Tongue Swab Xpert MTB/RIF Ultra Testing for Tuberculosis in Adolescents: A Cross-sectional Study of Diagnostic Accuracy and Acceptability

Abstract

Background

Improved diagnostics are needed for tuberculosis (TB) among adolescents. Tongue swab (TS) molecular testing is a promising strategy for TB testing. We evaluated diagnostic accuracy and acceptability of Xpert MTB/RIF Ultra (“Xpert”) using TSs for TB detection in adolescents.

Methods

We conducted a cross-sectional diagnostic accuracy study in southern Vietnam. Adolescents aged 10–19 years who were recommended to undergo investigation for TB were eligible. Participants provided TS and sputum samples and were surveyed regarding sampling experiences. TS samples were tested on Xpert, with sputum tested on Xpert and liquid culture. We utilized 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 TB and many (129/225 [57%]) did not exhibit symptoms of presumptive TB. All adolescents could provide a 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% confidence interval [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 TSs were considered favorably.

Conclusions

TS Xpert sensitivity and diagnostic yield were relatively low among adolescents recommended for TB 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.

In our diagnostic accuracy study of tongue swab Xpert MTB/RIF Ultra among adolescents recommended for tuberculosis investigation in Vietnam, sensitivity was relatively low, but specific was excellent. Acceptability of tongue swab testing was high, suggesting tongue swab testing should continue to be improved for this age group.

Graphical Abstract

Graphical Abstract.

This graphical abstract is also available at Tidbit: https://tidbitapp.io/tidbits/tongue-swab-xpert-mtb-rif-ultra-testing-for-tuberculosis-in-adolescents-a-cross-sectional-study-of-diagnostic-accuracy-and-acceptability

With over 10 million incident cases each year, but only 8 million cases notified to national tuberculosis programs (NTPs) [1], tuberculosis (TB) persists as a serious global health challenge. Adolescents, defined by the World Health Organization (WHO) as people aged 10–19 years [2], are estimated to experience approximately 17% of new TB cases globally [3]. WHO's child and adolescent “TB Roadmap” policy document calls for adolescent-specific TB strategies and recognizes that their age-specific needs must be met to improve health outcomes [4]. However, a lack of adolescent-specific data may prevent countries from implementing approaches that are tailored to adolescents’ physical and social needs.

The importance of including children and adolescents in evaluating new diagnostic tools and approaches was highlighted in a recent consensus report aiming to accelerate the pipeline for pediatric TB diagnostics [5]. In particular, the document emphasizes that tests utilizing a pathogen-based measurand should be prioritized for evaluation in young people, as the test target is consistent across ages and demographics. This rationale extends to the need to investigate alternative sample types, like tongue swabs (TSs), with widely available assays, such as Xpert MTB/RIF Ultra (Cepheid, USA; hereafter “Xpert”) [6].

Promisingly, WHO has recently recommended the use of TSs for TB testing on low-complexity automated nucleic acid amplification tests, including Xpert, when sputum and other respiratory samples cannot be obtained [7]. The recommendation is applicable to adults and adolescents, although data informing the recommendation were mainly from adults with presumptive TB disease. A systematic review of molecular tests for TB using oral swabs included 20 studies, though only 6 recruited participants in the adolescent age range [8]. Since the review's publication, molecular testing using TSs has proven to be most sensitive and specific, and primary data are accumulating quickly, as evidenced by WHO's new policy recommendation. However, an evidence gap persists regarding diagnostic accuracy of TS molecular testing in the adolescent age group. To our knowledge, no published work has focused specifically on TS molecular TB testing in an adolescent population. This means that adolescents may be having adult recommendations applied to them, without knowing if they are appropriate or fit for purpose [9].

Therefore, we aimed to estimate the diagnostic accuracy and yield of TSs on Xpert MTB/RIF Ultra for detecting TB among adolescents aged 10–19 years in Vietnam. Additionally, we aimed to investigate adolescents’ experiences and acceptability of this testing scheme.

