Background. Patients may experience multiple episodes of bacterial meningitis. Information from large studies of recurrent meningitis is limited. We evaluated the incidence of recurrent bacterial meningitis and the distribution of causative organisms in The Netherlands.
Methods. Data for patients with bacterial meningitis were prospectively collected nationwide for the period 1988–2005. Recurrent meningitis was defined as an episode of meningitis that either occurred ⩾28 days after a previous episode or occurred <28 days after a previous episode but was caused by a different pathogen or different subtype of the same pathogen.
Results. Of 18,915 patients, 202 (predominantly male) patients had recurrent bacterial meningitis (P<.01). Prevailing causative organisms were Streptococcus pneumoniae (40% of cases), Neisseria meningitidis (22%), and non-type b Haemophilus influenzae (9%). Pneumococci serotypes included in the heptavalent vaccine caused only 36% of cases of recurrent pneumococcal meningitis. The proportion of episodes caused by meningococcus serogroups W135, Y, and Z was higher among patients with recurrent meningitis than among those with nonrecurrent meningitis (odds ratio, 12.8), and the proportion caused by nontypeable and type f H. influenzae was also higher among patients with recurrent meningitis (odds ratio, 3.8 and 5.6, respectively).
Conclusions. In The Netherlands, the prevalence of recurrent bacterial and fungal meningitis is low. The distribution of causative microorganisms differs between cases of recurrent meningitis and cases of nonrecurrent meningitis; this could be associated with vaccination.
Bacterial meningitis is a life-threatening infectious disease with a reported case-fatality rate of 3%–30%, depending on the pathogen isolated (Neisseria meningitidis, Streptococcus pneumoniae, and Haemophilus influenzae type b [Hib] are the most common) and the age at which the meningitis occurs [1, 2]. Also, in developed countries such as The Netherlands, bacterial meningitis remains a major cause of death during the neonatal period. Patients who survive often experience neurological sequelae, such as sensorineural hearing loss, seizures, and cognitive impairment [1–3].
The implementation of conjugate vaccination has changed the epidemiology of bacterial meningitis in children since the 1990s. Routine vaccination of children against Hib has virtually eradicated Haemophilus meningitis from the developed world [2, 4, 5]. The incidence of meningitis due to serogroup C meningococci also decreased substantially after the introduction of routine vaccination .
Recently, the heptavalent pneumococcal vaccine was included in the Dutch vaccination program, which involves vaccination at the ages of 2, 3, 4, and 11 months. Reduction of pneumococcal bacterial carriage after the introduction of vaccination has been demonstrated in other countries. In a study from northern California, the rate of pneumococcal otitis and invasive pneumococcal disease due to any of the 7 serotypes included in the vaccine also decreased after the implementation of a vaccination program .
The majority of patients with bacterial meningitis will experience only 1 episode of bacterial meningitis. However, bacterial meningitis may recur in some patients. Such recurrences may be related to neurosurgical procedures or CSF leakage or may be attributed to an underlying primary or acquired immunodeficiency 8–15. In addition to these patient-related factors, environmental factors could also contribute to an elevated risk of meningitis.
Reported data on the exact incidence of recurrent meningitis and the variety of causative microorganisms are scarce. A recently published study provided some insight into local incidences of recurrent meningitis . Thus far, nationwide studies have, to our knowledge, not been performed. Here, we evaluated the epidemiologic characteristics and causative organisms for both primary and recurrent bacterial meningitis over an 18-year period, during which vaccines against Hib (1993) and serogroup C meningococci (2002) were introduced in the national vaccination program but just prior to the introduction of pneumococcal vaccination (2006).
Subjects. We used data from the prospective nationwide surveillance on bacterial meningitis performed by The Netherlands Reference Laboratory for Bacterial Meningitis (NRLBM). The NRLBM was officially established in 1975 as a collaboration between the Department of Medical Microbiology, Academic Medical Center (Amsterdam), and the National Institute of Public Health and the Environment (Bilthoven) to collect nationwide isolates recovered from CSF and blood samples from patients who had proven meningitis (i.e., patients had positive CSF culture results and possibly had positive blood culture results) or who had bacteremia and suspected meningitis (i.e., patients had positive blood culture results alone). By estimation, the NRLBM receives >90% of isolates recovered from patients with meningitis in The Netherlands (16 million inhabitants) . We analyzed the isolates recovered during the 18-year period 1988–2005. Information on the comorbidity of recurrent meningitis and on the immunological status of patients is not included in the NRLBM's surveillance.
