Scott, Lancer A. MD*; Tintinalli, Judith E. MD, MS, FACEP†; Brewer, Kori L. PhD‡; Salgado, Cassandra D. MD, MS*
The failure of clinicians to promptly diagnose and treat bacterial meningitis can result in significant mortality and morbidity. All cause mortality for adults with meningitis, for example, is approximately 17% (range, 7%-30%), and nearly 18% of patients diagnosed with pneumococcal meningitis may experience long-term complications such as dizziness, fatigue, and gait disturbances.1-3 The use-albeit limited-of cerebrospinal fluid (CSF) analysis to differentiate viral from more serious forms of meningitis (bacterial or fungal) has been well documented in the literature.4-6
Unfortunately, lumbar punctures (LPs) can cause patient discomfort and result in serious complications such as infection, bleeding, headache, and cerebral herniation. The incidence of cerebral herniation is both controversial and rare.5 However, the incidence of post-LP headache is reported as high as 30%.7
Clinical guidelines have not been developed to identify those patients whose signs and symptoms are associated with a low likelihood of meningitis and thus may not need LP. The purpose of this study was to describe the incidence and etiology of meningitis in emergency department (ED) LPs and any clinical criteria that can help physicians select which adult patients would not need an LP because the likelihood of meningitis is low.
MATERIALS AND METHODS
A retrospective electronic chart review was performed to identify patients aged 16 to 50 years who received an LP in the ED between March 17, 2003, and October 22, 2003, for suspected meningitis. Older adults were excluded because they more commonly have atypical pathogens due to relative immunosuppression and unique environmental factors such as nursing homes. For the purposes of this study, meningitis was defined as positive gram stain, culture, or DNA analysis for bacterial, viral, fungal, or parasitic organism infection of the CSF.
Patients were identified through the ED clinical database using the International Classification of Diseases, Ninth Revision (ICD-9) procedure code for "lumbar puncture," E870.5. The 6-month period was selected for several reasons: (1) the period covered 3 seasons, (2) a new electronic medical record system had been introduced and documentation compliance had been verified by the beginning of the study, and (3) the 6-month period provided sufficient number of LPs and CSF samples for a pilot study analysis. Variables collected were chief complaint, clinical signs and symptoms, comorbidities, results of CSF analysis, ED length of stay, and return ED visits within 7 days of the LP. Lumbar punctures performed primarily for subarachnoid hemorrhage were excluded from the study because it was felt that clinical signs, symptoms, and comorbidities were commonly different between meningitis and subarachnoid hemorrhage patient subgroups.
The project was approved by the University of North Carolina School of Medicine Committee on the Protection of Rights of Human Subjects and completed at the University of North Carolina Hospital's Department of Emergency Medicine in Chapel Hill, NC. University of North Carolina Hospital is a large tertiary facility serving a wide range of patients from both urban and rural settings across the Southeast. The hospital is located near Durham and Raleigh, NC, with a combined population of just more than 1.5 million. The ED census at the time of the study was approximately 38,000, with an admission rate just more than 40%.
A hospital-wide, electronic database was the primary source of data for this study. Patients' presenting signs, symptoms, and comorbidities were determined from the Emergency Medicine Attending's electronic note. The ED length of stay was determined using a separate ED administrative database. The ED clinical and administrative databases received daily quality control checks by the ED data manager to ensure that the study data had not been tampered with or otherwise manipulated by those not involved in the study. The length of stay for patients receiving an LP was measured. The average length of stay for adult patients receiving an LP was compared with the average length of stay for all adult patients not receiving an LP in the ED during the same period.
Cerebrospinal fluid data collected included gram stain, total nucleated cell count (>5 total nucleated cell/mm3 was considered elevated by the hospital laboratory), culture, or DNA analysis. DNA analysis included techniques such as polymerase chain reaction to identify herpes simplex virus and immunofluorescence antibody test to identify Rickettsia rickettsii. Cerebrospinal fluid samples with pleocytosis, or an elevated total nucleated cell count, were further subdivided into those with lymphocytic or polymorphonuclear lymphocytic (PMN) predominance based on which cell line accounted for 50% or more of the total nucleated cell population. Other CSF data included abnormal central nervous system (CNS) glucose (<50 or >75 mg/dL), abnormal protein (<15 or >45 mg/dL), or an elevated red blood cell count.
