This study used a convergent mixed methods study design because we concurrently collected qualitative and quantitative studies, analyzed qualitatively and quantitative data independently, and interpreted the results together [24]. Our study participants were persons living with drug-sensitive tuberculosis (persons with tuberculosis) recruited from the Tuberculosis Clinic in the Mbarara Regional Referral Hospital (MRRH) in southwestern Uganda, which provides care to an estimated 600 persons with tuberculosis annually. At the Tuberculosis Clinic, the newly diagnosed persons with tuberculosis receive free tuberculosis medication and are counseled about the benefits of tuberculosis medication. Directly observed therapy is not used for monitoring medication adherence at MRRH due to the costs involved for both persons with tuberculosis and the health care workers. Instead, persons with tuberculosis are given the 2HRZE/4HR tuberculosis treatment regimen (2HRZE: isoniazid, rifampin, pyrazinamide, and ethambutol for 2 months; 4HR: isoniazid plus rifampin for 4 months) every 2 weeks for the first 2 months, at which time they return to the Tuberculosis Clinic for a sputum conversion check. Those who become smear negative continue with isoniazid and rifampin only for a period of 4 additional months in the continuation phase. Those with positive test result receive GeneXpert to exclude rifampicin resistance; subsequent treatment is then individualized. Treatment is also extended up to a full year to cater for missed medication pickups or doses.

Between May 2022 and June 2022, we purposively recruited 21 participants (persons with tuberculosis) according to the following inclusion criteria: (1) being aged 18 years and older, (2) being infected with tuberculosis and undergoing a first-line 6-month course of standard antituberculosis therapy (per above) and counseling per clinic records, (3) owning a cellphone for personal use, and (4) being able to use SMS text messaging. Participants who were unwilling or unable to give consent did not have a cell phone and were unable to use SMS text messages were excluded. A subset of participants for focus group discussions (FGDs) to achieve relatively balanced gender representation were purposively sampled. We also purposively enrolled 9 participants who used DATs in our previous, above-noted study [11] who had completed a first-line 6-month course of antituberculosis therapy. The selection of the 9 participants was purposive based on (1) our perceptions about their openness in discussing qualitative issues based on our previous experience with them in the DATs study, (2) the functionality of their phones at a time of calling them to request their study participation, (3) their proximity to MRRH where we recruited participants for this study, (4) as well as their willingness to participate in this study. The main aim of including the 9 participants in this study was to learn from their practical experiences unlike the other newly recruited 21 participants who had no practical experience in using DATs.

As indicated in Figure 1, the My Mobile Wallet intervention is composed of (1) a real-time medication monitor (Wisepill Technologies, Cape Town, South Africa) to monitor medication adherence, (2) SMS reminders to support persons with tuberculosis in taking their medication as prescribed, and (3) a WiseCash component for sending mobile money incentives for transport to the clinic and motivating medication adherence.

The real-time adherence monitor (evriMED1000 internet-enabled pillbox) is a device that sends signals when opened (records a date-and-time stamp) as a proxy for taking medication. The evriMED1000 is a tabletop dispenser, designed to easily hold a month’s supply of tuberculosis medication. It is an affordable scalable tool, which allows real-time adherence management by automatically synchronizing with the Wisepill cloud service. The evriMED1000 dispenser consists of 2 hardware components, namely the medication container (Figure 2) and the electronic module (Figure 3). This modular design allows the electronic module to be reused and the container to be replaced if needed. The electronic module slots into the container so that the indicator light-emitting diodes are visible through the front of the container. An internal modem and SIM card enable the device to send a real-time mobile signal to a secure web server (hosted in South Africa). General packet radio service maintains the data in transit until the web server acknowledges receipt, which minimizes possible data loss because of power failure or lack of internet connectivity. In the event of inadequate mobile network coverage, the monitor stores openings in flash memory and sends them when the network becomes available. The monitor also transmits a daily “heartbeat” that indicates current battery life, remaining airtime balance, and signal strength as indication of its functionality. The monitor can be charged using electricity or a solar device. Its battery life is 6 months.