Microbiology. Strains of bacteria isolated from patients with bacterial or fungal meningitis, with the exception of Mycobacterium species, were submitted to the NRLBM by hospital laboratories throughout the country. The strains were identified and typed as described elsewhere [18, 19].
Definitions. All episodes of meningitis confirmed by a positive CSF culture result, as well as all episodes of suspected meningitis confirmed by positive results of blood cultures alone, were considered to be bacterial meningitis and were included in the analyses. In addition to episodes of bacterial meningitis, we also included episodes of fungal meningitis (due to Cryptococcus neoformans and Candida species) for comparative analyses of male predominance (see Results).
We excluded episodes that were confirmed on the sole basis of recovery of a pneumococcal isolate from a blood specimen, because a substantial proportion of episodes of pneumococcal bacteremia are possibly attributable to pneumonia in such cases. Episodes involving Mycobacterium species were excluded from the analysis because of uncertainty regarding the notification of mycobacterial meningitis. Furthermore, mycobacteria account for only a negligible proportion of cases of bacterial meningitis in The Netherlands (n<10).
We identified different episodes of meningitis in a single person as entries in the database that had the same name, initials, and date of birth but a different date of sampling. We defined recurrent meningitis as an episode of bacterial (or fungal) meningitis that met the criteria described above and that satisfied 1 of the following statements: an episode of bacterial meningitis that occurred in the same person ⩾28 days after a previous episode or an episode in the same person that occurred <28 days after a previous episode but that was caused by a different pathogen or a different subtype of the same pathogen.
Statistical analysis. Incidences were calculated using population figures obtained from Statistics Netherlands (Centraal Bureau voor de Statistiek; http://www.cbs.nl) with the use of StatLine . For the statistical analysis of our data, we used SPSS, version 12.0.1 (SPSS). Differences between groups were evaluated using Fisher's exact test or the χ2 test for categorical data, as appropriate. A P value <.05 was considered to be statistically significant.
Primary bacterial meningitis. During 1988–2005, a total of 19,163 episodes of bacterial meningitis in 18,915 individuals were identified by the NRLBM. Figure 1 presents the annual incidence of primary bacterial meningitis. Two peaks can be distinguished: 1989–1993 and 2001–2002. The highest incidence occurred in 1989 (8.2 episodes per 100,000 inhabitants). In 2005, the incidence reached its lowest level during our observation period (4.9 episodes per 100,000 inhabitants).
table 1 presents the baseline characteristics of patients at the time of diagnosis of bacterial meningitis. Sex was recorded for 17,843 (94.3%) of the 18,915 patients; the proportion of male patients was significantly higher than that of female patients (9503 [50.2%] vs. 8340 [44.1%] of 18,915 patients; P<.001). Age was recorded for 18,844 patients (99.6%); the median age at the time of diagnosis of bacterial meningitis was 7 years (interquartile range, 1–42 years). Three incidence peaks occurred (figure 2): during young age (age, 0–8 years), adolescence (age, 13–18 years), and older age (age, 60–92 years).
The most common causes of nonrecurrent bacterial meningitis were N. meningitidis (49%), followed by S. pneumoniae (19%) and Hib (11%) (table 2). Together, these 3 species accounted for 79% of episodes of nonrecurrent meningitis.
Recurrent bacterial meningitis. Of the 19,163 episodes recorded, 450 (2.3% of all episodes) occurred in 202 patients (1.1% of all patients) who experienced >1 episode of bacterial meningitis during the study period. Of these patients, 169 (84% of patients with recurrent bacterial meningitis) had 2 episodes, 25 (12%) had 3 episodes, 5 (3%) had 4 episodes, 2 (1%) had 5 episodes, and 1 (<1%) had 7 episodes. Of the 202 patients who met the criteria for recurrent bacterial meningitis, 178 (88%) had experienced a second episode of bacterial meningitis ⩾28 weeks after the previous episode, whereas 24 patients (12%) had experienced another episode <28 weeks after the onset of a previous episode, although the episode was due to a different pathogen. None of the episodes of recurrent bacterial meningitis that occurred <4 weeks after a prior episode were caused by a different subtype of the same pathogen.
