Healthcare Transition for Youth Living With HIV: Outcomes from a Prospective Multi-site Study

: Purpose: Youth living with HIV (YLHIV) in the United States (U.S.) account for nearly one-third of new HIV infections and face significant barriers to care engagement; only 25% are virally suppressed. Healthcare transition (HCT) from pediatric/adolescent to adult-oriented care can be particularly disruptive. Accordingly, we prospectively examined HCT processes at 14 distinct geographical sites across the U.S. Methods: We collected Audio Computer-Assisted Self-Interviews data and abstracted electronic medical records from 135 HCT-eligible YLHIV at baseline and 9-month follow-up. Descriptive analyses and multilevel modeling were conducted. Data also included qualitative interviews with 28 adolescent and 30 adult providers across 14 adolescent and 20 adult clinics, respectively. Interviews were analyzed using the constant comparative method; this analysis focused on specific HCT recommendations. Results: At baseline, youth were primarily age 24 (78.8%), male (76.8%), black (78.0%), identified as a sexual minority (62.9%), had attended an HIV appointment in the past 3 months (90.2%), had Medicaid for insurance (65.2%), and were always or mostly always adherent to their antiretroviral therapy (65.9%). At the 9-month follow-up only 37% of YLHIV

YLHIV. The study incorporates mixed method data from YLHIV as well as adolescent and adult clinic providers, all key players in HCT success, to provide a comprehensive and nuanced description of experiences with transition of YLHIV.

Study overview
Data for the mixed method Comprehensive Assessment of Transition and Coordination for HIVinfected Youth as they Move from Adolescent to Adult Care (CATCH) study were collected from YLHIV and pediatric/adolescent and adult clinic providers across 14 Adolescent Medicine Trials Network for HIV/AIDS Interventions (ATN) sites across the United States (U.S.) [4], [14]. Participants received a $25 gift card. Institutional review boards at the University of North Carolina Greensboro and participating ATN sites approved study protocols.
Youth data YLHIV were eligible if they were behaviorally infected, preparing to transition to adult HIV care within 6 months (as reported by their healthcare provider), spoke English or Spanish, and received care at an ATN clinic. Youth provided informed consent and signed a bidirectional release of electronic health record information. Study staff recruited 156 adolescents; 21 declined to participate. Audio Computer-Assisted Self-Interviews (ACASI) data were collected at the adolescent clinic from the 135 YLHIV at baseline (August 2015-February 2016) and at 9-month follow-up (May-November 2016). Electronic health record data were abstracted to confirm whether a youth had successful HCT (defined as at least one adult clinic appointment during the study period) and viral load (from the laboratory results closest to the ACASI data collection). YLHIV data in these models were collected at baseline to predict HCT at follow-up.
Individual-level HCT-related variables [16] included demographic, psychosocial, care engagement, and adolescent clinic services. Demographic variables included being 24 years old (typical HCT age at these clinics [20]) versus younger, current male gender versus nonmale gender (i.e., female or transgender woman), Hispanic versus non-Hispanic, and no insurance versus insurance. Adherence to antiretroviral therapy was dichotomized as always or mostly always (≥95%) adherent versus less consistent adherence using a validated selfreport measure [21]. Psychosocial variables included the two-item disclosurerelated stigma scale (Likert scale: 1/strongly disagree to 4/strongly agree): "I am very careful who I tell that I have HIV," and "I worry that people who know I have HIV will tell others." Higher summed scores indicated higher levels of perceived stigma. Psychological distress was measured by the Brief Symptoms Inventory Scale. Raw scores were summed across 18 items related to feelings of depression (e.g., feeling blue), somatization (e.g., faintness or dizziness), and anxiety (e.g., feeling hopeless about the future) in the past 7 days, then converted to T-scores (M = 50, SD = 10). T-scores greater than 64 denoted psychological distress. YLHIV care engagement was defined as HIV appointment attendance in the past 3 months. Services YLHIV received in the adolescent clinic were dichotomous (yes/no) and included whether staff discussed potential transition challenges, provided support in choosing adult clinic, and discussed differences between adolescent and adult health systems.

