Immersive virtual reality for assessing teamwork and clinical task performance: a multicenter pre-post study (Preprint)

BACKGROUND Immersive multi-user virtual reality (VR) offers standardized environments for interprofessional teamwork (IPT) training. However, evidence on observer-based assessment of teamwork performance in VR, its short-term maintenance, and its relationship with clinical task performance remains limited. OBJECTIVE To examine changes in observable IPT performance following structured teamwork instruction in immersive multi-user VR, its association with self-reported IPT perceptions, and the relationship between observed teamwork and clinical task performance across repeated sessions. METHODS In this multicenter pre-post study at four German academic institutions, 84 medical and nursing students completed three immersive VR emergency simulations in interprofessional dyads (days 1, 8, and 15), with a 30-minute TeamSTEPPS-based intervention before day 8. Blinded video ratings using the VR-adapted Team Performance Observation Tool (vTPOT) assessed observable teamwork. Self-reported perceptions were measured using the TeamSTEPPS Teamwork Perceptions Questionnaire (TPQ), and clinical task performance was captured through automated scenario-specific scores. Changes and associations were analyzed using mixed-effects models and correlation analyses, including a multiple-membership model accounting for changing dyads. RESULTS Of 126 scheduled dyadic sessions, 104 provided complete vTPOT and clinical performance data; 124 individual TPQ responses were available. Overall interrater reliability of the vTPOT was high (intraclass correlation coefficient 0.90). Adjusted vTPOT scores were higher than baseline immediately after the intervention (day 8: β=0.94, SE=0.14; P<.001) and at 1-week follow-up (day 15: β=0.75, SE=0.14; P<.001), with increases across all five domains. Self-reported IPT perceptions were also higher at follow-up (β=0.39, SE=0.13; P=.003). The association between dyad-averaged TPQ and vTPOT scores was weak at baseline (r=0.15, 95% CI −0.21 to 0.46) but improving at follow-up (r=0.49, 95% CI 0.11 to 0.74). Mean clinical task performance increased descriptively from 32.7% at baseline to 42.0% after the intervention and 50.8% at follow-up. In the multiple-membership model, higher vTPOT scores were associated with higher clinical task performance (β=11.18 percentage points per vTPOT point, 95% CI 5.18 to 16.81; P=.001). Communication, leadership, and situation monitoring showed the most consistent domain-specific associations with clinical task performance. CONCLUSIONS Immersive multi-user VR provided a standardized environment in which observable interprofessional teamwork and clinical task performance could be assessed together across multiple institutions and clinical scenarios. Observer-rated teamwork captured information not fully reflected in participants’ self-reported perceptions and was associated with clinical task performance. Further controlled studies should examine the specific contribution of VR-based approaches and the durability and transfer of observed performance changes.