Professional Documents
Culture Documents
Recurrent Pneumococcal Meningitis in Children: A Multicenter Case-Control Study
Recurrent Pneumococcal Meningitis in Children: A Multicenter Case-Control Study
The Pediatric Infectious Disease Journal • Volume 38, Number 9, September 2019 www.pidj.com | 881
Copyright © 2019 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Darmaun et al The Pediatric Infectious Disease Journal • Volume 38, Number 9, September 2019
Statistical Analyses
The frequency of RPM cases in the national database was
first calculated. Then, characteristics of children were compared
between the first PM episode in cases and the PM episode in con-
trols. The distribution of cases and controls over the study period
was described. All PM episodes in cases were compared with the
PM episode in controls. Cases and controls were compared for
patient outcomes and risk factors for PM. Quantitative data were
analyzed by Student t test or nonparametric Mann-Whitney U test
as applicable. Categorical data were analyzed by χ2 or Fisher exact
test. The analysis was performed with the software Epi-Info 6.04fr
(US Centers for Disease Control and Prevention, Atlanta, GA) and
Stata SE 13.1 (Stata Corp, College Station, TX). P <0.05 was con-
sidered statistically significant.
RESULTS
FIGURE 1. Details of the selection of cases (RPM) and From 2001 to September 2015, we collected data on 1634
controls (SPM) from the ACTIV national pediatric meningitis cases of PM in 1596 children (median age: 1.0 years [interquar-
database. tile range: 0.5–4.2]; male/female ratio: 1.39). In total, 24 (1.5%)
patients presented RPM, with 2 or more PM episodes. The median
time between the first and the second episode was 21 months (inter-
chain reaction results in CSF.4,5 Cases were all children 18 years
quartile range: 6–37). The distribution of first PM episodes in cases
of age or younger, with RPM defined as the presence of at least
(and controls) was homogeneous over the study period, with a peak
2 episodes of PM with an interval between episodes of at least in 2007–2008 (see appendix, Supplemental Digital Content 1;
28 days identified in the national database during the study http://links.lww.com/INF/D477). These 24 patients were matched
period. Each case was matched to 2 randomized controls identi- to 48 controls with an SPM. Thus, we included 72 children (median
fied in the national database (Fig. 1) who had SPM in the year age: 3.9 years [2.5–7.5]; male/female ratio: 1.25) with a total of 110
of the first PM episode in the case and were born the same year. episodes of PM (Fig. 1).
The winter season was defined by the period between Decem- The comparison of the first PM episode in cases with the
ber and March. Risk factors for PM identified in the literature PM episode in controls and all PM episodes between cases and
were CSF leakage, presence of a ventricular shunt, congenital controls is presented in Table 1. The frequency of ENT infections
or acquired ID, any type of asplenia and presence of a cochlear was higher in controls than cases. RPM cases were less frequent
implant.4,15 Risk factors for CSF leakage were trauma (surgical than SPM cases during the winter season (27% vs. 48%, P = 0.03).
or not), inflammation, congenital factors and neoplasm.19 For Cases and controls did not differ in the proportion of identified
patient outcomes, deaths were those resulting from the men- PCV serotypes (15/47 [32%] vs. 18/34 [53%], P = 0.06). In these
ingitis episode; deafness was defined by a significant hearing patients, PCV serotypes were no longer identified after 2010. Sero-
loss and needing regular follow-up with an ear, nose and throat type 24F was the most common non-PCV serotype identified (see
(ENT) specialist and hearing aids. Neurologic sequelae included appendix, Supplemental Digital Content 2; http://links.lww.com/
epilepsy, sensory-motor deficiency and mental deficiency, all INF/D478). Cases and controls did not differ in penicillin suscepti-
requiring follow-up in a neuropediatric unit. bility (69% vs. 81%, P = 0.11). Seizures overall were less frequent
in cases than controls (11% vs. 29%, P = 0.02). The only difference
Process of Data Collection in blood tests was for blood cultures, less often positive in first PM
Cases and controls were identified in the ACTIV/GPIP episodes for cases than controls (50% vs. 76%, P=0.05). Among
national database. For each PM episode, we collected data on 38 RPM cases, 8 (21%) occurred in 6 children with antimicrobial
anamnestic (sex, age and season at admission, presence of siblings, prophylaxis (penicillin V).
