Diagnostic accuracy and acceptability of self- and health worker-collected tongue swabs for <i>Mycobacterium tuberculosis</i> complex detection in adults in South Africa

Abstract

Abstract Tongue swabs (TSs) are a non-invasive specimen type for the detection of Mycobacterium tuberculosis complex (MTBC) and can expand access to testing for individuals unable to produce sputum. This study evaluated the diagnostic performance and user acceptability of self-collected and health worker (HW)-collected tongue swabs using the Xpert MTB/RIF Ultra (Ultra) assay and assessed participant perspectives on self-collection. In this prospective, cross-sectional study, symptomatic and asymptomatic adults under investigation for TB were enrolled from a high HIV prevalence setting. Each participant provided both a self-collected and a HW-collected TS, which were tested using Ultra. Ultra TS results were compared to liquid culture as the reference standard and sputum Ultra as a comparator. Participant perspectives on self-collection were captured via questionnaires. Sensitivity on Ultra for both self- and HW-collected TSs was 68% (95% CI:51.9-81.9), compared to liquid culture. This sensitivity was significantly higher than that of sputum smear microscopy (46%, 95% CI: 30.7–62.6; McNemar’s p = 0.003 ). Tongue swab sensitivity was lower than sputum Ultra (80.5%; p&lt;0.001) and decreased with low bacillary loads. Importantly, TSs enabled MTBC detection in six participants unable to produce sputum. Most participants (&gt;90%) found self-collection instructions easy to follow, reporting high confidence and comfort, and trust in results from self-collected TSs. This study demonstrates that self-collected TSs perform comparably to those collected by health workers for TB detection using Ultra and are both feasible and acceptable in a high TB/HIV burden setting. To maximize impact, clear training instructions and robust linkage to care remain critical priorities. Importance This study supports the use of tongue swabs (TSs) as a non-invasive alternative for tuberculosis diagnosis, particularly for individuals unable to produce sputum. When tested on the Xpert MTB/RIF Ultra assay, self-collected TSs performed comparably to health worker-collected swabs, yielding a 68% sensitivity (95% CI: 51.9–81.9) relative to liquid culture. This sensitivity was significantly higher than that of sputum smear microscopy (46%, 95% CI: 30.7–62.6; McNemar’s p = 0.003 ). Importantly, TSs successfully detected Mycobacterium tuberculosis complex in six patients who could not provide sputum, underscoring their clinical utility in expanding diagnostic access. Furthermore, high user acceptability (&gt;90%) confirms that self-collection is both feasible and trusted by patients in high TB/HIV burden settings. To maximize real-world impact, implementing clear training instructions and establishing robust linkage to care, especially following negative results, remain critical programmatic priorities.