Role of Empiric Parenteral Antibiotics Prior to Lumbar ... · meningitis. One form is "hyperacute"...
Transcript of Role of Empiric Parenteral Antibiotics Prior to Lumbar ... · meningitis. One form is "hyperacute"...
REVIEWS OF INFECTIOUS DISEASES • VOL. 10, NO.2. MARCH-APRIL 1988© 1988 by The University of Chicago. All rights reserved. 0886-0162/88/1002-0009$02.00
Role of Empiric Parenteral Antibiotics Prior to Lumbar Puncturein Suspected Bacterial Meningitis: State of the Art
David A. Talan, Jerome R. Hoffman,Thomas T. Yoshikawa, and Gary D. Overturf
From the Department of Medicine, Division of EmergencyMedicine, Olive View and UCLA Medical Centers, UCLASchool of Medicine; the Geriatric Research, Education and
Clinical Center (GRECC), Veterans Administration; and theDepartment of Pediatrics, UCLA School of Medicine,
Los Angeles, California
The performance of lumbar puncture (LP) in patients with suspected meningitis is oftendelayed if, for example, the clinical presentation suggests a need for prior computed tomographic (CT) scan or if patients are initially examined at settings with limited clinicalfacilities. The role of empiric parenteral antibiotic therapy prior to LP under these circumstances has not been critically analyzed. Review of the literature suggests that in casesof bacterial meningitis (1) the existing data are inadequate to assess the effect of a shortdelay of therapy on mortality and morbidity; (2) a short period of antibiotic therapy priorto LP does not change cerebrospinal fluid (CSF) white blood cell count, protein, or glucose; (3) the yield of CSF gram stain and culture may be somewhat reduced by a shortperiod of antibiotic therapy, but these tests often remain positive; and (4) adjunctive tests,including blood cultures and CSF antigen tests, can often independently identify the bacterial meningopathogen. The available evidence suggests that if bacterial meningitis issuspected and LP must be delayed, intravenous antibiotics are warranted before CSF isobtained.
Bacterial meningitis is a life-threatening but potentially treatable disease. Therefore, it is of critical importance to establish rapidly the etiologic diagnosisand institute effective antibiotic therapy. In uncomplicated cases, this can be accomplished in a relatively short time by performing a lumbar puncture(LP) and analyzing the CSF for cell count and bacteria by gram stain. However, circumstances oftenarise in which an LP cannot be performed immediately. LP has been associated with serious complications and death in patients with elevated intracranial pressure [1-5]. Thus, in patients with signsassociated with increased intracranial pressure, suchas papilledema or focal neurologic deficits, it maybe necessary to defer an LP until a computed tomographic (CT) scan can be obtained. A recent seriessuggested that the "frequent" practice of obtaininga CT scan prior to LP contributed to a significantdelay in the treatment of bacterial meningitis [6]. Pa-
Received for publication 5 March 1987 and in revised form6 August 1987.
The authors acknowledge the expert manuscript preparationof Ms. Lois McArthur.
Please address requests for reprints to Dr. David A. Talan,Department of Emergency Medicine, Olive View Medical Center, 14445 Olive View Drive, Sylmar, California 91342.
365
tients are commonly seen first in an outpatient setting in which there may be inadequate facilities foracute care or inadequate technical or laboratorycapabilities to permit an LP to be done. The timeinvolved in transferring a patient to an appropriatefacility delays the definitive evaluation by LP. In situations such as these, the clinician must decidewhether to treat immediately with empiric parenteralantibiotics or wait to initiate treatment until CSF canbe obtained for analysis and culture.
Because bacterial meningitis is rapidly progressiveand potentially fatal, it has been proposed that, ifthe diagnosis is suspected and the LP will be delayed,intravenous antibiotic therapy should be institutedimmediately without obtaining CSF for microbiologic studies. McGee and Kaiser [7] state that "thefirst consideration in a patient with the acute presentation of meningitis is therapy, not specific diagnosis." However, a basic caveat of traditional infectious disease practice is that adequate cultures shouldbe obtained prior to the initiation of antibiotic treatment so that the etiologic pathogen(s) can be isolatedand appropriate treatment begun. Should the specific bacterial etiology be obscured by prior antibiotic therapy, causing sterilization of CSF cultures,the patient may be committed unnecessarily to acourse of expensive and potentially toxic therapy. In
366
addition to the risk and cost of unnecessary treatment of nonbacterial meningitis, there is the chancethat therapy may not be efficacious because of thepresence of organisms such as penicillin- or chloramphenicol-resistant pneumococci [8] or ampicillinor chloramphenicol-resistant Haemophilus influenzae [9, 10].
Therefore, in cases in which there must be a delayin obtaining eSF, the decision to initiate treatmentprior to an LP in order to maximize therapeutic efficacy comes at the possible expense of a decreaseddiagnostic accuracy. The benefits - or potential detriment - of such an approach have not been criticallyanalyzed. To better understand this problem we addressed three questions: Does a delay in administering antibiotics affect mortality and/or morbidity?Does administration of antibiotics prior to an LPobscure the diagnosis of bacterial meningitis? Arethere other methods of diagnosing bacterial meningitis if the eSF is made sterile by antibiotics?
Effect of Delay in Administration of Antibioticson Mortality and/or Morbidity
Bacterial meningitis is a potentially rapidly fatal disease for which the only form of effective treatmentis specific antimicrobial therapy. Thus, it seems reasonable to assume that a delay of even a few hours
Talan et af.
in the administration of antibiotics could have anadverse effect on prognosis. Nevertheless, to answerthis question definitively, a clinical study would berequired in which patients with suspected bacterialmeningitis were randomized to receive antibiotics either immediately or only after some defined periodof delay. For obvious ethical reasons such a studycannot be done. Therefore, there are no existingstudies that directly examine the effect of any delayof antibiotics on the prognosis of bacterial meningitis. Only indirect information relevant to this question is available. Several studies have compared themortality and morbidity of patients with bacterialmeningitis who present to the hospital after a variable duration of symptoms [11-27]. It is importantto note that these studies examine the effect on prognosis of a potential delay of therapy of 1-2 days,whereas circumstances that preclude immediateevaluation of suspected meningitis by LP at mostrequire several hours.
Table 1summarizes studies in which mortality wasevaluated in patients with various etiologies of bacterial meningitis in relation to the duration of symptoms prior to presentation to the hospital. Moststudies did not demonstrate a statistically significantrelation between mortality and duration of symptoms [11-16]. In the study by Olsson et al. [17] of43 patients with pneumococcal meningitis, there was
Table 1. Bacterial meningitis: duration of illness vs. mortality.