risk factors for IPD), vaccinal (PCV7, PCV13 data), clinical (ENT The rate of abnormal psychomotor development before the
infection, seizure, coma, septic shock, purpura, assisted mechani- first episode of PM was more frequent in cases than controls when
cal ventilation), biologic (blood cell count, C-reactive protein considering all episodes (47% vs. 22%, P=0.01) (Table 1). Other
and procalcitonin levels, CSF data, search for ID), microbiologic short-term outcome and sequelae did not differ between cases and
(method of pneumococcal identification, serotype if available and controls (Table 2).
antimicrobial susceptibility) and therapeutic (antimicrobial and At the first PM episode, cases and controls did not differ in
steroid treatment) features. Additional data on the outcome (com- identified predisposing factors for PM (Table 3). ID was searched
plications, search for and treatment of CSF leakage, intensive care in 48% of all PM episodes and was more frequent in cases than
unit admission, sequelae and death) were retrospectively collected controls (58% vs. 29%, P = 0.01). Only 1 additional ID was
from the pediatric units. detected, in controls. At the end of all episodes, ID was searched
in 88% of cases (see table, Supplemental Digital Content 3; http://
Microbiology links.lww.com/INF/D479). The IDs were congenital (n = 4; 1 Down
S. pneumoniae was identified by standard methods in the syndrome and 3 functional asplenia) and acquired (n = 3; 1 HIV
microbiology laboratory of each hospital. Isolates were serotyped infection, 1 myelodysplastic disease and 1 chemotherapy for brain
at the National Reference Center for pneumococci by latex agglu- tumor). Brain imaging was performed in 86% of all PM episodes,
tination with antisera provided by the Statens Serum Institute with no differences between cases and controls. After the first epi-
(Copenhagen, Denmark). Antimicrobial susceptibility testing for sode, brain imaging was considered normal in 32/59 PM episodes
penicillin and ceftriaxone or cefotaxime was performed at National (54%), with no differences between cases and controls (55% vs.
Reference Center for pneumococci as previously described.1 54%, P = 0.93).
Copyright © 2019 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
The Pediatric Infectious Disease Journal • Volume 38, Number 9, September 2019 Recurrent Pneumococcal Meningitis
TABLE 1. Description of the Population With RPM (Cases) Compared With 48 Children With
SPM Episodes (Controls)
At the first PM episode, CSF leakage was identified as a less frequent than SPM cases in winter, with fewer concomitant
predisposing factor in 25% of cases and 11% of controls (P=0.12) ENT infections. CSF leakage was highly frequent in RPM cases
(Table 3). At the end of all episodes, CSF leakage was identified when compared with SPM cases (83% vs. 10%, P < 10−7). No ID
in 20 cases and 5 controls (83% vs. 10%, P<0.01). Nine (64%) was found in RPM cases, but 15% of controls had an ID previously
cases of congenital CSF leakage were located in the anterior cra- known or discovered after the PM episode (P = 0.09).
nial fossa. The predisposing factor was still unclear in 4 cases, 3 of The rate of RPM in our study is close to data previously
whom died of the PM. reported. In 1999, Drummond et al17 reported a 1.3% prevalence
of recurrent bacterial meningitis in children. A multicenter study
in Denmark performed between 1980 and 2013 found 59/2418
DISCUSSION (2.4%) children with recurrent IPDs, 24 of whom experienced 1 or
The rate of RPM in children was 1.5%, representing 3.8% of more PM episodes.12 We show that RPM cases were less predomi-
1634 PM cases over a 15-year period. RPM cases were significantly nant in winter when compared with SPM cases. The predominant
Copyright © 2019 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Darmaun et al The Pediatric Infectious Disease Journal • Volume 38, Number 9, September 2019
Cases Controls
Variables (n = 24) (n = 48) P
occurrence in winter of IPDs is reported in the literature, with a study by Gaschignard et al,6 ID was found in 26/163 patients with
peak in February in the northern hemisphere.6,20,21 Several studies IPD (16%; 95% CI, 11–22) but in none of the 11 with RPM. In the
have shown a link between seasonal viral infections (influenza, res- study by Mason et al,8 10/108 recurrent IPD cases were RPM; of
piratory syncytial virus) and the occurrence of PM21 and IPD.20,22–24 these 108 cases, 15 experienced more than 2 IPDs, and of these 15,
Because RPM was linked to the presence of CSF leakage, a winter- only 1 had RPM, with Mondini deformity. In the study by Ingels et
promoting factor was probably not necessary. al,12 among the 10 patients with at least 3 IPD episodes, only one
Many clinical cases have been published on the association had RPM that was related to CSF leakage. Seven of the remaining
between CSF leakage and RPM25–28 but not as systematically as 9 cases of recurrent pneumococcal bacteremia had an identified
our series. In the recurrent IPD cases in the study by Ingels et al,12 ID.12 In another Spanish study of 23 recurrent IPD cases in 10
only 4/24 patients (17%) with 1 or more PM episodes showed patients, 2 cases exhibited ID.7
CSF leakage. Alsina et al7 described 3 RPM cases in their recur- The short-term evolution of RPM was comparable to that
rent IPD cohort: all had CSF leakage. More generally for recur- of SPM. From the first recurrence, the PM was even less clinically
rent IPDs, CSF leakage was identified in 7%–75% of patients7,8,12 severe in cases than controls (fewer seizures, less ventilatory assis-
but sometimes never.9 The original feature of our study is that tance), which could be explained by earlier in-hospital medical
CSF leakage was the only source of infection found when the IPD advice by parents who recognized the first meningitis symptoms
was an RPM. early. The absence of differences in neurologic outcome and death
In our population, most congenital CSF leakage cases between cases and controls may also be related to this early medical
(58%) were diagnosed after the second PM episode. Therefore, alertness by parents of children with RPM. However, it may also
for the others, there was a real delay in the initial diagnosis of be due to the exclusion of patients who died or to a lack of power
CSF leakage. However, 33% of cases had a predisposing history of because of the small number of patients.
