Effects of virtual reality mindfulness interventions on depression and anxiety: a meta-analysis

ABSTRACT

Background: Depression and anxiety are common mental health problems that frequently co-occur with trauma-related disorders and often develop following traumatic or stressful events, substantially impairing quality of life. Mindfulness-based interventions have been shown to reduce these symptoms; however, their widespread implementation is often limited by poor accessibility and low adherence. Virtual reality (VR) may enhance mindfulness delivery through immersive experiences, but the overall efficacy of VR-based mindfulness interventions and the factors influencing their effectiveness remain unclear.

Objective: This exploratory meta-analysis aimed to evaluate the effectiveness of VR-based mindfulness interventions in reducing depressive and anxiety symptoms and to explore potential moderators (population, region, immersion level, and intervention duration).

Methods: A systematic literature search was conducted in PubMed, Web of Science and other databases up to 2025. Randomized controlled trials (RCTs) investigating VR-based mindfulness interventions for depression and/or anxiety were included. Meta-analyses were conducted using RevMan 5.4, calculating standardized mean differences (SMDs) with 95% confidence intervals (CIs).

Results: Fifteen RCTs involving 864 participants were included. VR-based mindfulness interventions were associated with significant reductions in depressive and anxiety symptoms. Subgroup analyses indicated larger effect sizes among older adults (SMD = −1.40, p = .005) and in studies conducted in Eastern regions (SMD = −1.07, p < .00001). Both immersive and semi-immersive VR formats were effective. Interventions of short duration (≤4 weeks) and moderate duration (>6 to ≤8 weeks) showed benefits, and session lengths of ≤30 min were associated with greater symptom reduction (SMD = −0.93, p = .005).

Conclusions: This exploratory meta-analysis suggests that VR-based mindfulness interventions are associated with reductions in depression and anxiety symptoms, moderated by intervention duration and regional cultural context. These findings support the potential of VR-based mindfulness as an accessible adjunctive approach, though high-quality studies with longer follow-up are needed to confirm sustained effects and improve generalizability.

HIGHLIGHTS

• Virtual reality mindfulness interventions were associated with significant reductions in depression and anxiety symptoms across randomized controlled trials.

• Greater effects were observed among older adults, participants from Eastern regions, and interventions with shorter session durations.

• Both immersive and semi-immersive virtual reality formats demonstrated beneficial effects on emotional outcomes.

1. Introduction

Globally, depression and anxiety disorders have become major public health concerns, with high recurrence rates and limited long-term remission posing persistent challenges to existing mental health service systems (Kuyken et al., 2016). Depression and anxiety not only present as independent mental disorders but also frequently occur as comorbid or secondary symptoms of posttraumatic stress disorder (PTSD) or other trauma-related conditions. A history of trauma is a well-established risk factor for mood and anxiety disorders, with many patients developing persistent depressive or anxiety symptoms following traumatic events (Kessler et al., 2005; Wang et al., 2023).

Given the persistent challenges in managing trauma-related depression and anxiety, mindfulness-based interventions have gained prominence as a valuable therapeutic option. Mindfulness, originating from Buddhist contemplative traditions, refers to a mental state characterized by purposeful attention to present-moment experiences with a nonjudgmental and accepting attitude (Kabat-Zinn, 2015). Mindfulness-based therapies developed from this concept have demonstrated effectiveness in reducing the severity of depressive and anxiety symptoms across diverse populations (Hofmann & Gómez, 2017). These interventions emphasize structured and systematic practices aimed at enhancing awareness of moment-to-moment psychological processes and internal experiences (Grossman et al., 2004). Among them, established approaches such as Mindfulness-Based Cognitive Therapy (MBCT) and Mindfulness-Based Stress Reduction (MBSR) have been shown to be associated with symptom alleviation (Querstret et al., 2020).

Despite their demonstrated efficacy, traditional mindfulness-based interventions continue to face limitations in real-world clinical implementation, including issues related to accessibility, engagement, and adherence (Pedro et al., 2021). First, traditional mindfulness training relies heavily on professional guidance and structured curricula, which limits access to standardized interventions for individuals living in regions with constrained healthcare resources (Ainsworth et al., 2023). Second, traditional mindfulness practices place substantial demands on individuals’ attentional regulation abilities and require relatively stable external environments, contributing to persistently high dropout rates, particularly among patients with impaired attentional functioning (van der Velden et al., 2015). In light of these challenges related to accessibility, adaptability, and adherence, the development of novel mindfulness intervention models that are flexible, scalable, and secure has become a central research focus in the prevention and treatment of depression and anxiety (Strauss et al., 2021).

