Preferences of health care workers using tongue swabs for tuberculosis diagnosis during COVID-19

Healthcare workers (HCW) who come into contact with tuberculosis (TB) patients are at elevated risk of TB infection and disease. The collection and handling of sputum samples for TB diagnosis poses exposure risks to HCW, particularly in settings where aerosol containment is limited. An alternative sample collection method, tongue swabbing, was designed to help mitigate this risk, and is under evaluation in multiple settings. This study assessed risk perceptions among South African HCW who used tongue swabbing in TB diagnostic research during the COVID-19 pandemic. We characterized their context-specific preferences as well as the facilitators and barriers of tongue swab use in clinical and community settings. Participants (n=18) were HCW with experience using experimental tongue swabbing methods at the South African Tuberculosis Vaccine Initiative (SATVI). We used key informant semi-structured interviews to assess attitudes toward two tongue swab strategies: Provider-collected swabbing (PS) and supervised self-swabbing (SSS). Responses from these interviews were analyzed by rapid qualitative analysis and thematic analysis methods. Facilitators included aversion to sputum (PS and SSS), perceived safety of the method (SSS), and educational resources to train patients (SSS). Barriers included cultural stigmas, as well as personal security and control of their work environment when collecting swabs in community settings. COVID-19 risk perception was a significant barrier to the PS method. Motivators for HCW use of tongue swabbing differed substantially by use case, and whether the HCW has the authority and agency to implement safety precautions in specific settings. These findings point to a need for contextually specific educational resources to enhance safety of and adherence to the SSS collection method.

INTRODUCTION 48 TB disease, caused by M. tuberculosis (MTB), remains a major global cause of morbidity 49 and mortality. Collection of the standard sample for TB diagnosis, sputum, presents safety risks 50 to health personnel. Moreover, the process is difficult for many types of patients, and sputum 51 testing is insensitive for certain types of TB. The availability of alternative, noninvasive 52 sampling methods, which can easily be collected outside of the clinic, would improve worker 53 safety, increase the efficiency of testing, and allow more active TB case finding in community 54 settings. [1-3] 55 We and others have shown that MTB DNA is deposited on the oral epithelium during 56 active TB, and can be detected by oral swab analysis (OSA). [4][5][6][7][8][9][10][11][12][13][14][15] In OSA, the dorsum of the 57 tongue is brushed with a sterile swab. The swab head with collected material is deposited into a 58 tube for MTB DNA detection by nucleic acid amplification testing (NAAT). [8] In this paper we 59 focus on the two methods of sample collection using tongue swabs: HCW (provider)-collected 60 swab (PS), or HCW-supervised self-swabbing (SSS). 61 Tongue swabs were developed in part to reduce the occupational health risks associated 62 with sputum collection.
[11] However, there may be new risks associated with tongue swabbing. 63 In particular, the emergence of the SARS-CoV-2 virus created new potential threats to HCW 64 collecting patient oral samples, and may have changed their attitudes toward such 65 procedures. [16] In early evaluations of tongue swabbing, samples were collected by HCW or 66 study personnel positioned in front of patients' faces. This presented infectious disease exposure 67 risks that were amplified by COVID-19. Therefore, a SSS approach, initially evaluated for 68 COVID-19 tongue swab collection [6], was adapted for TB tongue swab collection. 69 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 6, 2022. ; https://doi.org/10.1101/2022.12.06.22283185 doi: medRxiv preprint sites around the world. [11,17] Although qualitative assessments of preferences related to other 71 aspects of TB diagnosis have been reported [18,19], there have been few such assessments of 72 tongue swabs. Such studies are needed to understand the HCWs experience with tongue swabs in 73 real world settings and how this impacts their willingness to use the new method for TB 74 diagnosis.
[20] 75 Implementation science is the scientific study of methods and strategies to increase the 76 uptake of innovative, evidence-informed practice. A fundamental challenge of implementation 77 science is identifying contextual determinants (e.g., barriers and facilitators) and determining 78 which implementation strategies will address them. [ Participants were referred to the study team by their site manager, and all referents who 103 met the inclusion criterion were approached by the study team by email to invite them to 104 participate. We continued recruitment until we observed saturation among meta themes, and 105 additional interviews were determined to be unlikely to illuminate major new insights. [23-26] 106 An interview guide was developed a priori and reviewed by the study team and local 107 investigators to ensure question clarity, local/cultural appropriateness, and robustness of the 108 interview guide to address all study objectives. The interview guide was informed by the 109 Extended Parallel Process Model (EPPM), [27] (Figure 1). EPPM is a risk communication 110 theory that posits that individuals take action to control danger when they perceive that the 111 severity and susceptibility are high and that they are competent to take mitigating action. We 112 explored perceived self-efficacy (confidence in ability, knowledge, and skills to perform the 113 method), perceived response efficacy (belief that the method works/works better than other 114 methods), perceived threat susceptibility (belief that the HCW themselves are susceptible to 115 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted December 6, 2022. ; https://doi.org/10.1101/2022.12.06.22283185 doi: medRxiv preprint hazard). We explored how these constructs interacted with HCW willingness to perform specific 117 sample collection methods in specific contexts. 118 All interviews were audio recorded. Because travel was restricted during the COVID-19 119 pandemic, interviews were conducted remotely over Zoom in English. Interview audio files were 120 professionally transcribed by TranscribeMe! (San Francisco, CA) and reviewed for accuracy. 121 Interview transcripts were read and re-read by the primary interviewer (RC). An interview 127 contact sheet was created for each interview, summarizing key takeaways related to each of the 128 study objectives. These contact sheets were shared back with participants for confirmation of 129 accuracy. This process, known as member checking, provides an opportunity to enhance 130 qualitative credibility.
[29] We then entered data from each of the interview contact sheets into a 131 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted December 6, 2022. ; https://doi.org/10.1101/2022.12.06.22283185 doi: medRxiv preprint matrix. The matrix was used to synthesize key takeaways across interviews to identify 132 commonalities and counterpoints. 133 Next, transcripts were coded and thematically analyzed. Deductive codes were 134 developed based on the interview guide and underlying theory (EPPM) and research objectives. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

