J7ournal ofNeurology, Neurosurgery, and Psychiatry 1993;56:845-858

NEUROLOGICAL EMERGENCY

Management of raised intracranial pressure

J D Pickard, M Czosnyka

EpidemiologyRaised intracranial pressure (ICP) is the finalcommon pathway for many intracranial prob-lems (table 1) and has a profound influenceon outcome. For example, of the 3-500 000patients with head injury seen in Accidentand Emergency Departments in the UnitedKingdom per annum, 20% are admitted ofwhom 10% are in coma (2% of all attenders).Over 50% of these have an intracranial pres-sure greater than 20 mm Hg.'2 A total of80% of patients with fatal head injuries (4%of all patients with head injuries admitted)show evidence of a significant increase inintracranial pressure at necropsy. Some 35%of severe head injuries die and 18% are leftseverely disabled at enormous financial andemotional cost to the family and community.Similarly, 20 per 100 000 per year are admit-ted with intracerebral haematoma and 10-12per 100 000 per annum with subarachnoidhaemorrhage. The average regional neurosur-gical unit serving a population of two millionwill manage 200 patients per annum withbrain tumours, some 15 patients with chronicsubdural haematoma, and a similar numberof patients with a cerebral abscess and 50patients with hydrocephalus.3 In comatosechildren the incidence of raised ICP was 53%of those with head injuries, 23% with anoxic-ischaemia damage, 66% with meningitis,57% with encephalitis, 100% with masslesions and 80% with hydrocephalus.4 Thereis a considerable risk in all such patients ofsecondary brain damage with long termsevere disability if raised intracranial pressureis not recognised and managed appropriately.

AcademicNeurosurgical Unit(Box 167), Level 4, ABlock, Addenbrooke'sHospital, Cambridge,CB2 2QQ, UKJ D PickardM Czosnyka

PathophysiologyResting ICP represents that equilibrium pres-sure at which CSF production and absorptionare in balance and is associated with an

equivalent equilibrium volume of CSF. CSFis actively secreted by the choroid plexus atabout 0-35 ml/minute and productionremains constant provided cerebral perfusionpressure is adequate. CSF absorption is a

passive process through the arachnoidgranulations and increases with rising CSFpressure:

CSF drainage = CSF pressure-sagittal sinus pressureoutflow resistance

(Reference 22).

The 'four-lump' concept describes mostsimply the causes of raised ICP: the mass,CSF accumulation, vascular congestion andcerebral oedema (table 2).>7 The descriptionof a patient with raised ICP as having cere-bral congestion, vasogenic oedema etc, canonly be a working approximation, albeit use-ful, until our rather crude methods of assess-ment are refined. In adults, the normal ICPunder resting conditions is between 0-10mmHg with 15 mmHg being the upper limitof normal. Active treatment is normally insti-tuted if ICP exceeds 25 mmHg for more than5 minutes although a treatment threshold of15-20 mmHg has been suggested to improveoutcome.8 In the very young, the upper limitof normal ICP is up to 5 mmHg.9 Smallincreases in mass may be compensated for byreduction in CSF volume and cerebral bloodvolume but, once such mechanisms areexhausted, ICP rises with increasing pulsepressure and the appearance of spontaneouswaves (plateau and B waves).10 There is anexponential relationship between increase involume of an intracranial mass and theincrease in intracranial pressure at leastwithin the clinically significant range.

Cerebral perfusion pressure is commonlydefined as: CPP = mean arterial blood pres-sure-mean ICP; mean ICP closely approxi-mates to mean cerebral venous pressure. Ascerebral perfusion pressure (CPP) falls withincreasing ICP, ICP pulse pressure increases(fig 1).1 12 Firstly, the brain is less compliantor stiffer and a given pulsatile cerebral bloodvolume load provokes a bigger pressureresponse. Secondly, the pulsatile componentof cerebral blood flow increases with decreas-ing CPP. Cerebral arteriolar dilatation tomaintain cerebral perfusion ('autoregulation')may be involved. The lower limit of CPPwhich will permit autoregulation, when ICPis raised, is about 40 mmHg. However, thereis a paradox: the level of CPP below whichoutcome after severe head injury and associ-ated parameters deteriorate is of the order of60-65 mmHg (MAP < 80 mmHg; ICP > 20mmHg). Conventionally any elevation of ICPrequires treatment if CPP is below 60 mmHgin adults for over 5 minutes. This paradoxmay partly reflect the 'split brain' problem:autoregulation of CBF to changes in CPPand response to changes in arterial carbondioxide tension (PaCO2) may be impairedfocally, leaving intact reactivity in other areas

845

Pickard, Czosnyka

Table 1 Some common causes of raised ICP

Head InjuryIntracranial haematoma (extradural, subdural and intracerebral)Diffuse brain swellingContusion

CerebrovascularSubarachnoid haemorrhageIntracerebral haematomaHydrocephalusCerebral venous thrombosisMajor cerebral infarctHypertensive encephalopathy (malignant hypertension, eclampsia)

HydrocephalusCongenital or acquiredObstructive or communicating

Craniocerebral disproportionBrain "Tumour" (cysts; benign or malignant tumours)

20 HydrocephalusMass effectOedema

'Benign' Intracranial HypertensionCNS Infection

MeningitisEncephalitisAbscess

Cerebral malaria2' Hydrocephalus

Metabolic EncephalopathyHypoxic-ischaemicReyes syndrome etc.Lead encephalopathyHepatic comaRenal failureDiabetic ketoacidosisBurnsNear drowning

HyponatraemiaStatus epilepticus

(Adapted from references 4, 5, 6, 7)

of the brain. If vasospasm is present, an evenhigher perfusion pressure may be required toprovide adequate levels of cerebral bloodflow. One interesting phenomenon revealedby transcranial Doppler (which reflects flowin large vessels) and laser Doppler (whichreflects tissue perfusion) is the change in flowand perfusion during the cardiac cycle: dias-tolic perfusion pressure may be below thenormal limit of autoregulation whilst systolicis above (vide infra).

Total cerebral blood flow may be increasedor decreased in areas with absent reactivity.Hyperaemia is non-nutritional 'luxury perfu-sion' where CBF is in excess of the brain'smetabolic requirements13 and accompaniedby early filling veins on angiography and 'redveins' at operation. Cerebral vasodilatorssuch as carbon dioxide will dilate 'normal'arterioles, increase ICP and may run the riskof reducing flow to damaged areas of brain(intracerebral 'steal'). Inverse 'steal' is one

Table 2 Mechanisms of raised ICPA Mass Lesions Haematoma, abscess, tumour.B CSFAccumulation Hydrocephalus (obstructive and communicating) and

including contralateral ventricular dilatation fromsupratentorial brain shift.

C Cerebral Oedema Increase in brain volume as a result of increased watercontent.1 Vasogenic-vessel damage (tumour, abscess,contusion)2 Cytotoxic-cell membrane pump failure(hypoxaemia, ischaemia, toxins; myelinoclastic).3 Hydrostatic-high vascular transmural pressure (lossof autoregulation; post intracranial decompression).4 Hypo-osmolar-hyponatraemia5 Interstitial-high CSF pressure (hydrocephalus)

D Vascular (congestive) brain Increased cerebral blood volumeswelling -arterial vasodilatation (active, passive)

-venous congestion/obstruction.

15

IEE0L

0)

IEEa-

.: I

I.' .on*t~~~I

II

40

0 80ICP (mm Hg)

Figure 1 Relationship between mean ICP and amplitudeof the ICP waveform in two patients. In the lower trace,there is an upper breakpoint in this relationship when CPP(M;AP-ICP) is less than 30 mmHg.

reason for the treatment of raised ICP byhyperventilation: an acute reduction ofPaCO, vasoconstricts normal cerebral arteri-oles and thereby directing blood to focallyabnormal areas.

Normally, cerebral blood flow is coupled tocerebral oxidative metabolism via multiplemechanisms involving local concentrations ofhydrogen ions, potassium and adenosine forexample. Status epilepticus leads to grosscerebral vasodilatation and intracranial hyper-tension as a result of greatly increased cere-bral metabolism and local release ofendogenous vasodilator agents. Depression ofcerebral energy metabolism by anaesthesiaand hypothermia may reduce cerebral bloodflow and ICP where there is a large area ofthe brain with reasonable electrical activity14and where normal flow-metabolism couplingmechanisms are intact as indicated by a rea-sonable CBF CO2 reactivity.'5

Spontaneous waves of ICP are associatedwith cerebrovascular dilatation. Cerebralblood volume increases during plateau waves(ICP > 50 mmHg for more than 5 minutes)and may be the result in some cases of inap-propriate autoregulatory vasodilatation inresponse to a critical fall in CPP but certainlynot in all cases (fig 2).16 TCD has revealedthat middle cerebral artery (MCA) flowvelocity increases pari-passu with B waves(0-5-2/min) of ICP (fig 3).17

Finally, gradients of ICP may developwhen herniation occurs-transtentorial, sub-falcine and foramen magnum. Blockage tothe free flow of CSF between intracranial

846

Management of raised intracranial pressure

AMP116(mm Hg) Km-

40 ^ | 01:40

ICmPm g 34-85

120

ABP 86-23(mm Hg)

T| ~~~~~~~~01:40

CPP513(mm Hg) _ 13

30

1 hour

compartments leads to a much greater andmore rapid rise in ICP in the compartmentharbouring the primary pathology and henceto the final common sequence of transtentor-ial and foramen magnum coning. When ICPequals arterial blood pressure, angiographicpseudo-occlusion occurs, reverberation, sys-

tolic spikes or no flow may be seen on TCD(fig 4). Patients will often satisfy the formalclinical criteria for brainstem death, for whichTCD is not a substitute.1819

There is a complex interaction between theproperties of the CSF and the cerebral circu-lations that may be modelled (fig 5).20 21 Therelative contributions of abnormalities of CSFabsorption and cerebral blood volume may beapproximated by calculating the proportionof CSF pressure attributable to CSF outflowresistance and venous pressure from Davson'sequation (ICP = CSF formation rate x out-flow resistance + sagittal sinus pressure).22Phenomena such as the interaction ofautoregulation to changing CPP with PaCO2may be quantified.2'

30ICP(mm Hg) -15_

495ABP(mm Hg)85

45FVx(cm/s)30

115HR(b/m)105

1 minuteFigure 3 B waves ofICP in a head-injured patient and their relationship tovariations in middle cerebral arteryflow velocity compared to fluctuations in apressure (ABP) and heart-rate (HR).

