J7ournal ofNeurology, Neurosurgery, and Psychiatry 1994;57:405-4 15

NEUROLOGICAL MANAGEMENT

Meningitis

H P Lambert

Three organisms predominate in causingcommunity-acquired bacterial meningitis,Neisseria meningitidis, Haemophilus influenzae,and Streptococcus pneumoniae. Meningococcalmeningitis is most common in childhood butalso affects adolescents and adults. A total of1302 cases were notified in England andWales in 1992, 71% of them caused by groupB strains against which no vaccine is yetavailable. Haemophilus influenzae meningitischiefly affects children less than five years old,although a few older patients are encoun-tered. In 1992, 1089 blood or CSF isolates,or both, were recorded and a seven-yearstudy in the Oxford region has estimatedthe cumulative risk by the fifth birthday as1:800.1 Invasive haemophilus infections arealready showing a gratifying reduction fol-lowing the introduction of HIB conjugatevaccines in October 1992.The pattern of pneumococcal meningitis is

more complex. It is seen at all ages but affectsparticularly the extremes of life and is alsoespecially associated with immune defectssuch as asplenia, as in sickle cell disease, andwith fractures of the skull or congenitaldefects allowing entry of bacteria to the CNS.This infection carries a high mortality, rarelyrecorded as less than 20%. The overall mor-tality from meningococcal and haemophilusmeningitis is 5-6% in Britain, but the pictureis often complicated by inclusion of acutemeningococcal septicaemia, with a muchhigher death rate than that seen in meningitisalone. The overall mortality from childhoodmeningitis over a 10-year period inNottingham2 was 11-6% and at least one in10 of the survivors suffered permanentsequelae; these consequences of meningitisare discussed later. In developing countriesthe incidence and mortality from meningitisare often much higher than in wealthiercountries.

Meningitis caused by other bacteria is lesscommon but important, as these infectionsoften cause serious difficulties in diagnosisand management. The need to consider otheragents has increased with the increasingnumber of patients with immune defects,especially HIV infection, and the associatedresurgence of tuberculosis. To these bacterialcauses must be added the problem ofcryptococcal meningitis in association withHIV infection. In nosocomial meningitis,

Gram-negative organisms such as Escherichiacoli, Klebsiella and pseudomonas have to beconsidered in the differential diagnosis, asalso does Listeria monocytogenes. A similar dis-tribution of organisms is seen in neonatalmeningitis, a special problem not dealt within this review.

Clinical diagnosisIn most patients the initial diagnosis ofmeningitis, or at least its possibility, isobvious from the combination of systemicand neurological features. Malaise, fever,severe headache, photophobia, and neckstiffness are common and sometimes con-sciousness is disturbed. The discussionwill therefore mainly be concerned with diffi-culties in diagnosis and, as so often, these aregreatest at the extremes of life. The problemof diagnosis in infancy is beyond the ambitof this review, but similar difficulties aresometimes encountered in the elderly.Consciousness may become rapidly depressedand soft neurological signs may lead diagnos-tic thoughts in the direction of a cerebro-vascular accident. Fortunately meningismis a sign not easily masked and is usually, butnot always, maintained even in comatosepatients.

Meningococcal disease may present inseveral atypical and deceptive ways. At oneextreme, fulminating meningococcal septi-caemia, the illness may run its short coursefrom onset to death with no element ofmeningitis. At the other end of the scale ofseverity patients, especially children, mayexperience a period of hours, or even days, offebrile illness with no meningeal featuresbefore the symptoms and signs of meningitisappear. This sequence has often led tomedicolegal problems when a claim ofdelayed or missed diagnosis may rest on thispoint. One especially deceptive presentationof meningococcal meningitis, seen in adoles-cents and young adults, is as acute mania,directing the diagnosis towards acute psy-chosis, possibly drug-induced.3 Neck stiffnessmay not yet have developed or may be impos-sible to test for.The rash of meningococcal septicaemia,

when present, is a valuable feature, as thecombination of petechial-purpuric rash andmeningitis is virtually diagnostic for this

69 Christchurch Road,London SW14 7ANH P LambertCorrespondence to:Professor Lambert

405 on 17 M

arch 2019 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.57.4.405 on 1 April 1994. D

ownloaded from

Lambert

organism. The rash may vary from a fewinconspicuous petechiae to extensive purpurawith skin necrosis. Non-purpuric rashes-andoccasionally petechial ones-are seen in somepatients with enteroviral rashes, but thisaetiology should never be assumed, as theearly rash of meningococcal disease may bemacular and unimpressive.

Focal neurological signs are relativelyuncommon in pyogenic meningitis in pre-viously healthy patients. Ocular palsiesusually resolve within days or weeks. In sharpcontrast, labyrinthine damage, which oftendevelops early in the illness, is usually per-manent. The frequency of convulsions variesgreatly between different forms of meningitis,and is much higher in infancy than in olderage groups. They are most frequent inpneumococcal meningitis, less common inhaemophilus meningitis, and uncommon inmeningococcal disease.4 Fits are also dis-tinctly uncommon in community-acquiredmeningitis in adult life, in patients with nounderlying neurological abnormality. Whenfocal signs, convulsions, or disturbance ofconsciousness are prominent in the clinicalpicture, the possibility of an encephalitis asthe primary diagnosis should be considered.

In community-acquired meningitis inpreviously healthy patients there is often nospecific aetiological clue such as rash, recentotitis media, or head injury. In such cases thepatient's age may be the sole diagnosticpointer, as haemophilus meningitis is rareafter the first few years of life, meningococcalmeningitis is the most common form atschool age and in young adults, whereas inthe elderly S pneumoniae becomes morefrequent. There are many exceptions to thesegeneralisations, especially now that many rea-sonably well people in the community mayhave various forms of immune suppression,notably HIV infection or treatment withimmunosuppressive drugs, and may presentwith uncommon forms of meningitis. Earlyand accurate laboratory diagnosis thereforeremains important and should be achieved asoften as possible.

LUMBAR PUNCTUREUntil recently, suspicion of meningitis wasaccepted as an almost invariable indicationfor lumbar puncture, although it was alwaysrecognised that it should be avoided ordeferred if there was suspicion of an alter-native diagnosis that might mimic meningitis,such as brain abscess, cerebral haemorrhage,encephalitis, or posterior fossa tumour. In thelast few years, however, concern about thepossibility of coning has led to a morecautious approach to lumbar puncture inmeningitis. The risk of coning is variouslyestimated. A recent retrospective survey, froman Australian paediatric referral centrereceiving a preponderance of complicated orseriously ill children, noted cerebral hernia-tion in 19 of 445 children (4.3%).5 Therewas a strong suggestion that lumbar punctureled to this complication in some patients,although it can certainly occur in patients

with meningitis when no lumbar puncturehas been done. On the other hand, there aregreat advantages in making an accurate andearly diagnosis, and a Gram-stained smear ofthe CSF deposit is still the most commonway in which this can be achieved. Certainly'blind' treatment is often successful but if theresponsible organism is unusual or antibiotic-resistant, or the patient's progress is unsatis-factory in any way, failure of identificationcreates an uncertain and unsatisfactorysituation.The dilemma can be resolved by defining

genuine risk factors for coning while accept-ing that lumbar puncture is normally indi-cated in suspected meningitis. A commoninjunction is to avoid lumbar puncture ifthere is suspicion of raised intracranialpressure, a distinctly unhelpful precept asintracranial pressure is raised in nearly everypatient with meningitis.6 More realistic guide-lines can be set using known clinical corre-lates of impending coning, namely coma orrapidly increasing depression of conscious-ness, focal neurological signs, and tonic orprolonged fits. Fits are common in childhoodmeningitis in some communities and are notan invariable contraindication to lumbarpuncture although, as Mellor points out in athoughtful review,7 it is sensible to deferlumbar puncture for 30 minutes after a fitbecause of the transient cerebral oedema thataccompanies it. Papilloedema is a contraindi-cation but is very rare in meningitis and itspresence should in any case indicate a widerdiagnostic sweep.

