OUP user menu

Pneumococcal meningitis in childhood: a longitudinal prospective study

Pasquale Pagliano, Ugo Fusco, Vittorio Attanasio, Marco Rossi, Annalisa Pantosti, Marco Conte, Francesco Saverio Faella
DOI: http://dx.doi.org/10.1111/j.1574-695X.2007.00324.x 488-495 First published online: 1 December 2007


After implementation of programmes for active immunization against Haemophilus influenzae b, Streptococcus pneumoniae and Neisseria meningitidis became the most common agents of bacterial meningitis in childhood. Over a 9-year period, children showing clinical and laboratory findings of meningitis on the basis of their positive cultures of blood or cerebro-spinal fluid (CSF) for S. pneumoniae were enrolled. Predisposing conditions, clinical and laboratory findings, and microbiological and imaging studies were considered. Meningitis-related death or neurological sequelae defined an unfavourable outcome. Sixty-four patients met the inclusion criteria. Thirty-one (48%) children had predisposing conditions to pneumococcal meningitis. Fever and neck stiffness were the main symptoms; 14 patients (22%) reported seizures before admission. Twenty-one patients required treatment in the intensive care unit (ICU). Streptococcus pneumoniae strains were penicillin susceptible in 54 cases (84%). Forty-eight children (75%) showed complete recovery. Two patients (3%) died, and 14 (22%) had sequelae. Patients with a low CSF cell count, low neutrophils, early admission to ICU or infection by penicillin-nonsusceptible strains of S. pneumoniae had an unfavourable outcome more frequently. Low blood neutrophils, low CSF cell count, early admission to ICU and infection by penicillin-nonsusceptible strains are the main factors predicting an unfavourable outcome in children with pneumococcal meningitis.

  • meningitis
  • CSF
  • children
  • outcome
  • Streptococcus pneumoniae
  • penicillin susceptibility


The epidemiology of bacterial meningitis in childhood has changed significantly during the past few years, mainly because of programmes for routine immunization of infants with conjugate Haemophilus influenzae b (Hib) vaccine. In developed countries, where immunization programmes against Hib were strongly sustained, Hib meningitis has declined dramatically and Streptococcus pneumoniae and Neisseria meningitidis have become the most common agents of bacterial meningitis in childhood (Schuchat et al., 1997; Dawson et al., 1999).

Despite continuous improvements in antibiotic therapy and intensive treatment, pneumococcal meningitis remains a severe and life-threatening disease. In fact, large trials performed during the past 20 years, involving adult or paediatric cases, report case–fatality rates ranging between 4% and 16% and neurological sequelae in over 30% of the survivors (Kornelisse et al., 1998; Arditi et al., 1998; Auburtin et al., 2002; Kastenbauer & Pfister, 2003).

Identification of the factors predicting the outcome of pneumococcal meningitis is of great value as it permits the identification of patients at a higher risk of intracranial complications who may benefit from early diagnosis and intensive treatment. Among the baseline characteristics, the prognostic value of penicillin susceptibility was suspected on the basis of some case reports describing the poor efficacy of monotherapy with third-generation cephalosporins, when penicillin- or cephalosporin-resistant strains were involved, but was not confirmed by adult or paediatric studies of patients with pneumococcal meningitis (Bradley & Scheld, 1997; Kaplan, 2002; Kellner et al., 2002; McMaster et al., 2002).

The aim of the present study was to investigate the baseline clinical and laboratory features of childhood pneumococcal meningitis and their influence on outcome, and also to evaluate the role of penicillin susceptibility.


This was a prospective study of children with pneumococcal meningitis referred to the Department of Emergency and Infectious Diseases at the D. Cotugno Hospital (Naples, Italy) between January 1997 and December 2005. Approval for the study was obtained from the internal review board, and all patients' legal representatives gave written informed consent to participate in the study.


