All participants provided written informed consent (including permission to audio-record the sessions for transcription and coding purposes) in English or Luganda as they preferred and were informed of their freedom to withdraw at any time during the interview. The institutional review boards approved the study at the University of Georgia, Office of Research (STUDY00004974), and Makerere University Higher Degrees, Research and Ethics Committee, in Uganda (Protocol #562). All participants were offered an equivalent of US $5.00 in Uganda Shillings at the end of the interviews as a refund for travel expenses and compensation for their time.

There is considerable evidence suggesting that theories of behavior change are useful for informing the effective design, uptake, and adoption of new health interventions by health care workers and patients for the management of TB [31]. Generally, before changes in behavior occur, intended users have to perceive it as valuable to their lives and accept the intervention. We adopted some constructs from the Technology Acceptance Model, which is an information systems framework for understanding how users accept and eventually use new technology [32]. The model posits that attitudes, perceived usefulness, and ease of use of technology predict the intention to use the technology, which subsequently correlates with actual use [32]. The strength of the Technology Acceptance Model is that it can be used to explore views from a broad spectrum of users, including patients and health care providers.

The study was conducted in Kampala, the capital city of Uganda, from April to May 2018. This work was accomplished through collaborative efforts between researchers at the Makerere University School of Public Health, the University of Georgia, and the Uganda National TB Program staff. Kampala district has the highest TB burden in Uganda, with nearly 25% of annually reported TB cases occurring in the metropolitan area of the capital city [9]. In Kampala, care for patients with active TB is delivered through public and designated private clinics supervised by the Kampala Capital City Authority, with an oversight from the Uganda National TB Program. Diagnosis and treatment services for TB are provided free of charge to all patients.

We conducted a cross-sectional qualitative study of 5 focus group discussions involving a total of 30 participants. The focus groups were composed of a minimum of 6 participants, as recommended for qualitative studies [33,34]. Two groups comprised patients with TB stratified by sex, 1 group had caregivers, 1 group had TB health care providers (nurses and clinicians), and 1 group had trained DOT community volunteer workers. The caregiver category was broadly defined as any person (a spouse, other family member, or a friend) who regularly provided supportive care, such as accompanying the patient during the TB clinic visits. A DOT community worker was a volunteer who was previously trained by the National TB Program and designated to support DOT in the community for patients attending a specific TB clinic. The patient focus groups were stratified by sex to create more homogeneous groups to foster an environment for free discussion. Owing to cultural norms in the Ugandan settings, men tend to dominate conversations and women tend to be less inclined to speak in the presence of men. Therefore, sex stratification was important whenever possible, and the focus groups of health care providers and DOT community workers were not stratified by sex, as the job positions and number of workers were somewhat fixed. We used a hypothetical scenario to describe the step-by-step process of VDOT, as none of the participants had prior experience with it. We then gathered participants’ perceptions of the potential benefits and barriers.

The hypothetical scenario involved the following: a description coupled with a demonstration of how VDOT works were offered to the participants before the start of interviews in the focus group discussions. The VDOT system, as shown in Figure 1, has 3 main components: a patient-facing side, which is composed of a smartphone app for recording and submitting videos; a secure Health Insurance Portability and Accountability Act–compliant cloud server that stores the encrypted videos; and a provider-facing, computer-based log-in system with a dashboard where submitted medication videos are watched and adherence is confirmed. Daily dosing can be monitored, and individual or aggregated reports can be generated within the system. The system also sends automatic text reminders to patients’ phone numbers. A second reminder was sent 8 hours later if the medication video was not received in the system. An example of these messages is “It’s time to take your pills and send a video. Taking your pills will help you to get cured.” A translated version of the message is also sent in a local dialect (Luganda), which is predominantly used in the Kampala region.

A purposive sample of adult male and female patients with TB with their caregivers was invited to participate in the focus groups. Participants were approached face to face with the help of a clinic nurse at the TB clinic under the National TB Program. Patients with TB were eligible if they were aged ≥18 years and were receiving treatment for at least 2 months under the usual in-person DOT to ensure that they were no longer infectious. Consenting participants were recruited from 4 public TB clinics. Community DOT volunteers were also selected according to the participating TB clinics to which they were attached. All the participants who were invited agreed to participate in the study.

