Purposive sampling was used to ensure diversity in sociodemographic characteristics, clinical experience, and psychological distress, including varying levels of anxiety and depressive symptoms. Participants were eligible if they (1) were men aged ≥18 years, (2) self-identified as MSM, (3) had a confirmed HIV diagnosis, (4) had access to WeChat, and (5) provided informed consent. Individuals were excluded if they had hearing or language impairments or had participated in psychological treatments (eg, psychotherapy or support groups) or other intervention-based research within the previous three months.

Trained research assistants with prior experience in qualitative interviewing and CBT conducted semistructured, in-depth interviews lasting approximately 60‐90 minutes. The interview guide explored participants’ experiences related to HIV diagnosis, stigma and disclosure, social support, coping with sexual minority and HIV-related stress, treatment adherence, and preferences for the digital psychological intervention. The complete interview guide is provided in Table S1 in Multimedia Appendix 1.

Before the interview, participants completed standardized measures to characterize their baseline psychological symptom profiles. Anxiety was measured using the Generalized Anxiety Disorder 7-item (GAD-7) scale, in which scores >9 and >14 indicate moderate and severe anxiety, respectively (Cronbach α=0.90) [31]. Depression was measured using the Patient Health Questionnaire-9 (PHQ-9), with scores >9, >14, and >19 indicating moderate, moderately severe, and severe depression, respectively [32]. Sociodemographic information (eg, age, ethnicity, marital status, and education), behavioral, and clinical characteristics (eg, sexual behavior and substance use) were also collected.

Audio recordings were transcribed verbatim and analyzed using thematic analysis supported by NVivo 11.0 (QSR International). Two researchers (WW and XL) independently reviewed and coded the transcripts following the six-step approach proposed by Braun and Clarke [33]. To ensure that qualitative findings directly informed intervention adaptation, the analysis followed a framework-guided thematic approach focused on 3 broad domains relevant to the ESTEEM intervention framework: psychological distress, underlying cognitive schemas, and behavioral patterns. Specific themes and subthemes emerged inductively through iterative coding and constant comparison and subsequently served as targets for intervention adaptation in phase 2. WW and XL (project principal investigator) met regularly throughout the analytic process to compare interpretations, refine the coding framework, and resolve discrepancies through consensus discussion. Recruitment and analysis proceeded concurrently, and no substantially new themes emerged in the final interviews, suggesting adequate thematic saturation for the study objectives. Descriptive statistics were used to summarize participants’ demographic, clinical, and psychological characteristics, including means, SDs, ranges, and frequency distributions. GAD-7 and PHQ-9 scores were categorized according to established cutoff values to describe symptom severity.

Content adaptation was guided by integrating phase 1 qualitative findings with the theoretical principles of LGBTQ-affirmative CBT underlying the ESTEEM framework. Recurrent HIV-related cognitive, emotional, and behavioral patterns were mapped to corresponding CBT strategies, including cognitive restructuring, cognitive reattribution, flexible thinking, communication rehearsal, and psychoeducation. A multidisciplinary research team iteratively adapted the original 10-session manual by identifying content to retain, delete, modify, or add, and subsequently developed a revised manual. Content modifications were documented in an adaptation matrix, and representative examples are summarized in Table S4 in Multimedia Appendix 2.

After the manual was revised, the adapted intervention was then developed into a WeChat Mini Program in collaboration with a technology company. The platform included dedicated interfaces for users, counselors, and administrators. The prototype underwent theater testing (n=5 participants) to optimize the user interface and content delivery flow. Finally, an expert panel comprising HIV researchers, counselors, and community-based organization (CBO) staff reviewed the intervention to assess its theoretical fidelity, cultural sensitivity, and clinical safety.

