Barriers and facilitators of antiretroviral therapy (ART) adherence habit formation in Sub-Saharan Africa: Evidence from a qualitative study in Kampala, Uganda

Abstract

Rationale:

In 2020, nearly 40 million people lived with HIV/AIDS (PLWHA) worldwide, of whom 70% were receiving antiretroviral therapy (ART). Two-thirds of PLWHA reside in Sub-Saharan Africa (SSA), where rates of viral load suppression are often suboptimal and frequently attributed to low ART adherence. Strong pill-taking habits are often reported as a key strategy among those who successfully maintain medication adherence, yet not enough is known about the barriers and facilitators in SSA to pill-taking in response to the same contextual cue, which is a necessary step in the habit formation process.

Objective:

To address this knowledge gap and to inform a subsequent intervention to promote context-dependent repetition, called anchoring, we used a formative qualitative approach to collect in-depth narratives about barriers and facilitators of the anchoring intervention for establishing ART pill-taking habits at the Mildmay Hospital in Kampala, Uganda.

Methods:

We conducted interviews with 25 randomly selected patients starting ART, 5 expert patients, and 10 providers at Mildmay, and performed a rapid analysis to inform the intervention in a timely manner.

Results:

We found that pill taking in response to the same contextual cue, or anchor, was threatened by stigma and food insecurity and that the COVID-19 pandemic exacerbated these barriers. We also determined that important linguistic changes were needed to the instructional materials and reminder messages in the subsequent intervention to avoid words and phrases with negative connotations for this target population.

Conclusions:

Several important barriers and facilitators to context-dependent pill taking in Uganda were identified through our formative research that helped to inform important revisions to our subsequent intervention. These findings underscore the importance of understanding local barriers and facilitators when designing and planning interventions, particularly when implementing theory-based intervention approaches that have yet to be tested in a new setting.

INTRODUCTION

Globally, nearly 40 million people are living with HIV/AIDS (PLWHA), and by the end of 2020, it was estimated that over 27 million PLWHA were receiving antiretroviral therapy (ART) (UNAIDS, 2020). Two-thirds of PLWHA reside in Sub-Saharan Africa (SSA), many of whom have been initiated on ART (UNAIDS, 2020). However, increased access to HIV care has not consistently translated into optimal patient outcomes, such as HIV viral load suppression. Regional SSA estimates show that only 58% of all PLWHA and 87% of PLWHA on ART are virally suppressed (Marsh et al., 2019). In Uganda, only 64% of all PLWHA and 88% of PLWHA on ART are virally suppressed (Marsh et al., 2019). Health interventions are urgently needed to increase these rates of viral suppression in SSA in order to reduce the health and economic burden of HIV in this region.

A major factor for suboptimal rates of viral load suppression globally is poor adherence to the ART regimen (Bukenya et al., 2019; Galea et al., 2018; Inzaule et al., 2016). While a wide range of economic (e.g., transportation distance (Bajunirwe et al., 2020), cost, food security (Anema et al., 2009)), social (e.g., family members’ and healthcare providers’ adherence attitudes and support (Stadler, 2021)), and structural (e.g., access to HIV care; (Moriarty et al., 2018)) barriers to ART adherence have been identified in the literature, even when these barriers are overcome, behavioral barriers remain as a key contributor to ART nonadherence (Crim, 2020; Stecher et al., 2021a; Stecher and Linnemayr, 2020; Zogg et al., 2012). Even among PLWHA who are initiating ART treatment, a period when motivation and attention to adherence are heightened, forgetfulness, and declining motivation are commonly reported behavioral reasons for ART nonadherence (Hodgson et al., 2014). Specifically in Uganda, researchers have identified behavioral barriers as an important determinant of ART nonadherence, and have stressed that: “it is necessary to examine how treatment-related routines develop as a particular set of practices” (Winchester et al., 2017). Therefore, we developed a behavioral intervention for ART treatment initiators to combat forgetfulness and declining motivation in order to maintain high ART adherence and viral suppression, and we sought input at this formative stage from participants to adapt it to local conditions (Kiwanuka et al., 2021) and to receive general feedback on the intervention overall as well as its components.

