Young adulthood is a distinct developmental period in the life course, typically defined as ages 18‐29 years, and characterized by identity exploration, instability, and self-focus [1]. Young adulthood represents the highest-risk period for the onset of mental illness and symptoms of mental ill health [2]. Postsecondary students are an important subgroup of this population, as they predominantly fall within the young adult age range and experience transition-related stressors (eg, academic pressure, adjustment to new social environments, and disruptions to health behaviors) that increase the risk of mental ill health [3-7]. Large-scale international survey data [8-10] indicate that a substantial proportion of students report symptoms of depression and anxiety, as well as challenges related to social inclusion (eg, loneliness) and engagement in health-risk behaviors (eg, low PA [physical activity] and poor sleep). In response, policy initiatives position postsecondary settings as critical contexts for supporting young adult mental health, given their potential to address behavioral and social determinants of mental health (eg, health behaviors and social connectedness) across campus life, institutional culture, and academic systems [11,12].

Postsecondary settings include universities, colleges, and other educational institutions offering formal learning opportunities beyond secondary education (eg, vocational schools) [13]. To address postsecondary student mental health, there is an international priority to integrate whole-campus approaches [11,12,14]. Whole-campus approaches extend beyond the provision of primary clinical care to include community-based prevention and early intervention services to target social, cultural, and structural influences on student mental health [11,12]. Evidence-based prevention and early intervention initiatives can be conceptually grounded within clinical staging models, which are increasingly implemented across global mental health contexts [15]. Clinical staging models represent mental health along a multidimensional continuum from health to illness, distinguishing early (subclinical) stages from later stages characterized by severe and persistent mental illness [15]. Mental health reflects the capacity to realize one’s potential, cope with stressors, and participate in society, whereas mental ill health refers to subclinical experiences of distress and impaired functioning, and mental illness captures more severe, persistent, and clinically diagnosable conditions [15,16]. Prevention and early intervention initiatives are often offered in the community during early clinical stages, where the purpose is to target social determinants of and risk factors for mental ill health (eg, social disconnection, health-risk behaviors, and distress) associated with the onset and progression of mental illness [15,17]. Yet, gaps in the implementation and scale-up of prevention strategies across postsecondary campuses are evident and include evidence-based programming approaches that are sustainable and accessible to students [18].

A promising, sustainable, and accessible community-based prevention approach is on-campus PA programming tailored toward student mental health [19,20]. On-campus PA programming has the potential to support whole-campus priorities, including social (eg, community engagement and social connection), structural (eg, accessible, inclusive PA spaces and referral pathways), and cultural (eg, reducing barriers to help-seeking) influences on student mental health [21,22]. Evidence for the effectiveness and feasibility of PA programs for student mental health and for complementing primary treatment options is growing [20,23]. This is supported by practice guidelines recommending PA as an alternative and complementary therapeutic approach to support physical health, reduce mental ill health symptoms, and promote mental health across clinical and nonclinical community-based settings [24,25]. Importantly, findings from a systematic review on PA interventions for postsecondary student mental health offer evidence of effectiveness on outcomes including depression, anxiety, and stress [23]. This body of literature highlights the importance of improving student access to support structures (eg, structured and tailored mental health PA programs) to encourage participation in PA as an evidence-based, therapeutic intervention for both clinical and nonclinical populations [23,26]. Improving PA support structures reflects a broader movement in mental health care, aimed at enhancing access to lifestyle-based therapeutic interventions [24]. Indeed, to better address postsecondary student mental health, researchers support a need for expanded treatment options, beyond traditional modalities (eg, therapy and medication), with a focus on evidence-informed prevention interventions offered in the community, such as PA programming [27,28]. Aligned with clinical staging models, PA interventions can be offered to promote well-being within the campus community, as well as integrated within the range of treatment approaches in campus mental health care, where care offered is proportionate to the stage of illness and guided by shared decision-making [15,24].