Mortality (070)
Years of study No. of Duration of illness[reference] patients Organism Overall (days) p*
<2 >2
1949-1973 [11] 349 All 26 24 25 NS1950-1960 [17] 43 PN 65 57 80 NS1952-1964 [12] 53 PN 37 28 31 NS1966-1976 [13] 875 All 11 9 11 NS1966-1976 [14] 164 PN 19 16 18 NS
<1 >1
1950-1960 [18] 107 All 40 47 23 <.051953-1971 [19] 468 HI 8 11 5 <.051971-1976 [20] 207 PN 45 71 50 <.05
Unspecified
1948-1973 [16] 79 PN 11 NS1949-1959 [15] 133 PN 17 NS
NOTE. All = Hemophilus injluenzae, Streptococcus pneumoniae, Neisseria meningitidis, and other bacteria; PN = S. pneu-moniae; HI = H. influenzae.
* P value determined by y} test. NS = not significant (i.e., P >.05).
Antibiotics/Lumbar Puncture in Meningitis
a mortality of 57010 in patients who presented with<2 days of symptoms compared with 80010 in patientswith >2 days of symptoms. However, this association did not reach statistical significance (X2
, P> .05).This study was remarkable for an unusually highoverall mortality. However, this may have been because of the inclusion of an older population withexclusively pneumococcal meningitis, since both advanced age and a pneumococcal etiology are associated with a poorer prognosis [11]. Also, 72% ofthe patients had a major coexisting illness. Theauthors concluded that "delay in seeking treatmentappears to be primarily responsible for the highermortality in this group of patients."
In contrast to the report by Olsson et al. [17] thestudies by Carpenter and Petersdorf [18], Davis etal. [19], and Baird et al. [20] yielded opposing results;that is, patients with a shorter duration of symptomssuffered a significantly greater mortality. Carpenterand Petersdorf [18] studied 107 children and adultswith various etiologies of meningitis, includingStreptococcus pneumoniae, Neisseria meningitidis,and H. inJluenzae. Seventeen (47%) of 36 patientswho were admitted within the first 24 hours of theirillness subsequently died. On the other hand, patientswho were hospitalized 1-7 days after the onset ofsymptoms had a mortality of only 23% (16 of 71cases, P < .05). The authors speculated that thisdifference in mortality was secondary to the phenomenon of patients with more severe infections seeking earlier hospitalization.
Davis et al. [19] studied 468 children with H inJluenzae meningitis. Death occurred in 24 (11%) of220 children whose duration of symptoms prior toadmission was <1 day. This mortality was at leasttwice (5%, 12 of 248) the death rate in children whoseduration of symptoms prior to admission was >1 day(P < .05). In this study, prior treatment with antibiotics also was associated with a significantly lowermortality. These investigators suggest that theremight be at least two clinical forms of H. inJluenzaemeningitis. One form is "hyperacute" and characterized by a short-course, fulminant illness with arapid progression of clinical signs and symptoms.This syndrome is similar to that of severe meningococcal meningitis. The other form of H. influenzae meningitis is a milder illness, with a more gradualonset of symptoms, that is less likely to be recognized as meningitis by a physician on the first visitand has a lower mortality. The possibility that priorantibiotic therapy ameliorated the severity of the ill-
367
ness and subsequently delayed presentation to thehospital was not mentioned.
Thus, a direct correlation or association betweenduration of symptoms and mortality for bacterialmeningitis has not been demonstrated consistently.To a great extent this is due to the inherent limitations involved in analyzing an outcome - namelymortality - that may be affected by multiple additional variables such as patient age, associated underlying illnesses, etiology and virulence of the infecting organism, and prior antibiotic therapy. Inorder to be interpreted, these variables must be identified and studied for their independent importanceby multifactorial analysis. The duration of illnessmay also be difficult to ascertain. In retrospectivestudies such as these, it is often difficult to determine accurately the duration of symptoms by reviewing hospital records. Also, it is impossible to determine when symptoms of meningitis - as opposed tothe prodromal infection - begin.
Several other retrospective studies with similarlimitations have attempted to evaluate the effect ofthe duration of symptoms prior to treatment on morbidity rather than mortality. Bohr et al. [13] studiedthe association between the duration of symptomsbefore treatment - specifically, hemiplegia, gaitataxia, intellectual dysfunction, epileptic fits, serious hearing deficits, and vestibular dysfunctionand subsequent morbidity in 875 patients with bacterial meningitis. There was a positive correlation between the duration of symptoms before treatmentand the rate of sequelae at discharge. These investigators also studied late neurologic sequelae in 164patients with pneumococcal meningitis [14]. Patientswho had symptoms for >48 hours prior to admission had a greater incidence of severe sequelae thandid patients who had symptoms for <48 hours. Herson et al. [21] found that of 73 children with H influenzae meningitis, those with symptoms of >3days' quration were more likely to have died or hadsevere morbidity (defined as blindness, hydrocephalus, institutionalization, microcephaly, quadraplegia,severe retardation, or uncontrolled seizures).
Several investigators have studied retrospectivelythe relation between the development of sensorineural hearing loss and the duration of symptoms priorto treatment. Nadol [22] studied the developmentof sensorineural hearing loss in 210 patients with various bacterial etiologies of meningitis who were followed only during the acute and recuperative phasesof the illness. While the average duration of symp-
368
toms was 32 hours in survivors without hearing loss,it was 47 hours in those with hearing loss (either partial or complete) (P < .05). Richner et al. [23] recalled children for audiometric testing who hadsuffered H. influenzae meningitis several years earlier. He found that while only 2070 of patients whoseH. influenzae meningitis had been treated within 48hours had evidence of sensorineural hearing loss,86% of those who had had symptoms for >48 hourshad demonstrable hearing loss (P < .05).
In contrast to the above data, Dodge et al. [24]reported a prospective evaluation of sensorineuralhearing loss in 185 children with bacterial meningitis and found that only 10.3% had persistent bilateralor unilateral hearing loss. This was not correlatedwith the duration of illness. Kaplan et al. [25] recently evaluated these data and those of anotherprospective series with regard to neurologic sequelaeand the prior administration of oral antibiotics inchildren with meningitis caused by H. influenzae.Children who were pretreated were more likely tohave paresis and sensorineural hearing loss at followup; the duration of illness prior to admission wassignificantly longer for pretreated than for untreatedchildren. While the investigators found no significant association between duration of illness prior toadmission and deafness (data evaluated by multiplelogistic regression analysis), children who were ill for>1 day prior to admission had a greater relative riskof developing severe sensorineural hearing loss thandid children ill for ~1 day.