E

U-

Figure 4 Reversal of middle cerebral arteryflow velocity(FV) in a patient who fudfils the criteria for brainstemdeath.

Monitoring techniquesA) ClinicalfeaturesIn the non-trauma patient, there may or may

not be a clear history of headache, vomitingand visual disturbance suggestive of papil-loedema or a VIth nerve palsy. The absenceof papilloedema does not exclude raised ICPin patients with acute or chronic problems:disc swelling was found in only 4% of headinjury patients, 50% of whom had raised ICPon monitoring.6 Even in the 1 990s, it isregrettable that a clear history of raisedintracranial pressure may be misinterpreteduntil the final denouement of disturbance ofconsciousness and pupillary abnormality or

apnoea presents. Only slowly has the dangerof lumbar puncture in the differential diagno-sis of neurological patients been appreciatedby the non-expert. Many of the later signs ofraised ICP are the result of herniation: moni-toring should detect raised ICP at an earlierstage and hence treat before irreversibledamage occurs.

B) CT scanningCT scanning may reveal not only a mass,hydrocephalus or cerebral oedema but alsoevidence of diffuse brain swelling such as

absent perimesencephalic cisterns, com-

pressed 3rd ventricle and midline shift.

22-4C) Invasive methods ofICP monitoringincluding CSF infusion testsThe gold standard of ICP monitoring, thatwas first introduced between 1951,2410 still

86.8 remains the measurement of intraventricular

fluid pressure either directly or via a CSFreservoir, with the opportunity to excludezero drift. Subdural fluid filled catheters are

reasonably accurate below 30 mm Hg. A total

35-6 of 25 mmHg for more than 5 minutes is theusual threshold level at which treatmentshould be instigated. Risk of infection,epilepsy and haemorrhage is less with sub-dural than intraventricular catheters but even

111-8 the latter should be less than 5% overall.

Catheter tip transducers are very useful par-

113:56 ticularly for waveform analysis, whetherplaced intraventricularly, subdurally or

similr.intracerebrally. Ventricular catheters permitrterial the therapeutic drainage of CSF in cases of

ventricular dilatation. In more chronic condi-

Figure 2 Relationshipsbetween arteral pressure(ABP), CPP ICP

amplitude (MP) duringa plateau wave.

Brain death

847

Pickard, Czosnyka

Extracranialcerebral r-

Figure 5 Hydrodynamic model of cerebral bloodflow and CSF circulation w;electrically equivalent circuit (for details, see Czosnyka, et al, reference 21).

Figure 6 Relationshipbetween ICP, CPP, ICPamplitude (AMP) andflow velocity (systolic FV,diastolic FVd andamplitude FVJ during aplateau wave. Theamplitude ofICP increaseswith ICP and decrease inCPP: flow velocityamplitude increases as aresult of the fall in diastolicflow velocity.

70ICP(mm Hg)20

70CPP(mm Hg)2030AMP(mm Hg)0

80 i

FVa(cm/s)50

120CPPs(mm Hg)0

60CPPd(mm Hg)0 _140FVs(cm/s)0

50FVd(cm/s)0

1 minute

tions of ventricular dilatation, where ICP isnot greatly raised, obstruction to CSF

Saggital absorption may be confirmed by CSF infu-slnus sion tests (ventricular or lumbar) taking care

("P"') to adapt the technique to the site of anyobstruction.2527 Twenty four hour intracra-nial pressure monitoring in patients with so-called normal pressure hydrocephalus mayreveal a high incidence of B waves duringsleep which is a very helpful prognostic signfor the outcome following shunting.2829Benign intracranial hypertension seldomrequires more than CSF pressure monitoringthrough a lumbar catheter or needle for anhour.

Considerable effort continues into theclose analysis of the ICP trace to determinewhether it is possible to reveal the mechanismof raised ICP and whether autoregulatoryreserve remains intact.30 It has beenproposed7 that congestion or vascular brain

PSS swelling may be present when the ratio of theamplitudes of the pulse and respiratory com-ponents of the ICP trace exceeds 2, whenthere is an increase in the high frequency cen-troid or when there is a high amplitude trans-fer function for the fundamental harmonicfrom ABP to ICP. Such a transfer function iscalculated from the first Fourier transform of

,ith the the digitised signal. However, continuousmultimodality monitoring is required to drawany safe conclusions and should include somemeasure of CBF (for example, TCD) andcerebral metabolism (for example, EEG,jugular venous oxygen). Indices of imminentdecompensation would be very helpful butvolume-pressure responses,31 pressure-volumeindices or definition of the contribution ofCSF outflow resistance to ICP,22 are not suit-able for routine clinical use.

D) Non-invasive ICP monitoringIt would be very helpful to monitor ICP or

hi. CPP without invasive catheters. Transcraniali Doppler, tympanic membrane displacement

and even skull compliance studies have beenadvocated. It would be very helpful to haveanswers to the following questions: What isthe cerebral perfusion pressure at any giventime? What is the relative contribution ofeach possible mechanism to raised ICP?What features may predict decompensation?Is it possible to have an on-line assessment of

I cerebrovascular reactivity either to changes inCPP (autoregulation) or CO2? Which therapy

* or cocktail is best suited to the sum of that* individual's 'split-brain' problems?fl A noninvasive method that monitored con-

tinuously both CPP and CBF autoregulatoryreserve would be very helpful in refining man-agement of the swollen brain.

TRANSCRANIAL DOPPLERAaslid's description of transcranial Dopplerin 1982 permitted bedside monitoring of oneindex of CBF, non-invasively, repeatedly, andeven continuously.32 33 The problem has beenthat it is a big tube technique that measuresflow velocity in branches of the Circle ofWillis, most commonly the middle cerebral

848

Management of raised intracranial pressure

artery. Changes in velocity may reflect eitherchanges in blood flow or in diameter of theinsonated artery. Unfortunately, diameterand flow may not change in complementarydirections and great care must be taken withthe interpretation.34 Low velocity may indi-cate low flow or arterial dilatation at constantflow. High velocity may indicate high flow orarterial constriction/vasospasm at constantflow. Considerable ingenuity has beenexpended in analysis of a TCD wave form.The amplitude of the flow velocity pulse wave(FVa) reflects pulsatile changes in regionalCBF and is dependent on the amplitude ofthe arterial pressure wave, regional cere-brovascular resistance, the elastance of thecapillary bed and the basal cerebral arteries.Aaslid suggested that an index of CPP couldbe derived from the ratio of the amplitudes ofthe first harmonics of the arterial blood pres-sure and of the middle cerebral artery velocity(detected by TCD) multiplied by mean flowvelocity. There is a reasonable correlationbetween the pulsatility index (peak systolic-end diastolic FV's/time averaged FVm) ofMCA velocity and CPP after head injury butabsolute measurements of CPP cannot beextrapolated.35 Nelson et aP6 have providedboth experimental and theoretical modellingevidence for three haemodynamic phases asCPP falls. Above the lower limit of autoregu-lation, falls in CPP are masked by arteriolardilatation, a fall in CVR and gradual increasein FVa so that CBF and FVm remain stable.During the transitional phase, CBF and FVmstart to fall gradually: CPP in diastole is closeto or below the critical closing pressure of thecapillaries so that the fall in FV in diastole isgreater and FVa increases further. Finallyautoregulation becomes exhausted with arapid fall in CBF and FVm a sharp decrease inFVa and an increase in CVR. Where autoreg-ulation is impaired throughout, CBF andFVm fall pressure-passively as CPP is reducedand there is no increase in FVa. The correla-tion coefficient between FV. and ABP mayprovide a continuous on-line index ofautoregulatory reserve: as autoregulationbecomes exhausted, correlation rapidly movesfrom negative to positive. The response of thecerebral circulation to stress, such as a periodof hypotension, hypercapnia or transientcarotid compression may also be assessed."21Hence, autoregulation and cerebrovascularreactivity may be assessed within the vascularterritory supplied by the insonated artery.Comparison of the changes in blood flowvelocity in a number of branches of the Circleof Willis with that in the cervical internalcarotid artery and with cerebral arterio-venous oxygen difference may help to distin-guish vasospasm from hyperaemia at least ona global basis. Vasospasm defined by TCDhas been associated with delayed cerebralischaemia after trauma.'7-'9 Such techniquescannot yet define the proportional contribu-tions of both vasospasm and hyperaemia inthe same or different parts of the brain inpatients who may have both. Depressed CBFCO, reactivity was found in one study to cor-

relate with very severe brain injury or withextensive focal lesions in MCA distribution.Knowledge of the integrity of the CBF CO2response was helpful in determining thepotential effectiveness of hyperventilation orbarbiturates for ICP control.40

TYMPANIC MEMBRANE DISPLACEMENT.ICP is transmitted via the cochlear aqueductto the perilymph of the cochlea providing thatthe aqueduct is patent. Perilymphatic pres-sure may be assessed indirectly by recordingdisplacement of the tympanic membrane dur-ing stapedial reflex contractions elicited by aloud sound.4' High perilymphatic pressuredisplaces the resting position of the stapesfootplate laterally, thereby allowing a higherdegree of motion in a medial direction, andresults in a more inward going tympanicmembrane displacement on stapedial con-traction. Low perilymphatic pressure willhave an opposite effect. A transducer probeattached to a head set is placed in thepatient's external auditory meatus and com-puter based instrumentation allows smallmovements of the tympanic membrane to bemeasured when 100OHz of increasing soundpressure level induces stapedial contraction.This very ingenious technique is useful inyounger patients with hydrocephalus orbenign intracranial hypertension on a sequen-tial basis, provided that a skilled audiologist isavailable. It does not provide an absolutemeasure of ICP. It is of no value in patientson ventilators receiving muscle relaxants. Thepatency of the cochlear aqueduct decreaseswith age and should be checked with a pos-tural test.

