Immersive Group Dance in Virtual Reality and Perceived Social Cohesion in Older Adults (VR-GD-COHESION)

This study is a two-arm, parallel-group randomized controlled trial (RCT) conducted in community and senior activity centers in Beijing, China. Participants are randomized (1:1) to either (a) an immersive VR-based group dance intervention or (b) a dose-matched conventional physical activity control program. Outcomes are assessed at three time points: baseline (pre-intervention), post-intervention (week 5), and 30 days after the intervention (FU30). Sex differences in intervention response are examined as a secondary objective.

Randomization and Allocation: Randomization is generated using a computer-based allocation sequence by an independent researcher. Allocation concealment is ensured using sealed, sequentially numbered opaque envelopes opened after baseline assessment. Due to the nature of the intervention, participant blinding is not feasible; however, outcome assessors and the statistician are blinded to group allocation. Stratified randomization by sex is applied to ensure balanced sex distribution across groups.

Participants: Community-dwelling older adults aged 65 years and older (both men and women) are recruited via community announcements and senior centers. After eligibility screening and written informed consent, participants complete baseline assessment prior to randomization. A total of 180 eligible participants are randomized (1:1) to the VR group dance intervention (n = 90) or the dose-matched control condition (n = 90).

Intervention - VR Group: Participants assigned to the VR group complete a 5-week immersive group dance program, delivered three times per week for a total of 15 sessions. Each session lasts approximately 20-30 minutes, including a brief warm-up, the main VR dance activity, and a cool-down period. The intervention is delivered using standalone Pico 4 Enterprise VR headsets (Pico Interactive, ByteDance, China), which provide inside-out motion tracking and do not require connection to an external computer. Headsets are fitted individually, with interpupillary distance, head strap position, and visual clarity adjusted before each session. The VR dance activities are delivered through a commercially available rhythm-based dance application installed on the headset, used in its standard form without modification by the research team. The program presents visually guided dance movements, rhythm cues, and real-time movement feedback within an immersive virtual environment, with progressive movement sequences involving upper-limb gestures, stepping patterns, weight shifting, and coordinated whole-body rhythmic movements. Sessions are conducted in small groups of approximately 6-8 participants under the supervision of trained research staff. Although each participant uses an individual headset, sessions are synchronized in timing so that participants perform the same dance routines together in the same physical room. Before the first session, participants receive a familiarization session explaining headset use, movement cues, and how to stop the activity if discomfort occurs. Intensity is maintained at a light-to-moderate level and progresses gradually across the 5-week period by increasing rhythm complexity, movement variety, and coordination demands, organized in three stages: weeks 1-2 (simple familiarization routines with slow rhythmic stepping, basic arm gestures, short movement sequences), weeks 3-4 (longer dance sequences, more frequent weight shifting, combined arm-leg coordination), and week 5 (more continuous group dance routines with greater rhythm complexity within a light-to-moderate intensity range). Exercise intensity is maintained through age-appropriate low-impact routines, short rest breaks when needed, the talk-test principle, and continuous observation of breathing, balance, fatigue, dizziness, nausea, and general discomfort by trained staff.

The "group" nature of the VR dance intervention is a physically co-present and temporally synchronized group format rather than a fully networked multi-user virtual environment. Participants do not interact with each other through avatars, shared virtual bodies, or direct communication functions inside the VR software. Instead, each participant wears an individual headset and completes the same rhythm-based dance routines at the same time, in the same supervised physical room, and within a shared group session structure. The social component is created through real-world co-presence, synchronized participation, shared task engagement, and instructor-led group organization. Participants start and end each session together, follow the same visual and auditory movement cues, perform parallel dance routines in synchrony, and have brief opportunities for social exchange before and after the VR activity. Participants are allowed to rest at any time, and research staff monitor balance, fatigue, dizziness, nausea, and general discomfort throughout each session. The physical training area is cleared of obstacles, and each participant has a marked safe movement space to minimize collision or fall risk. The same headset model, software platform, session duration, group size, supervision procedure, and progression principles are used throughout the intervention period to improve intervention consistency and replicability.

Control Group - Dose-Matched Conventional Physical Activity: The control group completes a conventional activity program matched for frequency and duration (5 weeks; 3 sessions/week; 15 sessions). Sessions include low-intensity walking and gentle mobility exercises under supervision, consistent with routine center activities. Sessions are delivered in small groups comparable in size to the VR condition (approximately 6-8 participants per session), with equivalent instructor supervision and structured opportunities for social interaction before and after each session. This design ensures comparable levels of group contact and social exposure across conditions.

Safety Monitoring and Adverse Event Procedures: Participant safety is monitored throughout all intervention sessions by trained research staff experienced in working with older adults and VR-based exercise programs. Prior to each session, participants are asked about any new health concerns, discomfort, or symptoms that might affect their ability to participate safely. During the intervention, participants are continuously observed for signs of fatigue, balance difficulties, dizziness, nausea, headache, visual discomfort, or other symptoms commonly associated with immersive VR exposure. Participants are instructed to immediately report any discomfort. If symptoms occur, the activity is paused and appropriate support is provided. Short rest periods, hydration, and seated recovery are available whenever needed. All VR sessions are conducted in a supervised exercise area cleared of obstacles, and each participant is assigned an individual movement space to minimize collision, loss of balance, or fall risk. Any adverse events or participant-reported symptoms are documented by research staff immediately following the session.

Withdrawal and Discontinuation Criteria: Participants are informed that they may withdraw at any time without penalty. Discontinuation criteria after randomization include: (a) occurrence of an adverse event or new medical condition rendering participation unsafe, (b) participant-initiated withdrawal, (c) inability to continue due to scheduling or personal reasons, or (d) failure to meet the predefined minimum participation threshold.

Statistical Analysis Plan: Analyses are performed using IBM SPSS Statistics (Version 26). Descriptive statistics are computed for all variables (mean ± SD for continuous variables; counts and percentages for categorical variables). Baseline equivalence between groups is assessed using independent-samples t-tests for continuous variables and chi-square tests for categorical variables. Normality of PCS total scores at each time point within each group is examined using Kolmogorov-Smirnov tests supplemented by inspection of distributional diagnostics (skewness, kurtosis, and Q-Q plots). Homogeneity of variance for between-group comparisons at each time point is tested using Levene's tests. The assumption of sphericity for the within-subject factor (Time) is evaluated using Mauchly's test; if violated, Greenhouse-Geisser-corrected results are reported. The primary analysis is a mixed-design repeated-measures ANOVA with Time (baseline, post-intervention, FU30) as the within-subject factor and Group (VR vs control) as the between-subject factor. The primary test of the intervention effect is the Group × Time interaction. Effect sizes for ANOVA effects are reported as partial eta squared (η²p), and Hedges' g is reported for follow-up between-group comparisons. Follow-up comparisons examine group differences at each time point, with Holm-Bonferroni adjustment for multiplicity. Sex differences are examined by including Sex as a between-subject factor in exploratory analyses (Group × Time × Sex interaction). Paired-samples t-tests examine within-group changes as exploratory analyses.