Bacterial meningitis in a quaternary NICU: A multiyear retrospective study

Abstract

Bacterial meningitis causes significant morbidity and mortality in infants. Lumbar punctures are often deferred until the results of blood cultures are known and sometimes not considered, making this population susceptible to a missed diagnosis. There are few studies describing the epidemiology of neonatal meningitis in quaternary neonatal intensive care unit settings. We describe the epidemiology of meningitis in a level IV neonatal intensive care unit; compare pathogens and rates of concordant bacteremia between infants with and without neurosurgical (NS) devices. Retrospective review of infants < 1 year of age in the Children’s Hospital of Philadelphia neonatal intensive care unit with bacterial meningitis (June 2007–October 2021). Analysis included summary statistics, Wilcoxon rank sum, Chi square, and Fisher exact tests. We identified 101 episodes of bacterial meningitis (95 infants). 9 infants died. At diagnosis, 26 infants (27%) had NS devices. Group B streptococcus (GBS) and Escherichia coli (E coli) were most common pathogens, however, coagulase-negative staphylococci and Staphylococcus aureus (S aureus) predominated among infants with NS devices. While 86% had positive blood cultures in the absence of a NS device, only 14% of episodes with NS devices had concomitant bacteremia (P < .0001). Although Group B streptococcus and E coli remain most prevalent overall, coagulase-negative staphylococci and S aureus were common pathogens in NS patients. Infants with NS devices rarely had concomitant bacteremia. Meningitis was diagnosed in the absence of a positive blood culture in 36% of episodes, underscoring the importance of developing guidance for lumbar punctures in infants evaluated for sepsis.

1. Introduction

Meningitis is a life-threatening disease, affecting 0.21 infants per 1000 live births, with a higher incidence in preterm and chronically hospitalized infants.^[1–4] Approximately 8% of affected infants die, and 20% to 50% of survivors experience sequelae such as seizures, and motor and cognitive delays.^[5–8] Populations at the highest risk for meningitis include preterm infants and those with a history of neurosurgical (NS) interventions and devices.^[9] Despite declines in mortality, morbidity has not improved since the 1970s.

While prior studies on neonatal meningitis have examined rates and characteristics in preterm infants, there is a paucity of studies that have examined neonatal meningitis in the quaternary neonatal intensive care unit (NICU) setting.^[10,11] Quaternary level care NICUs possess a high-risk cohort of infants, comprising preterm infants with complications necessitating transfer for a higher level of care, infants with congenital anomalies, and those needing surgical interventions (including NS devices and procedures, among others).^[12]

Challenges exist in describing the epidemiology of meningitis in infants. Culture of cerebrospinal fluid (CSF) is the traditional gold standard for diagnosis. Confirming a meningitis diagnosis may be difficult because a lumbar puncture (LP) is often deferred until results of blood cultures (BCs) are known and sometimes not considered, making this population susceptible to a missed diagnosis.^[10,13] Moreover, studies have demonstrated a lack of concordance between blood and CSF cultures, suggesting that meningitis may be overlooked if an LP is not considered independent of positive BC results.^[10,11,14] Infants with NS devices are at particularly high risk of infectious complications, although literature regarding this group has been sparse.^[15] As the pathogenesis of these infections is likely related to colonized or infected central nervous system hardware, bacterial etiology and selection of empiric antimicrobial regimens may differ among these infants.

We describe a multiyear retrospective cohort of infants with bacterial meningitis in a quaternary level care NICU to identify risk factors, causative organisms, short term outcomes, and rates of concordant bacteremia between infants with and without NS devices.

2. Materials and methods

2.1. Study population

As part of a retrospective case series, we reviewed medical records of infants <1 year of age in the Level IV NICU at the Children’s Hospital of Philadelphia (CHOP) from June 1, 2007 to October 31, 2021 who were diagnosed with bacterial meningitis. The CHOP NICU is a 102-bed quaternary referral unit primarily treating outborn infants with complex medical conditions. The patient population also includes inborn infants with surgical and other anomalies requiring subspecialty care who are delivered in the Special Delivery Unit at CHOP. All patients hospitalized in the NICU at CHOP were eligible for inclusion. The Institutional Review Board at CHOP approved this study under a waiver of informed consent.

