The DOT Selfie study is an open-label RCT. A total of 144 adult patients with a confirmed diagnosis of drug-susceptible TB who are initiating treatment or are within 1 month of treatment initiation at designated TB clinics will be randomized into 2 parallel groups. Of the 144 participants, 72 will be randomized into a control group to receive the usual-care DOT (UC-DOT), and 72 will be randomized into the intervention group to receive a smartphone with the VDOT app to video-record daily medication intake for review via a secure cloud system. All study participants will attend 2 monthly clinic visits at 2, 4, and 6 months (or at the end of treatment), during which clinical and sputum assessments will be performed.

Participants will be included if they are (1) new or retreatment patients with clinically or microbiologically confirmed TB who initiated treatment within 1 month (we chose to enroll both new and retreatment cases since adherence barriers exist in both groups), (2) aged 18 to 65 years, (3) planning to reside in Kampala, Uganda, for the entire period of 6 months while on treatment to facilitate close follow-up, (4) able to provide signed informed consent, and (5) able to speak and read English or Luganda (the local dialect). Participants will be excluded if they (1) are confirmed to have drug-resistant TB (MDR- or XDR-TB); (2) are very ill patients (ie, those who feel that they would not be able to withstand 2 hours of study procedures at enrollment; (3) have a cognitive, motor, visual, or hearing disability that prevents full participation in VDOT (eg, disabilities that prevent holding a phone or an inability to swallow medication as whole pills); and (4) do not have access to a power source to charge the smartphone or reside in areas without cellular network coverage.

The primary study site will be the Lubaga TB clinic in Kampala, Uganda. Other public clinics, such as Kawala, Kitebi, or Kisenyi, which provide free TB services, will be used as secondary sites to supplement study recruitment. The study clinic sites are all located about 10-15 km from Kampala city center and, together, treat approximately 1000 TB patients annually. We estimate that a total of 8-10 patients per week, or 32-40 patients per month, will be enrolled by the research team. Since the selected sites treat roughly 90-100 TB patients a month, we will use consecutive sample selection to enroll every eligible subject in each arm until the required sample size is achieved. With this recruitment rate, we expect to complete participant enrollment within 4-5 months. The primary recruitment strategy is to screen patients who are newly diagnosed or those that will visit the clinic within the first month of their treatment. Eligibility will be assessed by a trained research assistant who will describe the study to the eligible patient and answer any study-related questions. Patients who are interested in participating will be required to provide written consent, after which they will be invited to complete a baseline questionnaire (Multimedia Appendix 1). The baseline questionnaire will ask questions about (1) patients’ TB diagnosis and initial treatment; (2) socio-demographics and income; (3) experience with cellphones, smartphones, and technology; (4) transportation and other costs; (5) social and personal life; (6) knowledge of TB disease and treatment; (7) privacy and confidentiality concerns; (8) current family and friend support; and (9) perceived stigma from the community. All participants will be educated on TB, including treatment, side effects, and the need for adherence. Thereafter, they will be randomized to one of 2 groups. The participant flow is described in Figure 1.

The primary outcome measure is the level of adherence calculated as the fraction (proportion) of expected doses observed (FEDO) over the 6 months of treatment or by the end of the study following randomization. Since direct observation as a way of measuring medication adherence may be challenging to employ consistently in the UC-DOT group, we will also use self-reports and pill counts as indirect measures of adherence and then triangulate the results. Time-to-treatment completion will be assessed as the main secondary outcome. Other outcomes will be sputum smear conversion at 2 and 6 months or the end of treatment, clinical response as a measure of self-reported improvement in TB symptoms, and weight gain at 2, 4, and 6 months or the end of treatment. Other outcomes will be self-reported side effects, the occurrence of adverse events, and patient satisfaction at the exit of the study. Patient satisfaction will be evaluated in 2 ways: (1) overall satisfaction with VDOT or UC-DOT, and (2) satisfaction with specific areas, such as the VDOT technology and app use; information and training received; interaction with treatment supporters, health providers, or research staff; and privacy and confidentiality during medication intake. Additionally, at study exit, we will conduct individual qualitative interviews for both UC-DOT and VDOT participants to capture detailed information about participant experiences during the study.