E) Cerebral venous oxygen42 43

Cerebral arterio-venous oxygen content dif-ference should normally be 5-7 ml/dl. Valuesbelow 4ml/dl indicate cerebral hyperaemiawhile values above 9ml/dl indicate globalcerebral ischaemia. Jugular bulb oxygen satu-ration may be monitored, preferably continu-ously, with an indwelling catheter. Singlemeasurements of jugular venous oxygen areof little value given the many fluctuationsduring the day. Overenthusiastic treatment,that on occasion may induce cerebralischaemia, may be monitored with this tech-nique. Hyperventilation and barbiturate-induced falls in CPP have been shown inindividual patients to be counter-productive.An index of regional oxygen metabolism isrequired. Transcutaneous, transcranial nearinfra-red spectroscopy is completely non-invasive.4 Unfortunately, it is well provenonly in neonates and younger children andnot yet in older age groups.

F) Cerebral electical activityThe compressed EEG (cerebral functionmonitoring) is helpful in deciding whethercerebral metabolic depressants may be indi-cated in the treatment of intracranial hyper-tension.'4 Such drugs will obviously not behelpful if the EEG is flat or greatly reduced inamplitude.

849

Pickard, Czosnyka

Table 3 Indications for ICP monitoring

Head injury(a) being artificially ventilated:

-coma with compression of 3rd ventricle and/orreduction in perimesencephalic cisterns on CT,

-coma following removal of intracranial haematoma,-coma with abnormal motor response as the best

reaction,-coma with midline shift/unilateral ventricular

dilatation,-multiple injury including severe chest wall injuries.-early seizures not easily controlled.-refractory hyperpyrexia.

(b) uncertainty over surgery for small haematoma/multiplelesions.

Intracerebral and subarachnoid haemorrhage-coma-postoperatively following intraoperative complications,-hydrocephalus.

Coma with brain swelling-metabolic,-hypoxia/ischaemia,-infective (see Table 1).

Hydrocephalus and benign intracranial hypertension

(Adapted from references 4, 7, 42)

Management strategies1) Emergency resuscitation and diagnosisPatients who are rapidly deteriorating oralready in coma require immediate resuscita-tion if necessary with intubation and ventila-tion followed by a diagnostic CT scan. Anintravenous bolus of mannitol (0-5 gms/kgover 15 minutes may be required if there isevidence of coning such as pupillary dilata-tion.) Acute ventricular dilatation demandsimmediate ventricular drainage-bilateral ifthe lesion is midline. Hyperacute ventriculardilatation following subarachnoid haemor-rhage or in association with a third ventricularlesion need not be gross to cause death.Surgical clots require removal and abscessesrequire tapping.

2) Post-acute managementMany neurosurgical units worldwide stillmanage patients without the help or hin-drance of ICP monitoring. The nihilisticargument is that ICP monitoring has notbeen clearly shown to improve outcome andsequential CT scanning provides sufficientinformation. Treatment for raised ICP hasnot greatly advanced over the past twodecades and can be applied pragmatically.The alternative school of thought argues thatICP monitoring should be selective (table 3)based in part on the initial CT scan.However, in addition such monitoring is veryeducational and greatly assists general nurs-ing and medical care; transport of a patient

Table 4 Potential problems exacerbating raised intracranial pressure1 Calibration of ICP and arterial blood pressure transducers and monitors particularly to

check the zero reference point.2 Neck vein obstruction -inappropriate position of head and neck-avoid constricting

tape around neck.3 Airway obstruction -inappropriate PEEP,

-secretions, bronchospasm etc.4 Inadequate muscle relaxation-breathing against ventilation,

-muscle spasms.5 Hypoxia/hypercapnia6 Further mass lesion-rescan7 Incomplete analgesia, incomplete sedation and anaesthesia.8 Seizures9 Pyrexia10 Cerebral vasodilating drugs11 Hypovolaemia12 Hyponatraemia (often iatrogenic fluid overload)

between ICU and CT scanning too ofteninvolves well documented risks of hypoxiaand hypotension; ICP therapy should not beused as a blunderbuss but needs to be selec-tively targetted if it is not to be counter-pro-ductive, and new treatments are emergingvery rapidly. Clinical trials based on outcomestudies at six months in such heterogeneousgroups of patients may easily miss usefulbenefits.45 To treat raised ICP, it must first beidentified, avoidable factors prevented ortreated and finally active treatment instigated,hopefully based on our understanding of theindividual mechanism involved. ICP shouldbe treated before herniation occurs-clinicalsigns particularly in patients on a ventilatorare just too crude. Knowledge of ICP mayhelp with prognosis and counselling of rela-tives: in one series of diffuse head injuries,where ICP persistently exceeded 20 mmHg,almost all the patients died compared with amortality rate of 20% in those where ICPcould be kept below 20mmHg with treat-ment.46

3) Prevention of intracranial hypertension:general medical and nursing care-avoidablefactors.Simple measures need to be checked (table4).7 42 Ideally the position of the patientshould minimise any obstruction to cerebralvenous drainage by head-up tilt whilst avoid-ing any fall in cardiac output or carotid arter-ial blood pressure. Direct measurement ofglobal CBF and CPP suggests that head-uptilt of up to 30° is safe but careful scrutinyshould be kept of CPP in individualpatients.47-50

Hypovolaemia should be avoided, contraryto some historical teaching. The evidence ismost clear cut in patients after subarachnoidhaemorrhage: dehydration particularly whencoupled with hyponatraemia increases therisk of cerebral infarction.51 Patients with CTevidence of raised ICP are already at greaterrisk of hypovolaemia after SAH.52Dehydration increases the risk of hypo-volaemia which may be revealed only whenthe patient is given an anaesthetic agent, forexample, for an orthopaedic procedure or aspart of a regime to control raised ICP. Astable circulation must be maintained,53 ifnecessary with colloid and inotropes (dobuta-mine or dopamine for its renal sparingaction). However, overenthusiastic hyperten-sive-hypervolaemic therapy remains very con-troversial in the context of head injury with itsmultiple pathology and uncertainty over theintegrity of the blood-brain barrier.5456Systemic hypertension should not be treateddirectly with agents such as sodium nitro-prusside. Sodium nitroprusside impairsautoregulation and increases the risk ofboundary zone infarction.57 The cause ofhypertension, for example, pain or retentionof urine, should be looked for.The majority of neurosurgical patients with

hyponatraemia do not have inappropriatesecretion ofADH and it is unwise to use fluidrestriction to treat them even if they do.51 58

850

Management of raised intracranial pressure

A useful dictum in neurosurgery is that bloodvolume comes before plasma sodium levels.Moderate hyponatraemia impairs cerebro-vascular reactivity experimentally to bothhypercapnia and hypotension but does notaugment the cerebrovascular effects of experi-mental SAH.59 A spectrum of abnormalitiesgive rise to hyponatraemia following SAH, forexample: initial natriuresis leading to volumedepletion, ADH secretion stimulated both bystress and volume depletion, ADH-inducedwater retention, steroid- and sympathetic-induced effects on the kidney and possiblerelease of atrial natriuretic factor (both car-diac and cerebral in origin) and digoxin-likesubstance.

Seizures may be difficult to recognise whenthe patient is paralysed and ventilated.Episodes of pupillary dilatation with increasesin arterial blood pressure and ICP are sugges-tive.

Pyrexia not only increases cerebral metabo-lism and hence cerebral vasodilatation butalso cerebral oedema. Severe hypothermiawas used historically to treat raised ICP but ithas become clear more recently that mildhypothermia of a few °C only will reducecerebral ischaemia for reasons that are not yetclear."13 Hyperglycaemia should be avoided.There is considerable evidence that cerebralischaemia and infarction is made worse byhyperglycaemia and the use of high glucosesolutions is contraindicated unless there issignificant evidence of benefit in a particularmetabolic encephalopathy.6061

Osmotic diureticsIntravenous mannitol is invaluable as a firstaid measure in a patient with brain herniationas a result of raised ICP. Its more prolongeduse and mechanisms of action remain con-tentious issues. Osmotherapy began experi-mentally with hypertonic saline and thenurea, entering neurosurgical practice with30% urea in 1958. A maximum fall in ICPoccurs within 30 minutes of starting urea,and the effect lasts for up to three hours butwith the possibility of subsequent rebound.62Conceptually, the mechanism was thought tobe osmotic extraction of water across theintact blood-brain barrier acting as a semiper-meable membrane. Experimentally, ureashrank normal rather than oedematousbrain.63 Entry of urea into the oedematousbrain through a 'defective' barrier would takewater in and thereby account for rebound.Overenthusiastic bolus administration of anosmotic diuretic may cause abrupt systemichypertension, an increase in cerebral bloodvolume if autoregulation is defective or itsupper limit is exceeded and promote hernia-tion rather than the reverse.64More recent studies indicate that mannitol,

given time, removes water from both normaland oedematous brain, be it ischaemic orinterstitial (Marmarou's infusion model).6>7The oedematous area around many masslesions may still have an intact blood-brainbarrier at least to the conventional high

molecular weight tracers. The time course isslow and does not account for the immediateeffect of mannitol on ICP. In patients withperitumoural oedema, mannitol causes with-drawal of water mainly from brain areaswhere the barrier is impaired as judged by TiMRI and in vitro measurements of brainwater content.68 However, mannitol mayaccumulate in oedematous white matter withrepeated doses.69The more immediate effects of intravenous

