This review is a systematic review and meta-analysis. The methodology was conducted in accordance with the Cochrane Handbook for Systematic Reviews of Interventions [27]. The reporting of this review fully complies with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) 2020 statement and its extension checklists [28], while the search process adhered to the PRISMA-S (Preferred Reporting Items for Systematic Reviews and Meta-Analysis–Search) statement for searching [29]. The protocol for this review was prospectively registered in the International Prospective Register of Systematic Reviews (PROSPERO; registration no CRD42025645987), and the study was conducted without any deviations from the preregistered protocol. It should be noted that the primary outcome originally registered was perioperative anxiety. During the conduct of the review, as the understanding of perioperative stress responses evolved and relevant recent literature emerged, we recognized that depression, blood pressure, and heart rate are pathophysiologically interconnected with anxiety and are of comparable clinical importance for perioperative nursing decisions. Consequently, while maintaining the original study design, inclusion criteria, and analytical framework, we expanded the outcome set to include depression, blood pressure, and heart rate as secondary outcomes. All meta-analytic methods, heterogeneity assessments, and effect size syntheses were applied uniformly to the finalized set of outcomes.

Two reviewers (SW and ZY) independently conducted the initial screening of titles and abstracts, followed by a full-text review, strictly according to the inclusion and exclusion criteria. EndNote (version 21; Clarivate) software was used to remove duplicate records. Any discrepancies arising during the screening process were resolved through discussion or consultation with a third reviewer (HY) to finalize the list of included studies.

The 2 reviewers (SW and ZY) independently extracted data using a standardized data extraction form. The extracted information included basic study information (author, publication year, and country or region), participant characteristics (sample size, age, and type of surgery), intervention details (type of VR device, intervention content, timing of intervention, and duration of intervention), outcome measures (measurement tools, mean, SD, 95% CI, and sample size), and methodological characteristics of the studies (randomization method, implementation of blinding, and information related to risk of bias).

This review confirmed that all participants included in the analysis were mutually independent, with no instances of double-counting. For studies where multiple VR intervention arms shared a single control group, we followed the guidelines of the Cochrane Handbook for Systematic Reviews of Interventions [27] by evenly dividing the control group sample size while keeping the mean and SD unchanged. This approach prevents the inflation of statistical weight caused by double-counting the control group sample, thereby ensuring independence among comparisons.

Two investigators (SW and ZY) independently assessed the risk of bias for the included RCTs using the Cochrane Risk of Bias 2 tool (Cochrane Collaboration) [30]. The assessment domains included the randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result. The evaluation results were presented using a traffic light system (green=low risk, yellow=some concerns, and red=high risk) to generate a risk of bias summary and graph. Any discrepancies were resolved through consensus or by consulting a third investigator (HY).

The quality of evidence for each outcome indicator was evaluated according to the criteria established by the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) Working Group [31]. The grading dimensions encompassed study limitations, inconsistency, indirectness, imprecision, and publication bias. The certainty of the evidence was accordingly downgraded based on the specific conditions of each dimension. Ultimately, the grading results were presented using the standardized summary of findings table provided by the GRADE Working Group.

Potential publication bias was evaluated using funnel plots and the Egger linear regression test for the included studies. The trim-and-fill method was used if asymmetry was detected.

This study is a systematic review and meta-analysis based on previously published literature. As it did not involve primary data collection or direct interaction with human participants or animals, ethical review and approval by an institutional review board, as well as informed consent, were not required.

The literature screening process was strictly conducted in accordance with the PRISMA 2020 statement, and the detailed procedure is illustrated in Figure 1. A total of 3186 records were initially retrieved from all databases. After removing 1984 duplicate records using EndNote (version 21) software, 1202 papers remained. Two reviewers (SW and ZY) independently screened the titles and abstracts of these 1202 papers, excluding 769 irrelevant studies due to ineligible participants, inappropriate interventions, or irrelevant outcome measures, leaving 433 papers for full-text evaluation. During the full-text screening phase, an additional 250 studies were excluded for the following reasons: inappropriate interventions (121 studies), non-RCTs (97 studies), and missing data (32 studies). Finally, 42 RCTs were included in this meta-analysis [23,34-74]. Reasons for exclusion of full-text papers are provided in the PRISMA flow diagram in Figure 1.