Another indication for avoiding or delayinglumbar puncture is unrelated to the questionof coning. This is the need for urgent treat-ment of patients with established or threat-ened bacterial shock (usually meningococcalsepticaemia). The window of opportunitycloses rapidly in these patients and treatmentshould on no account be delayed in order todo a lumbar puncture.CT or MR scans are usually normal or

mildly and non-specifically abnormal inmeningitis and are not generally helpful indetecting coning.7

TUBERCULOUS MENINGITISIn countries with a high prevalence of tuber-culosis, tuberculous meningitis acts as one ofthe markers of the frequency of infection, andthe diagnosis is in question in any patient,especially a child, with meningitis. Wheretuberculosis has declined the diagnosis ishard to bear in mind, and the problem iscompounded by the often deceptive ways inwhich this disease presents. Its importance isagain increasing with the spread of HIVinfection and the high prevalence in manycountries of the poor social conditions andcrowding that are associated with tuber-culosis. The relationship of success in treat-ment to the stage at presentation has rightlybeen stressed.

There is usually a prodromal illness ofdays or weeks, with low fever, malaise, andheadache before meningeal features develop.

406 on 17 M

arch 2019 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.57.4.405 on 1 April 1994. D

ownloaded from

Meningitis

The multiple pathology of tuberculousmeningitis, with basal arachnoiditis, vas-culitis, infarction, and obstructive hydro-cephalus leads to an immense variety ofpossible focal features; most frequent arecranial nerve palsies and papilloedema. Spinalinvolvement may take a number of differentforms, and rare presentations include amainly encephalopathic picture. It is impor-tant, however, to remember that tuberculousmeningitis may present as a meningeal illnesswith no specific features, so that the diagnosismust be entertained in any patient with non-purulent meningitis. Some persistent mythsmust be abandoned; the onset may be acute,the CSF glucose is within the normal range inabout 20% of cases, and the tuberculin testinitially negative in a similar proportion.Because of these protean manifestations,laboratory diagnosis is particularly important(see below), but unfortunately tubercle bacilliare seen in the CSF in only a few cases,depending on the technical skills available,and cultures are not always positive.Indicative features in the clinical presentationinclude slow onset, contact or family history,ethnic origin, or immigrant status from anarea of high prevalence, while a primary lungcomplex or any evidence of miliary tuber-culosis will obviously virtually establish thediagnosis.

CRYPTOCOCCAL MENING1TISCryptococcal meningitis also varies widely inits mode of onset. In immunocompromisedpatients the course tends to be more rapidthan in the immunocompetent but in bothgroups onset may be insidious, with generalmalaise, low fever, and headache, This subtleform of presentation is common now thatHIV infection is the most frequent type ofimmune abnormality preceding cryptococcalmeningitis.8 Indeed, features indicative ofCNS involvement may be entirely absent andthe indications for lumbar puncture must bemore widely set in these patients than in theimmunocompetent. Lumbar puncture inHIV-infected patients should be preceded byCT because of the possibility of a silent masslesion. A small proportion of patients withcryptococcal meningitis show focal neuro-logical signs, including early visual loss.9 Insome patients the course of the illness ismuch more protracted, extending over manymonths before a diagnosis is reached, andsometimes greatly fluctuating in severity.

LISTERIA MENINGITISThis is most frequently seen in the neonatebut also has a predilection for the immuno-compromised and the elderly. Presentation isusually as an acute purulent meningitis withno particular distinguishing features, but thesubtle presentations mentioned in the contextof cryptococcal meningitis are seen here also.Neurological signs and disturbance of con-sciousness are common with a mixed menin-gitic and encephalitic picture, which is morecommonly seen with listeria than with otherforms of bacterial meningitis. Listeria also

causes a range of other CNS conditions,including single or multiple brain abscess,diffuse encephalitis and, rarely, brain stemencephalitis.

Laboratory diagnosisThis topic can be discussed only briefly,chiefly to stress its importance and the needfor efficient and prompt acquisition of correctspecimens by the clinician and for goodliaison between ward and laboratory. Gramstaining of the centrifuged CSF depositremains the gold standard for early diagnosis,with occasional help from other specimenssuch as smears from skin lesions. Later,results of blood and CSF cultures provide themain diagnostic yield. Many methods of earlydiagnosis have been developed, the most use-ful of which are antigen detection methods,especially latex particle agglutination, whichcan be employed using CSF, serum, or urine.These techniques are not generally superiorto traditional staining, but have definite valuein reducing the diagnostic gap in patients whohave been given antibiotics before admission.

Diagnosis of tuberculous meningitis stilldepends mainly on traditional stainingmethods on the CSF deposit, but newtechniques are being developed, notably thepolymerase chain reaction, which maygreatly improve the low sensitivity of currentdiagnostic methods.'0

Diagnosis of cryptococcal meningitis restslargely on cryptococcal antigen detection inCSF (but not blood), on India ink pre-parations of CSF to detect capsulatedCryptococcus neoformans, and on culture usinglarge volumes of CSF.

Treatment (see Table 1 for doses)Treatment of the three common forms ofpyogenic meningitis has been greatly com-plicated by increasing prevalence of severaltypes of antibiotic drug resistance. Until thesechanges became widespread, antibiotic treat-ment of meningitis caused by these organismscould be accomplished using two safe, easilyadministered and cheap agents, benzylpeni-cillin (or ampicillin), and chloramphenicol.Benzylpenicillin is still appropriate formeningococcal and pneumococcal disease inmost countries, but haemophilus meningitis,and meningitis of uncertain aetiology mustnow often be treated with extended spectrumcephalosporins, a particular tragedy indeveloping countries in which meningitis iscommon and in which the cost of these com-pounds is beyond the reach of publiclyfunded health budgets. It is therefore impor-tant to remember that penicillin and chloram-phenicol remain agents of first choice in manyareas, and may indeed be the only agentsavailable. Table 1 gives the dosages of anti-biotics commonly used.