While research demonstrating the effectiveness of tailored PA programs for student mental health is growing, a limitation in the literature is the predominant focus on reducing mental ill health symptoms, with limited exploration of how such programs impact broader mental health outcomes, including social inclusion indices [23,26]. Social inclusion is defined to encompass a sense of belonging, connection, and possessing the means and opportunity to participate in valued social and community roles and activities [29]. Leisure activities, including community-based PA engagement, are supported as important for promoting postsecondary students’ mental health, community engagement, social support, and feelings of social connectedness and belonging [30-32]. In addition, extracurricular activities that involve PA have been identified by students as important interventions for supporting social interaction and belonging [28,33]. Intervention research, however, has yet to comprehensively examine the influence of structured PA programming on social inclusion indices (eg, social connectedness, social support, and belonging) among students, with current research predominantly being qualitative or cross-sectional study designs [23,26,34]. Assessing the effects of structured PA interventions on social inclusion indices is important for shifting PA research beyond symptom reduction toward a more holistic understanding of the mental health benefits of PA among postsecondary students.

Finally, the growth of online distance learning and hybrid learning formats across international settings can create accessibility barriers, limiting students’ ability to engage with on-campus mental health programs and community activities [35-37]. Researchers examining virtually delivered lifestyle interventions among individuals living with mental illness support their utility for facilitating engagement in health behavior change [38,39]. However, it remains unclear whether virtual and in-person PA delivery yield similar mental health prevention benefits, as systematic review findings support limited research on the effectiveness of virtual PA delivery for mental health and social outcomes [40,41]. Further understanding the effectiveness of both in-person and virtual PA delivery has important implications for expanding PA programs on postsecondary campuses, as digital technologies have been identified as critical for supporting student mental health [42]. Indeed, virtual PA programming has been supported as a postpandemic priority and may offer greater flexibility to students, reduce delivery costs, and expand reach to distance learners, a student population group at increased risk of social isolation and mental ill health [43].

To address the gaps in the literature, data from a multiphase research project aimed at evaluating the effectiveness of a PA program (MoveU.HappyU) for postsecondary student mental health at the University of Toronto were analyzed in this study. By using a multiphase approach, a more comprehensive understanding can be gained of how PA programs impact mental ill health symptoms and social inclusion indices, and whether in-person and virtual delivery modes yield comparable outcomes. The overarching purpose of this study was to explore the influence of on-campus PA programming (virtual and in-person delivery) on mental ill health symptoms (ie, anxiety and depression), social inclusion indices (ie, social connectedness, emotional ties, and social relationship quality), and well-being. Three specific objectives were addressed: (1) to examine the direction and magnitude of pre-post change in mental ill health symptoms, social inclusion indices, and well-being; (2) to examine whether pre-post change in mental ill health symptoms, social inclusion indices, and well-being differ between virtual and in-person delivery; and (3) to examine the degree to which change in mental ill health symptoms and social inclusion indices predict change in well-being. Guided by social inclusion research [29], social well-being models [44], and previous research [45,46], it was hypothesized that (1) there would be significant mental ill health symptom reduction and improvements in social inclusion indices and well-being pre-post intervention; (2) there would be no difference in pre-post intervention change in mental ill health symptoms, social inclusion indices, and well-being between virtual and in-person program delivery; and (3) positive change in social inclusion indices would contribute unique variance to positive change in well-being compared to the contribution of mental ill health symptom reduction alone.

MoveU.HappyU has been implemented at the University of Toronto since May 2015 and is an ongoing research initiative. Guided by the Obesity-Related Behavioral Intervention Trials model [47], adaptations have been made to support scale-up and broader implementation of the program across 3 data collection phases (Table 1). Phase 1 involved in-person delivery to students, while Phase 2 included virtual delivery during COVID-19 campus closures and in-person delivery after campuses reopened. Phase 3 returned to fully in-person delivery. Phases 1 and 2 were implemented using quasi-experimental designs, while Phase 3 was conducted as a randomized controlled trial (RCT), examining the effectiveness of one-on-one in-person PA and group-based PA on mental health and social outcomes compared to a waitlist control. For this study, analyses were restricted to the one-on-one in-person PA intervention arm. Full details of the RCT design are provided in the published protocol [48], including PA intervention delivery procedures and materials, which were consistent across the phases.