Although a delay in definitive antibiotic therapywould seem logically to contribute to a poor out-
Talan et al.
come, the existing literature has not resolved this issue. Nevertheless, it is a reasonable concern that evena short delay of antibiotic therapy may be deleterious, particularly for the subpopulation with hyperacute bacterial meningitis. It is clear that mortalityand morbidity are not consistently predicted by theduration of symptoms prior to presentation, and itmay be difficult clinically to identify a high riskpopulation. However, if there were no risk of obscuring the diagnosis and preventing identificationof the bacterial etiology of meningitis, the immediate institution of antibiotics, prior to (delayed) LP,would appear to optimize a desirable therapeuticoutcome.
Effect of Administration of Antibiotics Prior toLumbar Puncture on the Diagnosisof Bacterial Meningitis
The traditional methods of diagnosing bacterialmeningitis include an analysis of CSF cell count, protein, glucose, gram stain, and culture. To what extent are the results of these tests affected by intravenous administration of antibiotics several hoursbefore an LP is performed? While no publishedstudy directly addresses this question, the answer canbe extrapolated from data from two sources: clinical investigations that have compared the results oftraditional CSF tests in meningitis patients who either did or did not receive antibiotics in prior outpatient treatment, and studies in which CSF has beenanalyzed serially during the administration of therapeutic antibiotics.
Table 2. Effect of prior outpatient antibiotic treatment on cerebrospinal gram-stain smears and cultures.
Reference
Positive smear Positi're culture
Organism Treated (010) Untreated (%) Treated (%) Untreated (%)
HI 84 92HI 82 90 94 98All 41 69§ 50 n§All 35 53§ 59 88§All 68 73 62 63All 68 84§ 68 95§All 37 54§
NOTE. HI = Hemophilus injluenzae; All = H. injluenzae, Streptococcus pneumoniae, Neisseria meningitidis, and other bacteria.* Includes only cases with positive CSF cultures.t Includes only cases with positive blood or CSF cultures, positive CIE or LA, or positive gram stain.t Includes only cases "typical" of bacterial meningitis (Le., increased CSF white blood cell count, increased protein, decreased glucose).§ P <.05.II Includes all cases of meningitis with CSF white blood cell count>10/mm3
•
Antibiotics/Lumbar Puncture in Meningitis
The term partially treated meningitis has been applied to those patients with bacterial meningitis whohave received some form of antibiotic therapy (usually oral) prior to a diagnostic LP; several investigators have studied the effect of partial treatment onCSF parameters in such patients [19, 25, 28-32].Studies of this nature are somewhat difficult to evaluate because of differing criteria used to define bacterial meningitis. Because CSF culture is the goldstandard for the diagnosis of bacterial meningitis,any attempt to study the effect of prior antibioticson the yield of CSF cultures must also evaluate thosepatients who have negative CSF cultures. Thus, someinvestigations included patients with negative CSFculture but in whom CSF protein, glucose, whiteblood cell count, and differential are "typical" or"consistent" with bacterial meningitis; others haveincluded all patients with >10 white blood cells/mm3
in the CSF. This is problematic because a variablenumber of cases of nonbacterial meningitis will beincluded depending upon the stringency of definition.
With an appreciation of the confounding methodologic problems in these studies, it is still possible todraw some conclusions. Table 2 summarizes studiesthat have analyzed the effect of prior outpatient antibiotic therapy on CSF gram-stained smears and cultures. The studies by Harter [28], Dalton and Allison [29], Winkelstein [30], and Jarvis and Saxena[31] defined meningitis by a positive culture (CSFor blood) or smear or negative cultures and smearbut "typical" CSF parameters consistent with bacterial meningitis. Criteria for typical CSF parametersvaried but in general included an elevated whiteblood cell count with a predominance of polymorphonuclear cells, elevated CSF protein, and decreased CSF glucose. Converse et al. [32] definedmeningitis by a CSF white blood cell count of >10cell/mm3
• Davis et al. [19] included only cases ofbacterial meningitis with positive CSF cultures, whileKaplan et al. [25] included only cases with positiveblood or CSF cultures, positive counterimmunoelectrophoresis (CIE) or latex agglutination (LA),or positive gram stain.
In general, partial (oral antibiotic) treatment resulted in a decrease in the number of positive CSFcultures of 40"/0-330"/0, and a decrease in CSF gramstains of 70"/0-410"/0. It is important to note that antibiotic treatment can also alter the accuracy of thegram-stain results; dead or injured gram-positive organisms may appear to be gram-negative [33]. These
369
studies did not consistently reveal that the identification of a specific bacterial agent of meningitis wasmore difficult than that of any other specific bacteria following antibiotic therapy. In the patients withpositive CSF cultures, the CSF cell counts, protein,and glucose were similar in patients with or withoutpartial treatment. Because of the methodologic problems described above, it is difficult to draw conclusions regarding the effect of prior treatment on CSFparameters in culture-negative patients.