To reduce potential confusion between abnormal CSF samples and normal samples that may have been contaminated by a traumatic LP, an elevated red blood cell count found in tube 1 of the CSF sample was not included in the study. Serum findings measured included white blood cell count (normal hospital value was 3800-10,800 cells/μL) and lymphocytic or PMN predominance based on which cell line accounted for 50% or more the white blood cell population.
The ability of each physical finding to rule in or out meningitis was determined through the calculation of sensitivity, specificity, and negative and positive predictive values. Differences in laboratory values found in patientswith meningitis compared with those without were determined using an unpaired t test, with P < 0.05 considered significant.
Overall, 217 adult LPs were identified during the study period. Fifty-three samples were excluded because no LP data were found (11 samples) or the LP was performed primarily to diagnose subarachnoid hemorrhage (42 samples).
Headache and fever were the most common presenting symptoms. Three quarters of patients (75.6%) reported headache, and nearly half (47.6%) of patients reported fever (Table 1). The most common comorbidity was HIV (11.6%), followed by a history of migraine headaches (4.9%) and a history of cancer or chemotherapy (4.9%) (Table 2).
One hundred ten adult CSF samples (67.1%) were normal and 54 (32.9%) were abnormal. There was no confirmed infection or disease in 151 CSF samples (92.0%). Eleven patients (6.7%) had meningitis confirmed by gram stain, culture, or DNA analysis. The causative agents included 55% viral (2 enterovirus, 2 herpes simplex virus, 1 varicella, and 1 coxsackie), 27% bacterial (1 coagulase-negative Staphylococcus aureus, 1 rickettsial, and 1 syphilis), and 18% parasitic/fungal (1 cysticercosis and 1 Cryptococcus neoformans). Two of the 11 confirmed cases of meningitis were from HIV-positive patients (1 syphilis and 1 C. neoformans; Fig. 1).
No single or combination of clinical criteria-including the absence of headache, fever, and altered mental status-ruled out CSF-positive meningitis (negative likelihood ratio = 0.91). Likelihood ratios for the classic triad of headache, neck stiffness, and altered mental status could not be calculated because no patient with CSF positive meningitis had all 3 signs or symptoms (Table 3). In addition, no serum finding was predictive of meningitis before LP (Table 4). Cerebrospinal fluid findings such as pleocytosis and lymphocytic predominance were strongly associated with a confirmed diagnosis of meningitis (P = 0.003 and 0.0004, respectively) (Table 5).
Patients receiving an LP stayed in the ED 48.6% longer than other ED patients seen during the same period (Table 6). Ten patients returned to the ED within 1 week complaining of headache. A follow-up visit for headache occurred in 6.1% of cases (Table 7).
Given the high mortality and morbidity associated with untreated bacterial meningitis, the prompt diagnosis and treatment of patients with suspected bacterial CNS disease is an important objective for all clinicians. Unfortunately, few consensus guidelines exist to assist clinician decision making when treating adults for presumptive meningitis in the emergency room. In fact, a National Guideline Clearinghouse search using the term "meningitis" yielded only 1 guideline fully relevant to diagnosis and treatment of adult meningitis. This guideline outlines selection of antimicrobial therapy in the setting of acute bacterial meningitis. Lumbar punctures are required in the guideline's algorithm.4
Lumbar punctures help clinicians differentiate bacterial meningitis from aseptic meningitis or other benign illnesses, yet the procedure can be time-consuming and cause significant complications. Our study found that patients who underwent LP had 50% longer ED stays when compared with other patients in the ED. In addition, approximately 1 in 20 patients returned to the ED within 1 week complaining of headache. Other studies have shown headache to occur 10% to 30% of the time after LP.7
The development of a prediction rule to differentiate bacterial meningitis from aseptic meningitis before LP may be helpful to both clinicians and patients. Unfortunately, we found only one substantive study testing a prediction rule for bacteria meningitis that did not include CSF analysis. The prediction rule criteria include duration of the main complaint, vomiting, signs of meningeal irritation, cyanosis, petechia, altered consciousness, and elevated serum C-reactive protein. Unfortunately, the study was limited to children between the ages of 1 and 15.8
The development of a prediction rule to diagnose meningitis (bacterial or aseptic) before LP may also be helpful. At least one large retrospective study has shown, for example, that the "classic triad" of fever, stiff neck, and altered mental status occurs in about two thirds of adult patients with meningitis.1 Other studies have found the incidence of fever, neck stiffness, and altered mental status to be much lower, with only 21% to 51% of patients reporting these symptoms.9,10
In their large systematic review, Attia et al11 suggested that no single clinical examination finding could rule in meningitis, and only the absence of fever, neck stiffness, and altered mental status could effectively rule out meningitis in most adult patients. In our study, no combination of clinical criteria ruled out meningitis. In addition, of our 11 patients with meningitis, 7 had fever, but none reported the classic triad as symptoms.