Specifically, all students received a 6-week, individually tailored, supervised PA program consisting of weekly 1-hour sessions. The 6-week duration is supported by evidence indicating that PA interventions of at least 4 weeks are effective for improving postsecondary student mental health, with moderate effects [26]. From a pragmatic perspective, the timeframe also allows for sufficient recruitment at the start of the semester to support program completion prior to final semester exams and academic breaks (eg, winter and summer holidays). Grounded in behavior change theory and empirical evidence on PA for mental health [46,54-56], each 1-hour session included 30 minutes of PA behavior change coaching and 30 minutes of PA training.

The analytical samples included 87 students for sample 1 (Phase 1 in-person delivery), 62 students for sample 2 (Phase 2 virtual delivery), and 92 students for sample 3, which included Phase 2 virtual delivery (n=62), Phase 2 in-person delivery (n=5), and Phase 3 in-person delivery (n=25). Students were unique to each phase, with no participant overlap across phases. To address objective 1, sample 1 and sample 2 were examined separately to understand program effectiveness for in-person and virtual delivery. To address objective 2, sample 3 was used to explore whether the mode of delivery influenced program effectiveness. To address objective 3, sample 1 and sample 2 were examined in separate models again to understand the unique influence of social inclusion indices on well-being, within in-person and virtual delivery.

Across the intervention phases, infrastructure provided by the Mental Health and Physical Activity Research Centre (MPARC) at the University of Toronto was used. In Phase 1, PA sessions were delivered by graduate students affiliated with MPARC. Program trainers were certified PA professionals with academic backgrounds in PA and health psychology. Training included a suicide prevention workshop, as well as graduate-level coursework in exercise psychology and behavior change coaching principles. In Phases 2 and 3, as the program was scaled, graduate students from MPARC and certified PA trainers from Sport and Recreation Services were engaged using a train-the-trainer model [57]. The research team collaborated with on-campus Sport and Recreation Services to develop and deliver standardized training focused on behavior change coaching and PA program delivery for mental health support.

For in-person delivery, sessions were delivered in private PA spaces in MPARC, conveniently located in the campus athletics and recreation center. Virtual delivery was implemented during campus closures related to the COVID-19 pandemic. Weekly sessions were delivered over secure, university-hosted Zoom (Zoom Video Communications) calls, with sessions tailored to students’ available space and equipment. Aside from delivery mode and setting, the intervention protocol (refer to the “Study Design” section) remained consistent across phases, with all sessions delivered synchronously.

Recruitment targeted undergraduate and graduate postsecondary students who were physically inactive and experiencing mental ill health. As the program evolved to support broader scalability, the recruitment and screening processes used to determine eligibility were refined accordingly.

In Phase 1, students were recruited through referrals from campus Health and Wellness Services, as well as through posters displayed in waiting areas at the campus mental health center. In alignment with campus mental health strategies promoting early access, self-referral, and the integration of nonclinical supports, broader recruitment strategies were implemented in Phases 2 and 3 [11,12,27]. Recruitment in Phases 2 and 3 occurred via outreach through Accessibility Services, Health and Wellness Services, and student life listservs. Digital recruitment materials (eg, email scripts and poster advertisements) provided information about the program and directed students to an online intake questionnaire. Interested participants completed an online intake questionnaire via a secure platform (REDCap [Research Electronic Data Capture]) [58] to initiate the intake process. In Phase 1, this questionnaire was used for scheduling purposes only, and eligibility was confirmed during an intake session, whereas in Phases 2 and 3, eligibility screening was conducted based on intake questionnaire responses, and eligible participants were invited to complete an intake session. During intake sessions, written informed consent was obtained, PA readiness was assessed, and baseline self-report measures were completed to assess demographic characteristics, PA, mental ill health symptoms, and social inclusion indices. Across intervention phases, recruitment occurred at the beginning of the winter, spring, and summer semesters, with participants starting the intervention in the early to midsemester period and finishing prior to the end of the semester.