From these data it can be assumed that if prioradministration of oral antibiotics decreases the yieldof CSF gram stains and culture, then the administration of intravenous antibiotics prior to LP wouldhave a similar effect. Experimental studies using rabbit meningitis models suggest that CSF cultureswould remain positive following treatment with ~
lactam antibiotics for 8-12 hours [34].Several studies in humans have also reported the
continued positivity of CSF cultures in some patientsrelatively early after the initiation of antibiotic therapy [35-48]. Wilson and Haltalin [35] studied 62 children with H. inJluenzae meningitis who had repeatLP performed 26 hours (mean) after initiation ofintravenous ampicillin therapy. Fourteen (23%) of62 children continued to have positive CSF culturesthat could not be attributed to antibiotic resistance.Ten of these 14 positive isolates could only be grownon Levinthal medium and not on standard chocolateagar medium. These children had a poorer clinicaloutcome with a higher frequency of neurologic sequelae. Feldman [36] studied 27 children with culture-positive bacterial meningitis caused by H. inf/uenzae (19 cases, one of which was a relapse), groupB streptococci (five cases), S. pneumoniae (threecases), and N. meningitidis (one case). Patients hadinitial LP followed by a repeat LP 24 hours after initiation of intravenous ampicillin or penicillin, aloneor in combination with chloramphenicol. Persistentpositive CSF cultures (using standard microbiologicmethods) were present in four (180"/0) of22 patients.Three isolates were H. inf/uenzae and one was groupB streptococci. The author examined CSF bacteriaconcentration and found that patients with persistent positive cultures had high initial bacterial concentrations. These patients also had organisms withhigh ("resistant") MICs to ampicillin and penicillinbut not to chloramphenicol when a correspondinglylarge inoculum was used for in vitro susceptibilitytesting. Despite this, antibiotic therapy was eventually effective. Overturf et al. [37] reported that, de-
370
spite good antibiotic susceptibility and a favorableclinical outcome, nine (38070) of 24 patients withH. influenzae meningitis treated with carbenicillinand one (6%) of 17 patients treated with ampicillinhad positive CSF cultures on standard culture mediaafter 12-24 hours of therapy. After 2-3 days oftherapy, however, all CSF cultures were sterile. Barsonet al. [38] reported on 50 children with bacterialmeningitis due to H. influenzae (42 cases), S. pneumoniae (four cases), N. meningitidis (three cases),and Streptococcus agalactiae (l case), who had LPrepeated 10.5-18 hours after starting treatment. Eight(33070) of 24 children treated with ceftriaxone andeight (40%) of 20 children treated with ampicillinand chloramphenicol had persistently positive cultures. In all except one case the persistent pathogenwas H. influenzae. Repeat CSF cultures were morelikely to be positive in cases with initial CSF bacterial colony counts of ~107 cfu/mL. In a series inwhich LP was repeated earliest, specimens were obtained 4-12 hours after the initiation of therapy [39].Del Rio et al. studied 78 cases of bacterial meningitis caused by H. influenzae (54 cases), S. pneumoniae (nine cases), and N. meningitidis (nine cases).Nine (43 %) of 21 patients treated with ceftriaxoneand 11 (58%) of 19 patients treated with ampicillinor chloramphenicol had persistent positive CSF cultures at 4-12 hours. In all patients except one, thepersistent pathogen was H. influenzae. There wasno significant difference between CSF bacterial concentration or susceptibility in those CSF specimensin which cultures remained positive and those inwhich they did not; clinical outcome was similar, despite the presence or absence of positive cultures.
These studies demonstrate that intravenous antibiotics will significantly decrease the frequency ofpositive cultures after initiation of therapy. However,even after 24 hours, as many as 38% may have positive cultures. As antibiotic killing of bacteria proceeds at a rate of (\)1.0 log/hour [34] one would predict that if LP were performed after only 1-2 hoursof intravenous antibiotics, it would be possible toobtain positive CSF cultures in the majority of patients. A study to address this question directly wouldbe feasible if, in patients with presumed bacterialmeningitis, CSF were sampled for culture at serialhours after parenteral antibiotics are initiated.
Because CSF white blood cell count, protein, andglucose will not be affected to a significant degree,administration of antibiotics prior to LP will notusually obscure the physician's initial impression of
Talan et al.
bacterial meningitis. Nevertheless, since the yield ofCSF culture and gram stain may be reduced, if thesetests are relied upon alone, it might be impossibleto make a definitive diagnosis of bacterial meningitis and identify the specific bacterial pathogen.
Alternative Diagnostic Methods If CSFIs Made Sterile by Antibiotics
At present three tests other than CSF culture andgram stain are available to establish the specific bacterial etiology in cases of meningitis. These are bloodcultures, CSF counterimmunoelectrophoresis (CIE),and CSF latex agglutination (LA) and coagglutination (CoLA). Blood cultures can be obtained easilyprior to the initiation of antibiotic therapy, evenwhen LP must be delayed. In (\)50% of cases of bacterial meningitis, blood cultures are positive for theetiologic agent (see table 3) [11, 12, 15, 17, 19, 31,49, 50]. The yield of positive cultures depends to agreat extent on the specific bacterial pathogen andits tendency to disseminate via the blood stream.Some cases of bacterial meningitis may occur as aresult of local extension and thus are associated withnegative blood cultures. Swartz and Dodge [49] reported positive blood cultures in 79% of cases ofH. influenzae meningitis, 56% of cases of S. pneumoniae meningitis, 33% of cases of N. meningitidismeningitis, 29070 of cases of f3-hemolytic streptococcal meningitis, and only 17% of cases of Staphylococcus aureus meningitis. On the other hand, Finland and Barnes [50] noted positive blood culturesin 63070- 86070 of cases of bacterial meningitis causedby these organisms. Thus, in patients with clinicaland other laboratory findings suggestive of menin-
Table 3. Frequency of positive blood cultures in casesof bacterial meningitis.
Ref- Organ- No. of cases with positive blood cultures/erence ism no. of cases with bacterial meningitis (010)
11 All 72/242 (30)12 PN 22/40 (55)15 All 60/133 (45)17 PN 23/43 (53)19 HI 152/349 (44)31 All 52/105 (50)49 All 79/153 (52)50 All 289/371 (78)
NOTE. All = Hemophilus injluenzae, Streptococcus pneumoniae, Neisseria meningitidis, and other bacteria; PNS. pneumoniae; HI = H. injluenzae.
Antibiotics/Lumbar Puncture in Meningitis
gitis to whom antibiotics are administered afterblood samples are obtained for culture but beforean LP is performed, the presence of bacteremia willestablish a presumptive diagnosis of bacterial meningitis.
CIE, LA, and CoLA are laboratory tests that canidentify CSF antigens of the commonest bacterialmeningopathogens - i.e., H. influenzae, S. pneumoniae, and N. meningitidis. Currently, prepared antisera against these organisms have an acceptable sensitivity and specificity. In various studies CIE iscapable of detecting antigen in 50070-90070 of patientswith meningococcal meningitis, 50%-100% of patients with pneumococcal meningitis, and 77070-90070of patients with H. influenzae type b meningitis,[51-61]. Nonspecific reactions are rare. Gram stainis similarly sensitive but may be difficult to interpret after antibiotic therapy is initiated. Recent reports suggest that LA and CoLA are much moresensitive than CIE with similar specificity [51, 54,62-64]. CIE, LA, and CoLA can be particularly useful in patients with meningitis in whom antibioticshave been administered (either orally or parenterally)before a diagnostic LP is performed. Bacterial antigens may persist in the CSF for several days afterantibiotic therapy. Bortolussi et al. [51] reported that,on repeat lumbar puncture, all patients with initiallypositive LA had a persistently positive test from 1to 6 days after the initiation of intravenous antibiotic therapy for various etiologies of bacterial meningitis. This has been confirmed by other authors [54,64]. Thus, the yield from CIE, LA, and CoLA wouldbe essentially unaffected if antibiotics were administered several hours prior to obtaining spinal fluid.