Against this backdrop, virtual reality (VR) technology – defined as a computer-generated, interactive three-dimensional environment typically delivered via head-mounted displays (HMDs), motion trackers, or large-scale projection systems – provides an innovative approach to optimizing traditional mindfulness training. VR enhances feasibility and engagement through several positive aspects: (1) immersive experiences that create a strong sense of ‘being there’ (presence), aligning with Attention Restoration Theory (ART) to facilitate restored focused attention (Ohly et al., 2016); (2) customizable scenarios that allow tailoring of virtual environments to individual preferences or therapeutic needs (e.g. calm nature scenes or guided imagery), promoting decentering – the process of detaching from negative self-labels (Blackmore et al., 2024); and (3) improved accessibility, enabling remote or home-based delivery without requiring physical presence in clinical settings (Freeman et al., 2017). This multisensory integration effectively minimizes external distractions and enhances users’ sense of presence, thereby reducing attentional dispersion and facilitating sustained awareness of present-moment physical and psychological states (Liu et al., 2023; Strauss et al., 2021).

Building on these VR advantages, two core mechanisms – immersive emotional regulation and decentering – underlie the therapeutic effects of VR-based mindfulness interventions (Failla et al., 2022). Immersive emotional regulation refers to VR’s multisensory stimulation aiding emotional state modulation (Colombo et al., 2021); decentering refers to shifting attention away from self-centered rumination through external virtual stimuli, thereby interrupting negative emotional cycles (Blackmore et al., 2024). Multiple studies support these dual pathways. For example, Zheng et al. (2024) conducted randomized controlled trial using self-report measures such as the Patient Health Questionnaire-9 (PHQ-9) and Generalized Anxiety Disorder-7 (GAD-7). Their findings showed that the immersion pathway in VR-based mindfulness reduces stress and anxiety, while the decentering pathway from mindfulness practice helps alleviate depressive symptoms. Similarly, Neus et al. (2024) conducted a quasi-experimental study using pre–post changes in the PHQ-9 scores and subjective patient reports to support the synergistic role of the dual pathways: immersive environments provide a safe emotional buffer, and decentering reduces rumination, associated with notable symptom improvement in depressed patients with comorbid physical illnesses.

However, not all studies have reported consistent benefits of VR-based mindfulness therapy for depressive and anxiety symptoms. For example, Modrego-Alarcón et al. (2025) found that, in non-clinical adults, improvements in core depressive symptoms assessed by the Depression Anxiety Stress Scale-21 (DASS-21) scale were smaller in the mindfulness-based practice plus VR (MBP + VR) group than in the mindfulness-based practice (MBP) group alone. Additionally, Marocco et al. (2025) noted that while VR-based mindfulness therapy can reduce negative emotions, it did not show significant enhancement in participants’ positive emotions or the curiosity dimension within mindfulness – a dimension closely linked to the ability to alleviate depression and anxiety.

Despite the growing interest in VR-based mindfulness interventions, the field remains in its developmental stage. Important questions regarding its efficacy and key implementation details still require in-depth exploration. To address these gaps, this exploratory meta-analysis aims to quantitatively evaluate the efficacy of VR-based mindfulness interventions on depressive and anxiety symptoms. Population characteristics (e.g. age) and geographical region (Eastern vs. Western) were selected as potential moderators because age influences mindfulness acceptance and engagement, while cultural and socioeconomic differences affect VR accessibility, attitudes toward mental health interventions, and response to mindfulness practices (Hofmann & Gómez, 2017; Yang et al., 2025).

Specifically, this study aims to conduct a meta-analysis integrating empirical research data on VR-based mindfulness therapy, with the following core objectives:

1. Quantitatively assess the efficacy of VR-based mindfulness therapy in improving depressive and anxiety symptoms compared to control conditions;

2. Explore the influence of population, geographical differences, VR device immersion levels, and intervention duration on treatment outcomes through subgroup analyses;

3. Evaluate study quality and risk of bias to provide scientific grounds for interpreting results.

2. Methods

2.1. Data sources and search strategies

This study systematically searched eight electronic databases, including PubMed, Web of Science, Scopus, Embase, CINAHL, PsycINFO, China National Knowledge Infrastructure (CNKI), and Wanfang Data. The search strategy combined subject headings and free-text terms, focusing on three core domains: virtual reality (e.g. ‘virtual reality,’ ‘VR’), mindfulness-based therapy (e.g. ‘mindfulness,’ ‘mindfulness therapy’), and depression and anxiety-related concepts (e.g. ‘depression,’ ‘depressive symptoms,’ ‘depressive disorder,’ ‘anxiety,’ ‘anxiety symptoms,’). These terms were integrated using Boolean operator ‘AND.’ The search was limited to publications in Chinese and English, because the research team is proficient in these languages, and these databases cover the majority of high-quality studies in this field. Additionally, to ensure a comprehensive literature search, this study manually reviewed the titles of articles listed in the references section of included studies. The search was conducted from the earliest available records in each database up to December 2025.