Focus group discussions were conducted in Luganda (a commonly spoken local dialect) for patients, caregivers, and community DOT workers and in English for the health care providers. The moderator (RT), a male qualitative interview expert, reviewed the purpose of the study and the agenda for the meeting. He then described the VDOT process and demonstrated how it works using a smartphone and the app. Participants were shown the special features of the app, password access to the app to ensure privacy, and how the videos were recorded and sent to the cloud system. In addition, the provider-facing dashboard and what happens during a video review session were shown on the computer. RT introduced the interview topics and posed questions, followed by probing. The assistant moderator (RK), a female social worker with vast experience in qualitative data collection, asked follow-up questions, audio-recorded the discussion, and captured nonverbal expressions to enrich the collected data. Interview guides were pretested and revised, and the final versions were used for the data collection process (see an example of the guide in Multimedia Appendix 1). RK and other trained research team members also administered a brief exit survey on demographics, phone ownership and use, and perceived acceptability rating. Focus group discussion sessions were conducted at a selected clinic over the weekends to minimize disruption of routine health care activities and ensure that nonparticipants were not present. Each session lasted between 75 and 125 minutes.

To ensure data integrity, RT transcribed the data verbatim from the audio-recordings in an iterative process within 48 hours of completing the interviews. During transcription, RK reviewed the transcripts for accuracy by playing back the audio-recordings while reading the transcripts. In case of a discrepancy, the transcripts were edited to match the audio-recordings. The transcripts were then imported into ATLAS.ti for thematic content analysis using an inductive approach. Two authors (RT and RK) independently analyzed 3 phases: data immersion, coding, and coding sort. In the first phase, the research team led by JS and EB reviewed the transcripts several times to achieve familiarity and identify emerging issues. In the second phase, the authors flagged relevant transcripts with appropriate descriptive words (codes). In the third phase, the reviewers met and harmonized the independent codes. The harmonized codes were then combined to form categories, and the categories were combined to form emergent themes and subthemes. A third independent reviewer (VK) with vast qualitative analysis expertise verified the codes and themes. The reporting of the study results was guided by the Consolidated Criteria for Reporting Qualitative Studies guidelines [35]. Data from the exit survey were summarized as frequencies, means, SDs, and percentages. Summary statistics of the acceptability rating scores on a scale of 1 to 10 are presented as means and SDs (Table 1). For this study, the scores were further categorized into 3 groups to reflect levels of acceptability as low (1-4), moderate (5-7), or high (8-10).

A total of 5 focus groups were conducted with a total of 30 participants, with an average age of 34.5 (SD 10.7) years and an age range of 20 to 62 years. All participants owned a cell phone, but more health care providers owned smartphones compared with other categories. Overall, experience with cell phone use was modest, with at least 60% (18/30) using their phones to take photographs, record videos, or use the internet. However, patients reported relatively lower use of these phone features of interest compared with health care providers. Participants expressed mixed views about acceptability, with ratings varying somewhat across categories. The mean score for acceptability was 8.2 (SD 1.87), with a range of 5 to 10 on a scale of 1 to 10. About 70% (21/30) of participants rated VDOT as high, with a score of >8, whereas 30% (9/30) perceived it as moderately acceptable. Detailed information on the baseline characteristics of the participants is provided in Table 1.

The results from the focus group discussions are broadly categorized into perceived benefits and barriers. Within each category, we present themes and subthemes that emerged across the 4 categories of participants, including patients, caregivers, community DOT workers, and health care providers.

Three themes and interrelated subthemes emerged from discussions on perceived benefits. These included easy monitoring and support of adherence, enhancing patient-provider communication, and saving time and money. The subthemes presented under ease of monitoring reflect an aspect of this theme. These perceived benefits of using technology for monitoring TB medication adherence could indicate the degree of acceptability of VDOT.

Although VDOT was generally perceived to have several benefits by the respondents, they highlighted some potential barriers to its use. A total of 4 themes and related subthemes emerged prominently from the discussion. These included concerns about limited technology skills among users, inadequate technical infrastructure to support VDOT use, costs related to using the technology, and disruption of domestic routines for female patients. Subthemes that reflect aspects of inadequate infrastructure are also presented. Perceived barriers to VDOT are likely to reduce the acceptability and speed of adoption during large-scale implementation. The detailed findings and related quotes are provided later.

Mixed perceptions of confidentiality, privacy, and other closely related issues emerged prominently in the discussions across all participant categories. Although some respondents believed that VDOT could potentially enhance privacy, others expressed fear that their privacy and confidentiality could be breached. Both ends of the spectrum are presented as subthemes to show the perceived benefits of and threats to privacy or confidentiality.

In this study, we explored perceived acceptability and benefits of and barriers to using VDOT among patients with TB, their caregivers, health care providers, and community DOT volunteer workers in Uganda. Although we used a hypothetical scenario of VDOT to interview participants who had no prior experience, we gathered very rich information about perceived benefits and barriers. We found that VDOT was perceived as acceptable by most participants, despite some differences between patients and providers. Prominent benefits related to the ease of patient monitoring; timely follow-up of missed doses; and enhancement of provider-patient communication, efficiency, and cost savings. The perceived barriers cited were limited technology skills; suboptimal technical infrastructure; costs of smartphones; internet access; and disruption of routine domestic work, especially for female patients. There were mixed perceptions about the impact of VDOT on privacy and confidentiality, particularly with unintended disease closure, potentially resulting in stigma. Similar perceptions of structural and privacy- and patient-related barriers to VDOT implementation have been previously reported [24,36,37]. For example, some studies have reported that video recordings and repeated text reminders are perceived as being overly intrusive [37,38]. In contrast, a study in India reported that patients with TB perceived VDOT as being more private than DOT [24]. The mixed findings underscore the need for a critical evaluation of potential barriers within local contexts. To our knowledge, this is the first study to examine the perspectives of key stakeholders in the use of VDOT in Uganda. More studies are needed to contribute to the sparse evidence base on the topic.

The perceived benefits of VDOT highlight aspects that can be enhanced to increase acceptability among users. The ease of monitoring, convenience, access to objective evidence of dosing, and facilitation of timely follow-up have also been reported in previous studies [23,25,37,39]. Health care providers and patients perceived VDOT as a cost- and time-saving approach in the long run compared with costs that are likely to be incurred with frequent travel when using in-person DOT. Similar findings were reported in a study of health care providers in the private sector in urban Vietnam [39]. In addition, 4 studies quantified the time spent on the treatment observation process and found a significantly larger saving in time and money compared with the usual DOT [21,36,40,41]. VDOT was also perceived to enhance health care provider–patient communication, which in turn affects patient engagement with their treatment [42]. The consistency in these positive aspects makes VDOT a promising patient-centered approach for TB disease management.

The perceived barriers to using VDOT were mostly related to technology usability skills, given the limited experience with smartphones and apps. The technology usability barrier is likely to affect mostly older patients and those without formal education. Similar findings have been reported in VDOT studies conducted in Vietnam, Cambodia, and South India [23,37,39]. In our experience with the VDOT pilot study, intensive training with clear instructions to patients helped them gain the required skills. Inadequate network connectivity and electricity are structural barriers that are more challenging to overcome at the patient or programmatic level. Failure to charge a dead phone battery is one of the most common reasons for missing videos in the VDOT system [43]. The use of solar power banks for charging smartphones might be a short-term solution where electricity is unstable. The absolute costs of smartphones with internet data plans have also been documented as potential barriers in other studies [23,41]. The cost of smartphones on the global market is gradually decreasing, thereby boosting ownership, even in LMICs [44]. Our VDOT pilot study in Kampala showed that 70% of the participants owned a smartphone [43]. However, smartphones may remain unaffordable for some patients with TB. One possible way to minimize inequitable access to digital interventions is to set up a loaner system where patients can borrow and return their phones on completion of treatment. We used a similar system in our VDOT pilot study in Uganda, which worked well [43]. Regarding the cost of the internet, preloading the phones with prepaid internet could be a good option, but it raises other concerns of misusing it for other personal activities. Other creative ways can be explored to address the cost of internet access to support patients’ video submissions. For example, public-private partnerships between the National TB Program and telecommunication companies might tap into resources designated for corporate social responsibility from the business side. Gender roles emerged as a potential issue, especially for women, but have been rarely cited [38], warranting the need for more research. Despite these perceived barriers, patients and health workers believed that VDOT would be more flexible, convenient, and patient-friendly if the main barriers are addressed. Cost and cost-effectiveness studies are needed to inform the implementation and scale-up of VDOT [44].

Privacy and confidentiality issues with the use of VDOT were raised as perceived benefits and potential threats. Previous VDOT studies conducted in India, the United States, and Vietnam have also reported similar mixed concerns with privacy [24,25,39]. The importance of preserving confidentiality, guarding against unintended disease disclosure, and stigma also emerged as prominent themes in studies on acceptance of mobile health (mHealth) interventions among people living with HIV in rural and urban Uganda [26,30,45]. Indeed, the use of any mHealth interventions in monitoring treatment should be ethically comparable with the standard of care [38]. Specifically, for VDOT, the major privacy concerns raised have been addressed by several security features such as a unique personal identification number to facilitate secure log-in into the phone app, video encryption, secure cloud server, and password-protected log-in to the health system dashboard for the health care provider. However, to minimize feelings of mistrust, providers need to reassure patients that caution should be taken to prevent intentional breaches. It is also important to understand the complex cultural perspectives related to perceptions of privacy and cater to them accordingly. For example, some patients in South India perceived repeated adherence text message reminders as intrusive [37]. Finally, patients should also be encouraged to be active players in the process of protecting their own information, for example, by finding a private place to record videos and not sharing their personal identification numbers. DiStefano and Schmidt [38] proposed a valuable framework to guide ethical planning, implementation, and evaluation when using mHealth interventions. The collective goal is to minimize stigmatization and preserve patient autonomy [38].