The distribution of causative pathogens for episodes of recurrent meningitis differed from that for episodes of nonrecurrent meningitis (P<.001) (table 2). The proportion of episodes caused by S. pneumoniae, non-type b H. influenzae, streptococci other than S. pneumoniae, group A streptococci, and coagulase-negative staphylococci was higher for episodes of recurrent disease than for episodes of nonrecurrent disease (P<.001 for each comparison). This was also true for fungal meningitis caused by C. neoformans but not for that due to Candida species (data not shown) (table 2).
Patients in whom the first episode of meningitis was caused by S. pneumoniae or N. meningitidis were more likely to be infected with the same pathogen in subsequent episodes than with a different pathogen. All 61 patients whose first episode of meningitis was due to S. pneumoniae and 32 (62%) of 52 patients whose first episode of meningitis was due to N. meningitidis experienced recurrent episodes of meningitis due to pneumococci and meningococci, respectively. In contrast, this was less likely to occur in patients in whom the first episode of meningitis was caused by H. influenzae. Of 26 patients with primary H. influenzae meningitis, only 7 (27%) had recurrent meningitis episodes solely caused by H. influenzae. This difference could possibly be explained by the overall smaller prevalence of H. influenzae meningitis, compared with meningitis due to the other 2 encapsulated pathogens.
Sex distribution for episodes of recurrent bacterial meningitis. Recurrent bacterial meningitis occurred more often in male patients than in female patients (118 [58%] vs. 78 [39%] of 202 patients; P=.005). The ratio of male patients to female patients in the recurrent group was 1.3-fold larger than that in the nonrecurrent meningitis population (1.5:1 vs. 1.1:1; relative risk for male vs. female patients, 1.3; 95% CI, 1.0–1.8). This indicates that the higher risk that male persons will experience bacterial meningitis is even more pronounced for recurrent meningitis.
Sex distribution varied with regard to the most prevalent causative microorganisms of recurrent meningitis (figure 3A). Recurrent pneumococcal meningitis episodes occurred more often in male patients than in female patients (117 male patients [67%] among 174 episodes). Recurrent episodes of N. meningitidis occurred almost equally among male and female patients (43 male patients [44%] among 97 episodes). In contrast, recurrent Hib meningitis was mainly observed among female patients (1 male patient vs. 11 female patients [8%]). The predominance of episodes of recurrent pneumococcal meningitis involving male patients could be ascribed mainly to recurrent episodes among adolescents and adults (i.e., those aged 13–60 years; 71 male patients [79%] among 90 episodes of recurrent pneumococcal meningitis vs. 63 male patients [54%] among 117 episodes of recurrent nonpneumococcal meningitis; OR, 3.2; 95% CI, 1.7–6.0) (figure 3B).
HIV infection is considered to be highly relevant with regard to the susceptibility to meningitis caused by C. neoformans ; this was underlined by the finding of male predominance among patients with cryptococcal meningitis (recurrent and nonrecurrent; data not shown). Because pneumococcal meningitis occurs more frequently among HIV-infected patients than does meningitis caused by other bacteria [21, 22], we hypothesized that, after the introduction of HAART, the sex difference among patients with pneumococcal meningitis would diminish. Therefore, we analyzed the sex distribution for episodes of pneumococcal meningitis in the pre-HAART era (1987–1995) with the sex distribution in the HAART era (1997–2005). The ratio of male to female patients changed dramatically, from 3.6:1 (indicating a male predominance) to 1.4:1 for episodes of recurrent pneumococcal meningitis, suggesting HIV infection as a possible cause of meningitis in a number of these patients. In contrast, for episodes of nonrecurrent pneumococcal meningitis, the ratios of male patients to female patients were 1.1:1 and 1.2:1 for the pre-HAART and HAART eras, respectively.
Age distribution for recurrent bacterial meningitis. When divided into relatively arbitrary age categories (0–3 months, 4–12 months, 1–5 years, 6–12 years, 13–25 years, 26–40 years, 41–60 years, and >60 years), some important characteristics were noticed for patients with nonrecurrent bacterial meningitis. As expected, Streptococcus agalactiae group B and Escherichia coli were the most prominent pathogens for patients aged 0–3 months (44.5% and 21.8%, respectively). The most common cause of recurrent bacterial meningitis in these children was E. coli (32%). Otherwise, S. pneumoniae was the most commonly isolated pathogen (33%–66% of cases), except for adolescents, in whom N. meningitidis was most frequently isolated.