CIE, LA, and CoLA have limitations that mustbe considered. While a positive test using these techniques is very helpful, a negative test does not necessarily exclude the diagnosis of bacterial meningitis.Currently there is no consistently reliable antiserumto detect the polysaccharide of N. meningitidis serogroup B, a serogroup that is responsible for (\)50%of meningococcal meningitis in the United States.Pneumococcal serotypes 7 and 14 are also not reliably detected. Because S. pneumoniae and N. meningitidis account for the vast majority of adult bacterial meningitis, CIE, LA, and CoLA are much lessreliable in adults than in children in whom H. influenzae is the commonest pathogen. CIE, LA, andCoLA assays are not useful in diagnosing gram-negative bacillary, staphylococcal, and listerial meningitis. These organisms are more commonly isolated
371
from immunocompromised hosts with meningitis,those patients with meningitis associated with endocarditis, or those patients developing meningitis following surgery; therefore, in such patients, CIE, LA,and CoLA should not be relied upon to establisha diagnosis. Most importantly, although CIE, LA,and CoLA may identify the bacterial pathogen, theydo not provide direct information about the antibiotic susceptibility of the organism.
Other tests that may be helpful but that are lessavailable include the enzyme-linked immunosorbentassay (ELISA) for capsular antigen and the limuluslysate assay for endotoxin. ELISA has compared wellto CIE and LA for the detection of H. influenzaetype b and pneumococcal antigen [65, 66]. The limuIus lysate assay detects endotoxin from gram-negativebacillary organisms (e.g., Escherichia coli, Klebsiellapneumoniae, and Pseudomonas aeruginosa) as wellas H. influenzae and N. meningitidis [67,68]. Mostreports indicate that the test is rapid and reliable.However, in one study by McCracken and Sarff [69],the limulus test failed to diagnose culture-positivegram-negative bacterial meningitis in 31OJ(} of thecases in neonates.
Discussion
The performance of LP in patients with suspectedmeningitis may be delayed when clinical situationsdictate a need for prior CT scan. The risks of LPand the indications for CT scanning prior to LP havebeen debated. There have been several reports ofacute neurologic deterioration immediately following LP in patients with increased intracranial pressure or intracranial mass lesions. Duffy [1] followedfor one year 30 patients with raised intracranial pressure who had neurologic deterioration during or following LP. These patients were subsequently foundto have intracranial mass lesions or infarction. Thirteen of these patients had loss of consciousness during or immediately after LP; the neurologic statusof the others deteriorated to various degrees over thenext 12 hours. The same author noted neurologicdeterioration following LP in seven of 55 patientswith subarachnoid hemorrhage [2]. All six patientswho were subsequently investigated had evidence ofherniation confirmed by necropsy or CT scan.Garfield, who described one of the largest series ofbrain abscess cases, demonstrated that in 140 patientsin whom LP was performed, 41 had "significant deterioration in the level of consciousness" during the
372
48 hours following the procedure [3]. Careyet al.[4] reported eight of 62 patients with brain abscesswho died within 36 hours following LP.
Other studies, .however, suggest that even in thepresence of papilledema the risk of herniation afterLP is small. Korein et al. [5] reported possible complications following LP in seven of 59 patients withraised intracranial pressure without papilledema andin one of 70 patients with papilledema (85 % ofwhom also had increased intracranial pressure).These data and a review of 348 additional cases fromthe literature revealed possible complications in only1.2% of patients with papilledema on whom LP wasperformed. Thus, it is clear that rapid and severe neurologic deterioration has been associated with theperformance of LP; because of the close temporalrelation, these reports strongly suggest that LP wasthe cause of the deterioration. The frequency withwhich these complications occur in various settings,however, cannot be easily assessed.
Because of reports such as these, it has beenrecommended that LP not be performed in the presence of signs of increased intracranial pressure orintracranial mass lesions. With the advent of CTscanning, these contraindications for LP have become indications for CT scan prior to LP. The exact role of CT scan prior to LP is particularly germane with regard to the evaluation of patients withsuspected meningitis because the findings of fever,headache, and stiff neck are features also commonto the diagnoses of intracranial abscess (includingsubdural empyema) and subarachnoid hemorrhage.However, which clinical signs should preclude LPis not completely clear.
In the series reported by Duffy [1] of the 30 patients whose neurologic status deteriorated after LP,five were subsequently diagnosed with cerebral abscess that was treated initially as meningitis. Thesefive deteriorated after repeat LP. While all of theother 25 patients had either papilledema or otherevidence of an intracranial space-occupying lesion,in only one of the five with suspected meningitis andactual brain abscess were focal neurologic signsreported; none of the five apparently had papilledema. In Duffy's series of patients with subarachnoid hemorrhage, five of the seven who worsenedafter LP were initially grade 2- defined as moderateto-severe headache, nuchal rigidity, and no neurologic deficit other than cranial nerve palsies [2].There was no mention of the presence or absenceof papilledema in these patients. These observations
Talan et al.
led Duffy to suggest that LP not be performed whenthere is a history of progressive headaches associatedwith changes in mental status and the presence offocal neurologic signs or papilledema. He expressedcaution about the use of LP when a syndrome suggestive of cerebral abscess was present - such as anear, nose, or sinus infection - followed by signs ofmeningitis. He also suggested that CT scan be theinitial investigation prior to LP in all cases of suspected subarachnoid hemorrhage.
Garfield [3] noted that convulsions, hemisphericsigns, or papilledema were present in 68% of his series of 200 patients with brain abscess, comparedwith 30% of his series of 50 patients with nontuberculous bacterial meningitis. Based on this findinghe stated that, in the presence of these signs, LP iscontraindicated. A current neurology textbook byAdams and Victor [70] states that "in patients withsuspected (raised) intracranial pressure, LP shouldbe preceded by CT scan whenever possible." McGeeand Kaiser [7] state that LP is contraindicated in thepresence of papilledema or focal findings (exceptophthalmoplegia) until a CT scan has excluded thepossibility of a mass lesion.