In Phase 1, eligibility was determined based on clinical judgment by campus mental health care providers and established program criteria. Students were required to be physically inactive, defined as engaging in <150 minutes of moderate-to-vigorous physical activity (MVPA) per week in accordance with global PA guidelines [59], and to be seeking mental health support. In Phases 2 and 3, participants were screened as insufficiently active based on a score of ≤23 metabolic equivalents of MVPA per week on the Godin Leisure-Time Exercise Questionnaire [60], consistent with established classifications of insufficient PA and indicative of not meeting global recommended PA guidelines [61]. Mental ill health eligibility in Phase 2 was determined by the following criteria: (1) scoring at or above established clinical cutoffs for mild depression (Patient Health Questionnaire-9 [PHQ-9] score >10 [51]) or anxiety (Generalized Anxiety Disorder-7 [GAD-7] score >5 [52]); (2) screening positive for eating pathology (SCOFF Questionnaire score ≥2 [62]); (3) reporting a diagnosed mental illness; or (4) rating mental health as “poor” on a single-item 5-point scale (1=poor to 5=excellent). In Phase 3, rather than emphasizing symptom thresholds, inclusion criteria included rating mental health as “poor,” “fair,” or “good” on a single-item 5-point scale (1=poor to 5=excellent) and seeking PA for mental health promotion. This approach was adopted to align with the importance of intervening early, before symptoms escalate, to promote mental health and prevent mental ill health [15].

Exclusion criteria across all phases included: (1) sufficiently active, (2) unsuccessful PA clearance (Physical Activity Readiness Questionnaire for Everyone+ [63]), and (3) not being a postsecondary student seeking mental health support. Students were eligible to participate in the study regardless of whether they were using additional forms of mental health treatment or support (eg, therapy and medication).

Following the intake session, eligible participants were matched with a program trainer and worked with the same trainer for the 6-week duration for both virtual and in-person PA delivery. Participants completed an online self-report questionnaire at 2 time points (baseline: T1 and postintervention: T2). For purposes of this study, participants were included if they had complete data at 2 time points.

Internal consistency of the measures was assessed and supported using an omega coefficient (ω) or a Spearman-Brown coefficient (ρ) for 2-item scales, with values greater than 0.70 [64].

The data were screened for outliers using z-scores (|z|> 3.00 SD), missing data, and assumptions of regression analyses [66] and ANOVAs [67]. Descriptive statistics, including means, SDs, and frequency counts, were conducted to describe the samples. For primary analyses, demographic variables were dichotomized to simplify interpretation and based on data distributions. Gender was dichotomized as women (1) and nonwomen (0), age as young adults (18‐25 y; 1) and adults (26‐50 y; 0), ethno-racial identity as White (1) and less represented ethno-racial groups (0), and mental illness as reported past-year diagnosis (1) and no reported past-year diagnosis (0).

To address objective 1, 2-tailed paired samples t tests were conducted to examine pre-post change in mental ill health symptoms, social inclusion indices, and well-being. Bonferroni correction was applied to control for multiple comparisons across a family of 5 tests (Bonferroni-adjusted significance level, α=.01). To address objective 2, and test for equivalency, between-group differences (delivery mode: virtual and in-person delivery) in baseline main study outcomes and dichotomized demographic characteristics were examined using chi-square tests and independent samples t tests. For the main analysis, a 2 (delivery mode: virtual and in-person) × 2 (time: T1 and T2) repeated-measures ANOVA was conducted to examine pre-post change in the intervention outcomes and to examine differences in change between virtual and in-person program delivery. To address objective 3, bivariate correlations were conducted to examine associations between change in primary outcomes using standardized residual change scores and dichotomized demographic variables, with Spearman correlations used for analyses involving dichotomous variables and Pearson correlations for continuous variables. Standardized residual change scores were calculated for social inclusion indices, mental ill health symptoms, and well-being by regressing postintervention (T2) scores on the baseline (T1) counterparts and saving the residuals. Hierarchical regression analyses were then conducted using standardized residual change scores. Given that the purpose of this research was to explore whether change in social inclusion indices explained additional unique variance in well-being than change in mental ill health symptoms, covariates were included in Step 1 of the hierarchical model, mental ill health symptoms were included in Step 2, and social inclusion indices were entered in Step 3.

Effect sizes were calculated to reflect the magnitude of the associations in the hierarchical linear regression model using standardized beta coefficients (β), to capture the magnitude of pre-post change using Cohen d, reported as Cohen d_z to reflect within-subject effects for the paired samples t tests, and partial eta squared (ηp²) for the ANOVA. Based on empirically derived guidelines in social psychology, Cohen d and β values of 0.15, 0.36, and 0.65 were interpreted as small, moderate, and large effects [68,69]. ηp² values of 0.01, 0.06, and 0.14 were indicative of small, moderate, and large effects, respectively [70]. All analyses were conducted using IBM SPSS version 28.0.1.0 (142).

