INTRODUCTION
Social anxiety disorder (social phobia) is characterized by an intense fear of being scrutinized or observed by others in various social situations, such as engaging in con-versation, eating or drinking in public, delivering speeches, or performing specific tasks in the presence of others [1]. It is characterized by excessive fear or anxiety of embarrass-ment or humiliation in social situations [2]. Individuals with social anxiety disorder are highly sensitive to negative evaluation and social rejection, and a significant pro-portion also experience symptoms of depression [1-3]. Ac-cording to the World Health Organization, social phobia is a major mental health disorder [3], affecting between 7∼ 17% of the global population [4,5]. Despite its prevalence, it is under-researched and known to impair social interactions and daily functioning significantly [5].
Pharmacotherapy and cognitive behavioral therapy (CBT) are the most widely used interventions for social phobia. Commonly used medications include selective serotonin reuptake inhibitors, monoamine oxidase inhibitors, and benzodiazepines [6]. These treatments are effective in reducing anxiety and physiological arousal [7]. However, long-term use of medications can lead to side effects and temporary relief, and CBT is more effective than pharmacotherapy in some cases [7,8]. Some patients re-lieve anxiety through drug abuse, which may lead to drug dependence [9]. Nonpharmacological treatments, such as CBT, are currently preferred as the first-line treatment for social phobia [10]. CBT involves exposing patients to anxiety-provoking stimuli to help them realize that feared outcomes are unlikely to occur [11,12].
With the global shortage of psychiatrists and mental health professionals, however, access to face-to-face cognitive-behavioral therapy (CBT) can be limited.[13]. Owing to the nature of the disorder, there is a lack of practice opportunities, making it difficult to reproduce therapy effectively [12]. Additionally, there are limitations to traditional CBT, which can be challenging to provide at the time patients desire [14].
Furthermore, during the recent COVID-19 pandemic, face-to-face interventions for patients with mental disorders were limited because of social isolation. However, the development of non-face-to-face interventions and medical technology has led to the active implementation of non-face-to-face CBT [15]. Non-face-to-face CBT is particularly helpful for patients with social phobia by alleviating concerns about going to therapy or what others might think [16].
Treating mental disorders using virtual reality (VR) offers a unique opportunity to expand exposure therapy within CBT [17]. Moreover, with technological advance-ments, VR image quality has improved, and the cost has become much lower than traditional psychological treatments [17]. While previous meta-analyses have shown that VRET (VR exposure therapy) has a larger effect size than control groups, can be more attractive to patients, and may be easier to disseminate [18], recent meta-analy-ses indicate that both VRET and traditional face-to-face exposure therapy are effective in treating specific phobias, with benefits observed both immediately after treatment, and during follow-up assessments.[19].
However, the effectiveness of virtual exposure therapy and the specific domains in which it is effective for patients with social anxiety disorder have not been statistically confirmed.
By synthesizing recent studies on patients with social phobia and presenting effect sizes, this study aimed to provide foundational data for offering nonpharmacological interventions to patients with mental disorders in a socially conducive environment where non-face-to-face in-tervention is becoming more active.
Therefore, the purpose of this study is to conduct a meta-analysis to determine the effect size of anxiety and psychological outcomes following VR-based exposure therapy interventions for patients with social anxiety disorder, through a systematic review. In particular, this study aimed to assess the effects of negative evaluation, which is one of the key variables related to anxiety and social anxiety in social phobia, as well as its impact on depression.
METHODS
As a systematic review and meta-analysis on the effects of VR intervention programs for social anxiety disorders, this study adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement [20]. The research question was framed using the PICO format:
P (Patient/Problem/Population): Patients with social phobia
I (Intervention): VR technology Intervention
C (Comparison): No VR technology Intervention
O (Outcomes): Variables related to mental health
1. Search Strategy
In January 2024, a comprehensive search was conducted across core databases recommended by Cochrane (Pub-Med, Embase, Cochrane Library, CINHAL, PsyINFO). The search terms used were related to social phobia and CBT, with the following query: [{(social AND phobia) OR (social AND anxiety) OR (social AND disorder)} AND ("cognitive-behavioral therapy" OR CBT)]. Only articles published from 2021 onwards, when non face to face interventions became widely recommended due to the COVID-19 pandemic, were included. The search was restricted to studies published in English, and meta-analyses targeting patients with social phobia were manually screened. Studies that did not provide sufficient data for analysis were excluded after contacting the authors via email. Two researchers (ML and SP) independently conducted the search and verification, and the identified studies were managed using EndNote 20 software.
