Effects of Virtual Reality Interventions on Balance in Community-Dwelling Older Adults: A Systematic Review

Objectives: This systematic review compared the effects of virtual reality (VR) interventions on balance in community-dwelling older adults with those of inactive and active comparators. Methods: PubMed, PEDro, CINAHL and Embase were searched from inception to January 2025, and the PubMed search was updated on 8 October 2026 for randomised controlled trials (RCTs) of VR interventions in ambulatory community-dwelling older adults with a mean age of 60 years or above in every study arm. Risk of bias was assessed with RoB 2. Because comparators, outcome instruments and available statistics were too heterogeneous for meta-analysis, findings were synthesised by comparator type and outcome domain using the direction of the between-group effect on 1 prespecified outcome per trial and domain, irrespective of statistical significance, and the certainty of evidence was rated with GRADE. Results: Thirty-five reports of 33 RCTs were included, 15 of them identified in the updated search; overall risk of bias was low in three trials, raised some concerns in 20 and was high in 10. Forty-eight further reports from the updated search are awaiting classification. Against inactive comparators, the prespecified clinical balance outcome favoured VR with a clear between-group difference in six of 17 comparisons; all in trials with some concerns regarding the risk of bias and none of the three trials at low risk of bias showed a clear difference. Against active comparators of similar dose, clear differences occurred in both directions (two favouring VR and two favouring the comparator, out of 13), so no consistent advantage for VR was shown; the trials were small and rarely reported effect estimates with confidence intervals, so equivalence could not be established. Four trials reported falls; in the largest, at low risk of bias, home-based stepping exergames reduced the rate of falls compared with an education control (incidence rate ratio 0.74, 95% confidence interval 0.56–0.98). A sensitivity analysis restricted to trials meeting the registered age criterion did not change these findings. Conclusions: VR-based training may improve balance compared with no training, and home-based stepping exergames probably reduce falls. There is no evidence that VR is superior to conventional balance training of comparable dose, but the available trials cannot show whether it is equivalent, and the certainty of the balance evidence is low to very low. VR can be considered an alternative format for delivering balance training, pending adequately powered comparative trials.