Video Abstract

Video Abstract

Close modal
BACKGROUND:

New biomarkers like procalcitonin and C-reactive protein may help design an accurate decision support tool used to identify children with pleocytosis at low or high risk of bacterial meningitis. Our objective was to develop and validate a score (that we call the meningitis score for emergencies [MSE]) to distinguish bacterial meningitis from aseptic meningitis in children with pleocytosis when initially evaluated at the emergency department.

METHODS:

We included children between 29 days and 14 years old with meningitis admitted to 25 Spanish emergency departments. A retrospective cohort from between 2011 and 2016 was used as the derivation set and a prospective cohort recruited during 2017 and 2018 was used as the validation set.

RESULTS:

Among the 1009 patients included, there were 917 cases of aseptic meningitis and 92 of bacterial meningitis. Using multivariable logistic regression analysis, we identified the following predictors of bacterial meningitis from the derivation set: procalcitonin >1.2 ng/mL, cerebrospinal fluid (CSF) protein >80 mg/dL, CSF absolute neutrophil count >1000 cells per mm3, and C-reactive protein >40 mg/L. Using the derivation set, we developed the MSE, assigning 3 points for procalcitonin, 2 points for CSF protein, and 1 point for each of the other variables. An MSE ≥1 predicted bacterial meningitis with a sensitivity of 100% (95% confidence interval [CI]: 95.0%–100%), a specificity of 83.2 (95% CI: 80.6–85.5), and a negative predictive value of 100% (95% CI 99.4–100.)

CONCLUSIONS:

The MSE accurately distinguishes bacterial from aseptic meningitis in children with CSF pleocytosis.

What’s Known on This Subject:

No single variable distinguishes bacterial from aseptic meningitis in children with cerebrospinal fluid pleocytosis. Combinations of several variables have been used to distinguish these two types of meningitis.

What This Study Adds:

The meningitis score for emergencies can be used to guide initial clinical decision-making in children with cerebrospinal fluid pleocytosis, without misclassifying children with bacterial meningitis. Including procalcitonin and C-reactive protein achieves a more accurate decision support tool.

Since the widespread introduction of conjugate vaccines, most cases of pediatric meningitis are aseptic, and viruses remain the most common cause.13  Children with viral meningitis, a self-limiting condition, require only supportive care. In contrast, although uncommon,4  bacterial meningitis carries high mortality and morbidity rates, underlining the importance of prompt and appropriate treatment. Classically, because it is often difficult to distinguish between bacterial and aseptic meningitis, when children are evaluated in the emergency department (ED), those with cerebrospinal fluid (CSF) pleocytosis are admitted to the hospital to receive antibiotics pending bacterial culture results.5 

No single variable distinguishes bacterial from aseptic meningitis, and overlaps found in values of variables between patients with aseptic and bacterial meningitis limit their discriminative ability when applied as univariate predictors. Thus, combinations of several variables have been used to distinguish these two types of meningitis. Among the clinical prediction rules, only the bacterial meningitis score (BMS)5  was found to accurately identify children with CSF pleocytosis at low or high risk of bacterial meningitis. This score may be helpful to guide clinical decision-making for the management of children presenting to EDs with CSF pleocytosis,6  and it is easy to apply.7  It was developed after the widespread introduction of a highly effective bacterial vaccine against Haemophilus influenzae type b and validated in a postpneumococcal conjugated vaccine cohort.6  Nevertheless, a few cases of bacterial meningitis may be missed with use of the BMS.8,9 

In recent decades, C-reactive protein (CRP) and procalcitonin have been added to screening tests in febrile children. Globally, CRP and procalcitonin levels are the parameters providing the most diagnostic value in feverish children used to detect those with serious bacterial infection.10  Procalcitonin offers some advantages to identify patients with invasive bacterial infections,1114  specifically meningococcal diseases,15  including meningitis.11,16  In a previous series, the replacement of the peripheral absolute neutrophil count (ANC) with procalcitonin significantly increased the specificity of the BMS.17 

Our main objective of this study was to develop and validate a score, which we have called the meningitis score for emergencies (MSE), to distinguish bacterial meningitis from aseptic meningitis in children with pleocytosis when initially evaluated at the ED. Our secondary objective was to compare the performance of the new decision support tool to the BMS.

Our hypothesis was that including procalcitonin and CRP would achieve a more accurate decision support tool for distinguishing bacterial from aseptic meningitis in children with CSF pleocytosis.

This was a cohort study of children between 29 days and 14 years old diagnosed with meningitis in 25 pediatric EDs that are members of the Infectious Diseases Working Group of the Spanish Society of Pediatric Emergencies.18,19 

To be included, the children were required to have CSF pleocytosis and data on all the following: blood and CSF bacterial cultures, white blood cell (WBC) count, serum CRP, and procalcitonin.

We excluded the following children: those who were <29 days old, critically ill, with purpura, and not previously healthy and/or treated with antibiotics within 72 hours before the lumbar puncture.

