Ethics statement

Interviews were conducted according to the principles expressed in the Declaration of Helsinki with written informed consent obtained from all participants. Approval was obtained from all necessary ethical bodies including the Baylor College of Medicine Children’s Foundation Eswatini (IORG0006978) and the Eswatini National Human Health Research Review Board, Baylor College of Medicine Institutional Review Board (FWA-00000286), Houston, Texas, USA. All participants gave written or verbal consent.

Study design

This was a qualitative study investigating clinicians’ KAP of TS on sputum and stool for DR-TB diagnosis in Eswatini. Data was collected using semi-structured interviews (SSIs) and analyzed using directed thematic content analysis. This study followed the COREQ checklist for reporting qualitative research (S1 Checklist).

Context

Eswatini is a low-resource setting with a GDP per capita of US$3,987 [26]. Nonetheless, the country’s spending on health is relatively high compared to other LMICs. In 2020, 6.5% of GDP was spent on healthcare, which is above the average spent by LMICs (5.6%), but almost half the amount spent on average by HICs (14%) [27]. Regardless, the country continues to have sub-optimal health outcomes, in part due to persistent poverty and inequality. In 2024, 52.1% of the population lived below the LMIC poverty line [28] and the most recent Gini index in 2016 was 54.6 [26].

In 2022, Eswatini had a high incidence of TB estimated at 325 (95% CI 215–502) cases per 100,000; and a moderate incidence of DR-TB at 19 (95% CI 9.2-28) per 100,000 [1]. Additionally, HIV prevalence in adults was among the highest in the world at 27.4% (27,28). Consequently, Eswatini had a high incidence of TB-HIV coinfection estimated to be 187 (95% CI 104–294) cases per 100,000 [1]. Despite the high estimated TB incidence, only 61% of this total were diagnosed with TB. Limited access to healthcare may contribute to this diagnostic gap because 75% of the population lives in rural areas [29]. To improve access to TB screening in rural areas, the Eswatini Ministry of Health has mobilized “TB Community Champions” to screen for TB.

If an individual is presumed to have TB by a healthcare worker, then specimens are collected and sent via the National Specimen Transport System for testing at one of 35 decentralized laboratory sites throughout the country. TB and RR-TB are primarily diagnosed by Xpert MTB/RIF Ultra (Ultra). If M. tuberculosis DNA is detected, then specimens are sent to the National TB Reference Laboratory (NTRL) where M. tuberculosis culture, pDST, and line probe assay (LPA) are performed. Although Ultra detects M. tuberculosis DNA and mutations in the rpoB gene associated with rifampicin resistance, the majority of rpoB mutations in Eswatini are not detected by Ultra, Mycobacterial growth indicator tube (MGIT) culture or line probe assay (LPA) [30–32].

To address this DR-TB diagnostic gap, in 2019 Eswatini’s NTCP integrated TS into the diagnostic algorithm for DR-TB. This pilot program began prior to the updated WHO recommendations about using TS to diagnose DR-TB. To support the implementation of sequencing, the Global Health Protection Program through the German Ministry of Health funded the purchase of two iSeq 100 Illumina Next Generation Sequencing machines, reagents, and the training of laboratory personnel at the National Tuberculosis Reference Lab (NTRL). With support from the Research Center in Borstel and Baylor Foundation, laboratory personnel were trained at the NTRL in 2021 and in 2022. Although the COVID-19 pandemic delayed the procurement of many supplies and the launch of the program, TS was functional at the NTRL in 2021. The Eswatini Ministry of Health implemented the use of TS at all 14 DR-TB sites in Eswatini using a centralized model for laboratory testing. In March 2022, the NTCP created a Clinical Advisory Committee (CAC) comprised of expert clinicians, laboratory personnel and public health officials, to guide clinicians’ use of TS results. As part of the CAC mandate, clinicians were offered training on the use of TS results. The implementation of TS was overseen by the technical advisor for the national Programmatic Management of Drug-Resistant Tuberculosis (PMDT) committee.

In this context of DR-TB diagnosis, Eswatini’s NTCP delivers DR-TB care in 14 decentralized treatment sites located across 4 regions of the country (Shiselweni, Manzini, Hhohho, Lubombo). Each region has a designated DR-TB doctor to support the clinicians at the treatment sites. In the Manzini region, there are four DR-TB sites; in Hhohho, there are four sites; in Shiselweni, there are three sites; and in Lubombo, there are three sites (Fig 1).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. DR-TB care sites in Eswatini.

Blue circles represent DS-TB sites, while large red circles represent DR-TB sites. Black circles identify the National TB Reference Lab (NTRL) and the National TB Control Program (NTCP). Outer circles indicate the geography classifications urban, peri-urban, and rural. DR-TB sites are labeled in the legend. Interviews were performed at all sites except those in italics in the legend.

https://doi.org/10.1371/journal.pgph.0004718.g001

Sampling and recruitment

Nationally representative purposive sampling was used to select DR-TB providers interviewed for this study. Our intent was to interview clinicians practicing in each region in Eswatini, inclusive of rural and urban environments, and primary and secondary health centers. Physicians and nurses working with DR-TB patients at any of the 14 DR-TB sites were eligible to participate. At a minimum, we aimed to interview one physician and one nurse per region in Eswatini (eight healthcare providers total); at a maximum, we aimed to interview one physician and one nurse at each DR-TB treatment site in Eswatini (21 healthcare providers; seven sites did not have a permanent doctor or nurse). We excluded physicians who were on leave at the time of the study.

Prior to implementation, we discussed and developed the study with the NTCP. We contacted regional DR-TB coordinators to set up interviews with healthcare providers. Regional coordinators, along with experienced clinician-scientists (DV and BB), connected the interviewer (MM) to the DRTB coordinator at the NTCP (a former DRTB care provider), and a total of 20 clinicians (9 physicians and 11 nurses) from all four regions. MM contacted all 21 potential interlocutors. The relations between the researchers and the study population are outlined in S2 Checklist. In obtaining consent, the study details and the researchers’ roles were described to the potential participants.

Data collection

We conducted 15 semi-structured interviews with the NTCP coordinator, eight doctors and eight nurses (in two interviews, both a doctor and a nurse were present). Recruitment occurred between January 29, 2024, and March 22, 2024. Written informed consent was obtained from all participants before participating in and recording the interviews. All interviews were conducted in English for 30 minutes by one researcher (MM) using an SSI guide (S1 Text). Interviews were conducted over encrypted video calls or in a private and quiet location at the participant’s place of work. No repeat interviews were conducted. SSIs stopped once data saturation was reached. Notes were taken during the interview and interviews were transcribed verbatim. Transcripts were not returned to participants. After the transcription of the interviews, all audio files were destroyed, and de-identified transcripts were stored on a secure Baylor College of Medicine server. Names were removed from transcripts prior to analysis.

Analysis

The analysis followed a thematic analysis approach, which is a systematic classification process defined by coding and identifying themes or patterns [33]. The software Taguette 1.4.1-50-geed050b [34] was used to organize and code the transcripts. The codebook contained several initial codes based on the interview topic guides, including KAP of DR-TB diagnosis, TS, and TS on stool. During the preliminary coding process, emergent themes were identified, added to the codebook, and included in the analysis. A second independent coder (AK) coded the transcripts. Codes were compared and any discrepancies were resolved. After the codebook was finalized, the transcripts were re-coded line by line. Coded transcripts were exported from Taguette into Excel. The coded transcripts were further coded for sub-themes and analyzed by MM using Excel (S1 Data). Maps were generated using the R package leaflet (version 1.9.4), and the qualitative schematic and network diagram were created in Adobe Illustrator.

Sociodemographic

Out of 21 potential interlocutors, 18 responded and three nurses did not respond. All clinicians that could be reached consented to being interviewed. One doctor consented to the interview but did not participate due to scheduling around clinical work. In the Hhohho region, 100% (3/3) of doctors and 100% (4/4) of nurses agreed to be interviewed. In the Manzini region, 50% (1/2) of doctors and 50% (1/2) of nurses agreed to be interviewed. In the Shiselweni region, 100% (3/3) of doctors and 100% (2/2) of nurses agreed to be interviewed. In the Lubombo region, 100% (1/1) of doctors and 50% (1/2) of nurses agreed to be interviewed. Ultimately, 16 healthcare workers from 12 DR-TB care sites, and the NTCP coordinator were interviewed (Fig 1). The Hhohho and Shiselweni regions are overrepresented in the study because there are more DR-TB sites in these regions.

Participants were classified based on the geography of their primary clinic, their specialty training or experience, their involvement in research, and their experience directing or developing healthcare programs in the country (Table 1). The clinical experience among the providers interviewed ranged from less than 10 months to over 13 years. There were 10 female healthcare providers and 7 male providers. All the nurses interviewed were self-identified as TB nurses. All the doctors were caring for TB patients, but some had additional specialized training (Table 1). A greater proportion of doctors had research experience compared to nurses (Table 1). We interviewed an equal number of doctors from rural, urban, and peri-urban areas, while a greater proportion of the nurses interviewed were from rural clinics. Three doctors had experience managing DR-TB programs in the country (Table 1).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Participant characteristics.

https://doi.org/10.1371/journal.pgph.0004718.t001

Themes

During the interviews, several themes emerged regarding the implementation of TS on sputum. These themes include the importance of a well-functioning lab, the need for decentralized services, varying levels of knowledge about TS, challenges facing the result alert system, frustration with workload and delayed results, variable attitudes about when to use TS, the benefits of the Clinical Advisory Committee, and concerns about sustainability. Further, the themes applying to stool specimens included a moderate amount of knowledge of stool as a specimen for TB diagnosis, mixed perceptions of stool as a specimen, and questioning attitudes towards the feasibility of stool as a specimen. Major themes and selected illustrative quotes are organized in Table 2.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Quotes illustrating themes.

https://doi.org/10.1371/journal.pgph.0004718.t002

Synthesizing these themes and the sociodemographic information, we found that healthcare providers’ clinical, research, and programmatic experiences influence their knowledge of DR-TB diagnosis; healthcare providers perceive challenges in diagnosing DR-TB; operational challenges influence healthcare providers’ knowledge, attitudes, and perceptions of TS; and providers are open to implementing TS on stool to diagnose DR-TB if more information is disseminated to them and their communities. These findings are discussed in further detail below.

Influence of clinical, research, and programmatic experience on healthcare providers’ knowledge of DR-TB diagnosis

Healthcare providers’ clinical, research, and programmatic experience influenced their knowledge of DR-TB diagnosis. All providers were familiar with RR/MDR/XDR-TB and were highly knowledgeable about their clinical roles in treating and diagnosing these forms of DR-TB. However, providers with more experience caring for DR-TB patients accordingly had greater knowledge of the specific tests and workflow involved in DR-TB diagnosis. One doctor with extensive experience described the diagnostic pathway as follows:

“All presumptive TB patients who come to our clinic, they give two sputums, one for GeneXpert and one for culture and LPA. So, if you do a GeneXpert, we are using a GeneXpert result saying they are resistant as a proxy to MDR. And sometimes, if you are lucky enough, you get LPA results, which is rare, but maybe after a month, you get culture results which can also confirm INH resistance… We also put them in our genotypic registration. Then, we also have those ones who we diagnose presumptive. Maybe they are close contacts to our current DRTB patients” (P11).

This provider, like several others, had over 5 years of experience treating patients with DR-TB and was involved in quality improvement research. While this doctor described the algorithm in detail, another with nurse with less training and fewer years of experience described the algorithm as follows: “we know from the plan that [when someone] tests positive for TB… we have to run to see if it’s a DR, if it’s resistant to one drug, then we move on to test for other drugs, but it’s clearly outlined” (P12). The different levels of detail in these descriptions highlight discrepancies in providers’ levels of knowledge of the specific tests involved in DR-TB diagnosis.

Additionally, providers described their patients as predominantly adult males living with HIV. Few had experience treating children with TB. As a result, most providers were more knowledgeable about the challenges facing adults and PLHIV with DR-TB than children. Further, clinicians involved in research and directing programs had a greater knowledge of trends in DR-TB diagnosis and TS in the country than clinicians who only focused on providing care. Most clinicians without research experience were aware of ongoing research on DRTB and TS in the country but didn’t have knowledge of specific research activities.

Healthcare providers perceive challenges in diagnosing DR-TB

Many providers recognized GeneXpert as a vital tool for TB diagnosis, but a limited one for DST. For example, one doctor with research experience noted: “Sometimes GeneXpert misses the Rifampicin-resistance pattern according to the mutation, I think I491F, something like that? – the lab can explain to you very well about that” (P9). While not all clinicians mentioned the Isoleucine 491 Phenylalanine (I491F) mutation, which is associated with rifampicin resistance and is not detected by GeneXpert, LPA, or MGIT, most clinicians highlighted that GeneXpert was limited by the fact that it detected “RIF-resistance only” (P1, P2, P5, P7, P8, P11, P13).

Beyond the limitations of diagnostic tools, healthcare providers underscored the pivotal role of a well-functioning laboratory. One provider noted, that “we used to say that, especially for DRTB, the lab is, in fact, the head and shoulder of DRTB care” (P1). For example, if the lab doesn’t send results to clinicians in a timely fashion, then patient care suffers. All providers emphasized that even if GeneXpert, culture, or LPA results are accurate, they take too long, likely due to human and material resource shortages, leaving the clinician to treat the patient empirically. A nurse noted that this was a common occurrence: “I don’t know if they will improve the turnaround time for the results, because sometimes you find that you have a client, but you can’t really start the treatment, because you don’t have the results, but you can see the client, and maybe he’s not responding well to the treatment” (P12). Healthcare providers exhibited varied attitudes toward updating empiric treatment based on delayed GeneXpert, culture, and LPA results. Some said they adopt a precautionary approach by treating the result that presents the “worst-case scenario,” while others said they rely more on clinical judgment to manage patients: “we will consider the clinical [factors], as I told you, we consider the patient exposure, the kind of exposure, the time of exposure, the clinical examination, chest X ray, and then we associate with other tests. Then I take my decision” (P9).

Additionally, clinicians suggested that geography presented additional barriers to DR-TB diagnosis. Some clinicians highlighted that the prior physical separation of DR-TB and DS-TB services may have exacerbated stigma and discouraged individuals from seeking timely treatment. One doctor noted that the initial centralization of DRTB care led to “issues with stigma, because we had a hospital setting which was like [only] for DRTB patients. So, we need[ed] to decentralize the services for access… We wanted to integrate it with drug-susceptible TB, but it felt like people [healthcare workers and the community] wanted them to be on their own” (P1). Furthermore, providers in rural areas emphasized that a lack of resources and the transport of specimens from rural clinics to a centralized laboratory can lead to delays in result reporting that impact their ability to provide the most effective treatment.

Despite this challenge, clinicians serving rural areas stressed how decentralizing the DR-TB program has made services more accessible. An unintended consequence of this decentralization has been to increase the independence of physicians at distant sites. Consequently, one physician with programmatic experience perceived clinicians in Eswatini as highly competent and commended their ability to manage DR-TB cases. Nonetheless, most healthcare providers perceived obstacles to DR-TB management that their clinical skills could not overcome, necessitating technical and systems-level change.