Study measurements will be done at baseline and follow-up at months 2, 4, and 6 to ensure close monitoring of all patients in treatment. All participants, regardless of the study group, will complete a questionnaire at these follow-up visits (Multimedia Appendix 2 and 3). The follow-up questionnaire will address issues such as (1) patients’ current use and any changes in experience with a cell phone, such as acquiring a personal cell phone or smartphone; (2) transportation and other costs to and from the TB clinic; (3) TB treatment experiences using VDOT or UC-DOT; (4) experiences when using a phone to record videos or when receiving support under the usual DOT; (5) satisfaction with TB and VDOT or UC-DOT; (6) ongoing support received during treatment; (7) side effects experienced; (8) difficulties with medication supply or transportation to the clinic; and (9) privacy, confidentiality, and stigma concerns. For internal validity, adherence will also be measured at follow-up visits by pill counts (when possible) and using a 10-item self-reported adherence questionnaire [36] (Multimedia Appendix 4).

In-depth interviews will be conducted at the exit of the study to gather detailed information to enrich the understanding of the quantitative data. This will involve intensive individual interviews with a small number of study participants to explore their general experiences, operations, processes, and outcomes that they perceive as resulting from their involvement in VDOT or usual-care DOT. We will also specifically probe the participants’ experiences of stigma, privacy, gender, and socio-cultural issues. To ensure representation of a wide variety of views, participants will be purposively selected based on sex, age, and observed experiences during the time in the study. For example, we will ensure both younger and older and male and female participants are included. In terms of treatment experiences, patients with low, moderate, and high adherence will also be selected accordingly. We will aim to interview a total of 20-30 patients with equal numbers from each study arm. The final number of participants to be interviewed will be determined based on the attainment of thematic saturation (ie, the point at which interviewing more respondents will yield no new information on the topics of interest) [37]. Interviews will be audio-recorded and transcribed. We will use thematic analysis to code, identify key patterns or themes, and then summarize the findings. ATLAS.ti software (version 9; ATLAS.ti Scientific Software Development GmbH) will be used in the coding and detailed processing of the data.

All study questionnaires will be administered using electronic forms on ODK (Open Data Kit) using laptops or tablet devices. Data in the form of videos collected from the VDOT system will be held on a secure HIPAA-compliant cloud server and will be backed up nightly locally in Uganda. Weekly data integrity reports will be run on the system using an automated routine. Monthly data exports will be made, and data will be merged into the master data file. The master data file will be stored at the University of Georgia on a password-protected network with access restricted to study personnel. This will be automatically backed up daily. The master data file will use the unique study number allocated at the time of randomization but will not include personal identifier information. This identifiable information will be held in a separate password-protected file on the same network drive in the form of a lookup table.

This trial is powered on the primary outcome measure and based on a comparison of the medication adherence level between the UC-DOT group and the VDOT group. The calculated sample size is 124 in order to achieve a chi-square test difference of 0.85 versus 0.63 (difference 0.22 and odds ratio 0.30) between the groups. We assumed an attrition rate of 14%, guided by published literature; this was based on a randomized controlled trial comparing a digital adherence intervention and usual care in Kenya in which the overall loss to follow-up rate was 11.7% (9.9% in the usual-care DOT arm and 1.76% in the digital intervention arm) [38]. By inflating the attrition rate to 14%, we erred on the higher side of attrition. In contrast, another RCT of asynchronous VDOT done in the United Kingdom had a higher attrition rate of 23%; however, the study population was very different from our current study in that it included 58% of homeless subjects [15]. At a 14% attrition rate, the estimated final sample size was 144 participants with 72 subjects per group. This would provide a power of 80% to detect a 22% difference in the primary outcome between the 2 comparison groups (85% VDOT vs 63% DOT) based on a 2-sided significance level of 5%. Sample size tables were used to estimate precision. All calculations were performed using the SAS statistical software package (version 9.3; SAS Institute).