The 42 included studies were conducted across multiple countries and regions, including China, Turkey, Spain, Iran, France, and others. As shown in Table 1, the studies covered various perioperative populations, including patients undergoing general, orthopedic, cardiovascular, obstetric and gynecological, and oral and maxillofacial surgery. The sample sizes of individual studies ranged from 19 to 423 participants, and all control groups received routine perioperative care.

Regarding intervention measures (Table 2), the presentation and content of VR interventions were highly diverse. Interventions were mainly classified into three categories: (1) natural and relaxing scenes (eg, underwater world, forest, beach, and VR experiences combined with mindful music); (2) perioperative procedural education (eg, first-person simulation of entering the operating room, anesthesia induction, and recovery to reduce fear of the unknown environment); and (3) immersive games or interactive rehabilitation training, with the intervention timing covering the preoperative period (eg, brief 10‐20 minutes intervention in the preoperative waiting area), intraoperative period (eg, continuous use for patients under local or neuraxial anesthesia), and postoperative rehabilitation phase.

Two reviewers (SW and ZY) independently assessed the risk of bias for the 42 RCTs using the Risk of Bias 2 tool. Overall, the included studies exhibited favorable methodological quality, and no study was rated as having a high overall risk of bias. The risk-of-bias plot (traffic light plot, Figure 2) revealed that several studies [23,34-74] were rated as having some concerns in the domains of the randomization process (Domain 1) and deviations from intended interventions (Domain 2). The main reasons included insufficient reporting of the specific methods used to conceal the random sequence and the difficulty in achieving complete blinding of participants and personnel due to the inherent nature of VR interventions. The GRADE quality-of-evidence ratings for each outcome measure are summarized in Table 3.

Seven studies [34,35,61,63,64,69,71] reported the effect of VR on depression in patients undergoing perioperative care, with a total sample size of 647 participants. The pooled effect size indicated that VR significantly reduced depression scores (SMD −1.26, 95% CI −1.71 to −0.81; P<.001). Significant heterogeneity was observed across studies (Q=25.93; P<.001; τ=0.423; τ²=0.179; I²=76.9%), and the 95% PI was −2.39 to −0.13. The forest plot is provided in Figure 4 [34,35,61,63,64,69,71].

Twenty studies [23,37-44,47,50,51,55,56,59,67-70,73] reported the effect of VR on heart rate in patients undergoing perioperative care, with a total sample size of 2781 participants. The pooled effect size showed that VR intervention significantly reduced heart rate (MD −4.45, 95% CI −5.94 to −2.97; P<.001). Significant heterogeneity was observed across studies (Q=123.20; P<.001; τ=2.629; τ²=6.914; I²=84.6%), and the 95% PI was −10.15 to 1.24. The forest plot is provided in Figure 7 [23,37-44,47,50,51,55,56,59,67-70,73].

A leave-one-out sensitivity analysis was performed to assess the robustness of the 5 outcome indicators: anxiety, depression, SBP, DBP, and heart rate. After sequentially removing individual studies, the pooled effect sizes were recalculated using the Hartung-Knapp-Sidik-Jonkman random-effects model. The results demonstrated that omitting any single study did not cause a directional change in the overall pooled effect sizes, and all pooled estimates remained statistically significant (P<.001), indicating that the meta-analysis results were highly robust. Forest plots of the sensitivity analysis are shown in Figure S1 in Multimedia Appendix 2.

Small-study effects were evaluated using funnel plots combined with the Egger linear regression test, which was performed only for outcomes with ≥10 included studies (anxiety, SBP, DBP, and heart rate):

None of the Egger tests for the above outcomes were statistically significant (P>.05), indicating no significant small-study effects. For depression (n=7), the Egger regression test suggested asymmetry in the funnel plot (t=−5.68; P=.002). Further sensitivity verification using the trim-and-fill method showed that the pooled effect size after adjustment was consistent with the original result (SMD −1.26, 95% CI −1.71 to −0.81; P<.05). This suggests that the observed funnel plot asymmetry was more likely caused by heterogeneity in VR intervention content and timing across studies, rather than publication bias, further confirming the authenticity and robustness of VR in improving perioperative depression. The results are shown in Figure S2 in Multimedia Appendix 3.