Meningococcal meningitis can be treatedwith equal efficacy by high dose penicillin orampicillin given intravenously or by chloram-phenicol, given by intravenous injection

407 on 17 M

arch 2019 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.57.4.405 on 1 April 1994. D

ownloaded from

Lambert

Table 1 Dosage of antibiotics commonly used in meningitis

Total dose per 24 hours

Adult Child Dose intervalAgent (g) (mg per kg) (hours) Route

Benzylpenicillin 14-4 180-300 4 IntravenousAmpicillin 12 200 4 IntravenousChloramphenicol 3 50-100 6 Intravenous

or oralCefotaxime 8 200 8 IntravenousCeftriaxone 4 80 24 IntravenousVancomycint 2 40 6 Intravenous

*Dosage unsuitable for neonates, tControl by blood levels required.

initially, and by mouth as soon as practicable.The role of cephalosporins was at first contro-

versial, as the earlier compounds were rela-tively ineffective, but the newer compounds,although most widely used in H influenzaemeningitis, are also of established efficacy inmeningococcal infections. An importantcaveat is the emergence of penicillin resis-tance in meningococci, especially in Spain,but also increasingly recorded elsewhere.The level of risk of encountering a resistantstrain that would necessitate a change frompenicillin or ampicillin must be kept underreview. One important difference betweenmeningococcal and other forms of meningitisis the relative ease with which the organismcan be eradicated. Although 10-14 daycourses of treatment have traditionally beenused in all forms of meningitis, shortercourses are fully adequate in meningococcaldisease. Successful trials have included dura-tions of seven, five, and even four days withpenicillin." Chloramphenicol is the bestchoice in patients allergic to penicillin. Infield conditions in tropical Africa, severalstudies have been made of treatment in one

or two doses using a long acting preparationof chloramphenicol in oil. For example, in a

large, randomised trial involving 528 patientsin Mali and Niger, the success of a two-dosescheme of this sort equalled that achieved byintravenous ampicillin given four times dailyfor five days, and was clearly much more

simple to administer. The mortality in allgroups in these trials was much higher thanthat found in the West, again demonstratingthe great impact of meningitis in these areas.'2

Penicillin is also still the best agent for thetreatment of pneumococcal meningitis, butpneumococci resistant to penicillin havebecome widespread and are now common

enough in several countries, including Spainand Hungary, to preclude the use of penicillinin treatment. The treatment of penicillin-resistant pneumococcal meningitis presents a

difficult and partly unsolved problem. Thechief alternative, chloramphenicol, may besuccessful but has failed to eradicate infectionin a number of cases. It has been shown, inwork from South Africa, that many penicillin-resistant pneumococci, although susceptibleto chloramphenicol on disc testing, requirehigh minimum bactericidal concentrations ofpenicillin, and that this form of partial resis-tance is associated with poor results of treat-ment.'3 Vancomycin, although a difficult drug

to use, has had some success, but a numberof treatment failures have led to the sugges-tion that it should be used in penicil-lin/chloramphenicol-resistant pneumococcalmeningitis only if high-dose cephalosporinshave failed or if the patient has anaphylacticreactions to fl-lactam agents.'4H influenzae meningitis has its main impact

in children less than four years old, but isseen occasionally in older children and adults.Ampicillin, formerly widely used alone ortogether with chloramphenicol, has now lostits role because resistance to its action is nowpresent in a large number of isolates, 15-25%in most countries. Chloramphenicol is stillthe best agent in many areas, but resistance isincreasing, and many ampicillin-resistantstrains are also resistant to chloramphenicol.For these reasons extended spectrumcephalosporins such as cefotaxime and ceftri-axone have become the chosen agents inhaemophilus meningitis or meningitis ofuncertain aetiology. Treatment may be modi-fied when laboratory results are available.Although active against the three commonpathogens,"5 cephalosporins have limited orno activity against some of the more unusualcauses of meningitis, notably listeria andsome of the less common Gram-negativeorganisms. If the patient has an underlyingdisease that places him or her at risk for theseunusual pathogens and treatment has to bestarted without microbiological information,the initial regimen should include ampicillintogether with the cephalosporin.

Antibiotic treatment before admissiondiminishes the prospects of obtaining apositive finding by CSF microscopy or cul-ture. Occasionally the CSF is altered moreprofoundly, with a lymphocytic picture. Ifpartly treated pyogenic meningitis is in ques-tion (as well as other diagnoses such as brainabscess and tuberculous meningitis), it wouldbe reasonable to begin provisional treatmentwith a cephalosporin in full dosage whileother investigations are in train.

TUBERCULOUS MENINGITISTuberculous meningitis stands apart fromother forms of meningitis in many importantways. Diagnosis can be difficult, the course ofthe disease variable and prognosis uncertain.These and other factors have led to a defi-ciency of the extensive controlled clinical tri-als that have established well validatedregimens for the treatment of pulmonary andsome forms of non-pulmonary tuberculosis.Treatment tends therefore to be based onlimited trial data, the empirical opinion ofphysicians with experience of the disease, andon knowledge of the pharmacokinetics ofantituberculous agents.'6 The broad relation-ship of prognosis to neurological deficitdemands that treatment should be started assoon as possible, even when there is stilluncertainty about the diagnosis. Fear thatdiagnosis will then be impossible to establishis largely unfounded, because it can often bemade from post-treatment specimens of CSF,or from later results of culture of pre- or

408 on 17 M

arch 2019 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.57.4.405 on 1 April 1994. D

ownloaded from

Meningitis

post-treatment specimens. Alternatively, adifferent diagnosis may later be established-for example, by virological findings that allowthe provisional treatment for tuberculosis tobe discontinued.The pharmacokinetics of antituberculous

agents in the CSF are well established.Pyrazinamide shows excellent penetrationinto the CSF. The CSF/serum ratio inChinese patients is about 75% two hoursafter the dose is given, and about 110% afterfive and eight hours. Isoniazid too penetrateswell, with concentrations in CSF approachingthose in serum. Rifampicin concentrations inCSF are 10-15% those in serum, repre-senting the unbound moiety of the drug.Streptomycin achieves adequate concentra-tion in the CSF after standard intramusculardosage only when the meninges are inflamed,and this is true also of ethambutol, givenorally. Ethionamide and prothionamide, bycontrast, penetrate the CSF well, whether ornot the meninges are inflamed, although theirvalue is limited by gastrointestinal and otherunwanted effects. Initial treatment shouldemploy at least a triple regimen of pyrazi-namide, isoniazid, and rifampicin. It is stilluncertain whether streptomycin is needed asan addition to this scheme. Workers in HongKong with great experience of the diseaserecommend its inclusion during the first twoor three months of treatment, and it shouldcertainly be used if rifampicin cannot beobtained because of its cost.'6 The three prin-cipal agents are all given by mouth in oncedaily dosage, or by gastric tube if the patientcannot swallow. The dose of pyrazinamide is35 mg/kg for a child, 2 g for an adult, that ofrifampicin 10 mg/kg. Isoniazid has customar-ily been given in higher dosages, 10 mg/kg,than are used in other forms of tuberculosis,but the conventional dose of 300 mg or 4-5mg/kg is probably adequate. Pyridoxine 10mg daily is given to prevent isoniazid neu-ropathy. Streptomycin is administered byintramuscular injection in a dose of 20 mg/kgto a maximum of 1 g daily but careful andcontinued attention should of course bepaid to renal function. Intrathecal treatmentcan no longer be recommended, althoughthere was evidence of its value in treatmentregimens which preceded the use ofrifampicin and pyrazinamide.The optimal duration of treatment is