Study design and procedures were approved by the University of Toronto Health Sciences Research Ethics Board (Phase 1 and Phase 2 protocol number: 31097; Phase 3 protocol number: 45228). Written informed consent was obtained from all participants prior to study enrollment. As part of the consent process, participants were provided with a detailed explanation of the study objectives, the voluntary nature of participation, their right to withdraw, and the potential risks and benefits of participation. Participant confidentiality was protected throughout the study. Data were deidentified and replaced with participant identification codes, and all data were reported in aggregate form. Any information linking identifying information to study identification codes was stored separately from research data and accessible only to authorized members of the research team. Across intervention phases, all students who expressed interest in the program, regardless of eligibility, were provided with information about available campus and community mental health and support services. Program PA trainers and staff were also trained to identify and respond to participants experiencing distress and to facilitate access to appropriate supports where needed.

Participant flow for the data collection phases (Phase 1, Phase 2, and Phase 3), including adherence and reasons for dropout can be found in Figures S1-S3 in Multimedia Appendix 1. Participant demographic characteristics for the analytical samples are presented in Table 2.

Missing value analyses at the case-level and item level were conducted separately across all 3 analytical samples. Data were missing at random as suggested by Little’s Missing Completely at Random Test for sample 1 (χ²₈₉₅=810.00, P=.98), sample 2 (χ²₅₇₄=1.26, P≥.99), and sample 3 (χ²₃₆₁=359.83, P=.51). Missing data at the item-level were low (<4%) and main study variables were handled using item-level median imputation and demographic variables were handled using list-wise deletion. Skewness (±2.00) and kurtosis (±7.00) values indicated a normal distribution of the data across the main study variables.

Descriptive statistics, including baseline between-group differences in delivery mode and means for study outcomes at T1 and T2, are presented in Multimedia Appendix 2. A significant baseline difference was observed for gender, with significantly fewer students identifying as women in the in-person delivery mode compared to the virtual delivery mode (χ²₁=5.27, P=.02). No additional baseline differences in demographic characteristics or study outcomes were observed between delivery modes. Given the baseline difference in gender and the diagnostically heterogeneous sample, gender and mental illness were included as covariates in the 2×2 repeated-measures ANOVA. To maintain model parsimony, no additional demographic characteristics were included as covariates.

Results of the 2 (group: virtual, in-person delivery) × 2 (time: T1, T2) repeated-measures ANOVA, controlling for gender and mental illness, are presented in Table 3 and illustrated in Figure 1 for mental ill health symptoms and well-being, and Figure 2 for social inclusion indices. Error bars in Figures 1 and 2 represent 95% CIs. There was a significant multivariate within-subjects effect of time. Univariate tests indicated significant reductions in mental ill health symptoms and improvements in social inclusion indices and well-being from T1 to T2, with a moderate effect for depression, well-being, social connectedness, and emotional ties, and a large effect for anxiety. There was no significant time × delivery mode, time × gender, or time × mental illness interaction effect. Multivariate between-subjects effects were significant for delivery mode. Univariate tests indicated that emotional ties significantly differed between delivery modes. Bonferroni-adjusted pairwise comparisons demonstrated an estimated marginal mean difference of −0.98 (SE 0.47), with lower emotional ties observed in the in-person delivery mode (mean 6.92, SE 0.39) compared to the virtual delivery mode (mean 7.90, SE 0.27). There were no significant multivariate between-subjects effects for gender or mental illness.

Correlation coefficients among changes in the main study variables and demographic characteristics are presented in Table 4. Based on prior literature and evidence from bivariate change score correlations, gender, ethno-racial identity, and mental illness were included as covariates in the hierarchical regression analyses for virtual and in-person delivery [71-73].