Serotype and serogroup distribution. The serotype or serogroup distribution of the 3 most prevalent causative bacteria among the episodes of recurrent meningitis (S. pneumoniae, N. meningitidis, and H. influenzae) was analyzed. The overall serotype distribution for episodes of recurrent pneumococcal meningitis was not significantly different from that for episodes of nonrecurrent meningitis. However, the 10 most common pneumococcal serotypes for all episodes of nonrecurrent meningitis were more often responsible for nonrecurrent than recurrent episodes (64.6% vs. 45.8%; OR, 0.5; 95% CI, 0.3–0.6) (table 3). The proportion of the serotypes covered by the 7-valent conjugated pneumococcal vaccine (4, 6B, 9V, 14, 18C, 19F, and 23F) was 46.9% and 35.7% for nonrecurrent and recurrent pneumococcal meningitis, respectively (OR, 1.6; 95% CI, 1.2–2.2).
Serogroup B meningococci were the predominant cause of nonrecurrent and recurrent meningococcal meningitis, accounting for 79% and 62% of cases, respectively (table 3). The proportion of the rare serogroups W135, Y, and Z was larger for episodes of recurrent meningococcal meningitis than for episodes of nonrecurrent meningitis (OR, 12.8; 95% CI, 7.7–21.5). In addition, serogroups A and 29E were not observed among episodes of recurrent meningitis.
The distribution of H. influenzae serotypes is shown in table 3. Type b was the main cause of nonrecurrent meningitis, accounting for 69% of all episodes of H. influenzae meningitis. The proportion of cases due to nontypeable H. influenzae was larger for episodes of recurrent meningitis, compared with nonrecurrent meningitis (31 [60%] of 52 vs. 816 [28%] of 2941 episodes; OR, 3.8; 95% CI, 2.2–6.7).
Because routine vaccination against Hib, which began in 1993, could have affected the serotype distribution for nonrecurrent meningitis , we compared the H. influenzae serotype distribution for recurrent versus nonrecurrent meningitis episodes in the prevaccination period (1988–1993) with that in 2 equally sized vaccination periods (1994–1999 and 2000–2005). The proportion of cases due to Hib for episodes of recurrent Haemophilus meningitis decreased in the 3 periods (50%, 10%, and 9%, respectively), concomitantly with a decrease in the proportion among episodes of nonrecurrent meningitis (91%, 48%, and 28%, respectively).
Our analysis of 19,163 episodes of bacterial meningitis in The Netherlands over an 18-year period (1988–2005) demonstrates that the encapsulated microorganisms S. pneumoniae (40%), N. meningitidis (22%), and non-type b H. influenzae (9%) caused 71% of the recurrent episodes of meningitis. Age and sex differences among patients with recurrent meningitis suggest differences in the mechanisms of disease and host defense against the causative microorganisms.
The proportion of episodes caused by S. pneumoniae, non-type b H. influenzae, coagulase-negative staphylococci, and C. neoformans is ∼2 times higher for recurrent meningitis than for nonrecurrent meningitis. S. pneumoniae was the most frequent cause of recurrent bacterial meningitis. Pneumococci are very diverse, with 90 recognized serotypes; many of these serotypes are found among the pneumococci that cause invasive disease. Although the serotype distributions for episodes of recurrent meningitis and for episodes of nonrecurrent meningitis were, to a large extent, similar, differences were observed in groupwise comparisons.
Recurrent pneumococcal meningitis is usually caused by intracranial seeding of bacteria via a direct communication with an extracranial source [23, 24]. Because clinical data were not included in the database, we can only speculate about the underlying etiology of recurrent disease. First, this condition is frequently manifested by intermittent CSF rhinorrhea of various etiologies [25, 26]. Second, closed-head injury resulting from road-traffic accidents and trauma is the main predisposing factor . This could, to some extent, explain the predominance of male patients among those with pneumococcal recurrent meningitis observed in our study. In support hereof is the fact that, in The Netherlands, approximately two-third of the victims of lethal traffic accidents are male . Other factors that possibly explain this difference are alcoholism and cigarette smoking. Because we do not have data on substance abuse in our population, we were not able to study these factors in greater detail. Finally, a predominance of male patients has also been reported for postcraniotomy meningitis, suggesting as-yet-unidentified factors involved in the host defense against pneumococci .
Immune mechanisms may contribute to an increased susceptibility to recurrent pneumococcal meningitis [15, 28, 29]. Primary immunodeficiencies may contribute to recurrent infection [8, 9], and many are x-linked [14, 15], but these immunodeficiencies are rare and present early. Common variable immunodeficiency is an acquired immunodeficiency with a peak incidence in young adulthood, but it has a normal sex distribution and is also infrequent, affecting ∼1 in 25,000 individuals [30, 31].
