Validation of a Clinical Prediction Rule for Distinguishing Bacterial and Aseptic Meningitis in Pediatric Patients.

Abstract

Introduction The treatments and prognosis of bacterial meningitis differ greatly from those of aseptic meningitis, making early identification and differentiation essential. Several different clinical prediction rules have been developed to distinguish bacterial meningitis from aseptic meningitis. We sought to validate one clinical prediction rule for pediatric patients utilizing a centralized data warehouse that collects daily data from 184 hospitals across the United States. Methods We retrospectively collected data on all patients aged 29 days to 14 years who presented to Hospital Corporation of America (HCA) Healthcare hospitals from January 1, 2016, to May 31, 2021, with a diagnosis of meningitis. Our study replicated the original study of the meningitis score for emergencies (MSE) for the pediatric clinical prediction rule and assigned 3 points for procalcitonin (PCT) >1.2 ng/dL, 2 points for CSFprotein >80 mg/dL, and 1 point for each of the other variables of C-reactive protein (CRP) >40 mg/L and CSF absolute neutrophil count >1000 cells per mm3. Patients were categorized either as having bacterial or aseptic meningitis. Using the clinical prediction rule, a calculation of the sensitivity, specificity, positive predictive value, negative predictive value, and receiver operating characteristic (ROC) curve was performed. Results The optimum test characteristic was found to have a score of ≥ 3, showing a sensitivity of 92.86% (95% CI, 83.3-100), a specificity of 65.22% (95% CI, 51.5-79), a positive predictive value of 61.90% (95% CI, 47.2-76.6), and a negative predictive value of 93.75% (95% CI, 85.4-100). The ROC curve from this study showed an area under the curve (AUC) of 0.7892 (95% CI, 0.681-0.897). Conclusion Our study validated a high sensitivity for distinguishing bacterial meningitis from aseptic meningitis, suggesting the clinical prediction rule has clinical utility as a predictive screening tool. Although the original MSE advised a cutoff score of ≥1, our study suggests that a score ≥3 would give the best test performance.