unknown. Because the disease is so serious, ithas been customary to use long courses oftreatment. This is perhaps illogical because,despite the devastating pathology of the con-dition, the bacterial population concerned issmall compared with that found, say, in cavi-tating pulmonary tuberculosis. Most authorscontinue treatment for one year, but shortercourses have been used, for example, anine-month regLmen.The role of steroids has been debated for

decades, but rigorous analysis by controlledtrial has never been achieved. A mixture ofopen studies, anecdotal evidence and possiblyshaky inference from the known effects ofsteroids on inflammatory processes has led to

their use in the more neurologically advancedgrades of the disease, in infants, in generallyvery ill patients, and in impending spinalblock, which may nevertheless progress evenduring their administration. The situationwith raised intracranial pressure in tuber-culous meningitis is more clear cut. CSFblock leading to hydrocephalus requiresurgent control by shunting. Clinicians work-ing in areas of high prevalence believe thatearly shunting is an important factor inimproving prognosis. Rising intracranialpressure from enlarging tuberculoma can becontrolled in most patients by high-dosedexamethasone, and surgical decompressionis rarely necessary.

CRYPTOCOCCAL MENINGITISBefore the advent of HIV infection andAIDS, cryptococcal meningitis was rarelyseen in Britain, and then mainly in associa-tion with defects of cellular immunity-forexample, in patients with lymphoma, thosereceiving substantial doses of steroids, and inpatients with sarcoid whether or not onsteroids. In some other countries a substantialnumber of patients with cryptococcal menin-gitis had no overt predisposing factors. Thispicture changed dramatically with the spreadof HIV infection and cryptococcal meningitisis now seen as one of the most common CNSinfections associated with AIDS. Treatmentis difficult and demanding. Failure andrelapse are common even in patients withoutHIV infection; in the presence of HIV, aswith other infections, eradication is especiallydifficult and for this reason much thought hasbeen given to devising practical methods oflong-term prevention as well as to improvingtreatment of established infections.

Agents available for chemotherapy includeamphotericin in various forms, flucytosineand some of the more recently developedtriazole compounds, of which fluconazole hasbeen most extensively studied. Trials com-pleted before the onset of HIV established thevalue of amphotericin, despite its formidabletoxicity, and showed how the unwantedeffects could be partly mitigated by using thesynergy, demonstrable in vitro and in vivo,between this agent and flucytosine. A majormulticentre study showed that a combinationof amphotericin in a dose of 0 3 mg/kg dailytogether with flucytosine in a dose of 150mg/kg daily given for six weeks gave results atleast as good as with amphotericin alone at0*4 mg/kg per day for 10 weeks. The patientsgiven combination therapy showed superiorresults at a non-significant level by a numberof criteria; eight of 34 patients in the combi-nation group died compared with 15 of 32given amphotericin alone.'7

Since this trial the methods of usingamphotericin have been to some extentrefined, with a trend to give larger doses,0 5-0-7 mg/kg daily, or 1 0-1 2 mg/kg onalternate days. Duration of treatment mayalso be varied. Four weeks may prove ade-quate for patients with no adverse prognosticfactors. These factors include underlying

409 on 17 M

arch 2019 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.57.4.405 on 1 April 1994. D

ownloaded from

Lambert

immune defects (and therefore all those withcryptococcal meningitis superimposed on

HIV infection), neurological deficit, a highcryptococcal antigen titre (>1:32) in the CSFand a low (<20 mm3) CSF leucocyte count.Treatment may, however, fail even in patientswith no identifiable adverse features.The onerous nature and limited success

of these methods of treatment is even more

notable in AIDS-associated cryptococcalmeningitis, in which marrow suppression bydisease or by other drugs used in therapy maylimit or prevent the use of flucytosine. Thepotential of triazole compounds, with lowertoxicity and the advantage of oral administra-tion, is therefore especially interesting in thisgroup of patients. After favourable earlyresults, a formal comparison of fluconazoleand amphotericin was made in a randomised,multicentre trial.'8 Fluconazole was given in a

dose of 200 mg daily, amphotericin at0A4-0 5 mg/kg daily. The gravity of this infec-tion is clearly revealed, with successful treat-ment in only 40% of 63 patients in theamphotericin group and in 34% of 131patients given fluconazole. There was no sig-nificant difference in overall mortality, butdeaths were more frequent in the fluconazolegroup during the first two weeks of treatmentand CSF culture reverted to negative more

slowly than in the amphotericin group.

Comments on this trial emphasised thatdosage of both drugs might now be thoughttoo low. Other schemes being evaluated use a

higher (0 7 mg/kg) dose of amphotericin withor without flucytosine at 100 mg/kg, either as

definitive treatment or for the first two weeksfollowed by fluconazole. Other studies involvehigher doses: 400 mg or 800 mg of flucona-zole. Favourable results are also beingreported with itraconazole,'9 although thiscompound does not reach the CSF indetectable concentrations.The prospects for preventing recurrence of

cryptococcal meningitis after initial successfultreatment are more promising. A comparisonof fluconazole 200 mg daily, with ampho-tericin 100 mg weekly showed clearly supe-rior results for fluconazole, with relapse in 2%compared with 18% in the amphotericingroup.20 The high relapse rate in patients withAIDS and other forms of immune suppres-

sion make it essential to use long-term pro-phylaxis following initial treatment. What isstill uncertain is whether, at least in areas

where cryptococcal meningitis is a common

feature of AIDS, primary prophylaxisshould be attempted. A retrospective studyreported only one patient with cryptococcalmeningitis of 329 given daily fluconazole as

against 16 in 329 historical "controls". Thistype of primary prevention may come to bevitiated by the development of fluconazoleresistance, already emerging in Candidaalbicans infections.

ROLE OF STEROIDSThere is now abundant evidence that endoge-nously released factors play an importantrole in causing the inflammatory changes of

meningitis, and increasing interest in thepossibility that measures aimed at inhibitingthis response might be beneficial.