156
Participant characteristics. Participants had from 5 to 20 years of experience collecting 157 samples for TB diagnosis. HCW participant demographic information is outlined in Table 1 Participants were also asked about their health history related to TB and COVID-19. Two of the 164 18 participants had a previous TB diagnosis and treatment. Both cases were reported as due to 165 occupational exposure. Three of the 18 participants had a previous COVID-19 diagnosis, with all 166 three cases reported as community exposure. 167 Lives and works in the same community 6 (33) Speaks more than one language at work 14 (78) Previous TB disease diagnosis as of April 2021; reported as occupationally acquired 2 (11) Previous COVID-19 diagnosis as of April 2021; reported as community acquired 3 (17) 169 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 6, 2022. ; https://doi.org/10.1101/2022.12.06.22283185 doi: medRxiv preprint CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 6, 2022. ; https://doi.org/10.1101/2022.12.06.22283185 doi: medRxiv preprint kind over sputum sample collection method. PS was the most preferred (9/15). Only 3 175 respondents preferred sputum sampling over tongue swabbing. Motivators of swabbing differed 176 substantially by use case, and whether the HCW has the authority and agency to implement 177 safety precautions in specific settings. Below, we discuss key facilitators and barriers to PS and 178 SSS use (Figure 2). 179

Facilitators 180
Aversion to sputum (PS and SSS). Almost all HCWs interviewed shared that they do 181 not like looking at and collecting sputum samples. HCWs must confirm the color and volume, 182 which was reported to be an unenjoyable process. They also shared that sputum can be messy if 183 it is not collected properly and if the jar is not closed properly. As one HCW who works in the 184 community and the clinic shared, "Sometimes the participants don't close the lid of the stuff on 185 the sputum jars and then when you get there, everything is out of the sputum jar. The swab is less 186 messy than sputum and you know that you get a sample that's not going to leak or something like 187 that." 188 HCWs who work with infants and children expressed population-specific challenges with 189 obtaining sputum. Standard methods for children include invasive and poorly tolerated 190 procedures such as gastric lavage. HCW expressed the need for another sampling method and 191 that they were interested in trying the PS method with infants and children. As one who works 192 with children in the clinic shared, "Inducing sputum production in kids is difficult for parents to 193 see their child go through and there is loss to follow up because parents don't want to put their 194 kids through that. It would be interesting to see if we can start to use swabs on children." 195 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 6, 2022. ; https://doi.org/10.1101/2022.12.06.22283185 doi: medRxiv preprint still during the sample collection process. 197 Perceived safety of the method (SSS). All participants expressed that they have the 198 knowledge to protect themselves from the risk of TB, and most also shared that they have the 199 knowledge and skills to prevent COVID-19 exposure at work. The majority of HCW interviewed 200 identified SSS as their preferred method of sample collection compared to both PS and sputum 201 sampling due to perceived safety. They reported perceptions of being able to keep a safe distance 202 while observing the patient to make sure they were performing the sample collection correctly. Overall, HCWs shared that they have comprehensive training for sputum and PS but that 231 they would like training on SSS particularly in response to the threats of COVID-19. Many 232 HCW expressed that they would like to be trained on how to facilitate SSS; they also shared 233 ideas for improving training using videos, animations, and photos to guide patients through SSS 234 regardless of literacy level or language spoken. As one HCW who conducts home visits and 235 works in the clinic shared, "I need to have videos. Nowadays, people are very lazy to