CSF leakage at the first episode (previous neurosurgery or inner- Although recruitment was multicenter for almost 15 years,
ear surgery). The risk of CSF leakage is estimated at 6.3% after RPM remained a rare event, with a small study sample (n = 24).
craniotomy in children29 and from 0.9% to 8.5% after transsphe- The number (but not the rate) may be underestimated, because
noidal endoscopic surgery.30,31 In our study, after the first PM epi- the completeness of the ACTIV database is estimated at 61%.18 In
sode, the source of infection remained unknown in 63% of cases addition, for children whose first episode occurred at the end of
and 75% of controls (P = 0.3). After all episodes, no source was the study period, recurrence could not be reported. To the best of
found for 4 of the 24 RPM cases (17%), including 3 who died of our knowledge, this is the only study that exclusively addressed
the PM and for whom investigations were not possible. The last RPM. Because of the study design and age matching, we could
patient had a full search for ID and underwent magnetic resonance not compare the occurrence of PM at different ages between the 2
imaging twice, with images examined at least twice to search for groups. However, the median age of patients with SPM in the whole
CSF leakage, which was not identified. No ID was found in cases, ACTIV register was much lower than that of patients with RPM (1
whereas 13% of controls had ID known before their PM. In the vs. 3.9 years).1 The PCV coverage rate may appear low; however,
Copyright © 2019 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
The Pediatric Infectious Disease Journal • Volume 38, Number 9, September 2019 Recurrent Pneumococcal Meningitis
children were included from 2001 to 2015, and PCV vaccination 2. Jit M. The risk of sequelae due to pneumococcal meningitis in high-income
was introduced in 2006, with progressively increasing vaccination countries: a systematic review and meta-analysis. J Infect. 2010;61:114–
124.
coverage. In total, 25 PM episodes (23%) occurred before 2006.
3. Christie D, Viner RM, Knox K, et al. Long-term outcomes of pneumo-
Among children with PM occurring after 2006, 77% were vacci- coccal meningitis in childhood and adolescence. Eur J Pediatr. 2011;170:
nated; 97% after 2011. Although the ACTIV data collection was 997–1006.
prospective, some additional data were retrospectively collected, 4. Bingen E, Lévy C, De la Rocque F, et al; Groupe des pédiatres et microbi-
which may have led to information bias in some files. However, the ologistes de l’Observatoire national des méningites. [Pneumococcal men-
Observatory’s questionnaires were completed when the PM episode ingitis in France: age and medical risk factors in children]. Arch Pediatr.
occurred, which limited information bias. Data for clinical obser- 2005;12:1187–1189.
vations and pre- or post-PM blood tests were not always complete. 5. Levy C, Varon E, Bingen E, et al. Pneumococcal meningitis in children in
France: 832 cases from 2001 to 2007. Arch Pediatr. 2008;15(suppl 3):S111–
Regardless, we returned to the source records for all patients to S118.
obtain the greatest accuracy of the data collected. Only 39% of all
6. Gaschignard J, Levy C, Chrabieh M, et al. Invasive pneumococcal dis-
patients and 58% of those with RPM were screened for ID. This ease in children can reveal a primary immunodeficiency. Clin Infect Dis.
