The effects of virtual reality on learning outcomes in health professions education: a systematic review and meta-analysis

Background Virtual reality (VR) is increasingly integrated into health professions education to address limitations of traditional instructional formats. However, whether VR delivers consistent advantages across different educational constructs remains variable. This systematic review and meta-analysis evaluated the effectiveness of VR-based interventions on objectively assessed learning outcomes compared with traditional instructional methods across health professions curricula. Methods We systematically searched PubMed, Embase, Scopus, ERIC, and Cochrane Central from inception through August 28, 2026, for randomized controlled trials comparing VR with traditional teaching methods. Standardized mean differences (SMDs) were calculated using Hedges’ g via DerSimonian–Laird random-effects models, with outcomes categorized into cognitive/knowledge and procedural/technical domains. Between-study heterogeneity was quantified using the I 2 statistic, and 95% prediction intervals were computed. Risk of bias was evaluated using Cochrane RoB 2, and evidence certainty was graded using the GRADE framework. Results Twenty-three randomized controlled trials (24 independent datasets, 1,202 participants) were included. VR-based training significantly improved procedural/technical performance compared with traditional methods (13 datasets; SMD = 0.90, 95% CI 0.43 to 1.37; p = 0.0002; I 2 = 80%; 95% prediction interval: -0.82 to 2.62). VR also yielded a significant improvement in cognitive/knowledge outcomes (11 datasets; SMD = 0.43, 95% CI 0.20 to 0.66; p = 0.0002; I 2 = 54%; 95% prediction interval: -0.21 to 1.07), with a significant difference between domains ( p = 0.005). The secondary overall pooled effect was likewise significant (24 datasets; SMD = 0.65, 95% CI 0.40 to 0.90; p < 0.00001; I 2 = 74%). Under the GRADE framework, certainty of evidence was rated as moderate for both domains, downgraded solely for inconsistency. Conclusions VR instruction is associated with higher average scores in objectively assessed learning outcomes than traditional methods, with more pronounced gains observed for procedural skills than for theoretical knowledge. However, because prediction intervals crossed zero and substantial heterogeneity was present, educational advantages may vary across future instructional settings. Trial registration Not applicable. This systematic review was not prospectively registered, and no formal review protocol was prepared prior to study execution.