Clinic-level data
Researchers conducted 58 semi-structured telephone interviews from August 2015 to June 2016 with adolescent clinic (n = 30) and adult clinic (n = 28) medical and social service providers directly engaged with HCT. Interviews occurred across 14 adolescent and 20 adult clinics, averaged 45 minutes in length (range = 22-78 minutes), and were professionally transcribed. The interview guide addressed processes, factors, and recommendations specific to HIV-related HCT [4].
Data also include a clinic checklist (adapted from the National Diabetes Education Program Transition Planning Checklist [22] and American Academy of Pediatrics [23] HIV Transition Recommendations) to capture clinic specific HCT-related strategies (e.g., life skills training and adolescent clinic staff attending the first adult clinic appointment of YLHIV), logistical characteristics (e.g., adolescent and adult clinic/provider associated with the same institution), and comprehensive patient services (e.g., mental health services, support groups, and/or reproductive health services).
Two HCT "Best Practices" scales were created from checklist data. The first, adolescent clinic best practices, included eight items from 13 adolescent clinics (one clinic was excluded as it did not transition any YLHIV during the study). Items were scored as yes/no and included (1) having a HCT protocol; (2) having a specific "transition" staff person; (3) having an adult provider come to the adolescent clinic to treat/meet youth before HCT; (4) having an adolescent clinic staff attend a youth's first adult clinic appointment; (5) providing information and support to youth regarding adult clinic options; (6) providing youth with information about insurance protocols (e.g., making insurance claims); (7) having one adult clinic versus multiple adult clinics available for HCT; and (8) whether the receiving adult clinic was within the same clinical space/medical system or a different medical system. Responses were summed and scores ranged from 0 to 8; the scale was dichotomized to compare adolescent clinics that practiced seven or eight best practice strategies versus adolescent clinics that practiced six or fewer strategies. Similarly, the adult clinic best practices scale included four items and asked whether an adult clinic had (1) a clinic HCT protocol; (2) a specific HCT staff person; (3) an adult provider who goes to the adolescent clinic to treat or meet youth prior to transition; and (4) an adolescent clinic staff who attends the first adult clinic appointment with youth. Responses were sum scored and dichotomized to compare adult clinics that practiced all four strategies compared with three or fewer.

Quantitative analyses
Descriptive statistics assessed youth and clinical characteristics. Youth were nested within specific clinics necessitating a multilevel modeling approach; mixed effects logistic regression was used to test associations between clinical practices and successful HCT of YLHIV. We used a stepwise approach to the multilevel model building [24]. Model 1 included a fixed and random intercept to assess if successful HCT significantly varied across clinics. We also computed intraclass correlations to warrant examining differences between clinics in HCT success using the following formula: where τ00 equals the log odds of YLHIV having a successful HCT. Model 2 added YLHIVspecific individual-level fixed effects to examine the relationship between individual-level factors and successful HCT. To assess the extent to which individual-and clinic-level variables explained additional variation in successful HCT, Model 3 includes Model 2 plus adolescent clinic best practices, and Model 4 includes Model 2 plus adult clinic best practices. We assessed change in −2 log likelihood (−2LL) across models; the best model was classified as the model that explained the most difference from Model 1 in −2LL change.
Given the relatively small number of clinics (N = 13), we used Fisher's exact and chi-square tests to compare adolescent and adult clinic best practice strategies across clinics where 50% or more of youth had successful HCT to clinics and where less than 50% of youth had successful HCT. We used the 50% cut-off, as it is higher than existing chronic disease HCT rates [25] and captures a meaningful target for care continuity for HIV-related HCT.
All analyses were conducted in SPSS version 24.0 (Armonk, NY) and SAS version 9.4 (Cary, NC). Statistical significance was defined as p <.1 to detect any potentially relevant effects despite a small clinical sample size.

