Recurrent bacterial meningitis in children in the Netherlands: a nationwide surveillance study

STRENGTHS AND LIMITATIONS OF THIS STUDY

  • Nationwide surveillance study to estimate the recurrence rate after culture-positive bacterial meningitis in children, conducted over a period of 32 consecutive years.

  • The Netherlands Reference Laboratory for Bacterial Meningitis receives approximately 90% of all cerebrospinal fluid isolates from patients with bacterial meningitis in the Netherlands.

  • Underlying comorbidities of the patients could not be assessed.

  • No information on outcome after the first meningitis episode was available, and characteristics do define recurrence were missing for some patients, leading to an underestimation of the recurrence rate.

  • No relationship between immunisation and the risk of recurrence was found, which could partly be due to our study type. To study cause–effect relationships, a randomised controlled trial would be needed.

Introduction

Bacterial meningitis is a life-threatening infection with substantial mortality and morbidity.1 2 Causative pathogens vary by age, geographic location and seasonality.3 The introduction of routine use of polysaccharide conjugate vaccines against Haemophilus influenzae type b, Neisseria meningitidis and Streptococcus pneumoniae in paediatric immunisation programmes has substantially reduced the incidence of bacterial meningitis.4–6 In neonates, Streptococcus agalactiae (Group B streptococcus) and Escherichia coli cause the majority of community-acquired cases.4 7 In older infants and children, S. pneumoniae and N. meningitidis are the most common pathogens.4 8–10

Recurrence rates of bacterial meningitis of 4–9% have been reported in adults, with a median of 4–5 years between the first and second episode.11–13 Predisposing conditions such as cerebrospinal fluid (CSF) leak in adults with recurrent meningitis are common, but up to 23–26% of the adults with recurrent meningitis do not have predisposing conditions.11 12 Because of the high risk, the European Society for Clinical Microbiology and Infectious Diseases guideline on bacterial meningitis recommends to immunise adult survivors of bacterial meningitis.14 Pneumococcal vaccination is advised for adults after a first episode of pneumococcal meningitis. Meningococcal, pneumococcal and H. influenzae type b vaccines are recommended after bacterial meningitis for patients with CSF leakage.9

Data on bacterial meningitis recurrences in children are limited. A recent single-centre study from China found a recurrence rate of 2.3% in children, in which the majority of recurrences were caused by S. pneumoniae.15 In a retrospective multicentre study in France on recurrent pneumococcal meningitis in children between January 2001 and September 2015, a recurrence rate of 1.5% was observed.16 Here, we report results of a nationwide surveillance study to estimate the recurrence rate of bacterial meningitis in children in the Netherlands. We evaluate patient characteristics and pathogen distribution of first and recurrent bacterial meningitis episodes, whether recurrence rates changed over time, and analysed the association of various characteristics of the initial episode with the risk of a recurrent episode.

Methods

Study population

We identified patients with a first episode of culture-positive bacterial meningitis before the age of 18 years, between 1 July 1987 and 30 June 2019, using the nationwide surveillance database of the Netherlands Reference Laboratory for Bacterial Meningitis (NRLBM).4 17 18 The NRLBM receives approximately 90% of all CSF isolates from patients with bacterial meningitis in the Netherlands.19 20 Episodes from patients with missing name, sex or date of birth were excluded. Recurrent bacterial meningitis was defined as a patient with a positive CSF culture >28 days after the previous positive culture, or a subsequent episode that was caused by a different bacterial pathogen. There was no age restriction for recurrent episodes. Causative pathogens other than N. meningitidis, S. pneumoniae, H. influenzae, E. coli and S. agalactiae were defined as ‘other’ pathogens. Date of onset was defined as the date of collection of the first positive culture of blood or CSF. When unavailable, the date the isolate was sent to or received by the NRLBM was used. Based on the child’s age of onset of bacterial meningitis and the epidemiological year, we determined if the child had been eligible for vaccines according to the Dutch National Immunisation Programme before the recurrent episode occurred (online supplemental table 1).

