Pediatric Bacterial Meningitis Workup

Bacterial meningitis is a medical emergency. A firm diagnosis is usually made when bacteria are isolated from the cerebrospinal fluid (CSF) and evidence of meningeal inflammation is demonstrated by increased pleocytosis, elevated protein level, and low glucose level in the CSF. Timely collection and processing of CSF and isolation of an organism allows optimization of choice of antimicrobial agent and duration of therapy. CSF chemistries and cytology vary, depending on the maturity and age of the newborn. ^[17]

A lumbar puncture (LP) may be contraindicated in some of the following conditions: unstable patients with hypotension or respiratory distress who may not be able to tolerate the procedure, brain abscess, brain tumors or other cause of raised intracranial pressure, and occasionally infection at the lumbar puncture site.

The Bacterial Meningitis Score, a clinical decision rule developed by Nigrovic et al, ^[18] has shown high accuracy and usability and continues to be evaluated with respect to its effectiveness as an aid to identify those children with CSF pleocytosis who are at low risk for bacterial meningitis. ^[1] The components of the score include the following:

• Positive CSF Gram stain

• CSF absolute neutrophil count of 1000/µL or higher

• CSF protein level of 80 mg/dL or higher

• Peripheral blood absolute neutrophil count of 10,000/µL or higher

• History of seizure before or at the time of presentation

Specific hematologic, radiographic (eg, computed tomography [CT] and magnetic resonance imaging [MRI]), and other studies assist in diagnosis.

Blood studies that may be indicated include the following:

• Complete blood count (CBC) with differential

• Blood cultures

• Coagulation studies

• Serum glucose

• Electrolytes

Measurement of the serum glucose level close to the time of CSF collection is helpful for interpreting CSF glucose levels and assessing the likelihood of meningitis.

Bacterial antigen studies can be performed on urine and serum and can be useful in cases of pretreated meningitis; however, a negative bacterial antigen study result does not rule out meningitis. The group B streptococcal (GBS) antigen test in urine is unreliable and should not be used to make a diagnosis of sepsis or meningitis.

Some data suggest that procalcitonin may be a useful biomarker for distinguishing bacterial meningitis from aseptic meningitis. Its use may enhance the sensitivity of the Bacterial Meningitis Score. ^{[19, 20, 18]} In a retrospective analysis of admitted patients with meningitis, Dubos et al found procalcitonin at a level of 0.5 ng/mL to have a sensitivity of 99% and a specificity of 83% for differentiating bacterial from aseptic meningitis. ^[19]

Definitive diagnosis is based on examination of CSF obtained via lumbar puncture. Opening and closing pressures should be measured in the cooperative patient. Similarly, the color of the CSF (eg, turbid, clear, or bloody) should be recorded. If the CSF is not crystal clear, administer treatment immediately without waiting for the results of CSF tests.

In a traumatic lumbar puncture, where bleeding occurs and the CSF is contaminated with blood, interpretation becomes especially difficult. In this situation, it is better to initiate treatment before the results of the CSF culture are available. In very bloody lumbar punctures, a drop of the fluid on the sterile dressing usually will produce a double ring if CSF is present. Generally, when in doubt, proceed with treatment and attempt the lumbar puncture again later.

In particular, if the patient shows signs of pending herniation, consider treatment without performing a lumbar puncture. The puncture can be performed later, when intracranial pressure (ICP) has been controlled and the patient is clinically stable. CT or MRI is helpful in managing patients who require control of ICP and herniation.

CSF studies

Perform total and differential cell counts, chemistries (ie, glucose and protein), Gram stains, and cultures on all CSF specimens. In a setting of antibiotic pretreatment, rapid bacterial antigen testing may be considered. Note that patients with both fulminant disease and poor immune response may not show cytologic or chemical changes in CSF. In about 2-3% of bacterial meningitis cases, bacterial cultures may be positive even when the Gram stain is negative and the cell counts and glucose and protein levels are normal.

White blood cell (WBC) counts higher than 1000/µL are usually caused by bacterial infections. Counts of 500-1000/µL may be bacterial or viral and call for further evaluation. Lower counts are usually associated with viral infections. 