Qualitative analyses
We examined staff's HCT-related recommendations using the constant comparative method [26], [27]. Two researchers independently conducted line-by-line coding on each transcript to create a codebook. This codebook was then summarized and refined into a coding matrix, which also incorporated themes from existing literature [28]. Two researchers applied the finalized codes to all transcripts using Atlas.ti 7.5 (Atlas.ti, Berlin, Germany), with greater than 90% inter-rater agreement. Coding disagreements were resolved through research team consensus. Table 1 highlights youth and clinic characteristics. A total of 132 youth were included in final analyses (three were excluded because of missing data on key variables). The majority of youth were age 24 (n = 104, 78.8%), male (n = 100, 75.7%), black (n = 103, 78.0%), identified as a sexual minority (n = 83, 62.9%), had attended an HIV-related care appointment in the past 3 months (n = 119, 90.2%), had Medicaid (n = 86, 65.2%), reported ≥95% adherence to their antiretroviral therapy medication (n = 87, 65.9%), and were virally suppressed (n = 92, 69.7%). Fewer youth identified as Hispanic (n = 24, 18.2%) or reported being distressed (n = 15, 11.4%). On average, youth reported moderate disclosure-related stigma scores (M = 1.81 out of 4, SD = .82). Approximately 37% (n = 49) of YLHIV successfully transitioned to adult care. The majority of youth felt that adolescent clinics prepared them for HCT: 83.3% (n = 110) reported that adolescent clinic staff discussed potential HCT challenges and provided support in choosing where to receive adult care, and 80.3% (n = 106) stated that adolescent clinic staff discussed differences between adolescent and adult health systems.

Individual and clinic characteristics
Among the 13 adolescent clinics, five employed all best practice strategies, whereas four of the 20 receiving adult clinics employed all best practice strategies.
Mixed effects logistic regression models Table 2 summarizes the mixed effects logistic regression model results. Model 1 (HCT assessment by site) indicated that successful HCT varied across adolescent clinic sites (τ00 = .99, z(12) = 1.55, p = .06), explaining about 23% of variability in the successful HCT rate. In Model 2, the addition of individual-level characteristics of YLHIV improved the model fit by a 22.2 difference (τ00 = 1.56, z(12) = 1.51, p = .07). There was a negative relationship between disclosure-related stigma and successful HCT (b = −.79, p = .03); as youth's disclosure-related stigma increased, the predicted odds of successful HCT decreased. There were no other significant relationships.
In Model 3-which added adolescent clinic best practices-using seven or eight best practices was associated with successful HCT, improving model fit by a 3.73 difference from Model 2 and 29.93 from Model 1 (τ00 = 1.02, z (11) = 1.38, p = .08). YLHIV who attended adolescent clinics that used seven or eight best practice strategies were 4.84 times more likely to have a successful HCT (b = −1.58, p = .04) compared with clinics with six or fewer strategies. Perceived disclosure-related stigma remained negatively associated with successful HCT (b = −.83, p = .02). Youth without health insurance were 6.74 times more likely to have a successful HCT (b = 1.91, p = .04). In Model 4-which added adult clinic best practices-having all four best practices was not associated with successful HCT (b = .59, p = .57).