Although much detail remains to be filledin, the relevant processes can be summarisedbriefly. In the case of meningococci andH influenzae, the initiating factor is endo-toxin, a lipopolysaccharide component of thecell wall released in the form of vesicles fromthe bacteria. Similar processes are induced inpneumococcal infection by the release ofother cell wall components, mainly teichoicacid and peptidoglycan. After a time lagof a few hours, pro-inflammatory cytokinesare induced, including tumour necrosis factor(TNF), and interleukins (IL) 1, 6, and 8.These and other factors produced frommacrophages and from platelets have beenshown in experimental systems to induce thechanges of acute inflammation in the CNS.An additional complicating factor is that neu-trophils, although an important component ofthe defence systems against pyogenic infec-tion, themselves contribute to the inflamma-tory process when, after adhesion andmigration, they degranulate and produce pro-inflammatory factors including reactive oxy-

21gen species.The link between inflammatory mediators

and the timing and intensity of CSF changesis well established in experimental work, andthere is increasing evidence of their relevanceto human meningitis. For example, the out-come of meningitis can be correlated withlevels of endotoxin and pro-inflammatorycytokines in the CSF.2223 These relationshipsare also clearly established in septicaemicmeningococcal infection, in which there is aclose relationship between plasma endotoxinlevels and prognosis, and between endotoxinlevels and levels of TNF-a, IL-1, and IL-6.24An especially taxing question is whether

lytic antibiotics, given to cure the infection,might themselves have an adverse effect bycausing rapid release of bacterial productsand thus provoking an increased inflamma-tory burst. In experimental models of menin-gitis it is clear that an increase of severalmediators, and a consequent inflammatoryburst, follows the administration of lyticantibiotics such as /?-lactams.2' This increasedinflammatory response can be significantlymitigated by dexamethasone if given at thesame time as, but not one hour later than, theantibiotic.26 Again, the evidence from humanmeningitis is necessarily more fragmentary. Inone study, eight children with H influenzaemeningitis had repeat lumbar puncture twoto six hours after their initial dose of antibi-otic, ceftriaxone. The second specimensshowed increased concentration of endotoxincorrelated with the decrease in viable bacte-ria, together with increase in lactate anddecrease in glucose concentration in theCSF. In some cases TNF concentration inthe CSF rose about four hours after startingtreatment.27 It is therefore possible thatdiminishing the endogenously mediatedinflammatory processes might be beneficialand that some or all of this benefit might

410 on 17 M

arch 2019 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.57.4.405 on 1 April 1994. D

ownloaded from

Meningitis

accrue from diminishing a possible adverseconsequence of essential treatment, thesudden lysis of large numbers of bacteria.

It is against this background that the cur-rent renewed interest in a role for steroids inpyogenic meningitis has arisen, after a20-year lapse since their first trial.2829 Thefirst favourable evidence of a beneficial effecton outcome was gained from two trialsinvolving 200 children, predominantly withhaemophilus meningitis.30 In one trial theantibiotic used was cefuroxime (not nowconsidered entirely satisfactory as thecephalosporin of choice in meningitis), in theother ceftriaxone. In both, patients wererandomly allocated to receive placebo ordexamethasone in a dose of 0415 mg/kg sixhourly for four days. The dexamethasonegroup showed a significant increase in thespeed with which abnormal CSF findingsresolved, and, at follow up, substantial hear-ing loss was found in 15 placebo and threedexamethasone recipients; hearing aids wereneeded in 12 and one respectively. A smallertrial of 60 patients gave similar, althoughnon-significant results.3' Notable reductionsin the IL-1 ,B concentration were also demon-strated, together with improved prognosis, indexamethasone-treated children. An openstudy in Egypt involved adults and childrenwith meningitis, alternate patients being givendexamethasone. Seven of 52 patients withpneumococcal meningitis given dexametha-sone died, compared with 22 of 44 controls.Sequelae were also less common in the dex-

32amethasone group.Two more substantial controlled trials in

childhood meningitis have taken account,in their design, of the importance of timing inexperimental models. Children in CostaRica33 were given placebo or dexamethasonein the same dose as in the previous studies(0- 15 mg/kg six hourly for four days) butbeginning 15-20 minutes before the first doseof the antibiotic, cefotaxime. At 12 hoursthe dexamethasone-treated group showedimprovement compared with the controls inCSF pressure and in inflammatory indicesand cytokine concentrations in the CSF. Atfollow up seven of 51 dexamethasone-treatedchildren (14%), and 18 of 48 controls (38%)had neurological or audiological sequelae(relative risk 3-8, 95% CI 1-3 to 11-5). ASwiss trial"4 differed in the dose and durationof dexamethasone administration, 0-4 mg/kg12 hourly for two days, starting 10 minutesbefore the first dose of antibiotic, ceftriaxone.Follow up at 3, 9, and 15 months showedsequelae in three of 60 dexamethasone recipi-ents (5%) and nine of 55 placebo recipients

Table 2 Neurological and audiological sequelae of meningitis. Trials ofdexamethasone

Neurological sequelae Moderate or severe(no. (%)) deafness (no. (%/))

Reference No. in study Dexamethasone Placebo Dexamethasone Placebo

30,31 260 3(4) 9(12) 11(9) 21(19)33 101 5(10) 15(31) 3(6) 7(16)34 115 3(5) 5(9) 3(5) 7(13)

(16%) relative risk 3-27, 95% CI 0 93 to11-47. Table 2 summarises the results ofthese important studies.The trials also showed few adverse effects

in the dexamethasone groups, mainly a higherincidence of gastrointestinal bleeding, usuallydetected only by investigation rather than bya significant clinical event. Moreover, cases ofviral meningitis inadvertently treated withdexamethasone showed no adverse effects. Itis reasonable to conclude that steroids, givenin high dosage for a short time, and initiatedearly and preferably shortly before startingantibiotics, have a beneficial effect on out-come in childhood haemophilus meningitis,with lesser but suggestive evidence of benefitin other forms and at other ages.

Several other modes of damping theinflammatory response and its adverse patho-physiological consequences have been suc-cessful in experimental systems. They includenon-steroidal anti-inflammatory agents,monoclonal antibodies against some of thecytokines involved, and antibody inhibitingleucocyte adhesion to endothelium. Nomethod of modulating the inflammatoryprocess, however, other than administrationof steroids, is yet available for use in humanmeningitis.

Other aspects ofmanagementMight any aspects of therapy other thanantibiotics and steroids help to mitigate thecontinued high mortality and residual mor-bidity associated with bacterial meningitis?Fits must, of course, be brought under con-trol as promptly as possible, and mannitolinfusions are of established value if signs ofincreasing intracranial pressure appear.Changes in the cerebral circulation in menin-gitis may be important, and certainly inexperimental models, the inflammatoryprocess leads to a complex interaction ofbrain oedema and raised intracranial pressurewith changes in cerebral blood flow andperfusion. Especially notable is loss of auto-regulation so that the cerebral blood flowbecomes passively responsive to the systemicblood pressure.'5 Few measurements of thecerebral circulation have been made inhuman meningitis, and these largely ininfants and children. Two important studiesby non-invasive techniques are notable. Gohand Minns36 made serial measurements ofcerebral blood flow velocity using transcranialDoppler ultrasound. They found an increasein flow velocity as meningitis resolved with adecrease in the final resistance index, suggest-ing a decrease in cerebral perfusion duringthe acute phase of the illness. Mannitol infu-sion, by reducing intracranial pressure,increased cerebral perfusion pressure with aresultant decreased resistance index andincrease in blood flow velocity. Anotherstudy, involving seriously ill children,employed stable xenon CT.'7 This showeddiminished blood flow in only a fewpatients (five as against 18 with normalcerebral blood flow) but did reveal marked

411 on 17 M

arch 2019 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.57.4.405 on 1 April 1994. D

ownloaded from

Lambert

regional variations in flow changes. Deter-minations in artificially ventilated patients atdifferent pCO, tensions showed that hyper-ventilation to low pCO, tension could reduceregional blood flow below ischaemic thresh-olds. In these and other studies cerebralautoregulation appeared to be generally pre-served and pressure passivity was only seen insome very ill children with grossly abnormalneurological signs.Some tentative conclusions relevant to

clinical management may be drawn fromthese studies. It would seem sensible, inpatients seriously ill with meningitis, to aim atavoiding fluctuations of blood pressure ineither direction which might lead to changesin cerebral blood flow. In particular, hyper-ventilation with the aim of reducing intra-cranial pressure may sometimes provedisadvantageous and it may be preferable tomaintain the pCO, within the normal range.A common error in management is to

limit fluid intake unduly with the aim ofcontrolling inappropriate secretion of anti-diuretic hormone and, it is supposed, thusreducing the likelihood of cerebral oedema.It has been shown in childhood meningitis,however, that the observed high levels ofarginine vasopressin are an appropriateresponse to hypovolaemia and that levelsreturn to normal when fluid replacement isachieved.'8 Fluid deficits in meningitisshould be corrected with the appropriatereplacement fluids and normal maintenancerequirements provided.