The total WBC count cannot definitely distinguish between bacterial and other causes. At one time, it was generally believed that a predominance of polymorphonuclear leukocytes (PMNs) pointed to bacterial meningitis, but this has been an unreliable indicator; bacterial meningitis may also present with a lymphocytic predominance. Attempts to differentiate bacterial and aseptic meningitis on the basis of percentage and absolute number of premature neutrophils (ie, bands) have not yielded diagnostic results. ^[21]

The use of a corrected ratio of WBCs to red blood cells (RBCs)—that is, 1:500—or the percentage of neutrophils to “normalize” the cell count was shown to have limited utility in predicting which patients would have meningitis. The “corrected CSF” was shown to underestimate the true WBC count, causing clinicians to underdiagnose borderline meningitis cases. Formulas to adjust the WBC count have not increased the specificity or sensitivity of CSF analysis in traumatic lumbar punctures in neonates. ^[22]  

The CSF protein concentration is usually elevated in bacterial meningitis (greater than 50 mg/dL), but it is also elevated by a traumatic lumbar puncture. The CSF glucose concentration is usually reduced in bacterial meningitis. A normal CSF glucose level should be higher than two thirds of the serum glucose level; a CSF level lower than 50% of the serum level is suggestive of bacterial meningitis. In patients with very early disease, however, CSF protein and glucose values may be within the reference range. 

A study by Thomson et al that included CSF profiles from 7,766 infants ≤60 days old reported that in infants ≤28 days of age, CSF WBC counts and protein concentrations were higher and CSF glucose concentrations lower than in infants 29 to 60 days of age. ^[40]

A Gram stain of cytocentrifuged CSF may reveal bacterial morphology. The CSF should be plated immediately onto a chocolate and blood agar media. Smears of petechial lesions may reveal microorganisms on Gram stain. Although Gram stain may aid in diagnosis, the diagnosis may be missed in up to 30-40% of cases of culture-proven disease. The sensitivity of a positive Gram stain is 67%. ^[23]

Examination of a buffy coat smear also may reveal intracellular microorganisms. The results of a retrospective cohort study found that WBC counts in the CSF of febrile infants without bacterial or enteroviral infection are lower than was previously reported. ^[24]

Even when CSF results are otherwise normal, the fluid should still be sent for culture. Both N meningitidis meningitis and S pneumoniae meningitis are known to give normal CSF results. In an evidence-based article, Ray et al found that meningitis may still exist in 10% of children who have normal CSF analysis. ^[25] Their recommendation is to treat any child with antibiotics if there is a risk of bacterial meningitis.

Several tests based on the principle of agglutination are available for the detection of bacterial antigens in body fluids. Bacterial antigen detection can be carried out in samples of CSF, blood, and urine. A negative result, however, does not rule out bacterial infection. Antigen detection tests are most helpful in patients with partially treated meningitis in whom bacteria may not grow from CSF but antigens persist in body fluids. Antigen detection in urine is particularly helpful in such circumstances because urine can be concentrated severalfold in the laboratory.

Several gram-negative bacteria and higher serotypes of S pneumoniae have capsular antigens that cross-react with H influenzae type b polyribophosphate. Capsular antigens of group B meningococcus cross-react with K1-containing Escherichia coli. Gram stains of CSF are more sensitive than these rapid diagnostic tests for the detection of N meningitidis.

A study by Ye et al suggested that pediatric bacterial meningitis can be differentiated from similar disorders by assessing the concentration of IL-6 in the CSF and by evaluating the CSF/blood IL-6 ratio. The study found that these values were significantly higher in pediatric patients with bacterial meningitis than in healthy controls and in children with viral encephalitis, epilepsy, or febrile convulsions. The study also found that following treatment of the meningitis, CSF IL-6 levels fell by a significant amount, indicating that IL-6 concentrations may be a means of determining the efficacy of bacterial meningitis therapy. ^[26]

Partially treated meningitis

Many children receive antibiotics before definitive diagnosis is made. As a rule, a few doses of oral antimicrobial agents, or even a single injection of an antibiotic, do not significantly alter CSF findings, including bacterial cultures, especially in patients with H influenzae type b (Hib) disease. Oral antibiotics have never convincingly been shown to render patients with bacterial meningitis CSF culture–negative.

CSF cultures may become sterile rapidly if the pathogen was a pneumococcus or meningococcus, though cellular changes, an increase in protein, and low glucose levels persist. In such cases, CSF, blood, and urine should be tested for bacterial antigens; however, the presence of a negative antigen result does not entirely rule out a bacterial source.

In cases where antibiotic administration leads to CSF sterilization, polymerase chain reaction (PCR) testing may have a role to play in identifying the pathogen. PCR testing is able to identify the pathogen quickly and accurately and does not require a large number of organisms; however, it does require further validation in this setting.

Nigrovic et al found that Gram stain results (WBC count and absolute neutrophil count) in CSF were not affected by pretreatment with antibiotics; however, the rates of positive CSF culture and blood culture were lower with antibiotic pretreatment. ^[27] After pretreatment with antibiotics for 12 hours or longer, the patients had higher CSF glucose levels and lower CSF protein levels.

CT and MRI may reveal ventriculomegaly and sulcal effacement (see the images below).