2.2. Inclusion and exclusion criteria

Following the PRISMA guidelines for systematic reviews and meta-analyses (Page et al., 2021)，this study established clear and rigorous literature screening criteria. Inclusion criteria were formulated based on the Participants, Interventions, Comparisons, Outcomes, and Study Design (PICOS) format, as detailed below: (a) Study subjects: No restrictions on population; (b) The experimental group received virtual reality-based mindfulness therapy; (c) The control group could include conventional treatment, traditional mindfulness training, or other interventions; (d) Outcomes primarily focused on quantifiable assessments of depression and anxiety; (e) Study types were restricted to randomized controlled trials to ensure high-quality evidence.

Exclusion criteria are as follows: (1) Non-Chinese or non-English language publications; (2) Full-text unavailability; (3) Duplicate publications; (4) Case reports, qualitative studies, non-randomized designs, conference abstracts, and studies with incomplete data reporting.

2.3. Data extraction

To ensure standardization and consistency in data collection, this study employed a pre-designed data extraction form. Two systematically trained researchers independently performed the data extraction. Extracted content encompassed core elements including: study background information (authors, publication year, and country); participant characteristics (sample size, gender and age distribution); intervention and control design details (VR device type, immersion characteristics, virtual scenario content, intervention duration and frequency, and control measures); outcome measures and assessment tools (with a focus on depression and anxiety scales); follow-up status; and study limitations.

Immersion characteristics were coded according to Jiang et al. (2022): with explicit operationalization: (1) Immersive VR: Head-mounted displays (HMDs) integrated with full-field 3D visualization and real-time motion tracking (e.g. Oculus Quest 2, HTC VIVE), which isolate users from the physical environment to create a strong sense of ‘presence’; (2) Semi-immersive VR: Partial sensory setups without motion tracking (e.g. large-screen projection systems, interactive rehabilitation equipment like IREX), delivering virtual content via non-isolating interfaces; (3) Non-immersive VR: Desktop-based 2D/3D interfaces (not included in subgroup analyses due to insufficient eligible data). Coding details are provided in Table 1.

Table 1.

Characteristics of the included studies (N = 15).

		Participants			Intervention group
Author(s)	Country	Sample size	Age	Disorder Status	Intervention methods	Device	Frequency & duration	Control group	Main outcomes (instruments)
Burrows et al. (2023)	USA	N = 9, IG = 5, CG = 4	IG:59.67 ± 14.12 CG:59.50 ± 15.07	YES	VRM + PARx	Oculus Quest 2 (immersive)	Six sessions, 20-25min each, twice a week, 2 weeks	PARx	Depression (PROMIS), Anxiety (SAS)
Chen and Hongyan (2019)	China	N = 112, IG = 56, CG = 56	IG:54.6 ± 3.6 CG:54.9 ± 3.8	YES	MT + VRT	IREX (semi-immersive)	Ten times a week, 8 weeks	Usual care	Depression (SDS)
Dai et al. (2025)	China	N = 50, IG = 25, CG = 25	IG:69.48 ± 7.8 CG:69 ± 7.23	YES	VR + MBSR	RX-AP (semi-immersive)	1 h each，three times a week, 8 weeks	Usual care	Depression(HAMD),Anxiety(HAMA)
Chavez et al. (2020)	America	N = 29, IG = 8, CG1 = 11, CG2 = 10	IG:21.9 CG1:21.7 CG2:21.2	NO	VR meditation	Oculus Go (immersive)	10 min each, one time	Audio meditation/VR imagery	Anxiety(STAI)
Lee et al. (2023)	Korea	N = 61, IG = 33, CG = 28	IG:32.94 ± 11 CG:31.32 ± 10.37	YES	VBM	Oculus Rift CV1 (immersive)	Eight sessions, 30 min each, once a week, 8 weeks	VR	Depression(BDI)
Li et al. (2018)	China	N = 102, IG = 49, CG = 53	IG:56.76 ± 6.84 CG:57.14 ± 10.33	YES	MT + VRT	IREX (semi-immersive)	Eight sessions, 80-100 min each, once a week, 8 weeks	Usual care	Depression(SDS)
Liu et al. (2024)	China	N = 42, IG = 15, CG1 = 13, CG2 = 14	IG:20.9 ± 2.5 CG1:20.5 ± 2.7 CG2:20.4 ± 2.5	NO	VRM	HTC VIVE (immersive)	Twelve sessions, 10 min each, three times a week, 4 weeks	Audio mindfulness/mindfulness	Anxiety(SAS)
Modrego-Alarcón et al. (2025)	Spain	N = 36, IG = 12, CG1 = 15, CG2 = 9	IG:46.75 ± 10.08 CG1:49.27 ± 12.81 CG2:51.31 ± 8.24	NO	MBP + VR	Broomx MK360 immersive projection system (semi-immersive)	80 min each, once a week, 6 weeks	MBP / Relaxation	Depression(DASS-21)
Navarro-Haro et al. (2019)	Spain	N = 39, IG = 19, CG = 20	IG:45.05 ± 8.17 CG:45.4 ± 13.74	YES	MBI + VR	Oculus Rift DK2 (immersive)	Twice a week, 6 weeks	MBI	Depression(HADS), Anxiety(HADS)
Neus et al. (2024)	Spain	N = 26, IG = 13, CG = 13	IG:49.85 ± 2.09 CG:50.77 ± 5.01	YES	IVR + MBI, cognitive training and physical exercise	MK360 system (semi-immersive)	1 h each, twice a week, 8 weeks	Usual care	Depression(PHQ-9), Anxiety(GAD-7)
Xu et al. (2026)	China	N = 75, IG = 25, CG1 = 25, CG2 = 25	IG:19.52 ± 19.6 CG1:19.76 ± 11.75 CG2:20.68 ± 13.65	NO	VR-MG	HTC VIVE (immersive)	5 min each, one time	MG/CG	Anxiety(STAI)
You (2025)	China	N = 98, IG = 49, CG = 49	IG:45.36 ± 5.27 CG:46.05 ± 5.39	YES	MT + VR	-	8 weeks	Usual care	Depression(HAMD),Anxiety(HAMA)
Zhang et al. (2024)	China	N = 60, IG = 30, CG = 30	34.38 ± 10.76	NO	Virtual Reality – Based Meditation	Model, PRO6DOF (immersive)	20 min each, once a day, 2 weeks	Usual care	Depression(CES-D), Anxiety(SAI)
Zhang (2023)	China	N = 84, IG = 42, CG = 42	IG:27 ± 4.22 CG:26.95 ± 4.17	YES	MT + VR	HTC VIVE VR (immersive)	30 min each, twice a week, 4 weeks	Relaxing Music VR	Depression(SDS), Anxiety(SAS)
Zheng et al. (2024)	China	N = 60, IG = 20, CG1 = 20, CG2 = 20	IG:20.2 ± 1.01 CG1:20.25 ± 0.91 CG2:20.35 ± 0.75	NO	MT + VR	A head-mounted display, a bracelet, and a set of headphones (immersive)	Seven sessions, 10-15 min each, 2 weeks	Relaxing Music VR/Blank Control	Depression(PHQ),Anxiety(GAD)

