Adherence to antiretroviral therapy (ART) is a critical factor contributing to low rates of viral suppression among youth living with HIV. At the end of 2015, there were approximately 60,300 youth living with HIV (YLH) aged 13 to 24 years in the United States [1], and this number continues to increase. In 2016, YLH made up 21% of new HIV diagnoses—the equivalent of 8451 new cases in that year alone [2]. National treatment guidelines recommend initiating treatment with ART as soon as an individual is ready [3]. Even with simpler, more potent, and better-tolerated medications, viral suppression is difficult to achieve in youth. Only 41% of YLH in 2014 received HIV-related care, 31% were retained in care, and 27% were virally suppressed, the lowest percentage of any age group [1].

Adherence to ART is critical to sustain health, reduce transmission [4], and minimize the development of ART resistance [5]. Nonadherence has both public health implications and personal health risks. High viral load (VL) increases the likelihood of viral transmission [6], and nonadherent individuals may be more likely to transmit drug-resistant strains of the virus [7]. Conversely, treatment adherence decreases the pool of infectious individuals when condom use is poor, even among youth aware of their HIV status [8].

The Centers for Disease Control and Prevention (CDC) released a statement in 2017 that further underscored the importance of ART adherence and VL suppression. The CDC’s statement acknowledged that individuals living with HIV who have an undetectable VL are not able to transmit the HIV virus [9]. This is commonly known by the slogan launched by the Prevention Access Campaign: Undetectable=Untransmittable, or U=U [10], and has ushered in a new era of treatment as prevention for HIV as well as contributed to reduced stigma for those living with HIV [11].

Youth, in particular, are frequently shown to be poorly adherent to ART [12-15]. In a study with ethnic and racial minority YLH, 39% reported suboptimal adherence [15]. In a recent review, only 51% of youth on ART achieved viral suppression of less than 400 copies/mL [16]. Data from the largest cross-sectional study to date found that only 37% of perinatally infected and 27.1% of behaviorally infected youth were virologically suppressed [17]. Of the YLH in the study, 30.5% reported engaging in unprotected sex in the last 90 days and 74.4% had detectable viremia. In addition, young men who have sex with men who have detectable VLs have been found to have higher rates of serodiscordant, condomless anal sex than those with undetectable VLs [18]. Finding effective adherence interventions for YLH to reduce HIV transmission is a public health imperative, and improved ART adherence is a particularly promising avenue.

A growing body of research has explored predictors of adherence to ART in YLH. Depression and anxiety have been consistently associated with poor adherence to ART. Adherence should be considered within the broader contextual issues present in the lives of youth, such as HIV stigma and disclosure, peer relations, and mental health and substance use [13,19,20]. Ethnic or racial minority status [15,21,22], lack of social support [15,22,23], and low self-efficacy [15,24] have also been associated with poor medication adherence in youth living with HIV. Poor medication adherence has been linked to financial and structural challenges such as housing instability [25] and lower level of education [26]. MacDonell et al [15] explored social cognitive predictors of ART adherence in a large, multisite sample of racial and ethnic minority YLH. Results indicated that social support, self-efficacy, psychological symptoms, and substance use were predictors of adherence. Taken together, this body of literature suggests that adherence interventions offering social support and promoting self-efficacy, while aiding with problem-solving of contextual barriers, may be beneficial.

Although it has been abundantly clear for over two decades that adherence to ART is a critical problem, the field has identified only a limited number of modestly successful interventions for YLH, including motivational interviewing [27,28], directly observed therapy [29], multisystemic therapy [30], and cell phone reminder calls [31,32]. These failed to demonstrate lasting impact on VL beyond the intervention. All the interventions, except the cell phone reminder pilot studies [31,32], require in-person sessions, which may be difficult for nonadherent youth to complete. Interventions that reach youth more frequently, using modern and youth-friendly means of intervention delivery such as cell phones, may hold promise for improving both short- and long-term adherence.

