Abstract
The introduction of stool as a readily obtainable sample and the recently developed simple one-step (SOS) stool processing method on Xpert MTB/RIF Ultra (Xpert Ultra) offer an opportunity for TB diagnosis in children. We conducted this study in secondary health facilities in Ethiopia, which are the first-level referral facilities for childhood TB diagnosis and treatment, with the aim to determine if stool-based TB diagnosis can be performed with a reasonable level of concordance with sputum tests using Xpert MTB/RIF Ultra. Eligible children 0-14 years old with presumptive pulmonary TB were asked to provide stools in addition to routinely requested sputum samples. We determined the level of agreement between the stool and sputum test results. Of the 373 children included in the study, 61% were <5 years of age and 56% were male. Thirty-six children (9.7%) were diagnosed with TB, and all started treatment. The rate of concordance between stool and sputum was high, with a kappa value of 0.83 (p < 0.001). There were more Xpert Ultra positive results on stool (n = 27 (7.2%)) than on sputum/NGA (n = 23 (6.2%)). Laboratories in secondary hospitals can perform stool-based TB diagnosis in children, with high concordance between stool and sputum test results reaffirming the applicability of the SOS stool method.