Note:‘Disorder Status’ indicates whether participants had any diagnosed mental disorder or physical illness (Yes = with disorder/illness; No = without disorder/illness, including healthy or subclinical populations). This is not limited to depression or anxiety diagnoses.

Abbreviations: IG, intervention group; CG, control group; VRM, virtual reality meditation; PARx, personalized activity prescription; MT, mindfulness training; VRT, virtual reality therapy; VR, virtual reality; MBSR, mindfulness-based stress reduction; VBM, virtual reality-based meditation; MBP, mindfulness-based programme; MBI, mindfulness-based intervention; IVR, immersive virtual reality; VR-MG, virtual reality mindfulness group; IREX, Interactive Rehabilitation Exercise; RX-AP, Virtual Reality Psychological Crisis Auxiliary Intervention System (developed by Shenyang Ruihai Yinuo Intelligent Technology); HTC VIVE, HTC Vive virtual reality device; MK360, immersive projection system developed by Broomx Technologies (transforms a room into a multi-sensory, interactive environment); Model PRO6DOF, model with 6 degrees of freedom device; PROMIS, Patient-Reported Outcomes Measurement Information System; STAI, State-Trait Anxiety Inventory; BDI, Beck Depression Inventory; DASS-21, Depression Anxiety Stress Scales-21; HADS, Hospital Anxiety and Depression Scale; PHQ-9, Patient Health Questionnaire-9; GAD-7, Generalized Anxiety Disorder Scale-7; CES-D, Center for Epidemiologic Studies Depression Scale; GAD, Generalized Anxiety Disorder; PHQ, Patient Health Questionnaire.

2.4. Risk of bias assessment

The risk of bias in the included randomized controlled trials was assessed independently by two reviewers using the Cochrane Risk of Bias tool (RoB). Disagreements were resolved by discussion or consultation with a third reviewer. The tool assesses bias across seven domains: (1) random sequence generation (selection bias), (2) allocation concealment (selection bias), (3) blinding of participants and personnel (performance bias), (4) blinding of outcome assessment (detection bias), (5) incomplete outcome data (attrition bias), (6) selective reporting (reporting bias), and (7) other bias. Judgments were low risk, unclear risk, or high risk for each domain.

2.5. Statistical analysis

All statistical analyses in this study were performed using RevMan 5.4 software. For outcome measures, continuous variables were estimated using standardized mean differences (SMD) with 95% confidence intervals (CI), while categorical variables were pooled using risk ratios with 95% confidence intervals. The primary effect size of interest was the SMD, as it is suitable for integrating results from different scales and accounting for sample variability.

Subgroup analyses were conducted to explore potential moderators of the intervention effects. Studies were divided into subgroups based on population, geographic region, VR immersion levels (e.g. semi-immersive, immersive), and intervention duration (Yang et al., 2025). Differences in effect sizes between subgroups were examined using the test for subgroup differences, providing preliminary evidence of moderation (Borenstein et al., 2021). Publication bias assessment is visually presented using funnel plots, supplemented by quantitative validation through Begg's test and Egger's test (Page et al., 2021). The significance level for all statistical tests was set at p < .05, which is a conventional threshold in statistics to balance Type I error and practical significance (Lakens, 2013).

