofnimodipine on outcome after head injury. ActaNeurochirurgica 1991;110:97-105.

96 Choi SC, Barnes TY, Bulloch R, et al. Temporal profilesof outcome in severe head injury. J Neurosurg 1994;81:169-73.

97 Hall K, Cope DN, Rappaport M. Glasgow outcomescale: comparative usefulness in following recovery intraumatic head injury. Archives of Physical Medicine andRehabilitation 1985;66:35-7.

98 Clifton GL, Kreutzer JS, Choi SC, et al. Relationshipbetween Glasgow outcome scale and neuropsychologi-cal measures after brain injury. Neurosurgery 1993;33:34-8.

99 Anderson SI, Housley AM, Jones PA, et al. Glasgow out-come scale, an inter-rater variability study. Brain Injury1993;7:309-17.

100 Malec JF, Smigielski JS, De Popolo RW, et al. Outcomeevaluation and prediction in a comprehensive integratedpost-acute operative brain injury rehabilitation pro-gramme. Brain Injury 1993;7:15-29.

101 Spivack G, Spettel CM, Ellis D, et al. Effects of theintensity of treatment and length of stay on rehabilita-tion outcome. Brain Injury 1992;6:419-34.

102 Hetherington H, Earlam RJ. Rehabilitation after injuryand the need for co-ordination. Injury 1994;25:527-31.

103 Macauley TB. Hisworical essays contribiuted to -the EdinburghReview. Hallam's Constitutional History, 1828.

104 Teasdale GM. Who cares about head injuries? BrJ HospMed 1985;33:67.

105 Rose J, Valtonen S, Jennett B. Avoidable factors con-tributing to death after head injury. BM3' 1977;2:615-8.

106 Jeffreys RV, Jones JJ. Avoidable factors contributing tothe death of head injury patients in general hospitals inMersey region. Lancet 1981;2:459-61.

539


http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from


head injury - journal of neurology, neurosurgery, and ... · head injury is now a patient with a...

Documents