Between 15–25% of all patients with meningitis will have seizure with fever either at the time of presentation or sometime during the course of the disease. Patients with meningitis almost always have symptoms other than seizure that suggest the diagnosis. Meningitis or encephalitis should always be excluded by examination of the spinal fluid if there is any question of the state of consciousness or meningeal irrigation.

Infection not involving the CNS with seizure precipitated by fever or toxins, such as shigellosis, pneumococcal bacteremia, or infection with human herpesvirus type 6 (HHV-6) is another category of causes. There was a lot of enthusiasm for HHV-6 as a cause of febrile seizures after it was discovered that one third of patients up to the age of two years presenting to the emergency department with the clinical syndrome of simple febrile seizure had HHV-6 infection.⁴⁴

However, a subsequent case-control study found evidence of acute HHV-6 infection in 15 (43%) of 35 patients with febrile seizures and in 15 (45%) of 33 controls.⁴⁵ The conclusion of these authors was that HHV-6 infection is not a major factor in the pathogenesis of febrile seizures. A more plausible interpretation might be that HHV-6 infection is a frequent cause of high fever in the age group at risk for febrile seizures, and that some children are predisposed to the development of seizures with fever.

Convulsion secondary to a specific cause such as lead encephalopathy or hypoglycemia can accompany fever caused by an infection and need to be excluded if suggestive clinical findings are present.

Idiopathic convulsive disorder (“epilepsy”) with seizure precipitated by fever is the proper diagnosis if there is an abnormal electroencephalogram (EEG) obtained at least a week after the seizure or if convulsions also occur without fever.

This is by far the most common cause of seizure with fever, occurring in approximately 4% of children between the ages of 6 months and 5 years. Simple febrile seizures are a benign condition whose main complication is recurrence, which happens in about one-third of patients. Recurrence risk is difficult to predict, but seems to be higher in children who present at a younger age.⁴⁶ The diagnosis of simple febrile seizure is fairly straightforward in older toddlers who have a classic history and physical examination, but it is more difficult in younger patients.

There has been much discussion about the need for lumbar puncture in the evaluation of seizure and
fever; clearly, the yield is low in cases where the diagnosis of simple febrile seizure is clinically suggested. However, meningitis can present with seizure and fever, and in young children, other signs of meningitis may be subtle. Because of this, the American Academy of Pediatrics (AAP) practice parameter recommends that excluding CNS infection by lumbar puncture be strongly considered in children younger than 12 months, and considered for children between 12 and 18 months of age.⁴⁷

A recent review of 503 cases of meningitis (97% of which were proven or suspected to be bacterial) showed that 115 presented with seizures (23%). Of these patients, 105 (91%) were obtunded or comatose, and thus obvious candidates for lumbar puncture. Of the other 10, 6 had nuchal rigidity, 1 had prolonged focal seizures, and 1 had multiple seizures and a petechial rash, all independent reasons to obtain spinal fluid by lumbar puncture. Two were suspected, on clinical grounds, of having viral meningitis.⁴⁸

In an older review of 152 children with purulent meningitis, 27 (18%) had fever and seizures.⁴⁹ Of these 27 children, 11 (41%) had no recorded meningeal irritation, no change of consciousness, and no bulging fontanelle; all of these children were less than 18 months of age. It is often stated that clinicians with experience can exclude meningitis on clinical grounds; however, science supporting this statement is absent. In fact, one of the authors of this last study argued that after years of experience it is still difficult to exclude meningitis in young children on clinical grounds alone, and he would do a lumbar puncture on all children under 16 months of age who present with fever and a convulsion.⁵⁰

In a retrospective review of 709 outpatients undergoing lumbar puncture, 225 (32%) had fever and a convulsion as the reason for the puncture.⁵¹ Only five had abnormal CSF, and most of these also had signs of meningeal irritation. A prospective emergency department study that allowed physicians to decide which patients required lumbar puncture (and thus biased the results toward finding a higher percentage of children with meningitis) found a total of 7 cases of meningitis (3 bacterial) in 102 patients who underwent lumbar puncture. There were 98 patients who were thought not to need the test. Most of the children with meningitis had lethargy, irritability, or vomiting; all had features of complex febrile seizure.⁵² The “catch-22” is this: those who clearly have simple febrile seizure need not undergo lumbar puncture; however, absence of meningitis is one of the criteria for establishing the diagnosis of simple febrile seizure. The author of Clinical Pediatric Neurology says that “a brief, generalized seizure from which the child recovers rapidly and completely is not caused by meningitis, especially if the fever subsides spontaneously…”⁵³ Certainly, lumbar puncture in a child with fever and a convulsion can be selectively obtained, and other findings must be considered.

Criteria suggested for electing lumbar puncture are:

An EEG is usually not indicated immediately after the convulsion.⁵⁴ It will often be abnormal even after a simple febrile convulsion, although an expert can often distinguish between a simple postictal abnormality and abnormalities suggesting epilepsy. If the EEG reveals epileptogenic activity several weeks after the seizure, the correct diagnosis is more likely to be “convulsive disorder precipitated by fever.” The pattern of activity seen on acute EEG is not predictive of recurrence risk.⁵⁵

Skull roentgenogram and blood glucose, calcium, sodium, or blood urea nitrogen (BUN) measurements are unlikely to reveal an abnormality and are not recommended unless there is some clinical basis for suspecting an abnormality.^54,56,57,58

Hospitalization is not recommended except for severe or multiple seizures or when the parents are too frightened or otherwise unable to observe the child.⁵⁴

Various antipyretic medications, given at the onset of fever and at a fixed interval throughout the course of a febrile illness, have been tested in prophylaxis against febrile seizures. Prospective randomized trials of acetaminophen⁵⁹ and ibuprofen⁶⁰ have failed to show a reduction in the number of
recurrences. Although antipyresis may have other beneficial effects, there appears to be no scientific basis for the use of antipyretics in the prevention of febrile seizures.