Daily pill-taking habits (or routines) are a commonly reported strategy used by those who successfully maintain high medication adherence (Brooks et al., 2014; Phillips et al., 2013), but are often difficult for people to form on their own. A habit is defined as a reflexive, or automatic, behavioral response to a contextual cue (Wood and Neal, 2016; Wood and Runger, 2016). As outlined in Gardner and Lally (2018), a reflexive habit is formed when an individual (a) has the motivation to act, (b) has the self-regulatory skills to act, (c) repeats the behavior over time, and importantly, (d) repeats the behavior in response to the same contextual cue. This habit formation process establishes a cue-behavior association that allows the behavior to be performed with little to no cognitive effort, and no longer relying on motivation, self-regulation, or deliberation to be executed (Gardner, 2015; Rothman et al., 2015). In this way, habits can circumvent many common behavioral barriers to maintaining healthy behaviors, which helps to explain why contextually-cued habits have been successful at maintaining dietary improvements (Kothe et al., 2015), oral contraception adherence (Murphy et al., 2018), increases in physical activity (Ellingson et al., 2019; Gardner et al., 2014), and high adherence to various other medical treatments (Bolman et al., 2011; Hoo et al., 2019; Yeh et al., 2019). Unfortunately, even after conditions (a) – (c) of the habit formation process are met, it is often difficult for people to perform context-dependent repetition long enough for the targeted behavior to become a habit, and in these cases, additional interventions are needed to help people establish a reflexive cue-behavior association (Lally et al., 2010; Lally and Gardner, 2013).

One common intervention technique for supporting context-dependent repetition is to anchor, or pair, a new behavior with an existing routine (Gollwitzer and Sheeran, 2006; Lally et al., 2010, p. 20210; Wood and Neal, 2016), but this approach has not been tested in SSA and thus the barriers and facilitators of this intervention in SSA are still unknown. The anchoring intervention asks participants to identify a stable existing routine that they consistently perform every day, and participants then write down a plan to perform the targeted behavior directly after their existing routine (Neal et al., 2013). In this way, the existing routine acts as the contextual cue for participants’ new habit. This approach has been shown to establish healthy habits, such as improved diet (Armitage, 2004; Gardner et al., 2014), increased physical activity (Prestwich et al., 2003), improved meditation (Stecher et al., 2021b), and improved medication adherence (Brown et al., 2009; O’Carroll et al., 2013). On average though, only 55% of participants in these anchoring interventions repeat the targeted behavior in the same context long enough to form a habit. Therefore, the intervention we plan to test in SSA will combine anchoring with additional reminder messages and rewards informed by behavioral economics to better maintain context-dependent repetition during the habit formation period (Stecher and Linnemayr, 2020). The anchoring intervention method has yet to be conducted in SSA though, so the barriers and facilitators of taking ART medication in response to the same existing routine in this region are unknown and may limit the success of our combined intervention approach (Jennings Mayo-Wilson et al., 2020; Neal et al., 2006; Wagner and Ryan, 2004).

To start filling this knowledge gap (and in preparation for a randomized controlled trial (RCT) to test our intervention), we used a formative qualitative approach to collect in-depth narratives about barriers and facilitators of HIV medication adherence habit formation among clients of the Mildmay Uganda Hospital (hereafter referred to as Mildmay), one of Uganda’s largest HIV clinics in Kampala and the site of our upcoming RCT. The anchoring intervention we are testing hinges on participants being able to identify an existing routine behavior that they consistently perform every day in the same context, such as brushing teeth in the bathroom in the morning or eating breakfast in the kitchen (Wood and Neal, 2016). Several important barriers may exist to implementing the anchoring intervention in Uganda though, such as a lack of consistent daily routines, social stigma, and/or a lack of privacy or ability to hide one’s HIV status. We therefore asked clients of Mildmay about the activities they routinely complete during the day, the location of these routines, how consistently these routines are performed, and who is typically present when performing these routines. We also asked healthcare providers at Mildmay questions about the existing adherence counseling materials and about their perceptions of clients’ routines and barriers to habit formation. This formative phase of our research was designed to help us understand how well the anchoring intervention can work in Uganda and to better adapt the intervention content (e.g., the anchoring plan instructional materials and text message reminders) for this setting. Qualitative methods have often been poorly and inconsistently utilized in the development of RCTs (Davis et al., 2019). Yet, qualitative data, such as our in-depth interviews, can shed light on individual and contextual dynamics, helping to address many of the complex challenges of RCTs for testing novel health interventions (Davis et al., 2019) and adapting culturally-tailored interventions (Montgomery, 2016; Pallmann et al., 2018; Sandelowski, 1996). This paper moves the field forward in this area by describing the barriers and facilitators of a habit formation intervention for ART treatment initiators in Uganda, and illustrates how qualitative exploration can be used to culturally tailor interventions for specific clinical and regional settings.