Fisher's Exact and Chi-square tests
The majority of adolescent clinics (n = 12, 92.3%) reported having an HCT protocol and an HCT staff person, providing youth with information and support related to options for adult care, and discussing insurance protocols with YLHIV. The majority also had a clinic staff member attend the first appointment with youth (n = 9, 69.2%) and had multiple adult clinic options (n = 7, 53.8%). Relatively few adolescent clinics had an adult provider come to the adolescent clinic to treat or meet youth before HCT (n = 4, 30.8%) or shared clinical space with adult clinic(s) (n = 3, 23.1%). There were no significant differences between these individual clinical variables and sites that transferred more than 50% of youth to adult care (see Table 3).    You know, if there's a life skills group somewhere in the process, whether it's on the adult side or the youth side, if there was a life skills group that they can join into so they can learn about that insurance, learn about budgeting or get the housing referrals before they even get deep into the clinic, I think that that would help as well. (Site X Clinic c1 (Xc1)-Adult Social Worker) Provide patient-centered care where expertise and experiences of YLHIV are acknowledged and valued in the HCT process We've learned to listen to them a lot more, let them guide how we start out transition process more so than us just deciding this is what needs to be done, this is what you should do, and this is how you do it. Because, of course, most youth have to have a buy-in. With most youth, it's kinda like even with little kids. They gotta think, "It's my idea." So an adult just telling me, "This is what you gotta do. Do this. Do this, do this." They tend to pull back a little bit or just don't wanna do it because you're telling them. But, if you can get them to feel like, "This is something I need to do. This is something I should wanna do. This is my idea. I'm determining how my care goes and my path," it tends to be a lot easier, a lot more smoother.

Clinic specific
Develop and implement a formalized HCT protocol …A structured protocol around transitioning, not just the, "Oh, well, you're about to turn 25 or 21, whichever year you transition, so let's figure out where you're going to go." It's really thinking about it early and taking the steps to make sure that the young person doesn't feel rushed in the process. (N2-Adolescent Social Worker) …The physicians themselves sort-of being more involved in the transition process…I mean, if money weren't an object, perhaps having one staff person, a social worker or someone specifically devoted to the youth. So sort-of have a transition within the transition where they transition to here…You'd have to have a physician who specifically wants to work with that population, the age group between just coming over from adolescent care to adult care. (Ra2-Adult Social Worker) Provide staff training on developmental stage and context of YLHIV to provide competent care I guess learning, training on how to be more…youth-friendly because they're adults, they're 25, but they're not 40 or 50 and really responsible like a 40-or 50-year-old…Just some kind of training on how to meet the young adult where they are and not expect them to be at a different level when they're not. (K1-Adolescent Social Worker) …An awareness that these kids, even though they're adults, they either may not be cognitively adults or that they are stunted…Or they're just going to have a hard time, bottom line, because of what's going on, as a normal young person trying to navigate life…And I think that there needs to be a little bit of flexibility… Just understanding that this is a true, special, and key population and having the flexibility to say,

Adolescent clinic staff
Adult clinic staff "Well, because they are having a hard time, let's try to be a little bit kinder, gentler in our approach." (Xb1-Adult Physician) Ensure patient and peer navigation are available at the clinic to support HCT for YLHIV I think that having youth find kind-of a representative or buddy in the clinic is key. Someone…that the youth is actually comfortable or has a good attachment to…Maybe the first person that they met. I think that that's something key to gain trust in the clinic and in the clinic setting. (T2-Adolescent Physician) I think that if your navigator is in his 60s, that's not-We have discussed having a younger youth navigator, peer navigators on site. So that's potentially something that we could do to make it more youth-friendly. (Wa1-Adult Physician) Provide comprehensive care services to support all healthcare needs of YHIV I think that it would be nice to have a lot more transition services. I know that we continue to offer…mental health services for six months after someone transitions…The one I'm thinking of [is] pharmacy support. Childcare. I think that there's just some additional services that would make it helpful. (Z3-Mental Health and Substance Abuse Counselor) …A lot of mental health services, even psychological services on site and social work on site. So I think those support services are critical because often, when the patients come in, if they have an issue, that can be the only time that we have to work with them sometimes, so to say, "Could you please come back tomorrow because I know my social worker will be here?" When they walk out the door, we've lost them for a while. And without having been able to help them. (Ia1-Adult Physician) Prioritize interclinic collaboration and data sharing to support HCT …Establishing a relationship with an adult case manager who is very hands-on early on. So someone who is youth-focused and who's willing to provide a little bit more intensive services…someone who's willing to follow up with the youth in the community, engage with them, with us, with the new provider, that's a big piece, and also having an adult provider who's collaborative…Really, what it comes down to is collaboration with the adult providers and being a little bit more flexible and youthoriented, especially during the initial transition. (Q2-Adolesent Social Worker) The other is similar data sharing that would be from one clinical entity to another. So, say [a youth] comes to me for care and then goes to [another clinic] for six months because they're closer…[The youth] appear like they're out of care, but they're actually in care. I don't know about that. And so those kinds of information sharing networks are critical at the network level…the systems level, to be able to then be in touch with who's out and who needs to be contacted and brought back in. (Xa1-Adult Physician) Similarly, the majority of adult clinics had an HCT protocol (n = 12, 92.3%) and an HCT staff person (n = 8, 69%). Fewer adult clinics reported that adult providers treated or met youth in the adolescent clinic before transition (n = 7, 38.5%) or that an adolescent staff person was present at the first adult appointment of YLHIV (n = 6, 38.5%). There were no significant differences between these individual clinical variables and the sites that transferred more than 50% of youth to adult care (see Table 4).