To create the derivation set, the patients were retrospectively included from December 31, 2016, backward to a maximum of 6 years (until January 2011), depending on when procalcitonin testing was introduced in each hospital. Regardless of the date when the procalcitonin levels started to be measured, the study period for each hospital was required to be a multiple of 12 months to avoid possible bias due to seasonal variations. In the retrospective phase of the study, patients were identified by using the electronic records of the hospitals included in the study. Data on patients and episodes were obtained from the electronic health records of the pediatric EDs and health system. Researchers were asked to review all episodes corresponding to children diagnosed with meningitis following the coding system of the Spanish Society of Pediatric Emergencies (based on International Classification of Diseases, Ninth Revision). Cases of children diagnosed with meningitis and found to have CSF pleocytosis were reviewed.

The validation set was created by prospectively including patients in 2017 and 2018. In the prospective phase, patients were identified by the physician responsible who completed the questionnaires after ED discharge for patients who were discharged from the ED and after hospital discharge for patients who were admitted to the hospital to obtain complete patient information and both ED and hospital outcomes.

For all the patients included, data for following variables were collected and entered onto structured data sheets: date of birth, sex, date of admission and discharge, previous and current medical history of conditions including seizures, duration and peak of fever at admission, antibiotic pretreatment, peripheral WBC count and ANC, CSF WBC count and ANC, CSF red blood cell count, CSF glucose, CSF protein, Gram-stain if obtained, blood and CSF culture results, results of polymerase chain reaction (PCR) tests (both bacterial and viral) if performed, and disposition on discharge. The laboratory values included were those obtained when the child arrived at the ED.

Specific electronic questionnaires were completed via Google Drive for all children included. Questionnaires, in addition to a study manual, were distributed to site investigators (ED physicians) before the initiation of the study to check the text understandability and appropriateness for data collection to ensure clarity of the final data collected. The completed questionnaires were then sent electronically to the principal investigator (S.M.).

Bacterial Meningitis

Bacterial meningitis was defined as patients with either one of the following two criteria: (1) identification of bacterial pathogen in CSF by growth in bacterial culture and/or, for Neisseria meningitidis or Streptococcus pneumoniae, genomic detection in CSF using the PCR technique (RealCycler MENE and RealCycler MENELI; Progenie Molecular, Valencia, Spain); or (2) the presence of CSF pleocytosis (≥10 WBCs per mm3) and either a positive blood culture result and/or, for N meningitidis or S pneumoniae, genomic detection in blood using the PCR technique. Certain bacterial species (including Staphylococcus epidermidis, Propionibacterium acnes, Streptococcus viridans, Corynebacterium spp., and other diphtheroids) isolated in otherwise healthy patients were considered to be contaminants.

Aseptic Meningitis

Aseptic meningitis included patients with CSF pleocytosis and negative CSF and blood bacterial cultures as well as negative genomic detection of N meningitidis and S pneumoniae using the PCR technique, if performed.

Pleocytosis

Pleocytosis was defined as having CSF WBCs ≥10 cells per μL, corrected for the presence of CSF red blood cells by using a 1:500 ratio of leukocytes to erythrocytes usually found in peripheral blood.20  We also corrected the CSF protein (for every 1000-cell increase in CSF red blood cells per mm3, we considered that CSF protein increased by 1.1 mg/dL).21 

Well-Appearing

We defined well-appearing children as those who had a normal pediatric assessment triangle in the case of EDs in which these data are systematically recorded and, otherwise, if the findings of the physical examination documented in the patient medical record indicated no clinical suspicion of sepsis. Descriptors that led to exclusion included but were not limited to “poor/bad general appearance,” “irritable,” “cyanosis,” “hypotonic,” and “cutis marmorata.”

Critically Ill Children

Critically ill children were those with severe mental disturbance, evidence of cerebral herniation, or need for respiratory or hemodynamic support.

Previously Healthy Children

Previously healthy children were defined as those without any of the following risk factors: (1) immunosuppression (associated with cancer, chronic renal failure, sickle cell disease, or being a transplant recipient); (2) the presence of a mechanical device (indwelling catheter, ventriculoperitoneal shunt, auditory prostheses); (3) history of an invasive diagnostic or therapeutic procedure in the previous 10 days; or (4) CSF fistula. In the derivation set, this was determined after reviewing the entire chart of the patient, not only ED documentation. In the validation set, after reviewing the entire chart and asking to the parents.

We included the patients identified retrospectively in a derivation set. Subsequently, the validation set comprised the patients recruited prospectively.

Derivation Set

We conducted a receiver operating characteristic (ROC) curve analysis including the following variables: CSF ANC, CSF WBC count, CSF protein, CSF glucose, serum CRP, serum procalcitonin, serum WBC count, and serum ANC. Those revealing an area under the receiver operating characteristic curve (AUC) higher than 0.90 were selected for the score. We used the Youden index to identify the optimal cutoff points for these variables. Lastly, variables independently associated with bacterial meningitis were ranked according to the magnitude of the β-coefficient.