3. Results

3.1. Literature screening results

The literature screening process revealed that after deduplication and title/abstract screening of the 2,362 articles identified in the initial search, a large number of irrelevant records were excluded. The remaining 276 articles proceeded to full-text evaluation. Following full-text review, 15 studies ultimately met all inclusion criteria and were incorporated into this analysis (Figure 1).

Figure 1.

PRISMA flow chart for the study screening process.

3.2. Basic characteristics included in the study

This study included 15 randomized controlled trials involving a total of 864 adult participants. The included studies were originated from four countries: China, South Korea, the United States, and Spain, reflecting the international attention garnered in this field. Key characteristics of the included studies are detailed in Table 1. Primary outcomes were anxiety and depression symptoms, assessed almost entirely by self-report scales, with only two studies using clinician-rated scales: the Hamilton Depression Rating Scale (HAMD) and the Hamilton Anxiety Rating Scale (HAMA). Among them, the Self-Rating Depression Scale (SDS) and Self-Rating Anxiety Scale (SAS) were the most widely used self-report instruments, appearing in three studies each.

3.3. Literature quality assessment

Quality assessment of the 15 included studies was conducted using RoB (Figures 2 and 3). Higher risks of bias were primarily observed in blinding of participants and personnel (performance bias) and blinding of outcome assessment (detection bias), attributable to the inherent challenges in blinding participants to VR interventions. Allocation concealment showed a relatively high proportion of unclear or high risk. Risks were generally low in random sequence generation, incomplete outcome data (attrition bias), and selective reporting, with some unclear risks noted. Other bias showed mixed judgments with some high risk. Overall, the included studies demonstrated moderate quality.

Figure 2.

Risk of bias domains for all eligible trials according to Cochrane's RoB.

Figure 3.

Overall risk of bias graph.

4. Meta-analysis results

This meta-analysis included 15 studies encompassing 32 effect size groups. Considering the differences in assessment tools used across studies, the standardized mean difference (SMD) was selected as the statistical measure for combining effect sizes. This approach eliminates measurement scale variations caused by different assessment tools, enabling the comparability and integration of effect sizes from diverse studies. Typically, I²≥75% is defined as high heterogeneity (Sumsuzzman et al., 2024). Beyond sampling error, other significant sources of variation exist, such as differences in subject characteristics and variations in the form of VR-based mindfulness interventions, which may lead to differences in effect sizes across studies. Therefore, this study employs a random-effects model for data analysis. Given the substantial heterogeneity among the included studies, estimating only the average effect is insufficient. The random-effects model accounts for both sampling error within studies and variation in true effects across studies, yielding results that better reflect the actual situation (Kronthaler & Held, 2025).

4.1. Depression

Compared with the control groups receiving routine care or traditional mindfulness training, VR-based mindfulness interventions were significantly associated with greater reductions in depressive symptoms. As shown in Figure 4, the SMD was −0.87, with a 95% confidence interval of [−1.47, −0.28] and a p-value of .004.

Figure 4.

Meta-analysis on depression.

4.2. Anxiety

This study included 12 anxiety-related investigations, as shown in Figure 5. Compared with control groups, VR-based mindfulness interventions were significantly associated with greater improvements in anxiety symptoms (SMD = −0.49, 95% CI [−0.89, −0.08], p = .02).

Figure 5.

Meta-analysis on anxiety.

4.3. Subgroup analysis

Subgroup analyses examined potential moderators of VR-based mindfulness effects on depression and anxiety symptoms, including population, geographic region, intervention immersion level, duration, and single-session length (see Figures 6–10).

Figure 6.

Subgroup analysis of the impact of age on depression and anxiety.

Figure 7.

Subgroup analysis of the impact of region on depression and anxiety.

Figure 8.

Subgroup analysis of the effects of immersion on depression and anxiety.

Figure 9.

Subgroup analysis of the impact of intervention duration on depression and anxiety.

Figure 10.

Subgroup analysis of immersive VR: Single intervention time<30 min/times vs ≥30 min/times.

4.3.1. Population

This study categorized included research into youth, middle-aged, and elderly subgroups based on the average age of participants (Luo, 2017). Subgroup analysis by population showed significant associations with reductions in depression and anxiety symptoms in the elderly group (SMD = −1.40, 95% CI [−2.26, −0.54], p = .001; I² = 73%). Statistically significant but smaller effects were observed in the youth group (SMD = −0.65, 95% CI [−1.18, −0.11], p = .02) and middle-aged group (SMD = −0.58, 95% CI [−1.15, −0.02], p = .04), with both groups exhibiting high heterogeneity. Overall, effects appeared descriptively larger among older adults. Detailed results are presented in Figure 6.

