<![CDATA[Introduction : Tuberculosis (TB) detection in children is challenging, particularly because of their paucibacillary nature. Healthcare facilities may also lack the necessary resources to diagnose this disease properly. Treatment-decision algorithm can be considered as a solution, but their performance need to be evaluated. Objective : To evaluate and compare the diagnostic performance of the IDAI scoring system and the WHO scoring algorithms (A and B) in diagnosing childhood tuberculosis, using bacteriological and/or histopathological examinations as the reference standard. Methods : A quantitative diagnostic study comparing TB scoring results with diagnostic reference. Performance was evaluated based on AUC value from the ROC curve, and agreement between screening methods was evaluated based on Intraclass Correlation Coefficient. Two independent researcher evaluated TB diagnosis in 65 children based on WHO treatment-decision algorithm (A and B version) and IDAI scoring system. Results of AFB, MRT (Xpert MTB/RIF) or FNAB evaluation was used as diagnostic reference. Results : IDAI scoring method has low agreement with both WHO algorithm (κ = 0.228 vs WHO A and 0.265 vs WHO B), with better agreement between the two WHO algorithms (κ = 0.516). The IDAI scoring system has 24.32% sensitivity, 78.57% specificity and diagnostic odds ratio (DOR) >1. Both WHO algorithm have 52.5-54.05% sensitivity, 32.14-36% specificity, and DOR <1. All screening methods showed weak diagnostic performance with AUC between 0.5–0.6 when compared with diagnostic reference, with no significant difference between methods. Conclusion : IDAI scoring system and both WHO algorithms had weak diagnostic performance for TB diagnosis in children]]>
