Supporting Data docx 51KB

New Clinical-Based Scoring Systems To Diagnose Tuberculous Meningitis In Children

Objective: Due to the low accuracy of culture techniques in bacteriological confirmation and the lack of head imaging facilities for diagnosing tuberculous meningitis (TBM), this study aim to establish scoring system consisting of clinical manifestations and simple laboratory examination to help diagnosing TBM in children. Method: Retrospective study using multivariable diagnostic predictive model with children aged 3 months to 18 years diagnosed as meningitis, hospitalized during July 2011 until November 2021 in a tertiary hospital. Diagnosis was made based on CSF analysis and head CT scan result, then subjects divided into TBM group and non-TB meningitis group. Calculation of odds ratio, logistic regression test, ROC curve analysis were used to analyse sensitivity and specificity of the scoring system. Result: From 10 variables which have statistical significance with TBM, 8 variables were obtained for establishing the predictive model to identify TBM. These variables divided into two scoring parts which both had good discrimination and calibration, the systemic scoring part (4 criteria, cut-off score >=3, sensitivity of 78.8%, specificity of 86.6% with AUC of 89.9% (p<0.001)) and the neurological scoring part (4 criteria, cut-off score >=2, sensitivity of 61.2%, specificity of 75.2% with AUC of 73.3% (p<0.001)). Furthermore, these scoring systems when used together and met the cut-off score respectively, can predict the diagnosis of TBM in children well (sensitivity 47.1%, specificity 95.1%, positive predictive value 90.9%). Conclusion: a clinical scoring systems consist of two parts, systemic score and neurological score, have good ability in predicting the diagnosis of TBM in children.