Abstract
BACKGROUND: Swab-based molecular tests are emerging as more accessible and lower-cost options for tuberculosis (TB) diagnosis. We conducted multi-country diagnostic accuracy evaluations of late prototype versions of Truenat MTB Ultima (MTB Ultima) and MiniDock MTB Test (MiniDock MTB). METHODS: We enrolled consecutive people aged ≥12 years with presumptive TB at outpatient clinics in India, Uganda, and Vietnam. We evaluated the diagnostic accuracy of MTB Ultima and MiniDock MTB when using tongue and sputum swabs against a sputum liquid culture-based microbiological reference standard. Additionally, we compared the diagnostic accuracy of the swab-based molecular tests with that of sputum Xpert MTB/RIF Ultra (Xpert) and sputum smear microscopy. FINDINGS: From January to September 2024, 1050 participants were included in the tongue swab MTB Ultima evaluation, 197 in the sputum swab MTB Ultima evaluation and 322 in the MiniDock MTB evaluations. With tongue swab-based testing, sensitivity was 77.9%, 95% CI: 70.3, 84.2 for MTB Ultima and 85.7%, 95% CI: 75.3, 92.9 for MiniDock MTB. Both were more sensitive than sputum smear microscopy (p < 0.001, McNemar's test, for both comparisons). With sputum swab-based testing, sensitivity was 93.6%, 95% CI: 82.8, 97.8, for MTB Ultima and 91.1%, 95% CI: 82.1, 95.9 for MiniDock MTB. Compared to sputum Xpert, sensitivity differed by -6.4%, 95% CI: -15.5, 2.7 for MTB Ultima and -3.0%, 95% CI: -8.6, 2.6 for MiniDock MTB. Specificity exceeded 98% for all index tests. INTERPRETATION: MTB Ultima and MiniDock MTB exceed minimum World Health Organisation accuracy targets for sputum- and non-sputum-based near-point-of-care TB tests and offer strong potential to make universal molecular testing for TB a reality. FUNDING: National Institutes of Health (U01AI152087), Gates Foundation, and Global Health Labs.