mannitol include a fall in whole blood viscos-ity with reduced red cell rigidity and corpus-cular volume, an increase in brain complianceand possibly cerebral vasoconstriction.7>7'Experimental perivascular administration ofmannitol evokes vasodilatation. The cerebralvasoconstriction with intravascular bolusadministration was short-lived-in the cat,both pial arteriolar diameter and ICPreturned to normal within 30 minutes andthereafter both increased, pari-passu withchanges in blood viscosity.72 Administrationover 15 minutes produced no change in pialarteriole or venular diameter in anotherstudy.73 Current studies in patients are usingtranscranial and laser Doppler to re-examinethese conflicting reports. Why should a sud-den change in blood viscosity evoke acutetransient vasoconstriction? Chronic changesin blood viscosity by plasma exchange, with-out alterations in haemoglobin or arterial oxy-gen content do not change steady-state CBFin humans.74 Patients with high plasma vis-cosity or with high viscosity due to largenumbers of white cells do not have low CBFvalues. In a series of patients with haemato-logical diseases but no evidence of cere-brovascular disease, arterial oxygen contentwas the major determinant of CBF-bloodviscosity per se had no significant effect onCBF.75 If a single blood vessel is considered,the apparent viscosity of blood diminishes inproportion to its radius as a result of the mar-ginal sheath of low viscosity and axial flow ofred cells.76 The width of this sheath is greatestrelatively in small vessels. Furthermore,apparent viscosity increases with falling veloc-ity. Hence with pial arterial dilatation, localblood viscosity will rise both because of theincreased proportion of red cells and as aresult of the reduction in flow velocity if tis-sue perfusion flow remains constant.Simplistically, according to Poiseuille, as vis-cosity is reduced deliberately, so the pressuregradient along the pial arteriole under obser-vation falls. Hence, the distal intravascularpressure increases if the proximal pressureremains constant. The distal end of the arteri-ole therefore constricts if autoregulatorymechanisms such as the Bayliss effect areintact. 'Viscosity' autoregulation shoulddepend on pressure autoregulation unlessthere is a separate endothelial mechanismthat is flow- or viscosity-sensitive.Alternatively, mannitol may transientlyincrease CBF, increase oxygen delivery andwash out local vasodilators such asadenosine.7' Vasoconstriction then follows.Extracellular hyperosmolarity is a potent

851

Pickard, Czosnyka

cerebral vasodilator and it is remarkable thatthe intravenous vasoconstrictor effect of man-nitol so completely dominates the acute cere-brovascular effect. If the viscosity mechanismis apposite, it will depend upon the distribu-tion gradient of intravascular pressures alongthe cerebrovascular tree which may not beeasy to predict with different pathologies andcerebral perfusion pressures. Certainly thereported effects of mannitol on CBF are noteasy to rationalise.77-80 In severely headinjured patients in whom autoregulation wasabsent, intravenous mannitol caused anincrease in cerebral blood flow and no reduc-tion in ICP.8' ICP was reduced in thosepatients where autoregulation was intact.However, in patients with unrupturedaneurysms and in the majority of whomautoregulation was intact presumably, manni-tol (bolus or infusion) increased CBF formany hours.80 More regional assessments ofCBF suggest that mannitol may stablise pHand CBF in regions of moderate but notsevere ischaemia.8' Other suggested mecha-nisms for the effect of mannitol includemovement of water from CSF into capillariesand scavenging of free radicals.8' Plasmahyperosmolality rapidly reverses the intersti-tial fluid pressure/CSF pressure gradient andthere is a rapid volume shift within 30 min-utes from CSF into brain tissue."'3Many attempts have been made to ratio-

nalise how much mannitol may be given andwhen, for more prolonged effects.8485 In prac-tice, mannitol tends to be given as an inter-mittent bolus whenever the individualpatient's ICP rises significantly above thethreshold of 25-30 mmHg. The effects ofmannitol may be potentiated by addingfrusemide.85 It is crucial to avoid dehydrationand latent hypotension with careful attentionto fluid balance. Colloid with an adequateplasma half life (albumin, hetastarch forexample) should be combined with carefulelectrolyte replacement. Another dose ofmannitol should not be given if osmolarityexceeds 330 mmol/l for fear of tubulardamage and renal failure. Repeated doses ofmannitol should not be given unless anICP monitor is in place. Some authorscontinue to recommend glycerol for pro-longed osmotherapy.86

Hyperventilation, the buffer THAM andindomethacinThe cerebral vasoconstrictor effect ofhypocapnia, induced by hyperventilation,does not persist much beyond a day, probablyin part because the bicarbonate bufferingmechanisms within the brain and cerebrovas-cular smooth muscle themselves readjust toreturn extracellular and intracellular pHnearer to the original values.87 This phenome-non has now been confirmed in vivo in nor-mal subjects by magnetic resonancespectroscopy. Neurosurgical patients withhealthy lungs and systemic circulation oftenhyperventilate spontaneously down to aPaCO, of 30 mmHg.8 More enthusiastic

hyperventilation may precipitate cerebralischaemia with EEG slowing, CSF lactic aci-dosis and a cerebral arterio-venous oxygencontent difference > 9 ml/dl. Arterial bloodpressure may be reduced by the combinationof dehydration and aggressive hyperventila-tion. Finally, weaning from the ventilator maybe more difficult and prone to "rebound"intracranial hypertension: the brain's buffer-ing mechanisms have to readjust back again.Finally, cerebrovascular reactivity to CO2may be absent in some patients after headinjury, but such reactivity is seldom measuredbefore hyperventilation is started. TCD is asimple way of assessing CO2 reactivity. In onestudy, controlled hyperventilation used pro-phylactically did not improve outcome butdid prolong the recovery phase.88 Hence thereis growing awareness that hyperventilation beused sparingly, for example, to treat persis-tent ICP waves.CSF lactate accumulation and CSF acido-

sis occur after head injury.89 90 Both severity ofinjury and the proportion of patients with apoor outcome are related to high and increas-ing CSF lactate levels. Cerebral tissue lacticacidosis is related to secondary brain damagefollowing a primary insult such as cerebralischaemia even if moderate acidosis per se hasno persisting effect on normal neurons.Akiota et a19' found that the intravenousbuffer tris hydroxymethyl aminomethane(THAM) ameliorated both the CSF acidosisand brain swelling following epidural ballooncompression of the brain in dogs. In 1970,Gordan and Rossanda92 suggested that hyper-ventilation might be beneficial as the result ofcompensation of CSF acidosis but only atvery low arterial pCO2 (20-25 mmHg) that isnow proposed to produce severe cerebralvasoconstriction and in some individualscerebral ischaemia. THAM, after intravenousadministration, equilibrates with the intracel-lular and extracellular spaces in the body aswell as with CSF. Evidence is accumulatingboth experimentally and in humans thatTHAM is at least as effective as mannitol inreducing experimental oedema in the brainand lowering ICP after head injury.93 94THAM reduces the demand for mannitol andCSF drainage. In the most recently publishedrandomised prospective clinical trial,94 a totalof 149 patients with severe head injury(Glasgow Coma Scale < 8) were randomlyassigned to either a control or a THAMgroup. Both groups of patients matched interms of clinical parameters including age,sex, number of surgical mass lesions, numberin each Glasgow scale stratum and the firstICP measurement. All patients were treatedby standard management protocols, intu-bated, mechanically ventilated, and main-tained in the PaCO, range of 32-35 mm Hgfor 5 days. THAM was administered as a 0-3M solution in an initial loading dose (bodyweight x blood acidity deficit, average 4-27cc/kg/hour) given over two hours, followed byconstant infusion of lml/kg/hour for five days.Outcome was measured at three, six and 12months post injury. Although analysis indi-

852

Management of raised intracranial pressure

cated no significant difference between thesetwo groups at three months, six months orone year, there was a difference regardingICP. The time that ICP was above 20 mmHgin the first 48 hours after injury was less inpatients treated with THAM. Also the num-ber of patients requiring barbiturate comawas significantly less in the THAM group(5-5% versus 18-4%). The authors concludedthat THAM ameliorated the deleteriouseffects of prolonged hyperventilation, wasbeneficial in ICP control and further study ofdose and timing of administration waswarranted.The cerebrovascular response to hyper-

capnia may be manipulated in other ways. Ithas been known since 1973 that the cere-brovascular CO2 response is blocked byindomethacin in doses that partly inhibitbrain cyclo-oxygenase activity in vivo.95Cerebral venous pressure is very significantlyreduced suggesting that ICP is reduced. Thatobservation was not used clinically because offear that indomethacin was inhibiting produc-tion of prostacyclin and that might be coun-terproductive. Certainly, cerebral oxygendelivery is seriously impaired whenindomethacin is given to very young animalsor to very pre-term infants undergoing treat-ment for patent ductus arteriosus.96 However,in five patients with injury with cerebral con-tusion and oedema in whom it was not possi-ble to control ICP by hyperventilation andbarbiturate sedation, indomethacin (bolusinjection of 30 mgs followed by 30 mg/hourfor 7 hours) reduced ICP below 20 mmHgfor several hours.97 CBF was reduced at twohours without any changes in cerebral arteri-ovenous oxygen or lactate differences. Rectaltemperature also fell from 38-6 to 37T3°C.Hence a more substantial trial ofindomethacin appears warranted, perhapsavoiding young children until further experi-ence is accumulated. Inhibition of nitric oxidesynthesis also blocks the cerebrovascular CO,response but may also increase focal cerebralinfarction in the rat middle cerebral occlusionmodel.