Broadly, digital adherence technology studies on VDOT, electronic pillbox or Medication Event Reminder Monitor, and 99DOTS performed in LMICs have shown promising acceptability patterns among TB populations [22,37,43,46-48]. General lessons can be learned and applied across interventions and populations. For example, high acceptance in recent studies of VDOT was related to perceptions of convenience, ease of use, and perception of better privacy, whereas lower acceptance was mostly tied to technology skills, app glitches, and cellular connectivity challenges [23,24,36]. In India, a qualitative study evaluating differences in acceptability of 99DOTS, a low-cost, phone call–based strategy for reporting doses, found that high acceptance was related to improved patient-provider communication and the convenience of reduced clinic visits, among others [48]. Similar to our study, low acceptance was related to concerns about cell phone access, technology literacy, and poor cellular connectivity. Health care providers specifically expressed concerns about inadequate training in the use of the technology, changes in workload, and a lack of needed 99DOTS supplies. Although the health care providers in our VDOT study only had a hypothetical exposure, it is possible that similar challenges could arise. In a qualitative study using the electronic pillbox in patients with TB in Vietnam, the technology was perceived to be useful. However, the study participants pointed out that it would be most beneficial as a medication reminder for older patients. The device was less acceptable for people who worked outside of their homes, as they thought that the device was inconvenient to carry around [47]. Concerns about stigma, disease disclosure, or other privacy issues and costs related to technology seem to be crosscutting but to varying degrees from one study to another. Further research is required to expand our understanding on this key area.

This study has several strengths and limitations that should be considered when interpreting the findings. To our knowledge, this exploratory qualitative study is among the first to document the perspectives of TB stakeholders in relation to VDOT in Uganda. It provided helpful insights into potential barriers and benefits that informed the design of the quantitative pilot study that subsequently followed [43]. One limitation of this study was that we used hypothetical scenarios of VDOT; therefore, respondents had no real-life experience using the technology intervention. This could have resulted in increased socially desirable responses and perhaps limited the ability of the respondents to envision potential facilitators and barriers comprehensively. Despite this limitation, respondents identified some critical issues such as cost of internet, the smartphone, and privacy or confidentiality concerns. The sample of respondents was limited to urban Uganda; therefore, findings may not necessarily be generalizable to all TB stakeholders in this local setting. The small number of focus group discussions could have fallen short of reaching the point of saturation. Overall, the findings of this study build on the sparse evidence on the acceptability of digital adherence technologies and their use in TB management in LMICs. More robust research is still needed to explore various aspects of acceptability, feasibility, and ethical issues related to VDOT. Other future research priorities have been highlighted as accuracy, clinical effectiveness, and cost-effectiveness of digital technologies [44].

In the real world, successful implementation of VDOT will require both patients and health care providers to be equipped and willing to use the system. Users must have an adequate level of skills to operate the smartphone, mobile app, and technology system on the back end for health care providers. The baseline characteristics showed that the cell phone technology experience was higher among health care providers than among patients. This will require intensive training and perhaps run-in periods needed to allow users to gain a baseline functional level of skills at the beginning of a VDOT monitoring program. It is plausible that as individuals use VDOT, their level of comfort and usability skills will increase over time [49]. Special considerations must be made for subgroups such as older adults who may need more time to acquire new technology skills [50]. Any modifiable barriers must be addressed to ensure high uptake and sustained engagement with VDOT. Context- and culture-specific barriers for patients and health care providers must be evaluated before deploying the technology. The influence of digital technology use on gender roles emerged as a new aspect with little or no information from previous studies; therefore, further research is needed, especially in the African context.

VDOT was relatively acceptable and perceived as beneficial by most study participants despite the potential technical and cost barriers. There were mixed perceptions about privacy and confidentiality issues related to the use of VDOT. Although some participants thought it would increase patients’ autonomy and privacy, others indicated fears about unintended disclosure of one’s disease status that could lead to stigma. Future efforts should focus on training users, ensuring adequate technical infrastructure, assurance of privacy, and comparative cost analysis studies in the local context.