read. So 236 many people...are now very digitized to their phones. So maybe a free website that will give some 237 free education pertaining to certain signs and symptoms and diseases." Another, who also works 238 in both contexts, went on to say, "Cartoons would be great to help guide the patient through 239 swabbing with an explanation, then patients could successfully self-swab." 240 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 6, 2022. ; https://doi.org/10.1101/2022.12.06.22283185 doi: medRxiv preprint technology access realities and language needs. Some shared that videos could be played in a 242 clinic waiting room, but the expectation should be that all training be conducted at the clinic as 243 many patients lack smartphones or data necessary to stream the videos. One HCW shared that 244 training materials, including videos, should be available in English, Afrikaans, and Xhosa. with early morning sample collection before the patient eats, brushes their teeth, or goes to work. 263 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 6, 2022. ; https://doi.org/10.1101/2022.12.06.22283185 doi: medRxiv preprint samples, stating that patients may be shy and concerned with their early morning breath. HCW 265 shared that collecting samples during the daytime and not before brushing their teeth in the 266 morning may reduce the barrier that the patients feel when opening their mouth before they get a 267 chance to brush their teeth so early in the morning. 268 Control of the work environment in community settings. HCW in clinic settings 269 reported that they had control of their environment to implement training to mitigate risk while 270 collecting samples, were treated professionally by patients, and had authority to direct patient 271 behavior in the clinic. However, HCW in home visit settings shared that they did not have the 272 same control of their environment and often did not have the authority to control their patients' 273 actions especially when related to others in the household. As one who worked in both contexts 274 shared, "If you're at a patient's home, their kids, they will come and touch you. So, we are in high 275 risk at the home of the participant." 276 Female HCW expressed concern with being treated more like an "auntie" (a family friend 277 who doesn't challenge the norms of the household) than as a professional. HCWs expressed 278 greatest concern with being able to keep their distance and abide by safety standards when in the 279 home. The majority of the participants who make home visits shared examples of situations 280 when they did not have the authority or agency to protect themselves, which impacts their 281 agency to respond to perceived threats; this was exacerbated by cultural stigma, patient 282 discomfort with sample collection in the early morning, and the context of COVID-19. HCW 283 struggled to balance collecting the samples safely to prevent TB or COVID-19 transmission and 284 being culturally respectful when entering someone else's home. One HCW shared that they 285 wanted uniforms for their work that were washed at work and not taken home, outdoor facilities 286 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 6, 2022. ; https://doi.org/10.1101/2022.12.06.22283185 doi: medRxiv preprint for patients to produce sputum samples, and larger clinic spaces to accommodate social 287 distancing. 288

COVID-19 risk perception. Although all HCWs stated that sputum collection presented 289
challenges and that tongue swabbing would be preferred, they expressed a variety of barriers to 290 tongue swabbing related to their perceived risk of COVID-19 while collecting samples. More 291 than half of the HCW expressed that the threat of COVID-19 exposure impacted their 292 willingness to stand in front of the patient to collect the samples (PS). Some HCW reported that 293 this changed their preference to SSS. For instance, one HCW who works only in the clinic 294 shared, "Self-swabbing with me there guiding at a distance would be better now with COVID." 295 Another who works both in the community and clinic shared similar sentiments, "Before COVID 296

