BACKGROUND

Differentiating bacterial from aseptic meningitis in children is critical for optimal treatment. While symptoms overlap, bacterial meningitis demands immediate antibiotics. Traditionally, CSF culture has been the gold standard for diagnosis, but its yield has declined with widespread vaccination. Consequently, some children with negative cultures may still have bacterial meningitis. The Bacterial Meningitis Score (BMS), a validated clinical prediction rule, offers a valuable tool, particularly in resource-limited settings, to better identify high-risk children and guide more effective treatment strategies.

OBJECTIVES

To evaluate the clinical utility and diagnostic accuracy of the BMS in identifying pediatric patients at high risk for bacterial meningitis.

METHODOLOGY

This retrospective cross-sectional study included 75 pediatric patients (aged 29 days to 18 years) with suspected meningitis seen at the Emergency Room of the Pediatrics Department in Mariano Marcos Memorial Hospital and Medical Center from March to November 2023. Eligible patients underwent lumbar puncture for CSF analysis. The BMS, a five-variable clinical tool including CSF Gram stain, CSF absolute neutrophil count, CSF protein, peripheral absolute neutrophil count, and seizure, were used to classify patients as very low risk (BMS=0) or not very low risk (BMS ≥1).

RESULTS

The diagnostic performance of the Bacterial Meningitis Score (BMS) across different cut-off thresholds is as follows: At a cutoff of ≥1, sensitivity is 100%, specificity is 36.80%, positive predictive value (PPV) is 33.3% (95% CI: 22% – 46%), negative predictive value (NPV) is 100% (95% CI: 84.5% – 100%), positive likelihood ratio (LR+) is 1.58 (95% CI: 1.29 – 1.93), negative likelihood ratio (LR–) is 0 (95% CI: 0 – NaN), and Youden’s index is 0.36. For a cut-off of ≥2, sensitivity is 88.90%, specificity is 78.90%, PPV is 57% (95% CI: 39% – 73%), NPV is 95% (95% CI: 85% – 98%), LR+ is 4.21 (95% CI: 2.48 – 7.16), LR– is 0.14 (95% CI: 0.03 – 0.52), and Youden’s index is 0.67. At a cut-off of ≥3, sensitivity drops to 61.10%, specificity increases to 98.20%, PPV rises to 91% (95% CI: 64% – 98%), NPV is 88%(95% CI: 78% – 94%), LR+ is 33.94 (95% CI: 4.82 – 251.61), LR– is 0.39 (95% CI: 0.22 – 0.70), and Youden’s index is 0.59. Finally, at a cut-off of ≥4, sensitivity is markedly low at 5.56%, specificity is perfect at 100%, PPV is 100% (95% CI: 20% – 100%), NPV is 77% (95% CI: 66% – 85%), LR+ is not applicable, LR– is 0.94 (95% CI: 0.84 – 1.05), and Youden’s index is 0.056. The optimal cutoff based on Youden’s index (0.67) was BMS ≥2, providing a more balanced trade-off between sensitivity (88.90%) and specificity (78.90%).

CONCLUSION

The BMS is a highly sensitive initial screen for bacterial meningitis in children, but its low specificity at the ≥1 cutoff necessitates a more judicious approach. Employing a ≥2 cutoff (Youden index 0.67) significantly improves diagnostic accuracy, optimizing resource utilization and enabling targeted interventions. While definitive diagnosis requires confirmatory testing, the BMS strategically guides initial triage, particularly crucial in resource-constrained environments.

Human ; Pneumonia

One of the important prerequisites in developing health programs and policies for pediatric neurologic diseases is accurate information about the status of these diseases in the country. At present, there is no existing registry of pediatric neurological diseases in the country. The Child Neurology Society Philippines (CNSP) sought to develop and implement a national registry for pediatric neurologic diseases. Specifically, it aimed to 1.) determine a consensus for the purposes of the registry, 2.) develop an implementation system from data collection until information dissemination, 3.) install a system for evaluating data reliability, and 4.) establish a feedback system for improvement. Results showed that the CNSP members wanted a census of diseases, to use the data for research, to identify participants for a research study, and for epidemiological surveillance. Data collection is by filling up forms at the clinic level and its submission to the centralized encoder. Monthly and annual census of diseases are generated and e-mailed to each member who has subscribed to these reports. Custom reports may be produced when requested. Early challenges in the implementation of the database include negative personal preconceived ideas about the database, collecting data, possible ethical concerns, and logistical limitations.