BackgroundMedically underserved people with type 2 diabetes mellitus face limited access to group-based diabetes care, placing them at risk for poor disease control and complications. Immersive technology and telemedicine solutions could bridge this gap. ObjectiveThe purpose of this study was to compare the effectiveness of diabetes medical group visits (DMGVs) delivered in an immersive telemedicine platform versus an in-person (IP) setting and establish the noninferiority of the technology-enabled approach for changes in hemoglobin A1c (HbA1c) and physical activity (measured in metabolic equivalent of task [MET]) at 6 months. MethodsThis study is a noninferiority randomized controlled trial conducted from February 2017 to December 2019 at an urban safety net health system and community health center. We enrolled adult women (aged ≥18 years) who self-reported African American or Black race or Hispanic or Latina ethnicity and had type 2 diabetes mellitus and HbA1c ≥8%. Participants attended 8 weekly DMGVs, which included diabetes self-management education, peer support, and clinician counseling using a culturally adapted curriculum in English or Spanish. In-person participants convened in clinical settings, while virtual world (VW) participants met remotely via an avatar-driven, 3D VW linked to video teleconferencing. Follow-up occurred 6 months post enrollment. Primary outcomes were mean changes in HbA1c and physical activity at 6 months, with noninferiority margins of 0.7% and 12 MET-hours, respectively. Secondary outcomes included changes in diabetes distress and depressive symptoms. ResultsOf 309 female participants (mean age 55, SD 10.6 years; n=195, 63% African American or Black; n=105, 34% Hispanic or Latina; n=151 IP; and n=158 in VW), 207 (67%) met per-protocol criteria. In the intention-to-treat analysis, we confirmed noninferiority for primary outcomes. We found similar improvements in mean HbA1c by group at 6 months (IP: –0.8%, SD 1.9%; VW: –0.5%, SD 1.8%; mean difference 0.3, 97.5% CI –∞ to 0.3; P