Continuous CSF drainage and surgicaldecompressionExternal ventricular drainage via a catheter orreservoir is a rapid procedure in an emergen-cly in a patient with hydrocephalus.Biventricular drainage is required for 3rd ven-tricular lesions that occlude both Foramenaof Munro. Colloid cysts are best dealt with asthe primary procedure unless the patient is inextremis. Patients with communicatinghydrocephalus or benign intracranial hyper-tension may be temporarily controlled bylumbar drainage through an indwellingcatheter. It is unkind, unnecessary and lesseffective to use repeated lumbar punctures. Itis becoming recognised that permanent CSFdrainage via lumbar peritoneal shunts may becomplicated by secondary descent of thecerebellar tonsils in patients with no previousevidence of a Chiari malformation.98 In allcases of external drainage, CSF should be

drained gradually against a positive pressureof 15-25 cm H,O to avoid unrestraineddrainage. In the case of a posterior fossatumour, upward coning may be precipitatedif the supratentorial ventricles are drained tooprecipitately. In patients with a hemispheremass causing midline shift and contralateralhydrocephalus, drainage of that ventricle maymake the shift worse. In patients with diffusebrain swelling, the ventricles are small andnot always easy to cannulate. Stereotactictechniques are useful but not appropriate inan emergency. Even where ICP is controlledby drainage against a pressure 15-25 cmsH2O, such a ventricular catheter readilybecomes blocked and is seldom a satisfactorytechnique per se. CSF drainage alone is theoptimal method of controlling intracranialhypertension in patients with subarachnoidhaemorrhage where the cause is often distur-bance of the CSF circulation but there isprobably an increased risk of rebleeding. CSFdrainage is used as a diagnostic technique toassess patients in the poorer SAH grades.When they improve early surgery should beconsidered.

Removal of bone flaps or subtemporaldecompressions are performed much less fre-quently nowadays." Patients with largemeningiomas may have a smoother post-operative course if a flap is removed electivelyat the end of the operation rather than as anemergency a few hours later. Benign intracra-nial hypertension can be treated by a combi-nation of optic nerve sheath fenestration andlumbar peritoneal or cisterno-peritonealshunting: subtemporal decompressions arerarely indicated. Babies with complex formsof craniosynostosis may require craniofacialsurgery to expand the volume of the skull. Slitventricle syndrome for shunt-induced CSFoverdrainage may be managed by use ofsiphon control devices or programmablevalves-subtemporal decompression is sel-dom required. This procedure was sometimesfollowed by temporal lobe epilepsy.

There is a very restricted place for decom-pressive craniotomy following head injury andthere is the potential to do considerableharmn.9'00 With a very tight brain, openingthe dura induces herniation through thedefect. Cerebral venous drainage from theherniated brain obstructs and further brainswelling ensues with infarction. Experi-mentally, craniectomy facilitates formation ofhydrostatic brain oedema as might beexpected from consideration of Starling'sequation.'01 Craniectomy may be consideredin young patients without evidence of diffuseaxonal injury (high Glasgow Coma Score onadmission) and evidence of diffuse swelling.'02For paediatric encephalopathies, Kirkhammakes a case for performing decompressionearlier rather than later and certainly beforethe EEG disappears.'03

Steroids, free radical scavengers and thelazaroidsThe mechanism of the remarkable effect ofglucocorticoids such as dexamethasone on

853

Pickard, Czosnyka

focal, relatively chronic cerebral lesionsremains incompletely understood. Patientsdeteriorating with a cerebral tumour or anabscess rapidly improve within 24 hours. It isnot yet proven whether steroids help trau-matic cerebral contusions. ICP waves andcompliance improve together although meanICP and water content take days longer tosubside. Brain biopsy for tumour is muchsafer after at least three days of dexametha-sone (10-20 mgs loading dose; 4 mgs fourtimes a day thereafter) particularly whencombined with stereotactic biopsy tech-niques. Care should be taken to counselpatients about the side effects of steroids evenwith short courses.Much controversy has surrounded the use

of very high dose steroids in head injury butcareful controlled trials have shown no bene-fit and in one study the outcome of the treat-ment group was worse.'04-108 However, evenhigher doses started within a few hours ofinjury are currently under scrutiny. Thisrationale is based on trials in spinal cordinjury of methylprednisolone (30 mgs/kg/day)which showed a modest benefit in the groupwhere it was started within eight hours ofinjury.One purported mechanism of action for

steroids involves lipid peroxidation and freeradicals.'09110 Oxygen is needed for aerobiclife but it has toxic properties. All organismsare subject to oxidative stress as up to 2% ofoxygen consumed by the brain, for example,is used to form semi-reduced oxygen interme-diates: superoxide, hydrogen peroxide andhydroxyl free radicals. These may be used aspart of normal biochemistry or if the safetymechanisms fail-superoxide dismutase,catalase, glutathione peroxidase, glutathione,vitamin E and ascorbate-then such reactiveoxygen species may attack nucleic acids, pro-teins, carbohydrates and particularly lipids inthe brain. Ferrous iron from blood clots isalso active along with such reactive oxygenspecies. Cerebrovascular effects of acutehypertension and subarachnoid haemorrhagemay involve free radical mechanisms damag-ing the endothelium. Non-glucocorticoidsteroid analogues of methylprednisolone aswell as methylprednisolone itself weaklyinhibit lipid peroxidation. The 21-amino-steroids (the antioxidant family known as thelazaroids) are potent inhibitors of lipid perox-idation and have a vitamin E sparing effect."'Various experimental models of head andspinal injury and focal or global ischaemiahave shown a variable degree of protectionafter treatment with the lazaroid U-74006Fso that it is now undergoing large scale clini-cal trials in head injury and subarachnoidhaemorrhage. Recently, the steroid compo-nent of the lazaroid molecule has beenreplaced by the anti-oxidant ring structure ofvitamin E. U78517F has greater in vitro lipidanti-oxidant properties than U-74006F. It isinteresting that the iron chelator desferriox-amine may be helpful in treating the coma ofcerebral malaria and experimental vasogenicoedema. Early results in severe head injurieswith the oxygen radical scavenger-poly-

ethylene glycol conjugated superoxide dismu-tase-have recently been reported but a muchlarger trial is required to establish efficacy."12

Cerebral metabolic depressants:excitotoxic amino acid antagonistsThe cerebral metabolic depressant effect ofdeep hypothermia is now seldom used exceptduring cardio-pulmonary bypass and total cir-culatory arrest. Such a technique may beused for complex basilar aneurysms whereinterventional radiological techniques areinappropriate. Brain energy metabolism isdepressed more conveniently by hypnoticagents including barbiturates, etomidate,propofol, althesin and gamma hydroxybu-tyrate. Unfortunately all such agents haveside effects, the most relevant of which is sys-temic hypotension, often compounded bydehydration or hypovolaemia. It is essentialto maintain a normal arterial blood pressureand not allow CPP to fall. Central venouspressure monitoring is required. One factormaintaining CPP in patients with raised ICPmay be the Cushing mechanism-lower theICP and CPP falls.

Hypnotic agents depress cerebral oxidativemetabolism and hence lower CBF, CBV andICP. However, cerebral electrical activity andnormal coupling mechanisms between metab-olism and flow must be present if barbituratesare to lower ICP.14'542 Normal flow metabo-lism coupling mechanisms may be assessedby the cerebrovascular response to CO2.Short-term protection during aneurysmalsurgery with barbiturate or propofol is widelyused. Synergy with even moderate hypother-mia may be helpful provided MAP is main-tained.42 After initial reports of theeffectiveness of short acting barbiturates inlowering ICP after head injury, three con-trolled trials have failed to show any overallsignificant improvement of outcome orreduction in number of patients dying withintracranial hypertension."11l16 Such trialsinvolved heterogeneous groups of patients,however, and a treatment benefit in a subgroup may have been missed.

In the UK, althesin has been withdrawneven though its idiosyncratic allergic prob-lems would not have been a contra-indicationin the intensive care environment. Etomidateblocks steroidogenesis but it is apparently stillused in the USA as an intraoperative protec-tion agent by combining it with dexametha-sone postoperatively for a few days."17Gammahydroxybutyrate drops blood pres-sure and its administration involves a consid-erable sodium load."18 Propofol(di-isopropylphenol) is widely used but carehas to be taken to avoid hypotension.Propofol also has free radical scavengingeffects.' 9 The ideal hypnotic agent awaitsdevelopment. In the patient with cardiovascu-lar instability, lignocaine (1 -5mg/kg iv) mayhave a place in lowering ICP.'20 This dose isas effective as thiopentone (3mg/kg iv).

Epilepsy has long been known to increaseICP and increase the risk of cerebralischaemia as a result of a massive increase in

854

Management of raised intracranial pressure

cerebral electrical activity and oxidativemetabolism: both metabolic demand andCPP are embarrassed. Seizures must betreated aggressively. Over the last decade,however, a more focal phenomenon has beenrevealed: inappropriate hypermetabolism insmall areas of the brain in association withlocal release or failed reuptake of excitotoxicamino acids such as glutamate.'2' 122 Forexample, subdural haemorrhage in the rat isaccompanied by increased glutamate concen-trations in the hippocampus with increasedlocal cerebral glucose utilisation and lateischaemic brain damage. An NMDA antago-nist will protect against such damage'23 just asNMDA antagonists reduce delayed neuronalloss and the volume of cerebral infarctionafter middle cerebral artery occlusion, pro-vided such agents are given prophylacticallyor in some cases within an hour of occlu-sion.'24 125 The therapeutic window of oppor-tunity is very short unless it is surmised thatglutamate release occurs at various times aftera head injury, not just at the moment ofimpact. Such impressive experimental datahas led to Phase I and II trials of glutamatereceptor antagonists in patients after severehead injury. The results are awaited withkeen interest. Ironically, that old-fashionedtreatment for severe head injuries-rectalmagnesium chloride-is now known to be anoncompetitive NMDA receptor antagonistand reduces the infarct volume after middlecerebral occlusion in the rat.'26

Targetted therapyIt is clear that there are many causes of raisedICP. The Edinburgh school has attempted7to define which therapy should be selected foreach cause of intracranial hypertension. Theysuggest that hypnotic agents may be mostlogically targetted at younger patients withdiffuse congestive brain swelling with pre-served cerebral electrical activity, jugularvenous oxygen saturation over 75%, a pulse-respiratory ratio in the ICP trace of over 2,preserved CBF CO2 reactivity and absence ofa diastolic notch on TCD recordings of MCAvelocity. Mannitol may be best used inpatients of any age with focal lesions, a lowcerebral perfusion pressure and reasonablypreserved autoregulation. If arterial bloodpressure is low despite colloid then inotropicagents such as Dobutamine or dopamineshould be used. SAH with raised ICP is bestmanaged by CSF drainage accepting thatthere is probably an increased risk of rebleed-ing.