MATERIALS AND METHODS

Study Design and Setting

We conducted a multisite, cross-sectional diagnostic accuracy study. Participants were recruited from Can Tho and An Giang provinces in the Mekong Delta region of southern Vietnam. Both provinces have high incidence rates of TB, at over 120 per 100 000 [10]. Study recruitment occurred at the Can Tho Lung Hospital (formerly Can Tho Tuberculosis and Lung Diseases Hospital), a tertiary hospital at the edge of Can Tho City, Can Tho, and at Long Xuyen City Health Centre, a secondary hospital in Long Xuyen, An Giang. Participants also were recruited at decentralized, community screening events organized by the NTP, where close contacts of people with newly diagnosed TB were invited to undergo testing for TB. Consecutive adolescents in these settings were prospectively invited to participate in our study. All testing procedures were conducted at the Can Tho Lung Hospital laboratory by trained laboratory technicians.

Eligibility Criteria

Adolescents aged 10–19 years who were recommended to undergo investigation for TB were eligible to participate. Reasons for TB investigation include (1) living with a contact who had recently received a TB diagnosis, (2) presumptive disease based on signs/symptoms of TB, and (3) chest X-ray (CXR) abnormalities. Participants must have been capable of providing informed consent and must not have received TB treatment in the previous 12 months. Concurrent extrapulmonary TB was not an exclusion criterion, but the participant must have been recommended for pulmonary TB assessment.

Procedures

All participant activities occurred at a single study visit. Participants were asked to not eat or drink for 30 minutes before sample collection. Swabs (Copan FLOQswab 520CS01) were first collected by swabbing the back of the tongue for 15 (to a maximum of 30) seconds with a sweeping and rolling motion. Swab heads were snapped off into dry 2-mL cryovial tubes and stored in a cold box (2°C–8°C) for transport to the laboratory; all TSs were tested within 24 hours. Each participant was then coached to provide 2 sputum samples. Sputum appearance was documented by study staff as “mucopurulent,” “saliva,” “blood-tinged,” or “food debris” and samples were tested regardless of appearance. After providing TS and sputum samples, participants were asked by study staff to complete a short paper-based survey of questions structured using a Likert scale (strongly disagree, disagree, neutral, agree, strongly agree) pertaining to their experience of sample collection.

Separately, laboratory technicians who collected swabs and/or performed TS Xpert for the study each completed 1 short semi-structured interview. Technicians were asked prespecified questions developed by the study team regarding perceived TS sampling feasibility, processing, and running. Interviews were conducted at the Can Tho Lung Hospital in Vietnamese by a native Vietnamese-speaking team member (L. H. C.). Interviews were recorded and stored on a password-protected shared drive.

Our index test was TS Xpert Ultra. In brief, once brought to the laboratory, 700 μL of 2:1 Cepheid sample reagent:phosphate buffer solution was adding to each TS-containing cryotube. TS-buffer tubes were vortexed and then incubated for 15 minutes. Phosphate buffer (1.5 mL) was added to each Xpert cartridge. After TS incubation was completed, 0.5 mL of fluid from the TS-buffer tube was pipetted into the buffer-containing cartridge. Xpert tests were then run normally on the GeneXpert platform. At the time of index testing, laboratory staff were blinded to participants’ disease status and demographic information.

Digital or conventional CXR was completed following local protocols for each participant and were read by an experienced physician or TB doctor. Images were scored as “normal,” “abnormal, inconsistent with TB,” or “abnormal, consistent with TB.”

One sputum each was run on liquid culture (BD BACTEC MGIT, BD, USA) and on Xpert; all samples were tested fresh without freezing. The composite reference standard (CRS) was defined as either a positive result for “MTB detection” on sputum Xpert or confirmed growth of Mycobacterium tuberculosis (Mtb) on liquid culture, or both. “Trace” results were considered positive for analyses. Reference standard testing was performed according to manufacturer's instructions. Study data were managed using Research Electronic Data Capture (REDCap) hosted at the University of Sydney [11].

