Abstract This systematic review and meta-analysis aimed to evaluate the effects of virtual reality (VR) interventions on anxiety symptoms in women, based exclusively on randomized controlled trials (RCTs) conducted in female-only samples. As of August 15, 2025, RCTs written in English were systematically searched in databases including PubMed, Web of Science, PsycINFO, and the Cochrane Library. The inclusion criteria were as follows: VR interventions, including virtual reality exposure therapy, 360° VR videos, VR-based educational programs, and interactive virtual environments, delivered in at least one structured session; participants included women from various groups—such as pregnant and postpartum women, breast cancer patients, older women, and adolescents—and treatment effects were evaluated using standardized anxiety scales. A meta-analysis was performed in R (effect size: Hedges’ g), including subgroup analysis, meta-regression, and XGBoost machine learning to identify potential moderating variables. The risk of bias was evaluated using the Cochrane ROB2 tool, and publication bias was analyzed with funnel plots, Egger’s test, and methods like standardized residuals and Cook’s distance. Twelve RCTs involving 1,038 participants were included in the analysis.The random-effects model showed a moderate reduction in anxiety symptoms (standardized mean difference SMD = − 0.59, 95% confidence interval CI − 1.09 to − 0.08, P = 0.0240), although heterogeneity was high (I² = 90.4%). Sensitivity analysis and Egger’s test indicated the presence of outlier studies. After excluding these studies, the revised analysis showed that VR interventions significantly reduced anxiety symptoms in women (SMD = − 0.37, 95% CI − 0.69 to − 0.05, P = 0.0234), with a notable decrease in heterogeneity. Moderator analyses indicated that interventions were more effective when using blank-control designs (more effective than active interventions), when targeting breast cancer patients undergoing chemotherapy (SMD = − 1.12), and when the interventions frequency was once per week (SMD = − 0.38), with all results yielding P < 0.05. Additionally, meta-regression analyses showed that the type of anxiety assessment scale (β = 1.28, P < 0.001) and participant type (β = 1.11, P = 0.001) significantly influenced the effect sizes. An optimal implementation pattern was recommended based on restricted nonlinear regression analysis: one session per week, each lasting 60 min, with a total intervention dose ranging from 6.3 to 9.3 h. VR interventions showed a small but statistically significant effect with substantial heterogeneity in women, with more notable effects among breast cancer patients undergoing chemotherapy, studies using blank-control designs, and studies using the STAI-S scale. It is recommended that VR intervention parameters be customized individually for various female populations, considering cultural background and other factors to improve acceptability and effectiveness.
Lv et al. (Sun,) studied this question.