About one-third of patients will develop recurrence of febrile seizure, and half of those patients go on to a third episode. These recurrences can be decreased by continuous anticonvulsant therapy with either phenobarbital or valproic acid, but these medications are not without risk. Prophylaxis against recurrent febrile seizure with phenobarbital has been shown to lower achievement scores even 5 years out.⁶¹ Valproic acid is more effective at prevention of seizures,⁶² but carries the risk of idiopathic and irreversible severe hepatotoxicity.⁶³

In one study, diazepam, taken at the onset of fever, was also shown to decrease the incidence of febrile seizures; 39% of recipients, however, developed side effects of the medication.⁶⁴ Other studies failed to duplicate the beneficial effect.⁶² Some authorities believe that prophylaxis is even less likely to be needed if the convulsion is associated with roseola, shigellosis, or viral meningitis, because these illnesses have a tendency to be associated with convulsions. A lower recurrence rate among children whose first febrile seizure was associated with HHV-6 (the causative agent of roseola) has been confirmed by a prospective clinical trial.⁶⁵ A major factor to consider is that in the grand majority of cases, febrile seizure, even if it recurs, is a benign disorder with an excellent prognosis. The risk of most prophylactic regimens outweighs the potential benefits.

Exceptions could include children with neurologic disease, those with focal seizures, a family history of epilepsy, or a febrile seizure lasting more than 15 minutes or followed by a neurologic abnormality.⁶⁶

Occasionally the question arises as to whether seizure activity alone may be responsible for the finding of leukocytes in the CSF. In approximately 5% of cases, WBCs may be found in the CSF within 72 hours of a seizure, and most commonly within 12 hours of a seizure.^67,68,69 The maximum number of WBCs in the spinal fluid is usually less than 15 per mcL, but may rarely be as high as 80 per mcL.⁷⁰ Mildly increased protein may also be observed after seizures in about 10% of cases.⁶⁷ Postictal pleocytosis can occur after simple, complex partial, or generalized tonic-clonic seizures.⁶⁸ Clearly, this is a diagnosis of exclusion, and infectious causes should be pursued vigorously.

Atypical CSF findings include spinal fluid with cell count, glucose, and protein within normal limits, a situation where bacteremia “seeds” the meninges and the bacteria are culturable before the inflammatory reaction has occurred. A review found that 7 (3%) of 261 children with bacterial meningitis had CSF findings within normal limits when first seen.⁹⁵ All appeared sick enough to be hospitalized, and all but one child were immediately treated for sepsis, indicating that children with bacteremia producing positive CSF cultures and normal CSF findings usually appear sick enough to be hospitalized and treated for suspected septicemia. Another report described the atypical finding of cloudy spinal fluid with innumerable pneumococci and few leukocytes.⁹⁶

Low CSF cell counts (nonpurulent meningitis) may also occur with bacteremia. The prognosis in this situation depends on the prognosis for the bacteremic disease more than on the prognosis for meningitis if appropriate antibiotic therapy is given. Early meningococcemia or infective endocarditis may produce a “seeding” of the CSF with fewer than 100 leukocytes/mcL and normal CSF glucose and protein.

Prior antibiotic therapy is probably the most frequent factor causing delay in diagnosis of bacterial meningitis.⁹⁷ Previous immunization with the polysaccharide H. influenzae type b vaccine occasionally resulted in a more gradual onset in H. influenzae meningitis, presumably related to partial protection by antibodies stimulated by the vaccine. This clinical pattern has not been seen following receipt of the protein-conjugate Hib vaccine.

A predominance of mononuclear cells with low glucose and high protein can occur with Listeria or tularemia.⁹⁸ On the other hand, several virus infections (such as enteroviruses or La Crosse virus) may present with a leukocyte count above 1000/mcL, although the CSF glucose and protein are typically
normal or near normal (see section on purulent meningitis with negative culture).

Eosinophilic meningitis is discussed in the section on non-purulent meningitis.

As soon as S. pneumoniae is identified as the cause of meningitis and penicillin- and cephalosporin-resistance have been excluded by an oxacillin disk diffusion, minimal inhibitory concentration (MIC) test, and/or the e-test, penicillin alone is sufficient therapy and is less expensive than ampicillin. The dose is 250,000 units/kg per day divided into four to six doses. Penicillin-allergic patients can be treated with ceftriaxone, unless there is a history of anaphylaxis, in which case beta-lactam agents should probably be avoided. Vancomycin plus rifampin may be used in this unusual circumstance.

Patients infected with penicillin-resistant strains should be treated with a third-generation cephalosporin. In the event the isolate is intermediately or completely resistant to third-generation cephalosporins as well, vancomycin should be continued for the entire course of therapy. The cephalosporin should not be discontinued, as levels achieved in the spinal fluid may exceed the minimum inhibitory concentration of even “resistant” strains. For cases of meningitis caused by highly resistant strains, some experts advocate a repeat lumbar puncture 48–72 hours into therapy to document sterilization of the CSF.

Penicillin, ampicillin, or third-generation cephalosporins are effective. Penicillin is the drug of choice for susceptible strains. Usually, patients allergic to penicillin can be treated with a third-generation cephalosporin. Prophylaxis of household and other intimate contacts (such as daycare contacts) should be carried out as detailed in Table 21-9. There are several drugs that have been shown to eradicate carriage of meningococci; these drugs are effective prophylactic agents. Rifampin 600 mg twice a day for four total doses is effective in adults; children can be given 10 mg/kg/dose for four doses. Children less than one year should get 5 mg/kg/dose instead. A single dose of 500 mg of ciprofloxacin has been shown to be effective in adults and is considerably less cumbersome. One intramuscular dose of ceftriaxone is also effective,⁹⁹ and may be considered in cases where compliance to the other regimens is likely to be poor, or in situations where the other agents are contraindicated, as, for example, for prophylaxis of a pregnant woman.

A deficiency of the terminal components of complement is sufficiently common in systemic meningococcal infection that the patient should be screened for this disorder with a CH₅₀.¹⁰⁰ It is more common in patients of African-American descent.
A second case of invasive meningococcal infection is especially suspicious. The family should be screened if the patient has a complement deficiency.