CommunicationThe diagnosis of meningitis brings with itenormous fear and anxiety for many reasons.Will the patient die? Will the patient be braindamaged? Will other members of the familyor other people "catch" meningitis? All thesequestions must be discussed as often asnecessary. Especially difficult is the oft-expressed question, why did this particularperson develop meningitis? A simple accountof how these bacteria spread is often helpfulto the family and needs including with discus-sion of the patient's progress. To this must beadded the common anxiety of health servicestaff about the possibility of infection whentreating patients with meningococcal menin-gitis or septicaemia. The consultant in com-municable disease control must be informedas soon as possible so that preventive mea-sures (see below) can be rapidly brought intooperation. Close liaison with the laboratory isof course essential, to ensure the best possible

Table 3 Permanent hearing loss after bactenial meningitis

Permanent sensorineuralhearing loss ofany degree

Type of meningitis No. in study (no. (%))

Unselected 1175 113 (9-6)Haemophilus influenzae 876 99 (11*4)Meningococcal 398 30 (7-5)Pneumococcal 66 21 (31-8)

After Fortnum.

quality of specimens and thus the best chanceof identifying the causal organism and itsantibiotic sensitivity.

SequelaeMost previously healthy survivors of commu-nity-acquired meningitis recover completelybut an important minority is left with residuaranging from mild neurological or audio-logical defects to profound and lifelongdisability. Bacterial meningitis is the mostcommon cause of acquired sensorineuraldeafness. Its frequency has been analysed in athoughtful review by Fortnum39 acceptingonly published series with well-defined crite-ria avoiding the many possible confoundingfactors in such studies (table 3). Deafnesscomplicates meningitis in about 10% ofpatients, and will be bilaterally profound ortotal in 1-4%.

Pneumococcal meningitis is generallyreported to carry a much higher risk of deaf-ness than the other two common forms,although the difference may not be as great ashas been supposed. Certainly the risk of deaf-ness exists in all forms of meningitis and at allages, and no reliable predictors of this com-plication are available. Whether the trendtowards steroid treatment will make animpact on the frequency of deafness remainsto be seen.The likelihood of other serious long-term

sequelae is hard to estimate but may occur atabout the same frequency as severe deaf-ness-for example, four of 48 children whosurvived haemophilus meningitis in Walessuffered long-term neurological sequelae(8%).4

RELATION BETWEEN DURATION OF ILLNESSAND OUTCOMEThe relationship between clinical features andthe known pathophysiology of meningitis isimportant to our understanding of the dis-ease. The question is also important in themedicolegal context as actions about neuro-logical damage following meningitis oftencentre on a claim of missed or delayed diag-nosis. Inflammatory changes in the CNS arerelated to bacterial load, which suggests aconnection between prognosis and delayeddiagnosis, but attempts to analyse the pointreveal a more complex picture. One prospec-tive analysis4' showed that children with ahistory of less than 48 hours illness didsignificantly worse than those with longerhistories. A recent exhaustive examination of22 studies involving 4707 patients withmeningitis attempted to analyse the questionin more detail.42 Many of the studies wereunsuitable for formal meta-analysis, but theprevious finding that children with a slow andinsidious presentation had a better outlookthan those with acute illness was confirmed.At the other extreme, in a small subgroupwith fulminant meningitis, the influence ofantibiotics seemed minimal. Between theselies the group with clinically overt meningitisbut without septicaemic shock; data were

412 on 17 M

arch 2019 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.57.4.405 on 1 April 1994. D

ownloaded from

Meningitis

insufficient to analyse these, the very patientsin whom it would seem plausible that earlytreatment would be beneficial. These analysesreflect a familiar clinical observation, thatthere is often a substantial prodromal illnessof hours to days before features of meningitisappear, especially in children. Although theentire illness is subsequently designated asmeningitis, presumably the early phase corre-sponds with the bacteraemic illness precedinglocalisation to the meninges, or perhaps insome cases a preceding viral infection. Ineither case the host-parasite relationship ismore evenly balanced, and the illness lesssevere than in fulminant meningococcalsepticaemia.A further point bearing on the complex

relation between duration of illness and out-come is that deafness, the most commonadverse event leading to longterm disability,often develops very early in the course ofmeningitis.

PreventionVACCINATIONThe persistently high mortality and residualmorbidity from meningitis make it likely thatsubstantial further progress can be made onlyby improvements in prevention. Recent yearshave seen encouraging advances in vaccina-tion, with the promise of more to come.

Capsular antigen plays an important role inpathogenesis for all the common causes ofbacterial meningitis, and anticapsular anti-body is correspondingly important in protec-tion. Unfortunately children less than twoyears old show poor and transient responsesto these polysaccharide antigens.43 This seri-ous deficiency in the value of these antigensin immunisation programmes has been suc-cessfully addressed in the case ofH influenzaeby formulating conjugates of the capsularantigen with various protein moieties thatrender them fully immunogenic in infancy aswell as in older age groups. The problem iscomplex as different conjugates so far devisedvary in their immunological properties and intheir potency as antigens in infancy. Thereare also variations in host response dependenton ethnic, nutritional, and demographicfactors. Despite these intricacies, several con-jugate vaccines have been developed andwidely used in many countries includingGreat Britain.44 Their introduction has beenfollowed by rapid reduction in the incidenceof haemophilus meningitis and an additional,unexpected benefit, reduction of the nasopha-ryngeal carrier rate. In Finland, which hasbeen in the forefront of immunisation againsthaemophilus meningitis, this form of menin-gitis has been almost eliminated.45The great majority of invasive haemophilus

infections are caused by one serotype, groupb. The situation is more complex in meningo-coccal infection. Capsular polysaccharidevaccines have been developed against groupsA and C strains, and are available for use incontacts and for outbreak control when oneof these strains is responsible, for travellers to

areas of high prevalence, and for individualsat special risk of meningococcal disease.46The development of vaccine against menin-gococci of group B, the predominant group inBritain, has proved difficult, but severalapproaches to this problem are now beingpursued. One involves developing conjugatevaccines, another using outer membrane pro-teins (OMPs) rather than the polysaccharidecapsular material as antigen. As these OMPsare strain specific, this approach necessitatesthe development of multivalent vaccines.Trials of OMP vaccines have been achievedin several countries, including Norway,Brazil, and Cuba, with overall efficacy ratesvarying between 50% and 90%, but with lessefficacy in children.47 48

Pneumococcal vaccines49 are relevantmainly in the prevention of respiratory dis-ease and otitis media, but here too the suc-cessful development of conjugate vaccineagainst a high proportion of importantserotypes might in future contribute to ageneral vaccine against the main types of bac-terial meningitis. Alternatively, other anti-meningitis components may come to beadded to combinations, already being devel-oped, of diphtheria, pertussis, tetanus withHIB vaccine.