2.2. Eligibility criteria

The study population included patients whose cultures were obtained in the CHOP NICU as well as cases diagnosed at an outside hospital and then transferred to CHOP for further treatment. We included cases with confirmed bacterial meningitis defined by bacterial detection in CSF culture, who received a complete course of antibiotic therapy. CSF cultures positive for coagulase-negative staphylococci (CoNS) were included if they were regarded as true infection and received a full course of antimicrobial therapy. These decisions were made by the neonatologist caring for the infant in consultation with the pediatric infectious diseases team, based on the clinical context (presence or absence of NS device as well as clinical signs and symptoms and CSF parameters consistent with meningitis). Cases in which bacteria detected in CSF were regarded as contaminants and those with viral meningitis were excluded.

2.3. Study definitions

Bacterial meningitis was defined as the isolation of an organism from CSF culture and treatment for at least 7 days or death before completion of therapy. Recurrent meningitis was defined by bacterial growth in the CSF after an antibiotic-free period of at least 10 days following the first episode. Blood and CSF cultures were considered concordant if the same organism was identified in BCs within 5 days of the CSF culture. We defined a subgroup of infants with NS interventions and device placement prior to meningitis diagnosis (ventriculoperitoneal/atrial shunts, Ommaya reservoirs, external ventricular drains).

Infants were diagnosed with early onset meningitis at <7 days of age while late onset meningitis was diagnosed after 7 days of age. Antibiotic duration was defined as the total duration of parenteral antibiotic therapy. Patients whose antibiotic courses were truncated due to death were omitted from antibiotic duration calculations. NICU length of stay was defined as the total number of days an infant was in the NICU prior to discharge, death, or transfer to another unit or facility. Mortality was defined as any death prior to discharge from the NICU. Signs and symptoms of meningitis were obtained from manual chart review of clinician notes.

2.4. Data collection and statistical analysis

Data was retrieved from electronic medical records using a combination of automated extraction supplemented and validated by manual review of physician and outside hospital notes. Variables included in this analysis included patient demographics, laboratory data, medication administration records, NS interventions and devices, NICU length of stay, and mortality. Data regarding signs and symptoms and CSF parameters could only be accurately extracted for infants diagnosed while inpatients at CHOP, as these variables were difficult to retrieve from outside hospital records. CSF parameter values above the upper limit of reporting and below the lower limit of reporting were transcribed with the values of the upper and lower limits respectively. Descriptive data were calculated as frequency counts and percentages for categorical variables and median and interquartile range for continuous variables. Significance was determined by Wilcoxon rank sum test, Chi square and Fisher exact tests. All statistical tests were two-tailed, and differences were considered significant when the P-value was <.05. Analyses were performed using STATA version 17. All data were handled in accordance with CHOP Institutional Review Board guidelines.

3. Results

3.1. Patient population

Between June 1, 2007 and October 31, 2021, we identified 101 episodes of bacterial meningitis in 95 infants <1 year of age. Demographic characteristics of the cohort are shown in Table 1. 36 infants were diagnosed with meningitis at an outside hospital and referred to CHOP for treatment, while the remainder had cultures obtained in the CHOP NICU. 61% were preterm with a median gestational age of 35 weeks and a median postnatal age of 30 days. 6 infants had recurrent meningitis and 1 had polymicrobial meningitis. There were 12 episodes of early onset meningitis in the cohort (4 cases of group B streptococcus [GBS], 6 cases of Escherichia coli (E coli), and 1 each of Streptococcus pneumoniae, and Bacillus spp.) while the remainder were late onset disease.

Table 1.

Demographics of the cohort.

