Comparison of a Clinical Prediction Rule and a LAM Antigen-Detection Assay for the Rapid Diagnosis of TBM in a High HIV Prevalence Setting

Background/Objective: The diagnosis of tuberculous meningitis (TBM) in resource poor TB endemic environments is challenging. The accuracy of current tools for the rapid diagnosis of TBM is suboptimal. We sought to develop a clinical-prediction rule for the diagnosis of TBM in a high HIV prevalence setting, and to compare performance outcomes to conventional diagnostic modalities and a novel lipoarabinomannan (LAM) antigen detection test (Clearview-TBH) using cerebrospinal fluid (CSF). Methods: Patients with suspected TBMwere classified as definite-TBM (CSF culture or PCR positive), probable-TBM and non-TBM. Results: Of the 150 patients, 84 % were HIV-infected (median [IQR] CD4 count = 132 [54; 241] cells/ml). There were 39, 55 and 54 patients in the definite, probable and non-TBM groups, respectively. The LAM sensitivity and specificity (95%CI) was 31% (17;48) and 94 % (85;99), respectively (cut-point$0.18). By contrast, smear-microscopy was 100 % specific but detected none of the definite-TBM cases. LAM positivity was associated with HIV co-infection and low CD4 T cell count (CD4,200 vs..200 cells/ml; p = 0.03). The sensitivity and specificity in those with a CD4,100 cells/ml was 50 % (27;73) and 95 % (74;99), respectively. A clinical-prediction rule $6 derived from multivariate analysis had a sensitivity and specificity (95%CI) of 47% (31;64) and 98 % (90;100), respectively. When LAM was combined with the clinical-prediction-rule, the sensitivity increase