Although general guidelines can be given, the exact indications for CT scan prior to LP in a givenpatient suspected of meningitis are not clear. Moreover, the available data indicate that overt clinicalsigns are not always manifest when intracranial hypertension or mass lesions exist. Focal neurologicfindings have been noted in 15%-30% of patientswith proven bacterial meningitis [71], although thepercentage of patients with suspected meningitis whohave these findings may be less. Because of these uncertainties, the ready availability of CT scans, andthe present medicolegal climate, the CT scan is often employed and, perhaps, overutilized prior to LPin the evaluation of patients with suspected meningitis. It is our opinion that the CT scanning is particularly overutilized in the pediatric population, inwhich the likelihood of brain abscess and subarachnoid hemorrhage is very small relative to the frequency of meningitis. Further studies are needed todetermine the value or detriment of CT scan performed prior to LP in cases of suspected meningitis.
Another situation that may result in delay of LPis the initial evaluation of a patient at a setting whereLP cannot be performed. Even at our urban centers,patients are frequently referred to us with suspectedmeningitis in whom LP has not been performed. Patients are seen at offices, clinics, and urgent care
Antibiotics/Lumbar Puncture in Meningids
centers that may not have the LP t~quipmentavailable, may not have the time or the laboratory facilities to initiate a septic workup, or may not have thecapability to manage an acutely ill patient. Furthermore, the urgency for rapid transport of the patientto an acute-care facility may preclude further diagnostic evaluation, Le., LP. It is our opinion that inseverely ill patients, obtaining CSF is of low prioritycompared with initiating treatment and ensuring thatpatients are expeditiously moved to an acute-carehospital.
In situations such as these in which LP is delayedand bacterial meningitis is suspected, the clinicianmust decide whether or not to treat immediately withempiric parenteral antibiotics. While no study currently exists that directly addresse~, the potential advantages or disadvantages of giving immediate antibiotic therapy prior to LP, and while the indirectdata regarding this issue are inconclusive, it wouldbe reasonable to assume that a delay of therapy maybe deleterious. Moreover, direct and indirect evidencecurrently available suggests that immediate antibiotic therapy shortly before LP is unlikely to seriouslyhamper the clinician's ability to diagnose the specific etiology of bacterial meningitis. By combiningthe results of preantibiotic blood cultures, subsequent CSF cultures, and CSF, CIE, LA, or CoLA,it should be possible to make an etiologic diagnosisin the large majority of patients with bacterial meningitis, despite a short period of prior antibiotictherapy.
With regard to epidemiologic considerations, it isimportant to note that rv75f1Jo of all cases of bacterial meningitis occur in children less than 10 yearsof age, and approximately two-thirds of these casesare due to H. injluenzae. Compared to other meningopathogens, H. injluenzae appt~ars to be easier togrow from early postantibiotic CSF cultures, is morefrequently cultured from blood, and is identifiedwith greater sensitivity by CSF, CIE, LA, and CoLA.Thus, particularly in the pediatric population - inwhich most cases of suspected bacterial meningitisoccur - it is extremely unlikely that a single dose ofparenteral antibiotic would prevent the identification of the offending bacteria. In other populations,such as neonates, adults, and imrnunocompromisedpatients, the sensitivity of blood cultures and immunologic tests for meningopathogens that commonly infect these groups is less secure; thus, thereis a greater justification for obtaining CSF either be-
373
fore antibiotics are given or at least within 1-2 hoursof initiating therapy.
The necessity to establish a specific etiologic bacterial diagnosis is diminished because of the improved broad-spectrum coverage of new antibiotics,such as the third-generation cephalosporins. Yet, bacterial identification and antibiotic-sensitivity testing remain desirable because of the emergence ofantibiotic resistance among common meningopathogens - such as ampicillin- or chloramphenicol-resistant H. influenzae and penicillin- or chloramphenicol-resistant pneumococci [8-10] - and the increasingnumber of immunocompromised and postsurgicalpatients who develop meningitis with unusual bacteria such as Listeria monocytogenes and gram-negative organisms. In cases in which LP is delayed, thebacterial pathogen and its antibiotic susceptibilitycan usually be established, even after antibiotics areadministered, by blood cultures and subsequent CSFcultures. In cases in which cultures are sterile, CIEor LA may identify the bacterial pathogen, but theantibiotic sensitivity of the organism will not beknown. In this situation, treatment can be basedupon the known epidemiology of bacterial meningitis. Only rarely will both cultures and immunologictests not reveal the causative bacteria and mandatecontinued empiric broad-spectrum antibiotic coverage. Appropriate empiric antimicrobial regimens andappropriate dosages have been recommended previously [71]. Initiation of a combined regimen ofchloramphenicol and ampicillin in children less than5 years of age and a regimen of ampicillin or penicillin G alone in older children and adults is sufficientfor disease suspected to be caused by communityacquired meningopathogens. Alternatively, a singleinitial dose of a number of third-generation cephalosporins (cefotaxime, ceftizoxime, ceftazidime, orceftriaxone) could be utilized for these patients. Initial empiric therapy must be sufficiently broadenedto include coverage for community-acquired meningopathogens, staphylococci, L. monocytogenes, andgram-negative organisms (including Pseudomonasaeruginosa) in immunosuppressed hosts or in patients recovering from neurosurgical procedures (e.g.,a regimen of vancomycin and a third-generationcephalosporin).
Conclusion
Should empiric parenteral antibiotics be adminis-
374
tered prior to LP, the risk of obscuring the diagnosisof bacterial meningitis is small, as is the risk thatthe patient would receive inadequate antibiotic treatment. Evidence found in this review supports therecommendation that, if bacterial meningitis is suspected and LP will be delayed, intravenous antibiotics should be given immediately after blood samples are drawn for culture. Further research needsto be done to determine the conditions that lead todelays in the evaluation of patients with suspectedbacterial meningitis, the role of CT scanning in theinitial evaluation of these patients, and the effectsof early therapy on the ability to identify bacterialmeningopathogens.
