This study uses a grounded theory design, a qualitative research method aimed at developing a theory based on the data collected [20]. Initially, grounded theory methods were used to arrange and code the data, allowing the development of tentative categories. Reflexive thematic analysis, rooted in constructivist principles, then facilitated the identification and analysis of patterns across the data [20]. Through iterative coding and constant comparison, grounded theory methods helped in refining these patterns into themes, ultimately leading to an explanatory framework that elucidates the relationships among identified themes. The present researchers sought to develop a theory directly based on the data collected to understand the participants’ experiences. The research questions were as follows: How do therapeutic relationships and community dynamics differ across virtual and in-person care? How does acuity of mental health diagnoses influence these perceptions?

Participants included clinical staff (ie, those with social work degrees, professional counseling degrees, nursing degrees, Master of Arts degrees, mental health counseling degrees, alcohol and substance use degrees, and dietetic degrees) and clients seeking intensive outpatient mental health treatment in New York, NY, and Washington, DC. The study included 32 individuals (14 clients and 18 clinical staff members) broken into 6 homogenous groups, 3 of which were client groups and 3 of which were staff groups.

Groups did not exceed 4 to 6 individuals, and this was intentional to provide ample opportunity for engagement of each respondent in the groups. During the focus groups, the client respondents were in 1 of 2 locations: home or a mental health treatment center. However, all clinicians were on site at the treatment program for focus group interviews.

Client demographics, including age, race or ethnicity, and gender identity, and clinical credentials for staff were consistent with those previously reported in Cerrito et al [5]. From the point of view of the analyses conducted, the distribution of the sociodemographic factors remains important, as the sample was predominantly White (n=12, N=14, 86%), and majority were cisgender male and female (n=11, N=14, 79%), with a smaller portion identifying as transgender nonbinary (n=3, N=14, 21%), which may limit generalizability.

This study is approved by the institutional review board at Yale School of Medicine (IRB #2000032626). Clinical staff and clients who had sought intensive outpatient mental health treatment since the pandemic provided insight about their experience through qualitative focus groups. All participants provided informed consent prior to participation. Data were deidentified and stored securely to protect participant privacy and confidentiality. No financial compensation was provided for participation.

This research followed the methodology outlined in Cerrito et al [5]. Recruitment occurred via an email invitation outlining the study’s aims, the voluntary nature of participation, and the opportunity to contribute to advancing understanding of mediums of mental health care. Data collection occurred between March 20 and April 3, 2023, with participation invitations sent via email to staff and clients. Inclusion criteria were as follows: (1) at least 18 years of age, (2) are a client previously admitted to The Dorm or are a provider employed by The Dorm, and (3) able to read and understand the consent form. Exclusion criteria included the following: (1) are unfit to complete the survey due to medical or psychological constraints and (2) require a legally authorized representative.

In qualitative research, the social context of the participants’ lives, time, place, and biography are important [21]. Client participant perspectives were interpreted within a shared social and clinical context, including engagement in intensive outpatient mental health care during the pre- and postpandemic period, while clinical staff participants had experience delivering both in-person and virtual treatment modalities. This was also the subject of attention both during the interview and the analytical process.

The invitation to participate in this study was sent out in emails that included detailed descriptions of the subject of the study, its course, and the purpose of the data being solely for scientific purposes. All the respondents gave informed verbal consent to participate in the study at the start of the focus groups. To minimize interviewer preconceptions, focus groups were conducted by a qualitative researcher trained in neutral, open-ended interviewing techniques that use nonleading prompts. These prompts were consistent with Braun and Clarke’s reflexive thematic analysis approach, which emphasizes reflexivity, openness to participant meaning, and minimizing assumptions [22].

The first part of the interview guide was related to attendance and engagement in the 2 modes of care, and the second portion began to explore detailed topics, referring to a list of auxiliary questions, which had a semistructured form. To obtain the best quality of data possible, the researcher asked detailed questions about experiences and reflections related to engagement, friendships, rapport-building, trust, honesty, communication, and the benefits of virtual versus in-person. The questions were open and nonleading, and follow-up items were prewritten to ensure comprehensive dialogue, for example, “Tell me about attendance in virtual care settings.”