Supplemental material

Serogrouping of meningococcal isolates was performed by Ouchterlony gel diffusion.21 Pneumococcal isolates were serotyped by coagglutination and subtyped by capsular swelling (Quellung method) using specific antisera (Statens Serum Institute, Denmark). H. influenzae isolates were serotyped using slide agglutination with polyclonal rabbit antisera to types a–f. S. agalactiae isolates were serotyped with the agglutination test using monospecific antisera.22 23 E. coli isolates were classified by phenotyping methods using the O- and H-antigens and since 2012 by using whole genome sequencing.18 24

Statistical analysis

Differences in proportions were compared using the Fishers exact test, or the McNemar’s test when proportions were correlated.25

Two different approaches were used to evaluate first and recurrent meningitis episodes over time. For the first approach, we determined the incidence of first and recurrent bacterial meningitis episodes on a population level. The annual incidence of first meningitis episodes was calculated as the number of children with a first meningitis episode in an epidemiological year (1 July–30 June) divided by the number of children (<18 years old) in the Netherlands in that year.26 The annual incidence of recurrent episodes was calculated as the number of children with a recurrence in an epidemiological year divided by the number of children in the Netherlands in that year. To evaluate trends in the incidence of first and recurrent episodes over time, the median annual incidence was calculated for the epidemiological periods 1988–2003 and 2004–2019. Incidence rate ratios (IRRs) with 95% CIs) were calculated using the Epitools package.27

For the second approach, we determined the risk of developing a recurrence within 3 years after the first bacterial meningitis episode for an individual patient. Due to our long inclusion period, the variation in follow-up time is relatively high. Therefore, follow-up time for each patient was restricted to 3 years after the first meningitis episode. Patients with less than 3 years of follow-up after their first meningitis episode were excluded from these analyses.

To study associations between predictors and recurrent meningitis, Cox proportional hazards regression (Survival package) was used.28 The proportional hazards assumption was assessed using the scaled Schoenfeld’s residuals. Linearity was assessed graphically by plotting the Martingale residuals and predictors with a non-linear relationship were categorised. Both univariable and multivariable ORs with 95% CIs were estimated. Statistical tests were two-sided and p values below 0.05 were considered to denote statistical significance. Analyses were performed using IBM SPSS Statistics V.26.0 (SPSS, Chicago, Illinois, USA) and Rstudio V.4.2.1.

Results

We identified 9731 children with a first episode of bacterial meningitis between 1 July 1987 and 30 June 2019 (figure 1), of whom 4373 (44.9%) were female (table 1). The number of first bacterial meningitis episodes strongly decreased during the observation period (figure 2A). In the general population, the median annual incidence rate decreased from 11.81 (IQR 11.26–17.60) per 100 000 children in the period 1988–2003 to 2.60 (IQR 2.37–4.07) per 100 000 children in the period 2004–2019 (IRR 0.25, 95% CI 0.23 to 0.26, figure 3A).

Figure 1
Figure 1

Selection of patients. CSF, cerebrospinal fluid.

Figure 2
Figure 2

(A) First bacterial meningitis episodes in children in the Netherlands, 1988–2019. (B) Recurrent bacterial meningitis episodes in children in the Netherlands, 1988–2019.

Figure 3
Figure 3

(A) The incidence of first episodes per 100 000 children in the Netherlands (sum of patients with a first bacterial meningitis episode divided by the number of children (<18 years old) in the Netherlands, per year). (B) The incidence of recurrent episodes per 100 000 children in the Netherlands (sum of patients with a recurrent bacterial meningitis episode divided by the number of children (<18 years old) in the Netherlands, per year). (C) The risk of a recurrence over time (the risk of a recurrent bacterial meningitis episode within 3 years after experiencing a first bacterial meningitis episode, per year).