As alternatives to ceftriaxone listed in Table 9-4, cefotaxime can be used for initial empiric therapy, depending on cost and availability.⁷⁷

Other bacterial causes are usually related to trauma, the newborn period, or to some host defect. Recommended initial chemotherapy is shown in Table 9-5.

Nafcillin appears to be the best of the penicillinase-resistant penicillins for penetration of the CSF.¹⁰¹ Clindamycin penetrates CSF poorly. Bacteroides meningitis also has been successfully treated with oral metronidadole.¹⁰²

For the patient with uncomplicated bacterial meningitis, antibiotic therapy should be given for at least 10 days except for meningococcal meningitis, where 5–7 days is sufficient.⁷⁸ Patients should generally be afebrile for 48–72 hours before therapy is stopped.
If therapy is begun late, or if the prognosis is otherwise poor, 14 days of IV therapy should be given. Duration of therapy for gram-negative meningitis should probably be longer; 21 days is considered standard.

In our opinion, treatment of bacterial meningitis should be completed in the hospital by the intravenous route. When there is unusually severe illness, when the patient is a young infant, or when there is a delay in diagnosis and treatment, the clinician should choose the longer duration and the higher doses of antibiotics, using the surest route. CSF protein and WBC do not usually return to normal until well after therapy has been completed; normalization of CSF parameters should not be used as a criterion for duration of therapy.

The management of meningitis requires more than the choice of the best antibiotic, the best dose, and the best route. Indeed, antibiotic therapy of meningitis is relatively standardized; the anticipation, early recognition, and effective treatment of complications, particularly cerebral edema, must be emphasized.

The recognition of cerebral edema is based on progressive changes in several physical findings¹⁰⁸ (Table 9-7). A flow sheet is useful to follow the course (Fig. 9-1). In cerebral edema, the state of consciousness changes from alert but irritable, to lethargic but arousable, to stuporous, and finally to deep coma. Pupillary reflexes change from mid-position, equal, and reactive to light to dilated and sluggish and, finally, to dilated and fixed. The optic discs are usually not a useful guide, because rapid changes in pressure are often not reflected by anything more than minimal blurring of the discs. Such a minimal blurring in acute purulent meningitis is not a contraindication to a careful lumbar puncture.
Marked papilledema or choked discs implies a more dangerous situation, and a chronic illness such as lead encephalopathy should be considered, as discussed earlier under Indications and Risks. Eye movements change from fixation on distant objects when the neck is rotated, to rotation in concert with the head, as if staring (doll’s eye movement) late in the course of the disease. Response to pain changes from purposeful withdrawal of the extremity to which the painful stimulus is applied, to non-purposeful withdrawal and stiffening with decerebrate rigidity, to complete flaccidity.

The pattern of breathing is an important guide to increased intracranial pressure. In cerebral edema, the breathing pattern changes from regular to irregular to the Cheyne-Stokes pattern. Cheyne-Stokes respirations are characterized by periods of deep and rapid respirations alternating with periods of slow, shallow respirations or apnea. The late changes in breathing and eye movements are related to compression of the brainstem and are often not seen if the child receives early chemotherapy.

The fontanel may change from flat to bulging. Convulsions may occur. Patients with meningitis who have had convulsions should not be kept deeply sedated, because this obscures changes of consciousness that are a guide to the severity of cerebral edema. Lateral rectus palsy may be seen in severe or chronic cerebral edema and is not of localizing value.

There is one report of 5 children who developed cutaneous flushing, an obvious but transient reddening
of the skin, at the same time they experienced neurologic deterioration secondary to increased intracranial pressure. The epidermal flushing involved the upper chest, face, or arms, and lasted from 5–15 minutes. The exact origin of this response is unknown, but it is postulated that it may be a centrally mediated response to sudden elevations in ICP.¹⁰⁹

These changes associated with cerebral edema resemble the progression of stages during induction of general anesthesia and have also been described in patients with brain tumors when herniation is imminent. This progression can sometimes be reversed by drugs such as mannitol that create an osmotic gradient between the brain and the plasma. Osmolar agents rapidly remove water from all of the extravascular space, but it is the removal of water from the swollen brain cells that may be lifesaving. Increased cerebral blood flow also occurs during mannitol infusion and may be a factor in its cerebral effects.¹¹⁰

Corticosteroids such as dexamethasone also reduce cerebral edema, according to studies of patients with brain tumors.

Other agents have been shown to decrease intracerebral pressure in experimental meningitis in animal models, including the calcium-channel blocking agent nimodipine.¹¹¹

The use of mannitol or dexamethasone in patients with meningitis and progressive worsening of brain signs has not yet been proved effective in any prospective controlled comparative study. The evidence for its effectiveness is derived primarily from repeated observation of reversal of the signs of brain deterioration in individual patients when this therapy is given.

The fatal consequence of cerebral edema is cerebral herniation and typically occurs within 8 hours of admission.¹¹² This is the usual cause of death in the first 24 hours of H. influenzae meningitis.¹¹³ However, it is not always fatal when treated (e.g., with mannitol), as described later in this section. Prevention by limiting intravenous fluids, as described later, is better than having to treat the cerebral edema, but fluid restriction has no immediate effect in an urgent cerebral edema situation. In addition, cerebral perfusion pressure is dependent on adequate systemic blood pressure; fluid restriction is thus inappropriate in the patient with hypotension and shock.

The presence of cerebral edema should be recognized by clinical observations. The actual measurement of opening pressure at the time of lumbar puncture is much more easily accomplished in older patients who are lucent enough to be cooperative with the examination (i.e., those who likely have normal pressures). In smaller children and in babies, attempting to measure the CNS pressure by manometry increases the risk of needle manipulation and may produce intrathecal bleeding. This
blood in the spinal fluid may make the CSF cell count and protein results, which are of critical importance in a patient with suspected meningitis, difficult to interpret. Free back-flow is often not obtained if the patient is straining. Localized cerebral edema may be present in the absence of increased spinal fluid pressure as measured in the lumbar area.¹¹⁴ Finally, the finding of a normal opening pressure does not rule out the presence of increased pressure a few hours later, after intravenous fluids have begun to correct dehydration.