Specific physical examination findings associated with meningitis have been reviewed in the literature. For example, several studies have shown that Kernig and Brudzinksi signs for meningeal irritation are unreliable predictors of meningitis.12 In addition, Uchihara and Tsukagoshi13 observed that accentuation of the headache after rapidly moving head side to side may be moderately predictive of abnormal CSF. Our study did not measure the predictive value of these signs.
Traditional teaching suggests that an elevated CSF cell count (normal, <5 cells/mm3), markedly elevated CSF protein (normal, 9-58 mg/dL), a decreased CSF glucose (normal, 45-80 mg/dL), and a PMN predominance greater than 24 hours from the onset of illness can help predict bacterial meningitis.14 Other authors suggest that benchmarks such as these unreliably differentiate bacterial meningitis from aseptic meningitis.6,15-17 In their comprehensive review, Fitch and van de Beek state, "despite multiple retrospective models using logistic equations and other mathematical modeling, none (of the models predicting bacterial CNS disease) have proved robust enough for widespread clinical practice."5 In our study, CSF findings such as pleocytosis and lymphocytic predominance were strongly associated with meningitis. This is not surprising given the predominance of viral etiologies of meningitis in our cohort which are associated with lymphocytic predominant CSF.
The causative agents of meningitis in our study were very different from standard textbook descriptions of meningitis. For example, there were no confirmed cases of meningococcal or pneumococcal meningitis, 2 agents commonly reported in the literature.5,18 In addition, there was 1 case of coagulase-negative S. aureus, which is a rare cause of bacterial meningitis. It is unclear whether this culture positive case was due to a skin contaminant. Finally, 1 of 10 patients in the study was HIV positive. Because only 2 of the 11 confirmed cases of meningitis were from HIV-positive patients (1 syphilis and 1 C. neoformans), the overall impact of HIV on the study results was likely minimal.
Our findings are consistent with research by Powers, who reviewed LPs during a 2-year period at the University of Virginia in Charlottesville. Of 104 LPs, Powers19 reported only 1 case of bacterial meningitis and 13 cases of viral meningitis or encephalitis. In our study, nearly one third (27%) of the confirmed cases were treatable with antiviral agents, suggesting the need for adding antiviral agents to the standard regimen when treating presumptive meningitis on adults in the ED.
It is important to note the role antibiotics played in our study. The sensitivity of a gram stain deceases by as much as 20%, and the diagnostic yield of CSF culture is similarly decreased with previous antibiotic use.20,21 In our study, 27 (16.4%) patients received antibiotics before LP, suggesting that some cases of meningitis were unconfirmed. However, the incidence of antibiotic use was distributed evenly between abnormal and normal CSF samples (16.7% and 16.3%, respectively), suggesting that the impact of antibiotic use before LP was minimal.
Limitations for this study include the small sample size and its retrospective nature. Reliance on retrospective review of electronic notes could have led to misclassification bias, and LPs done primarily for subarachnoid hemorrhage could have been identified as meningitis, or vice versa. If the LP was not documented, cases would have been missed on the database search. Likewise, cases of meningitis that were missed because an LP was not done could not be identified by this study.
The administration of antibiotics before LP could have resulted in false-negative LP results. In addition, at least 1 of the CSF culture-positive cases may have been a contaminant. The period of the study excluded winter months raising the possibility that the CSF samples studied underestimate certain pathogens such as enteroviruses. Finally, this study may not be generalizable to community settings because 10% of the study population was HIV positive.
In this study of 164 adult LPs performed for possible meningitis, no clinical criteria were identified that could select which adult ED patients would not need an LP because the likelihood of meningitis is low. The absence of fever, neck stiffness, and altered mental status did not rule out adult meningitis. In addition, more research may be needed to evaluate the need for adding antiviral agents to the standard regimen when treating presumptive meningitis on adults in the ED.