Consecutive children affected by bacterial meningitis were evaluated if they tested positive either on urine or on cerebro-spinal fluid (CSF) for pneumococcal antigens. A case of pneumococcal meningitis was defined by characteristic clinical signs and symptoms (i.e. neck stiffness, a bulging fontanel, headache or vomiting), CSF pleocytosis (>10 cells µL−1), detection of pneumococcal antigens in either urine or in CSF samples or isolation of S. pneumoniae from blood or CSF cultures. If lumbar puncture was not performed at admission, the diagnosis of pneumococcal meningitis was based on clinical signs and symptoms and isolation of S. pneumoniae from blood cultures. The inclusion criteria for the study were: (1) positive blood or CSF cultures for S. pneumoniae; (2) availability of the S. pneumoniae susceptibility assay performed by an E-test; and (3) ages ranging from 1 month to 14 years. The exclusion criterion was HIV coinfection.

Data collection

Demographic data, detailed information about previous or underlying diseases, presenting signs and symptoms (fever, defined as a body temperature >38 °C, neck stiffness, focal neurological deficit, coma), findings of the CSF exam (cell count, protein and glucose concentrations) and evidence obtained from laboratory and neuroradiographic studies were recorded on admission. The data obtained from clinical evaluation, laboratory exams and neuroradiographic studies performed during hospitalization, as well as the findings of a complete clinical evaluation performed 8 weeks after treatment were recorded for each subject on a standardized case report form.

Microbiological studies

Blood and CSF samples were obtained for microbiological studies, including pneumococcal antigen detection before starting any antibiotic treatment. Blood and CSF samples were inoculated in a BACTEC Plus Aerobic/F (Becton-Dickinson) and were incubated in BACTEC 9240 device (Becton-Dickinson). Positive cultures were inoculated in sheep blood agar and chocolate agar and incubated in 5% CO2. The identification of the strains was confirmed by conventional methods (e.g. optochine susceptibility test). Susceptibility assays of penicillin, cefotaxime, rifampin, vancomycin and erythromycin were performed using an E-test with the Clinical Laboratory Standard Institute interpretative criteria for minimum inhibitory concentration (MIC). Susceptibility of the pneumococcal strains obtained during the course of the study was evaluated by the authors' laboratory and confirmed using an E-test by the reference laboratory of the ‘Istituto Superiore di Sanità’ in Rome, Italy, as part of a national surveillance programme of bacterial meningitis. The serotyping of the strains of S. pneumoniae was performed by the authors' Laboratory of Microbiology or by the Laboratory of Microbiology of the Istituto Superiore di Sanità (ISS), as reported elsewhere (Pantosti et al., 2000).

Imaging studies

A computed tomographic (CT) scan of the brain was scheduled at admission. During the hospital stay and the 8-week follow-up period, repeated CT or magnetic resonance imaging (MRI) scans of the brain were performed for patients with either positive findings on admission or suspicion of cerebral oedema, or in the presence of focal neurological deficit. A CT scan of the brain was performed immediately before discharge in all cases, regardless of the CT findings on admission.


A 14-day course of antimicrobial therapy was scheduled. During the first 3 years of the study, children received third-generation cephalosporin (ceftriaxone 100 mg kg−1 intravenously every 24 h or cefotaxime 100 mg kg−1 intravenously every 8 h), as empirical therapy. In this period, a multidrug therapy containing third-generation cephalosporin and vancomycin (15 mg kg−1 intravenously every 6 h) or rifampin (8 mg kg−1 intravenously every 12 h) was administered only to comatose or immunodepressed patients. During the last 6 years of the study, because of the increasing diffusion of penicillin-nonsusceptible strains of S. pneumoniae, study protocol changed (Pantosti et al., 2000; Whitney et al., 2000). Therefore, the empirical therapy was changed to third-generation cephalosporin plus rifampin or vancomycin (vancomycin was given additionally to children aged 6 months or less). When microbiological studies were available, antibiotic treatment was continued on the basis of the susceptibility test, and monotherapy with ceftriaxone or cefotaxime was administered only to children with pneumococcal meningitis due to a penicillin-susceptible strain. Besides antibiotic therapy, children received dexamethasone (0.15 mg kg−1 intravenously every 6 h). The latter was administered 1 h before the antibiotics on four consecutive days starting on admission.


An adverse clinical outcome was defined as meningitis-related death or evidence of neurological sequelae in survivors. The latter included motor deficit, hearing loss, behavioural or language disturbance and hydrocephalus. These conditions were diagnosed by means of a complete clinical exam of sensory and motor functions performed 8 weeks after the end of the therapy. The clinical examination was integrated by audiological evaluation and CT or MRI scan of the brain, when required.

