In-home tongue swab testing for tuberculosis screening among household contacts in South Africa: a prospective cohort study.

Abstract

Household contact investigation for tuberculosis is limited by reliance on clinic-based testing and sputum-dependent diagnostics. To inform household-level screening strategies, we estimated the diagnostic accuracy of in-home molecular testing using tongue swabs compared with sputum, and quantified linkage to care among household contacts with confirmed tuberculosis. We conducted a prospective cohort study at 26 government health clinics in South Africa. We recruited index patients with drug-sensitive pulmonary tuberculosis and their household contacts. Participants aged 18 years or older and not on tuberculosis treatment within the past 6 months were eligible. Household contacts provided tongue swabs and, where possible, sputum specimens during household visits. Xpert MTB/RIF Ultra on GeneXpert Omni or Edge devices were used. Tongue swabs were tested individually or in pools of up to three participants, whereas sputa were tested individually. The primary outcome was the estimated diagnostic accuracy (ie, sensitivity and specificity) of tongue swab testing compared with sputum using paired results, irrespective of pooling strategy. Secondary outcomes included performance of pooled tongue swab testing as a household-level screening strategy (defined by conditional detection of tuberculosis by pool size and the residual probability of underlying tuberculosis following a negative pooled result) and linkage to care (defined as the proportion of individuals with a positive sputum result presenting for clinic-based tuberculosis services and their time-to-clinic presentation). Between June 15, 2021, and Oct 29, 2024, 438 households and 909 household contacts were enrolled. The median age of household contacts was 39 years (IQR 28-55), 570 (62·7%) of 909 were female and 338 (37·2%) were male, 377 (41·5%) were symptomatic, 132 (14·5%) reported living with HIV, and 207 (22·8%) had a previous history of tuberculosis. Tongue swabs were obtained from 901 (99·1%) of 909 household contacts, whereas 292 (32·1%) produced sputum. Among 203 paired tongue swab-sputum results, sensitivity of tongue swab testing was 61·9% (95% CI 38·4-81·9) and specificity was 100% (98·9-100). Among individually tested tongue swabs paired with sputum results (58 paired tests), sensitivity was 100% (95% CI 47·8-100) and specificity was 100% (93·3-100). Among pooled tests in which at least one contributing participant was sputum-positive, five (55·6%) of nine for two-swab pools and three (42·9%) of seven for three-swab pools tested positive. Among pooled tests with a negative result, the residual probability of an underlying sputum-positive case was four (4·4%) of 91 for two-swab pools and four (9·5%) of 42 for three-swab pools. Sputum testing identified 29 (3·2%) of 909 household contacts with tuberculosis (all previously undiagnosed). Following referral, 25 (86·2%) of 29 presented for clinic-based care within a median of 1 day (IQR 1-2). In-home molecular testing using tongue swabs enabled near-universal specimen collection and microbiological testing among household contacts. When implemented as a pooled, household-level screening approach, tongue swab testing identified tuberculosis among contacts unable to produce sputum and reduced the need for universal individual testing. However, pooling was associated with a measurable residual risk of missed disease following negative pooled results. These findings indicate that efficiency gains from pooled tongue swab testing must be balanced against the residual risk of missed cases. US National Institutes of Health, Australian Department of Foreign Affairs and Trade, and the UK Foreign, Commonwealth and Development Office.