The performance of the score was then tested in the validation set. Specifically, the clinical decision rule derived from the derivation set was applied to the validation cohort. We used the AUC as a measure of the discriminatory performance. To establish the diagnostic accuracy of the score, sensitivity, negative predictive value (NPV), and negative likelihood ratio were determined.

Comparison With the BMS

We compared sensitivity, specificity, positive predictive value, and NPV of the two scores and the number of cases of bacterial meningitis missed including all the patients in the validation set in which the Gram-stain was performed. We also compared the AUC of the new score and BMS.

All the analyses were conducted by using SPSS version 23 (IBM SPSS Statistics, IBM Corporation).

Derivation Set

In a multicenter study including children with meningitis conducted by the Research Network of the Spanish Society of Pediatric Emergencies,17  the prevalence of bacterial meningitis was 6.2%. To achieve an accuracy of 5.0% in the estimation of percentages with a normal 95% bilateral asymptotic confidence interval (CI), we needed to include 90 patients in the study; hence, assuming a 10% drop-out rate, we needed to recruit 100 patients diagnosed with bacterial meningitis.

In the aforementioned study,17  we diagnosed 1 child >2 years of age with bacterial meningitis per 20 000 ED episodes, and 1 patient per 51 000 ED episodes would have met the inclusion criteria for the study. On the basis of these data, we would have required ∼5 100 000 ED episodes. Considering that in the current study, we also included infants between 2 months and 2 years old, the number of ED episodes was expected be somewhat lower.

Validation Set

We estimated that it would be sufficient to include 30 children diagnosed with bacterial meningitis in a 2-year period.

Role of the Funding Source

We received no funding.

We obtained overall approval from the Clinical Research Ethics Committee of Basque Country. Informed consent was required for participants in the prospective phase of the study.

Globally, we registered 5 167 945 ED presentations corresponding to children <14 years old in 25 pediatric EDs. Among these, 1509 patients aged between 29 days and 14 years old had pleocytosis and were diagnosed with meningitis (1341 cases of aseptic meningitis and 168 of bacterial meningitis), but 488 were excluded (412 with aseptic meningitis and 76 bacterial meningitis). Hence, finally, we included 1009 patients between 29 days and 14 years old (917 with aseptic meningitis and 92 bacterial meningitis). Bacterial meningitis was caused by the following pathogens: N meningitidis in 38 cases (41.3%), S pneumoniae in 35 (38.5%), group B Streptococcus in 5 (5.5%), Streptococcus pyogenes in 4 (4.3%), Enterococcus faecalis in 2 (2.2%), H influenzae in 2 (2.2%), Escherichia coli in 1 (1.1%), Listeria monocytogenes in 1 (1.1%), Salmonella typhimurium in 1 (1.1%), Streptococcus bovis in 1 (1.1%), Kingella kingae in 1 (1.1%), and Fusobacterium necrophorum in 1 (1.1%). Of these patients included, 819 (758 aseptic meningitis and 61 bacterial meningitis) were in the derivation set and 190 (159 aseptic meningitis and 31 bacterial meningitis) were in the validation set (Supplemental Fig 2).

The main characteristics of the patients included are shown in Table 1.

TABLE 1

Characteristics of the Patients Included in the Derivation and Validation Sets

VariableDerivation Set (n = 819)Validation Set (n = 190)
Female, n (%) 311 (38.0) 69 (36.3) 
Age, mo 48 (36–72) 48 (27–84) 
Highest temperature at home, °C 38.2 (37.7–38.9) 38.4 (37.8–39.0) 
Temperature at the ED, °C 37.4 (36.9–38) 37.2 (36.5–37.9) 
Duration of fever, h 24 (12–24) 24 (12 – 48) 
Blood test results   
 Procalcitonin, ng/mL 0.13 (0.09–0.30) 0.10 (0.05–0.33) 
 CRP, mg/L 8.0 (2.1–23.0) 8.1 (3.0–34.7) 
 WBC count, ×103 cells per mm3 12 100 (9260–15 400) 12 690 (9300–17 000) 
 ANC, cells per mm3 9540 (7100–12 600) 9755 (6391–13 242) 
CSF test results   
 WBC count, cells per mm3 126 (50–310) 163.0 (57–464) 
 ANC, cells per mm3 43 (10–148) 72 (10–246) 
 Bacterial meningitis, n (%) 61 (7.4) 31 (16.3) 
VariableDerivation Set (n = 819)Validation Set (n = 190)
Female, n (%) 311 (38.0) 69 (36.3) 
Age, mo 48 (36–72) 48 (27–84) 
Highest temperature at home, °C 38.2 (37.7–38.9) 38.4 (37.8–39.0) 
Temperature at the ED, °C 37.4 (36.9–38) 37.2 (36.5–37.9) 
Duration of fever, h 24 (12–24) 24 (12 – 48) 
Blood test results   
 Procalcitonin, ng/mL 0.13 (0.09–0.30) 0.10 (0.05–0.33) 
 CRP, mg/L 8.0 (2.1–23.0) 8.1 (3.0–34.7) 
 WBC count, ×103 cells per mm3 12 100 (9260–15 400) 12 690 (9300–17 000) 
 ANC, cells per mm3 9540 (7100–12 600) 9755 (6391–13 242) 
CSF test results   
 WBC count, cells per mm3 126 (50–310) 163.0 (57–464) 
 ANC, cells per mm3 43 (10–148) 72 (10–246) 
 Bacterial meningitis, n (%) 61 (7.4) 31 (16.3) 

Except for sex and bacterial meningitis, data are expressed as median (interquartile range).