The iESTEEM was designed as a counselor-assisted intervention rather than a fully automated self-guided program. Participants completed the 10 intervention modules at a weekly cadence while receiving asynchronous counselor support integrated into each session [36]. Counselor involvement followed a standardized protocol with two primary functions: (1) skill reinforcement, whereby counselors reviewed cognitive restructuring assignments and provided individualized feedback to reinforce CBT skills; and (2) responsive support, whereby participants’ questions and concerns were addressed within 48 hours. This structure ensured ongoing therapeutic support while maintaining the flexibility of self-paced learning. Counselors were also responsible for responding to automated risk alerts generated by the platform and implementing the study safety protocol when indicated.

To ensure intervention fidelity, all project counselors had prior experience working with sexual minority populations and possessed professional CBT certification, and all research assistants received standardized training before pilot implementation. Training covered the original ESTEEM model, HIV-specific adaptations, platform operation, and procedures for reviewing assignments, responding to participant inquiries, and managing the risk-alert system.

The 2-week pilot was designed to evaluate preliminary feasibility and usability, including participants’ engagement with core intervention components and the operational feasibility of the counselor-assisted Mini-Program. The pilot involved 10 MSMLWH representing the spectrum of psychological distress identified in phase 1 and five project counselors. The 2-week duration was selected to allow participants to experience the complete intervention workflow and interact with all 10 modules. During the pilot, participants engaged with the platform’s core features, including module learning, skill practice, assignment completion, and asynchronous communication with counselors. Long-term adherence, sustained counselor workload, or implementation outcomes across the full 10-week intervention were beyond the scope of this pilot study.

After the 2-week pilot, separate focus group interviews (30‐60 minutes) were conducted with MSMLWH participants and counselors. The evaluation focused on three domains informed by previous implementation and digital mental health studies [37,38]: (1) perceived utility and contextual relevance, including perceptions of the intervention’s relevance to participants’ lived experiences and the usefulness of its therapeutic skills; (2) system usability, including barriers and facilitators to engagement, privacy protections, and workflow efficiency; and (3) implementation feasibility and acceptability, including therapeutic alliance, willingness to continue using and recommend the program, and suggestions for future optimization. The complete interview guide is provided in Table S2 in Multimedia Appendix 3. Feedback from participants and counselors was documented and used to identify areas for future optimization of intervention content, platform functionality, and implementation procedures prior to the ongoing randomized controlled trial.

Feasibility and engagement were summarized descriptively using objective platform analytics, including login frequency, cumulative login duration, counselor-message interactions, learning duration, and module-level access. Continuous variables were summarized using means, SDs, medians, and ranges or IQRs, as appropriate. Given the small sample size and formative nature of the pilot, changes in GAD-7 and PHQ-9 scores were examined descriptively and visualized as exploratory symptom trajectories without inferential efficacy testing. Qualitative feedback from participants and counselors was summarized according to the predefined evaluation domains to identify areas for intervention and platform optimization.

The Mini-Program incorporated a backend risk-alert system that automatically notified counselors when participants’ routine psychometric monitoring scores exceeded predefined clinical thresholds. Counselors were instructed to review the alert, contact the participant through the platform, conduct a brief risk assessment, and provide follow-up supports or referral as needed. Urgent safety concerns were escalated to the principal investigator and clinical team according to the study safety protocol.

This study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Xiangya School of Nursing, Central South University (approval number 2023110901). All participants provided electronic informed consent through the secure iESTEEM Mini-Program and were informed that participation was voluntary and could be withdrawn at any time without penalty. Data were anonymized and stored on an encrypted, password-protected server. Participants received compensation of 50 RMB (US $7.43) for completing the phase 1 interviews and 50 RMB for completing the phase 3 pilot assessment.

Participant characteristics are summarized in Table 1. Twenty MSMLWH participants were relatively young (mean 23.25, SD 3.08 years), predominantly highly educated (19/20, 95%), and had been living with HIV for 7.25 (SD 4.14) months since diagnosis. Baseline psychological screening revealed a substantial mental health burden, with 30% (6/20) of participants reporting moderate-to-severe anxiety symptoms and 60% (12/20) reporting moderate-to-severe depressive symptoms.