METHODS

Sample and Recruitment

Incentives and ReMINDers to Improve Long-term Medication Adherence (INMIND) is a mixed methods RCT involving both qualitative (interviews) and quantitative (survey, electronic health records, and electronic pill bottle cap readings) data collection [1R34MH122331–01A1]. This qualitative study draws on formative data from interviews conducted to refine the proposed INMIND intervention before implementation. During the interviews, participants were shown draft materials pertaining to the intervention and then offered feedback. INMIND proposes to test a novel intervention for maintaining high ART adherence among PLWHA initiating ART treatment at Mildmay by turning daily ART pill-taking into a contextually-cued habit. Specifically, this is a 3-armed randomized controlled trial with two intervention groups. All participants (including in the control group) will receive information about the importance of habits, as is part of the standard adherence counseling for treatment initiators at Mildmay, and will create personalized ART adherence anchoring strategies. The anchoring strategy involves identifying an existing routine that can serve as a daily contextual cue for ART pill-taking, and then writing down an implementation intention based on the chosen existing routine (e.g., “After eating breakfast in the morning, I will take my ART medication”). In the first intervention group, participants will additionally receive daily text message reminders for three months to keep ART adherence and their anchoring strategy salient (high on their mental priority list). In the second group, we add small rewards for taking ART pills within a two-hour time window that corresponds to participants’ personalized anchoring strategy to increase the immediate benefit of context-dependent ART pill-taking.

In this formative qualitative phase, we recruited 25 treatment initiators, 5 expert patients, and 10 providers to participate in in-person semi-structured interviews conducted at Mildmay. We focused on these three categories because they each bring a different perspective to our research questions regarding barriers and facilitators of context-dependent ART pill-taking habits. Treatment initiators were defined as those who had been diagnosed with HIV at Mildmay and had been receiving care from Mildmay for less than six months prior to the interview date. We expected these participants to offer valuable information about the habit formation barriers and facilitators at the start of the regimen. Expert patients could have been diagnosed and receiving care at Mildmay for more than six months before the interview date but needed to demonstrate high medication adherence, which was identified by Mildmay health care providers. We included these patients in the study so that they can share lessons learned from their ability to maintain ART adherence successfully. Finally, providers were eligible for these formative interviews if they interacted with ART treatment initiators frequently at Mildmay. We spoke with providers to elicit their perceptions on the anchoring intervention’s ability to promote context-dependent ART pill-taking based on their experiences with this patient population.

Using the clinic’s electronic health records database as a sampling frame, we identified all patients who met our eligibility criteria for treatment initiators. Participants were selected randomly and added to a list that specified the dates of their upcoming appointment at the clinic. On the day of their clinic visit, a study coordinator approached each potentially eligible treatment initiator and introduced the study to them. Those who consented to participate proceeded with the interview either immediately after their clinic appointment or at another time that was more convenient for them. In total, 36 patients were randomly selected and approached about the study, of whom 28 consented and agreed to participate, and 25 completed both parts of the interview. The other 8 refused to participate, citing reasons such as lack of time or interest, inability to respond to questions, or a fear of being recorded. Of those who consented, 3 did not complete their interviews and were excluded from the study. Expert patients typically work at the clinic as part of a peer support program maintained by the clinic. They were recommended for participation by the head of the peer support program. All 5 expert patients and 10 providers approached for this study consented and completed the full interview. All recruitment, data collection, and analytic procedures were approved by the RAND and the Mildmay Institutional Review Boards. Participants were offered 30,000 Uganda Shillings (approximately $8.50) for their complete participation of this qualitative phase of the study.

Data Collection

All interviews were conducted in person at Mildmay between April and June 2021. The team developed two sets of interview protocols: one focused on contextual barriers and facilitators of ART habit formation and another focused on intervention development. These protocols were tailored for clinic patients and providers respectively, such that we deployed four protocols for data collection. The protocol on contextual factors comprised the following topics: (1) existing pill-taking behaviors; (2) adherence habit formation facilitators; (3) adherence habit formation barriers; (4) and adherence during the COVID-19 pandemic. The protocol on intervention development sought input on: (1) study informational flyers; (2) language of text message reminders; (3) attitudes and perceptions of rewards; and (4) integration of the intervention into the clinical workflow (for providers only). Each of the resulting four protocols was designed to last 30–45 minutes, and each participant was asked to complete both the contextual and the intervention protocols. To manage participant burden, interviewees were given the option to do the two parts of the interview consecutively or to schedule the second interview at another time. The team members based at Mildmay translated the protocols into Luganda, so that participants who were not comfortable speaking English could speak Luganda instead.

The interviews were conducted by two Ugandan study coordinators, a 37-year-old female with an undergraduate degree and a 46-year-old male with a Master’s degree, who were knowledgeable and experienced conducting research on HIV care among this patient population. They were bilingual (English and Luganda), had prior training in qualitative interviewing, and prior experience with qualitative data collection. The two interviewers received a refresher training in qualitative interviewing and had several weeks of intense practice with the protocols. Supplement 1 provides the four protocols in full in English. All interviews were audio-recorded, transcribed verbatim (for those conducted in English), or translated and then transcribed (for those conducted in Luganda). For context, it is important to note that these interviews were conducted at a time when the death toll of the COVID-19 pandemic in Uganda was rising. The country had experienced lockdowns the year before and the situation deteriorated during our interview timeframe such that Uganda re-entered a partial lockdown just as the interviewers finished data collection in June 2021. The research team decided to continue with in-person data collection on the basis that the clinic was following the standard operating and safety procedures mandated by the government during the partial lockdown.

