Answers are grounded in this paper and cite their sources.
Background: Pediatric tuberculosis (TB) remains difficult to diagnose. Culture, the gold standard, has limited sensitivity in children (~40-60%). We analyzed 452 symptomatic children ≤15 years who underwent chest x-ray (CXR), respiratory sample testing with Xpert Ultra, smear microscopy and culture, tuberculin skin test (TST), and interferon gamma release assay (IGRA) in Nairobi, Kenya. Methods: Three Bayesian latent class models (BLCMs) were used: Model 1 assumed conditional test independence; Model 2 incorporated random effects to account for shared biological (Smear, geneXpert Ultra/RIF and culture) or immunologic (IGRA, TST) test dependence; Model 3 extended Model 2 by adjusting for clinical covariates and NIH TB signs and symptoms. We estimated TB prevalence, test sensitivities/specificities and covariate effects. Results: Prevalence estimates under Models 1 [20.2% (15.9–24.7)] and 2 [24.5% (19.6–30.0)] were higher than under Model 3 [14.8% (11.0–19.3)]. All the three model-based prevalence estimates were considerably higher than the 9.9% of children identified through microbiological testing alone. Sensitivity and specificity were broadly consistent across models. Model 3 was considered preferred because it accounted for dependence among the tests and adjusted prevalence and sensitivity for clinical covariates. Under this Model 3, culture and GeneXpert MTB/RIF Ultra demonstrated moderate sensitivity (56.9% and 63.2%, respectively), while smear microscopy showed lower sensitivity (29.8%). CXR achieved moderate sensitivity (79.7%) but limited specificity (40.2%). IGRA and TST displayed sensitivities of 63.7% and 50.1%, and specificities of 81.5% and 65.8%. Older age was positively associated with TB (log-odds ratio 1.09; 95% CrI: 0.14–2.18). Conclusion: BLCMs estimated TB prevalence and test accuracies without a gold standard. Sensitivity analyses confirmed robustness to priors. Radiography demonstrated moderate sensitivity while culture and GeneXpert retained excellent specificity with improved test performance with older age.
No takes yet. Share an insight, caveat, or question.
Mecha et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: