Data were obtained from a multimethod evaluation of the Strategic, Multisite, Initiative for the Identification, Linkage and Engagement in Care Program (hereafter called the Care Initiative). The care initiative originated in a formal partnership of the National Institutes of Child Health and Human Development, Centers for Disease Control & Prevention, and The Adolescent Medicine Trials Network for HIV/AIDS Interventions (ATN). The care initiative was developed to improve LTC for HIV-positive adolescents by improving collaboration with local health departments and community partners, and by supporting outreach workers solely dedicated to facilitating adolescent linkage to care (Tanner et al., in press
The data included 124 semi-structured qualitative interviews collected between February 2010 and October 2011 from 15 ATN clinics across 13 cities in the US that provide HIV-related care to adolescents (Straub et al., 2007
). Two qualitatively trained researchers conducted interviews with ATN staff involved in linkage to care processes (e.g., physicians, nurses, linkage to care outreach workers, social workers, case managers, and program staff). Interviews lasted approximately one hour each. Interviews used a topic guide that focused on site-specific organization and LTC processes; facilitators and barriers to LTC; relationships with local health departments and community partners; and the scale-up of the care initiative LTC process. Specific inquiries addressed potential solutions to barriers to LTC. Interviews were conducted in personnel offices or a private space at the clinics, and informed consent was obtained prior to each interview. Interviews were digitally recorded, transcribed verbatim, and managed using Atlas ti 6.2 (Muhr, 2004
). The Institutional Review Board at the Johns Hopkins Medical Institutions and each ATN site approved the study protocol.
We used a Dynamic Social Systems model (Latkin, Weeks, Glasman, Galletly, & Albarracin, 2010
) to guide the identification of structural factors associated with linkage to care, specifically the influence of macro-, meso-, and micro-level structures (Latkin et al., 2010
). This model emphasizes the social and dynamic qualities of structural factors that influence HIV programs, and postulates three key structural dimensions that affect care: resources; influence and control; and, contextual factors (Latkin et al., 2010
). Within the current study, the macro-level structure refers to the sociopolitical, economic, and cultural context, as well as larger social institutions that shape linkage to care more broadly. The meso-level structural factors include systems that work within the more proximal institutions, within which individuals are involved (for example, clinic dynamics and physical space). Micro-level structural factors refer to the immediate social and physical context within which interactions among individuals take place (for example, between providers and patients; Latkin et al., 2010
). Pairing this model with qualitative research has allowed us to explore issues related to adolescents and linkage to care, which have been missing in earlier studies.
To assess the barriers and facilitators to linkage to care at each of the 15 ATN clinics, transcripts were analyzed using the constant comparative method (Glaser & Strauss, 1967
). A list of thematic codes was first created based on the existing linkage to care literature and interview guide; two team members read and coded each interview transcript to create an initial code dictionary. These team members cross-coded a random sample of 33% of transcripts to refine the code dictionary, which was subsequently reviewed by other team members. Final coding was conducted during a sequence of weekly meetings to develop additional codes and resolve discrepancies. We constructed analytical memos on these processes, discussed the memos, and refined the coding matrix based on these discussions of the data (Glaser & Strauss, 1967
). Following the procedures of the constant comparative method, we searched for negative cases to explore potential exceptions to the themes surrounding barriers and facilitators to linkage to care, modified and developed the coding matrix as needed, and returned to the data for additional comparisons (Glaser & Strauss, 1967
).A random sample of 20% of the interviews was again cross-coded to assess consistency; any discrepancies in coding were discussed among the investigators and resolved.