4.3.2. Geographic region

Subgroup analysis by geographic region (Eastern vs. Western countries) showed reductions in depression and anxiety symptoms for studies from Eastern countries (SMD = −1.07, 95% CI [−1.49, −0.64], p < .00001), while studies from Western countries showed no significant reduction (SMD = 0.12, 95% CI [−0.12, 0.36], p = .33). Subgroup differences were observed (p = .0002; see Figure 7). Overall, greater improvements were associated with studies from Eastern countries.

4.3.3. Immersion level

Subgroup analysis by immersion level (immersive vs. semi-immersive) was conducted on 30 effect sizes (two studies did not report VR device types). Significant reductions in depression and anxiety symptoms were observed for both immersive VR (SMD = −0.50, 95% CI [−0.88, −0.12], p = .01) and semi-immersive VR (SMD = −0.50, 95% CI [−0.90, −0.10], p = .01), with high heterogeneity in both subgroups. Overall, there were no significant differences in effects on depressive and anxiety symptoms between immersive and semi-immersive VR formats. Detailed results are shown in Figure 8.

4.3.4. Duration

Subgroup analysis based on intervention duration revealed that intervention length served as a significant moderator of effect, with statistically significant differences between subgroups (p = .0002). Short-term interventions (≤4 weeks) were associated with significant improvements (SMD = −0.72, 95% CI [−1.15, −0.28], p = .001; I² = 85%). Interventions lasting >4 to ≤6 weeks showed no significant improvements (SMD = 0.19, 95% CI [−0.09, 0.48], p = .18). Long-term interventions (>6 to ≤8 weeks) also showed significant benefits (SMD = −1.13, 95% CI [−1.79, −0.46], p = .001). Overall, short-term (≤4 weeks) and longer-term (>6 to ≤8 weeks) interventions were associated with greater improvements in depressive and anxiety symptoms. Overall, short-term and longer-term interventions were associated with greater improvements in depressive and anxiety symptoms. Detailed results are presented in Figure 9.

4.3.5. Single-session length

Subgroup analysis by single-session length (<30 min vs. ≥30 min) showed that sessions <30 min significantly improved depressive and anxiety symptoms (SMD = −0.93, 95% CI [−1.58, −0.28], p = .005; I² = 88%). In contrast, sessions ≥30 min showed a trend toward improvement, but the pooled effect size did not reach statistical significance (SMD = −0.36, 95% CI [−0.75, 0.04], p = .08; I² = 79%). Overall, single-session lengths <30 min were associated with greater improvements in depressive and anxiety symptoms. See Figure 10.

4.4. Sensitivity analysis

This study conducted a sensitivity analysis by sequentially excluding individual studies to assess the robustness of the results. The results indicate that after sequentially excluding each study, the variation in the heterogeneity index (I²) remained limited, suggesting that the overall level of heterogeneity was not significantly influenced by any single study. This supports the core finding of this meta-analysis – VR-based mindfulness interventions were associated with significantly greater improvements in depressive and anxiety symptoms compared to control groups – demonstrating robust validity independent of any specific study's outcome.

Although the initial analysis suggested high heterogeneity among studies, the sensitivity analysis still indicated that the estimated pooled effect size remained stable.

4.5. Publication bias

For depression and anxiety outcome measures, this study systematically assessed publication bias using three methods: Begg's rank correlation test, Egger's linear regression, and visual funnel plot analysis. The funnel plot, plotting effect size (SMD) on the x-axis and standard error on the y-axis, showed a broadly symmetrical distribution of study points, forming a typical inverted funnel shape with no evident asymmetry (Figure 11).

Figure 11.

Funnel plot.

5. Discussion

5.1. Core research findings and clinical implications of effect sizes

A meta-analysis of 15 randomized controlled trials demonstrated that VR-based mindfulness therapy produced a large effect size for depressive symptoms (SMD = −0.87) and a moderate effect size for anxiety symptoms (SMD = −0.49). From a clinical assessment perspective, this indicates that post-intervention depression and anxiety scores (e.g. HAMD, HAMA) were reduced by approximately 0.66 standard deviations compared with control conditions.

These findings suggest that VR-based mindfulness interventions may produce clinically meaningful improvements in emotional symptoms. Immersive virtual environments may facilitate these improvements by enhancing attentional engagement and present-moment awareness – core mechanisms underlying mindfulness-based interventions (Riva et al., 2019; Seabrook et al., 2020). Specifically, the strong sense of presence created by VR may help individuals with depression disengage from rumination and negative self-focus (Blackmore et al., 2024), while for anxiety, the controlled and customizable virtual settings may reduce hyperarousal and support gradual emotional regulation (Colombo et al., 2021). By minimizing external distractions and providing structured sensory cues, VR environments appear to promote sustained attention during mindfulness practice, thereby interrupting negative emotional cycles common in both conditions. These findings highlight the potential therapeutic value of immersive VR environments for supporting mindfulness practice and emotional regulation.