Mannitol, urea, and dexamethasone appear to be effective in reducing cerebral edema from causes other than meningitis. The use of these agents has not been adequately studied in meningitis.

Mannitol has some theoretical advantages: there is no commitment to continuous therapy, and there is no risk of adrenal suppression or gastric ulceration or of masking fever, as is the case with dexamethasone. Moreover, there is no preparation time because of addition of diluents required, and no confusion about possible renal insufficiency, as in the case of urea. Urea is seldom used.

Mannitol is usually given intravenously as a 25% solution, 0.25–0.5 g/kg/dose, over a 20- to 30-minute period, for trauma.¹¹⁵ A review of supportive therapy recommended mannitol at 0.5–2.0 g/kg/dose over a 30-minute period, repeated as necessary.¹⁰⁷

A decrease in CSF pressure has been noted in brain tumors within 1/2 hour of starting the infusion, with the lowest pressure reached after 2–4 hours (Fig. 9-2). Pressure returned to previous levels by 6–10 hours. Mannitol can be repeated as soon as 4–6 hours after the last infusion if signs of cerebral edema are again noted. However, after about 48 hours, or six to eight doses, the patient’s long-term prognosis is likely to be unchangeable. The decrease in cerebral edema produced by mannitol or urea is only temporary, and the result depends on improvement of the disease process itself.

There has been some controversy over the role of fluid restriction in the prevention of cerebral edema in patients with meningitis. Prospective studies have shown that as many as 88% of patients with meningitis develop the syndrome of inappropriate antidiuretic hormone secretion (SIADH).¹¹⁶ Furthermore, studies done in patients who were never fluid restricted showed that the development of SIADH correlated with poor neurologic outcomes.¹¹⁷ However, the diagnosis of SIADH should never be made in a patient who is dehydrated (because the release of ADH in the setting of dehydration is not inappropriate). Restriction of fluids in
the setting of even mild dehydration may decrease cerebral perfusion. A prospective study of 50 children with meningitis suggested a worse outcome in patients who were fluid restricted.¹¹⁸

After restoring circulatory perfusion, restricting fluid therapy to approximately two-thirds maintenance in patients with hyponatremia would be a rational approach. Fluid restriction should be continued only until it can be demonstrated that the patient does not have SIADH, after which fluids are liberalized. This can usually be accomplished within 24–36 hours by careful monitoring of the patient’s input, output, serum sodium concentrations, and urine specific gravities and osmolalities. Serum sodium concentration should be monitored at least twice every 24 hours, especially in patients with documented sodium levels less than 130 mg/dL. Use of less than one-half the calculated maintenance dose as a preliminary guide may contribute to low blood volume, hypotension, or cerebral vascular sludging and venous thrombosis and is therefore to be avoided.

Often, children with meningitis have poor circulatory perfusion when first seen, as judged by poor capillary refilling in nail beds and cool extremities. This is not the same as septic shock, which is a later, more severe, situation. Treatment consists of a rapid intravenous infusion of Ringer’s lactate, 10–20 mL/kg/dose, repeated as necessary, to restore circulatory blood volume and improve brain perfusion.

Shock is usually recognized by low systemic blood pressure and a fast, weak pulse. The extremities may be warm, so that septic shock is sometimes called “warm shock.” Slow filling of the capillary nail beds may be a useful guide to septic shock.

The therapy of septic shock is controversial and is discussed in more detail in Chapter 10. The optimal therapy is plasma volume expansion, which can be done using plasma or 5% albumin, with an estimated initial dose of 25 mL/kg/dose. Patients with meningitis and shock need to receive enough fluid to keep systolic blood pressures adequate, and to keep urine output above 0.5 mL/kg/hour. Central venous pressure should be monitored. Pressor agents such as dopamine can be used to support blood pressure when needed but are not a substitute for adequate filling volume.

The total amount of plasma or 5% albumin given should be based on the blood pressure and on the central venous pressure. When the blood pressure is low or unobtainable, blood volume replacement should be given until blood pressure is adequate or central venous pressure is high.

The patient with meningococcemia is much more likely to develop disseminated intravascular coagulation (DIC) with purpura than is a patient with bacterial meningitis of another cause. This problem is discussed further in Chapter 10.

Congestive heart failure, manifested by rapid pulse, enlarging tender liver, or pulmonary edema, is not always caused by over-treatment of endotoxin shock by excessive intravenous fluids. Myocarditis, presumably secondary to endotoxin, has been demonstrated in some cases by autopsy findings of petechiae, cell infiltrates, and muscle fiber necrosis. Digoxin may be of value.

Purulent pericarditis may also occur, especially as a complication of H. influenzae meningitis. Small, self-limited effusions were detected in 20% of 100 consecutive cases.¹¹⁹ They were usually not significant clinically.

Decreased serum sodium concentration may be noted as early as the time of admission to the hospital, although it usually appears after the patient has had some intravenous therapy. This hyponatremia is probably secondary to brain disease and is mediated by an “inappropriate” excessive secretion of ADH. Use of 0.5 normal saline as the maintenance intravenous fluid is reasonable, with continued fluid restriction. Mannitol should be avoided when the patient is severely hyponatremic, because it increases sodium losses. It is an error to focus on correction of serum sodium and the mechanism of inappropriate ADH secretion to the neglect of the basic complication of cerebral edema and its treatment.

In a patient with meningitis, convulsions may occur because of cerebral edema, hyponatremia, subdural effusion, fever, and, by unexplained mechanisms, the disease itself. It is important to consider all possible underlying causes that might be improved by treatment other than anticonvulsants. Barbiturates in low to moderate doses may actually improve the state of consciousness in a patient with constant seizures. However, high doses of barbiturates or other anticonvulsants can obscure the persistence of a physiologic disturbance that should be corrected, such as cerebral edema or subdural effusions.

