Virtual Reality Versus In-Person Simulation of Sepsis for Medical Students: A Non-Inferiority Cluster Randomised Study (Preprint)

<sec> <title>BACKGROUND</title> Virtual reality (VR) simulation- using head-mounted displays to present a computer-generated, three-dimensional, interactive environment - may be a cost-effective alternative to in-person (IP) medical simulation training. However, studies directly comparing learning outcomes have demonstrated mixed results and mainly focused on knowledge or skill acquisition rather than integrated practice. </sec> <sec> <title>OBJECTIVE</title> This randomised comparative pilot study aimed to evaluate the effectiveness of VR versus IP simulation in developing sepsis management skills among final-year medical students, addressing a gap in medical education evidence. </sec> <sec> <title>METHODS</title> Final-year medical students at Great Western Hospital, UK, participated in both in-person (IP) and virtual reality (VR) simulation sessions featuring sepsis scenarios. Session order was randomised, determining study group assignment. Participants underwent an additional video-recorded "assessment" IP simulation of septic shock management either between or after both scheduled sessions. Questionnaires were completed between scenario completion and debriefing across all sessions. Performance was evaluated using a modified Queens' Simulation Assessment Tool (mQSAT) by facilitating study authors across all sessions, with the assessment simulation additionally evaluated by blinded assessors. Primary outcomes included mQSAT scores, recognition of septic shock, and identification of critical care needs. Analysis of covariance (ANCOVA) was conducted to detect differences in mQSAT scores between groups, with simulation modality as the independent variable and number of simulations/debriefs prior to assessment as the covariate. Binary outcomes between groups were analysed using binomial tests. </sec> <sec> <title>RESULTS</title> 32 participants were recruited and allocated to one of four groups based on completed simulation sessions prior to assessment: IP only (IP-Assess, n=10), VR only (VR-Assess, n=6), IP then VR (IP-VR-Assess) and VR then IP (VR-IP-Assess). No statistically significant differences in mQSAT scores were detected between any groups for any domain, nor for recognition of septic shock. For recognition of need for critical care, participants who completed VR simulation only prior to assessment were more likely to recognise need for critical care than those who completed in-person simulation only (3/5 vs 1/7, P=.01). </sec> <sec> <title>CONCLUSIONS</title> This study demonstrates feasibility of the proposed trial method, and provides insight into likely effect sizes for design of further studies. Measured learning outcomes were similar across groups, regardless of which simulation modalities were used prior to assessment. Our findings are in keeping with non-inferiority of VR simulation versus IP simulation for measured educational outcomes, which is reassuring for the ethical conduct of further studies comparing VR and IP simulation. </sec>