When compared with established mindfulness-based interventions, the effects observed in this meta-analysis appear relatively substantial. Previous studies have reported effect sizes of approximately 0.34 for depression and 0.48 for anxiety in MBCT programmes (Nandarathana & Ranjan, 2025; Wang et al., 2025). Although methodological differences across studies should be considered, the larger effect sizes observed in VR-based mindfulness interventions may reflect the enhanced experiential engagement provided by immersive environments (Xie et al., 2025).

5.2. The mechanism of action of VR-based mindfulness therapy

The present meta-analysis found that VR-based mindfulness interventions were significantly associated with reductions in both depressive and anxiety symptoms. These findings appear to be supported by multiple interacting mechanisms rather than a single technological factor.

To better understand these mechanisms, the following paragraphs examine how immersive environments and decentering processes may contribute to the observed effects.

This pattern is consistent with previous research suggesting that immersive technologies may enhance attentional focus, emotional regulation, and experiential engagement, which are fundamental processes in mindfulness practice (Riva et al., 2019; Seabrook et al., 2020). For example, Chandrasiri et al. (2020) reported that VR-based mindfulness training significantly increased mindfulness levels, and this increase was negatively correlated with anxiety scores, supporting the role of immersive environments in facilitating attentional disengagement from maladaptive rumination. Similarly, the controlled and customizable nature of VR scenarios has been shown to aid emotional regulation by providing a safe buffer against hyperarousal while allowing gradual exposure to emotional stimuli (Colombo et al., 2021; Javanbakht et al., 2024).

Taken together, these mechanisms suggest that VR may enhance the therapeutic effects of mindfulness interventions beyond traditional approaches. By minimizing external distractions and enabling user-controlled pacing, VR environments appear to support the interruption of maladaptive rumination in depression and the reduction of hyperarousal in anxiety. Such integration of immersive technology with mindfulness practice may therefore contribute to greater accessibility and effectiveness of emotion regulation strategies in clinical settings (Blackmore et al., 2024; Xie et al., 2025).

5.3. Sources and interpretation of inter-study heterogeneity

This study found significant heterogeneity across research on VR-based mindfulness therapy (I² = 83%), suggesting that despite overall efficacy, substantial variation exists.

First, intervention duration was a key moderator. Subgroup analysis showed that short-term interventions (≤4 weeks) demonstrated significant improvements in depression and anxiety (SMD = −0.66, p = .0002), potentially suitable for crisis or acute-phase adjunctive therapy. Intermediate-term interventions (＞4 to ≤6 weeks) showed no significant effect (SMD = 0.19, p = .18). This may reflect a transitional phase in which the novelty of VR has worn off, but core mindfulness skills – such as attentional control and decentering – have not yet been fully internalized (Garland et al., 2017). In contrast, longer interventions may allow sufficient time for skill consolidation and habit formation, enabling participants to integrate mindfulness techniques into daily life and thereby restore therapeutic benefits (Querstret et al., 2020).

Secondly, regional cultural context represents another significant source of heterogeneity. Eastern studies showed higher effect sizes for depression alleviation (SMD = −1.07, p < .00001), while Western studies did not reach significance (SMD = 0.12, p = .33). This difference may arise from culturally adapted settings (e.g. bamboo forests) in Eastern research, which may yield higher presence and efficacy (Shinohara et al., 2021). Taken together, these findings suggest that cultural familiarity with mindfulness practices, along with differences in participant characteristics, may contribute to the larger intervention effects observed in Eastern countries.

Notably, VR immersive level showed no significant moderating effect (p = .83). The key may lie in matching device features, content, and user tolerance – immersive devices increase presence but carry a higher risk of motion sickness in approximately 15% of elderly users, while semi-immersive devices showed no significant difference in outcomes with a lower risk of motion sickness (Seabrook et al., 2020).

Additionally, single-session length may influence outcomes. Sessions ≤30 min were associated with significant effects, while ≥30 min were not. This suggests shorter durations may sufficiently activate immersion and regulation mechanisms, whereas longer sessions could lead to fatigue or motion sickness. Prior VR research supports this association between shorter sessions and better completion/emotional outcomes (Seabrook et al., 2020). Single-session dosage thus appears as another critical heterogeneity source.

5.4. Clinical significance and practical recommendations

Based on the aforementioned research evidence, the clinical application of VR-based mindfulness therapy may involve establishing a systematic pathway that prioritizes two key dimensions: population stratification and intervention protocols. Regarding population stratification, it is suggested to establish differentiated application positioning based on clinical characteristics: For patients with moderate-to-severe depression or anxiety, VR-based mindfulness may serve as an adjunct to medication or conventional psychotherapy, using short-term (2-4 week) intervention to alleviate acute symptoms (Lee et al., 2024). For individuals with mild-to-moderate or subclinical depression and anxiety, it could be considered as a first-line non-pharmacological intervention, utilizing a standardized 8-week programme to achieve stable symptom improvement. For those with mobility limitations, community-based populations (e.g. individuals receiving care in community settings rather than specialized clinics), or younger groups with low acceptance of traditional mindfulness, the high accessibility and immersive experience of VR technology may have potential to help expand mental health service coverage (Yang et al., 2025). Evidence from previous mindfulness research suggests that repeated short sessions may facilitate emotional regulation and skill acquisition (Howarth et al., 2019). In terms of intervention design, personalized configurations are recommended based on patients’ clinical needs and intervention stages to balance efficacy and adherence (Xie et al., 2025).

