A novel diagnostic model for tuberculous meningitis using Bayesian latent class analysis.

Area of Science:

• Neurology
• Infectious Diseases
• Medical Diagnostics

Background:

• Diagnosing tuberculous meningitis (TBM) is challenging due to the absence of a gold standard test.
• Current microbiological tests for TBM have limited sensitivity, and clinical assessments can be subjective.
• A reliable diagnostic model is needed for early TBM detection, especially before microbiological results are available.

Purpose of the Study:

• To develop and validate a diagnostic model for tuberculous meningitis (TBM).
• To improve early detection of TBM in patients with suspected brain infections.
• To assist clinicians in identifying patients at high risk for TBM.

Main Methods:

• A logistic regression diagnostic model was fitted using data from 659 individuals with suspected brain infections.
• Latent class analysis was employed to estimate unknown values from three mycobacterial tests (Ziehl-Neelsen smear, Mycobacterial culture, GeneXpert).
• A simplified model and scoring table were developed for early screening and validated internally.

Main Results:

• Factors associated with TBM included HIV, miliary TB, long symptom duration, and high cerebrospinal fluid (CSF) lymphocyte count.
• The full and simplified diagnostic models demonstrated high sensitivity and specificity (e.g., 86.0% sensitivity and 79.0% specificity for the full model).
• The model accurately identified patients at higher risk for TBM, with performance metrics validated internally.

Conclusions:

• The developed diagnostic model demonstrates reliable performance for TBM detection.
• This model can serve as a clinical decision support tool to identify patients at high risk of TBM.
• The accurate and well-calibrated models can aid both clinical practice and future research in TBM diagnosis.