Using data from an ongoing and longstanding PA program tailored toward postsecondary student mental health (MoveU.HappyU), a comprehensive evaluation of the utility of in-person and virtual PA delivery for reducing mental ill health symptoms and promoting social inclusion was conducted. Findings from this study support the effectiveness of both in-person and virtual PA programming for reducing mental ill health symptoms (anxiety and depression) as well as promoting social inclusion indices (social connectedness, social relationship quality, and emotional ties) and well-being among postsecondary students. Reductions in mental ill health symptoms were associated with improved well-being, and improvements in social inclusion indices further contributed meaningfully to well-being change following program completion. The results provide support for the importance of further understanding intervention targets to address social inclusion indices alongside symptom reduction, indicating that PA programs can support both clinical and social dimensions of student mental health recovery.

Drawing on the broader mental health recovery literature [74,75], these findings support recovery-oriented approaches to mental health, which emphasize not only clinical recovery (ie, symptom reduction) but also social recovery (ie, rebuilding relationships, fostering belonging, and participating meaningfully in community life). Consistent with hypothesized objective 1 pre-post effects, in-person PA delivery was effective in improving social relationship quality and emotional ties with a medium effect, and in improving well-being and decreasing symptoms of anxiety and depression with a large effect. Similarly, virtual PA delivery led to improvements in emotional ties and social connectedness with a moderate effect, as well as improvements in well-being and reductions in anxiety and depression symptoms with a large effect. While PA interventions are often framed in clinical or behavioral health contexts, these results suggest that PA may also function as a social intervention by supporting students in rebuilding social networks and fostering a sense of social connectedness [25,76,77]. Notably, a substantial proportion of students were using primary forms of mental health treatment (eg, therapy and medication), suggesting that PA was often used alongside existing supports. This is consistent with models of care in which PA is positioned as a complementary component of mental health care and emphasizes that PA should be offered alongside primary treatments to support physical health, symptom management, and psychosocial functioning [24]. Future research is needed to better understand how PA can be integrated into postsecondary mental health care (eg, through social prescription and referral to community-based PA programs) to support mental health and social inclusion outcomes [24,78].

Aligning with the hypothesized results for objective 2, which examined whether intervention effects differed by delivery mode, there were no differences in pre-post change over time between virtual and in-person PA for reducing anxiety and depression symptoms and enhancing social connectedness, emotional ties, and well-being. These findings are consistent with prior evidence demonstrating beneficial effects of PA on mental ill health symptom reduction and well-being improvements with moderate-to-large effects [23,26,79]. These results further address the limited research examining social outcomes in PA-based mental health interventions [23,34,80], suggesting that improvements may also extend to social inclusion outcomes with moderate effects. These findings provide novel support for the mental health prevention value of virtual and in-person PA delivery, as limited research has examined whether mental health and social inclusion benefits differ based on delivery format [81]. However, given the quasi-experimental design of this study, further research is needed to replicate these findings using randomized controlled designs. Overall, future research is needed to better understand virtual delivery as an alternative to in-person PA delivery for students who may face barriers to in-person engagement due to geographic distance, personal preferences, scheduling conflicts, disability, or mental ill health–related challenges such as social anxiety [82-84].

In addition, the observed mental health and social inclusion benefits did differ based on mental illness diagnosis, despite a diagnostically heterogeneous sample, where a substantial proportion of students reported a past-year mental illness, supporting the transdiagnostic potential of PA programs (ie, defined as their capacity to address shared risk factors of mental illness across different mental illnesses and diagnostic criteria [85]). While evidence for the transdiagnostic potential of PA is promising, further research is needed to examine how program effects vary by diagnostic profiles and symptom severity within prevention-oriented and transdiagnostic approaches to better understand for whom PA interventions delivered virtually and in-person are most effective [85,86]. Practically, the transdiagnostic potential of PA may be particularly relevant in postsecondary contexts where students often present with a range of mental health concerns (diagnosed or undiagnosed) and where access to individualized, diagnosis-specific intervention is often limited [87,88].