Data Analysis

Given the need for this qualitative data to inform our intervention development within a short timeline, and given the added complexities resulting from conducting in-person research during COVID-19, we employed a rapid assessment (RA) approach (Johnson and Vindrola-Padros, 2017) that precluded the iterative calculation of metrics like inter-coder reliability (Cohen, 1960), percent agreement (McHugh, 2012), and saturation (Saunders et al., 2018). In the context of this study, RA was especially advantageous because it facilitated the rapid identification of issues relevant to the development of the intervention (Brahmbhatt et al., 2009). The second author worked iteratively with the project leader and one of the interviewers to develop the rapid coding approach. The coder was trained in qualitative methods in the context of health services research and behavioral science but had no significant prior exposure to the target population in this study or to the subject of ART adherence habit formation. Therefore, the analytic process did not draw on any prior assumptions or expectations from prior work.

In the first step of the analytic process, we developed an Excel template to compile summary notes from each interview, including illustrative quotes (Gale et al., 2019). The template contained pre-identified domains from the interview protocols. Yet, this process also allowed for some degree of inductive analysis based on dimensions from the interview content; that is, when interviewees offered new perspectives beyond what had been expected in the protocol, such as fear. Data were extracted into Excel by the second author, while the two interviewers from our partner site cross-checked the extraction against their own notes taken during the interviews.

In the second step of the analysis, we consolidated the summaries by participant type (i.e., treatment initiators, expert patients, and providers). This consolidation facilitated comparisons across the three groups, and the identification of variation within broad thematic categories. This approach helped us offset, to a degree, the depth sacrificed with the rapid analysis, by capturing both what the interviewees’ perceptions were overall and how facilitators and barriers varied within and across participants (Lin, 1998).

RESULTS

Treatment initiators (TI) were aged between 21 and 63 years (mean age 29). Sixty-eight percent of TI were female. Providers (P) included physicians, counselors, nurses, social workers, and clinic leadership, with experience ranging between 1.5 and 20 years. Mean provider age was 36, and 70% were male. Expert patients (EP) were aged between 39 and 54 years (mean age 49) and were mostly female. Their experience with ART ranged between 6 and 17 years among the EP sample. While we did not collect income information from treatment initiators or expert patients in this study, our prior research at Mildmay has found that patients’ average monthly disposable income is roughly $66.63 (SD 63.23), and at this income level, only 7.74% report experiencing food insecurity in the past year (Stecher et al., 2021a).

Below we report two main thematic categories corresponding to the protocols used, each with several subcodes: 1) contextual barriers and facilitators of ART adherence habit formation; and 2) intervention feedback (e.g., study flyer design and feedback on the reminder messages).

Contextual barriers and facilitators of ART adherence habits

Limited opportunities for anchoring activities

For most expert patients and treatment initiators, their homes were the most comfortable and convenient place to take their daily ART medications on time and in privacy. Home was where most patients reported having reliable access to food, water, and timekeeping devices. However, some providers and treatment initiators said it was important to be able to take pills on the move, especially given patients’ lifestyles (e.g., fieldwork assignments, traveling for social engagements, or commuting between urban and rural areas). For some clients, such lifestyles had disrupted their ability to form a habit of taking their ART because they were often away from their familiar context and cues: “On long journeys, I forget” (TI 24). At the same time, other patients reported successfully building a pill-taking habit within their fluctuating daily schedules by planning ahead and traveling with their pills. One nurse explained, “Those who are adhering well, they move with their drugs. Like if they move away from home maybe for a week, they carry their seven pills, they take their pills properly packed in small polythene bags” (P 5). This suggests that alternative strategies to anchoring daily ART adherence may be necessary for these patients who experience inconsistent daily routines.

With regards to time of day, participants were split on the advantages of taking one’s pill in the morning or evening, with important implications for the chance to develop a healthy pill-taking habit. For some, especially those who were employed, mornings typically felt rushed and under the pressure to get to work on time, which had increased their likelihood of forgetting to take their ART medication. Another potential disadvantage of mornings was that some patients were not in the habit of eating breakfast, which made it more difficult to take a pill that often has fewer side effects when taken with food. For many others though, the morning offered enough behavioral anchors, such as taking a bath, getting dressed, or having tea, which they saw as good anchors for pairing with pill taking, as illustrated by this treatment initiator: “You wake up in the morning, you wash your face, and you get what to eat. Then you take medicine. That is the reason as to why I decided in the morning” (TI 21). Conversely, for other initiators and expert patients, evenings were often described as a relaxed time of day when they expected to be reliably at home with access to the main meal of the day and other behavioral anchors: “Nine pm works for me, because I would be done with taking a bath, and even in the stomach there is something” (EP 4). As such, these patients perceived that evenings would make it easier for them to anchor their ART adherence. For an intervention aiming to support ART adherence habits through anchoring, this heterogeneity in the timing of existing behavioral routines in people’s lives points to the need to allow for flexibility and personalization when identifying an existing routine for anchoring.