As indicated in Figure 4, WiseCash consists of five main elements: (1) “Sense Wisepill” which is a web-based software app that allows the configuration and management of the real-time adherence monitor, including linking participants to their respective real-time monitors; (2) an event forwarding and adherence software program that enables the receipt of the real-time monitor openings as adherence events from the Wisepill server; (3) a clinic appointment and adherence database to store and track participant data and adherence; (4) My Mobile Wallet web app to allow the management of participant clinic appointments and payments; and (5) payment and notification apps to implement the actual payments and send notifications to participants in the local language (Runyankole).

WiseCash includes a script that daily checks participant appointments, evaluates user adherence, and queues and makes payments and payment notifications to participants when due. Yo! Uganda Limited (Kampala, Uganda) is a technology solutions company that offers a single application program interface (API) to provide payments to SIM cards for 2 telecom providers (ie, MTN and Airtel). WiseCash is hosted by Xneelo Limited, a web-hosting company in South Africa. Additionally, Yo! Uganda Limited also provides a second API for the SMS reminders. A number of security elements are included in WiseCash including Yo! Payment platform passwords for the research team members to manage the My Mobile Wallet intervention; API keys with Wisepill data stored on a secure server; and a digital authorization by the research staff for payments triggered by the My Mobile Wallet intervention.

We first oriented each participant about the My Mobile Wallet intervention components by presenting to them an initial version of the intervention as described earlier. We explained and demonstrated how the real-time adherence monitor (Wisepill) works, including how it monitors medication and sends a signal to researchers every time it is opened, how the monitor makes an alarm to remind medication taking, and how to open and close it to put in or retrieve medication. Participants who had never used the monitor were asked to practically demonstrate how it works. We then demonstrated how the intervention could send daily SMS reminders to remind persons with tuberculosis take medication. We also explained how the intervention will send monthly mobile money (US $8) to phones belonging to persons with tuberculosis for transporting them to the clinic at recruitment until the end of their 6-month treatment and monthly medication adherence incentives (US $1.5) to those who meet a ≥90% medication adherence as ascertained from a real-time adherence monitor. The ≥90% medication adherence was considered because adherence less than 90% has been reported as a risk factor for unfavorable treatment outcomes [25].

We used FGDs to solicit participants’ (persons with tuberculosis and patients who had tuberculosis) perceptions about the initial version of My Mobile Wallet intervention. The development of questions for the FGDs (Multimedia Appendix 1) was informed by the Unified Theory of Acceptance and Use of Technology model with specific focus on participants’ perceived usefulness of the intervention, as well as the anticipated challenges in using the intervention [26]. Subsequent versions of the intervention were developed using a UCD approach, which is an iterative technology design process in which users’ needs are continuously solicited and included in the design and development process in order to generate products that can easily work in specific contexts [27,28].

With UCD, technologies are designed and tailored toward users’ needs (rather than the users having to adjust their behaviors to accommodate the technology) to enhance usefulness, usability, and use of technology. An iterative refined version of the intervention was developed based on feedback from participants until the beta version of the intervention suited their needs. Specifically, we conducted a series of 5 FGDs over a period of 6 months to develop the intervention. FGDs were focused on assessing how participants would feel about using the intervention, including (1) anticipated benefits and barriers; (2) usage of mobile phones, mobile money, and SMS text message; (3) preferred content of mobile phone messages; and (4) appropriateness of the intervention for all persons with tuberculosis in the region versus those with demonstrated poverty (eg, means testing). FGDs lasted between 90 and 120 minutes and were conducted in Runyankole, which is digitally recorded, transcribed, and translated to English. Participants tested the initial version of intervention (eg, to withdraw money from it, receive SMS text message, and open the real-time monitor) and provided feedback. We also administered surveys (Multimedia Appendix 2) composed of both closed and open-ended questions. Information elicited by surveys include participants’ cell phone use, preferred frequent and content of SMS text message, timing for sending transport incentives, and anticipated barriers in using the intervention.