Table 1

Patient characteristics

Age of onset was based on the difference between the date of birth and the culture date (6082 (62.5%)), the date the isolate was sent to (2441 (25.1%)) or received by (1208 (12.4%)) the NRLBM. There was a bimodal age distribution, with the first and largest peak in the first months of life and a second smaller peak around 15 years of age. The first episode of bacterial meningitis was caused by N. meningitidis in 5064 (52.0%), S. pneumoniae in 1736 (17.8%), H. influenzae in 1652 (17.0%), S. agalactiae in 555 (5.7%), E. coli in 298 (3.1%) and ‘other’ pathogens in 426 (4.4%) cases. Most prevalent meningococcal strains were serogroup B (82.7%) and C (14.9%). Pneumococcal episodes were most often caused by serotypes 14 (13.6%), 6B (10.4%), 18C (9.0%), 7F (8.8%), 19F (7.1%) and 19A (5.3%). Most prevalent H. influenzae strains were serotype b (91.4%) and non-typeable (6.9%). S. agalactiae serotypes were serotype III (75%) and IA (12%).

Of 9731 children with a first episode of meningitis, 63 (0.6% (95% CI 0.5% to 0.8%)) contracted recurrent meningitis comprising 85 recurrent episodes. Of these 63 children, 32 (51%) were female (table 1). Seventy-seven recurrent episodes in 56 patients occurred more than 28 days after the previous episode. Eight recurrent episodes in seven patients occurred <28 days but were caused by a different pathogen. A single recurrence occurred in 51 (81%) patients and 12 (19%) patients had more two or more recurrent episodes during the observation period. The absolute number of any recurrence per year (figure 2B) and the median annual incidence rate of recurrences per 100 000 children did not change during the observation period: from 0.06 (IQR 0.02–0.11) in the period 1988–2003 to 0.03 (IQR 0.00–0.06) in the period 2004–2019 (IRR 0.65 (95% CI 0.39 to 1.1), figure 3B).

Age of onset was based on the culture date (67 (78.8%)), the date the isolate was sent to (11 (12.9%)) or received by (7 (8.2%)) the NRLBM in the 85 episodes. In children with a recurrent episode of bacterial meningitis, the median age of the first recurrence was 62 months (IQR 12–140). The median duration between the first and recurrent episode was 8 months (IQR 2–23, table 1). Of all patients with a recurrence, 83% of the first recurrent episode occurred within 3 years and 97% occurred within 8 years. For an individual patient, the risk of a second episode in the first 3 years after the first episode of bacterial meningitis increased: 0.44 (95% CI 0.30 to 0.67) per 100 children in the period 1988–2001 to 0.90 (95% CI 0.56 to 1.4) per 100 children in the period 2002–2016 (IRR 2.1, 95% CI 1.2 to 3.6). The individual risk of a recurrence within 3 years is shown in figure 3C.

Most common causative pathogens of the recurrent episodes were S. pneumoniae in 44 (52%), N. meningitidis in 14 (16%), H. influenzae in 7 (8%) and E. coli in 5 cases (6%, table 1). ‘Other’ pathogens included Staphylococcus epidermidis (n=4, 5%), viridans streptococci (n=2, 3%), Klebsiella pneumoniae (n=2, 3%), Klebsiella oxytoca (n=1, 1%), Enterobacter cloacae (n=1, 1%), Enterococcus faecalis (n=1, 1%), Pseudomonas aeruginosa (n=1, 1%), Streptococcus dysgalactiae (n=1, 1%), Streptococcus pyogenes (n=1, 1%) and ‘mixed’ (n=1, 1%, H. influenzae and S. pneumoniae in culture). There were no recurrences after S. agalactiae meningitis. Most meningococcal recurrences were caused by serogroup B (n=9, 64%) and serogroup Y (n=4, 29%). Pneumococcal recurrences were most often caused by serotype 19F (n=6, 14%), 23F (n=4, 9%) and 6B (n=4, 9%). Of the recurrences caused by H. influenzae, 6 (86%) were of the non-typeable strain and 1 (14%) was serotype b. Three (60%) of five E. coli recurrent isolates carried the K1 capsular antigen.

Thirty-two (51%) of 63 children had one or more recurrent episodes that were caused by the same pathogen as the first episode: 19 by S. pneumoniae, 8 by N. meningitidis and 5 by E. coli (online supplemental figure 1). Three (16%) of 19 S. pneumoniae recurrences were caused by the same serotype (14, 18C and 19F). This was the case for 4 (50%) of 8 N. meningitidis recurrences (2 by serogroup B, 1 by C and 1 by Y). Three (60%) of five E. coli recurrences were caused by the same O-group (two by O18 and one by O16). All five patients with recurrent E. coli meningitis were younger than 90 days at the time of the first episode and recurrences occurred within 35–48 days.

