Best Evidence Topics
  • Send this BET as an Email
  • Make a Comment on this BET

Normal CSF: Does it exclude meningitis?

Three Part Question

In [a child with suspected meningitis] the [initial CSF findings normal], is that reassuring enough to [exclude bacterial meningitis]?

Clinical Scenario

A 2-year-old girl presented with 1-day history of temperature, off food and “not herself”. Clinical examination showed a slightly irritable child with a temperature of 38.8°C, mildly congested throat and doubtful neck stiffness, with no other apparent focus of infection. Urine was clear. A full septic screen was performed including lumbar puncture. C reactive protein was 38; cerebrospinal fluid (CSF) showed glucose 3.6 mmol (blood glucose 4.8 mmol), protein 0.6 g/l, white blood cell count 4 with no bacteria on Gram staining. The child was observed with a diagnosis of viral illness without any antibiotics. We wondered: can meningitis occur with initial normal CSF?

Search Strategy

This review was aimed to obtain all available information with the intention of providing a perspective for resolving this issue. A non-systematic search of the world literature was carried out using English as the main language in July 2008 using the Cochrane Library, PubMed, Medline (1950 to date), Embase (1974 to date), and Cinhal (1982 to date) via the NHS healthcare database.
The following key words were used (where “ADJ” is the operator for “adjacent within 5 words”): meningitis, bacterial ADJ meningitis, CSF, cerebrospinal ADJ fluid, spinal ADJ fluid, normal, no ADJ abnormalities, restricted to newborn and children.

Search Outcome

The search totalled 61 hits. Relevant papers were studied along with the cross-references. A total of six cases series and four case report publications were identified as eligible for analysis. All papers included children with culture-positive bacterial meningitis, and their charts were reviewed retrospectively to ascertain cases with initial normal CSF. We excluded publications where children had received antibiotic treatment before the lumbar puncture was undertaken.

Relevant Paper(s)

Author, date and country Patient group Study type (level of evidence) Outcomes Key results Study Weaknesses
Coll M-T et al,
n = 82 Age: 4–106 months and a 69-year-old adult. Symptom duration: 5–24 h. 2b Population-based prospective studyIncidence of meningitis with clear CSFCSF results (expressed in range) and organisms grown: Normal CSF in eight (9.7%) cases. Neisseria meningitides in all cases: 8 . Cell count 0–8/mm3. Glucose 2.5–4.05 mmol/l. Protein 0.14–0.4 g/l. Only cases of meningococcal meningitis were reviewed. The patient characteristics and symptoms did not differ significantly when groups with normal CSF and abnormal CSF were compared
Rapkin RH,
n = 5 Age: <1 yearCase reports Grade 4Cases of bacterial meningitis with clear CSFCell count 0–12 cells/mm3 Sugar 3–7.1 mmol/lProtein 0.1–1.0 g/lRepeat lumbar puncture: meningitis in all

S pneumoniae: 1H influenzae: 1Beta-haemolytic streptococcus: 1Listeria: 1Escherichia coli: 1
Children were initially treated as minor illness after the full infection screen, but repeat LP showed features of meningitis in all cases
Soskolne E,
South Africa
n = 4 Age: 6 months – 4 yearsCase reports Grade 4Cases of bacterial meningitis with clear CSFCell count 0 in 3 cases; 13 in 1 caseGlucose 2.8–4.9 mmol/lProtein 0.2–0.3 g/lRepeat CSF: meningitis