This review aimed to comprehensively evaluate the efficacy of VR interventions on psychological emotions and physiological stress in patients undergoing perioperative care. Consistent with our initial hypotheses, the findings from 42 included RCTs confirm that, compared to conventional perioperative care, adjunctive VR interventions significantly reduce patients’ anxiety and depression levels. Furthermore, the results validate our hypothesis regarding objective physiological metrics, demonstrating that VR effectively lowers SBP and DBP while decreasing heart rate. These dual regulatory benefits offer strong, evidence-based support for VR as a safe and efficacious nonpharmacological intervention for stabilizing perioperative stress.

Subgroup analyses showed that the anxiolytic effect of VR was more pronounced in Asian populations. The 95% PI indicated that the effect of VR on depression was stable in most clinical settings, whereas its effects on anxiety, blood pressure, and heart rate were influenced by heterogeneity and might not be significant in some specific contexts. Sensitivity analyses confirmed the high robustness of the results. Assessment of small-study effects revealed no significant small-study effects for any outcome except depression.

This study demonstrated that VR technology effectively alleviates anxiety and depression in patients undergoing perioperative care, which is consistent with findings from previous relevant systematic reviews [75]. The core mechanisms underlying the psychological stress-relieving effect of VR are mainly attributed to distraction theory and information processing theory [76]. On the one hand, the perioperative environment (eg, alarms from monitoring equipment and unfamiliar instruments) serves as a direct stressor triggering anxiety and depression in patients [77]. By constructing a highly immersive 3D audio-visual environment, VR can effectively redirect patients’ limited attention, disconnect them from the negative surrounding environment, and thereby reduce the development of depressive symptoms [76,78]. On the other hand, for anticipatory anxiety induced by the unknown nature of surgery, preoperative full-process scenario simulation via VR (eg, first-person experience of anesthesia induction) can significantly reduce information asymmetry between physicians and patients [20]. When patients develop a clear understanding of the treatment procedure and gain a sense of control, their inner uncertainty and helplessness are substantially diminished [49].

Furthermore, subgroup analyses revealed an important finding: the anxiolytic effect of VR intervention was more significant among patients in Asian regions. This is consistent with the cross-cultural findings reported by Streuli et al [79]. Possible reasons include that in Asian cultures, patients tend to express negative emotions less directly; VR, as an implicit and nonconfrontational intervention, better matches the psychological acceptance of this population [80]. Most studies [23,35,37,39,47-49,51,56,58,66] conducted in Asian regions used VR scenes featuring natural landscapes, which may enhance patients’ immersive experience and adherence to interventions. In addition, intraoperative interventions exhibited extremely low heterogeneity. This may be attributed to the fact that patients under local or neuraxial anesthesia during surgery are more susceptible to distraction by VR content, thus reducing between-study heterogeneity [39,41].

At the physiological level, this review confirmed that VR effectively reduces blood pressure and heart rate, findings highly consistent with those of the single-center clinical trial by Ugras et al [23]. Patients undergoing perioperative care are under severe psychological stress due to fear and uncertainty, which directly activates the hypothalamic-pituitary-adrenal axis and sympathetic nervous system, leading to massive catecholamine release, followed by tachycardia, peripheral vasoconstriction, and hypertension [81]. Through its powerful psychological soothing effects, VR interrupts this psychosomatic stress cycle. Functional magnetic resonance imaging studies have confirmed that immersive VR experiences not only regulate the balance of the autonomic nervous system (ie, reducing sympathetic activity and enhancing parasympathetic tone) but also significantly decrease activation in brain regions involved in pain and emotional processing, such as the anterior cingulate cortex, insula, and somatosensory cortex [82]. Such inhibitory effects at the central neural level represent the fundamental physiological basis for VR to stabilize perioperative hemodynamic parameters.

The findings of this review carry important implications for clinical nursing practice. First, as key providers of perioperative care, health care professionals act as a bridge between technology and patients in VR interventions [83]. They are required not only to operate the equipment but also to assess patients’ digital literacy, medical history, and preferences before intervention to select the most suitable VR content (eg, relaxing natural scenes or cognitive educational videos). During intervention implementation, they should closely monitor changes in patients’ heart rate and blood pressure, which serve as objective indicators for dynamically adjusting the intervention protocol [84].

