Fig 1.
PRISMA 2020 flow diagram illustrating the study selection process.
Records were identified from five electronic databases (PubMed, Web of Science, Scopus, SPORTDiscus, PsycINFO; total n = 456). After title/abstract screening and full-text assessment, six controlled trials met eligibility criteria and were included in the systematic review and meta-analysis.
Table 1.
Characteristics of included studies. RCT = randomised controlled trial; Non-RCT = non-randomised controlled trial; VR = virtual reality; UTR = Universal Tennis Rating; RoB = overall risk of bias (RoB 2 for RCTs; ROBINS-I for non-RCTs). Effect sizes are Hedges’ g with HKSJ-adjusted 95% confidence intervals.
Fig 2.
Forest plot of the overall meta-analysis comparing VR-based training with non-VR controls on racket sports performance outcomes (k = 6 studies; N = 401).
Effect sizes are Hedges’ g with Hartung–Knapp–Sidik–Jonkman (HKSJ) adjusted 95% confidence intervals. The pooled estimate was g = 0.78 (95% CI [0.41, 1.15]). The 95% prediction interval [−0.01, 1.57] is shown as a horizontal bar at the bottom of the plot. Heterogeneity: I² = 52.4%, τ² = 0.0597, p = 0.062. Square size is proportional to study weight; horizontal lines represent 95% CIs; the diamond represents the pooled estimate.
Fig 3.
Forest plot of the subgroup analysis by intervention type.
The Physically Active VR subgroup (k = 5) yielded a pooled effect of g = 0.78 (95% CI [0.31, 1.25]; I² = 61.9%). The single Perceptual-Cognitive VR study (Anguera et al., 2025; k = 1) is presented descriptively only (g = 0.81, 95% CI [−0.05, 1.67]); no heterogeneity statistics are reported for this single-study subgroup. Overall heterogeneity statistics (all k = 6) are reported at the bottom of the figure. Effect sizes are Hedges’ g with HKSJ-adjusted 95% confidence intervals.
Fig 4.
Funnel plot of standardised mean difference (Hedges’ g) against standard error for all included studies (k = 6).
The plot is presented for descriptive purposes only; with k = 6 studies, formal tests for small-study effects (e.g., Egger’s test) lack sufficient power and were not performed. Visual inspection suggests slight asymmetry, but no definitive conclusions regarding publication bias can be drawn from this plot alone.