Open Forum Infectious Diseases
Tongue swab Xpert MTB/RIF Ultra testing for tuberculosis in adolescents: a cross-sectional study of diagnostic accuracy and acceptability
By Emily Lai-Ho MacLean, Thuy Thu Ma, Long Huynh Chuong and 7 others.
Abstract
Abstract Background Improved diagnostics are needed for tuberculosis among adolescents. Tongue swab (TS) molecular testing is a promising strategy for tuberculosis testing. We evaluated diagnostic accuracy and acceptability of Xpert MTB/RIF Ultra (Xpert) using TS for tuberculosis detection in adolescents. Methods We conducted a cross-sectional diagnostic accuracy study in southern Vietnam. Adolescents aged 10-19 who were recommended to undergo investigation for tuberculosis were eligible. Participants provided TS and sputum samples and were surveyed regarding sampling experiences. TS was tested on Xpert, with sputum tested on Xpert and liquid culture. We utilised a composite reference standard of a positive sputum Xpert or culture result to define disease status. Sensitivity, specificity, and diagnostic yield were calculated for TS Xpert. Results From July to December 2025, we enrolled 225 adolescents. Most participants (157/225, 70%) were close contacts of a person recently diagnosed with tuberculosis and many (129/225, 57%) did not exhibit symptoms of presumptive tuberculosis. All adolescents could provide TS, while 116 (52%) could produce mucopurulent sputum. Tuberculosis prevalence was relatively low (12/225, 5.3%). TS Xpert sensitivity and specificity were 58.3% (90%CI: 35.6-78.0) and 99.5% (90%CI: 97.9-99.9), respectively. TS diagnostic yield among all diagnosed was 58.3% (7/12). TS sampling was highly acceptable to adolescents; the quickness and simplicity of collecting TS were considered favourably. Conclusions TS Xpert sensitivity and diagnostic yield were relatively low among adolescents recommended for tuberculosis investigation. All participants were able to provide a TS. Specificity was excellent. TS’s high acceptability indicates it remains a promising sample for diagnostic algorithms.