Statistical analysis

The Fisher exact test and a two-tailed χ2 test were used to compare qualitative variables. Quantitative data were expressed as medians [interquartile range (IQR)] and compared using the Mann–Whitney U-test. Two-tailed P-values of <0.05 were considered to be statistically significant. Variables achieving statistical significance at a 90% level in the univariate analysis were simultaneously considered by multivariate logistic regression analysis to determine independent factors of adverse outcome. The results of the multivariate analysis are presented as estimated odds ratio, with corresponding 95% confidence intervals. spss 13.0 software was used for statistical analysis.


Eighty-six children with pneumococcal meningitis were observed during the study period. Of these, 18 were excluded because of negative cultures despite positive CSF pneumococcal antigen, three because of lack of susceptibility data, and one because of HIV coinfection. Overall, 64 children were included in the study; the sex ratio (male : female) was 1.9 : 1, and the median (IQR) age was 26 (7–89) months. No patient had previously received antipneumococcal vaccine.

One or more predisposing conditions to pneumococcal meningitis were recognized in 31 (48%) children. Eighteen reported recurrent ear/sinus infections, seven had a CSF leak or experienced a severe head trauma or underwent neurosurgery (Table 1). Predisposing conditions were not found in 21 (68%) of 31 patients younger than 24 months and in 12 (36%) of 33 patients older than 24 months (P=0.01; odds ratio 3.7, confidence interval 1.2–11.9).

View this table:
Table 1

Predisposing conditions in 31 of 64 children with pneumococcal meningitis

ConditionsNumber of cases%
Recurrent ear or sinus infections1828
Common variable immunodeficiency35
Recent or remote head trauma35
Previous neurosurgery23
CSF leak23
Status post splenectomy12
Diabetes mellitus12
Bone marrow transplantation12
Chronic hepatitis12
  • CSF, cerebrospinal fluid.

The length of general symptoms of illness before the administration of intravenous antibiotic therapy ranged between 1 and 4 days (median 1 day). On admission, 60 (94%) patients had fever, 55 (86%) had neck stiffness and 16 (25%) were comatose. Seizures before admission were reported in 14 (22%) patients, 12 had generalized seizures and two had focal seizures. Twenty-one patients were admitted to the intensive care unit (ICU) because of a rapidly deteriorating clinical condition; 13 patients required mechanical ventilation. All ICU admissions occurred within 48 h from the diagnosis of meningitis.

Lumbar puncture was performed on admission in 61 (95%) children, while in three it was delayed because of CT evidence of increased intracranial pressure. The findings of the CSF exam revealed a high cell count, high protein and low glucose concentration in all patients (Table 2).

View this table:
Table 2

CSF exam findings and their relationship with disease severity

FindingsICU admission (19 patients)No ICU admission (42 patients)All cases
Proteins [mg dL−1; median (IQR)]338 (250–328)319 (181–482)320 (200–530)
Glucose [mg dL−1; median (IQR)]32 (5–61)17 (5–48)19 (19–45)
Cell count [ × 103 cells µL−1; median (IQR)]900 (150–4000)1900 (900–6650)1600 (600–6250)
  • CSF, cerebro-spinal fluid; ICU, intensive care unit.

CT scan performed on admission showed abnormalities in 14 (22%) patients: brain oedema in 14, arterial infarction in two, and sinus thrombosis in two.

Empirical therapy included third-generation cephalosporin monotherapy in 15 patients and a combined administration of cephalosporin plus vancomycin (six patients) or rifampin (43 patients) in the remaining 49 cases. Two patients had a maculopapular skin eruption because of an allergic reaction to ceftriaxone, respectively, 2 and 7 days after administration of the first dose. Both received meropenem (120 mg kg−1 day−1, divided in three doses) to complete the course of therapy. No other toxicity was reported, and, no case of bleeding occurred after dexamethasone therapy. Time from evidence of symptoms and start of i.v. antimicrobial therapy ranged between 24 and 70 h and was quite similar for cases infected by nonsusceptible strains of S. pneumoniae and for those infected by susceptible strains.