We used inductive content analysis [29] to derive categories describing and summarizing how participants perceived the intervention. Initially, authors A Musiimenta and WT reviewed and discussed 20% of transcripts for content relevant to participants’ perceptions about the intervention, anticipated benefits, and challenges. A Musiimenta and WT then assembled a codebook from the identified concepts, using an iterative process, which included developing codes to represent content, writing operational definitions, and selecting illustrative quotes. JH and SL reviewed and discussed the codebook. Following completion of the codebook, A Musiimenta and WT applied codes using NVivo (version 11; QSR International). Differences in coding were harmonized through discussion. We followed the Consolidated Criteria for Reporting Qualitative Research (COREQ-2) [30] in reporting qualitative results. A Mugaba and A Musiimenta used Stata (version 13; StataCorp) to describe study participants, cell phone use, and preferences regarding the intervention.

The institutional review committees of Mbarara University of Science and Technology (MUST-2021-102) and the Uganda National Council for Science and Technology (HS1688ES) approved this study. All participants provided signed informed consent before study participation.

Of 35 screened participants, 3 (9%) were excluded for not having a cell phone and 2 (6%) were excluded for being unable to use SMS text messages. Between May 2022 and June 2022, a total of 30 participants were enrolled in the study and participated in 5 FGDs (each FGD composed of 6 participants, totaling to 30) of whom 21 (70%) were still on tuberculosis treatment with median months on medication of 1 (IQR 1.0-3.5), while 9 (30%) were former persons with tuberculosis from our previous study with a median months after completing treatment of 30 (IQR 28.0-34.5) (Table 1). The majority of participants were living with HIV coinfection (n=26, 87%) and female (n=20, 67%). Most were in their early 40s (median age 44.5, IQR 32.3-52.5 years) and had not studied beyond primary level (n=19, 63%).

As indicated in Table 2, the majority of participants reported not sharing their phones (n=24, 80%), checking their phones for SMS text message more than once a week (n=24, 83%), preferring to receive daily SMS reminders to weekly SMS (n=23, 77%), and preferring to receive SMS reminders that are not easily related to tuberculosis (n=20, 67%). The majority of participants (n=25, 83%) preferred being sent mobile money transport incentives 1-2 days before clinic visit. Barriers anticipated in receiving money included automatic reduction of money by mobile phone service providers for participants who borrow money from the service providers (n=4, 13%), the possibility of using the money for other competing demands such as buying food (n=12, 40%), misusing the money by other people in case of phone sharing (n=4, 13%), and inability to read mobile money messages (n=8, 27%).

The My Mobile Wallet intervention composed of real-time adherence monitor, SMS reminders, and mobile money incentives was perceived as useful in supporting tuberculosis medication adherence in a low-resource setting. Participants expressed the perceived usefulness of the My Mobile Wallet intervention in terms of being reminded to take medication, supported with transport to the clinic, and money to meet other tuberculosis medication–related costs, all of which were perceived to imply care, which could create a sense of connectedness to health care workers. This could consequently cause participants to develop a self-perceived need to prove their commitment to adherence to health care workers who care for them, thereby motivating medication adherence.

Through a UCD process, we learned that the main concerns were (1) unintended status disclosure triggered by seeing the real-time monitor or hearing the alarm from the monitor or seeing the SMS reminders and (2) using the money meant for transport to the clinic for other competing demands. The main design considerations that arose in the process were (1) using language in the SMS text message that is confidential and not easily related to tuberculosis and (2) sending transport refund a day before the appointment to limit the time lag between receiving the money and visiting the clinic.

The main limitation of this study is that we asked participants about perceptions before they could use the intervention in real life, although 9 participants (30%) had used the real-time monitor and SMS text message in our prior study. Although we practically demonstrated the use of the intervention to all participants, 21 (70%) participants were not able to describe actual experiences using the device or medication-related SMS reminders or financial incentives as part of their daily routine.

Participants who had used the real-time monitor felt being cared for as a result of being monitored and sent an SMS reminder. This created a sense of connectedness to health care workers and consequently caused participants to develop a self-perceived need to prove their commitment to adherence to health care workers who cared for them. Participants referred to the real-time monitor as a friendly companion that can potentially offer welcoming medication monitoring compared to human beings whose mood may be dictated by circumstances.