Statistical Analyses

The primary outcome measures were sensitivity and specificity with accompanying 90% confidence intervals (CIs) for TS Xpert. Secondary outcomes included diagnostic yield of TS Xpert; rates of error and invalid results; acceptability of TS sampling by participants; and perceived acceptability of TS Xpert testing by laboratory workers. A precision-based approach was used to estimate sample size [12, 13]; to calculate 90% CIs with a width of ±20% around the sensitivity estimate, we estimated we would need 206 individuals. To account for possible unsuccessful test results, we aimed to recruit 225 participants.

Demographic features were summarized using descriptive statistics. Sensitivity and specificity were calculated following the intention-to-diagnose principle [14, 15], with unsuccessful results among reference standard–positive or –negative groups analyzed within that disease stratum only [14]. A complete case analysis (ie, individuals with valid positive or negative results on first test runs only, ie, not including results of repeat testing) was also conducted. Diagnostic yield was calculated among all diagnosed with TB via CRS and among all tested [16], and among those only producing salivary samples. Differences between participants’ experiences and acceptability of TS and sputum sampling were tested using Wilcoxon rank-sum tests. Tabular analyses were conducted using epiR [17]. All analyses were run using R version 4.5.1 [18].

Analysis of Interviews

Qualitative data from laboratory staff interviews were assessed through interview responses, reflecting the discussion guide (see Supplementary Material 1) and commonalities and differences between the 3 participants’ responses. We used these data to contextualize and interpret the quantitative data and did not perform any further abductive or interpretive analysis.

Patient Consent

All participants provided informed consent. Per local regulations, for adolescents aged 10–15 years, participant verbal assent and parental consent were obtained; for participants aged 16 and 17 years, written participant and parental consent were obtained; and participants aged 18 and 19 years provided their own written consent. Participants found to have TB on reference standard testing were referred for usual care in the Vietnamese NTP. Index test results did not inform usual care.

Ethical approval was obtained from the University of Sydney Human Research Ethics Committee (2024/HE001890), Australia, as well as the Institutional Review Board of Can Tho Lung Hospital (then known as Can Tho Tuberculosis and Lung Diseases Hospital) (483/QĐ–BVLBP), Vietnam.

RESULTS

Between 2 July and 26 December 2025, we enrolled 225 adolescents aged 10–19 years (Supplementary Figure 1), of whom 157 (70%) were close contacts of people with newly diagnosed TB, 63 (28%) presented with symptoms of presumptive TB, and 5 (2%) were recommended for further TB investigation due to abnormal CXRs.

Median (Q1, Q3) participant age was 16 years (13, 18 years) and a minority of participants (96/225 [43%)] were symptomatic (Table 1); cough was the most frequently reported respiratory symptom (Table 1), with a median duration of 1 week. Nonrespiratory symptoms (Supplementary Table 1) were not commonly reported (28/225 [12%]). All participants were investigated for pulmonary TB without concurrent extrapulmonary TB. Altogether, 15 participants (15/225 [6.7%]) had CXR abnormalities consistent with signs of TB, with 11 (11/15 [73%]) reporting concurrent cough. No one reported previously receiving TB treatment.

Table 1.

Study Population Features

Characteristic	Total Count (N = 225)
Age, y
Median (Q1, Q3)	16 (13, 18)
Minimum, maximum	10, 19
Gender
Girl/woman	102 (45%)
Boy/man	123 (55%)
HIV status
Positive	0 (0%)
Negative	4 (1.8%)
Not tested	221 (98%)
Comorbidities
None	222 (99%)
Epilepsy	1 (0.4%)
Tonsilitis	1 (0.4%)
Sinusitis	1 (0.4%)
Diabetes mellitus	0 (0.0%)
BMI, kg/m2
Median (Q1, Q3)	19.6 (17.4, 22.2)
Minimum, maximum	12.4, 38.7
Chest radiography results
Normal	198 (88%)
Abnormal, consistent with TB	15 (6.7%)
Abnormal, not consistent with TB	12 (5.3%)
Respiratory symptoms present
None	129 (57%)
Any	96 (43%)
Respiratory symptoms
Cough	80 (83%)
Breathlessness	7 (7.3%)
Hemoptysis	3 (3.1%)
Chest pain	4 (4.2%)
Other respiratory symptoms	28 (29%)
Nonrespiratory symptoms
None	197 (88%)
Any	28 (12%)

