Embedding virtual reality social skills training into return‐to‐work care for depression: A single‐arm feasibility pilot with exploratory autistic‐trait moderation

Aim To assess the feasibility of embedding virtual reality‐based social skills training (SST‐VR) into a return‐to‐work day‐care program for major depressive disorder (MDD) and to estimate pre–post change in self‐reported social skills, with exploratory moderation by autistic traits. Methods In this single‐arm, add‐on pilot at a Japanese psychiatric day‐care Re‐Work center, 20 adults with MDD (18 men) received six 90‐min SST‐VR sessions every 2 weeks over approximately 3 months plus the standard Re‐Work program. The primary outcome was Kikuchi's Social Skills Scale–18 (KiSS‐18); secondary outcomes were social adaptation, social anxiety, and depressive symptoms (self‐report). A linear mixed‐effects model tested the effects of Time (pre vs post), baseline Autism‐Spectrum Quotient (AQ; mean‐centered), and their interaction. Feasibility was assessed via attendance and attrition. Results Attendance was 94.2% with no attrition. KiSS‐18 increased from 49.7 ± 10.6 to 53.5 ± 12.4 ( p = 0.028; r = 0.56). The model showed significant effects of Time ( χ ²(1) = 8.11, p = 0.004) and Time × AQ ( χ ²(1) = 4.46, p = 0.035), suggesting smaller gains at higher AQ (exploratory; restricted AQ range). Emotional processing and stress management subscales improved; secondary outcomes showed no significant change. Bootstrap analyses were consistent with the mixed‐model findings. Conclusion SST‐VR was feasible. Because the single‐arm add‐on design cannot isolate SST‐VR–specific effects from concurrent care and nonspecific influences, the KiSS‐18 change is hypothesis‐generating. Controlled comparisons with Re‐Work alone in larger, more diverse samples should test efficacy, include objective behavioral and vocational outcomes, and prospectively evaluate moderation.