Streptococcus pneumoniae was isolated from CSF cultures in 16 patients, from blood cultures in seven patients and from both blood and CSF cultures in 41 patients. Overall, two strains were fully resistant to penicillin (MIC 3 µg mL−1) and eight showed intermediate resistance (MIC 0.1–1 µg mL−1). Two strains had intermediate resistance to cefotaxime (MIC 2 µg mL−1); both strains were fully resistant to penicillin. Twenty-one strains were resistant to erythromycin; none was resistant to vancomycin or rifampin. One (7%) of 15 patients observed during the first 3 years of the study and nine (18%) of 49 cases observed during the last 6 years were infected by penicillin-nonsusceptible strains of S. pneumoniae (P=0.28). All cases infected by nonsusceptible strains received combined treatment since admission.

Association through serotypes and outcome for 52 strains of S. pneumoniae are reported in Table 3. Serotypes were analysed by the Laboratory of Microbiology of the ISS in an earlier study and by the authors' Laboratory of Microbiology during the last period of the study. The overall coverage of pneumococcal conjugate vaccines (PCV) was 54% for the heptavalent (PCV-7) and nanovalent (PCV-9) vaccines, 57% for the nanovalent vaccine (PCV-11) and 67% for the 13-valent vaccine (PCV-13). Children infected by S. pneumoniae serotype 9 had the highest incidence of unfavourable outcome.

View this table:
Table 3

Serotype distribution and outcome of 52 patients with pneumococcal meningitis

SerotypeFavourable outcomeUnfavourable outcomeSequelae
6 A30
6 B31Hearing loss
9 N02Hearing loss, hemi paresis
9 V12Death, hemi paresis
11 A30
1442Hemi paresis, hearing loss
15 B20
18 A20
18 C31Cognitive impairment
19 A11Hemi paresis
19 F21Hearing loss
23 A11Hemi paresis
23 F41Cognitive impairment
  • NT, nontypeable.

The median (IQR) time of defervescence was 4 (2–6) days. No difference occurred in patients infected by penicillin-susceptible or -nonsusceptible strains of S. pneumoniae. During hospitalization, 11 (17%) patients had meningitis-associated intracranial complications: eight had subdural hygroma, two had hydrocephalus and one had sinus thrombosis. One of these cases needed permanent CSF drainage.

Forty-eight (75%) children did not show sequelae. Two (3%) children died: one 3 days after admission because of severe septic shock, and the other after 14 days because of cardiac arrest. The latter children had no electroencephalographic activity 48 h after admission. The case–fatality ratio was 2% for patients infected by penicillin-susceptible strains and 10% for those infected by nonsusceptible strains. One or more sequelae were reported in 14 (22%) patients (six were infected by penicillin-nonsusceptible strains). Seven reported paresis, four hearing loss, two hydrocephalus and two cognitive impairment.

Table 4 reports the baseline findings vs. outcome, as assessed by univariate analysis. The CSF cell count (P=0.001), peripheral white blood cells (P=0.001) and blood neutrophils (P=0.0001) were lower in patients with unfavourable outcome. ICU admission within 48 h from the onset of symptoms (P<0.00001) and infection sustained by a nonsusceptible strain of S. pneumoniae (P=0.012) resulted more frequently in an adverse clinical outcome. When variables achieving significance at a 90% level by univariate analysis were simultaneously evaluated by multivariate logistic analysis, only admission to ICU was associated with an increased risk of unfavourable outcome (odds ratio 23.6, 95% confidence limits 3.12–178.9, P=0.0001). To better observe the prognostic value of laboratory and microbiological findings, a multivariate model was used where the continuous variables achieving significance by univariate analysis were dichotomized according to the 25th percentile value. Both CSF cell count below 600 µL−1 (odds ratio 10.4, 95% confidence intervals 2.6–52.6, P=0.002) and penicillin nonsusceptibility (odds ratio 11.2, 95% confidence intervals 1.6–83.9, P=0.009) were predictive of an unfavourable outcome (Table 5).