I preferred to do the tongue swab but now with COVID I would prefer to do the SSS. It (SSS) is 297 also safer for the family members in the house… if it is done at the clinic then it is safer for the 298
other staff members like cleaners, security and register." 299 However, two HCWs who previously preferred PS reported that they preferred sputum 300 sampling over any self-swabbing methods in the context of COVID-19. As one HCW who 301 worked both in the community and the clinic shared, "For now, because of the pandemic, I 302 prefer the sputum over the (provider) swab. I don't want to be so close. But before the pandemic, 303 the (provider) swab was a good alternative to sputum." 304 A few participants described how they perceived that fear of COVID-19 and community 305 and occupation controls worked to reduce the spread of TB at work and in their community. The 306 COVID-19 pandemic has introduced new trainings to the curriculum such as social distancing 307 and reinforcement of use of PPE. During COVID-19, PPE has also become more readily 308 available, and adherence to PPE protocols has become stricter. Precautionary behaviors for TB 309 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 6, 2022. ; https://doi.org/10.1101/2022.12.06.22283185 doi: medRxiv preprint have had higher adherence due to fear of contracting COVID-19. In addition, the novelty of the 310 hazard of COVID-19 influenced their risk perceptions. HCW were more concerned with 311 contracting COVID-19 than TB at work. As one HCW who worked exclusively in the clinic 312 This study explored HCW willingness to use tongue swabs, especially with regard to 329 perceptions of risk of occupational exposure to TB and COVID-19. Some facilitators and 330 barriers reported by HCWs were germane to tongue swabbing in general, while others were 331 specific to different methods (PS or SSS). For both PS and SSS, aversion to sputum was 332 reported to support HCW willingness to use the methods. For SSS specifically, perceived safety 333 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 6, 2022. perceptions that SSS was safer for the HCWs themselves. Perceived patient self-efficacy has 347 been reported to influence HCW willingness to use other types of diagnostics or put themselves 348 at risk. [31] Accordingly, HCWs pointed to the need for educational materials for patients, 349 including illustrations and videos, to improve their capability to perform SSS correctly. This 350 solution has the potential to both improve implementation of SSS and support HCW willingness 351 to perform a safer and preferred method of sample collection (PS). As described by our study 352 participants, these materials should be made in multiple languages and should be available in the 353 clinic, as the patient population of interest may lack the technology and resources to review them 354 beforehand. HCW could be provided with digital educational materials (and airtime/data) to 355 show participants in their homes. Although this was not specifically suggested by HCW, it would 356 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 6, 2022. Our study had limitations. It was conducted in a single site in South Africa, which may 368 limit the generalizability. However, as noted above, this deep context-specific inquiry showcased 369 the importance of local context in the determination of facilitators and barriers to HCW 370 willingness to use tongue swabbing. Moreover, as participants were sampled from one of the first 371 sites with considerable experience using this approach, this allowed for early formative research 372 on implementation. Additional site-specific qualitative inquiries are thus recommended. 373 Qualitative interviews were performed over Zoom by a U.S.-based researcher (RC) in 374 participants' non-primary language, introducing opportunities for misinterpretation of both 375 questions and answers. To promote qualitative trustworthiness, we introduced a number of 376 credibility checks. First, our study team included in-country collaborators, who reviewed our 377 study protocols and interview guide to ensure question clarity and local and cultural 378 appropriateness. Interviews were recorded and professionally transcribed, and professional 379 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 6, 2022. ; https://doi.org/10.1101/2022.12.06.22283185 doi: medRxiv preprint capture. We also employed member checking [29], wherein we provided interviewees with a 381 summary of our key takeaways from their interview, aligned with our research objectives. 382 Participants were offered the opportunity to review and provide feedback on this interpretation. 383 In conclusion, interviews with 18 HCWs in South Africa supported the acceptability of 384 PS/SSS as a promising method for collecting TB diagnostic samples from patients while 385 reducing occupational risks. Interviewed HCWs preferred PS/SSS over sputum sampling, 386 attributed to their aversion to sputum and perceived relative safety of tongue swabbing, with SSS 387 being perceived as the safer option. However, some HCWs described that they were deterred 388 from using the preferred method (SSS) because of their lack of confidence in patients' ability to 389 collect the sample correctly. Accessible educational resources, both in the terms of language and 390 technology access, targeted at patients could thus improve SSS uptake and HCW occupational 391 health and safety. Disease-associated stigma, lack of control over the workplace environment, 392 and COVID-19 risk were also identified as barriers to tongue swab sampling. These findings 393 support the need for formative, context-specific implementation research in parallel with the 394 rollout of novel diagnostic sampling approaches. 395

ACKNOWLEDGMENTS 397
We are grateful to the HCWs who agreed to be interviewed for this study as well as the 398 patients, communities and clinical teams in Western Cape for their longstanding participation in 399 TB research. We are specifically grateful for the contributions of Danelle Van As, SATVI study 400 coordinator and research nurse who was essential for data collection, site management and 401 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted December 6, 2022. ; https://doi.org/10.1101/2022.12.06.22283185 doi: medRxiv preprint operations which facilitated this work. This work was supported by the Bill and Melinda Gates 402 Foundation (INV-004527, OPP 1213054), and by NIH grant U54EB027049. 403 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted December 6, 2022. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted December 6, 2022. ; https://doi.org/10.1101/2022.12.06.22283185 doi: medRxiv preprint