The accuracy of tuberculous meningitis diagnostic tests using Bayesian latent class analysis

Area of Science:

• Infectious Diseases
• Neurology
• Public Health

Background:

• Tuberculous meningitis (TBM) is a severe form of tuberculosis with significant mortality and disability.
• Delayed diagnosis of TBM is common due to the lack of definitive gold standard tests.
• Accurate diagnostic test performance data is crucial for effective TBM management.

Purpose of the Study:

• To estimate the prevalence of TBM in suspected patients.
• To determine the diagnostic performance (sensitivity and specificity) of four common tests for TBM.
• To compare the effectiveness of different diagnostic procedures for TBM.

Main Methods:

• Bayesian latent class analysis was employed to estimate TBM prevalence and test performance.
• Data from 1,213 patients suspected of TBM were analyzed.
• Sensitivity, specificity, and Area Under the Curve (AUC) were calculated for mycobacteria culture, Gene Xpert, protein levels, and leukocyte count.

Main Results:

• The estimated prevalence of TBM was 34.8% (95% CI: 28.8 - 41.3).
• The sensitivity of the culture test was 62.7% and Gene Xpert was 57.5%.
• Gene Xpert demonstrated high specificity (95.9%), while leukocyte count (AUC 76.0%) and protein levels (AUC 73.4%) showed moderate diagnostic capability.

Conclusions:

• This study provides crucial data on the performance of routine diagnostic tests for TBM.
• Understanding test accuracy and TBM prevalence can improve disease control strategies.
• Enhanced diagnostic information can lead to better treatment outcomes for TBM patients.