Intention-to-treat analysis will be conducted where all patients will be analyzed according to the arm to which they were originally randomized. VDOT treatment observations will be classified as completed if ingestion of all medicines is observed, or if videos are received but not viewable because of a technical complication (given that patients would have no control over whether videos were corrupted). The sensitivity analysis will consider only videos for which all medications are observed as completed. An adherence proportion will be observed for each individual in this study. Let p_ij be the adherence proportion for individual j (j=1, ..., n) in treatment group i (i=0 for VDOT and i=1 for UC-DOT). Then, let p₀ and p₁ equal to the average FEDO in the VDOT and UC-DOT arms, respectively. Similarly, let s₀=s.e.( p₀) be the standard error of p_0, and similarly, let s₁ be the standard error of p₁. The standard t test statistic will be used to calculate a P value based on the following statistic:

We expect that study participants may drop out or miss visits, thus producing missing data and compromising the ability to conduct an intent-to-treat analysis and draw conclusions about the causal link between the method of monitoring and medication adherence. Since there is no universal method of analyzing missing data [39,40], we will follow the guidelines of the National Research Council (NRC) on the handling of missing data in clinical trials [41]. We will focus on 2 critical elements: (1) careful conduct of the study to limit the amount of missing data, and (2) analysis that makes full use of information on all randomized participants. Careful attention will be paid to assumptions about the nature of the missing data underlying estimates of treatment effects. We will use approaches identified by Little and Rubin [42,43] based on 3 categories for classifying how missing data are generated: missing completely at random (MCAR), missing at random (MAR), and missing not at random (MNAR).

Since this pilot RCT study employs a low-risk intervention, is not blinded, and does not involve vulnerable populations, we plan to perform a limited scope of data and safety monitoring. The main focus will be on the safe execution of the study protocol as planned [44]. The principal investigators and co-investigators of the study have constituted an independent data monitoring committee to ensure that the trial is conducted according to the approved protocol, including participant recruitment, accrual and retention, participant risk versus benefit, adverse events, periodic assessments of data quality, timeliness, and other factors that may affect study outcomes. All adverse events will be reported to the University of Georgia Institutional Review Board and the Makerere School of Public Health Research Ethics Committee. Summary reports of adverse events will be made to the National Institutes of Health (NIH) in the progress report at the end of month 4, 6, and the final report, unless the nature of a particular event warrants reporting to NIH immediately.

Interim monitoring will be performed after enrolling 50% (36/72) of patients in the VDOT intervention and control groups. This will be done to ensure that the study is conducted according to the protocol and also to check on the level of adherence in both study groups. All analyses will be led by the study biostatistician. Stopping rules will not be applied to this pilot RCT because there is no evidence to inform the threshold for the clinically important difference (ie, the minimum magnitude of the treatment benefit large enough to offset the treatment harms) [44,45]. Stopping the trial would be justified only on the basis of strong evidence of net benefit or harm [45], conditions which are rarely met in small pilot trials like our study.

An adverse event in this study is any untoward medical occurrence in a patient that is temporally related to the research, whether or not it is related to the intervention or the patient's participation in the research. Data on all adverse events will be collected after informed consent has been provided and patients have been enrolled in the study. This will continue throughout the study. If a patient experiences an adverse event after providing informed consent but has not started to receive the study intervention, the event will be reported as not related to the study intervention. An adverse event that meets the criteria for a serious adverse event between study enrolment and the end of the study will be reported to the local institutional review board as a serious adverse event.

A serious adverse event for this study is any untoward medical occurrence that is believed by the investigators to be causally related to the study drug and either results in death, is life-threatening, results in severe or permanent disability, requires inpatient hospitalization, or is a significant hazard as determined by the data monitoring committee. Serious adverse events that occur after a patient is discontinued from the study will not be reported unless the investigators believe the event may have been caused by a study intervention or protocol procedure. This will be determined based on a temporal relationship to the study intervention and whether the event is unexpected or unexplained given the patient’s clinical course, previous medical conditions, and concomitant medications.

Potential serious adverse events will include lost to follow-up, death from tuberculosis, data security breaches, violence toward study personnel during patient interaction, complaints about the study from patients, or study clinics. These will be reported to the study coordinator and the study chief investigator, who will look into the matter and discuss it with the affected patient’s case managers, and findings will be reported to the trial steering committee and data monitoring committee.

Nonadherence to treatment in patients with TB is a serious obstacle to the realization of the World Health Organization’s End TB Strategy goals [46] because it is responsible for the emergence of multidrug resistance and prolonged periods of infectiousness [4,47]. Currently, there is a lack of universally feasible, acceptable, and effective strategies to monitor and support adherence to TB treatment. This creates a critical need to evaluate VDOT applications under more diverse conditions and settings to define their function and compare them with traditional approaches to treatment monitoring [48,49]. Mounting evidence suggests increased utility and effectiveness of accessible mobile technologies in enhancing the monitoring of patients with chronic diseases [48-50]. The increased utility can be attributed to the increasing affordability and reliability of smartphones in both high- and low-income settings, while the expansion of cellular and internet networks in developing countries is responsible for the effectiveness of mHealth technologies. Therefore, VDOT is a promising alternative way to mitigate the adverse impact of nonadherence on individuals, families, and communities, especially in underserved populations in Africa.