Variable	Total cohort (101 episodes, 95 patients)	No NS device (71 episodes, 69 patients)	NS device (30 episodes, 26 patients)
Patient level data (n = 95)
Birthweight, grams (median, 25th–75th)	2250 [1108–3139]	2570 [1400–3185]	1270 [943–2761]
Gestational age, weeks (median, 25th–75th)	35 [28–39]	36 [30–39]	28 [26–35]
Male (n, %)	59 (62%)	42 (61%)	17 (65%)
Female (n, %)	36 (38%)	27 (39%)	9 (35%)
White (n, %)	38 (40%)	32 (46%)	6 (23%)
Black or African American (n, %)	27 (28%)	17 (25%)	10 (38%)
Multiracial (n, %)	5 (5%)	5 (7%)	0 (0%)
Asian (n, %)	5 (5%)	5 (7%)	0 (0%)
Other (n, %)	20 (21%)	10 (14%)	10 (38%)
Mortality (n, %)	9 (9%)	7 (10%)	2 (8%)
Episode level data (n = 101)
Age at time of infection, days (median, 25th–75th)	30 [14–56]	19 [9–38]	68 [41–101]
Concurrent positive blood culture (n, %)	63 (64%)	59 (86%)	4 (14%)
Antibiotic duration, days (median, 25th–75th)	21 [15–28]	21 [15–29]	18 [14–25]
NICU length of stay, days (median, 25th–75th)	27 [11–74]	19 [18–49]	67 [34–149]

NICU = neonatal intensive care unit, NS = neurosurgical.

Twenty-six infants (30 episodes) had a NS device at the time of meningitis diagnosis. Infants with NS devices were more likely to be preterm and were likely to be older at the time of onset of meningitis. Infants with NS devices also had longer NICU length of stay, but similar antibiotic duration and mortality (Table 1). Overall, 9% of infants who experienced meningitis died before hospital discharge.

3.2. Bacterial pathogens

Overall, 65% of episodes were attributed to Gram-positive organisms. The most common pathogens were GBS (29%) and E coli (20%) (Table 2). In contrast, among infants with NS devices, CoNS (37%) and Staphylococcus aureus (S aureus) (30%) were the most frequent pathogens. Among the entire cohort, 63/101 episodes (64%) had both positive CSF and BCs (3 had unknown BC results). 59 of 71 episodes (86%) had positive BCs in the absence of a NS device, however only 14% of episodes in infants with NS devices had concomitant bacteremia (4/29 [with 1 unknown BC]; P < .0001). Episodes of GBS and E coli were frequently associated with positive BCs (90% and 79% episodes, respectively), while CoNS and Enterococcus rarely had concomitant bacteremia (15%, 40%). Similar to recent literature regarding early onset sepsis in neonates, E coli meningitis most commonly occurred in preterm infants, while GBS was commonly a disease of term infants (Table S1, Supplemental Digital Content, http://links.lww.com/MD/O219).^[16]

Table 2.

Pathogens isolated in CSF.

Pathogens	All episodes (101 episodes, 102 pathogens)	No NS device (71 episodes, 71 pathogens)	NS device (30 episodes, 31 pathogens)
Gram negative	35 (35%)	29 (41%)	6 (20%)
E coli	20 (20%)	18 (25%)	2 (7%)
Enterobacter spp.	3 (3%)	3 (4%)	0 (0%)
Citrobacter spp.	3 (3%)	3 (4%)	0 (0%)
Pasteurella spp.	2 (2%)	2 (3%)	0 (0%)
Pseudomonas spp.	2 (2%)	0 (0%)	2 (7%)
Klebsiella spp.	2 (2%)	1 (1%)	1 (3%)
Serratia spp.	1 (1%)	0 (0%)	1 (3%)
Salmonella spp.	1 (1%)	1 (1%)	0 (0%)
Ureaplasma spp.	1 (1%)	1 (1%)	0 (0%)
Gram positive	66 (65%)	42 (59%)	24 (80%)
GBS	29 (29%)	28 (39%)	1 (3%)
CoNS	13 (13%)	2 (3%)	11 (37%)
S aureus	12 (11%)	3 (4%)	9 (30%)
Enterococcus spp.	4 (4%)	2 (3%)	2 (7%)
S pneumoniae	3 (3%)	3 (4%)	0 (0%)
Listeria spp.	1 (1%)	1 (1%)	0 (0%)
Paenibacillus spp.	1 (1%)	1 (1%)	0 (0%)
Bacillus spp.	1 (1%)	1 (1%)	0 (0%)
S gallolyticus	1 (1%)	1 (1%)	0 (0%)
Polymicrobial (Enterococcus spp. + S epidermidis)	1 (1%)	0 (0%)	1 (3%)