CHEMOPROPHYLAXISThe risk of meningococcal and of haemo-philus meningitis is significantly higher incertain contact groups than in control popu-lations. For this reason chemoprophylaxis isused in these contacts to eradicate nasopha-ryngeal carriage of the causal organism andthus diminish the likelihood of disease.50 51

Although the alarm which is engendered bythe diagnosis of meningitis leads naturally togreat emphasis on treatment of contacts, ithas to be emphasised that chemoprophylaxisis a control measure of very limited value,compared with preventive vaccination, orearly diagnosis and treatment of cases. Thereare several reasons for this, chiefly that inmost patients with meningitis the source isunknown, and the patient could therefore nothave received chemoprophylaxis. In addition,spread of both H influenzae and the meningo-coccus between carriers may be quite slow,and the risk to contacts extends long beyondany practicable period of chemoprophylaxis."Moreover, sometimes the carrier state is noteradicated, and resistance may develop to theagent used.

Rifampicin is used for contacts of boththese forms of meningitis, although there areimportant differences in dose and duration(10 mg/kg 12 hourly for two days formeningococcal contacts, 20 mg/kg daily forfour days for haemophilus contacts, maxi-mum 600 mg per dose). Two other authenti-cated forms of chemoprophylaxis areavailable, ciprofloxacin in a single oral dose,or ceftriaxone as a single injection. In general,close family, household, and nursery contactsare given chemoprophylaxis, and also theindex case before leaving hospital, as,paradoxically, successful chemotherapy for

413 on 17 M

arch 2019 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.57.4.405 on 1 April 1994. D

ownloaded from

Lambert

meningitis does not eradicate the carrierstate.

It is most important that contact betweenhealth professionals and families of patientswith meningitis should include careful discus-sion of possible early features of meningitis,and the limitations of chemoprophylaxis.

Resource implicationsHospital management of meningitis requireswell trained and well staffed medical andpaediatric units with easy access to the rele-vant supporting services. Most essential isliaison with the laboratory on a 24-hour basis,and with the relevant clinical specialists,including especially neurological and infec-tious disease services. Access to a fully staffedintensive care unit may become essential forsome very ill patients, with facilities for safetransfer if this is not available on site. Outsidethe hospital, communication with the relevantspecialist in public health medicine shouldextend beyond the legal requirements ofnotification and should include detailed dis-cussion of possible contacts and control mea-sures. Audit of services for meningitis mightmeasure, in addition to mortality and mor-bidity, such factors as completeness of notifi-cation, duration of illness before admission,whether antibiotics were given before admis-sion, time between admission and definitivetreatment and appropriateness of treatmentregimens employed.

At a later stage, rehabilitation facilities willbe needed for patients with residual disability.The needs of children with hearing problemshave been well studied.53 Early skilled assess-ment is required because, although somehearing loss in the acute illness and duringrecovery is conductive and may be transient,early identification of sensorineural loss isessential so that rehabilitation can beginpromptly. It has been strongly argued that allchildren who have had meningitis shouldhave auditory assessment, initially 4-6 weeksafter discharge from hospital. This wouldinvolve 30-40 appointments annually for ahealth district with a population of 250 000.This requirement may well change with suc-cessful vaccination programmes, althoughthere is evidence that at present not allpatients who need assessment are referred.

1 Booy R, Hodgson SA, Slack MPE, Anderson EC, Mayon-White DT, Moxon ER. Invasive Haemophilus influenzaetype b disease in the Oxford region (1985-1991). ArchDis Child 1993;69:225-8.

2 Fortnum HM, Davis AC. Epidemiology of bacterialmeningitis. Arch Dis Child 1993;68:763-7.

3 Baldwin LN, Henderson A, Thomas P, Wright M. Acutebacterial meningitis in young adults mistaken for sub-stance abuse. BM.7 1993;306:775-6.

4 Bohr V, Hansen B, Jessen 0, et al, Eight hundred and sev-enty-five cases of bacterial meningitis. Part 1 of a threepart series. .7 Infect 1983;7:21-30.

5 Rennick G, Shann F, de Campo J. Cerebral hemiationduring bacterial meningitis in children. BM7 1993;306:953-5.

6 Minns RA, Engelman HM, Stirling H. Cerebrospinalfluid pressure in pyogenic meningitis. Arch Dis Child1 989;64:8 14-20.

7 Mellor DH. The place of computed tomography and lum-bar puncture in suspected bacterial meningitis. Arch DisChild 1992;67:1417-9.

8 Harrison MJG, McAllister RH. Neurologic complicationsof HIV infection. In: Lambert HP, ed. Infections of the

central nervous system. London: Edward Arnold 1991:343-60.

9 Johnson SRD, Corbett EL, Foster 0, Ash S, Cohen J.Raised intracranial pressure and visual complications inAIDS patients with cryptococcal meningitis. 7 Infect1992;24: 185-9.

10 Brisson-Noel A, Aznar C, Chureau C, et al. Diagnosis oftuberculosis by DNA amplification in clinical practiceevaluation. Lancet 1991 ;338:364-6.

11 Viladrich PF, Pallares R, Ariza J, Rufi G, Gudiol F. Fourdays of penicillin therapy for meningococcal meningitis.Arch Intern Med 1986;146:2380-2.

12 Pecoul B, Varaine F, Keita M, et al. Long-acting chloram-phenicol versus intravenous ampicillin for treatment ofbacterial meningitis. Lancet 1991 ;338:862-6.

13 Friedland IR, Klugman KP. Failure of chloramphenicoltherapy in penicillin-resistant pneumococcal meningitis.Lancet 1992;339:405-8.

14 Viladrich PF, Gudiol F, Linares J, et al. Evaluation ofvancomycin for therapy of adult pneumococcal menin-gitis. Antimicrob Agents Chemother 1991 ;35:2467-72.

15 Peltola H, Anttila M, Renkonen O-U and the FinnishStudy Group. Randomised comparison of chloram-phenicol, ampicillin, cefotaxime and ceftriaxone forchildhood bacterial meningitis. Lancet 1989;1: 1281-7.

16 Teoh R, Humphries M. Tuberculous Meningitis. In:Lambert HP, ed. Infections of the central nervous system.London: Edward Arnold 1991:189-206.

17 Bennett LE, Dismukes WE, Duma RA, et al. A com-parison of amphotericin B alone and combined withflucytosine in the treatment of cryptococcal meningitis.N EnglJ Med 1 979;301:126-31.

18 Saag MS, Powderly WG, Cloud GA, et al. Comparison ofamphotericin B with fluconazole in the treatment ofacute AIDS-associated cryptococcal meningitis. N EnglJ Med 1992;326:83-9.