The qualitative analysis identified 3 major themes of psychological distress, including persistent health anxiety, stigma internalization, and intimacy barriers (Themes 1‐3), as well as specific intervention needs and delivery preferences (Theme 4). Across themes, participants described how the coexistence of sexual minority stress and HIV-related stress generated maladaptive cognitive schemas that shaped emotional responses and coping behaviors. These findings informed the systematic adaptation of iESTEEM and are summarized in Figure 2.

Guided by the phase 1 findings, the iESTEEM was systematically adapted and iteratively refined to address the dual-stress context and HIV-specific maladaptive cognitive schemas identified among Chinese MSMLWH (Figure 3; Table S4 in Multimedia Appendix 2). The adaptation broadened the original ESTEEM focus from solely addressing sexual minority stress to encompassing both sexual minority stress and HIV-related stress. Overall, 53 universal scenarios were retained, 54 scenarios were modified to incorporate HIV-specific content, and 23 new scenarios were created based on participant interviews.

Representative adaptations included four major domains. First, to address the disclosure dilemma associated with HIV stigma and filial piety, the generic “Coming Out” module was replaced with a “Dual Disclosure” scenario titled “Yezi’s Story,” which guides users to practice disclosing their dual identities (MSM and HIV status) and reinterpret parental shock and fear as temporary emotional reactions rather than inevitable rejection. Second, to address the health-related catastrophizing, cognitive reattribution exercises were added to challenge the tendency to interpret minor physical symptoms (eg, fatigue or dizziness) as signs of immune failure or mortality. Third, to address medication-related distress and nonadherence, goal-setting and cognitive reframing strategies were integrated to reposition medication-taking from a painful reminder of illness into a symbol of self-care and health protection. Fourth, to address intimacy barriers, the “U=U” principle was incorporated into core belief modification exercises, and new “Intimacy Dilemma” scenarios (eg, Mr Cao’s story) were developed to help users distinguish their HIV status from their value as romantic partners. The complete scenarios are provided in Table S4 in Multimedia Appendix 2.

Building upon our previously adapted ESTEEM program for Chinese HIV-negative MSM (“Yi Si Tang”), the intervention for MSMLWH was renamed iESTEEM (Chinese: Ai Yi Si Tang) to reduce HIV-related stigma and enhance cultural relevance. The prefix “i” was intentionally selected for its phonetic resonance with “Ai (爱, meaning Love),” symbolizing self-love and empowerment while reinforcing the intervention’s emphasis on cognitive restructuring and positive identity reconstruction.

To address participants’ preferences for privacy, accessibility, and flexible learning identified in phase 1, iESTEEM was developed as a counselor-assisted, safety-monitored WeChat Mini-Program (Figure 3). The platform was designed to provide a highly private, user-friendly, and flexible digital environment that could seamlessly integrate into participants’ daily routines while minimizing concerns regarding unintended identity disclosure.

The front-end user interface consisted of 5 core modules: secure dual-verification login, structured course learning, assignment submission for skill consolidation, counselor messaging, and routine psychometric monitoring (Figures 4A-4F). To accommodate participants’ preferences for bite-sized, engaging learning, intervention content was delivered through multimodal materials, including short videos, graphics, and interactive case scenarios. This modular design enabled participants to engage with intervention content flexibly while maintaining structured progression through the 10-session program.

The back-end management system was designed to support monitoring, safety, and adherence (Figures 4G and 4H). The dashboard enabled administrators and counselors to monitor engagement indicators, including login duration, counseling duration, and assignment completion. An automated risk-alert system notified counselors when psychometric monitoring scores exceeded the predefined severity threshold and prompted follow-up safety assessments. Weekly reminders and asynchronous counselor messaging, with counselor responses provided within 48 hours, were implemented to promote adherence and ensure timely professional support.