Second, although the clinical efficacy of VR technology has been confirmed, its widespread implementation still faces cost-effectiveness considerations. The initial investments in hardware procurement, software customization, and operational training are relatively high [85]. Nevertheless, several health economic studies on digital therapeutics have demonstrated that, in the long run, VR interventions help reduce postoperative analgesic requirements, shorten anesthesia recovery time, and decrease overall length of hospital stay, thus yielding significant cost-effectiveness [86,87]. Therefore, hospital administrators should focus on the overall health economic value when introducing this technology.

Regarding the review process itself, no significant process-specific limitations were identified. This review is the first to comprehensively evaluate VR interventions in the perioperative setting by incorporating anxiety, depression, and objective hemodynamic parameters (blood pressure and heart rate). It overcomes the limitation of previous studies [24-26] that were only focused on anxiety alone and more comprehensively reveals the intervention effect of VR on both psychological and physiological stress during the perioperative period, which is consistent with the pathophysiological characteristics of perioperative stress [3,81]. Through subgroup analyses, the impacts of region and intervention timing on VR efficacy were identified, providing evidence for the clinical design of personalized VR intervention protocols—for example, VR can be prioritized for Asian patients, and standardized intraoperative VR programs can be adopted to reduce heterogeneity.

This review has several limitations that should be considered when interpreting the results. First, substantial clinical heterogeneity was observed across the included studies regarding types of VR devices (eg, head-mounted display models), intervention content (relaxation-based vs education-based), exposure frequency, and specific surgical types [23,49], which may have reduced the precision of the pooled effect sizes to some extent. Second, owing to the inherent physical characteristics of VR interventions, it was difficult for the included studies to achieve genuine double-blinding of patients and implementing nurses [20,75], which inevitably introduced risks of performance bias and measurement bias. Third, the number of studies included on depression was relatively small [34,35,61,63,64,69,71], leading to low-quality evidence. Therefore, the corresponding results should be interpreted with caution, and more high-quality studies focusing on VR interventions for perioperative depression are needed in the future.

Future studies should standardize the reporting of VR devices, intervention content, and exposure parameters and clarify the impact of different surgical types and intervention protocols on pooled effect sizes. To address bias arising from blinding limitations, it is recommended to strictly implement outcome-assessor blinding and include objective physiological indicators to ensure reliable results. Furthermore, researchers should conduct multicenter, large-sample randomized trials and extend follow-up periods to build high-quality evidence regarding the long-term efficacy of VR.

This systematic review and meta-analysis demonstrate that VR serves as a safe and effective adjunctive intervention that exhibits a dual regulatory mechanism in perioperative adult patients. Based on the 95% CIs, VR significantly alleviates average psychological distress (anxiety and depression) and stabilizes physiological indicators (SBP, DBP, and heart rate). However, while the CIs confirm a robust average benefit, the wide 95% PIs indicate that, due to substantial clinical heterogeneity, the true effect of VR for an individual patient in future clinical settings may vary widely. Furthermore, these findings should be interpreted cautiously; due to the inherent difficulty of double-blinding in VR interventions and the diverse VR contents and devices used, the GRADE assessment indicates a moderate-to-low certainty of evidence across all outcomes.

The innovation of this study lies in its comprehensive evaluation framework. Unlike existing systematic reviews that are predominantly limited to a single psychological metric (eg, anxiety alone) or focused on pediatric populations, this review innovatively integrates both psychological and objective physiological dimensions to comprehensively assess the perioperative psychosomatic stress response. Contributing to the field, this multidimensional assessment clarifies the dual regulatory mechanism of VR in mitigating perioperative stress. In terms of real-world implications, this review provides valuable, evidence-based guidance for health care professionals to implement personalized digital nonpharmacological interventions—such as prioritizing VR use for Asian perioperative populations, where the anxiolytic effect is notably pronounced. Ultimately, integrating VR into clinical practice can reduce reliance on pharmacological interventions, improve the patient care experience, and enhance the quality of postoperative recovery. Future multicenter, large-sample RCTs with standardized protocols and strict blinding are warranted to further elevate the quality of evidence.