CoNS = coagulase negative staphylococcus, CSF = cerebrospinal fluid, E coli = Escherichia coli, GBS = group B streptococcus, NS = neurosurgical, S aureus = Staphylococcus aureus, S gallolyticus = Streptococcus gallolyticus, S pneumoniae = Streptococcus pneumoniae, spp = species.

In all cases, initial antibiotic coverage was comprised of a third or fourth generation cephalosporin, most commonly cefepime or cefotaxime, along with vancomycin. Once bacterial speciation and antibiotic susceptibility information was available, further antibiotic choice was determined based on those data. 1 patient with a multidrug resistant organism and very long course of antibiotics (102 days) was included in our dataset.

In our cohort, 9 infants did not survive. 3 cases were associated with GBS, 2 with S aureus (1 Methicillin-sensitive S aureus, 1 Methicillin-resistant S aureus) and 1 each with E coli, Enterococcus, Salmonella, and CoNS. 6 infants had recurrent meningitis (Table S2, Supplemental Digital Content, http://links.lww.com/MD/O219). Among them, E coli was associated with 2 cases, 1 of whom died. The remaining cases of recurrent meningitis were caused by Pseudomonas, CoNS, GBS, and S aureus, although pathogens isolated in the first and second episodes were often not identical.

Among the infants with CoNS meningitis (12 isolates of Staphylococcus epidermidis, 1 S warneri, and 1 S haemolyticus), 12 of 14 had a NS device at the time of diagnosis. The 2 infants without NS devices had concomitant bacteremia and were adjudicated as pathogenic in conjunction with pediatric infectious diseases consultants. All infants in this group received a full course of antibiotic therapy.

3.3. Signs and symptoms of bacterial meningitis

Fever was the most common presenting symptom overall (Table 3). Nonspecific symptoms such as respiratory distress, irritability, tachycardia, and feeding intolerance were also frequently reported. Feeding intolerance and lethargy were significantly more common in infants without NS devices (P .003 and P .002 respectively). Neurological symptoms such as irritability, seizures, lethargy, and bulging fontanel manifested more commonly in term infants compared to preterm infants, though sample sizes were too small to determine statistical significance (data not shown).

Table 3.

Signs and symptoms of meningitis.

Symptom (%)	All episodes (n = 63)*	No NS device (n = 37)	NS device (n = 26)	P value
Fever	38 (60%)	24 (65%)	14 (54%)	.38
Respiratory distress	34 (54%)	18 (49%)	16 (62%)	.31
Irritability	26 (41%)	18 (49%)	8 (31%)	.16
Tachycardia	25 (40%)	16 (43%)	9 (35%)	.49
Feeding intolerance	21 (33%)	18 (49%)	3 (12%)	.003†
Lethargy	18 (29%)	16 (43%)	2 (8%)	.002†
Seizures	17 (27%)	11 (30%)	6 (23%)	.56
Apnea	17 (27%)	10 (27%)	7 (27%)	.99
Bulging fontanel	8 (13%)	6 (16%)	2 (8%)	.45†
Hypothermia	5 (8%)	4 (11%)	1 (4%)	.39†
Poor perfusion	4 (6%)	3 (8%)	1 (4%)	.64†

NS = neurosurgical.