Stigma

A commonly reported threat to the use of reminders and/or visual cues for pill taking was stigma. For some expert patients and treatment initiators, disclosing their HIV status meant that they were marginalized in their homes or local communities, with some even experiencing domestic violence and running the risk of losing their jobs. As a result, many hesitated to disclose the HIV status to friends, family, or work colleagues, which in turn burdened them with the need to hide the pills, the act of taking the pills, clinic phone-calls, and reminders to take the pill, rendering it difficult to consistently take their pills at the same time and/or location every day. One of the counselors explained that “people are hiding their medicine in the corners, hiding medicine in their cars, hiding the medicine in the kitchen. Others are just hiding medicine in the pillows” (P 7). As this treatment initiator described, the mental exhaustion from the ongoing deceit may mean that some forget or ignore the reminders when it is simply too risky to take the pill: “Once you start hiding, you are not going to take your medicine well. That is because when you hide, you are going to miss your appointment to come here. You are going to miss your time of taking the medicine, reason being you don’t want people to know” (TI 2). Another initiator talked about fearing her husband finding out about her HIV status: “At my place, my pills I take them when I am hiding. Reason being my husband doesn’t know it that I am on medication. Now when he sees the pills, we will have a fight” (TI 24). For habitual pill-taking, this means that not all patients can use visual cues, such as leaving the pill bottle out, or use digital reminders, such as phone alarms, if they have not disclosed to the people they live with.

Family support

Some expert patients and treatment initiators did describe successfully being able to use reminders to help with pill taking, such as phone alarms, radio and TV programs they consume regularly. In other cases, it was supportive friends and family who reminded them to take their pills on time, as this treatment initiator explained: “My wife is also one of my reminders, sometimes I sit in bed, and she asks me ‘have you taken your medication’?” (TI 13) For another interviewee it was her young son who played this key role: “I have my boy child who is 11 years old, I tell everything, and so sometimes even if I have not yet taken the pills, he tells me that ‘mom you have not yet taken your pills’ and he brings for me water and then I swallow them” (TI 11). Importantly, the ability of patients to use these forms of reminders and/or rely on friends or family requires that they not face stigma, so these strategies are not available for many ART treatment initiators.

Perceived absence of symptoms and negative ART side-effects

Some interviewees commented on the absence of immediate symptoms when ART is not taken as prescribed, noting that this may lull patients into a false sense of security that it is ok to not take the pills and take ‘pill holidays’, as this treatment initiator explained: “You really suffer from flu and say eh, I forgot to take drugs. You sneeze, and you remember quickly, but this one [ART] it gives us a difficulty. You have nothing paining you, but you are supposed to take it. If you have not yet got anything, it doesn’t be on your heart. In the beginning, you can be suffering from malaria and you shiver, you remember I have missed the medicine, even if it is minutes you remember” (TI 1). Similarly, one counselor said: “He [patient] will just think of medicine when he just starts like coughing, when it starts like the body reaction, the rashes” (P 7). In addition, both expert patients and treatment initiators explained how the experience of side effects at the beginning of their treatment challenged their ability or willingness to continue with ART. Experiencing dizziness, fatigue, and fever, one treatment initiator said that “it was somehow hard for me, I even missed it. I didn’t come back, and I got scared” (TI 25). Such side-effect induced breaks in pill-taking pose a threat to the success of our anchoring intervention which requires consistent pill-taking during the habit formation period.

Prior experience with long-term medication regimens other than ART was not a factor in this sample. Of the 25 treatment initiators, most had no prior experience with taking long-term medication. Of the few who mentioned previous regimens for issues such as infections, asthma, tuberculosis, no one talked about having learned a pill-taking routine. Instead, the key lesson for these patients was their conviction that taking medication can help them feel better. Many described the act of taking ART as an act of survival, such as this expert patient who said that: “In taking this medicine, the fact that I take it well, it has helped me to stay alive [and] well” (EP 2). Several treatment initiators reiterated this point, for example: “I am supposed to take it so that I can be alive. It means this medicine is like food, when you don’t eat food, you die” (TI 4). Another one admitted that “I know that when I swallow them then I will get a better life than when I don’t take them” (TI 12). According to providers and treatment initiators, key drivers of this attitude included the patients’ coming to terms with the diagnosis, and an intrinsic value of life, as these treatment initiators stated: “It started with me by accepting that with me taking this, I have a life there” (TI 15), and “I have to take this because I love my life” (TI 23). Contrary to the experience of older adults in other settings who often take multiple medications in a day and thus have an opportunity to think about ways to take their pills as prescribed,³¹ it does not seem that in our sample a significant fraction are already taking pills for other chronic conditions that may have helped them form healthy pill-taking habits, and/or serve as an anchor behavior to take ART pills.