low rate may be explained by the lack of ID searches in the early 2014;59:244–251.
years of the study, because the relation between IPDs and ID was 7. Alsina L, Basteiro MG, de Paz HD, et al. Recurrent invasive pneumococcal
not well established at that time. The case–control design of the disease in children: underlying clinical conditions, and immunological and
study was chosen to limit bias and was the most appropriate for microbiological characteristics. PLoS One. 2015;10:e0118848.
such rare events. 8. Mason EO Jr, Wald ER, Tan TQ, et al. Recurrent systemic pneumococcal
In France, the implementation of PCV7 then PCV13 with disease in children. Pediatr Infect Dis J. 2007;26:480–484.
high vaccination coverage (>91% in 2011, http://invs.santepub- 9. King MD, Whitney CG, Parekh F, et al; Active Bacterial Core Surveillance
Team/Emerging Infections Program Network. Recurrent invasive pneu-
liquefrance.fr/) led to a 38% decrease in PM incidence (95% CI, mococcal disease: a population-based assessment. Clin Infect Dis.
−56.1 to −20.4) in children <15 years of age.32 After this period, 2003;37:1029–1036.
the PM incidence increased and was linked to a serotype replace- 10. Ku CL, Picard C, Jeurissen A, et al. IRAK4 and NEMO mutations in other-
ment by nonvaccine serotypes, especially the emerging serotype wise healthy children with recurrent invasive pneumococcal disease. J Med
24F, which started in 2015.32 In our study, among all 110 episodes Genet. 2007;44:16–23.
of PM, 33 (30%) cases were due to PCV serotypes, with no differ- 11. Orlicek SL, Herrod HG, Leggiadro RJ, et al. Repeated invasive pneumococ-
ences between cases and controls. After 2010, no PCV serotype cal infections in young children without apparent underlying immunodefi-
ciency. J Pediatr. 1997;130:284–288.
was found in patients receiving PCV13. During this period, PM
episodes were mostly due to serotype 24F in controls and serotypes 12. Ingels H, Lambertsen L, Harboe ZB, et al. Recurrent invasive pneumococcal
disease in children: epidemiological, microbiological and clinical aspects
15A and 15C in cases. Patients with underlying conditions, includ- from a Danish 33-year nationwide survey (1980–2013). Scand J Infect Dis.
ing CSF leakage, tend to be infected by non-PCV serotypes.33 2014;46:265–271.
With the data we present, it seems possible to propose a 13. Kahue CN, Sweeney AD, Carlson ML, et al. Vaccination recommenda-
hypothesis of a pathophysiologic mechanism of these RPM cases, tions and risk of meningitis following cochlear implantation. Curr Opin
different from SPM. An anatomic abnormality (CSF leakage) may Otolaryngol Head Neck Surg. 2014;22:359–366.
be the main factor in the development of RPM. Thus, an ENT ori- 14. Revest M, Michelet C. [Predisposing factors of community acquired
gin (otitis or other local infection) or the occurrence of a predispos- bacterial meningitis (excluding neonates)]. Med Mal Infect. 2009;39:
562–571.
ing winter viral infection may not be entirely necessary. The lower
15. Hénaff F, Levy C, Cohen R, et al. Risk factors of pneumococcal meningitis
level of positive blood cultures in RPM than SPM cases could mean in children older than 5 years old: data from a national network. Pediatr
that the CSF leakage would be sufficient to promote PM by loco- Infect Dis J. 2017;36:457–461.
regional diffusion with less bacteremia. 16. Østergaard C, Konradsen HB, Samuelsson S. Clinical presentation and
We emphasize the importance of searching for a predis- prognostic factors of Streptococcus pneumoniae meningitis according to the
posing factor (ie, ID or CSF leakage) in any episode of PM. CSF focus of infection. BMC Infect Dis. 2005;5:93.
leakage should be searched for especially when the PM occurs 17. Drummond DS, de Jong AL, Giannoni C, et al. Recurrent meningitis in the
outside the winter period and/or without any concomitant ENT pediatric patient–the otolaryngologist’s role. Int J Pediatr Otorhinolaryngol.
