From 2019 to 2020, we conducted a national multicenter retrospective study of children (<15 years of age) diagnosed with PM across seven geographical divisions of China. A total of 33 tertiary Grade A hospitals in 23 provinces (27 cities) participated in this study. Of the 33 collaborating hospitals, 13 are in East China, 6 in North China, 4 each in Central and Northwest China, 3 in Southwest China, 2 in South China, and 1 in Northeast China.

All patients enrolled met the standards of PM diagnosis established by the World Health Organization (WHO) (Organization WH., 2003). The PM diagnostic criteria of PM were as follows: 1) clinical signs and symptoms (persistent or recurrent fever, bulging fontanelles, neck stiffness or painful neck, vomiting, headache, altered mental status, seizures, or focal neurologic signs) that comply with meningitis; 2) CSF examination revealing at least one of the following (turbid appearance; CSF leukocytosis (>100 cells/μL); leukocytosis (10–100 cells/μL AND either an elevated protein (>1.0g/L) or decreased glucose (<2.2 mmol/L). If CSF protein and glucose results are not available, diagnose using the first 2 conditions (turbid appearance or leukocytosis >100 cells/μL). 3) S. pneumoniae was identified by microbiological testing (positive blood or CSF culture, positive CSF antigen detection, positive CSF mNGS testing) or Gram staining and microscopy of the CSF.

Based on the International Classification of Diseases Tenth Revision diagnostic codes and Systematized Nomenclature of Medicine codes, these cases were identified by retrieving the electronic medical records for the discharge diagnoses “acute purulent meningitis” or “acute bacterial meningitis” or “central nervous system infection” or “intracranial infection” or “meningitis” in each collaborating hospital from January 2019 to December 2020. Only the first medical record was included for analysis if the patient was hospitalized multiple times for the same diagnosis. Medical records, laboratory, and microbiological data of all enrolled cases were systematically assessed and recorded by locally trained pediatric researchers using a standard case report form and entered into the Good Clinical Data Management System.

We retrospectively collected data including baseline information on demographics, clinical characteristics, the time of admission and discharge, craniofacial and spinal anatomical abnormalities associated with central nervous system infection, causative microorganisms, laboratory findings, cranial imaging, treatment, antibiotic susceptibility test results, and outcomes.

All enrolled cases with PM diagnosis underwent LP after admission and cerebrospinal fluid (CSF) examinations were conducted, encompassing leukocyte and red blood cell count, differential leukocyte count, CSF biochemical, CSF culture, Gram staining and microscopy), as well as blood routine test, blood cultures, and cranial imaging. CSF and blood specimens were cultured, and S. pneumoniae was identified using standard methods as previously described. Microbiological specimens were cultured by BACT/ALERT 3D 240 automatic blood culture instrument (Mérieux, France), and S. pneumoniae were identified by the automatic bacterial identification system (VITEK Compact 2, France) at each surveillance center. Drug susceptibility testing of S. pneumoniae to penicillin was supplemented with the E-test method (Sangji et al., 2012). The culture and identification procedures are followed following the National Clinical Laboratory Procedures (Hong et al., 2015). Normal reference ranges of CSF parameters adhered to the guidelines outlined in the National Clinical Laboratory Procedures (Hong et al., 2015). In this study, antigen detection and molecular biological detection methods were selectively employed in only some of the PM cases that were in serious status requiring earlier identification of pathogens or patients with empirical antibiotics therapy failed or patients with negative initial CSF Gram stains whose CSF and(or) blood cultures at 72 h incubation were negative (including BinaxNOW for Streptococcus pneumoniae detection and mNGS).

The interpretation of positive CSF mNGS results followed the guidelines established by the Expert Consensus on the Use of Cerebrospinal Fluid Metagenomic next-generation sequencing for Infectious Diseases of the Central Nervous System (Infectious Diseases and Cerebrospinal Fluid Cytology Group of the Neurology Section of the Chinese Medical Association, 2021).

The identification of S. pneumoniae isolates and antibiotic susceptibility tests were conducted by an automatic bacterial identification system (VITEK Compact, France) or Optochin Discs (OXOID, UK) according to the National Guide to Clinical Laboratory Procedures. Antimicrobial susceptibility results of S. pneumoniae isolates were interpreted according to the standards of Clinical and Laboratory Standards Institute (Clinical and Laboratory Standards Institute, 2021).

