Effectiveness of virtual reality therapy in the treatment of anxiety disorders in adolescents and adults: a systematic review and meta-analysis of randomized controlled trials

The study addresses whether virtual reality therapy (VRT), including exposure, distraction, and mindfulness-based applications, effectively reduces anxiety symptoms in adolescents and adults diagnosed with anxiety disorders. Anxiety disorders are prevalent and often comorbid, notably generalized anxiety disorder and panic disorder, causing significant functional impairment and reduced quality of life. While pharmacotherapy and cognitive behavioral therapy (CBT) are effective, access and uptake of CBT remain limited, motivating exploration of alternative or adjunctive interventions. VR technology provides immersive, interactive environments that can divert attention, facilitate exposure to feared stimuli under controlled conditions, and augment psychotherapeutic techniques. Despite increasing clinical use, rigorous evidence on VR’s effectiveness for anxiety disorders is limited and heterogeneous. This systematic review and meta-analysis aims to synthesize randomized controlled trials evaluating VRT’s impact on anxiety outcomes in adolescents and adults.

The authors outline the spectrum of anxiety disorders and summarize established treatments (pharmacotherapy, CBT, mindfulness-based interventions, exercise, music therapy, rTMS), noting comparable efficacy between pharmacotherapy and psychotherapy. VR has evolved with improved immersion, interactivity, and accessibility, and has been applied to pain management, PTSD, social anxiety, phobias, and perioperative settings. Prior meta-analyses have supported VR exposure therapy for anxiety-related disorders but evidence remains mixed, particularly for perioperative anxiety, potentially due to variability in scales (e.g., STAI, YPAS, APAIS, STOA) and intervention designs. The paper situates VRT within this context, highlighting gaps in high-quality, focused RCTs on adolescents and adults with defined anxiety disorders and clinical procedures, motivating the present synthesis.

Design and registration: Systematic review and meta-analysis conducted per PRISMA 2020 guidelines and registered with PROSPERO (CRD42024574772). Search strategy: Comprehensive searches in PubMed, Web of Science, Embase, CINAHL, Scopus, and Cochrane databases from inception to September 2024; reference lists of included studies were also screened. Search terms combined subject headings and keywords related to anxiety (e.g., anxiety disorder), populations (adolescent, adult), and interventions (VR, virtual reality therapy, virtual reality exposure). Eligibility criteria: Randomized controlled trials including adolescents or adults diagnosed with anxiety disorders; no restrictions on gender, illness course, or ethnicity. Outcomes included clinically validated anxiety scales at post-treatment and follow-up (e.g., HAMA, SAS, BAI, STAI variants). Experimental groups received VR-based interventions (e.g., VRT, VRET, VR games, VR mindfulness); controls received non-VR interventions (e.g., CBT, mindfulness therapy, treatment as usual, health education, placebo, waitlist). English-language articles only. Study selection and data extraction: Literature imported to EndNote 20; duplicates removed; two researchers independently screened titles/abstracts and full texts against criteria; data extraction performed independently and cross-checked (study characteristics, participants, interventions, controls, outcome data). A third researcher resolved disagreements; authors were contacted for missing information where feasible. Quality appraisal: Risk of bias assessed using the Cochrane Risk of Bias tool across selection, performance, detection, attrition, reporting, and other biases. Studies judged as low, unclear, or high risk; overall grades A (low risk), B (partial low risk), C (high risk). Double blinding was often infeasible due to VR equipment, contributing to performance and detection bias risks. Statistical analysis: Review Manager 5.4 used for meta-analyses; continuous outcomes pooled as standardized mean differences (SMD) with 95% confidence intervals due to varied scales. Heterogeneity assessed with Cochrane Q and I². Fixed effects applied when P > 0.1 and I² ≤ 50%; random effects applied otherwise. Sources of heterogeneity explored via descriptive and subgroup analyses (age, sex, anxiety type, scale, country, intervention type). Sensitivity analyses and publication bias: Stata 16 used for sensitivity analyses; funnel plots and Egger’s test assessed publication bias.