Cell phones may be a powerful tool for HIV prevention and treatment intervention [33]. Interventions delivered via cell phone may offer an advantage over traditional, in-person interventions in cost, flexibility, and ease of adapting the intervention to the participant. Our pilot study found that daily phone call reminders, with incentives for 80% intervention adherence, were both acceptable and feasible for nonadherent youth [32]. Cell phone intervention delivery allows for tapering of the frequency of calls or texts in response to the needs of each participant, which could help sustain the impact of the intervention over a longer period of time for a lower cost. However, few interventions utilizing cell phones as a stand-alone mode of intervention (vs as part of a larger intervention) have been tested, and even fewer have targeted YLH.

Electronic reminders, particularly text messages, have been found in the short term to impact health behavior, including medication adherence of adults with a range of chronic conditions [34,35]. Text message adherence reminders have demonstrated some success in adults with HIV in Africa [36,37]. Recent studies have demonstrated the efficacy of text messages for adherence support and retention in care among US adolescents and young adults living with HIV, although the majority of studies were pilot studies [38,39]. An uncontrolled pilot study of 25 moderately nonadherent youth aged 14 to 29 years (baseline adherence=75%) using personalized, interactive, daily text messages demonstrated significant improvements in self-reported adherence in a 24-week intervention but lacked the power to detect VL changes [40]. Building on this pilot work, a randomized, controlled crossover study [41] utilizing personalized text messages in moderately nonadherent (missed 1+ dose in the last week or 4+ in the last month) youth aged 16 to 29 years found significant improvement in those reporting >90% adherence at 6 months compared with controls (60.5% vs 51%); improved adherence was maintained at 12 months [41].

Although interventions using text messaging to improve adherence in YLH show promise, cell phone–delivered interventions are not limited to text messages. Cell phone–delivered interventions that use voice calling may offer some of the advantages of text messaging (eg, youth-friendly delivery and modern applicability) but retain more of the important elements of human interaction (eg, social support and alliance). A recent qualitative study supports the use of cell phones as a strategy to maintain adherence to antiretroviral refill appointments at a public HIV clinic in Nigeria [42]. A pilot of cell phone support with incentives (CPS-I) in a cohort of highly nonadherent youth demonstrated significant improvements in adherence and VL during the 24-week intervention and 24-week postintervention [43]. Although promising, this study was small, and 7 of the 19 intervention subjects were unable to adhere to the strict intervention requirements [43].

This study utilizes a Sequential Multiple Assignment Randomized Trial (SMART) design. SMART designs have numerous advantages over traditional trial designs. These are among the newest generation of improved clinical trial designs and methods and are used to inform the development of adaptive treatments or interventions [44,45]. This design is a cost-effective and methodologically rigorous way to maximize clinical utility and real-world implementation in the resulting adaptive intervention. Adaptive interventions are interventions in which the type or dosage of the intervention (eg, number of texts sent) is adjusted based on participant characteristics or response (eg, VL). Thus, an adaptive intervention allows treatment to be tailored to the specific needs of each participant. A SMART design involves multiple intervention stages, with each stage corresponding to one of the critical decisions involved in the adaptive intervention.

The intervention is guided by the conceptual model of supportive accountability [46]. This model was developed to guide research into human support components of mobile health (mHealth) interventions. The model is based on the premise that human support increases adherence through accountability to a coach (in the intervention, an adherence facilitator or AF) who is perceived as trustworthy, knowledgeable, and benevolent. Accountability should involve clear, process-oriented expectations that the patient is involved in determining (eg, reporting adherence, problem-solving). The effect of accountability may be moderated by patient motivation so that patients with higher intrinsic motivation may actually require less support.

The process of support is also mediated by the mode of communication (eg, phone, text messages, and computer), with different advantages and disadvantages for each mode. There is evidence that “lean media,” or those modes of communication with less face-to-face contact and fewer visual social cues, may be associated with more positive, even idealized, attributions of communication partners. This is because people tend to form stronger impressions based on more limited social and interpersonal cues [47]. Interactions via lean media, including CPS and text messages, have the potential to foster social accountability toward improved adherence.