In the ROC curve analysis, CSF ANC, CSF protein, serum CRP, and serum procalcitonin revealed an AUC >90% (Supplemental Fig 3).

Using the Youden index, we established the following cutoff points: serum procalcitonin of 1.2 ng/mL, serum CRP of 40 mg/L, CSF protein of 80 mg/dL, and CSF ANC 1000 per µL (Supplemental Fig 4). The bivariate comparison of predictors is shown in Table 2.

TABLE 2

Bivariate Comparison of Predictors in the Derivation Set Between Those With Aseptic Meningitis and Bacterial Meningitis

β-Coefficient95% CISignificance
CSF ANC >1.000/µL 11.136 1.86–66.61 .008 
Serum CRP >40 mg/L 16.271 3.50–75.62 .000 
CSF protein >80 mg/dL 31.491 6.81–145.67 .000 
Serum procalcitonin >1.20 ng/mL 65.606 14.80–290.93 .000 
CSF WBC count >500/µL 1.591 0.06–41.61 .780 
Serum ANC >10.000/mL 0.297 0.33–2.65 .277 
Serum WBC count >16.000/mL 1.09 0.00–2.97 .994 
CSF glucose >50 mg/L 0.943 0.12–7.51 .956 
β-Coefficient95% CISignificance
CSF ANC >1.000/µL 11.136 1.86–66.61 .008 
Serum CRP >40 mg/L 16.271 3.50–75.62 .000 
CSF protein >80 mg/dL 31.491 6.81–145.67 .000 
Serum procalcitonin >1.20 ng/mL 65.606 14.80–290.93 .000 
CSF WBC count >500/µL 1.591 0.06–41.61 .780 
Serum ANC >10.000/mL 0.297 0.33–2.65 .277 
Serum WBC count >16.000/mL 1.09 0.00–2.97 .994 
CSF glucose >50 mg/L 0.943 0.12–7.51 .956 

We developed the MSE on the basis of the results of the logistic regression (Table 3).

TABLE 3

Multivariate Logistic Regression Analysis

β-Coefficient95% CIP
Serum procalcitonin >1.20 ng/mL 484.50 161.46–1453.87 <.0001 
Serum CRP >40 mg/L 66.02 31.05–140.38 <.0001 
CSF ANC >1000/µL 73.18 36.10–148.33 <.0001 
CSF protein >80 mg/dL 117.80 52.55–264.06 <.0001 
β-Coefficient95% CIP
Serum procalcitonin >1.20 ng/mL 484.50 161.46–1453.87 <.0001 
Serum CRP >40 mg/L 66.02 31.05–140.38 <.0001 
CSF ANC >1000/µL 73.18 36.10–148.33 <.0001 
CSF protein >80 mg/dL 117.80 52.55–264.06 <.0001 

We assigned points to the 4 variables on the basis of the relative magnitude of the β-coefficient. Specifically, 3 points were assigned to the serum procalcitonin level higher than 1.2 ng/mL, 2 points to CSF protein >80 mg/dL, and 1 point for each of the other variables (CSF ANC >1000 per µL and serum CRP >40 mg/L) (Table 4). The range of the resulting MSE was thus 0 to 7 points.

TABLE 4

MSE

PredictorPoints
PresentAbsent
Serum procalcitonin >1.20 ng/mL 
Serum CRP >40 mg/L 
CSF ANC >1000/µL 
CSF protein>80 mg/dL 
PredictorPoints
PresentAbsent
Serum procalcitonin >1.20 ng/mL 
Serum CRP >40 mg/L 
CSF ANC >1000/µL 
CSF protein>80 mg/dL 

We tested the performance of the MSE in the validation set. An MSE ≥1 predicted bacterial meningitis with a sensitivity of 100% (95% CI: 89.0%–100%) and a specificity of 77.4 (95% CI: 70.3–83.2). Similar values were found when applying the MSE to both derivation and validation sets (Table 4). Distributions of all the patients included in the study with bacterial and aseptic meningitis related to the value of the MSE are shown in Fig 1. No children diagnosed with bacterial meningitis in the derivation or validation sets had an MSE <2.

FIGURE 1

Distribution of children with bacterial and aseptic meningitis by MSE.

FIGURE 1

Distribution of children with bacterial and aseptic meningitis by MSE.