- Included evidence: 33 RCTs (published 2019–2024) from 18 countries, totaling 3182 participants; 27 two-arm and 6 three-arm studies. - Overall effect: VR therapy significantly reduced anxiety compared with non-VR controls [SMD = -0.95; 95% CI: -1.22 to -0.69; Z = 7.05; P < 0.00001], with substantial heterogeneity (I² = 91%; P < 0.00001). - Sensitivity analysis: 11 studies notably influenced outcomes. Removing two studies reduced heterogeneity (I² = 44%; P = 0.02) with a still significant pooled effect [SMD = -0.76; 95% CI: -0.86 to -0.67; Z = 15.59; P < 0.00001]. - Subgroup analyses (Table 2): • Sex: Male (I² 80%; SMD -0.81 [-1.06, -0.55]; Z = 6.17; P < 0.00001); Female (I² 91%; SMD -0.95 [-1.22, -0.69]; Z = 7.05; P < 0.00001). • Age: Adolescents (I² 96%; SMD -0.95 [-1.48, -0.42]; Z = 3.84; P = 0.0005); Middle age (I² 74%; SMD -0.79 [-1.09, -0.48]; Z = 5.07; P < 0.00001); Young old (I² 92%; SMD -1.23 [-1.92, -0.55]; Z = 3.54; P = 0.0004). • Scale: STAI/SATI (I² 87%; SMD -0.84 [-1.09, -0.60]; Z = 6.83; P < 0.00001); Other scales (I² 94%; SMD -1.34 [-2.11, -0.58]; Z = 3.44; P = 0.0006). • Country: Asia (I² 94%; SMD -1.08 [-1.46, -0.69]; Z = 5.54; P < 0.00001); Europe (I² 66%; SMD -0.56 [-0.82, -0.30]; Z = 4.19; P < 0.0001); North America (I² 0%; fixed-model; SMD -0.89 [-1.29, -0.49]; Z = 4.33; P < 0.0001). • Diagnostic categories: Perioperative anxiety (I² 93%; SMD -1.05 [-1.44, -0.66]; Z = 5.26; P < 0.00001); Perinatal anxiety (I² 86%; SMD -0.96 [-1.46, -0.47]; Z = 3.83; P = 0.0001); Examination-related anxiety (I² 72%; SMD -0.89 [-1.29, -0.49]; Z = 4.34; P < 0.0001); Other related anxiety (I² 60%; SMD -0.48 [-0.84, -0.12]; Z = 2.62; P = 0.009). • Control interventions: Active controls (I² 77%; SMD -0.50 [-0.98, -0.01]; Z = 2.00; P = 0.005); Inactive controls (I² 92%; SMD -1.02 [-1.31, -0.72]; Z = 6.75; P < 0.00001). • Experimental interventions: VR distraction therapy (I² 73%; SMD -0.42 [-0.86, 0.01]; Z = 1.90; P = 0.06, not statistically significant); VR exposure therapy (I² 92%; SMD -1.02 [-1.31, -0.72]; Z = 6.79; P < 0.00001). - Publication bias: Funnel plot showed some asymmetry; Egger’s test did not indicate significant publication bias (P = 0.731). - Risk of bias: Among 33 studies, 1 low risk, 16 unclear risk, 16 high risk; performance and detection biases were common due to infeasibility of double blinding with VR equipment.

The meta-analysis demonstrates that virtual reality-based interventions reduce anxiety symptoms in adolescents and adults across diverse clinical contexts, addressing the need for accessible adjuncts or alternatives to established therapies like CBT. Effects were robust overall, though heterogeneity was high, reflecting variability in VR modalities (distraction vs exposure), intervention dosages, devices, co-interventions (e.g., music, mindfulness), outcome scales, and clinical populations (perioperative, perinatal, examination-related, public speaking, social anxiety, bipolar-related). Subgroup results suggest stronger effects for VR exposure therapy compared to distraction, consistent with theory that controlled exposure in immersive environments facilitates habituation and desensitization to feared stimuli. Active control comparisons yielded smaller effects than inactive controls, indicating that part of VR’s benefit may be specific beyond general engagement or education. Geographic and demographic differences (e.g., age bands, sex, country) may reflect contextual factors, sample characteristics, or implementation differences. Conflicting findings in perioperative anxiety literature underscore the influence of outcome measurement variability (e.g., different STAI versions, perioperative-specific scales) and intervention timing/design. The discussion highlights VR’s therapeutic mechanisms (attention diversion, controlled exposure, mindfulness facilitation) alongside potential adverse effects (cybersickness, ocular strain, blue-light exposure, dissociation), emphasizing the importance of presence, safety, and user experience in design and clinical protocols. Overall, the findings support VRT as an effective, emerging modality for anxiety management while underscoring the need for standardized methods, rigorous designs, and integration of physiological measures to clarify mechanisms and optimize clinical utility.

Virtual reality therapy is a promising and effective intervention for reducing anxiety symptoms in adolescents and adults, though it does not uniformly surpass conventional treatments. Given the limited number and variable quality of included studies and substantial heterogeneity, further high-quality, large-sample RCTs across specific anxiety subtypes are warranted. Future work should standardize VR evaluation and treatment protocols, incorporate physiological measures, assess real-world clinical feasibility and long-term outcomes, explore the role of presence, and reduce reliance on therapist involvement (e.g., via apps or guided recordings) while mitigating adverse effects through improved device design and usage guidelines.

- High heterogeneity across studies (devices, content, dosage, timing, populations, outcome scales) complicates interpretation. - Risk of bias concerns: frequent performance and detection bias due to inability to double blind VR interventions; many studies with unclear or high risk of bias. - Publication bias cannot be ruled out (some funnel asymmetry), though Egger’s test was non-significant. - Variability and inconsistency in anxiety measurement tools (different STAI versions and perioperative-specific scales) may influence results. - Limited long-term follow-up and durability of effects, especially for distraction-based VR. - Potential adverse effects of VR (e.g., nausea, vertigo, ocular strain, dissociative symptoms) not uniformly assessed. - Some inconsistencies in reported search timeframe; overall reliance on studies from the last five years may limit generalizability. - Small samples and limited representativeness in certain subgroups/geographies.