The iESTEEM protocol underwent iterative refinement through theater testing and expert consultation to enhance usability, ecological validity, and cultural relevance. User feedback from theater testing informed several user-centered refinements, including simplification of the login process, increased font size, and strengthened engagement mechanism—the system was programmed to provide automated encouraging prompts and explicit reminders of counselor availability when users faced challenges, thereby aiming to enhance adherence. Expert feedback further informed adaptations that address culturally salient stressors, particularly marriage pressure and the family disclosure dilemma. The finalized intervention content and the technically optimized platform then proceeded to phase 3 to evaluate their feasibility and usability.

A total of 10 MSMLWH and 5 counselors participated in the 2-week pilot study and postintervention focus group interviews. As shown in Table S5 in Multimedia Appendix 5, these MSMLWH faced considerable psychological distress, with 60% (6/10) reporting moderate-to-severe anxiety and 70% (7/10) reporting moderate-to-severe depression.

Objective platform analytics demonstrated active and sustained engagement with the iESTEEM Mini-Program during the 2-week pilot period (Table 2). Participants logged into the platform 14.1 (SD 6.7) times per person and spent 3.83 (SD 2.25) hours per person on the platform, completed 134.3 (SD 103.1) minutes of learning activities per person, and engaged in 2.7 (SD 3.0) message interactions per person.

Engagement with individual modules was concentrated in the intervention’s introductory, psychoeducational, and cognitive components (Table 3). All participants accessed the introductory module, and 90% (9/10) accessed modules 2‐5. Engagement gradually declined in later behavioral modules, with 70% (7/10) accessing module 6 and 40% (4/10)‐50% (5/10) accessing modules 7‐10.

Although the pilot was not designed or powered to evaluate efficacy, descriptive trends suggested reductions in anxiety symptoms over the study period (Figure 5). The GAD-7 scores decreased from 8.9 (SD 2.3) at baseline to 7.2 (SD 3.0) at week 2. In contrast, the PHQ-9 scores remained relatively stable (baseline: mean 6.8, SD 3.5; week 2: mean 6.7, SD 4.4).

Participants and counselors endorsed the contextual relevance and perceived utility of iESTEEM. Participants reported that newly developed scenarios, particularly the “Dual Disclosure” and “Somatic Hypervigilance” exercises, closely reflected their lived experiences, describing them as “finally speaking our language” compared with generic HIV-related materials. Participants also valued the self-paced videos and cognitive exercises, which provided practical strategies for managing anxiety, stigma, and relationship concerns and were subsequently reinforced through individualized asynchronous counselor feedback.

The WeChat Mini-Program was perceived as highly accessible, private, and acceptable. Participants unanimously highlighted the advantages of the no-download, WeChat-embedded format. It reduced barriers to use and enabled an “invisible” mode of engagement that enhanced psychological safety by minimizing concerns regarding unintended disclosure, identity exposure, and HIV-related stigma. In addition, the bite-sized learning format facilitated integration of therapeutic activities into fragmented periods of daily life with minimal disruption to work and social routines.

Counselors emphasized the clinical utility of the backend management system. The dashboard enabled real-time monitoring of engagement metrics and automated identification of participants experiencing elevated psychological distress, facilitating timely follow-up and individualized support through the integrated risk-alert system. Counselors particularly valued the contextual cross-referencing functions, which allowed seamless navigation between participant assignments and relevant intervention modules and substantially improved the efficiency and precision of asynchronous counseling.

Overall satisfaction was high. All participants expressed willingness to continue using the platform and stated they would recommend iESTEEM to peers. One technical challenge involving delayed counselor notifications was identified during the pilot. However, participants reported that the high quality of the therapeutic alliance and the sincere “human touch” of the counselor’s detailed responses largely offset the negative impact of the notification delays. Identification of this issue directly informed subsequent technical refinement for the future full-scale trial.