1999;48:199–208.
infection. Any recurrence of PM should always lead to an active
search for CSF leakage. A literature review proposed the follow- 18. Perrocheau A, Doyle A, Bernillon P, et al. Estimation du nombre total de
méningites à pneumocoque de l’enfant par la méthode capture-recapture à 3
ing recommendations34: β-transferrin test for clear rhinorrhea; sources, France, 2001–2002. Bull Epidemiol Hebd. 2006:2-3,16–19.
if positive, cerebral and facial mass computed tomography scan 19. Hegazy HM, Carrau RL, Snyderman CH, et al. Transnasal endoscopic
first and cerebral magnetic resonance imaging with cisternogra- repair of cerebrospinal fluid rhinorrhea: a meta-analysis. Laryngoscope.
phy second. Searching for and treating the source of infection is 2000;110:1166–1172.
imperative to limit the risk of neurologic consequences second- 20. White AN, Ng V, Spain CV, et al. Let the sun shine in: effects of ultra-
ary to the PM recurrence. violet radiation on invasive pneumococcal disease risk in Philadelphia,
Pennsylvania. BMC Infect Dis. 2009;9:196.
21. Talbot TR, Poehling KA, Hartert TV, et al. Seasonality of invasive pneumo-
ACKNOWLEDGMENTS coccal disease: temporal relation to documented influenza and respiratory
We are grateful to all the pediatricians and microbiologists syncytial viral circulation. Am J Med. 2005;118:285–291.
who collected the data via the National Observatory of Bacterial 22. Opatowski L, Varon E, Dupont C, et al. Assessing pneumococcal
Meningitides, as well as to the employees of Association Clinique meningitis association with viral respiratory infections and antibiot-
ics: insights from statistical and mathematical models. Proc Biol Sci.
et Thérapeutique Infantile du Val de Marne and the National Refer- 2013;280:20130519.
ence Center for pneumococci for their help. 23. Weinberger DM, Grant LR, Steiner CA, et al. Seasonal drivers of pneumo-
coccal disease incidence: impact of bacterial carriage and viral activity. Clin
REFERENCES Infect Dis. 2014;58:188–194.
1. Levy C, Varon E, Picard C, et al. Trends of pneumococcal meningitis in 24. Weinberger DM, Harboe ZB, Viboud C, et al. Pneumococcal disease sea-
children after introduction of the 13-valent pneumococcal conjugate vaccine sonality: incidence, severity and the role of influenza activity. Eur Respir J.
in France. Pediatr Infect Dis J. 2014;33:1216–1221. 2014;43:833–841.
Copyright © 2019 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Darmaun et al The Pediatric Infectious Disease Journal • Volume 38, Number 9, September 2019
25. Kajimoto M, Hasegawa S, Nomura S, et al. An 11-year-old girl with recur- 30. Hendricks BL, Shikary TA, Zimmer LA. Causes for 30-day readmis-
rent bacterial meningitis due to liquorrhea caused by bone malformation of sion following transsphenoidal surgery. Otolaryngol Head Neck Surg.
the skull base. Brain Dev. 2014;36:637–640. 2016;154:359–365.
31. Soudry E, Turner JH, Nayak JV, et al. Endoscopic reconstruction of sur-
26. Ben-Shoshan M, DeRowe A, Grisaru-Soen G, et al. Recurrent meningitis gically created skull base defects: a systematic review. Otolaryngol Head
and cerebrospinal fluid leak-two sides of the same vestibulocochlear defect: Neck Surg. 2014;150:730–738.
report of three cases. Eur J Pediatr. 2007;166:269–272. 32. Ouldali N, Levy C, Varon E, et al; French Pediatric Meningitis Network.
27. Ramesh V, Sankar J, Krishnamurthy B. Recurrent pneumococcal meningitis Incidence of paediatric pneumococcal meningitis and emergence of new
serotypes: a time-series analysis of a 16-year French national survey. Lancet
in a child with congenital common cavity malformation of the inner ear. Infect Dis. 2018;18:983–991.
Clin Pediatr (Phila). 2014;53:397–398.
33. Cohen R, Varon E, Béchet S, et al. Comparative impact of pneumococcal
28. Motojima T, Fujii K, Ishiwada N, et al. Recurrent meningitis associated with conjugate vaccines on pneumococcal meningitis according to underlying
a petrous apex cephalocele. J Child Neurol. 2005;20:168–170. conditions. Vaccine. 2016;34:4850–4856.
34. Muzzi E, Battelino S, Gregori M, et al. Life-threatening unilateral hear-
29. Hosainey SA, Lassen B, Helseth E, et al. Cerebrospinal fluid disturbances ing impairments. Review of the literature on the association between
after 381 consecutive craniotomies for intracranial tumors in pediatric inner ear malformations and meningitis. Int J Pediatr Otorhinolaryngol.
patients. J Neurosurg Pediatr. 2014;14:604–614. 2015;79:1969–1974.
Copyright © 2019 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.