^*Only patients diagnosed with meningitis at CHOP were included (n = 63).

^†Fisher exact test used when cell values <5.

3.4. CSF parameters

Median white blood cell count (WBC) in the infants with meningitis was 281 cells/mm (range 1–30,750/mm³), with protein 295 mg/dL (range 19–3000 mg/dL) and glucose 22 mg/dL (range 20–102 mg/dL). There were no significant differences in CSF parameters between infants with and without NS devices (Table 4). Among the infants in the non-NS device group, 7 had a WBC ≤ 30/mm³ while 6 infants in the NS group had normal CSF WBC counts (Table S3, Supplemental Digital Content, http://links.lww.com/MD/O219).

Table 4.

CSF parameters.

CSF parameter, median (range)	No NS device (38 episodes)	NS device (25 episodes)	P value
WBC (cells/mm3)	292 (3–19,560)	263 (1–30,750)	.46
Protein (mg/dL)	365 (52–1763)	295 (19–3000)	.68
Glucose (mg/dL)	27 (20–102)	20 (20–63)	.21

CSF = cerebrospinal fluid, NS = neurosurgical, WBC = white blood cells.

4. Discussion

Bacterial meningitis, though infrequent, is associated with significant impact to infant outcomes. In this retrospective case series, we described the epidemiology of bacterial meningitis over a >10-year period in a quaternary NICU. While GBS and E coli remain the most prevalent pathogens overall, CoNS and S aureus are common pathogens in patients with NS devices. Meningitis was diagnosed in the absence of a positive BC in 36% of infants, highlighting the importance of considering an LP in infants evaluated for infection. This was especially true for infants with NS devices, who rarely had concomitant positive BCs.

Studies of meningitis in infants have mostly focused on preterm populations, and have demonstrated marked variation in decision making around when to perform LPs.^[1,11] Wiswell and colleagues demonstrated that selective approaches to performing LPs could result in missed diagnoses of meningitis.^[17] In a large cohort of very low birthweight infants from 15 NICUs, Stoll et al demonstrated that up to one-third of preterm infants with meningitis had an absence of concordant positive blood cultures.^[10] The Australasian study group demonstrated that 62% of infants who died from early onset sepsis or suspected infection did not have an LP performed.^[18] Similarly, a very recent study by Brumbaugh et al noted the decreasing frequency of obtaining LPs during sepsis evaluations as well as the absence of positive BCs in 16% of low birthweight infants with meningitis.^[14] However, there is a paucity of literature on infants admitted to quaternary NICUs, including those experiencing NS interventions.

In this study, we outline the clinical courses of infants with meningitis admitted to a Level IV NICU and describe the pathogens implicated. In agreement with other studies, the majority of infants in our study had late onset meningitis which may reflect the decreasing incidence of early onset infections in the era of adherence to intrapartum prophylaxis guidelines.^[11,16] The predominant pathogens also correlated with those generally reported in the literature, as GBS and E coli were most frequently implicated.^[11,19] Of note, more than a quarter of the episodes in our study occurred in patients with NS devices, highlighting the high risk conferred on these patients by these interventions. S aureus and CoNS predominated in this group which is consistent with prior adult and pediatric literature despite the paucity of large, multicenter prospective studies.^[15,20] Our data further support the selection of empiric antimicrobial regimens based on the presumed pathogenesis of infection and the pathogens most likely to be associated with each mode of pathogenesis (metastatic spread to meninges following bacteremia vs ventriculitis or shunt infection).^[19]

Importantly, we found a high rate of discordance between meningitis and bacteremia, which was accentuated in the NS group. For the NS group, pathogenesis of meningitis is likely related to colonization of devices, prior surgical sites, followed by translocation of skin flora and subsequent infection; the lack of concordant bacteremia is therefore plausible, as the bloodstream is less likely to be the source of infection.^[20] In non-NS patients, it is assumed that most cases of meningitis occur via bacteremic spread to the meninges. However, there were several episodes in this group of infants where meningitis occurred in the absence of concurrent bacteremia raising the possibility of primary central nervous system infection.^[14] These findings highlight the importance of timely performance of LP in order to accurately diagnose meningitis.