View this table:
Table 4

Base-line findings in respect to outcome

FindingsFavourable outcome (48 patients)Unfavourable outcome (16 patients)P
Age<24 months20110.09
Age >24 months295
Treatment within 48 h34120.51
Treatment after 48 h144
ICU admission8130.00001
No ICU admission403
CSF cell count # [cells µL−1; median (IQR)]2000 (1100–6500)550 (142–1325)0.001
CSF glucose [mg dL−1; median (IQR)]31 (5–46.5)10 (3–22)0.13
CSF proteins [mg dL−1; median (IQR)]347 (200–1090)315 (254–512)0.39
Blood leucocytes [ × 103 cells µL−1; median (IQR)]22.5 (16.1–27.5)14.1 (7.5–19.0)0.001
Blood neutrophils [ × 103 cells µL−1; median (IQR)]19.1 (13.4–24.6)12.6 (6.1–16.1)0.0001
  • P are calculated by Mann–Whitney U-test and by Fisher's exact test as appropriate.

  • The analysis was done for 61 patients, because three, respectively one with favourable and two with unfavourable outcome did not undergo to lumbar puncture at admission.

  • PNSP, Penicillin-nonsusceptible Streptococcus pneumoniae; PSSP, penicillin-susceptible Streptococcus pneumoniae; CSF, cerebro-spinal fluid; ICU, intensive care unit.

View this table:
Table 5

Prognostic value of laboratory and microbiological findings as assessed by multivariate analysis

FindingsOdds ratio95% CIP
CSF cell count<600 µL−110.42.6–52.60.002
Penicillin nonsusceptibility11.21.6–83.90.009
White blood cell count<13 850 µL−11.9450.017–225.90.784
Blood neutrophils<11 950 µL−11.9450.017–225.90.784
  • Continuous variables achieving significance by univariate analysis were dichotomised according to 25th percentile value.


Pneumococcal meningitis in childhood is a life-threatening disease determining a high incidence of neurological sequelae in survivors. The case–fatality ratio and sequelae reported in this study were lower than those observed in other European and North-American series (Arditi et al., 1998; Auburtin et al., 2002; Kellner et al., 2002; McMaster et al., 2002). Many factors, including wide use of combined antimicrobial therapy and administration of dexamethasone, may have favoured the better outcome (Bradley & Scheld, 1997; Kaplan, 2002; McMaster et al., 2002; Buckingham et al., 2006).

The prognostic value of the CSF cell count emerging from the analysis of the present data may be sustained on the basis of clinical and experimental evidences and may have a therapeutic impact. The association between CSF cell count and outcome was reported in a retrospective study of adult pneumococcal meningitis and was related to the serotype involved in another retrospective study considering the outcome of pneumococcal meningitis cases sustained by serotypes 1, 3 and 9 V (Kastenbauer & Pfister, 2003; Ostergaard et al., 2004). Similar findings were reported in a model of pneumococcal meningitis demonstrating that the bacterial growth within CSF and the outcome were influenced by the increase of peripheral neutrophils obtained by granulocyte colony-stimulating factor or by the attenuation of CSF pleocytosis obtained by fucoidin (Brandt et al., 2004; Brandt et al., 2005). On the basis of these clinical and laboratory findings, it may be supposed that an insufficient inflammatory response against pneumococcal antigens during the first hours of the disease permits a bacterial overgrowth within the CSF. In these cases, when antimicrobial therapy is administered, the considerable bacterial lysis results in more pronounced inflammatory changes within the meninges and vascular complications are favoured. Children with a relatively low CSF cell count have to receive high-dose steroids and combined antimicrobial therapy to reduce the bacterial load, meningeal inflammation and, consequently, neurological complications as quickly as possible (Kornelisse et al., 1995; Müller et al., 1995; Ries et al., 1997; Waterer et al., 2001).