References
1. Duffy GP. Lumbar puncture in the presence of raised intracranial pressure. Br Med J 1969;1:407-9
2. Duffy GP. Lumbar puncture in spontaneous subarachnoidhaemorrhage. Br Med J (Clin Res) 1982;285:1163-4
3. Garfield J. Management of supratentorial intracranial abscess: a review of 200 cases. Br Med J 1969;2:7-11
4. Carey ME, Chou SN, French LA. Experience with brain abscesses. J Neurosurg 1972;36:1-9
5. Korein J, Cravisto H, Leicach M. Reevaluation of lumbarpuncture: a study of 129 patients with papilledema or intracranial hypertension. Neurology 1959;9:290-7
6. Byran CS, Reynolds KL, Crout L. Promptness of antibiotictherapy in acute bacterial meningitis. Ann Emerg Med 1986;15:544-7
7. McGee ZA,·Kaiser AB. Acute meningitis. In: Mandell GL,Douglas RG Jr, Bennett JE, eds. Principles and practiceof infectious diseases, 2nd ed. New York: John Wiley andSons, 1985:560-73
8. Jacobs MR, Koornhof HJ, Robins-Brown RM, StevensonCM, Vermaak ZA, Freiman I, Miller GB, Witcomb MA,Isaacson M, Ward 11, Austrian R. Emergence of multiplyresistant pneumococci. N Engl J Med 1978;299:735-40
9. Kenny JF, Isburg CD, Michaels RH. Meningitis due to Haemophilus inJluenzae type b resistant to l-I"\'~ ampicillin andchloramphenicol. Pediatrics 1980;66:14-6
10. Uchiyama N, Green GR, Kitts DB, Thrupp LD. Meningitisdue to Haemophilus injluenzae type b resistant to ampicillin and chloramphenicol. J Pediatr 1980;97:421-4
11. Hodges GR, Perkins RL. Acute bacterial meningitis: an analysis of factors influencing prognosis. Am J Med Sci 1975;270:427-40
12. Weiss W, Figueroa W, Shapiro WH, Flippin HE Prognosticfactors in pneumococcal meningitis. Arch Intern Med1967;120:517-24
13. Bohr V, Hansen B, Jessen 0, Johnsen N, Kjersem H, Kristensen H, Nyboe J, Rasmussen N. Eight hundred and seventyfive cases of bacterial meningitis. Part I of a three partseries: clinical data, prognosis, and the role of specializedhospital departments. J Infect 1983;7:21-30
14. Bohr V, Paulson OB, Rasmussen N. Pneumococcal menin-
Talan et al.
gitis: late neurologic sequelae and features of prognosticimpact. Arch Neurol 1984;41:1045-9
15. Quaade F, Kristensen KP. Purulent meningitis: a review of658 cases. Acta Med Scand 1962;171:543-50
16. Laxer RM, Marks MI. Pneumococcal meningitis in children.Am J Dis Child 1977;131:850-3
17. Olsson RA, Kirby JC, Romansky MJ. Pneumococcal meningitis in the adult: clinical, therapeutic, and prognosticaspe~ts in forty-three patients. Ann Intern Med 1961;55:545-9
18. Carpenter RR, Petersdorf RG. The clinical spectrum of bacterial meningitis. Am J Med 1962;33:262-75
19. Davis SD, Hill HR, Feigl P, Arnstein EJ. Partial antibiotictherapy in Haemophilus injluenzae meningitis: its effecton cerebrospinal fluid abnormalities. Am J Dis Child 1975;129:802-7
20. Baird DR, Whittle HC, Greenwood BM. Mortality from pneumococcal meningitis. Lancet 1976;2:1344-6
21. Herson ve, Todd JK. Prediction of morbidity in Hemophilusinjluenzae meningitis. Pediatrics 1977;59:35-9
22. Nadol JB Jr. Hearing loss as a sequela of meningitis. Laryngoscope 1978;88:739-55
23. Richner B, Hof E, Prader A. Hearing impairment followingtherapy of Haemophi/us inJluenzae meningitis. Helv Paediatr Acta 1979;34:443-7
24. Dodge PR, Davis H, Feigin RD, Holmes SJ, Kaplan SL,Jubelirer DP, Stechenberg BW, Hirsh SK. Prospectiveevaluation of hearing impairment as a sequela of acutebacterial meningitis. N Engl J Med 1984;311:869-74
25. Kaplan SL, Smith EO, Wills C, Feigin RD. Association between preadmission oral antibiotic therapy and cerebrospinal fluid findings and sequelae caused by Haemophilusinjluenzae type b meningitis. Pediatr Infect Dis 1986;5:626-32
26. Sell SHW, Merrill RE, Doyne EO, Zimsky EP Jr. Long-termsequelae of Hemophilus injluenzae meningitis. Pediatrics1972;49:206-11
27. Wiebe RA, Crast FW, Hall RA, Bass JW. Clinical factorsrelating to prognosis of bacterial meningitis. South MedJ 1972;65:257-64
28. Harter HP. Preliminary antibiotic therapy in bacterial meningitis. Arch Neurol 1963;9:343-7
29. Dalton DH, Allison MJ. Modification of laboratory resultsby partial treatment of bacterial meningitis. Am J ClinPathol 1968;49:410-3
30. Winkelstein JA. The influence of partial treatment with penicillin on the diagnosis of bacterial meningitis. J Pediatr1970;77:619-24
31. Jarvis CW, Saxena KM. Does prior antibiotic treatment hamper the diagnosis of acute bacterial meningitis? An analysis of a series of 135 childhood cases. Clin Pediatr 1972;11:201-4
32. Converse GM, Gwaltney JM Jr, Strassburg DA, Hendley JO.Alteration of cerebrospinal fluid findings by partial treatment of bacterial meningitis. J Pediatr 1973;83:220-5
33. Wilson GS, Miles AA. The biological characters of bacteria:morphology. In: Topley and Wilson's principles of bacteriology and immunity. 5th ed. Baltimore: Williams andWilkins, 1964:43
34. ScheId WM. Rationale for optimal dosing of beta-Iactam an-
Antibiotics/Lumbar Puncture in Meningi:is
tibiotics in therapy for bacterial ffii~ningitis. Eur 1 ClinMicrobiol 1984;3:579-91
35. Wilson HD, Haltalin KC. Ampicillin in Hemophilus influenzae meningitis: clinicopharmacologic evaluation of intramuscular vs. intravenous administration. Am 1 Dis Child1975;129:208-15
36. Feldman WE. Concentrations of bacteria in cerebrospinalfluid of patients with bacterial menilgitis. 1 Pediatr 1976;88:549-52
37. OverturfGD, Steinberg EA, UndermalJ. AE, Wilkins 1, Leedom 1M, Mathies AW lr, Wehrle PE Comparative trialof carbenicillin and ampicillin therapy for purulent meningitis. Antimicrob Agents Chemother 1977;11:420-6