During the interview, all respondents were consistently encouraged to share their experiences as openly and honestly as possible. The interviewer used nonjudgmental language and flexible pacing to ensure that all respondents had a chance to contribute. Much effort was made to create optimal comfort and safety in the space for this purpose.

Throughout the data analysis process, the interpretative perspectives of clients and clinical staff were prioritized, consistent with grounded theory methodological assumptions. The analysis began with grounded theory techniques like open coding and constant comparison, which helped in the initial organization of the data [23]. Reflexive thematic analysis was then used to further analyze and report patterns, guided by Braun and Clarke’s 6-phase framework [21]. This structured approach included familiarization with data, generating initial codes, and identifying themes [21,22]. Constant comparison was used to refine these themes into conceptual categories, contributing to the development of theoretical insights into the impacts of different care mediums [21]. For instance, themes such as client engagement, friendship, and rapport were identified as central to the sense of community. Through constant comparison, these themes were linked to broader constructs like therapeutic alliance and community dynamics, leading to theoretical insights into the differential effects of virtual and in-person care on these constructs.

The authors remained open to the depth of the transcripts and were responsive to emerging insights throughout the analytic process. The stages of this research study were adapted from Braun and Clarke’s 6-phase thematic analysis framework, which include familiarization with data, generating initial codes, searching for themes, reviewing themes, and defining and naming themes, and the final phase 6, writing the report [22].

Thematic analysis in this study went in the following stages.

The in-depth qualitative focus group interviews were recorded on Zoom (Zoom Video Communications) and lasted up to 1 hour. As a result of the 6 focus group interviews (3 client groups and 3 staff groups of 4 to 8 individuals per group), Zoom-recorded audio material of 4 hours 40 minutes was obtained, which was then subjected to transcription. Groups were labeled (ie, client groups 1, 2, and 3; clinician groups 1, 2, and 3) solely for organizational purposes and do not reflect hierarchy or analytic distinctions. The 6 interviews, 3 from clients and 3 from clinical staff, were rich in depth and content. Following iterative engagement with the data, it was identified that saturation was reached once no new themes emerged, and there was no need to conduct repeat interviews. Iterative engagement consisted of repeated transcript review, reflexive memoing, and ongoing refinement of themes as analysis progressed, in line with Braun and Clarke’s reflexive thematic analysis [22]. Saturation was achieved when the data no longer yielded new insights or properties relevant to the emerging theoretical categories, and all core themes were fully developed and defined conceptually [22].

The identified codes that emerged were considered of greatest analytical importance due to their salience across interviews and relevance to the research questions. Analytic priority was given to codes that captured meaningful clinical patterns, rather than just those that appeared most frequently. The identified codes were as follows: client engagement, friendship, rapport, acuity, ADLs, client diagnosis, and use of clinical skills. These codes created 3 themes: community (client engagement, friendship, and rapport), clinical presentation (acuity, ADLs, and client diagnosis), and use of clinical skills.

Finally, researchers looked at the use of their clinical skills and how the medium of care (ie, virtual vs in-person) impacts this, if at all, within the context of individual and group sessions. Use of clinical skills was both the theme and code for this category, with particular topics coming up around the importance of engagement, rapport building, and body language being a cornerstone for using clinical skills.

For example, engaging clients is a skillset that clinicians highlighted as difficult when comparing virtual and in-person therapeutic spaces. Clinicians reported that client engagement skills are negatively impacted in virtual spaces:

Fostering therapeutic alliance and rapport is another clinical skill that appeared in the transcripts when discussing the use of virtual versus in-person therapeutic spaces. Clinicians found that rapport building was easier in person. For example:

Clients and clinicians also noted that picking up on body language cues, an important domain of identifying signs and symptoms of mental health concerns, is more challenging in a virtual space. Being able to only see one’s head prevents the clinician from accessing important nonverbal cues (ie, tapping fingers and legs shaking; Textbox 4).