Close modal

Finally, we compared the performance of the MSE to the BMS (Table 5) and the AUC of the new score and BMS (Supplemental Fig 5).

TABLE 5

Comparison of the Accuracy of the BMS and the MSE for Distinguishing Bacterial and Aseptic Meningitis

ScoreSensitivity, % (95% CI)Specificity, % (95% CI)PPV, % (95% CI)NPV, % (95% CI)Bacterial Meningitis Missed, % (95% CI)
Validation set (n = 190)       
 BMS       
 ≥1 93.5 (77.1–98.8) 50.3 (42.3–58.3) 26.9 (19.0–36.3) 97.6 (90.6–99.6) 6.5 (1.8–20.7) 
 ≥2 87.1 (69.2–95.8) 93.7 (88.4–96.8) 73.0 (55.6–85.6) 97.4 (93.0–99.2) 12.9 (5.1–28.9) 
 MSE       
 ≥1 100 (89.0–100) 77.4 (70.3–83.2) 46.3 (34.9–58.1) 100 (97.0–100) 0 (0–11.0) 
All the patients (derivation + validation set) (n = 1009)       
 BMS       
 ≥1 96.7 (90.1–99.2) 51.3 (48.0–54.5) 16.6 (13.6–20.1) 99.4 (98.0–99.8) 3.3% (1.1–9.1) 
 ≥2 90.2 (81.8–95.2) 95.3 (93.7–96.5) 65.9 (56.8–73.9) 99.0 (98.0–99.5) 9.8 (5.2–17.6) 
 MSE       
 ≥1 100 (95.0–100) 83.2 (80.6–85.5) 37.4 (31.4–43.8) 100 (99.4–100) 0 (0–5) 
ScoreSensitivity, % (95% CI)Specificity, % (95% CI)PPV, % (95% CI)NPV, % (95% CI)Bacterial Meningitis Missed, % (95% CI)
Validation set (n = 190)       
 BMS       
 ≥1 93.5 (77.1–98.8) 50.3 (42.3–58.3) 26.9 (19.0–36.3) 97.6 (90.6–99.6) 6.5 (1.8–20.7) 
 ≥2 87.1 (69.2–95.8) 93.7 (88.4–96.8) 73.0 (55.6–85.6) 97.4 (93.0–99.2) 12.9 (5.1–28.9) 
 MSE       
 ≥1 100 (89.0–100) 77.4 (70.3–83.2) 46.3 (34.9–58.1) 100 (97.0–100) 0 (0–11.0) 
All the patients (derivation + validation set) (n = 1009)       
 BMS       
 ≥1 96.7 (90.1–99.2) 51.3 (48.0–54.5) 16.6 (13.6–20.1) 99.4 (98.0–99.8) 3.3% (1.1–9.1) 
 ≥2 90.2 (81.8–95.2) 95.3 (93.7–96.5) 65.9 (56.8–73.9) 99.0 (98.0–99.5) 9.8 (5.2–17.6) 
 MSE       
 ≥1 100 (95.0–100) 83.2 (80.6–85.5) 37.4 (31.4–43.8) 100 (99.4–100) 0 (0–5) 

PPV, positive predictive value.

In the validation set, 2 patients with BMS = 0 were diagnosed with bacterial meningitis: a 1-month-old boy with meningitis with Streptococcus agalactiae (no seizure, negative Gram-stain result, CSF protein = 74.5 mg/dL, CSF ANC = 127 per µL, peripheral ANC = 2590 per µL, serum procalcitonin = 92.4 ng/mL, and serum CRP = 122.7 mg/L) and a 3-year-old girl with meningococcal meningitis (no seizure, negative Gram-stain result, CSF protein = 60 mg/dL, CSF ANC = 900 per µL, peripheral ANC = 9600 per µL, serum procalcitonin = 20.50 ng/mL, and serum CRP = 123.7 mg/L).

In this large multicenter study, the MSE accurately identified bacterial and aseptic meningitis in children between 2 months and 14 years old with CSF pleocytosis, without misclassifying those with bacterial meningitis. As shown in the validation set, the inclusion of procalcitonin and CRP levels improves the accuracy of a previous validated decision support tool for this context, the BMS.

We recommend that physicians admit and give antibiotics to all febrile children with pleocytosis and MSE ≥1. On the other hand, a less conservative management can be considered for those well-appearing, previously healthy febrile children without purpura with pleocytosis and MSE <2 who have no received antibiotics within 72 hours before the lumbar puncture. In this set of patients, discharge without antibiotics may be recommended if adequate outpatient follow-up can be ensured.

