Abstract
BACKGROUND: Childhood tuberculosis (TB) diagnosis remains challenging, partly because children cannot provide sputum. This study evaluated the diagnostic accuracy of the Simple One-Step (SOS) stool method with Xpert MTB/RIF Ultra (Xpert-Ultra) for childhood TB compared to culture and Xpert-Ultra on a respiratory sample (RS) and clinical diagnosis. It also assessed the feasibility and acceptability of stool testing according to laboratory staff, and caregivers' sample preference. METHODS: We enrolled children (≤10 years) with presumptive pulmonary tuberculosis in Ethiopia. RS was tested using Xpert-Ultra and culture; stool samples were tested using the SOS stool method with Xpert-Ultra. Laboratory staff and caregivers' opinions were assessed using standardised questionnaires. RESULTS: Of the 898 children enrolled, 792, 832 and 794 were included for assessing the diagnostic accuracy of SOS stool with Xpert-Ultra against culture, RS Xpert-Ultra and clinical diagnosis, respectively, yielding sensitivity estimates for SOS stool with Xpert-Ultra of 69.1% (95% confidence interval (CI) 56.0-79.7%), 76.8% (95% CI 64.2-85.9%) and 59.0% (95% CI 47.9-69.2%), respectively. The specificity was ≥98.8% for all comparisons. The rate of non-determinate test results was 2.8% after one repeat test. According to laboratory staff, stool collection was feasible and acceptable and the SOS stool method was easy to perform. Most caregivers (75%) preferred stool for TB diagnosis over RS. CONCLUSION: This study shows that SOS stool Xpert-Ultra testing offers a good alternative to RS testing for TB in children who cannot spontaneously produce a sputum sample and would otherwise need to undergo invasive procedures to obtain RS for diagnosis.