ChildrenThe management of raised ICP in childhoodmust take account of a number of factors.4 127The critical values for ICP, ABP and CPP arelower, the younger the child. The normal ICPin the newborn is probably of the order of2-4 mmHg. ABP at birth is about 40 mmHg,80/55 by one year and 90/60 during the earlyschool years. CPP rises from 28 mmHg at28-32 weeks gestational age of 37-5 mmHg

at normal full-term. In the neonate, muchlower CBF values may be tolerated forlonger. Many of the pathologies differ fromthe adult including birth asphyxia, posthaem-orrhagic ventricular dilatation, craniocerebraldisproportion and the many metabolic andinfective encephalopathies. Hyperaemia playsa greater role as a cause of raised ICP in chil-dren after head injury than in adults.'28 129 TheNIH Traumatic Data Bank of severe headinjuries revealed that diffuse brain swellingoccurs twice as often in children (aged 16years or younger) as in adults. A total of 53%of children with diffuse swelling died com-pared with a mortality rate of 16% in thosewithout. The skull may expand in childrenwhere fusion of the sutures has not occurred.Controlled trials of therapy in the variouspathologies are made difficult by the verysmall numbers of patients seen in each cen-tre. As with adults, there is a wide diversity ofopinion on the use of barbiturate coma,steroids and mannitol. Even fluid restrictionstill has its advocates in the neonates, and forinappropriate secretion ofADH. The dangersof fluid restriction, based on assuming thatSIADH is the common cause of hypona-traemia after intracranial insults, have beendocumented very clearly in studies of adultsafter SAH.

SummaryThis review has been written at an unfortu-nate time. Novel questions are being asked ofthe old therapies and there is an abundanceof new strategies both to lower ICP and pro-tect the brain against cerebral ischaemia. Inthe United Kingdom, the problem is toensure that appropriate patients continue tobe referred to centres where clinical trials ofhigh quality can be undertaken. One of thesuccess stories of the past decade has beenthe decline in the number of road accidentsas a result of seat belt legislation, improve-ments in car design and the drink/drivinglaws. Hence, fortunately there are fewerpatients with head injuries to treat and it iseven more important that patients are appro-priately referred if studies to assess efficacy ofthe new strategies are not to be thwarted. Thenihilistic concept that intensive investigationwith ICP monitoring for patients with diffusehead injury or brain swelling following evacu-ation of a haematoma or a contusion has noproven beneficial effect on outcome, requiresrevision. A cocktail of therapies may berequired that can be created only whenpatients are monitored in sufficient detail toreveal the mechanisms underlying their indi-vidual ICP problem.

Ethical problems may arise over howaggressively therapy for intracranial hyperten-sion should be pursued and for how long.There has always been the concern that cra-nial decompression or prolonged barbituratecoma may preserve patients but with unac-ceptably severe disability. Some patients maybe salvaged from herniating with massivecerebral infarction with the use of osmother-apy but is the outcome acceptable?"30 Similar

855

Pickard, Czosnyka

considerations apply to some children withmetabolic encephalopathies. Where such con-siderations have been scrutinised in patientswith severe head injury, the whole spectrumof outcomes appears to be shifted so that thenumber of severe disabilities and persistentvegetative states are not increased. However,it is important to be sensitive to such issuesbased on experience of the particular cause ofraised intracranial pressure in a given agegroup.

We are indebted to Dr J M Turner, Mr E Guazzo and Mr PKirkpatrick for their comments on the manuscript

1 Jennett B, Teasdale G. Management of head injuries.Philadelphia: F A Davies. 1981.

2 Miller JD, Jones PA, Dearden NM, Tocher JL. Progressin the management of head injury. Br J Surg 1992;79:60-64.

3 Pickard JD, Bailey S, Sanderson H, Rees M,Garfield JS. Steps towards cost-benefit analysis ofregional neurosurgical care. BMJ 1990;301:629-35.

4 Minns RA. Problems of intracranial pressure in child-hood. Clinics in developmental medicine 113/114.London: MacKeith Press, 1991:1-458.

5 Langfitt TW. Increased intracranial pressure. ClinicalNeurosurgery 1969;16:436-71.

6 Miller JD. Normal and increased intracranial pressure.In: Miller JD, ed. Northfield's surgery of the centralnervous system 2nd ed, London: Blackwell, 1987:7-57.

7 Miller JD, Dearden NM. Measurement, analysis and themanagement of raised intracranial pressure. In:Teasdale GM, Miller JD, eds. Current neurosurgeryEdinburgh: Churchchill Livingstone, 1992:119-56.

8 Saul TG, Ducker TB. Effect of intracranial pressuremonitoring and aggressive treatment on mortality insevere head injury. J Neurosurg 1982;56:498-503.

9 Welch K. The intracranial pressure in infants. JNeurosurg 1980;52:693-9.

10 Lundberg N. Continuous recording and control ofventricular fluid pressure in neurosurgical practice.Acta Psych Neurol Scand 1960;36,Suppl 149:1-193.

11 Avezaat CJJ, von Eijndhoven JHM, Wyper DJ.Cerebrospinal fluid pulse pressure and intracranialvolume-pressure relationships. 7 Neurol NeurosurgPsychiatry 1979;42:687-700.

12 Czosnyka M, Laniewski P, Batorski L, et al. Remarks onamplitude-pressure characteristic phenomenon. In:Hoff JT, Betz AL, eds. Intracranial pressure VII. Berlin:Springer Verlag 1989:255-9.

13 Obrist WD, Langfitt TW, Jaggi JL, et al. Cerebral bloodflow and metabolism in comatose patients with acutehead injury. Relationship to intracranial hypertension.J Neurosurg 1984;61:241-55.

14 Bingham RM, Procaccio F, Prior PF, Hinds CJ.Cerebral electrical activity influences the effects of eto-midate on cerebral perfusion pressure in traumaticcoma. BritJAnaes 1985;57:843-8.

15 Nordstrom CH, Messeter K, Sundbarg G, et al. Cerebralblood flow, vasoreactivity and oxygen consumptionduring barbiturate therapy in severe traumatic brainlesions. J Neurosurg 1988;68:424-3 1.

16 Rosner MJ, Becker DP. Origins and evolution of plateauwaves. Experimental observations and theoreticalmodel. J Neurosurg 1984;60:312-24.

17 Newell DW, Aaslid R, Stooss R, Reulen HJ. The rela-tionship of blood flow velocity fluctuations to intracra-nial pressure B waves. J Neurosurg 1992;76:415-21.

18 Kirkham FJ, Levin SD, et al. Transcranial pulsedDoppler ultrasound findings in brainstem death. JfNeurol Neurosurg Psychiatry 1987;50: 1504-1513.

19 Hassler W, Steinmetz H, Pirschel J. TranscranialDoppler study of intracranial circulatory arrest. J7Neurosurg 1989;71:195-201.

20 Marmarou A, Shulman K, Larmorgese J.Compartmental analysis of compliance and outflowresistance of the cerebrospinal fluid outflow system. JfNeurosurg 1975;43:523-34.

21 Czosnyka M, Pickard JD, Whitehouse H, Piechnik S.The hyperaemic response to a transient reduction incerebral perfusion pressure-a modelling study. ActaNeurochirurgica (Wien) 1992;115:90-7.

22 Marmarou A, Maset AL, Ward JD, et al. Contribution ofCSF and vascular factors to elevation of ICP inseverely head-injured patients. J Neurosurg 1987;66:883-90.

23 Czosnyka M, Harris NG, Pickard JD. CO2 cerebrovas-cular reactivity as a function of perfusion pressure-amodelling study. Acta Neurochirurgica 1993;121:159-65.

24 Guillaume J, Janny P. Manometrie intracranienne con-tinue. interet de la methode et premiers resultats.Revue Neurologique 195 1;84:131-42.

25 Katzman R, Hussey F. A simple constant-infusionmanometric test for measurement of CSF absorption. IRationale and method. Neurology 1970;20:534-44.

26 Borgesen SE, Gjerris F. The predictive value of conduc-tance to outflow of CSF in normal pressure hydro-

cephalus. Brain 1982;105:65-86.27 Czosnyka M, Batorski L, Laniewski P, Maksymowicz W,

Koszewski W, Zaworski W. A computer system for theidentification of the cerebrospinal compensatorymodel. Acta Neurochir (Wien) 1990;105:112-6.

28 Symon L, Dorsch NWC, Stephens RJ. Pressure waves inso-called low-pressure hydrocephalus. Lancet 1972;2:1291-2.

29 Pickard JD. Teasdale GM, Matheson M, Wyper DJ, etal. Intraventricular pressure waves-the best predictivetest for shunting in normal pressure hydrocephalus. In:Shulman K, et al, eds. Intracranial pressure IV. Berlin:Springer-Verlag 1980:498-500.

30 Laniewski P, Czosnyka M, Maksymowicz W.Continuous analysis of the intracranial pressure wave-form as a method of autoregulatory reserve assessment.ICP VIII. Berlin: Springer Verlag (in press).

31 Miller JD, Garibi J, Pickard JD. Induced changes in cere-brospinal fluid volume. Effects during continuousmonitoring of ventricular fluid pressure. Arch Neurol1973;28:265-9.

32 Aaslid R, Markwalder T-M, Nornes H. Non-invasivetranscranial Doppler ultrasound recording of flowvelocity in basal cerebral arteries. Jf Neurosurg 1982;57:769-74.