Focal or lateralized signs observed during the first few days of purulent meningitis have several possible etiologies.¹²⁰ Usually, mild focal signs such as slight asymmetry of strength or reflexes improve with observation and specific antibiotics and presumably have a vascular or inflammatory basis. However, hemiparesis or total paralysis may be caused by bleeding from DIC, indicating potential residual damage. Focal signs can also be caused by subdural effusions or asymmetric cerebral edema. CT scanning is indicated in the patient with focal neurologic signs.

In young children with bacterial meningitis subdural effusions are very common, occurring in 44 (39%) of 113 children aged 1–18 months old in one study. Only one child had a documented subdural empyema.¹²¹

Long-term follow-up (median 5.5 years) demonstrated no increased incidence of seizures, hearing loss, neurologic deficits, or developmental delay. The authors concluded that specific invasive therapy is not indicated in infants with meningitis and subdural effusion who are otherwise improving. The rare child with a subdural empyema will usually not improve clinically on antibiotics alone or will have an initial improvement followed by relapse. In the older child, whose cranial sutures are closed, subdural effusions are potentially more serious, especially if large or if associated with midline brain shift. Neurosurgical consultation for possible drainage should be obtained immediately.

Except for gram-negative meningitis in the neonatal period, brain abscess is extremely rare in the first week of purulent meningitis. Operative drainage is not usually necessary unless intracranial pressure is increasing.

As with other severe infections, meningitis is frequently accompanied by the anemia of inflammation.¹²²

One retrospective study of neonates with meningitis concluded that a markedly abnormal EEG during the acute phase of meningitis correlated with long-term poor neurologic outcome.¹²³ The study looked at 75 EEGs in 29 neonates; babies with normal or near-normal EEGs had good outcomes. This study also suggested that in a population of neonates, EEG may be useful for detecting subtle or subclinical seizure activity.¹²³

The availability of CT and magnetic resonance imaging (MRI) scanning has revolutionized CNS diagnosis. Scanning is useful for detecting brain abscess or subdural effusion (neither of which usually needs to be drained), large lateral ventricles (implying ventricular obstruction and ventriculitis), small lateral ventricles (cerebral edema), and diminished attenuation or focal hemorrhage (neither particularly treatable nor closely correlated with prognosis).¹²⁴ Imaging may be useful for the patient who is not responding as expected after several days of adequate therapy; to look for cerebral edema when there is no clinical emergency and a chronic process is suspected; and to evaluate possible hydrocephalus. It is also useful for the patient with focal neurologic signs. However, routine brain scanning in all patients with meningitis is not indicated: one prospective study of serial CT scans in meningitis concluded that clinical management was not influenced by scan results, which failed to reveal any significant abnormalities not suspected on neurologic examination.¹²⁵ Another study of 58 children with meningitis showed that positive findings with obvious clinical significance were found in
only 6 (10%) of 58 CT scans, all of which were occasioned by complex or prolonged seizures or prolonged fever.¹²⁶ The procedure demonstrates what clinicians and pathologists have always known: purulent meningitis exerts a profound effect on the brain parenchyma as well as on the meninges. If imaging is deemed necessary, contrast-enhanced MRI is probably the best imaging modality.¹²⁷

Sensorineural hearing loss is the most common adverse neurologic outcome in children who recover from bacterial meningitis. This complication occurs in about 10% of patients, with about half of these cases being bilateral.¹²⁸ It is most common with meningitis due to S. pneumoniae (31%), as compared with N. meningitidis (11%) or H. influenzae (6%).¹²⁸ It appears to be unrelated to the number of days of illness before admission or the type of antibiotic therapy but is correlated with ataxia, severe neurologic deficits, or initial CSF glucose of less than 20 mg/dL.¹²⁸ It is thought that hearing loss is secondary to inflammation that results not only from the infection itself, but also from the bacterial products released by lysis during antibiotic therapy.

Pretreatment with dexamethasone decreases the concentrations of some inflammatory mediators and leads to decreased hearing loss in animal models of experimental meningitis, especially meningitis caused by H. influenzae. Some clinical studies also documented better hearing outcomes in patients pretreated with dexamethasone versus placebo.¹²⁹ Most of the patients in these trials were suffering from H. influenzae infection, as the trials were carried out when Hib was the most common pathogen. Even though some aspects of the pathogenesis of disease are similar, extrapolation of data obtained from children with H. influenzae meningitis to those with S. pneumoniae or N. meningitidis infection may not be straightforward. A small, prospective trial of dexamethasone vs. placebo in children with pneumococcal meningitis showed a trend toward better audiologic outcomes in the dexamethasone recipients.¹³⁰ Limitations of the study include that the numbers were small and that ampicillin/sulbactam was the antimicrobial therapy employed. It has already been demonstrated that hearing outcomes vary with differing antimicrobial regimens.

A large, prospective, multicenter trial demonstrated that hearing loss, when it occurred, was present within the first several hours of illness.¹³¹ There is also the theoretical concern about the ability of dexamethasone to tighten the blood-brain barrier that is opened due to the inflammation of meningitis. This would normally be considered good, as it decreases the transport of plasma proteins, etc.; however, in this era of increasingly resistant S. pneumoniae isolates, the concern is that the transport of large molecules such as vancomycin would also be decreased. In a rabbit model of meningitis, CSF concentrations of vancomycin have been shown to be lower in rabbits that received concomitant dexamethasone.¹³²

Most experts advise pretreatment with dexamethasone in cases where the suspicion of H. influenzae infection is high (e.g., contact with a known case, gram-negative rods on Gram stain, patient from a group that gets no vaccinations for religious reasons). A recent randomized, placebo-controlled trial in adults with bacterial meningitis demonstrated decreased morality in the dexamethasone-treated group.¹³³ However, the use of steroids in the treatment of bacterial meningitis remains controversial. The preponderance of the clinical evidence suggests that benefit for patients with pathogens other than H. influenzae, if indeed a benefit exists, is small enough to be clinically difficult to demonstrate. The risk of having meningitis secondary to a multiply-resistant pneumococcus that requires vancomycin therapy, on the other hand, is tangible.

