Abstract
Background/Objectives: Stool-based GeneXpert testing has become a useful approach for diagnosing pediatric pulmonary tuberculosis (PTB). This study compared two stool-processing methods, centrifugation-based processing (CBP) and simple one-step (SOS), for detecting PTB in children using Xpert MTB/RIF Ultra (Ultra). Methods: Children with presumptive PTB were screened cross-sectionally, and stool samples were collected and tested with Ultra using the CBP method from March 2022 to December 2024 across seven divisions of Bangladesh. A subset of stool samples (n = 281) that tested positive (n = 191) and negative (n = 90) by the CBP method were re-tested again with the same sample by Ultra using the SOS method. The results of the Ultra with SOS-processed stool were compared with the CBP method to evaluate overall agreement and detection efficiency across different bacterial burdens. Results: The SOS method detected 97 of 191 CBP-positive samples, resulting in a positive percentage agreement of 50.8% (95% CI: 43.5-58.1). All 90 Ultra-negative stool were also negative by the SOS method, yielding a negative percentage agreement of 100% (95% CI: 96.0-100.0). Overall agreement between the methods was 66.6% (Kappa: 0.398). The SOS method detected 100% of high- (4/4) and medium- (7/7), 97.3% (36/37) of low-, and 83.3% (35/42) of very-low-bacterial-burden samples, but only 14.9% (15/101) of the trace-detected samples that were identified by the CBP method. Conclusions: Stool testing with Ultra using the SOS processing method missed a significant number of the most prevalent form of child TB-the 'trace-detected' category identified by the CBP method. For increased detection of childhood TB nationwide, the national program should prioritize the use of Ultra on stool samples processed by the CBP method.