Initial CSF sterileRepeat CSF Neisseria meningitides: 4
In all four cases, lumbar puncture was repeated 12–48 h after the initial tap. All showed CSF suggestive of meningitis
Rosenthal J et al,
n = 107 Age: ≤13 years Symptom duration: <24 h2c Retrospective case reviewIncidence of meningitis with clear CSFNormal CSF in seven (6.5%) cases N meningitides: 3 . Cell count 0–5/mm3 H influenzae: 2 Glucose 2.2–4 mmol/l Escherichia coli: 1 Protein 0.1–0.6 g/l . S pneumoniae: 1Three children died in the normal CSF group. Case fatality rate was higher than those with initial abnormal findings (43% vs 10%)
Rao SP et al,
Meningitis in sickle cell disease n = 21 Non-sickle cell disease n=24 Age: <3 years2c Retrospective case reviewIncidence of meningitis in CSF with normal cell countSickle cell disease Normal cell count in six (28%) cases S pneumoniae: 5 . Cell count <5/mm3 H influenzae: 1 Glucose: normal in four cases Non-sickle cell disease. Protein: normal in four cases S pneumoniae: 1. Non-sickle cell disease Normal CSF cell count in one (4.1%) case Author did not elaborate the patient characteristics. Repeat lumbar puncture in sickle cell group showed pleocytosis in three cases. Two deaths were reported in the same group
Polk DB et al
n = 261 Age: 21 days – 18 months Symptom duration: 6–48 h2c Retrospective case reviewIncidence of meningitis with clear CSF Normal CSF in seven (2.7%) cases H influenzae: 2 Cell count 0–9/mm3 Streptococcus pneumoniae: 2 . Glucose 2–6.2 mmol/l Group B Streptococcus: 2. Protein 0.1–0.4 g/l N meningitides: 1 All patients observed in hospital as viral infection. One received amoxicillin prior to lumbar puncture
Teyssier et al,
n = 150 Age: >24 months2c Retrospective case reviewIncidence of meningitis with minimal/no CSF abnormalityNearly normal CSF in 18 (12%) cases No pleocytosis N meningitides: 5 . No pleocytosis in six cases (4%) H influenzae: 1 pleocytosis. CSF pleocytosis in 12 cases (8%) N meningitides: 11. CSF: blood glucose >0.5–14 cases S pneumoniae: 1. CSF protein in normal range: 17 casesIn strongly suspected cases CSF was repeated within a few hours and noted to be purulent
Hegenbarth MA et al,
Cases only Haemophilus influenzae meningitis n = 379 Age: 5–31 months2c Retrospective case reviewIncidence of meningitis with clear CSF Normal CSF in four (1.05%) cases H influenzae in all cases: 4 . Cell count 0–4/mm3 Glucose 2.5–5.2 mmol/l Protein 0.1–0.2 g/lBecause of normal CSF results treatment was delayed in all cases. One had high-frequency hearing loss
Moore CM et al,
n = 4 Age: 4 months – 3 yearsCase reports Grade 4Cases of bacterial meningitis with clear CSFCell count 0–14 cells/mm3Glucose 2.3–4.1 mmol/lProtein 0.09–0.2 g/lGram stain: no bacteria

Haemophilus influenzae: 2 Marchantia polymorpha: 1 Streptococcus pneumoniae: 1
Cases were treated as respiratory tract infections. One case had subdural empyema. In two cases LP was repeated and showed purulent CSF
Onorato et al,
n = 1 Age: 5 yearsCase reports Grade 4Cases of bacterial meningitis with clear CSFCell count 0Glucose 6.5 mmol/lProtein 0.3 g/l

N meningitides: 1
One isolated case report along with three similar cases in adults. Treatment was started after 8 h of hospital attendance and the child died after 40 h. Repeat LP in all cases showed meningitis


The majority of meningococcal infections occur in infants <5 years of age, with a peak incidence in those <1 year of age. However, because of increased immunisation coverage, the overall incidence of meningococcal infection in the UK is gradually declining (total laboratory-confirmed cases under 15 years in England and Wales has decreased from 1472 in 2000–2001 to 834 in 2007–2008). However, the incidence of type B meningococcal disease, as well as meningitis by other organisms still remains high, particularly among children under the age of 4 years. This is still a potentially devastating illness of childhood. Meningococcal disease has a case fatality rate of approximately 10%; however, more deaths are caused by septicaemia than by meningitis (HPA).

Early diagnosis and treatment is the key in reducing morbidity and mortality from meningitis, but this can be difficult because of the relatively non-specific symptoms and signs particularly in young children. Lumbar puncture is considered the definitive test; however, its sensitivity is not 100%. So, is it ever appropriate to start an antibiotic in a suspected case of meningitis even when the initial lumbar puncture is showing no abnormality? Or can antibiotics be safely withheld in these cases in children considering the wide prevalence of viral infections in this age group?

