Abstract
BACKGROUND: Tongue swabs are a promising specimen for tuberculosis (TB) diagnosis. In a previous study using a consensus protocol, tongue swabs tested with Xpert MTB/RIF Ultra (Xpert Ultra, Cepheid, USA) outperformed sputum smear microscopy, but a substantial proportion (6.1%) of results were non-actionable (e.g., invalid/error). We evaluated a revised protocol in four high TB burden countries. METHODS: Participants aged ≥12 years with presumptive TB were enrolled from outpatient clinics in the Philippines, South Africa, Nigeria, and Zambia. Tongue swabs were processed using Sample Reagent (SR, Cepheid, USA) diluted 2:1 with phosphate buffer or phosphate-buffered saline and tested with Xpert Ultra. Diagnostic performance was assessed against a culture-based microbiological reference standard and compared to sputum-based tests. RESULTS: From March to November 2024, 1168 participants were enrolled (median age 37 [IQR 28-48] years; 46.7% female, 21.8% living with HIV, 18.5% culture-confirmed TB). The proportion of nonactionable results was 5.6% overall, but was less than 4% in all countries except South Africa (15.4%). Tongue swab sensitivity was 66.0% (95% CI 59.0-72.5); specificity was 99.6% (95% CI 98.9-99.9). CONCLUSION: The revised protocol yielded low error rates at most sites and moderate sensitivity, supporting tongue swabs as an alternative when sputum is unavailable.