Procalcitonin and CRP have commonly been used in the management of children at risk for serious bacterial infection.10  In fact, procalcitonin and CRP have shown a better performance than WBC count and ANC to distinguish serious bacterial infections from viral illnesses. In addition, procalcitonin offers some advantages in the identification of patients with invasive bacterial infections,1114  specifically patients with meningococcal diseases,15  including meningitis.11,16  In line with this, in a previous series, the replacement of the peripheral ANC with procalcitonin significantly increased the specificity of the BMS.17  The BMS includes 5 dichotomous predictors (seizures during or before presentation, the ANC in peripheral blood and in CSF, CSF protein, and CSF Gram-stain).5  In our study, 2 patients with bacterial meningitis had a BMS of 0. One of them was a 3-year-old girl with meningococcal meningitis. This supports that some children with meningococcal meningitis may not be correctly classified with the BMS.8  This is worrisome because a significant percentage of patients with meningococcal meningitis may not develop a rash22 ; it also supports the inclusion of procalcitonin, which has a good performance in identifying patients with meningococcal infection. The second missed patient was a 1-month-old boy with S agalactiae meningitis. In young febrile infants, procalcitonin has shown a better performance than traditional tests in identifying infants at high risk of invasive bacterial infections, including meningitis.12  Nevertheless, considering that few infants <2 months with bacterial meningitis might be misclassified, it was later suggested that the BMS be used in children >2 months of age.6,23 

Like Nigrovic et al,5  we excluded certain patients for our study: infants <29 days old, those critically ill, those with purpura, and those not previously healthy. This seems reasonable. In fact, when a physician strongly suspects bacterial meningitis, the administration of intravenous antibiotics should not be delayed. On the other hand, we also excluded, as did Nigrovic et al,5  children treated with antibiotics within 72 hours before the lumbar puncture because it would be difficult to know whether they truly had aseptic meningitis.

Gram-stain is an excellent tool for distinguishing bacterial form aseptic meningitis. The CSF Gram-stain result is positive in ∼75% of cases of bacterial meningitis.24  When the Gram-stain result is positive, the specificity is higher than 97%.25  In addition, CSF Gram staining is rapid, inexpensive, and well validated for detecting bacteria. We decided not to include the Gram-stain in the MSE score for two reasons. First, it is not available 24 hours a day and 7 days a week in all EDs. In addition, it seems difficult not to put a child with CSF pleocytosis and a positive Gram-stain result on antibiotics. In fact, broad-spectrum antimicrobial therapy should be continued until CSF culture results are available.26  We recommend prescribing antibiotics to a child with pleocytosis and a positive Gram-stain result regardless of the value of the MSE. Nevertheless, we did not find any cases of children diagnosed with bacterial meningitis and a positive Gram-stain result and an MSE <2.

This study has certain limitations. We only included children with pleocytosis considered by Nigrovic et al5  suitable to be assessed by using a score in the ED. Hence, this score should not be applied to certain patients: those who are <29 days old, critically ill, with purpura, not previously healthy, and/or treated with antibiotics within 72 hours before the lumbar puncture. On the other hand, we conducted the study in 25 Spanish EDs. It is possible that the distribution of the causes of bacterial meningitis varies in other countries and would alter the performance of the MSE and BMS. Nevertheless, we consider that the distribution of the main pathogens involved in bacterial meningitis is likely to be similar in other countries with similar vaccination policies and coverage and that the MSE is able to help identify a population suitable for outpatient management without antibiotics. In addition, we first developed this clinical prediction rule and, afterward, we performed a prospective validation, which is typically the preferred approach for such tools despite this being difficult in the case of bacterial meningitis because of its rarity in high-income countries. Nevertheless, external validation would be appropriate to confirm our results. On the other hand, in the derivation set, it is possible to have missed some patients with fever and CSF pleocytosis not coded as “meningitis.” It was not possible for all the hospitals to check all the lumbar punctures performed in the derivation set, but all blood and CSF cultures were checked and no child with identification of a bacterial pathogen in the CSF or blood was missed. In addition, we think that not coding meningitis for children with pleocytosis is not expected in Spanish EDs included in the Infectious Diseases Working Group of the Spanish Society of Pediatric Emergencies. We do not think that this may bias the results of the study. Finally, the distribution of aseptic and bacterial meningitis varied in the derivation and validation sets. Nonetheless, we do not believe that this is likely to have influenced the results and main conclusion of the study.

MSE accurately distinguishes bacterial from aseptic meningitis in children with CSF pleocytosis. The inclusion of procalcitonin and CRP improves the performance of the BMS. The MSE can be used to guide initial clinical decision-making in children with CSF pleocytosis without misclassifying children with bacterial meningitis.