Apart from head injury and CSF leakage, HIV infection was considered to be relevant with regard to the predominance of male patients among persons with recurrent meningitis in our study. The incidence of invasive pneumococcal disease among HIV-infected patients has been reported to be 46–100 times greater than that for the general population, and recurrences are common [12, 22, 32, 33]. The evaluation of pneumococcal recurrent meningitis in the pre-HAART and HAART eras revealed a dramatic change in the ratio of male patients to female patients, from 3.6:1 to 1.4:1 for episodes of recurrent pneumococcal meningitis, suggesting that underlying HIV infection makes a possible contribution to the data observed.
Our study shows that the 7-valent vaccine (introduced in The Netherlands in April 2006) covers 46% and 36% of the serotypes that caused nonrecurrent and recurrent pneumococcal meningitis, respectively. In other countries, the introduction of 7-valent vaccination resulted in the replacement of vaccine serotypes with nonvaccine serotypes (e.g., serotypes 11, 15, 19A, 29, and 33), many of which are penicillin resistant [34–38]. However, the positive effect of herd immunity, as demonstrated after vaccination against Hib and against pneumococci [39–41], may outweigh the negative impact of replacement. Herd protection may be particularly beneficial to individuals at risk because of potential reductions in the efficacy of immunization, as indicated for HIV-infected individuals [12, 42, 43]. Therefore, the overall incidence of recurrent pneumococcal meningitis may decrease over the coming years in both vaccinated and unvaccinated individuals.
The proportion of cases due to nontypeable H. influenzae and to H. influenzae type f was higher for the recurrent meningitis group than for the nonrecurrent meningitis group. It was recently found that, in cases involving H. influenzae type f, ∼50% of the patients were immunocompromised . In 1993, the Hib-conjugated vaccine was introduced in the Netherlands, diminishing the incidence of Hib meningitis [1, 2]. Comparison of the distribution of serotypes among episodes of recurrent and nonrecurrent meningitis in pre- and postvaccination periods indicated that the incidence of meningitis caused by H. influenzae type b among the population determines its occurrence among episodes with recurrent disease.
Serogroup B was the major cause of meningococcal disease in The Netherlands during the study period . In addition, episodes of recurrent meningococcal meningitis were mostly caused by serogroup B meningococci (62%). In contrast, the proportion of cases due to the rare serogroups W135, Y, and Z among 99 episodes of recurrent meningococcal disease was higher than that among the episodes of nonrecurrent meningococcal disease, whereas serogroups A and 29E were only seen in cases of nonrecurrent meningitis. Patients with recurrent meningococcal disease may have defects in the complement system [46–49]. At least 13 of these 99 episodes occurred in 6 individuals with a complement defect . Screening tests to assess the activities of classic (CH50) and alternative (AP50) pathways are simple and reliably identify patients if another cause of recurrent meningococcal disease seems less probable. We believe that screening of complement defects should be performed for all patients who present with recurrent meningococcal disease.
The database used in this study allowed us to describe the nationwide epidemiology. The risk of initial meningitis for the general population could be calculated. Because the collection took place over several years, we were also able to study changes over time. Patients with bacterial meningitis were identified through the finding of a causative agent. However, follow-up data for patients were unavailable. For this study, we only included episodes of culture-proven bacterial or fungal meningitis, to ensure that we did not include nonmeningitis illnesses or meningitis with a viral etiology. This could have resulted in underreporting, especially with regard to patients for whom treatment was started before a culture specimen was obtained. We also did not have information about underlying diseases in the patients with bacterial meningitis.
In summary, of 18,915 patients included in the study cohort, 202 patients (1.1%) experienced a recurrence of meningitis, mostly due to encapsulated bacterial pathogens. In general, an immunological defect of some sort may be suspected in the event of a recurrence of meningitis if no CSF leak is found. We found clear differences in age- and sex-associated susceptibility among the patients with recurrent bacterial meningitis. More research is needed to elucidate these differences.
We thank the staff of The Netherlands Reference Laboratory for Bacterial Meningitis for their cooperation and aid in setting up the database. We are indebted to Dr. Cees Fijen, who let us use his personal database of complement-deficient families, and to Brid Ryan and Drs. Cees Fijen and Diederik van de Beek, who critically read the manuscript. We also thank the anonymous reviewers for their helpful comments in further improving the manuscript.
Potential conflicts of interest. All authors: no conflicts.