We defined the clinical situation on the discharge day as the clinical outcome. “Cured” was characterized by relief of clinical symptoms and signs, negative CSF detect results, and the absence of cranial imaging complications during hospitalization (Sun, 2002). “Disease improvement” was considered as relief of clinical symptoms and signs, negative CSF culture, normalization of blood inflammatory indicators, approximately normal CSF leukocyte count, CSF protein and (or) glucose level not returned to normal level, and no progress in cranial imaging complications during hospitalization (Sun, 2002). “Unhealed” was defined as no improvement in clinical symptoms, abnormal CSF test results, and the presence of neurological complications observed on cranial imaging. “Death” was defined as a fatality occurring during hospitalization or within 3 days of discharge from the hospital after the cessation of treatment. Relapse was defined as the recurrence of bacterial meningitis caused by the same pathogenic bacteria after the completion of antimicrobial therapy from the initial episode (Subspecialty Group of Neurology et al., 2019). Adverse outcomes included unhealed, relapse, death, being discharged automatically, abandoned treatment, or being transferred to other hospitals for further treatment.

Predisposing factors of PM are underlying diseases and conditions that predispose a person to bacterial meningitis, including cerebrospinal fluid nasal/otorrhea, head trauma, intracranial or ear malformation, previous history of bacterial meningitis, perforation of the tympanic membrane, sacrococcygeal furs and cochlear implants, organ transplantation, and tumor-related diseases, basic metabolic disorders, congenital organ malformations, hereditary diseases, chronic diseases (van de Beek et al., 2016). Extracranial infectious diseases included pneumonia, mastoiditis, sinusitis, and otitis media (Infection Group of the Chinese Society of Pediatrics et al., 2018).

SPSS 19.0 software (IBM Corporation, Armonk, NY, USA) was used for statistical analysis. Categorical data were expressed in frequency and percentage, and the chi-square test and Fisher’s exact test were used to analyze the statistical difference. Continuous data were expressed as median (Q1, Q3) for non-normal distribution and Kruskal–Wallis H-test was used to analyze the statistical difference, with the P value set at less than 0.05 with statistical significance.

During Jan 1, 2019, and Dec 31, 2020, 167 children (<15 years old) with PM were enrolled for screening. 7 cases (4.2%) were excluded due to hospital-acquired, post-surgical, post-traumatic meningitis and two episodes for one person. A total of 160 PM patients met the inclusion criteria, including 105 (65.6%) in 2019 and 55 (34.4%) in 2020. There were 82 cases (51.3%) in East China, 30 cases (18.8%) in Southwest China, 15 cases (9.4%) in North China, 14 cases (8.8%) in Northwest China, 10 cases (6.3%) in Central China, 7 cases (4.4%) in Northeast China and 2 cases (1.3%) in South China. Among 160 PM children, 103 (64.4%) were males and 57 (35.6%) were females, with a male-to-female ratio of 1.8:1. The median age at presentation was 1 year and 3 months, with cases distributed as follows: 1 case (0.6%) under 28 days old, 5 cases (3.1%) between 28 days and 3 months old, 63 cases (39.4%) between 3 months and 1 year old, 46 cases (28.8%) aged 1 to under 3 years, 28 cases (17.5%) aged 3 to under 5 years, and 17 cases (10.6%) aged 5 to under 15 years. The median hospitalization time was 23 (17,30) days, and the hospitalization cost was 42 (28, 62) thousand yuan. In 2020, both the hospitalization time and cost were significantly lower than those in 2019 (Z=-3.113, -3.469; P=0.002, 0.001) ( Table 1 ). The onset time of PM formed two peaks, with the highest incidence in November, December, and January of the subsequent year, followed by another peak in April, May, and June ( Figure 1 ).

55 cases (34.4%) had one or more predisposing factors for bacterial meningitis, including 17 cases (10.6%) of CSF nasal/otorrhea, 13 cases (8.1%) of cranial trauma, 12 cases (7.5%) of intracranial or ear malformation, 9 cases (5.6%) of previous history of bacterial meningitis, 6 cases of leukemia. Of these, 1 case each (0.6%) was tympanic membrane perforation, sacrococcygeal fur sinus, cochlear implantation, hemophagocytic syndrome, and post-liver transplantation. Seven children had more than one predisposing factor for bacterial meningitis.

113 patients (70.6%) had one or more extracranial infections at the initial stage of PM, including 81 patients (50.6%) with pneumonia, 29 with mastoiditis (18.1%), 23 with sinusitis (14.4%) and 19 with otitis media (11.9%). There were no significant differences in the proportion of predisposing factors of bacterial meningitis and extracranial infectious diseases among different onset years (χ² = 0.397, 0.095; P=0.529, 0.758) ( Table 1 ).

The main clinical manifestations of PM children are presented in ( Table 1 ). Fever (147 cases, 91.9%) was the most common symptom. 33 patients (20.6%) were hospitalized with a critical presentation on admission, including respiratory failure (23 cases, 14.4%), coma (20 cases, 12.5%), mechanical ventilation (13 cases, 8.1%), and septic shock (7 cases, 4.3%). PM critical presentations were predominant in the less than 1-year-old group (21 cases, 63.6%), including 12 cases of coma, 12 cases of mechanical ventilation, and 6 cases of septic shock. Three (1.9%) children were admitted with multiple organ failure, all of whom were less than 3 years of age. One child with PM developed a cerebral hernia and died on the day of admission. The proportion of critical presentations such as respiratory failure and coma in the 2020 group was significantly lower than that in the 2019 group (χ² = 5.418,3.804; P=0.020, 0.048) as shown in ( Table 1 ).