We acknowledge Mariano Plana (Barbastro), Ana Fernández Lorente (Hospital Universitario Basurto), Laura Míguez-Martín (Hospital Universitario de Cabueñes), Jose Angel Muñoz Bernal (Hospital Universitario Donostia), Diana Martínez Cirauqui (Complejo Hospitalario de Navarra), Javier Melgar Pérez (Hospital de Dénia), Sara Pons (Hospital Universitaro Dr Peset), María José Martín Díaz (Hospital Infantil Universitario Niño Jesús de Madrid), Aristides Rivas Garcia (Hospital General Universitario Gregorio Marañón), Mª Ángeles García Herrero (Hospital Universitario Príncipe de Asturias de Alcalá de Henares), Irene García de Diego (Hospital Universitario del Tajo), Carlos Miguel Angelats (Hospital Francesc de Borja de Gandía), Isabel Durán Hidalgo (Hospital Regional Universitario de Málaga), Laura Minguell (L’Hospital Universitari Arnau de Vilanova de Lleida), Silvia García (Hospital Universitario Cruces), Ana Isabel Mohedas Tamayo (Hospital Universitario de Fuenlabrada), Alberto Barasoain (Hospital Universitario Fundación Alcorcón), Helvia Benito (Hospital Universitario Río Hortega), Nuria Gilabert Iriondo (L’Hospital Universitari Son Espases), Esther Crespo Rupérez (Hospital Virgen de la Salud Toledo), Santos García (La Paz Regional Hospital), Virginia Gómez Barrena (Hospital Universitario Miguel Servet), Nuria Cortés Alvarez (Hospital Universitario Mútua Terrassa), Laia Sánchez Torrent (Hospital General, Parc Sanitari Sant Joan de Déu), Sandra Moya Villanueva (Hospital Universitari Parc Taulí), Susanna Hernández-Bou (Hospital General, Parc Sanitari Sant Joan de Déu), Mª Ángeles Martín (Instituto Valenciano de Pediatrïa, Unidad de Pediatrïa Integral Quiron, Quirón Valencia Hospital), Esther Lera Carballo (Vall d’Hebron University Hospital), Maria Teresa Alonso (Hospital Universitario Virgen del Rocío), Amalia Pérez (Hospital de Zumárraga), Marta Velazquez (Hospital de Terrassa, Consorci Sanitari de Terrassa).

Dr Mintegi conceptualized and designed the study, supervised data collection, analyzed the data, and wrote the initial draft of the manuscript; Dr García collaborated in the design of the data collection system and critically revised the manuscript; Drs Martín and Benito reviewed the design of the data collection system and critically revised the manuscript; Dr Arana-Arri collaborated in the design of the study, analyzed the data, and critically revised the manuscript; Dr Fernandez reviewed the design of the data collection system, supervised data collection, and critically revised the manuscript; Dr Hernández-Bou reviewed the design of the data collection system, coordinated the inclusion of emergency departments from the Meningitis Group of the Spanish Society of Pediatric Emergencies, and critically revised the manuscript; and all authors approved the final manuscript as submitted.

FUNDING: No external funding.