The recurrence rate was highest after a first meningitis episode with ‘other’ pathogens and E. coli: 2.6% (95% CI 1.5 to 4.6) and 2.4% (95% CI 1.1 to 4.8), respectively. The recurrence rate was 1.4% (95% CI 1.0 to 2.1) after pneumococcal meningitis, 0.2% (95% CI 0.1 to 0.4) after meningococcal meningitis and 0.5% (95% CI 0.3 to 1.0) after H. influenzae meningitis. The proportional hazards assumption was met for the outcome of recurrence after 28 days, but not for the combined outcome of a recurrence after 28 days or a second episode with another pathogen within 28 days. We therefore present the results for Cox regression for a recurrence after 28 days only. Several characteristics were associated with recurrence in the multivariable analysis: age above 5 years at the first meningitis episode (OR 3.6 (95% CI 1.5 to 8.3)) and a first episode due to pathogens other than N. meningitidis (table 2). Being eligible for vaccination was not associated with the risk of recurrence.

Table 2

Factors associated with bacterial meningitis recurrence

Of all patients with a first meningitis episode, 883 (9%) were eligible for H. influenzae type b (Hib) vaccination after the first meningitis episode, 7245 (74%) for meningococcal serogroup C (MenC), 1480 (15%) for serogroup A, C, W and Y (MenACWY), 115 (1%) for 7-valent pneumococcal conjugate vaccine (PCV7) and 66 (1%) for 10-valent pneumococcal conjugate vaccine (PCV10). The median follow-up duration was 9.0 years (IQR 4.3–12.8) after potential Hib vaccination, 24.1 years (IQR 19.3–28.0) after MenC vaccination, 9.0 (IQR 4.3–12.8) after MenACWY vaccination, 11.1 years (IQR 9.4–12.0) after PCV7 vaccination and 6.8 years (6.2–7.3) after PCV10 vaccination. The median follow-up for the children who were not eligible for vaccination was 25.9 years (IQR 21.2–28.8).

Thirty (35%) of the 85 recurrent episodes in 23 patients were caused by a pathogen and subtype that is covered by the vaccines that are currently included in the Dutch National Immunisation Programme (online supplemental table 1). Of these 23 patients with a recurrence due to a vaccine-covered pathogen, 13 patients (67%) were eligible for vaccination before the recurrence occurred. However, in 12 of those patients, the disease-causing pathogen of the recurrence was not included in their previous vaccination. Only one child with a recurrent episode was, based on its age, eligible for vaccination for the corresponding pathogen before the recurrence occurred. This patient would have received the PCV7 vaccine at the age of 2, 3, 4 and 11 months. The first meningitis episode occurred at the age of 5 years and 5 months and was due to S. pneumoniae (serotype 35). The recurrence occurred at the age of 5 years and 8 months, with S. pneumoniae (serotype 19F). There was no difference in the risk of recurrence between patients who were eligible for vaccination before their recurrence (29 (0.79%) of 3692) and patients who were not eligible (34 (0.65%) of 6039, p=0.18).

Compared with PCV10, the 13-valent pneumococcal conjugate vaccine (PCV13; additional coverage of serotype 3, 6A and 19A) would have covered an additional two recurrences due to S. pneumoniae. A combination of vaccines against Hib, MenACWY, meningococcal serogroup B (MenB) and PCV13 offered after the first meningitis episode would cover 41 (48%) instead of 30 (35%) recurrences (p<0.01, (online supplemental figure 2). The potential of increased coverage of the recurrent episodes with additional vaccines was highest for the subgroups of patients with a first episode due to S. pneumoniae, N. meningitidis or H. influenzae. A combination of vaccines against Hib, MenACWY, MenB and PCV13 would cover 37 (56%) of all 66 recurrences in this subgroup, compared with 4 (21%) of all 19 recurrences that occurred after a first episode due to a different pathogen than S. pneumoniae, N. meningitidis or H. influenzae (p=0.01).