As it is currently practiced in Uganda, in-person DOT has the potential to be stigmatizing for patients, as it may result in unwanted public disclosure of patient disease status. This fear of—or reality of—stigmatization can negatively impact patient autonomy in regard to treatment-seeking behavior and treatment adherence [51]. However, mHealth technologies such as VDOT are likely to be less stigmatizing as they do not require health workers to visit patients at home or work. The remote nature of interaction minimizes the chances of unintended exposure of private information. The flexibility of making video recordings at the patient’s desired time and place greatly enhances privacy [46]. Patients agreed that VDOT promotes autonomy and a sense of control over their health [18]. Nonetheless, even with the use of VDOT, the potential for stigma, unintended disclosure, and a loss of autonomy could still be present given that videos contain identifiable information like the patients’ faces. At the study exit, we will conduct qualitative interviews with study participants selected to discuss their varying experiences with both VDOT and UC-DOT successes, concerns, and failures that may have affected their adherence to treatment during the study.

This protocol describes the design and methods of a randomized control trial assessing the effectiveness of using VDOT plus incentives to improve medication adherence in TB treatment in Uganda. This RCT leverages the accessible features of mobile phone technology to deliver VDOT remotely, in a less intrusive and less cumbersome way, than in-person DOT. It explores a patient-led management approach that is more convenient, enabling patients to take medications on their schedule.

The proposed study is innovative because it is the first time that an RCT will be conducted to evaluate VDOT in an African setting. The control participants receiving usual-care DOT will provide an important comparison that will enrich our understanding and inform future interventions using VDOT and other mHealth programs. Fundamentally, Kampala, Uganda (the proposed site of the pilot study), has a significantly higher incidence of TB disease (201 cases per 100,000 in 2017) than any of the other locations where VDOT has been evaluated to date [2].

We anticipate that a few smartphones may be lost or stolen throughout the study duration. This study will document any phone losses and replace phones to allow participants to continue in the study. Phones may run out of battery power and turn off during video recording. To minimize this issue, patients will be trained and encouraged to ensure that phones are always charged nightly or before recording videos. Interruptions in the electrical power supply may prevent patients from charging their phones to take videos when taking their medications. As such, some adherence records may be incomplete; information gaps will be filled by using alternate data sources such as self-reports [36]. Cellular network connectivity problems may limit VDOT recordings by patients; unless patients travel to remote, rural areas of Uganda, we expect this situation to be rare. Although pill counts and self-reports are indirect measures of adherence, they are acceptable proxies in the absence of more direct demonstrations of adherence [36]. As described above, VDOT has less potential to infringe upon patient autonomy and cause stigmatization compared to the in-person DOT. Lastly, the results of the study will be generalizable to TB patient populations in urban settings in the context of the ongoing COVID-19 pandemic. However, infrastructural and technical limitations may prevail against the implementation of VDOT or similar technology interventions in rural settings.

The significance of this study is at least 3-fold. First, the expected positive impact is the potential to adapt and scale a technology-based alternative to the DOT strategy, one that is contextualized to the local setting, convenient, and cost-saving for patients and the health care system. This RCT is highly relevant because the setting (Kampala, Uganda) has a much higher incidence of TB than any of the other locations in which VDOT has been evaluated to date. Second, it will provide the first RCT data on the use of VDOT in an African country. According to the NIH, “there is a need to stimulate research utilizing mHealth tools aimed at the improvement of adherence to treatment, effective patient-provider communication, and self-management of chronic diseases in underserved populations” [52]. Third, this pilot RCT will be a first step to meeting the urgent need to develop new mHealth tools and test existing ones for contextual acceptability, feasibility, and efficacy to mitigate the serious public health consequences of medication nonadherence. Our long-term goal is to revolutionize patient monitoring, improve patient-provider communication, and promote self-management by utilizing mobile health tools that are contextualized to the African setting.