Healthcare transition-related recommendations
The adolescent and adult clinic providers focused the majority of their HCT recommendations and strategies at the clinic level (summarized in Table 5). These recommendations included staff training around adolescent development, creating formal HCT protocols, strengthening communication and data sharing between the adolescent and adult clinics, and offering more comprehensive HCT-related care services to support YLHIV. The few youth-specific recommendations centered on providing support and skill development for youth to manage their own healthcare (e.g., life skills and being assertive in the healthcare environment) and allowing youth to guide their own individualized HCT plan (e.g., listening to their unique needs).

Discussion
This prospective study across 14 ATN clinical sites highlights the complex set of individual-and clinic-level factors associated with HCT. Although being "transition eligible" was an inclusion criterion, the majority of YLHIV (63%) did not successfully transition. The youth who did not transition may have been unsuccessful for a variety of reasons. Some likely never left the adolescent clinic; in qualitative interviews, some providers reported keeping youth past the age of 24; others may have returned to care in the adolescent clinic despite being "ready" for transfer, and additional YLHIV were lost to care. This rate of successful transition-37%-is low, particularly given the importance of maintaining viral suppression for individual and public health [29]. This lower rate may be due to our follow-up occurring at 9 months as opposed to 1 year. It might also be that ATN clinical sites are particularly well-resourced, "one stop shops," which make it difficult for YLHIV to leave a clinical space that provides wrap-around and social services such as mental health, housing, and job training. More work is needed to comprehensively understand both the operationalization of transition "readiness" and the longer term HCT trajectory given the low rates of HCT and the high level of disengagement after 1 year [18], [19].
At the individual-level, disclosure-related stigma was associated with HCT. Youth who reported higher levels of HIV disclosure-related stigma (i.e., were more fearful of telling people they had HIV) were less likely to have successful HCT. Receiving care at a new clinic requires youth to repeatedly disclose to new individuals (e.g., receptionist, nurse, doctor, and social worker), a process that often involves sharing traumatic experiences. This forced retelling is necessary for a successful HCT, but for YLHIV who particularly fear disclosure, it may serve as a barrier to care engagement in a new clinic [30]. Research highlights the importance of a positive experience with first disclosure [31], suggesting a critical role for adult clinics in supporting YLHIV (especially those with fears around disclosure) during HCT. Further, adolescent HIV clinics are often situated within general adolescent clinics, so youth can attend without friends or family knowing why. In contrast, many adult HIV clinics are HIV-only, which may increase the chance that family, friends, and/or other community members of YLHIV might learn about their status through unintentional disclosure [2].
Not having health insurance at baseline was positively associated with successful HCT. This finding, although unexpected, may have occurred for the following reasons. First, of the 16 youth who lacked insurance at baseline, 50% reported having insurance at follow-up, which suggests that they were able to receive insurance through the adult clinic. Of the remaining eight youth who lacked insurance at baseline, 31.3% (n = 5) were lost to follow-up and 18.8% (n = 3) remained uninsured. Second, clinics who cover youth through Ryan White may be forced to transfer YLHIV before age 25, whereas those with private insurance have fewer restrictions. Lastly, perhaps because YLHIV without insurance are deemed particularly vulnerable, adolescent clinic staff may have engaged in additional measures (e.g., going with the youth to the first adult clinic appointment or, when possible, finding a clinic within the same medical system) to help facilitate their HCT. Future work should examine the role of insurance (e.g., parental insurance and disclosure) on HCT trajectories.