     
  • ANC

    absolute neutrophil count

  •  
  • AUC

    area under the receiver operating characteristic curve

  •  
  • BMS

    bacterial meningitis score

  •  
  • CI

    confidence interval

  •  
  • CRP

    C-reactive protein

  •  
  • CSF

    cerebrospinal fluid

  •  
  • ED

    emergency department

  •  
  • MSE

    meningitis score for emergencies

  •  
  • NPV

    negative predictive value

  •  
  • PCR

    polymerase chain reaction

  •  
  • ROC

    receiver operating characteristic

  •  
  • WBC

    white blood cell

1
Sawyer
MH
.
Enterovirus infections: diagnosis and treatment
.
Pediatr Infect Dis J
.
1999
;
18
(
12
):
1033
1039; quiz 1040
2
Rubio
G
,
Mintegui
S
,
Gaztelurrutia
L
,
Sánchez
J
.
Meningitis by enterovirus in pediatrics. Clinical characteristics and virologic diagnosis [in Spanish]
.
Enferm Infecc Microbiol Clin
.
1998
;
16
(
1
):
14
18
3
Lee
BE
,
Chawla
R
,
Langley
JM
, et al
.
Paediatric Investigators Collaborative Network on Infections in Canada (PICNIC) study of aseptic meningitis
.
BMC Infect Dis
.
2006
;
6
:
68
4
Sáez-Llorens
X
,
McCracken
GH
 Jr.
.
Bacterial meningitis in children
.
Lancet
.
2003
;
361
(
9375
):
2139
2148
5
Nigrovic
LE
,
Kuppermann
N
,
Malley
R
.
Development and validation of a multivariable predictive model to distinguish bacterial from aseptic meningitis in children in the post-Haemophilus influenzae era
.
Pediatrics
.
2002
;
110
(
4
):
712
719
6
Nigrovic
LE
,
Kuppermann
N
,
Macias
CG
, et al;
Pediatric Emergency Medicine Collaborative Research Committee of the American Academy of Pediatrics
.
Clinical prediction rule for identifying children with cerebrospinal fluid pleocytosis at very low risk of bacterial meningitis
.
JAMA
.
2007
;
297
(
1
):
52
60
7
Dubos
F
,
Korczowski
B
,
Aygun
DA
, et al
.
Distinguishing between bacterial and aseptic meningitis in children: European comparison of two clinical decision rules
.
Arch Dis Child
.
2010
;
95
(
12
):
963
967
8
Nigrovic
LE
,
Malley
R
,
Kuppermann
N
.
Meta-analysis of bacterial meningitis score validation studies
.
Arch Dis Child
.
2012
;
97
(
9
):
799
805
9
Dubos
F
,
De la Rocque
F
,
Levy
C
, et al;
Bacterial Meningitis Study Group
.
Sensitivity of the bacterial meningitis score in 889 children with bacterial meningitis
.
J Pediatr
.
2008
;
152
(
3
):
378
382
10
Van den Bruel
A
,
Thompson
MJ
,
Haj-Hassan
T
, et al
.
Diagnostic value of laboratory tests in identifying serious infections in febrile children: systematic review
.
BMJ
.
2011
;
342
:
d3082
11
Dubos
F
,
Korczowski
B
,
Aygun
DA
, et al
.
Serum procalcitonin level and other biological markers to distinguish between bacterial and aseptic meningitis in children: a European multicenter case cohort study
.
Arch Pediatr Adolesc Med
.
2008
;
162
(
12
):
1157
1163
12
Gomez
B
,
Bressan
S
,
Mintegi
S
, et al
.
Diagnostic value of procalcitonin in well-appearing young febrile infants
.
Pediatrics
.
2012
;
130
(
5
):
815
822
13
Luaces-Cubells
C
,
Mintegi
S
,
García-García
JJ
, et al
.
Procalcitonin to detect invasive bacterial infection in non-toxic-appearing infants with fever without apparent source in the emergency department
.
Pediatr Infect Dis J
.
2012
;
31
(
6
):
645
647
14
Fernández Lopez
A
,
Luaces Cubells
C
,
García García
JJ
,
Fernández Pou
J
;
Spanish Society of Pediatric Emergencies
.
Procalcitonin in pediatric emergency departments for the early diagnosis of invasive bacterial infections in febrile infants: results of a multicenter study and utility of a rapid qualitative test for this marker
.
Pediatr Infect Dis J
.
2003
;
22
(
10
):
895
903
15
Carrol
ED
,
Newland
P
,
Riordan
FA
,
Thomson
AP
,
Curtis
N
,
Hart
CA
.
Procalcitonin as a diagnostic marker of meningococcal disease in children presenting with fever and a rash
.
Arch Dis Child
.
2002
;
86
(
4
):
282
285
16
Dubos
F
,
Moulin
F
,
Gajdos
V
, et al
.
Serum procalcitonin and other biologic markers to distinguish between bacterial and aseptic meningitis
.
J Pediatr
.
2006
;
149
(
1
):
72
76
17
Garcia
S
,
Echevarri
J
,
Arana-Arri
E
,
Sota
M
,
Benito
J
,
Mintegi
S
;
Meningitis group of RISEUP-SPERG
.
Outpatient management of children at low risk for bacterial meningitis
.
Emerg Med J
.
2018
;
35
(
6
):
361
366
18
Velasco
R
,
Martínez
A
,
Benito
FJ
.
Analysis of the scientific production of the Spanish Society of Paediatric Emergencies (SEUP) in the period 2011-2017 [in Spanish]
.
An Pediatr (Barc)
.
2018
;
89
(
6
):
391
392
19
Benito Fernández
J
,
Luaces Cubells
C
,
Gelabert Colomé
G
, et al
.
An update of the diagnostic coding system by the Spanish Society of Pediatric Emergencies [in Spanish]
.
An Pediatr (Barc)
.
2015
;
82
(
6
):
442.e1
-
442.e7
20
Bonsu
BK
,
Harper
MB
.
Corrections for leukocytes and percent of neutrophils do not match observations in blood-contaminated cerebrospinal fluid and have no value over uncorrected cells for diagnosis
.
Pediatr Infect Dis J
.
2006
;
25
(
1
):
8
11
21
Nigrovic
LE
,
Shah
SS
,
Neuman
MI
.
Correction of cerebrospinal fluid protein for the presence of red blood cells in children with a traumatic lumbar puncture
.
J Pediatr
.
2011
;
159
(
1
):
158
159
22
Gangoiti
I
,
Valle
JR
,
Sota
M
,
Martinez-Indart
L
,
Benito
J
,
Mintegi
S
.
Characteristics of children with microbiologically confirmed invasive bacterial infections in the emergency department
.
Eur J Emerg Med
.
2018
;
25
(
4
):
274
280
23
Kowalsky
RH
,
Jaffe
DM
.
Bacterial meningitis post-PCV7: declining incidence and treatment
.
Pediatr Emerg Care
.
2013
;
29
(
6
):
758
766–768
24
Davis
LE
.
Acute bacterial meningitis
.
Continuum (Minneap Minn)
.
2018
;
24
(
5
):
1264
1283
25
Feldman
WE
.
Concentrations of bacteria in cerebrospinal fluid of patients with bacterial meningitis
.
J Pediatr
.
1976
;
88
(
4, pt 1
):
549
552
26
Tunkel
AR
,
Hartman
BJ
,
Kaplan
SL
, et al
.
Practice guidelines for the management of bacterial meningitis
.
Clin Infect Dis
.
2004
;
39
(
9
):
1267
1284

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

Supplementary data