Food insecurity

Food insecurity—disruptions in regular meals and reductions in food intake—emerged as a significant perceived impediment to habit formation across providers, expert patients, and newly diagnosed patients. As this nurse explained, some patients site their lack of food as a reason for not taking their pills: “We have seen cases like some of them don’t have jobs, they cannot find something to eat, and they give an excuse that I cannot take this medication without food” (P 5). Expert patients pointed out that food insecurity may exacerbate ART medication side effects, because when the pills are ingested without food patients may experience physical weakness and dizziness. Lack of understanding of the relationship between ingesting pills incorrectly and side effects may in turn lead to incorrect assumptions that the pills do not work, as this expert patient said: “a person thinks that now I am on medication, why do I get so weak, unending dizziness? But you need to eat, you need to drink on time” (EP 5). Similar narratives emerged from treatment initiators, such as this one who said “They [pills] would get to my head, and I be as if I am drunk, but I realized it later that I don’t drink water a lot and even eating, I wasn’t eating” (TI 7). As an important determinant of pill-taking, food insecurity will need to be considered when evaluating the efficacy of our anchoring intervention and potentially countered in future habit formation interventions.

COVID-19 pandemic

Some of the pre-existing issues, such as food insecurity and stigma inside the home, were exacerbated by the COVID-19 pandemic. Loss of employment and confinement to overcrowded homes has made it challenging for treatment initiators to stick to their pill taking habits. Participants also spoke about several new challenges that emerged during the pandemic that were perceived to undermine habit formation. First, impoverished treatment initiators were most affected by the sudden increase in transportation costs, as bus fares to the clinic became prohibitively expensive. As a result, they faced disruption in procuring ART refills. In addition, the clinic itself lacked the resources needed to provide transportation outreach to remote patients, as this nurse explained: “We didn’t have transport, we couldn’t offer services. The ambulance was allowed to move to everywhere, but who could put the fuel to visit every client or who could go to the clients?”(P 4) Another issue with implications for the success of anchoring in establishing an ART adherence habit may be the rise in mental health problems associated with the pandemic. As this clinician explained, “our psychiatric clinic is also growing. They lose hope, they go into remorse, and they became hopeless. So the fact that COVID brought in depression, the pill taking habits have also changed” (P 9). Just as food insecurity, healthcare access and mental health are two important barriers to our anchoring intervention that need to be considered in our evaluation and when designing future interventions in similar settings.

Intervention feedback

Recommendations for the study flyer design

In the interviews, we also asked for feedback on a leaflet we designed to explain the importance of habits for maintaining pill-taking and to explain the anchoring intervention strategy. An important question emerged regarded the use of the word “routine” versus “habit”: when asked about what comes to mind when they hear the word “routine,” many spoke about something that must be done every day, regularly, like eating or brushing one’s teeth. When offering examples of “routines,” many participants included time references, such as “every day at the same time,” “something you do every day or after every particular time,” or “every day at 8.” Definitions of “habit” similarly included an element of daily or weekly repetition, but they also notably came with discussions about its mixed or negative valence. As this social worker explained, a habit can be either good or bad and it is “something you do, it may not be the best thing to do, but it has become part of your life, it is controlling you. … It may be a habit to take your pill at a bad time” (P 3). In some instances, participants suggested that habit may indicate someone’s character, especially when it is a bad habit, as this clinical leader noted: “when you say habit it means a character, any character of a certain person that is his habit. He comes late, that is his habit, he does like this. It is both negative and positive” (P 6). Expert patients and treatment initiators in particular emphasized the negative connotation of habit, as illustrated by this quote from an initiator, who used examples of criminal behavior to make the point: “A habit is bad, it is something bad you do, for me I would have understood it like that, like stealing, abusing people, or you have sat badly that is a bad habit too, you look badly, you have a bad habit” (TI 7). Considering that our original study flyer repeatedly used the word “habit” when describing the anchoring strategy and the value of ART adherence habits, this was important and formative feedback that we have integrated into our RCT study materials.

Treatment Initiation Procedures

Interviewees provided helpful information about how patients at the clinic are typically being initiated on their ART treatment. Providers described that initial instruction about the medication regimen does not focus on the location where patients may take their pills, so long as they take it at the same time every day. That is because it may be difficult for clients with hectic schedules to anticipate where they may be at a given time. As this nurse put it, “You can’t say the same location, people work, you know” (P 2), while a clinic leader noted “the time is almost fixed. If you say I will be taking at 10, then you adhere to 10, it doesn’t matter whether you are home or you are at a work place” (P 6).

