Evaluation of the efficacy of support groups in the metaverse for Ukrainian refugees: randomised clinical trial.

Abstract

Millions of Ukrainians displaced since 2022 experience high rates of depression, anxiety and distress, with major barriers to in-person mental healthcare. Immersive avatar-based virtual reality environments may offer scalable psychosocial support while preserving key therapeutic elements. To evaluate the efficacy of avatar-based virtual reality support groups versus in-person support groups and a waitlist control among Ukrainian refugees in Romania. In this 3-arm randomised controlled trial, 156 participants were assigned to avatar virtual reality groups, in-person support groups or a waitlist control. Assessments were conducted at baseline, mid-intervention, post-intervention and at 3-month follow-up. Primary analyses used clustered intention-to-treat longitudinal mixed models with cluster-robust standard errors. Complementary baseline-adjusted analysis of covariance and sensitivity analyses were also performed. Most between-group comparisons favoured the interventions but many lacked statistical significance, probably owing to attrition and clustering. In-person groups showed the strongest benefits, in particular, they were associated with greater perceived social support versus waitlist at post-intervention and reduced anxiety in end-point analyses. Avatar virtual reality effects were smaller and non-significant. Effects were largely attenuated at follow-up. Sensitivity analyses excluding participants with high levels of baseline trauma identified stronger treatment effects. In-person support groups demonstrated the strongest short-term psychosocial benefits, especially for perceived social support. Delivery of the avatar virtual reality intervention proved feasible, but the intervention did not show consistent superiority or equivalence to in-person groups. The findings support further refinement of immersive interventions with improved retention strategies and trauma-sensitive adaptations, particularly across populations with varying digital familiarity.