The spread of antibiotic-resistant pneumococcus strains with high-level penicillin resistance and multidrug resistance represents a concrete threat. Indeed, when penicillin-nonsusceptible strains were involved, about half of the cases reported an unfavourable outcome, unless an antimicrobial therapy active against penicillin-nonsusceptible strains of S. pneumoniae was firmly adopted. The association between penicillin susceptibility and outcome has not been reported in any previous study of pneumococcal meningitis, but was reported in studies evaluating the outcome of patients affected by pneumococcal pneumonia, as confirmed by a large meta-analysis (Tleyjeh et al., 2006). It is believed that the larger definition of unfavourable outcome adopted in this study, including both mortality and sequelae, gave more statistical efficacy to the sample and permitted the observation of the influence of penicillin susceptibility on outcome. However, because other factors, such as the pneumococcal strain involved and the patients' underlying conditions, may influence the outcome of pneumococcal meningitis, the prognostic value of penicillin susceptibility needs further investigation by larger multicentre trials (Bradley & Scheld, 1997; Aronin et al., 1998; Cabellos et al., 2000; Kaplan, 2002; Kellner et al., 2002; Martinez-Lacasa et al., 2002; McMaster et al., 2002; Ostergaard et al., 2004; Tunkel et al., 2004; van de Beek et al., 2004).

Although children affected by pneumococcal meningitis frequently suffer a transient hearing loss and regain normal hearing over the weeks or months following meningitis, a number of them still show hearing sequelae. The present study highlighted an incidence of hearing loss lower than other studies of patients receiving third-generation cephalosporin and vancomycin. As an association between precocious vancomycin administration and hearing loss was reported, it is believed that, in the cases in this study, the wide use of rifampin, whose penetrability within the inner ear and the cochlear aqueduct is relatively high, may have favoured the better outcome (Bhatt et al., 1993; Richardson et al., 1997; Arditi et al., 1998; Buckingham et al., 2006). These findings, coupled with the relatively low number of patients showing an unfavourable outcome reported in this study, suggests that the use of rifampin in combination with ceftriaxone or cefotaxime is a potentially attractive therapeutic strategy for children with pneumococcal meningitis. Of note, rifampin penetration within CSF is not affected by concomitant use of dexamethasone, and experimental evidence demonstrates lower mortality rates and lower levels of CSF lipoteichoic acids in animals with pneumococcal meningitis receiving rifampin (Nau et al., 1999; Gerber et al., 2003).

In an immunocompetent host, immune defence mechanisms, mainly secretory IgA and an efficacious antibody- and complement-mediated phagocytosis, do not permit the invasion of pneumococcus in the intravascular space and its survival within the bloodstream. Younger children show a relatively lower ability to contain the infection and frequently acquire pneumococcal meningitis by an uncontrolled haematogenous diffusion of S. pneumoniae. In this study, there is no evidence of extrameningeal foci of infection (e.g. sinus or middle ear infection) in 21 (68%) of 31 children aged 24 months or less. This finding is a strong argument for the administration of the PCV during the first 2 years of life, because it promotes an efficient cell-mediated immune response against more common strains of S. pneumoniae (Janoff et al., 1999; Black et al., 2000; Hausdorff et al., 2000; Koedel et al., 2002; Obaro & Adegbola, 2002; Klugman et al., 2003; Whitney et al., 2003; Bogaert et al., 2004). PCV is not routinely administered by the local national health system, and heptavalent vaccine-type strains of S. pneumoniae accounted for 54% of serotyped strains. It is estimated that the large use of PCV-7 (the only conjugate vaccine available in Italy) had prevented at least half of the cases observed in children aged 24 months or less (Lebel et al., 2003). The value of the pneumococcal vaccine is greater if it is considered that serotypes covered by PCV-7 are involved in the more severe cases and that many younger children having bacterial meningitis experience neurobehavioural sequelae, also if evaluated many years after the illness (Anderson et al., 2004).

In conclusion, pneumococcal meningitis in childhood is a severe but preventable disease, either by conjugate vaccine or early treatment of extrameningeal foci of infection. CSF and haematological findings are useful in identifying more severe cases requiring prompt intensive care. On the basis of the incidence of penicillin-nonsusceptible strains and the beneficial effect of dexamethasone on mortality and neurological sequelae, use of combined protocols, containing drugs with a good blood–brain barrier penetrability regardless of meningeal inflammation and steroids use, such as rifampin or linezolid, may become an excellent therapeutic strategy (McIntyre et al., 1997; Yogev & Guzman-Cottrill, 2005; Faella et al., 2006).


The authors have reported no conflict of interest.


  • Editor: Ewa Sadowy


View Abstract