Finally, eye movement desensitization and reprocessing (EMDR) was mentioned multiple times as a modality that clinicians felt strongly about being conducted in person. The below quote summarizes the consensus:

While virtual care is acknowledged by the study participants as a valuable tool for maintaining treatment continuity, particularly in the face of logistical barriers (ie, distance, scheduling, school, or work), participants repeatedly emphasized the relational and clinical limitations innate to virtual platforms. This study uncovers, from the perspective of client and clinician focus groups, that in-person treatment is more supportive and fosters engagement in treatment, interpersonal connection (ie, therapeutic alliance and peer-to-peer), and the ability to exercise clinical skills than virtual settings.

Clinicians reported that their ability to use essential clinical skills (ie, engaging clients, fostering therapeutic alliance, and reading body language) is hindered in virtual settings. Certain modalities, such as EMDR, were considered more effective in person due to the emotional intensity they evoke and the importance of clinicians closely monitoring clients for safety in this type of therapeutic modality. While existing literature supports the feasibility of EMDR in virtual settings, there is a need for more robust methodological research to examine efficacy [22]. This study supports in-person treatment as the ideal medium for complex modalities, such as EMDR, and high acuity cases of all diagnoses, given the emotional and relational nuance that is missing when treatment is conducted online.

In examining how therapeutic relationships and community dynamics differ across virtual and in-person care, it was found that loss of connection occurs when therapy switches to a virtual format. Clients described distractibility and a reduction in therapeutic and peer-to-peer accountability when in virtual groups. Body language, nonverbal communications, and water-cooler conversations between sessions were some of the factors that were reported as being essential for therapeutic alliance and peer connection; yet, these were also reported to mainly occur in an in-person setting. Participants’ reports made clear that there are qualities of in-person care that help build rapport and foster therapeutic alliance, naturally, in a way that only virtual settings can forcibly achieve. Existing literature supports the finding that virtual mental health treatment limits access to nonverbal cues, body language, and the type of connection that is central to therapeutic and peer alliance [5,17,24].

Acuity of mental health diagnoses also strongly influenced participant perceptions of the medium of care. When discussing clinical presentation, clinicians and clients admitted to the absence of and difficulty in noticing subtle cues (eg, smell and fidgeting) that are essential for picking up on signs and symptoms of mental health concerns. For this reason, in-person care was reported as a necessity for clients with higher acuity mental health concerns and complex diagnoses, such as active eating, substance use, or thought disorders.

This study extends previous work that examines the benefits and drawbacks of virtual mental health care [5]. The ability to connect, be present, and experience therapeutic alliance as well as support, particularly for young adults, is understood to be most effective when in person. While each setting offers distinct advantages, participants emphasized that clients with higher clinical acuity and complex diagnoses experience more effective support, safety, and peer-to-peer as well as therapist-to-client connection when in person. Certain diagnoses may need more support and the added element of in-person monitoring for safety, due to the complexity and/or acuity of one’s mental health condition. These findings underscore the importance of accounting for therapeutic and diagnostic factors when deciding the medium of care, as this can offer an effective framework for optimizing client engagement and experience.

The following limitations are consistent with Cerrito et al [5]. In summary, these include the following: (1) the analyzed group is not representative of the broader populations of mental health providers and young adults seeking mental health treatment, and future research should examine these hypotheses in a larger, more diverse population. (2) Considering factors that qualify a client for virtual versus in-person care is warranted in future research, as we understand that these results represent a specific group of individuals at a particular point in history (ie, post–COVID-19 pandemic), from large urban settings (ie, New York, NY, and Washington, DC), and social contexts (ie, predominantly high socioeconomic status). (3) Due to the necessary exclusion criteria, the sample may not fully represent all vulnerable populations. Individuals were excluded if they were unable to complete the survey due to medical or psychological limitations or if participation required a legally authorized representative.

The results of this study outline how the medium of care (ie, virtual vs in- person) impacts therapeutic experience from the perspective of both client and clinician. It is understood that critical domains such as community, interpersonal relationships, and clinicians’ use of clinical skills are affected by the medium of care, and key factors such as clinical presentation (ie, acuity and diagnosis) must be accounted for when choosing a particular mode of care for a client.