Multiple studies have demonstrated that social support is a strong predictor of good adherence to ART [43,48-50], and retention to HIV care is predicted by clients’ perceptions of providers as engaging and validating [51]. Although overall social support was predictive in these studies, specific aspects such as instrumental support (ie, practical assistance) and informational support (advice or problem-solving) were found to be predictive of adherence in adults with HIV [50]. The content of conversations in the CPS study [43] is designed to validate the importance of adherence, prompt problem-solving (through informational support), and provide instrumental assistance (through referrals) to address barriers as they emerge. This intervention utilizes social support constructs to provide tailored conversations to improve both short- and long-term adherence. Social support theory also suggests that an ongoing alliance could be protective against depression, substance use, and general stress [52-55].

It is increasingly important in the face of competing demands for health care resources to establish not just the efficacy of interventions but also their relative economic value. The National Institutes of Health (NIH) established cost-effectiveness analysis as a key priority in 2015 [56]. By integrating cost-effectiveness analysis into a SMART trial, we can simultaneously determine not only the most effective sequences but also whether the added costs of potentially more effective sequences are worthwhile. Within an implementation science framework, costs include those associated with executing implementation strategies as well as those associated with service delivery as uptake changes [57]. Understanding the potential barriers and facilitators to intervention sequences for widespread implementation can provide additional critical information about treatment sequences. Many have argued that the science-practice gap is inflated by the bias toward step-wise progression of research from development, to efficacy, to effectiveness, to implementation, and that hybrid designs can maximize clinical implementation earlier in this developmental process [58]. Thus, we can greatly enhance the potential for intervention scale-up by comparing interventions in real-world settings and identifying effective sequences of treatments and their cost-effectiveness ratio while simultaneously studying the context of implementation in a Hybrid 1 Effectiveness-Implementation trial [58,59].

In addition, helping people living with HIV maintain an undetectable VL is in itself a way to reduce sexual transmission of HIV, as highlighted by U=U. Therefore, not only is this intervention a potentially cost-effective way to achieve VL suppression in YLH, but it may also prevent future cases of HIV transmission to those who are HIV negative. This potentially saves the health care system hundreds of thousands of dollars per person. A recent study found that the medical cost saved from preventing a single HIV infection is US $229,800 [60]. This rises to US $338,400 if those living with HIV entered into the care cascade early on and stayed in care [60]. This figure grows higher when taking into account secondary infections avoided and societal costs saved (eg, social services and housing, patient and family time, productivity, physical tolls, and emotional distress) [60]. It is clear that VL suppression is a key part of saving medical costs related to HIV care, both for the patient themselves and for the public.

The aim of this study is to test an adaptive adherence intervention, which utilizes 2 mHealth intervention designs, in an effort to promote adherence to ART and maintain VL suppression in YLH from across the United States. Both interventions are delivered remotely, utilizing a central research center. In addition, this study aims to increase understanding of the context for wide-scale implementation of this type of intervention as well as to understand the benefit of incentives for nonresponders and tapering of interventions for responders.

The primary hypotheses are as follows:

Secondary aims include assessing the following concepts, all central to the Scale It Up program described in this issue [61], within the different intervention sequences within the SMART design: (1) cost-effectiveness; (2) the 5 components of the Self-Management Model over time (see ATN: Scale It Up overview paper in this issue); and (3) the barriers and facilitation throughout Exploration, Preparation, Implementation, and Sustainment (EPIS) Model phases (identified in ATN 153 EPIS protocol paper in this issue).

This study is part of the Scale It Up program as described in the overview paper in this issue [61] and employs a SMART design with repeated measures (NCT03535337). Approximately 190 YLH participants are being recruited, consented, enrolled, and randomized to either CPS or SMS (short message service). CPS is provided on all weekdays except holidays. Weekend and holiday adherence are assessed at the first CPS contact with the participant following the weekend and/or holiday. SMS is provided every day, including weekends and holidays. Both intervention groups receive 3 months of intervention, followed by a second randomization, which is be based upon response to the intervention.

Participants who respond to the intervention, that is, those with a suppressed VL (<200 copies/mL) in either group, are randomized to receive either 9 months of follow-up without intervention (standard care) or 3 months of tapered intervention (CPS-T or SMS-T twice a week) based on their previous intervention modality, followed by 6 months of follow-up without intervention (standard care; see Figure 1).

