Diagnostic Accuracy of Cartridge-Based Nucleic Acid Amplification Test (CBNAAT) for the Rapid Diagnosis of Tuberculous Meningitis: A Prospective Hospital-Based Study

Abstract

Background: Prompt treatment is crucial for tuberculous meningitis and diagnosing it through traditional cerebrospinal-fluid tests is either slow or of very low sensitivity. The purpose of this study is to assess the diagnostic accuracy and operational utility of cerebrospinal fluid (CSF) CBNAAT for the rapid diagnosis of TBM. We use a simulated dataset to carry out the study as no patient data were provided. In our sample of 160 analysable adult patients suspected for meningitis, a composite reference standard classified study participants into cases of TBM with varying degrees of certainty (definite [n=28], probable [n=44], and possible [n=26]), and non-TBM cases [n=62]. Definite and probable TBM cases were of primary interest. CBNAAT results showed a total of 49 true positives, 23 false negatives, 6 false positives, and 82 true negatives. In our study, the sensitivity and specificity of CBNAAT were 68.1% (95% confidence interval) and 93.2% (85.9% - 96.8%), respectively. Sensitivity improved from 50.0% to 79.5% if at least 4 mL CSF was submitted. Median turnaround time was 2.2 hours for CBNAAT, 6.2 hours for smear, and 18.2 days for culture. AUC was 0.864 for a model that included combined CBNAAT. In conclusion, CBNAAT is a rapid, relatively high-specificity test for TBM, that provides rule-in evidence, but is an incomplete test for the rule-out diagnosis of TBM. Clinical, CSF, and imaging data should also be used in conjunction with TBM CBNAAT and culture. All observations reported have been simulated and must be replaced with real data that is prospectively verified prior to any publication or clinical use.