Discussion

In this nationwide surveillance study of children with culture-positive bacterial meningitis, covering a 30-year period and including 9731 children, we showed that the recurrence rate of bacterial meningitis in children is 0.6% (95% CI 0.5% to 0.8%). The most common causative pathogen of the recurrent episodes was S. pneumoniae (52%) followed by N. meningitidis (16%). For patients with a recurrence within 28 days, most recurrences were caused by the subgroup of ‘other’ pathogens. The recurrence rate in our study is lower than those previously reported, ranging from 1.3% to 2.3%.15 29 These previous studies included also patients without culture-confirmed bacterial meningitis which may have inflated the number of recurrences.15 29 These studies did not note the number of culture-confirmed patients.15 29 Our recurrence rate for pneumococcal meningitis is similar to the 1.5% that has previously been reported.16 The overall risk for recurrence in children is considerably lower than the recurrence rate for adults, ranging from 1.1% to 6%.11–13 30 The difference between reported rates for recurrences of children and adults may be due to longer follow-up times for adults in retrospective cohort studies, differences in predisposing factors and lower rates of vaccination or waning immunity after vaccination in childhood.31–34

Children with a first episodes caused by ‘other’ pathogens had the highest recurrence rate. The majority of these recurrences were also caused by these ‘other’ pathogens and half of them occurred within 28 days. Given the aetiology and short time interval between episodes, these patients might have suffered from hospital-acquired bacterial meningitis, that is, neurosurgical procedures, or head trauma.35 36 The recurrence rate was also high after E. coli meningitis (2.4%). Most of these recurrences were also due to E. coli and occurred within 7 weeks. Recurrences have been reported in 6–21% of infants with E. coli meningitis, occurring within 1–28 days after completion of antibiotic treatment.34 37–40 The aetiology of these recurrences and optimal prevention strategies remain unclear. Hypothetically, such high recurrence rates with the same pathogen could be caused by delayed sterilisation of CSF, and therefore, repeated lumbar puncture and longer antibiotic treatment duration will need to be evaluated.40 Larger clinical cohorts including bacterial sequencing studies to differentiate between continued infection or reinfection are needed. Using the definition of >28 days between both episodes for recurrence, a first episode in older children and first episodes due to other pathogens than N. meningitidis were associated with higher risks of recurrence.

The impact of the implementation of different vaccines in the Dutch National Immunisation Programme is demonstrated by the declined incidence rate of first episodes over time in our study. After introduction of several conjugate vaccines, the annual incidence of meningitis caused by these pathogens decreased, as expected.3 4 We did not find a concurrent decline in annual incidence of recurring bacterial meningitis episodes. There was an increase in the risk of recurrence for an individual patient after experiencing a first meningitis episode. No association was found between introduction of vaccines in the Dutch National Immunisation Programme and the risk of recurrent meningitis. This indicates that vaccination is more effective in preventing a first than a second episode, which might be due to the fact that a substantial proportion of children with a recurrence have underlying comorbidities. Most recurrences are due to S. pneumoniae or N. meningitidis, especially after a first episode of H. influenzae, pneumococcal or meningococcal meningitis. Although the recurrence rate in children is lower than in adults, a more than 1 in 200 chance of bacterial meningitis is still considerably higher than in the general population. Therefore, it seems reasonable to maximise coverage against S. pneumoniae or N. meningitidis after a first episode of bacterial meningitis. For instance, by adding PCV13, PPV23, MenB and MenACWY vaccines to routine immunisation programmes for patients who experienced a first episode of bacterial meningitis. PPV23 alone is poorly immunogenic in children younger than 2 years of age.41 However, when PPV23 is administered in children previously vaccinated with PCV7, PPV23 has the ability to induce stronger antibody responses.42 The Dutch Society of Paediatrics recommends replacement of PCV10 by PCV13 and additional vaccination with PPV23 after a first bacterial meningitis episode.43 Also, additional MenB vaccination is recommended after a first meningococcal meningitis episode or known CSF leak.43 The benefit of vaccination remains uncertain, as controlled studies on the effect of vaccination in recurrent meningitis patients are not available. On the other hand, the chance of developing a recurrence is substantially higher than the general risk of developing bacterial meningitis. Also, vaccination is minimally invasive and the risk for adverse events is low. Therefore, maximising coverage against the most common pathogens, including S. pneumoniae, may contribute to risk reduction. The Spanish Association of Paediatrics recommends MenACWY vaccination for patients with a history of invasive meningococcal disease (IMD), regardless of the serogroup.44 Other European countries and the European Centre for Disease Prevention and Control do not recommend additional vaccination for IMD survivors.45–52 Vaccines against E. coli are currently not available for clinical use, but phase 1 and 2 studies in adults are ongoing.53 Another strategy to reduce the risk of recurrent meningitis is to eliminate or treat risk factors. Recurrence of bacterial meningitis has been associated with several underlying comorbidities.15 16 54 Previous studies report predisposing conditions in 83–93% of young children with a recurrent bacterial meningitis.15 16 Most prevalent predisposing conditions include congenital or traumatic causes of CSF leakage and occur in children with a recurrence due to S. pneumoniae.15 16

