Delayed Retention of Technical Surgical Skills Following Virtual, Augmented, and Mixed-Reality Simulation: A Systematic Review and Meta-analysis
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Extended-reality simulation is widely used for technical surgical training, yet immediate gains do not establish that skills persist once practice stops.
Objective
To characterize delayed technical-skill outcomes after virtual, augmented, or mixed-reality training and estimate the comparative effect at the first eligible uncontaminated delayed assessment.
Methods
PubMed/MEDLINE, Europe PMC, CENTRAL, ERIC, and ClinicalTrials.gov were searched from inception through May 5, 2026. Eligible studies enrolled surgical learners or practitioners, used virtual, augmented, or mixed reality for technical training, and measured objective performance at least seven days later. Pure retention was distinguished from maintenance and reacquisition. One endpoint per independent study family was selected. Standardized mean differences were expressed as Hedges’ g, oriented so positive values favored extended reality. Paule-Mandel random effects with Hartung-Knapp confidence intervals were primary.
Results
Thirty-four reports representing 33 studies (1,186 participants; 21 randomized trials) from 12 countries were included. Six studies (162 participants in selected arms) were pooled. The summary effect was g = 1.39 (95% CI -0.94 to 3.72; I² = 88.0%; τ² = 4.53); the 95% prediction interval was -5.04 to 7.81. Excluding the figure-derived estimate reduced the effect to g = 0.61 (95% CI -0.76 to 1.99). GRADE certainty for the primary outcome was very low. No pooled study assessed an interval of 90 days or longer.
Conclusions
Delayed technical advantages after extended-reality simulation were observed in some settings, but the comparative evidence does not define a precise or generalizable retention effect. Substantial heterogeneity, sparse transfer and longer-term evidence, comparator differences, and influence from one figure-derived estimate warrant cautious interpretation.