Given the prevailing stigma and discrimination that is associated with living with tuberculosis in Uganda [31], such positive perceptions could empower participants to cope with social isolation and other challenges of having tuberculosis, especially among those who do not have adequate social support, those initiating treatment who are not yet accustomed to taking medication, and those living with HIV coinfection who are more likely to experience double stigma from both tuberculosis and HIV. Experiences of unintended HIV status disclosure were reported by the people living with HIV or AIDS who used the real-time device and SMS reminders in the same setting [32]. Noteworthy, however, these technologies can also facilitate intended HIV status disclosure in order to get social support [33]. To reduce the potential effects of the tuberculosis status disclosure that can potentially emanate from using the intervention, strategies to ensure privacy and confidentiality (eg, using SMS text message that is not easily related to tuberculosis) and reduce stigma and discrimination were incorporated into the intervention.

Participants felt that providing financial incentives in the form of mobile money could relieve persons with tuberculosis of the transport burdens associated with clinic visitations for pill refill, thereby eliminating the need to walk long distances to the clinic or borrow money to get transport or miss pill refill appointments. Lack of transport to the clinic is one of the major challenges constraining tuberculosis medication adherence in Uganda [34]; 2 in every 10 people (21% of the population) live on US $1.25 a day or less [35], making saving for health difficult amid other immediate competing interests. To reduce the possibility of using the money meant for transport to the clinic for other competing demands, participants proposed sending transport refund a day before the appointment. Participants also reported that financial incentives could promote medication taking in order not to miss the money involved, which could be used to meet other costs related to tuberculosis treatment such as nutritional supplements and food. This is important given that in Uganda, tuberculosis or HIV high burdened country, over 53% of tuberculosis-affected households experience tuberculosis costs, which are above 20% of their annual household expenditures, and the main cost drives are nonmedical expenses such as transport to the clinic and food [36]. Because tuberculosis disproportionately affects the poorest in the population, its poverty-aggravating effects are felt more by those who are already vulnerable. Moreover, having tuberculosis is associated with other negative socioeconomic consequences, such as divorce, school interruptions for children, and social exclusion, in addition to the hardship of deteriorating health. Incentive-based interventions can potentially overcome the poverty-based structural barriers to tuberculosis treatment, although research in this area is limited. Incentive-based interventions can also potentially be integrated with routine health care especially given the evidence that an incentive as small as US $1 can improve tuberculosis cure and reduce loss to follow-up among persons with tuberculosis in rural Uganda [37]. This can potentially prevent drug resistance, thereby relieving the country of the financial burden associated with MDR tuberculosis treatment—Uganda has over 16,000 patients with MDR tuberculosis [35]. The country currently spends US $105 (on transport refund and food incentives) per month on each patient with MDR tuberculosis from the time they are discharged (after the initial 6 months of hospitalization) until treatment completion (1 year). This translates to a cost of US $1260 per patient for the 12 months of treatment. Importantly, treatment for MDR tuberculosis leads to productivity loss for patients and their caregivers, which negatively affects the economic development of the country.

Using a mixed methods approach to collect both closed- and open-ended data, this study identifies important insights (about perceived usefulness, concerns, and design considerations) from patients with tuberculosis that can inform the development of real-time adherence monitoring technologies linked to mobile money–based financial incentives. In sum, real-time adherence technologies complemented with mobile money–delivered financial incentives could potentially provide acceptable approaches to remind, motivate, and support patients to adhere to taking their medication especially in settings where directly observed therapy short course is difficult to implement as well as low-resource settings where poverty-based structural barriers to tuberculosis treatment are prevalent.

As DATs proliferate, using user-centered approaches to understanding how they are perceived by users is critical to developing tailored interventions that are useful, usable, and acceptable in supporting tuberculosis medication adherence. Findings from this formative study enabled the design of the My Mobile Wallet to support tuberculosis care and treatment adherence, which is currently being implemented for initial feasibility and acceptability assessment. After this initial assessment, further refinements will be done before the technology can be further tested for large-scale feasibility, acceptability, and impact in a planned randomized control trial (NCT05656287).