19 British Society for Antimicrobial Chemotherapy Work-ing Party. Antifungal chemotherapy in patientswith acquired immunodeficiency. Lancet 1992;340:648-51.

20 Powderly WG, Saag MS, Cloud GA, et al. A controlledtrial of amphotericin B to prevent relapse of cryp-tococcal meningitis in patients with the acquiredimmunodeficiency syndrome. N Engl 7 Med 1992;326:793-8.

21 Quagliarello V, Sheld WM. Bacterial meningitis: patho-genesis, pathophysiology, and progress. N Engl I Med1992;327:864-72.

22 Mustafa MM, Lebel MH, Ramilo 0, et al. Correlation ofinterleukin 1,B and cachectin concentrations in CSF andoutcome from bacterial meningitis. _7 Pediatr 1989;115:208-13.

23 Arditi M, Manogue KR, Kaplan M, Yogev R.Cachetin/tumour necrosis factor and platelet-activatingfactor concentrations and severity of bacterial meningitisin children.. Infect Dis 1990;162:139-47.

24 Brandtzaeg P, Kierulf P, Gaustad P, et al. Plasma endo-toxin as a predictor of multiple organ failure and deathin systemic meningococcal disease. . Infect Dis 1989;159:195-204.

25 Tauber MG, Shibl AM, Hackbarth CG, Larrick JW,Sande MA. Antibiotic therapy, endotoxin concentrationin CSF, and brain edema in experimental E. coli menin-gitis in rabbits. Y Infect Dis 1987;156:456-62.

26 Mustafa MM, Ramilo 0, Mertsola J, et al. Modulation ofinflammation and cachectin activity in relation to treat-ment of experimental H. influenzae type b meningitis.Y Infect Dis 1989;160:818-25.

27 Arditi M, Ables L, Yogev R. Cerebrospinal fluid endo-toxin levels in children with H. influenzae meningitisbefore and after administration of intravenous ceftriax-one. Y Infect Dis 1989;160: 1005-11.

28 de Lemos RA, Haggerty RJ. Corticosteroids as an adjunctto treatment in bacterial meningitis. A controlled clini-cal trial. Pediatrics 1969;44:30-4.

29 Belsey MA, Hoffpauir CW, Smith MHD. Dexamethasonein the treatment of bacterial meningitis: the effect ofstudy design on interpretation of results. Pediatrics1969;44:503-13.

30 Lebel MH, Freij BJ, Syrogiannopoulos GA. Dexa-methasone therapy for bacterial meningitis. Results oftwo double-blind, placebo-controlled trials. N Engl 7Med 1988;319:964-71.

31 Lebel MH, Hoyt MJ, Waagner DC, Rollins NK, FinitzoT, McCracken GH. Magnetic resonance imaging anddexamethasone therapy for bacterial meningitis. Am.7Dis Child 1989;143:301-6.

32 Girgis NI, Farid Z, Mikhail IA, Farrag I, Sultan Y,Kilpatrick ME. Dexamethasone treatment for bacterialmeningitis in children and adults. Pediatr Infect Dis .71989;8:848-5 1.

33 Odio CM, Faingezicht I, Paris M, et al. The beneficialeffects of early dexamethasone administration in infantsand children with bacterial meningitis. N Engl 7 Med199 1;324:1526-3 1.

34 Schaad UB, Lips U, Gnehm HE, Blumberg A, Heinzer I,Wedgwood J. Dexamethasone therapy for bacterialmeningitis in children. Lancet 1993;342:457-61.

35 Tureen JH, Dworkin RJ, Kennedy SL, Sachdeva M,Sande MA. Loss of cerebrovascular autoregulation inexperimental meningitis in rabbits. .7 Clin Invest1990;85:577-81.

36 Goh D, Minns RA. Cerebral blood flow velocity monitor-ing in pyogenic meningitis. Arch Dis Child 1993;68:111-9.

414 on 17 M

arch 2019 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.57.4.405 on 1 April 1994. D

ownloaded from

415

37 Ashwal S, Stringer W, Tomasi L, Schneider S, ThompsonJ, Perkin R. Cerebral blood flow and CO2 reactivity inchildren with bacterial meningitis. Jf Pediatr 1990;117:523-30.

38 Powell KR, Sugarman LI, Eskenazi AE, et al.Normalisation of arginine vasopressin concentrationswhen children with meningitis are given maintenanceplus replacement fluid therapy. Jf Pediatr 1990;117:515-22.

39 Fortnum HM. Hearing impairment after bacterial menin-gitis: a review. Arch Dis Child 1992;67:1128-33.

40 Howard AJ, Dunkin KT, Musser JM, Palmer SR.Epidemiology of Haemophilus influenzae type b invasivedisease in Wales. BMJ 1991;303:441-5.

41 Kilpi T, Anttila M, Kallio MJT, Peltola H. Severity ofchildhood bacterial meningitis and duration of illnessbefore diagnosis. Lancet 1991;338:406-9.

42 Radetsky M. Duration of symptoms and outcome inbacterial meningitis: an analysis of causation and theimplications of a delay in diagnosis. Pediatr Infect Jf1992;11:694-8.

43 Peltola H, Kayhty H, Virtanen M, Makela PH. Preventionof Haemophilus influenzae type b bacteremic infectionswith the capsular polysaccharide vaccine. N Engl Jf Med1984;310: 1561-6.

44 Cartwright KAV. Vaccination against Haemophilusinfluenzae b disease. BMJ 1992;305:485-6.

45 Peltola H, Kilpi T, Anttila M. Rapid disappearance ofHaemophilus influenzae type b meningitis after routinechildhood immunisation with conjugate vaccines. Lancet1992;340:592-4.

46 Frasch CE. Vaccines for prevention of meningococcal dis-ease. Clin Microbiol Rev 1 989;2(Suppl 1):S134-8.

47 Bjune G, Hiby EA, Grnnesby JK, Arnesen 0. Effect of anouter membrane vesicle vaccine against group Bmeningococcal disease in Norway. Lancet 1991 ;338:1093-6.

48 Cassio de Moraes J, Perkins BA, Camargo MCC, et al.Protective efficacy of a serogroup B meningococcalvaccine in Sao Paulo, Brazil. Lancet 1992;340:1074-8.

49 Broome CV, Breiman RF. Pneumococcal vaccine-past,present and future. NEngl7Med 1991;325:1506-8.

50 Cartwright KAV, Begg NT, Hull D. Chemoprophylaxisfor Haemophilus influenzae type b. BMJ 1991;302:546-7.

51 Jones DM. Control of meningococcal disease. BMJ 1989;298:542-3.

52 Cooke RPD, Riordan T, Jones DM, Painter MJ.Secondary cases of meningococcal infection amongclose family and household contacts in England andWales, 1984-7. BMJ 1989;298:555-8.

53 Fortnum H, Davis A. Hearing impairment in childrenafter bacterial meningitis: incidence and resourceimplications. BrJ7Audiol 1993;27:43-52.

Meningitis

on 17 March 2019 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.57.4.405 on 1 A

pril 1994. Dow

nloaded from