Hearing loss, if it is going to occur, is present within 48 hours of presentation. Children with bacterial meningitis should thus have their hearing tested prior to discharge from the hospital or shortly thereafter. If it is normal, no further testing is necessary. If abnormal, repeat testing should be performed, because about one-third of children will regain hearing over the ensuing 6 months. For infants under about 1 year of age, routine electrical auditory testing is advisable. Behavioral hearing testing by a pediatric audiologist in a soundproof room is the best method for a cooperative child.

The subdural space normally contains no fluid. Subdural effusions occur in about a third of children with meningitis, according to subdural taps done routinely.¹²¹ The best explanation of the pathogenesis of subdural effusion is that protein enters the subdural space from dural blood vessels, which are abnormally permeable in meningitis. The protein then brings in fluid from the vessels by osmotic action. In early papers, it was suggested that subdural effusions are more likely to occur when relatively large amounts of spinal fluid are removed at lumbar puncture. Although this has not been clearly documented, the theoretical risk has resulted in the consensus that the volume of fluid removed at lumbar puncture should be kept to the minimum necessary for study (a total of 2–3 mL).

Subdural effusions are least frequent after meningococcal meningitis¹³⁵ and, after correction for age, are about equally frequent in H. influenzae and S. pneumoniae meningitis.

A subdural effusion is not likely to occur until several days after the onset of purulent meningitis but rarely is found on the first day that meningitis is recognized, particularly if there has been preceding antibiotic therapy.

Because small asymptomatic effusions are common, and because asymptomatic moderate effusions typically resolve without drainage, they need not be looked for and drained routinely, as was done in the past. Effusions are often blamed for persistent fever, vomiting, or other complications of meningitis; subdural effusions are common enough that you might expect to find them in patients with these clinical features. Causation is more difficult to establish. Prospective studies suggest that the neurologic outcome is no worse for those with effusions than for those without.¹²¹ Effusions presumably also occur in older children with closed fontanels and usually resolve without drainage. Subdural punctures are rarely done when small effusions are detected on older infants, because the procedure is much more invasive if the anterior fontanel is closed. CT scans demonstrate that even large effusions may resolve completely.¹³⁷

Clinical indications to obtain CT scans consist primarily of persistent or recurrent neurologic abnormalities after 48 hours of therapy. These include focal neurologic signs, continuing lethargy, seizures, bulging fontanel, or increase in head circumference.¹³⁸ Usually, the fever is higher than expected for the day of the illness. CT scans can only detect effusions of about 30 mL or larger, but smaller effusions are unlikely to be clinically significant.¹³⁸ Transillumination is a simple, less expensive way to detect the thin layer of fluid and can be used to supplement the CT scan (Fig. 9-3).

If an effusion is found concurrently with any of the above neurologic findings, subdural needle aspiration through an open fontanel should be considered to see if improvement occurs. A second aspiration is usually not indicated unless the first aspirate is positive on culture or unless neurologic findings are relieved and then recur later to an equally severe degree. Large volumes of fluid (25
mL or more) should not be removed rapidly, because this may result in rapid expansion of the compressed brain, which may be manifested by pallor, tachycardia, or even apnea. Persistent subdural effusions have been treated by neurosurgical shunting of the fluid from the subdural space into the peritoneal cavity. However, the more conservative approach of waiting for absorption of the fluid may be just as effective in most cases.

This term is used when the effusion is purulent, containing more than 5000 leukocytes/mcL. It is more likely to be associated with a positive culture and severe neurologic damage.¹³⁸ Subdural empyema can also occur as a complication of severe sinusitis or mastoiditis without presenting as a purulent meningitis.¹³⁹ In contrast to patients with simple effusions, patients with subdural empyema usually present with focal signs, especially hemiparesis. Imaging may reveal shift of the midline structures, requiring an immediate neurosurgical referral for drainage.

A follow-up LP should be performed 3–4 days into therapy in neonates with gram-negative meningitis and in patients infected with drug-resistant pneumococci, and should be considered in anyone who has a poor response to therapy. Neonates with bacterial meningitis of any etiology should have a repeat LP at the end of therapy. However, routine follow-up examination of the spinal fluid is unnecessary as a test of cure.^77,140 Sometimes, equivocal results on the follow-up examination as the result of laboratory variation or a traumatic (bloody) tap prolong hospitalization and lead to another lumbar puncture, both of which are clinically unnecessary.

Patients with neurosyphilis generally get a follow-up LP at 6 months and those with cryptococcal meningitis are “re-tapped” two weeks into therapy. Meningitis caused by multidrug resistant Mycobacterium tuberculosis is a vexing problem; follow-up LP may be required on more than one occasion.

Persistent pleocytosis alone should not be used as a reason for prolonging therapy beyond 14 days. Repeat lumbar puncture to be sure the CSF glucose has risen to normal is not essential, if the response has otherwise been adequate. The frequency distributions of glucose, protein, and leukocyte counts show wide variation after successful treatment of meningitis.¹⁴⁰

The most frequent causes of fever beyond the expected range are drug fever, unrelated infections, phlebitis, arthritis, and unknown etiology.¹⁴¹ Sometimes prolonged fever is diagnosed and evaluated based on an unrealistic expectation of the rapidity of defervescence. Prolonged and even secondary fevers (fever occurring after a period of defervescence) are not uncommon in children with bacterial meningitis, even in the absence of complications. A brain CT scan is not automatically indicated. Several studies have documented that approximately 10–13% of patients with meningitis have fever beyond 8 days, and about 15–20% develop secondary fevers.¹⁴² The presence of prolonged or secondary fever does not correlate with adverse neurologic outcome, nor does it correlate with relapse or recrudescence.¹⁴³ If the child is neurologically well, and progressive improvement in clinical condition (other than fever) is noted, watchful waiting is the best approach. Plotting a graph of the child’s fever curve may also reveal a decrease in the temperature index, which is encouraging. Small undetected collections of serous fluid containing antigen-antibody complexes may be present in a joint, pleural space, pericardial space, or intracranially. This theoretical explanation is based on the detection of larger fluid collections in some patients with persistent fever.

