We report the findings of a prospective laboratory diagnostic accuracy study to evaluate the sensitivity, specificity, and predictive values of the Xpert MTB/RIF Ultra assay for Mycobacterium tuberculosis detection in fresh stool specimens from children under 15 years of age with confirmed tuberculosis (TB) disease from Dushanbe, Tajikistan. Six hundred eighty-eight (688) participants were enrolled from April 2019 to October 2021. We identified 16 participants (2.3%) with confirmed TB disease, defined as ≥1 TB sign/symptom plus microbiologic confirmation. With the Xpert MTB/RIF Ultra assay for stool, we found a sensitivity of 68.8% (95% CI, 46.0 to 91.5) and a specificity of 98.7% (95% CI, 97.8 to 99.5) in confirmed TB disease. Our results are comparable to other published studies; however, our cohort was larger and our confirmed TB disease rate lower than most. We also demonstrated that this assay was feasible to implement in a centralized hospital laboratory in a low-middle-income Central Asian country. However, we encountered obstacles such as lack of staffing, material ruptures, outdated government protocols, and decreased case presentation due to COVID-19. We found eight patients whose only positive test was an Xpert Ultra stool assay. None needed treatment during the study; however, three were treated later, suggesting such cases require close observation. Our report is the first from Central Asia and one of a few from a low-middle-income country. We believe our study demonstrates the generalizability of the Xpert MTB/RIF Ultra assay on fresh stool specimens from children and provides further evidence supporting WHO's approval of this diagnostic strategy. IMPORTANCE The importance of this report is that it provides further support for WHO's recent recommendation that fresh stool is an acceptable sample for GeneXpert TB testing in children, especially small children who often cannot produce an adequate sputum sample. Diagnosing TB in this age group is difficult, and many cases are missed, leading to unacceptable rates of TB illness and death. In our large cohort of children from Dushanbe, Tajikistan, the GeneXpert stool test was positive in 69% of proven cases of TB, and there were very few false-positive tests. We also showed that this diagnostic strategy was feasible to implement in a low-middle-income country with an inefficient health care delivery system. We hope that many more programs will adopt this form of diagnosing TB in children.
Keywords: Mycobacterium tuberculosis; Xpert MTB/Rif Ultra; pediatric infectious disease; stool.