Data are presented as No. (%) unless otherwise indicated.

Abbreviations: BMI, body mass index; Q, quartile; TB, tuberculosis.

Tongue swabs were collected from all participants. At least 1 mucopurulent sputum was collected from 116 participants (51.6%), while 109 could only produce saliva. Younger adolescents aged 10–14 years (54/93 [58.1%]) were more likely than adolescents aged 15–19 years (55/132 [30.3%]) to only produce salivary samples (P = .0221).

The CRS classified 12 (5.3%) adolescents as TB positive, including a trace-positive 13-year-old with CXR abnormalities consistent with TB. TS Xpert produced 1 error (0.4%); repeat testing was not possible. Sputum Xpert also produced 1 error; repeat testing using a second sputum sample yielded a negative result.

Following the intention-to-diagnose principle, TS Xpert sensitivity was 58.3% (90% CI, 35.6%–78.0%) while specificity was 99.5% (90% CI, 97.9%–99.9%) (Table 2). Complete case analysis estimated TS Xpert sensitivity of 58.3% (90% CI, 35.6%–78.0%) and specificity of 100.0% (90% CI, 98.7%–100.0%) (Table 2). Among adolescents who could produce a sputum sample, sensitivity was 60.0% (90% CI, 35.2%–80.6%) and specificity remained 100.0% (90% CI, 97.5%–100.0%) (Table 2), and for those only producing salivary samples, sensitivity was 50.0% (90% CI, 12.1%–87.9%) with specificity of 99.1% (90% CI, 96.0%–99.8%). One participant tested positive on sputum Xpert for rifampicin resistance; TS Xpert did not detect this case of TB. Supplementary Table 3 shows exploratory analyses of TS Xpert diagnostic accuracy stratified by age and by respiratory symptom status. Sensitivity was higher among older adolescents (15–19 years) than in their younger counterparts (10–14 years). Among participants not reporting respiratory symptoms, TS Xpert did not detect any cases of TB (Supplementary Table 3).

Table 2.

Diagnostic Accuracy of Tongue Swab Xpert Ultra Among All Participants in the Intention-to-Diagnose Population; Complete Case Analysis Population; Among Participants Producing Mucopurulent Sputum Only; and Among Participants Producing Salivary Sputum Samples Only

Analysis	Swab Xpert Result	Composite Reference Standard Result
		TB Positive	TB Negative
Intention-to-diagnose analysisa (N = 225)	Positive	7	0
	Negative	5	213
	Unsuccessfulb	0	1
	Total classified	12	213
Complete case analysisc (n = 222)	Positive	7	0
	Negative	5	210
	Unsuccessful	NA	NA
	Total classified	12	210
Mucopurulent sputum (n = 116)	Positive	6	0
	Negative	4	106
	Unsuccessfulb	0	0
	Total classified	10	106
Salivary sputum (n = 109)	Positive	1	0
	Negative	1	106
	Unsuccessfulb	0	1
	Total classified	2	106

Abbreviations: NA, not applicable; TB, tuberculosis.

^aWhen estimating sensitivity, unsuccessful results among disease-positive are excluded from numerator and included in denominator. When estimating specificity, unsuccessful results among disease-negative are included in numerator and in denominator.

^bFor participants requiring repeat testing due to a reference standard component test initially producing an unsuccessful result, the final classification based on at least 1 valid result was used.