Treatment initiators also recalled being told to pick the time that works for them and make sure they take their pills every day at that same time, although they seemed to have received little instruction on how to make sure they take their pills in that manner. This quote was typical of most clients: “He asked me, how do you prefer to take your medicine, at what time, things like that. And he told me the importance of following. You stick on time and he told me that even if you have missed your time, any time you remember you take” (TI 1). This finding points to a need to provide more instruction on the ways of establishing a healthy pill-taking habit that is not purely time based.

Acceptability and feasibility

Providers agreed on the importance of helping clients build ART pill-taking habits, and also on the need to harness technological advancements to help clients sustain healthy pill-taking habits. There was also agreement on the proposed intervention’s acceptability and feasibility in the context of the clinic and the patient population it serves. Moreover, providers and patients offered practical recommendations with regards to the use of text messages to form habits. For example, they described how sharing mobile phones within households or among friends was a common practice in Uganda, therefore suggesting that prior to being enrolled in the study, clients ought to be asked about their level of comfort with receiving health messages on their phone.

Perceived Role of Rewards

All participants were asked about their perceptions of the role of incentives in supporting context-dependent repetition. Among providers and expert patients, the views were mixed, with many (n=7 providers and n=3 expert patients) fully embracing the use of financial rewards as external motivation for taking pills in response to a contextual cue, explaining that some patients would regard the reward as a mark of appreciation for their efforts and an indication that providers really care about patients’ success. However, a few (n=3 providers and n=2 expert patients) noted that incentives can do little to mitigate existing social problems, such as food insecurity and stigma, and might even distort client motivations, as this counselor explained: “It will weaken the client’s mind, in that I am going to take my medicine simply because I want that money, not targeting good health. The moment you stop giving them money, they will stop, thinking that they are hurting you, yet they are hurting themselves.” Among treatment initiators, the predominant view was positive (n=19), framing these rewards as a source of relief against food insecurity and transportation: “It excites me somehow because I wouldn’t be expecting it. It can help me because I can withdraw it, I get something to eat and take the medicine.” Nontheless, the link between the rewards and motivating context-dependent repetition was less clear in the treatment initiator narratives.

DISCUSSION

While habits are often mentioned as key strategies for successful chronic disease management, many HIV patients in Uganda struggle to form ART adherence habits on their own (as demonstrated by the low rates of viral suppression) and the barriers and facilitators of the anchoring intervention for promoting context-dependent repetition and establishing habits have not been investigated in this setting. To address this knowledge gap and to refine a subsequent RCT to test an anchoring intervention among ART treatment initiators at Mildmay in Uganda, we conducted qualitative interviews with treatment initiators, expert clients, and providers to find out what they think can help or hinder the success of our ART adherence habit formation intervention.

Our qualitative work uncovered several important barriers of context-dependent repetition through anchoring that have not been reported in studies conducted outside of SSA and that may need to be considered when developing an anchoring intervention in this setting. First, several patients reported having inconsistent daily routines and/or having jobs that required unexpected and extended travel away from the home. For these patients, the lack of a consistent daily routine in the same context threatens the success of anchoring intervention, which relies on participants being able to identify an existing and stable routine (often at home) to pair with ART pill-taking. In the absence of an existing routine at home, even if participants have sufficient motivation, self-regulation, and initial ART adherence, they may struggle to repeat ART adherence in the same context and thus will not establish a habit. It will be crucial for us to evaluate how our intervention impacts participants with these inconsistent daily routines, and future research should also aim to develop alternative habit formation intervention approaches that can help participants with inconsistent routines identify a stable context for forming a habit.

Our interviews also provided important reminders about the barriers of stigma, food insecurity, and the COVID-19 pandemic for maintaining context-dependent ART adherence in our research setting. For example, opportunities for anchoring pill-taking were reduced when patients had to manage stigma among family members or friends. Specifically, stigma makes it difficult to use alarms or visual reminders as adherence anchors, which are often used in other anchoring studies for less stigmatized chronic conditions (Stawarz et al., 2020, 2016) because many HIV patients hide their pill-bottles or take the pills out of the original container to be able to carry them around without arousing suspicion. Food insecurity is another important barrier to context-dependent ART adherence since many patients reported a desire to take ART medications with food to avoid nausea, and unfortunately, the prevalence of food insecurity is relatively high in Uganda. The COVID-19 pandemic exacerbated the rates of food insecurity in Uganda, with many individual losing employment, and the pandemic also introduced new barriers to context-dependent pill-taking, such as being confined to overcrowded homes with relatives who may not be aware of a patient’s diagnosis throughout the strict lockdowns in Uganda. It is imperative that our RCT consider these additional factors (i.e., stigma and food insecurity) when evaluating our intervention’s efficacy, which will likely have different effects based on the number of these barriers a given patient is experiencing when trying to take their ART pills in the same context.