Bacterial meningitis presenting with normal CSF is uncommon but a well-recognised phenomenon. Although the incidence is variable, it can be as high as 10%(Coll)(proportion of meningitis cases that had normal CSF). “Normal” CSF in meningitis does not correlate with the age of the child or the subsequently demonstrated organism. It is related to the duration of illness, and incidence is higher when lumbar puncture is performed within the first 24 h of illness (before the inflammatory response has developed). Hence, Rapkin et al and Soskolne et al, suggested repeating the lumbar puncture after 24–48 h in clinically suspected cases. Rapkin et al also suggested repeating lumbar puncture at least in blood culture-positive cases if the initial CSF is clear.

Increased rates of complications including deaths have been reported in the group with initial normal CSF compared with initial abnormal CSF (death (43% vs 10%)(Rosenthal); (12.5% vs 3.1%)(Coll). In special circumstances such as sickle cell disease, one case series of 21 episodes of meningitis has reported the incidence with normocellular CSF response as high as 28% (Rao). The author suggested a sluggish inflammatory response owing to cellular defect in granulocyte chemotaxis in cases of sickle cell anaemia.

We are aware that our literature search will have specifically identified studies which have examined meningitis where the CSF showed no abnormalities. This biased search means that the prevalence shown is highly unlikely to reflect the true proportion of meningitis cases in which there is normal CSF, and the proportion of normal CSF results that are cases of missed meningitis.

Accepting this, the present review re-emphasises that normal CSF does not exclude the presence of bacterial meningitis. With the wider availability of PCR technique the diagnosis of meningitis even in early stages has improved in recent years. However, in the paediatric population, the complications arising from delayed or inadequately treated meningitis are serious. Hence it is appropriate to start antibiotic therapy in cases in whom clinical presentation is sufficiently concerning to warrant a lumbar puncture. In more doubtful cases, where there is still clinical concern about meningitis despite a normal CSF, a repeat of the lumbar puncture after a short period of time should be strongly considered.

Editor Comment

Some papers are published in imperial units and these have been converted into SI units for consistency.

Clinical Bottom Line

Normal cerebrospinal fluid (CSF) may not exclude bacterial meningitis in the early stage – Grade B.

Antibiotic use is justified when meningitis is strongly suspected even if the CSF results are normal – Grade B.

Repeat lumbar puncture should be undertaken when initial CSF shows no abnormality but there remains clinical concern – Grade C.


  1. Heath Protection Agency. Background information meningitis/meningococcal reviewed 23 Feb 2009. – meningitis (accessed 11 Mar 2009).
  2. Health Protection Agency. All laboratory confirmed cases of invasive meningococcal disease by serogroup, age and epidemiological year. Last reviewed 18 Feb 2009. – epidemiological data (accessed 11 Mar 2009).
  3. Coll M-T, Uriz M-S, Pineda V et al. Meningococcal meningitis with ‘normal’ cerebrospinal fluid. J Infect 1994;29:289–94.
  4. Rapkin R. Repeat lumbar puncture in the diagnosis of meningitis. Pediatrics 1974;54:34–7.
  5. Soskolne EI, Kuhn SH, Van Der Merwe PL, et al. The need of repeat lumbar puncture. S Afr Med J 1977;51:395–6.
  6. Rosenthal J, Golan A, Dagan R. Bacterial meningitis with normal cerebrospinal fluid findings. Isr J Med Sci 1989;25:186–8.
  7. Rao SP, Schmalzer E, Kaufman M et al. Meningitis in patients with sickle cell anemia: normocellular CSF at diagnosis. Am J Pediatr Hematolo Oncol 1983;5:101–3.
  8. Polk DB, Steele RW. Bacterial meningitis presenting with normal cerebrospinal fluid. Pediatr Infect Dis J 1987;6:1040–2.
  9. Teyssier G, Rayet I, Legall C, et al. Meningites. Bacterial meningitis with initially clear CSF (article in French). Pediatrie 1984;39:635–8.
  10. Hegenbarth MA, Green M, Rowley AH, et al. Absent or minimal cerebrospinal fluid abnormalities in Haemophilus influenzae meningitis. Paediatr Emerg Care 1990;6:191–4.
  11. Moore CM, Ross M. Acute bacterial meningitis with absent or minimal cerebrospinal fluid abnormalities. Clin Pediatr 1973;12:117–8.
  12. Onorato IM, Wormser GP, Nicholas P. “Normal” CSF in bacterial meningitis. JAMA 1980;244:1469–71.