Aligning with the hypothesized results for objective 3, reductions in mental ill health symptoms were associated with improved well-being, and changes in social inclusion indices contributed additional, meaningful variance in well-being change in both virtual and in-person PA delivery. In partial support of the hypotheses, positive change in emotional ties and social connectedness were associated with improvement in well-being in the virtual group, while only emotional ties, but not social relationship quality, were associated with well-being improvement in the in-person group. The inconsistent findings may reflect differences in the nature of the constructs [89,90]. Specifically, emotional ties and social connectedness reflect subjective experiences of closeness, belonging, and feeling emotionally supported, whereas social relationship quality represents a more general appraisal of satisfaction with one’s social relationships [90]. Researchers have suggested that feelings of closeness and connection may be more strongly associated with well-being than general assessments of one’s social life, relationships, and activities [89-91]. Further research is needed to explore how different dimensions of social inclusion uniquely contribute to well-being across PA delivery modes, with a practical implication for prioritizing PA program design elements (eg, peer PA mentors, group identity–building activities, and goal setting with social emphasis) that foster meaningful relationship building and social connectedness to support well-being [32,92,93].

In addition, reductions in depression symptoms were associated with improved well-being across in-person and virtual delivery. However, reductions in anxiety symptoms were only associated with improved well-being in the in-person group. The inconsistent results may reflect contextual factors related to global events during the virtual delivery period or inherent aspects of the virtual format. Specifically, virtual delivery occurred during the early stages of the COVID-19 pandemic, a time marked by heightened and persistent anxiety related to health concerns, uncertainty, and isolation [94,95]. These contextual stressors may have limited the extent to which reductions in anxiety translated into perceived improvements in well-being, even if symptoms decreased. From a theoretical perspective, virtual PA may be less effective in activating the mechanisms linked to well-being change and anxiety symptom reduction, such as real-world social interaction and opportunities to face anxiety-provoking situations in supportive environments, which may be more readily facilitated in face-to-face environments [96,97]. Future research should clarify whether virtual and in-person PA delivery activate different mechanistic pathways linking well-being change with symptom reduction and social inclusion improvements. In turn, future research focused on mechanisms of action could inform the optimization of virtual and in-person PA programs as a mental health prevention approach on postsecondary campuses.

Despite the strengths of this study, including a multiphase design and comparison of PA delivery modes, there are limitations that need to be acknowledged. Given the single-arm design, causal inferences are not possible, and it remains unknown whether the observed outcomes are directly attributable to the intervention or influenced by other factors such as the passage of time, external events, or participant heterogeneity across intervention phases. For example, individual variability in workload fluctuations, mental ill health symptom severity, and concurrent treatment were not accounted for, and future research using person-centered data analysis approaches is needed to account for the effects of within-person variability [98]. Moreover, while program outcomes over time did not differ based on gender, the samples across the intervention phases predominantly identified as women, which is a limitation consistent with the broader literature [23]. Further research should explore strategies to engage men and assess whether engagement and intervention outcomes differ among underserved student groups (eg, racialized, 2SLGBTQ+, and students living with a disability) to ensure PA programs are effective, inclusive, and accessible in their reach and impact [99,100]. Furthermore, data collection during virtual delivery occurred in the early stages of the COVID-19 pandemic, which may limit the generalizability of the findings due to contextual stressors such as heightened health-related anxiety, uncertainty, and social isolation [94,95]. Finally, this study only examined social inclusion indices related to social relationships and social connectedness. Further research is needed to understand the role of systemic social determinants of health in lifestyle intervention research, including safe and stable housing, access to services, and food and financial security [101].

Mental health recovery and treatment are often conceptualized and measured primarily through symptom reduction, with insufficient consideration of social recovery, a limitation that is similarly reflected in the PA and mental health literature [23,74]. This study contributes novel insights into the effectiveness of a 6-week one-on-one supervised in-person and virtual PA program for reducing symptoms of mental ill health and supporting social inclusion indices among postsecondary students. In postsecondary contexts where loneliness and social disconnection are increasingly prevalent, PA interventions that promote social inclusion and reduce symptoms of mental ill health may play a critical role in supporting student well-being and reducing vulnerability to a range of mental ill health concerns [102]. In addition, with the rise in distance-based learning on postsecondary campuses, these findings provide practical implications for further understanding the effectiveness of virtually delivered PA as an alternative delivery format to in-person PA, particularly among students who face barriers to in-person engagement. Overall, the findings provide support for the effectiveness of virtual and in-person PA delivery for targeting shared risk factors (eg, low social connectedness and poor psychological well-being) that contribute to mental ill health across clinical and nonclinical populations.