^cWhen estimating sensitivity and specificity, individuals missing a valid result on first run of each test are removed from the calculation.

TS Xpert diagnostic yield among all participants diagnosed with TB was 58.3% (7/12) (Table 3). Among those only producing salivary samples, the diagnostic yield among all diagnosed was 50.0% (1/2). The diagnostic yields of TS Xpert and first sputum Xpert among all tested for TB were 3.1% (7/225) and 4.5% (11/225), respectively, a difference that was not statistically significantly different (P = .449).

Table 3.

Diagnostic Yield of Tongue Swabs and Sputum Using Xpert Ultra Among Adolescents

Diagnostic Yield	Value
Diagnostic yield of TS Xpert among all diagnosed (n = 12)	58.3% (7/12)
Diagnostic yield of TS Xpert among all diagnosed who cannot produce sputum (n = 2)	50.0% (1/2)
Diagnostic yield of TS Xpert among all tested (n = 225)	3.1% (7/225)
Diagnostic yield of initial sputum sample Xpert among all tested (n = 225)	4.5% (11/225)
Diagnostic yield of TS Xpert among all tested who cannot produce sputum (n = 109)	0.9% (1/109)

Abbreviations: TS, tongue swab; Xpert, Xpert MTB/RIF Ultra.

Examining the observed test result patterns (Table 4) shows that TS Xpert results were positive only among participants who had positive sputum Xpert, positive culture, CXR abnormality, and reported symptoms of TB. Notably, TS Xpert results for Mtb detection were negative among 2 adolescents who were positive on both sputum Xpert and culture. TS Xpert was more frequently positive when sputum Xpert semi-quantitative result was high (4/4) or medium (1/2) versus low, very low, or trace (1/4) (Supplementary Table 2).

Table 4.

Frequency of Observed Test Result and Symptom Patterns

Tongue Swab Xpert	Sputum Xpert	Sputum Culture	Chest X-Ray	Respiratory Symptoms	Frequency
(−)	(−)	(−)	Normal	None	118
(−)	(−)	(−)	Normal	Any	78
(−)	(−)	(−)	Abnormal not c/w TB	None	6
(−)	(−)	(−)	Abnormal not c/w TB	Any	6
(−)	(−)	(−)	Abnormal, c/w TB	None	4
(−)	(−)	(−)	Abnormal, c/w TB	Any	1
(−)	(−)	(+)	Abnormal, c/w TB	Any	1
(−)	(+)	(−)	Normal	None	1
(−)	(+)	(−)	Normal	Any	1
(−)	(+)	(+)	Abnormal, c/w TB	Any	2
(+)	(+)	(+)	Abnormal, c/w TB	Any	7

For participants requiring repeat testing due to a test initially producing an unsuccessful result, the final valid test result was included.

Abbreviations: (−), negative result; (+), positive result; c/w, consistent with; TB, tuberculosis; Xpert, Xpert MTB/RIF Ultra.

Regarding adolescents’ experiences of providing TS and sputum samples, the time spent to produce a sample, physical comfort level, and complexity of TS collection compared favorably to sputum collection (Table 5). As Xpert is not optimized to run on TSs, we also asked whether participants would find it annoying to provide a sample for repeat testing (ie, if the initial test failed); over 97% strongly disagreed or disagreed that they would find this annoying. TS sampling and sputum were found to be acceptable to most participants, although many more adolescents “strongly” agreed that TS collection was acceptable (P < .001).

Table 5.