Additionally, our qualitative work uncovered facilitators of ART adherence habits that are similar to those reported in the literature for interventions targeting other medication adherence habits. Specifically, patients reported a preference for pill-taking in the privacy of their own home (Stawarz et al., 2020, 2016) and that patients’ ability to understand and manage their HIV symptoms and medication side-effects would be important motivators for building cued pill-taking habits (Emilsson et al., 2017; Mutanana et al., 2020; Tedla and Bautista, 2016). Additionally, both providers and expert patients were optimistic about the potential benefits of external rewards for supporting context-dependent repetition, which helps to further motivate the design of our subsequent ART adherence habit formation intervention. Finally, some patients reported being able to successfully use phone-based or clock alarms and other personally relevant reminders that had helped them overcome forgetfulness when starting ART, which aligns with findings in other settings where stigma is less common (Stawarz et al., 2020).

The findings also confirmed that the proposed theory-based intervention was perceived to be acceptable and feasible in this setting, but identified several ways in which our proposed methods need to be tailored for PLWHA in Kampala, Uganda. Specifically, participants offered useful guidance on linguistics when introducing habit formation to clients at the clinic, as well as important cultural information. For example, for the two intervention groups, the intervention relies on consistent personal access to cell phones, which may conflict with the local practice of sharing cell phones with family and friends. For our planned intervention, this means that participants must confirm that they are comfortable receiving health-related reminders on their cell phones. Additionally, we learned that our reminder messages needed to be ‘coded’ (i.e. do not mention words such as HIV, ART, or medication pill-taking). These will help participants protect their HIV treatment status. A final important intervention design lesson was to change the word “habit” to “routine” in all study fliers and other instructional materials, as routines have a more positive connotation in this setting. This is an important revision to our initially planned study materials that will help to mitigate the potential harm of our language, and suggests that further revisions to the language commonly used in the health habits literature should be considered. For example, habit researchers should consider replacing the term “adherence” with “routinization,” which focuses more on the psychological processes of habits as opposed to a broader range of barriers that may be outside of an individual’s control. More broadly, our findings underscore the importance of utilizing in-depth interviews to help researchers more fully understand issues from prior literature, to critique pre-existing theoretical constructs, and to incorporate the role of culture and dynamic life contextual factors into the application of theory-based interventions in new settings (Bishop, 2015; Kiwanuka et al., 2021).

Strengths and Limitations

This study benefited from our ability to randomly sample patients and providers at Mildmay, which increases the representativeness of our findings to other clients of this facility, but this research is not without limitations. First, the experiences of the studied sample may not capture all experiences of people in the broader population. However, Mildmay provides HIV care services to people both in and around Kampala and these services are free of charge to all patients, so we believe our results are likely representative of low-income residents of most urban and peri-urban areas of Uganda. Second, the analytical approach is a rapid analysis rather than full coding, which means there were fewer opportunities for iterative team discussions around thematic interpretation and fewer opportunities for cross-checking facts to reduce coder bias (Johnson and Vindrola-Padros, 2017). The advantage of our approach is that we were able to conduct the formative phase in a timely fashion to directly improve our subsequent RCT with minimal delay.

CONCLUSIONS

In one of the first qualitative studies investigating the barriers and facilitators of the anchoring interventions for establishing ART pill-taking habits among HIV-positive patients initiating treatment in Uganda, we find both common factors to previous studies as well as factors unique to this setting. Specifically, participants reported a preference for pill-taking at home, which has been commonly observed in other medication adherence studies. Importantly, several participants reported having inconsistent daily routines, confronting stigma within the household, and combating food insecurity, all of which need to be addressed in successful anchoring interventions in this setting. These barriers could also limit the success of similar habit formation approaches that promote context-dependent behavioral repetition in other resource-constrained environments. We also determined that important linguistic changes were needed to the instructional materials and reminder messages in the subsequent intervention to avoid words and phrases with negative connotations for this target population. These findings underscore the importance of understanding local barriers and facilitators when designing and planning interventions, particularly when implementing theory-based intervention approaches that have yet to be tested in a new setting.

Highlights.

• Context-dependent pill taking is threatened by stigma and food insecurity in Uganda.

• Taking pills in one’s home facilitates contextually-cued repetition.

• The COVID-19 pandemic exacerbated existing barriers to context-dependent pill taking.

• We identified regional barriers and facilitators to context-dependent repetition.

• Culturally adapting theory-based interventions is crucial for success.