In summary, this study supports the potential efficacy of VR-based mindfulness therapy in alleviating depressive and anxiety symptoms (Milasi et al., 2025), identifies intervention duration and regional/cultural factors as key sources of heterogeneity, and offers preliminary direction for clinical optimization. Furthermore, depression and anxiety often co-occur with or arise as sequelae of traumatic stress exposure, with trauma history as a well-established risk factor for mood and anxiety disorders. From a trauma-informed care perspective, VR-based mindfulness may offer potential benefits through its controlled, customizable immersion, enabling safe emotional regulation, gradual exposure to distressing stimuli, and decentering from trauma-related rumination while minimizing risks of hyperarousal or re-traumatization (Javanbakht et al., 2024). Clinicians may adapt protocols with trauma-sensitive features, such as neutral starting scenes, user-controlled pacing, exit options, and therapist-guided debriefing.

6. Limitations and outlook

6.1. Limitations

This study has several limitations. First, the number of randomized controlled trials was relatively limited, and some studies had small sample sizes, which may affect the stability of the results. Additionally, due to the characteristics of VR intervention technology, most studies found it difficult to implement blinding for participants and intervention providers, potentially increasing the risk of performance and detection bias. Second, a proportion of the included studies were conducted in non-clinical populations and relied primarily on self-report measures to assess depressive and anxiety symptoms. These factors may limit the generalizability of the findings to clinically diagnosed populations and to outcomes measured through clinician-administered instruments. Therefore, caution is warranted when generalizing the present findings to clinically diagnosed populations or routine clinical treatment settings. Third, although a systematic search of grey literature was conducted, no additional studies meeting the inclusion criteria were identified. Consequently, the possibility of publication bias cannot be fully excluded, as unpublished or negative trials may exist that could influence the pooled effect sizes. Fourth, the financial cost of VR hardware and related accessibility barriers in low-resource settings remains important practical limitations that may hinder widespread clinical implementation. Finally, although subgroup analyses were conducted to interpret sources of heterogeneity, residual heterogeneity remains high. This suggests the presence of other unaddressed factors, such as VR scenario design, interaction complexity, therapist support intensity, and individual differences.

6.2. Future research directions

Future research should aim to recruit larger and more diverse samples, preferably through multicenter collaborations, to enhance statistical power and robustness. Procedures for assessor blinding and outcome evaluation should be more clearly defined, and extended follow-up periods (e.g. 6–12 months) should be incorporated to assess sustained effects. Intervention protocols also require further optimization to enhance adaptability and sustainability; for example, content could be tailored to baseline characteristics, such as gamified tasks for younger users or simplified, low-stimulus environments for older populations. Moreover, studies should expand population and scenario coverage to better define applicability boundaries – that is, the specific populations, clinical contexts, and conditions under which VR-based mindfulness interventions demonstrate optimal efficacy and safety. Future studies should prioritize populations that were either underrepresented or showed high heterogeneity in the current meta-analysis, including elderly patients, individuals with comorbid physical illnesses, pediatric/adolescent groups, and those with PTSD or trauma-related conditions. These groups warrant further investigation, with particular attention to practical challenges such as device operability, motion sickness risk, and scenario adaptability.

7. Conclusion

This meta-analysis synthesizes evidence from current randomized controlled trials, indicating that VR-based mindfulness therapy is associated with significant improvements in both depression and anxiety symptoms. Subgroup analyses further identified intervention duration and regional cultural context as key moderating factors influencing efficacy, providing important evidence for the precision application of VR-based mindfulness therapy. Despite limitations such as high heterogeneity among included studies and limited methodological quality in some research, and potential generalizability issues related to non-clinical samples and self-report measures, VR-based mindfulness therapy suggests clear clinical value and potential for broader adoption. From a trauma-informed perspective, it may offer particular promise in addressing comorbid trauma-related symptoms. Future research should focus on optimizing intervention designs, deepening investigations into its mechanisms of action, and conducting more rigorous trials with extended follow-up to validate its sustained effects. This will advance the broader and standardized application of VR-based mindfulness therapy within multi-tiered mental health service systems.

Funding Statement

This work was supported by the Shanghai Education Scientific Funding: This work was supported by the Shanghai Education Scientific Research Project (Research on Digital Media Narrative Design Teaching Based on Data Analysis of Virtual Reality Psychological Therapy, C2026151), and by the Shanghai Dianji University Graduate Student Supervisor Team Development Program (Virtual Reality Immersive Experience Therapy Integration and Innovation Design Team).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.