33 Newell DW, Aaslid R. Transcranial doppler. New York:Raven Press, 1992.

34 Martin JL, Perry S, Pickard JD. Cerebral blood flow andDoppler flow velocity: different responses to threevasodilators. Jf Cereb Blood Flow Metab 1991;11:suppl2:S455.

35 Chan KH, Miller JD, Dearden NM, Andrews PJD,Midgley S. The effect of cerebral perfusion pressureupon changes in middle cerebral artery flow velocityand jugular venous bulb oxygen saturation after severehead injury. JfNeurosurg 1992;77:55-61.

36 Nelson RJ, Czosnyka M, Pickard JD, Maksymowicz W,Perry S, Martin JL, Lovick AHJ. Experimental aspectsof cerebrospinal hemodynamics: the relationshipbetween blood flow velocity waveform and cerebralautoregulation. Neurosurgery 1992;31:705-10.

37 Chan KH, Dearden NM, Miller JD. The significance ofposttraumatic increase in cerebral blood flow velocity:a transcranial Doppler ultrasound study. Neurosurgery1992;30:697-700.

38 Chan KH, Dearden NM, Miller JD, Midgley S, PiperIR. Transcranial Doppler waveform differences inhyperemic and nonhyperemic patients after severe headinjury. Surg Neurol 1992;38:433-6.

39 Chan KH, Miller JD, Dearden NM. Intracranial bloodflow velocity after head injury: relationship to severityof injury, time, neurological status and outcome. JfNeurol Neurosurg Psychiatry 1992;55:787-91.

40 Schalen W, Messeter R, Nordstrom CH. Cerebralvasoreactivity and the prediction of outcome in severetraumatic brain lesions. Acta Anaesthesiol Scand1991;35:113-22.

41 Reid A, Marchbanks RJ, Martin R, Pickard JD, BatemanN, Brightwell R. Mean intracranial pressure monitor-ing by an audiological technique-a pilot study. JNeurol Neurosurg Psychiatry 1989;52:610-2.

42 Campkin TV, Turner JM. Neurosurgical anaesthesia andintensive care. 2nd ed. London: Butterworth, 1986.

43 Robertson CS, Narayan RH, Gokaslan ZL, et al.Cerebral arteriovenous oxygen difference as an esti-mate of cerebral blood flow in comatose patients. JNeurosurg 1989;70:222-30.

44 Wyatt JS, Cope M, Delpy DT, Wray S, Reynolds EOR.Quantification of cerebral oxygenation and haemody-namics in sick newborn infants by near infrared spec-troscopy. Lancet 1986;2:1063-6.

45 Jennett B, Teasdale G, Galbraith S, Pickard JD, et al.Severe head injuries in three countries. Jf NeurolNeurosurg Psychiatry 1977;40:291-8.

46 Miller JD, Butterworth JF, Gudeman SK, et al. Furtherexperience in the management of severe head injury. JNeurosurg 1981;54:289-99.

47 Ropper AH, O'Rourke D, Kennedy SK. Head position,intracranial pressure and compliance. Neurology 1982;32:1288-91.

48 Rosner MJ, Coley IB. Cerebral perfusion pressure, ICPand head elevation. JNeurosurg 1986;65:636-41.

49 Feldman Z, Kanter MJ, Robertson CS, et al. Effect ofhead elevation on ICP, CPP and CBF in head injuredpatients. J Neurosurg 1992;76:207-1 1.

50 Kanter MJ, Robertson CS, Sheinberg MA, et al.Changes in cerebral hemodynamics with head elevatedvs head flat. In: Avezaat C, ed. Intracranial pressure VIII(In press).

51 Hasan D, Vermeulen M, Wijdicks EFM, et al. Effect offluid intake and hypertensive treatment on cerebralischaemia after subarachnoid haemorrhage. Stroke1989;20:1511-5.

52 Nelson RJ, Roberts J, Rubin C, Walker V, Ackery DM,Pickard JD. Association of hypovolaemia after SAHwith computer tomographic scan evidence of raisedintracranial pressure. Neurosurgery 1991;29:178-82.

53 Marmarou A, Anderson RL, Ward JD, et al. Impact ofICP instability and hypotension on outcome in patientswith severe head trauma. JNeurosurg 1991;75:S59-66.

54 Muizelaar JP. Induced arterial hypertension in the treat-ment of high intracranial pressure. In: Hoff JT, BetzAL, ed. Intracranial pressure VII. Springer-Verlag,1989:508-10.

55 Rosner MJ, Rosner SD. Cerebral perfusion pressuremanagement of head injury. In: Avezaat C, ed.

856

Management of raised intracranial pressure

Intracranial pressure VIII. (In press)56 Shimoda M, Oda S, Tsugane R, Sato 0. Intracranial

complications of hypervolemic therapy in patients withdelayed ischemic deficit attributed to vasospasm. JNeurosurg 1993;78:423-9.

57 Fitch W, Pickard JD, Tamura A, Graham DI. Effects ofhypotension induced with sodium nitroprusside on thecerebral circulation before, and one week after, thesubarachnoid injection of blood. .7 Neurol NeurosurgPsychiatry 1988;51:88-93.

58 Nelson PB, Seif SM, Maroon JC, Robinson AG.Hyponatraemia in intracranial disease: perhaps not thesyndrome of inappropriate secretion of antidiuretichormone (SIADH). _7 Neurosurg 198 1;55:938-4 1.

59 Nelson RJ, Perry S, Burns ACR, Roberts J, Pickard JD.The effects of hyponatraemia and subarachnoid haem-orrhage on the cerebral vasomotor responses of therabbit. .7 Cereb Blood Flow Metab 199 1;11:661-6.

60 Myers RE. Anoxic brain pathology and blood glucose.Neurology 1976;34:345.

61 Marie C, Bralet J. Blood glucose level and morphologicalbrain damage following cerebral ischaemia.Cerebrovascular and Brain Metab Reviews 1991 ;3:29-38.

62 Reed DJ, Woodbury DM. Effect of hypertonic urea oncerebrospinal fluid pressure and brain volume. 7 Physiol(Lond) 1963;164:252-64.

63 Pappius HM, Dayes LA. Hypertonic urea-its effect onthe distribution of water and electrolytes in normal andedematous brain tissue. Arch Neurol 1965;13:395-402.

64 Ravassin P, Abou-Madi M, Archer D, Chiolero R,Freeman J, Trop D, de Tribolet N. Changes in CSFpressure after mannitol in patients with and withoutelevated CSF pressure. _7 Neurosurg 1988;69:869-76.

65 James HE, Harbaugh RD, Marshall LF, Shapiro HM.The response to multiple therapeutics in experimentalvasogenic edema. In: Shulman K, Marmarou A, MillerJD, et al, eds. ICP IV. Berlin, Springer-Verlag. 1980;272-6.

66 Rosenberg GA. Brain fluids and metabolism. Oxford:Oxford University Press, 1990.

67 Inao S, Kuchiwaki H, Wachi A, et al. Effect of mannitolon intracranial pressure-volume status and cerebralhaemodynamics in brain oedema. Acta Neurochirurgica1990;Suppl5l :401-3.

68 Bell BA, Smith MA, Kean DM, et al. Brain water mea-sured by magnetic resonance imaging: correlation withdirect estimation and change following mannitol anddexamethasone. Lancet 1987;1:66-9.

69 Kaufmann AM, Cardoso ER. Delayed cerebral accumu-lation of mannitol in vasogenic edema. In: Avezaat C,ed. Intracranial pressure VIII. (In press).

70 Burke AM, Quest DO, Chien S, et al. The effects ofmannitol on blood viscosity. . Neurosurg 1981;55:550-3.

71 Miller JD, Leech P. Effects of mannitol and steroidtherapy in intracranial volume pressure relationships..7Neurosurg 1975;42:274-81.

72 Muizelaar JP, Wei EP, Kontos HA, Becker DP.Mannitol causes compensatory cerebral vasoconstric-tion and vasodilatation in response to blood viscositychanges. JNeurosurg 1983;59:822-8.

73 Auer LM, Haselsberger K. Effect of intravenous manni-tol on cat pial arteries and veins during normal andelevated intracranial pressure. Neurosurgery 1987;21:142-6.

74 Harrison MJG. Influence of haemocrit in the cerebralcirculation. Cerebrovascular and Brain Metab Reviews1989;1:55-67.

75 Brown MM, Wade JPH, Marshall J. Fundamentalimportance of arterial oxygen content in the regulationof cerebral blood flow in man. Brain 1985;108:81-93.

76 Purves MJ. The Physiology of the cerebral circulation.Cambridge: Cambridge University Press, 1972.

77 Bruce DA, Langfitt TW, Miller JD, et al. Regional cere-bral blood flow, intracranial pressure, and brain metab-olism in comatose patients. _7 Neurosurg 1973;38:131-44.

78 Johnston IH, Harper AM. The effect of mannitol oncerebral blood flow. An experimental study. .7Neurosurg 1973;38:461-71.

79 Mendelow AD, Teasdale GM, Russell T, et al. Effect ofmannitol on cerebral blood flow and cerebral perfusionpressure in human head injury. _7 Neurosurg 1985;63:43-8.

80 Jafar JJ, Johns LM, Mullan SF. The effect of mannitol oncerebral blood flow..7 Neurosurg 1986;64:754-9.

81 Muizelaar JP, Lutz HA, Becker DP. Effect of mannitolon ICP and CBF and correlation with pressureautoregulation in severely head-injured patients. _7Neurosurg 1984;61:700-6.

82 Meyer FB, Anderson RE, Sundt TM, Yaksh TL.Treatment of experimental focal cerebral ischaemiawith mannitol. Assessment by intracellular pH, corticalblood flow and EEG. JTNeurosurg 1987;66:109-15.

