No added value of interferon-γ release to a prediction model for childhood tuberculosis

Area of Science:

• Pediatric infectious diseases
• Respiratory medicine
• Diagnostic accuracy studies

Background:

• Accurate diagnosis of active tuberculosis (TB) in children exposed to TB is challenging.
• The utility of combining clinical features with interferon-γ release assay (IGRA) for TB diagnosis in this population remains unclear.

Purpose of the Study:

• To develop and validate a clinical prediction model for active TB disease in symptomatic, TB-exposed children.
• To assess the incremental value of IGRA testing in improving diagnostic accuracy.

Main Methods:

• Prospective recruitment of 150 HIV-negative, symptomatic children exposed to TB.
• Development of a clinical prediction model using backward stepwise logistic regression.
• Internal validation and assessment of model discrimination using receiver operating characteristic curves (AUC).

Main Results:

• A parsimonious clinical model including age <5 years and lymphadenopathy achieved an AUC of 0.70.
• Active TB disease was diagnosed in 35 children (23%); other respiratory infections in 115 (77%).
• A positive IGRA result did not significantly improve the model's discriminatory ability (c-statistic 0.72 vs. 0.70, p=0.644).

Conclusions:

• A clinical algorithm incorporating age and lymphadenopathy can identify active TB in symptomatic, TB-exposed children.
• Interferon-γ release assay testing does not enhance the diagnostic performance of this clinical prediction model.