Influence of operational challenges on healthcare providers’ knowledge, attitudes, and perceptions of TS

Since sputum-based TS was introduced in Eswatini, operational challenges have shaped providers’ knowledge, attitudes, and perceptions of each step in the implementation pathway (Fig 2). First, providers have gained varying levels of knowledge, mostly related to its clinical applications. For example, one doctor (P9) stated that “I can just give you the clinical part. That’s the only thing that I can explain about the sequence thing” and another nurse described “what I understand is that with sequencing, it’s a further step which is more detailed compared to the GeneXpert, the line probe assay, and the culture and DST. It gives more details about the resistance.” While nearly all the healthcare workers described their knowledge of TS in terms of its utility as a diagnostic tool, one nurse described his knowledge of why TS is a more effective diagnostic: “what I know is that there’s whole genome sequencing where we use the entire genome for the TB… the DNA of the microbe, where we’re looking at resistant patterns throughout the entire genome in order to identify any mutations rather than being restricted to certain regions where in this country [there are] certain mutations” (P10). This nurse was referring to the I491F mutation, which is prevalent in Eswatini and not detected by traditional pDST or any tests completed on the GeneXpert platform. Similarly, other clinicians mentioned that TS is more effective than Xpert Ultra or phenotypic liquid DST in identifying this resistance mutation, without naming the specific amino acids.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Implementation of TS.

The schematic outlines perceived challenges and successes of sputum-based TS implementation. Numbers indicate areas where healthcare workers perceived strengths (green) and weaknesses (red) in the implementation pathway. 1) Case detection is limited in communities due to stigma and centralized DRTB care sites. 2) Delayed arrival of sputum specimens at the National TB Reference Lab due to difficult specimen collection and problems with transportation. 3) Delayed and inconsistent results generation due to limited reagents and laboratory personnel. 4) Delayed action on results due to a lack of a laboratory alert system. 5) Increased workload and frustration due to paperwork and multiple regimen changes. 6) Improved patient care because of clinical guidance from the clinical advisory committee. 7) Improved patient care because of more accurate resistance predictions and improved turnaround times compared to traditional modalities. 8) Need for more information to be disseminated by the National TB Control Program.

https://doi.org/10.1371/journal.pgph.0004718.g002

Although we didn’t examine healthcare providers’ knowledge of TS before its introduction in Eswatini, clinicians state that they have gained knowledge through NTCP workshops and experience with using TS clinically. Seven of the interviewed healthcare workers mentioned attending a “workshop” to learn about TS for DR-TB diagnosis. One doctor described that they learned about sequencing at the workshop, but that “you can’t integrate all those things in one day or two days” and the learning is “about what you call actually progressive training, progressive learning”(P15). That progressive learning seems to have come through practical clinical experience. The program director at the NTCP noted that they had seen an improvement in knowledge and the use of results from clinicians over the course of the last two years: “now we feel like [physicians] are forthcoming. [They will come to us and say]—“we have a client who’s not doing well—could it be a mutation that we are not picking up?” …[Physicians] can now confidently talk about sequencing” (P1). Providers’ exposure to and use of TS has increased their knowledge of TS which has increased their comfort in using the technology (Fig 3a).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Network diagram of clinicians’ knowledge, attitudes, and perceptions of using TS for DRTB diagnosis.

A) sputum-based TS. B) stool-based TS. Light grey nodes represent knowledge held by clinicians, while dark grey nodes represent a lack of knowledge. Light green nodes and edges represent positive perceptions, while light pink nodes and edges represent negative perceptions. Dark green and maroon edges and nodes represent attitudes shaped by positive or negative perceptions.

https://doi.org/10.1371/journal.pgph.0004718.g003

Although providers’ experiences have increased their knowledge, their perceptions of TS have been mixed due to operational challenges (Fig 2). Most providers described result delivery as inconsistent because of machine failures, human error, shortages of resources, and unreliable result delivery. For example, providers mentioned that effective treatment decisions are often delayed because results are sent by email and there is no standardized result alert system. A nurse said “if you don’t check your emails here and there, you may not know whether you’ve received [the results]... So, the patient is almost, it’s actually done [with] treatment, it’s only then that I’m receiving sequencing results” (P10). In this system, it’s easy to miss results that could inform more effective treatment. One clinician said that recently the laboratory addressed this problem by calling physicians in advance of sending the results. Despite this isolated report of improvement, most clinicians attributed patient frustration with treatment regimen changes, and clinician frustration with an increase in their workload to delayed TS result reporting. For example, one doctor remarked: “it’s just frustrating on the patient side, because now they’ve been changing treatment about three times now.”