Our study has several limitations. First, the NRLBM receives around 90% of all CSF samples from children with positive CSF cultures in the Netherlands. While this represents a substantial inclusion rate, approximately 10% of children with culture-positive bacterial meningitis will have been missed. Our estimate of recurrence will therefore be a conservative estimate. Second, underlying comorbidities of the patients could not be assessed. However, the high diversity of less common causative pathogens in our cohort, including pathogens that are considered to be more associated with nosocomial infections, suggests the presence of severe underlying comorbidities in some of these patients.36 Third, we used national surveillance data, and no information on outcome after the first meningitis episode was available. Therefore, our recurrence rate is an underestimation because the recurrences can only have occurred in patients who survived the first episode. Furthermore, characteristics to define a recurrence (name, date of birth, sex) were missing for some patients. This might have led to missed recurrences in patients who were excluded. Fourth, information on treatment type and duration was not available. Therefore, some recurrences might actually have been relapses due to inadequate treatment. Fifth, the actual immunisation status was unknown. It is likely that most children in our cohort who were eligible for vaccination were actually vaccinated, since the uptake of the Dutch National Immunisation Programme has been approximately 90–95% over the past decades.55 However, our cohort could be subject to some selection bias, as the risk of bacterial meningitis is highest in children who are not vaccinated. Sixth, the long inclusion and follow-up period of our cohort could lead to an overestimation of the IRR of recurrent episodes: children with a first bacterial meningitis episode <2004 have a longer follow-up time and this logically increases the risk of developing a recurrence, compared with children with a first bacterial meningitis episode ≥2004. We attempted to account for this by performing a second analysis for a subcohort of children with a follow-up time limited to 3 years. Last, we did not find a relationship between immunisation and the risk of recurrence, which could partly be due to our study type. To study cause–effect relationships, a randomised controlled trial would be needed.

In conclusion, the recurrence rate of bacterial meningitis in children in the Netherlands over the past three decades was at least 0.6%. Age over 5 years and a first episode due to pathogens other than N. meningitidis were associated with a higher risk of recurrence. While the incidence rate of first episodes decreased substantially over time, this was not the case for recurrent episodes. The risk of experiencing a recurrent bacterial meningitis episode increased over time for an individual patient who survives a first episode. We found no clear association between immunisation and the risk of recurrence. However, expanding vaccination after a first meningitis episode could be beneficial, considering the relatively high risk of developing a recurrence after a first bacterial meningitis episode.

Data availability statement

Data are available upon reasonable request. Although Dutch data protection regulations do not allow sharing of individual participant data, datasets with selected aggregated data are available upon reasonable request. Proposals should be directed to the corresponding author. Data requestors will be asked to sign a data access agreement.

Ethics statements

Patient consent for publication

Ethics approval

The NRLBM is a nationwide surveillance database. Data from this database that was used for the current manuscript were anonymised. Therefore, individual patients are not traceable. Patient consent is not necessary for this type of study under Dutch Law.

Acknowledgments

We thank the technicians of the Netherlands Reference Laboratory for Bacterial Meningitis (NRLBM) for the collection and analyses of the bacterial isolates.

This post was originally published on https://bmjopen.bmj.com