Antibiotic therapy need not be continued beyond
the previously recommended times if the clinical condition is satisfactory.

Hydrocephalus may be communicating or obstructive. Obstructive hydrocephalus can occur within a few days of the onset of the illness if there is thick pus in the ventricles that blocks CSF flow out of the ventricular system, especially in newborns or very small infants. Obstructive hydrocephalus occurring early in the illness is usually manifested by acutely increased intracranial pressure, with slow pulse, rising blood pressure, and apnea. Emergency treatment consists of insertion of a needle into the ventricles to remove CSF and relieve pressure.

Communicating hydrocephalus usually is not noted until 2 weeks or more after the onset of the illness. It occurs more frequently in patients with a delay in beginning therapy. It may also occur in young infants with meningococcal meningitis, where the onset of the disease may be slow even without modification by preceding antibiotic therapy. Hydrocephalus is sometimes first suspected by noting a “setting sun” appearance of the eyes (Fig. 9-5), and an enlarging head can be confirmed by daily measurement of the head circumference. A CT scan or sonogram is useful to determine whether the ventricles are dilated throughout (communicating) or if there is obstruction of the aqueduct with a small fourth ventricle. Hydrocephalus is a more common complication of tuberculous or cryptococcal meningitis.

Severe intellectual damage is probably the most dreaded complication of meningitis. Other neurologic deficits that may result include seizures, paralysis, and deafness. Many mechanisms can contribute
to this damage. Cerebral anoxia with subsequent infarction may occur because of shock or apnea or increased intracranial pressure. Direct infectious or toxic destruction of brain tissue may also occur.

Statistical data are usually not helpful in discussing the prognosis of an individual patient with the parents. Bad prognostic factors include coma, hypothermia, shock, age less than 12 months, anemia, and seizures that are intractable or last (or have their onset) greater than 72 hours into antibiotic therapy. In most patients with sequelae, functional improvement tends to occur to at least some extent over time.

With CT scanning, these findings are recognized more often, but no specific therapy is available. Infarctions are more common with meningitis caused by S. pneumoniae. Hemiparesis may be present but sometimes resolves with time. Occlusion of the internal carotid artery has been reported.¹²⁰

Sterile arthritis representing a presumed antigen-antibody reaction has been recognized as a complication of infections such as hepatitis B, yersiniosis, salmonellosis, meningococcemia, and gonorrhea. During the course of H. influenzae meningitis, a sterile high-protein arthritis with fever is occasionally observed, which typically responds promptly after removal of the fluid. Arthritis during the first few days, on the other hand, is usually septic.^144,145

Transient or permanent cortical blindness, quadriplegia, acute endocarditis, and endophthalmitis are basically not preventable. Brain abscess can occur, especially in neonatal enteric bacterial meningitis.¹⁴⁶ Movement disorders, such as athetosis, as well as convulsive disorders, can occur.¹⁴⁷

This is presumed to be the usual cause of purulent meningitis with a negative spinal fluid culture. In many cases, the culture is negative because of preceding antibiotic therapy,¹⁴⁸ and in a few cases, cultures may be negative because of improper collection or delay in delivery to the laboratory. This is discussed further in the following section on aseptic meningitis syndrome. Listeria monocytogenes may not be positive on culture until 72 hours of incubation.

The CSF culture may be negative, or a skin or other “contaminant” bacterium may be found, so the culture is regarded as negative when the source is a congenital dermal sinus. Any child suffering from meningitis who is noted to have a midline dimple over any part of the vertebral column should have an MRI performed to look for the presence of an associated dermal sinus. If present, the sinus tract should be resected after recovery from the meningitis to prevent a second episode.

Anaerobes are a very rare cause of bacterial meningitis and would result in a negative conventional culture.¹⁴⁹ Anaerobic meningitis is usually associated with prior CNS or otolaryngologic surgical procedures or penetrating trauma. The diagnosis is sometimes suspected by the finding of pneumocephaly on imaging studies.¹⁵⁰ The principal predisposing cause in a child would be chronic sinusitis or chronic otitis media. An illustrative case of an 18-year-old with infectious mononucleosis who developed Lemierre’s syndrome and associated meningitis has been reported. The blood culture grew Fusobacterium necrophorum, but the CSF culture grew Prevotella bivia. CT scan showed pan-sinusitis.¹⁵¹ Another report tells of a previously healthy child who developed Fusobacterium necrophorum meningitis secondary to purulent otitis media with the same organism.¹⁵² A 3-year-old who injured her eye with a toothbrush developed meningitis secondary to Veillonella parvula, a gram-negative anaerobic coccus found in the mouth flora.¹⁵³

Acute sinusitis or other bacterial infection near the meninges may produce more than 1000 WBC/mcL, predominately neutrophils, whereas the glucose and protein are usually normal or nearly normal, and the Gram stain typically reveals no organisms. In the absence of prior antibiotics, parameningeal
infection should be suspected, and head imaging should be obtained.

Amebic meningitis is exceedingly rare. However, the physician should be aware that purulent meningitis with a negative culture can be due to amebae such as Naegleria species.¹⁵⁴ The patient sometimes has a history of swimming in lake water, which may be the source of the ameba. It has been suggested that the route of inoculation is through the nose to the olfactory bulbs. The CSF cell count is usually in the purulent meningitis range, with a predominance of neutrophils. Erythrocytes are often present and may provide a useful clue. The spinal fluid glucose may be normal or slightly depressed. The CSF protein is usually elevated.