More research is needed to risk stratify adult patients and improve the utilization of LPs in the ED. Although nearly one third of LPs were abnormal, no cases of pneumococcus or meningococcus were identified. Considering the discomfort, time, and complications associated with an LP and considering over two thirds of LPs were completely normal, improved clinical guidelines to better select patients for LPs would be clinically useful.
1. Durand ML, Calderwood SB, Weber DJ, et al. Acute bacterial meningitis in adults. A review of 493 episodes. N Engl J Med. 1993;328:21-28.
2. van de Beek D, de Gans J, Spanjaard L, et al. Clinical features and prognostic factors in adults with bacterial meningitis. N Engl J Med. 2004;351:1849-1859.
3. Bohr V, Paulson OB, Rassmussen N. Pneumococcal meningitis, late neurological sequelae and features of prognostic impact. Arch Neurol. 1984;41:1045-1049.
4. Tunkel A, Hartman BJ, Kaplan SL, et al. Practice guidelines for the management of bacterial meningitis. IDSA Guidelines. Clin Infect Dis. 2004;39:1267-1284.
5. Fitch MT, van de Beek D. Emergency diagnosis and treatment of adult meningitis. Lancet Infect Dis. 2007;7(3):191-200.
6. Segreti J, Harris A. Acute bacterial meningitis. Infect Dis Clin North Am. 1996;10:797-809.
7. Johnson KS, Sexton DJ. Lumbar puncture: technique; indications; contraindications; and complications. UpToDate. 2007. Available at: http:///www.uptodate.com
. Accessed October 10, 2007.
8. Oostenbrink R, Moons KM, Donders AT, et al. Prediction of bacterial meningitis in children with meningeal signs: reduction of lumbar punctures. Acta Paediat. 2001;90:611-617.
9. Sigurdardottir B, Bjornsson OM, Jonsdottir KE, et al. Acute bacterial meningitis in adults. A 20-year overview. Arch Intern Med. 1997;157:425-430.
10. Pizon AF, Bonner MR, Wang HE, et al. Ten years of clinical experience with adult meningitis at an urban academic medical center. J Emerg Med. 2006;30:367-370.
11. Attia J, Hatala R, Cook DJ, et al. Does this adult patient have meningitis? The rational clinical examination. JAMA. 1999;822(2):175-181. Similar findings by Newman DH. Clinical assessment of meningitis in adults. Ann Emerg Med. 2004;44(1):71-73.
12. Thomas KE, Hasburn R, Jekel J, et al. The diagnostic accuracy of Kernig's sign, Brudzinski's sign and nuchal rigidity in adults with suspected meningitis. Clin Infect Dis. 2002;35:46-52.
13. Uchihara T, Tsukagoshi H. Jolt accentuation of headache: the most sensitive sign of cerebrospinal fluid pleocytosis. Headache. 1991;31:167-171.
14. Greenlee JE, Carroll KC. Cerebrospinal fluid in CNS infections. In: Scheld WM, Whitley RJ, Durack DT, eds. Infections of the Central Nervous System. 2nd ed. Philadelphia, PA: Lippincott-Raven; 1997:907.
15. Graham TP. Myth: cerebrospinal fluid analysis can differentiate bacterial meningitis from aseptic meningitis. CJEM. 2003;5(5):348-349.
16. Negrini B, Kelleher KJ, Wald ER. Cerebrospinal fluid findings in aseptic versus bacterial meningitis. Pediatrics. 2000;105:316-319.
17. Powers WJ. Cerebrospinal fluid lymphocytosis in acute bacterial meningitis. Am J Med. 1985;79:216-220.
18. Schuchat A, Robinson K, Wenger JD, et al. Bacterial meningitis in the United States in 1995. N Engl J Med. 1997;337(14):970-976.
19. Powers RD. Emergency department lumbar puncture: clinical application and utility of laboratory testing. J Emerg Med. 1987;5(6):516-520.
20. Greenlee JE. Approach to diagnosis of meningitis. Infect Dis Clin North Am. 1990;4:583-598.
21. Martin JB, Tyler KL, Scheld WM. Bacterial meningitis. In: Tyler K, Martin JB, eds. Infectious Diseases of the Central Nervous System (Contemporary Neurology Series). Philadelphia, PA: FA Davis; 1993:176-187.
© 2008 Lippincott Williams & Wilkins, Inc.