Due to the operational challenges limiting the use of TS, some clinicians have limited experience with TS. Clinicians’ amount of experience with the technology and their perception of resource availability appear to inform their comfort with the tool and confidence in the results. While providers had received some information about TS, most did not consider themselves to have “specialized training.” Consequently, clinicians expressed different attitudes about when to use TS to inform patient care, from using it early and often to as a last resort. One doctor believed that TS should be used for all patients— “you[‘d] just have a result that is pointing to the possible mutations that are available and you tailor-make the regimens” (P11). At the other end of the spectrum, two clinicians thought that TS should only be used as a last resort when patients don’t respond well to treatment— “we must make sure that we are really stuck” (P15), or “make it a last resort, but use it” (P12). Between these attitudes, the majority of clinicians believed that TS should be used as a second test after rifampicin resistance is detected by GeneXpert (P6, 9, 10, 14) (Fig 3A).

Further, when TS identified additional resistance compared to GeneXpert, LPA, or MGIT, clinicians with limited TS experience preferred to rely on clinical judgment, while those with more experience used TS results. Despite differences in opinion about when to use TS, most clinicians found TS results easy to interpret and appreciated the expertise of the clinical advisory committee when they were unsure about results. One doctor noted: “even though we might have, you know, those personal feelings that the patient may be responding to the previous regimen… we do practice what the committee suggests, because we do think the panel is well experienced and well informed about everything. We consider genome sequencing results.” Providers unanimously highlighted the clinical advisory committee as improving DR-TB care (Fig 2).

Although obstacles have challenged the implementation of TS, most providers commented that TS improves DR-TB diagnosis. Clinicians noted improvements in the turnaround time of results compared to pDST and improved resistance predictions. They emphasized that these improvements in DR-TB care are worth the added work. Accordingly, providers were open to the expansion of TS from a pilot to an official program, but they wanted to ensure its sustainability, by evaluating its cost-effectiveness, improving the timeliness of results, and training more clinicians and laboratory personnel (Fig 2).

Providers are open to implementing TS on stool to diagnose DR-TB if more information is disseminated to them and their communities

Although most providers did not have experience diagnosing TB in children, they were aware of the use of stool as a specimen for diagnosing TB in children (Table 2B; Fig 3B). Notably, providers perceived stool as an essential specimen for TB diagnosis in children, providing an alternative where traditional specimen types may fail, and offering quicker results due to easier collection. While providers knew less about TS on stool, they perceived it as a promising tool to increase drug resistance detection using this specimen type (Table 2B).

Providers anticipated potential confusion and resistance among patients regarding stool-based diagnosis due to traditional beliefs about stool testing (Table 2B). Similarly, there were mixed perceptions among providers about their peers’ acceptance of stool-based testing for DR-TB. Some believed that clinicians would appreciate the introduction of this specimen to reduce invasive testing, while others questioned healthcare providers’ willingness to collect stool (Table 2B).

In addition to disagreement about the acceptability of using stool, providers felt there was insufficient information about the feasibility of TS on stool. While clinicians deferred the assessment of technical feasibility to laboratory personnel, they were concerned about the operational feasibility of implementing such a test given the challenges they identified with the current TS implementation (Table 2B).

Despite these concerns, most providers were open to the use of stool for TS. Providers noted the potential benefits of stool-based testing for pediatric DR-TB diagnosis and expressed interest in expanding its use to diagnose adults. Nonetheless, they mentioned a desire for more information and comprehensive training before implementation (Fig 3B).