This diagnosis can be made by isolating the ameba from the brain or by observing the motile organism in the spinal fluid (Fig. 9-6). Specific chemotherapy with amphotericin B, miconazole, and rifampin has cured several cases,^154,155 but the prognosis is extremely poor. In animal models, passive antibody prior to infection is protective,¹⁵⁶ and antibody given intracisternally as therapy prolongs survival but does not prevent death.¹⁵⁷

In this situation, there are no CSF findings of purulent meningitis except a smear showing bacteria, which may be traced to commercial lumbar puncture kits.¹⁵⁸ False positive gram staining due to contamination of the ethanol the slides were stored in has been reported. If contaminated gentian violet is placed on a hot slide, false positive Gram stains may result; this problem is solved by placing gentian violet on slides that have cooled.¹⁵⁹

Hemorrhage or thrombosis of the brain occasionally results in a CSF leukocytosis beyond what can be explained on the basis of the red blood cells present. It should be remembered that erythrocytes become hemolyzed in the CSF, resulting in the supernatant CSF becoming xanthochromic.

As discussed later in this chapter, herpes simplex encephalitis can resemble bacterial meningitis with negative bacterial cultures.

Mycoplasma pneumoniae produces CNS involvement in approximately 1 in 1000 patients. The primary manifestation is usually encephalitis, but meningitis and meningoencephalitis may also be seen. In meningitis secondary to Mycoplasma pneumoniae the cells may be predominantly neutrophils and the protein may be elevated, but the glucose is normal.¹⁶⁰ Typically, there is a preceding definite middle or lower respiratory infection, as discussed in Chapter 8.

Mycoplasma hominis is carried in the genital tract of 12–50% of pregnant women. Prevalence of infection and frequency of antibody titers increase with increasing parity.¹⁶¹ M. hominis cannot be seen on Gram stain and is difficult to isolate on routine microbiology laboratory media. Waites et al. found evidence of CNS infection with M. hominis in 5 of 100 predominantly preterm babies being evaluated for meningitis. In 4 babies, the organism was repeatedly isolated over several weeks.¹⁶² In another study of infants with good prenatal care, M. hominis was isolated from 9 (3%) of 318 infants who underwent
lumbar puncture. Most cleared the infection without specific therapy. In all these studies, the lack of a control group of healthy infants makes interpretation of the findings difficult.

This is an uncommon syndrome of recurrent aseptic meningitis, usually seen in adults and rarely teenagers, characterized by cloudy spinal fluid, usually with a neutrophilic pleocytosis and normal glucose and protein. The presence of “Mollaret cells,” once thought to be endothelial but now known to be of the monocyte/macrophage family, is pathognomonic. This syndrome is now thought to be secondary to HSV-2. This conclusion is based on the PCR finding of HSV-2 DNA in the CSF of three patients with the disorder and on the known propensity of herpes group viruses to cause recurrent disease.¹⁶³

Several cases of aseptic meningitis after receipt of high-dose immune globulin have been reported. Bacterial meningitis is often suspected initially because of the neutrophilic predominance.^164,165 In a review of 11 cases,¹⁶⁵ the mean CSF leukocyte count was 1123 per mcL (median, 451) with a mean of 74% neutrophils (median 87%).

After the red blood cells have been lysed, the white blood cells and protein may remain for several days after an intraventricular hemorrhage. The CSF glucose also is usually depressed early or several weeks later, and this may persist. The finding of xanthochromic fluid is the principal clue to this cause of suspected meningitis with negative cultures.

When a patient develops nonpurulent meningitis while receiving an antibiotic, the most reliable clue to bacterial meningitis may be a significantly decreased CSF glucose. In this situation, the protein may also be elevated. Because prior antibiotic therapy may prevent recovery of the infecting organism on culture, the frequency of this etiology is unknown, but it is probably the most frequent cause of hypoglycorrhachia in children.¹⁸¹ This situation is discussed in detail in the section on purulent meningitis.

Early bacterial meningitis may also produce this CSF pattern. Listeria monocytogenes produces nonpurulent meningitis with predominately lymphocytes in a small proportion of cases. The CSF glucose is almost always low and the protein almost always high in such cases.¹⁸²

Bacterial meningitis can also present as nonpurulent meningitis with a normal CSF glucose, as discussed later.

Tuberculous meningitis presents subacutely in 3 clinical stages. Stage 1 consists of nospecific signs such as fever and headache. In stage 2 the patient develops mental confusion and early focal signs, such as cranial nerve palsies. In stage 3 the patient progresses to obtundation and coma. The prognosis is directly related to the clinical stage at the time therapy is begun. Thus, it is critical to make the diagnosis as early as possible.

Tuberculous meningitis usually results in hypoglycorrhachia. In a series of 405 children with tuberculous meningitis from 1987–1998, 80% had depressed CSF glucose levels. CSF protein was greater than 100 mg/dL in 65%. Half of patients
had less than 60 WBC per mcL, 45% had 60–500 WBC/mcL, and 5% had more than 500 WBC/mcL. Just over half had evidence of pulmonary involvement, and only 16% had a positive tuberculin skin test (TST). Nearly 70% had a positive family history of tuberculosis.¹⁸³

Another study reported a higher percentage of pulmonary involvement: 87% of 214 children with tuberculosus (TB) meningitis in that series had an abnormal chest x-ray.¹⁸⁴ Family or exposure history was found in 66%, and the TST was greater than 10 mm in only 30%.

A study comparing 110 children with tuberculous meningitis to 94 patients with non-TB aseptic meningitis syndrome identified 5 clinical features that suggested TB:

When tuberculous meningitis is suspected, three factors should be evaluated: TST, chest roentgenogram, and exposure history (including past travel to or residence in a high-risk area). If the TST is negative (less than 5 mm), the chest roentgentogram is normal, and there is no history of exposure to tuberculosis (or residence in high-risk areas), then tuberculous meningitis is much less likely. In contrast to past teaching, pulmonary tuberculosis that has been present long enough to calcify (more than about 6 months) may still be associated with dissemination.¹⁸⁵ Thus, even if the three factors listed above are negative, tuberculous meningitis should be considered if the CSF glucose is low and there is a CSF lymphocytosis.