S. pneumoniae detection samples and results are shown in ( Table 1 ). Of the 35 children with positive CSF mNGS tests, the positive rates of cerebrospinal fluid cultures and blood cultures were 10.3% (87 PM cases performed mNGS test.) and 5.7%, respectively; the rate of both blood and CSF cultures was 1.1%, and the rate of positive results for CSF mNGS only was 25.3%. Among the 21 children with positive S. pneumoniae antigen test in CSF, the positive rate of CSF culture and blood culture was each 10.3% (78 PM cases performed S. pneumoniae antigen test.), the positive rate of both blood and CSF culture was 3.8%, and the positive rate of CSF S. pneumoniae antigen test result was only 6.4%.

Intracranial complications were identified through cranial magnetic resonance imaging and (or) cranial CT scans during hospitalization in 69 (43.1%) children, of which 17 (25%) had more than 1 intracranial imaging complication. The common intracranial complications were subdural effusion and (or) empyema in 43 cases (26.9%), hydrocephalus in 24 cases (15.0%), brain abscess in 23 cases (14.4%), cerebral hemorrhage in 8 cases (5.0%), and other cerebral vascular disorders (including encephalomalacia, cerebral infarction, and encephalatrophy) in 13 cases (8.1%). Subdural effusion and (or) empyema and hydrocephalus manifested predominantly in children under the age of one in 90.7% and 83.3%, respectively, which was shown in ( Table 1 ).

The rates of cured and disease improvement in PM cases were 22.5% and 66.3%, respectively. 18 (11.3%) children had adverse outcomes, including 5 (3.1%) children who died during hospitalization (two patients were <1 year old, one patient 2 years old, and two patients >5 years old) and 4 early deaths (within 1 week of hospitalization); two children had relapses (one case within 3 weeks and 1 case after 3 weeks, both in 2019); five cases not cured, and six cases of automatic discharge and (or) abandonment of treatment.

The initial CSF leukocyte count was documented in 147 (91.9%), the initial glucose concentration of CSF in 156 (97.5%), and the initial protein concentration of CSF in 152 (95.0%) of the 160 children diagnosed with PM.

The initial leukocyte counts of CSF were normal in two children, both of whom were younger than 1 year of age (10 months of age and 5 months and 19 d, respectively). CSF cultures of one child were positive for S. pneumoniae and of one child were positive for both blood and cerebrospinal fluid S. pneumoniae cultures. Two cases with normal initial leukocyte counts in CSF had a decrease in the initial glucose concentration of CSF (1.1 mmol/L, 1.0 mmol/L, respectively), and an increase in the initial protein concentration of CSF (3.0 and1.6 g/L, respectively).

Among the 20 children with an initial CSF glucose concentration of ≥2.8 mmol/L, 12 (60.0%) exhibited positive CSF cultures, while three (15.0%) had positive results in both blood and CSF cultures, as well as in CSF mNGS. Additionally, both CSF culture and CSF mNGS were positive in 1 case, and only CSF mNGS was positive in 1 case (5.0%). The initial CSF leukocyte count was elevated in 20 cases, of which 10 cases were >1000×106/L, 2 cases were >500∼1000×106/L, 5 cases were 100∼500×106/L, and 3 cases were <100×106/L; the initial CSF protein concentration was >1000 mg/L in 10 cases, (450∼1000) mg/L in 4 cases, and normal in 6 cases. The initial CSF parameters of the children and pre-discharge from the hospital are shown in ( Table 2 ).

Among 157 S. pneumoniae strains isolated, only 91, S. pneumoniae antimicrobial susceptibility data were registered (84 from CSF, 7 from blood) ( Table 3 ). Seven blood-derived S. pneumoniae strains were determined for antimicrobial susceptibility according to the breakpoint of parenteral administration of meningitis strain. We found that 68 S. pneumoniae isolates underwent penicillin susceptibility testing: 11 (16.2%) with susceptibility and 57 (83.8%) with resistance to penicillin. The susceptibility rates of S. pneumoniae to levofloxacin, moxifloxacin, rifampicin, and chloramphenicol were 81.5%, 82.4%, 96.2%, and 91.3%. The susceptibility rates of S. pneumoniae to cefotaxime, meropenem, and ceftriaxone were 56.1%, 51.1%, and 63.5, respectively. No S. pneumoniae isolates were resistant to vancomycin, linezolid, or ertapenem. S. pneumoniae were completely resistant to erythromycin.