Participants who do not respond to the intervention, that is, those with a VL ≥200 copies/mL at the 3-month study visit, are rerandomized to CPS or SMS; however, an incentive is added to both arms. This incentivized CPS or SMS (CPS-I or SMS-I) provides the opportunity to explore the role of incentives with each intervention modality. After 3 months of CPS-I or SMS-I, participants receive 3 months of tapered intervention (CPS-T or SMS-T twice a week) based on their most recent intervention modality followed by 3 months of follow-up without intervention (standard care). Those who are unable to provide a documented VL result at the 3-month time point are considered nonresponsive to the intervention and are rerandomized to an incentive arm.

All participants complete a Web-based, computerized survey, which measures self-reported adherence, substance use, depression, and other mediators and moderators of adherence at baseline and every 3 months thereafter until study completion (12 months for both responders and nonresponders). In addition, a VL result is obtained from participants or their care provider, or they can provide a blood sample for a study-sponsored VL assay.

Participants are compensated US $40 after completing the baseline and each follow-up assessment for a total of US $200 for all study assessments. Participants in the CPS-I or SMS-I arms who reach 75% monthly adherence to calls or text responses receive an additional US $50 for up to 3 months (up to US $150). As all activities are conducted remotely, compensation is provided via electronic gift cards (eg, Amazon, Target, Walmart).

The AFs are staff members who are not licensed professionals or graduate students who have completed clinical externship training. AFs should (1) be able to interact and engage with youth; (2) be knowledgeable about HIV infection and its treatments, including side effects and their management; (3) be skilled in interpersonal communication and be able to display empathy for the participant; (4) trained in motivational interviewing; (5) maintain a professional relationship and not invade the personal boundaries of the participant; and (6) be trained in providing a link between the participant and their main resource for care, as a means of providing referral services for necessities and counseling, if needed.

All AFs attend a 2-hour webinar designed to familiarize AFs with the purpose of the study, definition of adherence, role of the AF, HIV basics, cultural humility, building rapport and effective communication, legal issues, and protocol review. Following the training, AFs participate in a minimum of 2 telephone role plays during which they practice an initial phone call with a participant and a daily phone call. These training calls last about 30 min, including the role play and feedback. AFs are also encouraged to practice role plays with each other to obtain not only more time in practice but also to experience the role of being the participant during the cell phone support call.

A fidelity evaluation is conducted to assess whether the AF is collecting the correct information and providing an appropriate level of support. This is to ensure fidelity to intervention delivery and to detect areas of the AF training that need to be improved. A clinical coordinator conducts this fidelity evaluation.

All intervention phone calls between the AF and participant in this study are digitally recorded utilizing an amplifier and recorder, and all digital recordings are saved on a secure network for review. The clinical coordinator reviews 20% of all audio files for each AF during the initial 3 months. The audio files are randomly selected from all audio files saved by the AFs. If 90% of reviewed files are found to be adherent to the requirements listed above, the clinical coordinator only reviews 10% of all audio files after the initial 3 months. The recordings are assessed for adherence to the phone call script, advice, referrals, centering of participants in discussion, order of content discussed, and appropriate length of call.

This study is highly innovative in that it focuses on a critical area for intervention—adherence to ART in YLH—and goes beyond what has been done previously in several key ways: (1) it directly compares 2 successful, potentially sustainable, and youth-friendly modes of mHealth youth adherence intervention delivery; (2) the project uses a SMART design to explore a number of key issues surrounding sequencing mHealth interventions in a cost-effective manner that would be practical in clinical settings; (3) the CPS intervention uniquely addresses multiple barriers to ART adherence in real time using a social support framework; and (4) the study explores the clinical and public health costs of sequencing of interventions based on viral suppression and sexual transmission risk.

