Abstract
BACKGROUND: Tuberculosis (TB) diagnosis remains difficult in children under 5 years of age (under-5s), who have high TB morbidity and mortality rates. In a high-burden TB setting, we investigated the diagnostic characteristics of Xpert MTB/RIF Ultra testing of tongue swabs (TS-XU) collected from under-5s. METHODS: In a masked, prospective, observational study, tongue swabs were collected from enrolled hospitalized under-5s deemed high risk for TB disease who were categorized into 1 of the following: confirmed, unconfirmed, or unlikely TB. RESULTS: Of 201 enrolled under-5s, 11 (5.5%) had confirmed TB, 53 (26.4%) unconfirmed TB, and 137 (68.2%) unlikely TB. TS-XU testing reported "Mtb detected" in 116 (57.7%) of 201 under-5s: positive results were "trace" (90/116, 77.6%), "very low" (21/116, 18.1%), and "low" or "medium" (4/116 [3.4%] and 1/116 [0.8%], respectively). There were no "high" TS-XU results. When trace results were presumed negative, TS-XU sensitivity was 17.2% (95% CI, 7.9%-26.4%) and specificity 89.1% (95% CI, 83.8%-94.3%), and TS-XU detected Mtb in 15 (10.9%) of 137 children with unlikely TB. Our data showed that TS-XU, in addition to routine TB testing, increased TB detection rates by 19.2%. CONCLUSIONS: Despite the difficulty of interpreting trace-positive results, TS-XU testing increased TB detection rates in hospitalized under-5s with presumptive TB.