Clinic-level factors also influenced HCT success. Specifically, adolescent clinics that employed seven or eight best practices were almost five times more likely to have more than 50% of their YLHIV successfully transition to adult care. Individual practices were not significantly associated with HCT, highlighting the importance of "bundled services" to support YLHIV. This finding builds on previous research focused on YLHIV care linkage and engagement [32]-since HCT requires that youth re-link and re-engage. It further indicates that providers' qualitative recommendations focused on clinical-level factors such as data sharing, clinics' patient population, and the type of provider that completes the linkage/engagement are relevant to care continuity of YLHIV [32] and underscore the importance of engaging in concrete structural change within a clinic to ensure that processes exist to facilitate successful HCT of YLHIV.
HIV-related HCT is a complicated process especially within the different pediatric/adolescent and adult medicine subcultures [4]. The qualitative data identified useful HCT recommendations to support YLHIV that acknowledged the need of incorporating a developmental perspective within the compounding challenges of also living with HIV. Accordingly, the majority of recommendations focused on changes for adolescent clinics (e.g., less handholding, improved data sharing, and refinement of streamlined HCT processes) and/or adult clinics (e.g., staff training on adolescent development and more flexibility) to better support HCT. Many of these changes can be implemented across multiple clinic settings-some quickly and easily (e.g., provide basic life skills education). In contrast, some of the proposed structural changes (e.g., shared medical records) may be more laborious but could also dramatically improve care engagement and health outcomes for YLHIV [16], [30].

Strengths and limitations
Several study limitations should be considered. First, the participating ATN sites are in urban areas with relatively high HIV prevalence among adolescents. HCT may be affected by different issues in cities or in rural areas with lower HIV prevalence [33]. Second, the follow-up period was limited to 9 months, so it is possible that some of the YLHIV may have had a successful HCT within a year-or even longer. Thus, future work should extend follow-up periods to examine HCT trajectories, especially as the operationalization of HCT success is complicated and we measured only one (essential) component of HCT-"transfer" or adult care linkage [14], [16], [30]. Initial linkage does not ensure long-term engagement [8]: only about 50% of youth who transferred were retained in adult care after 1 year [18], [19]. Finally, although this geographically diverse study included 13 distinct clinical sites that supported HCT for YLHIV, this sample size is small for the purposes of multilevel modeling, which resulted in reduced power for examining the influence of clinical best practices. This suggests that our clinic-level findings (from well-resourced and HCT supportive clinics) are particularly conservative and that the actual effect size of employing seven or eight adolescent clinic best practices is likely much larger. This is the first study to prospectively examine HIV-related HCT from the perspective of YLHIV, as well as clinic staff and providers from adolescent and adult clinics. Results suggest that involvement from both the adolescent and adult clinics is critical to provide coordinated care, thus highlighting the importance of cultivating interclinic connections to support YLHIV during HCT [4]. The data allow for the identification of possible intervention leverage points to support HCT; future research needs to examine long-term HCT trajectories to develop interventions that both YLHIV and providers at adolescent and adult clinics can use to facilitate successful HCT. Addressing the multiple factors affecting HCT is essential for ensuring continuous care engagement for YLHIV.