Participant Experiences With Sampling and Sample Acceptability

Feature	Tongue Swab (N = 225)	Sputum (N = 225)	P Valuea
Time spent collecting sample was far too long
Strongly disagree	183 (81%)	95 (42%)	<.001
Disagree	33 (15%)	38 (17%)
Neutral	3 (1.3%)	8 (3.6%)
Agree	6 (2.7%)	69% (31%)
Strongly agree	0 (0%)	15 (6.7%)
Sample collection was very physically uncomfortable
Strongly disagree	203 (90%)	176 (78%)	<.001
Disagree	17 (7.6%)	23 (10%)
Neutral	2 (0.9%)	8 (3.6%)
Agree	1 (0.4%)	15 (6.7%)
Strongly agree	2 (0.9%)	3 (1.3%)
Providing sample made me feel very awkward
Strongly disagree	185 (82%)	169 (75%)	<.130
Disagree	20 (8.9%)	31 (14%)
Neutral	4 (1.8%)	5 (2.2%)
Agree	13 (5.8%)	15 (6.7%)
Strongly agree	3 (1.3%)	5 (2.2%)
Sample collection process was really complicated
Strongly disagree	200 (89%)	174 (77%)	<.001
Disagree	18 (8.0%)	25 (11%)
Neutral	4 (1.8%)	4 (1.8%)
Agree	2 (0.9%)	19 (8.0%)
Strongly agree	1 (0.4%)	4 (1.8%)
Repeat sampling would be very annoying to me
Strongly disagree	187 (83%)	158 (70%)	<.001
Disagree	24 (11%)	28 (12%)
Neutral	8 (3.6%)	11 (4.9%)
Agree	4 (1.8%)	21 (9.3%)
Strongly agree	2 (0.9%)	7 (3.1%)
Overall, providing this sample was acceptable to me
Strongly disagree	6 (2.7%)	6 (2.7%)	<.001
Disagree	4 (1.8%)	11 (4.9%)
Neutral	1 (0.4%)	10 (4.4%)
Agree	53 (24%)	82 (36%)
Strongly agree	161 (72%)	116 (52%)

Data are presented as No. (%).

^aWilcoxon rank-sum tests were used to calculate P values.

We completed semi-structured interviews with 3 laboratory technicians (see Supplementary Material 1). Overall, TS collection and testing on Xpert was highly acceptable to technicians. They also perceived TS collection as acceptable to sampled adolescents, who could readily understand the procedure and sit still for its duration. Some adolescents seemed “slightly shy” or perhaps “queasy,” but the process was perceived as being easier for participants than obtaining a sputum sample. As one technician noted: “after tongue swab, patients complain about difficulty expectorating sputum when doing [sic] sputum next, but no complaints about tongue swab.”

There were logistical challenges to untangle regarding transporting and testing samples within 24 hours, particularly when there were high volumes of TSs for testing from contact screening events. The technicians suggested that developing a method to preserve TSs for over 24 hours would be welcome. Some concern was expressed regarding infection risk to the swab collector, and technicians felt that the risk to them of being infected by Mtb during TS collection was higher than when collecting sputum. Using a droplet barrier, standing to the side of the participant while swabbing, wearing personal protective equipment (gown, face mask, gloves), and working in a well-ventilated environment were felt to mitigate, if not obviate, risks.

Preparing the TS sample for Xpert testing was found to be slightly more complex than sputum preparation, and the risk of contamination during TS manipulation was seen as lower than with sputum. One technician suggested that a premade buffer at the correct dilution would be welcome and quicker, and suggested setting up clear processes for managing Xpert cartridges of different samples (eg, using separate trays, batching).

DISCUSSION

In our consecutively recruited population of adolescents recommended to undergo TB investigation, TS Xpert had low sensitivity, with excellent specificity. Additionally, TS sampling was found to be highly acceptable to participants and laboratory staff.

Molecular testing for TB using TSs is a major advance in the fight to end TB and find the “missing millions,” as TSs could be collected from, essentially, all individuals, even if other respiratory samples are unavailable. Although we did not observe any participants with positive TS Xpert results who could not produce sputum, larger or higher TB prevalence studies may provide insight in the rate at which TS molecular testing would be positive in the absence of sputum. We observed lower sensitivity with TS Xpert than has been reported previously [19, 20], which may be partially attributable to the sample processing methods. As Xpert was not designed to run with TSs, research is ongoing to determine the optimal sample preparation protocol with the lowest limit of detection, and it is possible that a different dilution of reagent-buffer solution may have improved sensitivity [21]. It also may reflect a possible spectrum effect in our population, as less than half of participants were reportedly symptomatic. Additionally, our study population's median age was notably lower than that of most other published studies [8, 19, 20], and participants may have had correspondingly lower Mtb bacillary loads.