No single sign or symptom or combination of signs and symptoms was universally present, highlighting the lack of a specific presentation of meningitis in infants. Consistent with prior literature, the signs of bacterial meningitis in infants are essentially indistinguishable from those of bacterial sepsis rather than specific to meningeal inflammation.^[2] However, we can also reasonably suggest that it is prudent to consider an LP early when an infant possesses a NS device or is presenting with symptoms such as seizures, irritability, lethargy, a bulging fontanel, or fever in the first 1 to 2 months of life.

Despite their differing pathogenesis, CSF parameters were not significantly different in infants with and without NS devices. In both groups, the range of WBCs included several cases with normal CSF cell counts (WBC < 30/mm³). Our data are supported by previous work by Garges et al in which CSF parameters were unreliable to exclude the diagnosis of bacterial meningitis.^[21] Similar to our findings, several studies noted considerable overlap between normal and abnormal values for CSF parameters in both term and preterm infants, reinforcing the value of CSF culture prior to antibiotic administration.^[19,22,23] Interestingly, a recent study of bacterial meningitis from our group noted similar metabolic profiles in infants with and without NS devices despite differing infecting pathogens.^[24]

Our study has several strengths. It is one of few studies to closely examine a complex population of infants and draw comparisons between infants with and without NS devices. Although it is commonly recognized that infants with NS devices are at higher risk of infection, this group has not been intensively studied in the past. This study is strengthened by the inclusion of a large cohort of hospitalized infants with rich information regarding meningitis risks, organisms, and concordance with BC results. We used robust electronic medical records data to present data from a >10-year period. Our limitations include data derived from a single center such that results may not be generalizable beyond quaternary level care NICUs, or NICUs with different demographics and geographical locations. Our population encompassed infants who received LPs and were hospitalized in the CHOP NICU; as such we do not have data on the frequency with which CSF is obtained in infants evaluated for sepsis. Further, we could only extract complete data for episodes of meningitis occurring during admission to the CHOP NICU. We also acknowledge that there may be intrinsic inaccuracies in the retrospective design of our study, although cases were included based on review of clinical context and review of consultation with the infectious diseases team.

Our study highlights that meningitis, while infrequent, may cause significant morbidity and mortality in NICU patients. A high index of suspicion in patients with NS devices or concerning symptoms, coupled with an early LP, is likely important for accurate and timely diagnosis and treatment. Our study uncovers variation in decision-making around LPs in quaternary NICUs with complex case mix. Additional studies are needed to explore whether guidelines for consideration of LPs during evaluations for infection could help decrease clinical practice variation. Future studies should also closely examine brain imaging and neurologic outcomes for these patients to more fully understand the long term impact of meningitis.

Author contributions

Conceptualization: Olivia O’Hearn, Lakshmi Srinivasan, Mary Catherine Harris.

Data curation: Olivia O’Hearn, Lakshmi Srinivasan, Matthew Devine, Mary Catherine Harris.

Formal analysis: Olivia O’Hearn, Lakshmi Srinivasan, Mary Catherine Harris.

Investigation: Olivia O’Hearn, Lakshmi Srinivasan, Matthew Devine, Mary Catherine Harris.

Methodology: Olivia O’Hearn, Lakshmi Srinivasan, Mary Catherine Harris.

Project administration: Lakshmi Srinivasan, Mary Catherine Harris.

Software: Matthew Devine.

Supervision: Mary Catherine Harris.

Writing – original draft: Olivia O’Hearn, Lakshmi Srinivasan, Mary Catherine Harris.

Writing – review & editing: Olivia O’Hearn, Lakshmi Srinivasan, Matthew Devine, Mary Catherine Harris.