83 Takagi H, Saito T, Kitahara T, Morii SM, Ohwada J,Yada K. The mechanism of the ICP reducing effect ofmannitol. In: Ishii S, Nagai H, Brock M, eds. ICP V.Berlin: Springer-Verlag, 1983:729-33.

84 Smith HP, Kelly DL, McWhorter JM, et al. Comparisonof mannitol regimes in patients with severe head injuryundergoing ICP monitoring. .7 Neurosurg 1986;65:820-4.

85 Roberts PA, Pollay M, Engles C, Pendleton B, ReynoldsE, Stevens FA. Effect on intracranial pressure of

furosemide combined with varying doses and adminis-tration rates of mannitol. . Neurosurg 1987;66:440-6.

86 Smedema RJ, Gaab MR, Heissler HE. A comparisonstudy between mannitol and glycerol therapy in reduc-ing ICP. In: Avezaat, ed. Intracranial pressure VIII. (Inpress).

87 Muizelaar JP, van der Poel HG, Li Z, Kontos HA,Levasseur JE. Pial arteriolar vessel diameter and CO2reactivity during prolonged hyperventilation in the rab-bit. J Neurosurg 1988;69:923-7.

88 Ward JD, Choi S, Marmarou A, et al. Effect of prophy-lactic hyperventilation on outcome in patients withsevere head injury. ICP VII. In: Hoff IT, Betz AL, eds.Berlin: Springer-Verlag, 1989:630-4.

89 Enevoldsen EM, Cold G, Jensen FT, et al. Dynamicchanges in regional CBF, intraventricular pressure,CSF pH and lactate levels during the acute phase ofhead injury. J Neurosurg 1976;44:191-214.

90 De Salles AAF, Kontos HA, Becker DP, et al. Prognosticsignificance of ventricular CSF lactic acidosis in severehead injury. _7 Neurosurg 1986;65:615-24.

91 Akiota T, Ota K, Matsumato A, et al. The effect ofTHAM on acute intracranial hypertension. An experi-mental and clinical study. In: Beks JWF, Bosch DA,Brock M, eds. Intracranial pressure III. Berlin: Springer-Verlag, 1976:219-33.

92 Gordan E, Rossanda M. Further studies in cerebrospinalfluid acid-base status in patients with brain lesions.Acta Anaesth Scand 1970;14:97-109.

93 Gaab MR, Seegers K, Goetz C. THAM (tromethamine,"tris-buffer"): effective therapy of traumatic brainswelling. In: HoffJT, Betz AL, eds. Intracranial pressureVII. Berlin: Springer-Verlag, 1989:616-9.

94 Wolf AL, Levi L, Marmarou A, et al. Effect of THAMupon outcome in severe head injury: a randomizedprospective clinical trial. _7 Neurosurg 1993;78:54-9.

95 Pickard JD, Mackenzie ET. Inhibition of prostaglandinsynthesis and the response of baboon circulation tocarbon dioxide. Nature New Biology 1973;245:187-8.

96 Edwards AD, Wyatt JS, Richardson C, et al. Effects ofindomethacin on cerebral haemodynamics in verypreterm infants. Lancet 1990;335:1491-5.

97 Jensen K, Ohrstrom J, Cold GE, Astrup J. The effects ofindomethacin on intracranial pressure, cerebral bloodflow and cerebral metabolism in patients with severehead injury and intracranial hypertension. ActaNeurochir (Wien) 1991;108:116-21.

98 Sullivan LP, Stears JC, Ringer SP. Resolution ofsyringomyelia and chiari malformation by ventriculo-atrial shunting in a patient with pseudotumour cerebriand a lumbo-peritoneal shunt. Neurosurgery 1988;22:744-7.

99 Editorial. Cranial decompression. Lancet 1988; 1: 1204.100 Gower DJ, Lee KS, McWhorter JM. Role of subtempo-

ral decompression in severe closed head injury.Neurosurgery 1988;23:417-22.

101 Umezewa H, Shima K, Chigasaki H, Ishii S. Effects ofthe pressure gradient on hydrostatic brain edema. In:Hoff JT, Betz AL, eds. Intracranial pressure VII. Berlin:Springer-Verlag, 1989:953-6.

102 Rittierodt M, Gaab MR. Traumatic brain swelling andoperative decompression: a prospective investigation.In: Avezaat C, et al, ed. Intracranial pressure VIII.Berlin: Springer-Verlag, (In press).

103 Kirkham FJ, Neville BGR. Successful management ofsevere intracranial hypertension by surgical decompres-sion. Developmental Medicine Child Neurol 1986;28:506-9.

104 Cooper PR, Moody S, Clark WK, et al. Dexamethasoneand severe head injury: a prospective double blind trial.. Neurosurg 1979;51:307-16.

105 Saul TF, Ducker TB, Salomon M, et al. Steroids insevere head injury. A prospective randomized clinicaltrial. _7 Neurosurg 1981;54:596-600.

106 Braakman R, Schouten HJD, Dishoeck BMV,Minderhoud JM. Megadose steroids in severe headinjury: results of a prospective double blind clinicaltrial. J Neurosurg 1983;58:326-30.

107 Giannotta SL, Weiss MH, Apuzzo MLJ, Martin E. Highdose glucocorticoids in the management of severe headinjury. Neurosurgery 1984;15:497-501.

108 Dearden NM, Gibson JB, McDowall DG, et al. Effect ofhigh dose dexamethasone on outcome from severehead injury. .7 Neurosurg 1986;64:81-8.

109 Floyd RA, Camey JM. Protection against oxidative dam-age to CNS by a-Phenyl-tert-butyl Nitrone and otherspin-trapping agents: a novel series of nonlipid freeradical scavengers. In: Marangos PJ, Lal H, eds.Emerging strategies in neuroprotection, Boston: Birkauser,1992:252-72.

110 Halliwell B, Gutteridge JMC. Free radicals in biology andmedicine. Oxford: Clarendon Press, 1985.

111 Hall ED. Lazaroids: novel cerebroprotective antioxi-dants. In: Marangos PJ, Lal H. Emerging strategies inneuroprotection. Boston: Birkauser, 1992:224-37.

112 Muizelaar JP, Marmarmou A, Young HA, et al.Improving the outcome of severe head injury with theoxygen radical scavenger polyethylene glycol-conju-gated superoxide dismutase: a Phase II trial. .Neurosurg 1993;78:375-82.

113 Welsh FA, Sims RE, Harris VA. Mild hypothermia pre-vents ischemic injury in gerbil hippocampus. .7 CerebBlood FHow Metab 1990;10:557-63.

114 Schwartz M, Tator CH, Rowed DW, et al. The

857

Pickard, Czosnyka

University of Toronto Head Injury Treatment Study: aprospective randomized comparison of pentobarbitatand mannitol. Canad JNeurol Sci 1984;11:434-40.

115 Ward JD, Becker DP, Miller JD, et al. Failure of prophy-lactic barbiturate coma in the treatment of severe headinjury. JNeurosurg 1985;62:383-8.

116 Eisenberg HM, Frankowski RF, Condant CG, et al.High dose barbiturate control of elevated intracranialpressure in patients with severe head injury.Neurosurg 1988;69: 15-23.

117 Batjer HH. Cerebral protective effects of Etomidate:experimental and clinical aspects. Cerebrovasc BrainMetab Rev 1993;5:17-32.

118 Leggate JRS, Dearden NM, Miller JD. The effects ofgammahydroxybutyrate and thiopentone on ICP insevere head injury. In: Miller JD, Teasdale GM, et aleds. Intracranial pressure VI. Berlin: Springer-Verlag,1986:754-9.

119 Murphy PG, Myers DS, Davies MJ, et al. The antioxi-dant potential of propafol (2,6-diisopropylphenol). BritJAnaes 1992;68:613-8.

120 Bedford RF, Persing JA, Pobereskin L, Butler A.Lidocaine or thiopental for rapid control of intracranialhypertension? Anaesth Anaig 1980;59:435-7.

121 Meldrum B. Protection against ischaemic neuronal dam-age by drugs acting on excitatory neurotransmission.Cerebrovascular Brain Metab Reviews 1990;2:27-57.

122 Choi DW. Methods of antagonizing glutamate neurotox-icity. Cerebrovasc Brain Metab Rev 1990;2:105-47.

123 Inglis FM, Kuroda Y, Bullock R. Glucose hypermeta-

bolism after acute subdural haematoma is amelioratedby a competitive NMDA antagonist. Neurotrauma1992;9:75-84.

124 Gill R, Foster AC, Woodruff GN. MK801 is neuropro-tective in gerbils when administered during the post-ischaemic period. Neuroscience 1988;25:847-56.

125 Park CK, Nehls DG, Graham DI, Teasdale GM,McCulloch J. The glutamate antagonist MK801reduces focal ischaemic brain damage in the rat. AnnNeurol 1988;24:543-55.

126 Izumi Y, Roussel S, Pinard E, Seylaz J. Reduction ofinfarct volume by magnesium after middle cerebralartery occlusion in rats. Cereb Blood Flow Metab1991;11:1025-30.

127 Levene M. Neonatal neurology. Edinburgh, ChurchillLivingstone. 1987.

128 Bruce DA, Alavi A, Bilaniuk, et al. Diffuse cerebralswelling following head injuries in children: the syn-drome of "malignant brain edema". Neurosurg1981;54:170-8.

129 Aldrich EF, Eisenberg HM, Saydjari C, et al. Diffusebrain swelling in severely head-injured children.Neurosurg 1992;76:450-4.

130 Harrison MJG. Is shrinking the brain a good thing aftercerebral infarction? In: Warlow C, Garfield J, eds.Dilemmas in the management of the neurological patient.Edinburgh: Churchill Livingstone, 1984:62-73.

131 Pullen RGL, De Pasquale M, Cserr HF. Bulk flow ofcerebrospinal fluid into brain in response to acutehyperosmolality. AmerJ Physiol 1987;253:538-45.

858