The low sensitivity of TS Xpert we observed demonstrates the importance of careful evaluation among subpopulations who need better-performing TB tests. New samples and assays must be rigorously assessed in young people [9] as well as other key groups (eg, people living with human immunodeficiency virus) [22], to ensure they will work as hoped. This is because the actual load of Mtb impacts test positivity [23], so an “accessible” sample alone may be insufficient to improve diagnosis in groups with smear-negative or paucibacillary disease [24]. The understandable tendency to chase new technological “silver bullets” [25] once available is probably not the ultimate cure for the persistent TB case detection gap. This trend was previously seen during the coronavirus disease 2019 (COVID-19) pandemic when major stakeholders recommended bidirectional testing for TB and COVID-19 [26, 27] without evidence; subsequently, prospective studies [28, 29] showed this to be an inefficient approach. Similarly, broadening the array of testable samples is an important step, but further evidence is needed to determine how to best incorporate novel samples, including TSs, into routine care pathways for adolescents [30].

Our findings on patient acceptability are broadly aligned with results of another multicountry survey, primarily among adults [31]. There, a majority (61%) of participants indicated a preference for TSs over sputum. Although we did not query head-to-head whether our participants preferred one sample over another, a larger share of adolescents “strongly agreed” that TSs were acceptable to them than sputum. In a qualitative study interviewing caregivers, healthcare workers, and NTP stakeholder regarding oral swab testing among young children (<5 years), swabs were considered highly acceptable. Swabs were viewed favorably as they were not invasive, painful, or time-consuming to collect [32]. Our participants also expressed an appreciation for the simplicity and short time required for TS collection, which, coupled with their high accessibility, may bode well for their deployment in screening. Future focused work is needed for this test use-case.

Our study had some notable strengths. To our knowledge, this is the first study describing performance of TS TB testing on any molecular platform specifically in adolescents. Designing our study such that only a single participant visit was required permitted excellent completion of the patient experience and acceptability survey. By utilizing a CRS comprised of both molecular and bacteriological tests, we can be reasonably confident in participants’ disease classification; however, residual misclassification, particularly false negatives, due to the imperfect nature of TB reference standards is possible. Our findings are likely generalizable to other high-TB-incidence settings, as Vietnam is classified as a country with a high burden of TB [33] and the population is relevant to those in whom TB testing is typically performed in countries with high TB burdens.

There were also limitations. We observed a relatively low event rate, resulting in low precision for sensitivity estimates; we acknowledge that TB prevalence was low, yielding wide CIs. This study is just beginning to fill a knowledge gap, and diagnostic accuracy estimates should be interpreted as preliminary. We have reported other salient outcomes, but more diagnostic evaluation in this group is necessary. Additionally, the clinical presentation of our study population was heterogenous, with approximately 50% presenting with symptoms. Among inpatients or a study population where a larger proportion of adolescents had presumptive TB, TS Xpert sensitivity may be higher, so our diagnostic accuracy findings may not be completely generalizable to all settings. However, our population does represent adolescents who would be recommended for TB investigation by most NTPs in countries with high TB burdens.

In a population of adolescents recommended for TB investigation, participant experiences were positive and sampling acceptability was high. TS Xpert sensitivity was relatively low and specificity was excellent, with TS Xpert mainly detecting cases of TB with relatively high bacillary load. Among those who could not produce mucopurulent sputum, sensitivity was poor. Further work should be undertaken to improve TS molecular testing in adolescents, including potentially as part of diagnostic algorithms.

Supplementary Data

Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.