Stool-based Xpert MTB/RIF Ultra assay as a tool for detecting pulmonary tuberculosis in children with abnormal chest imaging: A prospective cohort study

Objectives To evaluate the diagnostic efficacy of stool-based Xpert MTB/RIF Ultra assay versus other assays for the detection of paediatric pulmonary tuberculosis (PTB). Methods A prospective head-to-head comparative study was conducted from Dec 2017 to May 2019 in Shanghai Public Health Clinical Centre. Samples were collected from children (< 15 years) with abnormal chest imaging (X-ray or CT scan) results for the following tests: Ultra on stool sample (Ultra-Stool), Ultra on respiratory tract sample (Ultra-RTS), Xpert MTB/RIF assay (Xpert) on RTS (Xpert-RTS), acid-fast bacilli smear on RTS (AFB-RTS), and Mycobacterium tuberculosis (Mtb) culture on RTS (Culture-RTS). The results were compared with a composite reference standard. Results A total of 126 cases with paired results were analysed. Against a composite reference standard, Ultra-RTS demonstrated the highest sensitivity (52%) and specificity (100%). Ultra-Stool showed 84.1% concordance with Ultra-RTS, demonstrating 45.5% sensitivity and 94.7% specificity (kappa = 0.65, 95% CI= 0.51–0.79). The sensitivity of Ultra-Stool was similar to Mtb culture (45.5%, p = 1.000) and higher than AFB-RTS (27.3%, p < 0.05). Assay positivity was associated with age and infiltration range in chest imaging. Conclusions When RTS is difficult to obtain, stool sample-based Ultra is a comparable alternative.