Technology-based interventions offer tremendous advantages for delivery of brief interventions in terms of reach, ease of replication, anonymity, and cost. They also have an advantage in functionality or how quickly and easily they can be modified for a specific population. mHealth assessments and interventions have been shown to be powerful tools through which brief, targeted interventions can be delivered to people with HIV or at risk of acquiring HIV [75-77]. Texting and cell phone voice calls, in particular, have been linked to improved adherence to ART [41,43,78,79]. However, no study to date has directly compared the use of SMS with CPS for intervention delivery and efficacy. There are currently no specific evidenced-based CDC-supported ART adherence interventions for youth. Cell phones are a convenient and culturally relevant mechanism for intervention delivery, and socioeconomically disadvantaged youth and ethnic and racial minority youth report high rates of access and use [43,80-82].

This innovative SMART design [83] not only compares the efficacy of the proposed CPS and SMS interventions but also explores other key issues such as the role of incentives in participant engagement, the impact of tapering intervention frequency, and the potential cost-effectiveness of sequencing the interventions without incentives before providing incentives. This design allows us to explore potential benefits of sequencing our interventions in a way that mimics what would likely happen in a clinical setting (eg, start with the less intensive, less costly interventions, and if that fails, move to the more intensive and costly intervention with incentives). Finally, the design allows us to test an additional secondary aim, whether tapering the interventions and removing incentives will maintain improved adherence in those that reach a suppressed VL (in both arms). All youth adherence intervention studies to date have found that adherence declines once the intervention is over, and tapering an adherence intervention in youth has not yet been done. This will offer clinical sites a less intensive, low-cost option that could be employed indefinitely in highly nonadherent youth.

Another key intervention is the way the study overcomes the numerous barriers to ART adherence (eg, forgetting, not feeling like taking ART, medication reminds me of HIV) [84]. Effective adherence interventions need to address forgetting, but also promote motivation, reinforce positive reasons to take ART as prescribed, and offer solutions to barriers as they arise or change over time [85]. The CPS intervention is based on the model of social accountability, so that cell phones are used to provide instrumental and informational support, foster patient accountability to their AF, and promote real-time problem-focused coping.

A further key innovation is the use of a centralized model for recruitment and for providing the mobile intervention. The centralized model provides a single location in which multiple activities can occur. This can provide cost savings as staff members can be cross-trained to screen, recruit, and provide adherence support. This may be of benefit when case managers in individual clinics may be overloaded with providing direct care to their clients. In addition, nationwide recruitment through social media combined with the distribution of recruitment materials to HIV care providers such as Ryan White-funded clinics may also provide a cost-effective manner to recruit participants when it is getting more difficult to identify and recruit youth with an unsuppressed VL.

Finally, in the clinical setting, cost-effectiveness becomes critical to understanding how to implement sequences and tapering. Each of these adaptive intervention sequences has differing costs, and we will be able to help determine which intervention might be best for which groups (eg, behavioral acquisition vs perinatal acquisition, substance abuse, depression). From a broader public health perspective, it is clear that youth nonadherent to ART are frequently the same patients engaging in other risky behaviors such as condomless anal sex [18]. Thus, a successful intervention such as this one has the potential to lower the risk of HIV transmission. The assessment of sexual risk and viremia will allow us to estimate the cost benefits from reduced HIV transmission for each sequence.

There are some limitations to this study. Participants are not required to verify that they have been on ART for at least 3 months for study enrollment. Although participants must submit proof of an ART medication bottle or prescription, this does not necessarily have to be from at least 3 months ago. It is possible but unlikely that a subject would have inadequate time to suppress VL under 200 copies/mL if use has been less than 3 months.

This study also relies on mHealth intervention methods, specifically through phone calls and text messages, as a way to improve adherence. The requirement to be the sole owner of a device capable of sending and receiving calls and text messages, although necessary for confidentiality, may pose a barrier to some study participants. As a result, this study may not be accessible to all YLH, particularly those who are low income and do not have a mobile device or who may share a mobile device.

Finally, participants are required to be able to communicate in English. This means that YLH who are not yet fluent in English, such as those who have recently immigrated or who live in areas that are primarily non-English speaking, would not be eligible for this study. However, these are populations which could also benefit from ART adherence interventions if they were made to be accessible, and it is plausible that cultural differences in implementation may be missed.

Overall, this dynamic and adaptive study shows great promise. It offers a highly innovative SMART design to assess the efficacy of 2 mHealth interventions to promote adherence to ART.