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

The predictive value of a combination of clinical and radiological features with interferon-ã release assay (IGRA) for diagnosis of active tuberculosis (TB) disease among TB-exposed children is unknown. 150 symptomatic HIV-negative children (aged 3 months to 14 years), prospectively recruited through active contact tracing, were included. Backward stepwise logistic regression and bootstrapping techniques were used for the development and internal validation of a clinical prediction model for active TB disease. Model discrimination and incremental value of a positive IGRA test were assessed by area under the receiver operating characteristic curve (AUC). 35 (23%) children were diagnosed with active TB disease and started on treatment and 115 (77%) had other respiratory tract infections. A final parsimonious clinical model, comprising age <5 years (adjusted (a)OR 4.8, 95% CI 2.0-11.5) and lymphadenopathy on clinical examination (aOR 4.9, 95% CI 1.8-13.0) discriminated active TB disease from other disease with an AUC of 0.70 (95% CI 0.61-0.80). A positive IGRA result did not improve the discriminatory ability of the clinical model (c-statistic 0.72 versus 0.70; p=0.644). A clinical algorithm, including age <5 years and lymphadenopathy classified 70% of active TB disease among symptomatic TB-exposed children. IGRA does not add any discriminatory value to this prediction model.