38. Barson Wl, Miller MA, Brady MT, Powell DA. Prospectivecomparative trial of ceftriaxone vs. conventional therapyfor treatment of bacterial meningit is in children. PediatrInfect Dis 1985;4:362-8
39. Del Rio MA, Chrane D, Shelton S, McCracken GH lr, Nelson lD. Ceftriaxone versus ampicillin and chloramphenicol for treatment of bacterial meningitis in children. Lancet 1983;1:1241-4
40. Cogeni BL. Comparison of ceftriaxone and traditional therapy of bacterial meningitis. Antimicrob Agents Chemother 1984;25:40-4
41. McCrumb FR lr, Hall HE, Imburg ., Merideth A, Helmhold R, Defillo lB, Woodward TE. Treatment of Haemophilus influenzae meningitis with chloramphenicol andother antibiotics. lAMA 1951;145:469-74
42. Barrett FF, Taber LH, Morris CR, Stephenson WB, ClarkDl, Yow MD. A 12 year review of the antibiotic management of Hemophilus injluenzae meningitis. 1 Pediatr 1972;81:370-7
43. Prather GW, Smith MHO. Chloramphenicol in the treatmentof Haemophilus influenzae meningitis. lAMA 1950;143:1405-6
44. Ivler D, Thrupp LD, Leedom 1M, Wehrle PR, Portnoy B.Ampicillin in the treatment of acme bacterial meningitis.Antimicrob Agents Chemother 1963:335-45
45. Shackelford PG, Bobinski lE, Feigin RD, Cherry lD. Therapy of Haemophilus influenzae meningitis reconsidered.N Engl 1 Med 1972;287:634-8
46. Mathies AW lr, Leedom 1M, Thrupp LD, Ivler D, PortnoyB, Wehrle PF. Experience with ampicillin in bacterialmeningitis. Antimicrob Agents Chemother 1965:610-7
47. Fleming PC, Murray lDM, Fujiwc,ra MW, Prichard lS,McNaughton GA. Ampicillin in the treatment of bacterialmeningitis. Antimicrob Agents Chemother 1966:47-52
48. McCracken GH lr. The rate of bacteriologic response to an-timicrobial therapy in neonatal meningitis. Am 1 Dis Child1972;123:547-53
49. Swartz MN, Dodge PRo Bacterial meningitis: a review ofselected aspects. I. General clinical features, special problems and unusual meningeal reacti,)ns mimicking bacterialmeningitis. N Engl 1 Med 1965;272:779-87
50. Finland M, Barnes MW. Acute bact'~rial meningitis at Boston City Hospital during 12 selected years 1935-1972. 1Infect Dis 1977;136:400-15
51. Bortulussi R, Wort AJ, Casey S. The latex agglutination testversus counterimmunoelectrophoresis for rapid diagnosisof bacterial meningitis. Can Med Assoc 1 1982;127:489-93
375
52. Dirks-Go SIS, Zanen He. Latex agglutination, counter-immunoelectrophoresis, and protein A co-agglutination indiagnosis of bacterial meningitis. 1 CUn Pathol 1978;31:1167-71
53. Colding H, Lind I. Counterimmunoelectrophoresis in the diagnosis of bacterial meningitis. 1 Clin Microbiol 1977;5:405-9
54. Thirumoorthi MC, Dajani AS. Comparison of staphylococcal coagglutination, latex agglutination and counterimmunoelectrophoresis for bacterial antigen detection. 1 ClinMicrobiol 1979;9:28-32
55. Denis F, Samb A, Chiron lP. Bacterial meningitis diagnosisby counterimmunoelectrophoresis [letter]. lAMA 1977;238:1248-9
56. Olcen P. Serological methods for rapid diagnosis of Haemophi/us injluenzae, Neisseria meningitidis and Streptococcuspneumoniae in cerebrospinal fluid: a comparison of coagglutination, immunofluorescence and immunoelectroosmophoresis. Scand 1 Infect Dis 1978;10:283-9
57. Tugwell P, Greenwood BM, Warrell DA. Pneumococcal meningitis: a clinical and laboratory study. Q 1 Med 1976;45:583-601
58. Higashi 01, Sippel lE, Girgis NI, Hassan A. Counterimmunoelectrophoresis: an adjunct to bacterial culture in the diagnosis of meningococcal meningitis. Scand 1 Infect Dis1974;6:233-5
59. Whittle HC, Greenwood BM, Davidson NM, Tomkins A, Tugwell P, Warrell DA, zalin A, Bryceson ADM, Parry EHO,Brueton M, Duggan M, Oomen lMV, Rajkovic AD. Meningococcal antigen in diagnosis and treatment of groupA meningococcal infections. Am 1 Med 1975;58:823-8
60. Hoban Dl, Witwicki E, Hammond GW. Bacterial antigendetection in cerebrospinal fluid of patients with meningitis. Diagn Microbiol Infect Dis 1985;3:373-9
61. Kaplan SL. Antigen detection in cerebrospinal fluid - prosand cons. Am 1 Med 1983;75(Suppl IB):109-18
62. Whittle HC, Tugwell P, Egler Ll, Greenwood BM. Rapid bacteriological diagnosis of pyogenic meningitis by latex agglutination. Lancet 1974;2:619-21
63. Sippel lE, Hider PA, Controni G, Eisenbach KD, Hill HR,Rytel MW, Wasilauskas BL. Use of diretigen latex agglutination test for detection of Haemophilus influenzae,Streptococcus pneumoniae, and Neisseria meningitidis antigens in cerebrospinal fluid from meningitis patients. 1Clin Microbiol 1984;20:884-6
64. Leinonen M, Herva E. The latex agglutination test for thediagnosis of meningococcal and Haemophilus injluenzaemeningitis. Scand 1 Infect Dis 1977;9:187-91
65. Pepple 1, Moxon RE, Yolken RH. Indirect enzyme-linkedimmunosorbent assay for the quantitation of the typespecific antigen of Haemophilus influenzae B: a preliminary report. 1 Pediatr 1980;97:233-7
66. Harding SA, ScheId WM, McGowan MD, Sande MA.Enzyme-linked immunosorbent assay for detection ofStreptococcus pneumoniae antigen. J Clin Microbiol1979;10:339-42
67. Ross S, Rodriguez W, Controni G, Korengold G, Watson S,Khan W. Limulus lysate test for gram-negative bacterialmeningitis: bedside application. lAMA 1975;233:1366-9
68. lorgensen lH, Lee lC. Rapid diagnosis of gram-negative bac-
376
terial meningitis by the limulus endotoxin assay. 1 ClinMicrobiol 1978;7:12-7
69. McCracken GH lr, Sarff LD. Endotoxin in cerebrospinalfluid: detection in neonates with bacterial meningitis.lAMA 1976;235:617-20
Talan et al.
70. Adams RD, Victor M. Principles of neurology. 3rd ed. NewYork: McGraw Hill, 1985:10
71. Overturf GO. Pyogenic bacterial infections of the CNS. Neurologic Clinic 1986;4:69-90