2. Selection Process
Six steps were followed for the final selection of literature. First, two researchers (ML and SP) imported 830 studies from electronic databases and five additional studies from manual searches into EndNote 20. Second, 78 du-plicate studies were identified and removed using the "Find Duplicates" feature in EndNote, which automati-cally generated a list of duplicates for manual verification and deletion by the researchers. Third, ML and SP independently reviewed the titles and abstracts of the re-maining 757 studies. Fourth, any discrepancies between the independent reviews were resolved through discussion to select full-text articles for further assessment. Fifth, after reviewing the titles and abstracts, 745 studies were excluded, leaving 12 full-text articles for full evaluation based on the inclusion criteria. These inclusion criteria were assessed based on the study's PICO framework (Patient/Problem/Population: patients with social phobia, Intervention: VR technology intervention, Comparison: no VR technology intervention, Outcomes: variables related to mental health). Six studies were excluded during the full-text review process, and six studies were ulti-mately selected for analysis. Finally, ML and SP verified the selected studies, and extracted the relevant outcome data.
To ensure quality, only articles published in peer-re-viewed journals were included. The detailed flow of the selection process is shown in Figure 1.
3. Data Extraction
Information and data from the six studies that met the inclusion criteria is presented in Table 1, including the author, publication year, country, study design, participants (characteristics, primary diagnosis, mean age, education level, etc.), characteristics of the CBT intervention (content, duration, process design, provider, virtual method), outcomes (main measurement variables and tools), and results (statistical values of the effectiveness of CBT). Overall, Table 1 provides a systematic review of the in-formation on CBT research.
4. Statistical Analysis
Statistical analysis was conducted using SMD (Standar-dized Mean Difference) with a 95% confidence interval (CI) to calculate the overall effect sizes of all meta-analyses using Review Manager (version 5.3) [21]. SMD=0.2 indicates a small effect size, g=0.5 and g=0.8 indicate medium and large effect sizes, respectively. The advantage of using SMD is that it accounts for potential baseline differences and does not rely solely on posttreatment differences between groups [22]. Statistical heterogeneity between the studies was systematically evaluated using the I2 and Q statistics. When I2=0∼50%, there was no heterogeneity; when I2=50∼75%, there was moderate heterogeneity; and when I2=75∼100%, there was high heterogeneity. A p-val-ue of <.10 indicates significant heterogeneity [23].
This meta-analysis divided the results into primary and secondary outcomes. The primary outcome involved analyzing the effect sizes of studies reporting on anxiety disorders. Anxiety was categorized into subgroups for further analysis, including general anxiety, social anxiety, and relationship anxiety. The secondary outcome focused on examining the effects on psychological discomfort, spe-cifically analyzing depression and negative emotional evaluation. This approach allowed for an evaluation of the differences across specific types of anxiety disorders as the primary outcome, and the impact on psychological factors and emotional states as the secondary outcome.
5. Quality Assessment
The quality of the included studies was assessed using the Cochrane Collaboration's risk-of-bias tool (RoB) for randomized controlled trials (RCTs) [24]. The ROB tool was employed to assess the risk of bias across five domains: (1) bias arising from the randomization process, (2) deviations from intended interventions, (3) missing outcome data, (4) measurement of the outcome, and (5) selection of the reported result. Each domain was rated as having either "low," "some concerns," or "high" risk of bias. This assessment was conducted by two independent re-viewers, and in cases of disagreement, discussions were held to resolve the differences, adjusting for potential bias in the estimation of effect sizes.
RESULTS
1. Characteristics of Included Studies
Of the six studies included in this meta-analysis, two were from Korea, two from the United Kingdom, one from the United States, and one from Switzerland. A total of 286 participants were included: 144 in the experimental group and 142 in the control group. These participants were patients diagnosed with anxiety disorders such as social anxiety (n=4) or performance anxiety (n=2). The six VR interventions included in this review expose users to various scenarios (classroom, coffee shop, phone situations, public speaking, business life, etc.), offering interventions where users can react, converse, and deliver speeches in these situations. The study duration varied from 2 weeks to 10 weeks (Appendix 1). The studies were conducted between 2021 and 2023, which was a challenging period for patients with mental disorders, owing to the social isolation imposed during the COVID-19 pandemic.
2. Methodological Quality and Risk of Bias
Using the Cochrane Collaboration's RoB tool [25], the individual studies were rated as having a low, high, or unclear risk of bias. RoB assessment was performed using Review Manager (version 5.4). Overall, the six included studies had a low risk of bias.
Specifically, four of the six studies had a low risk of bias in all domains, whereas two studies had an unclear risk in some areas owing to the lack of detailed methods for the randomized control trials (RCTs). In particular, the risk of bias in performing the interventions was commonly rated as unclear. In addition, the completeness of the outcome data was not clearly reported in the domain of incomplete outcome reporting.
The report on the quality assessment is presented in Figure 2.
3. Heterogeneity and Consistency Test
I2 and Q statistics (based on the full design-by-treatment interaction random-effects model) were used to evaluate the included studies' heterogeneity. The overall heterogeneity of the six included studies was high, with an I2 value of 68%.
4. Effect Size of Primary Outcome
The results of the analysis of the effect size on anxiety, the primary outcome of the VR intervention study conducted among patients with anxiety disorders, are shown in Figure 3. The analysis of the effect size of the VR intervention study showed an SMD of -0.51 (95% CI=-0.85 to -0.17), indicating a medium, statistically significant effect. However, the overall heterogeneity of the six included studies was high, with an I2 value of 68%, leading to the need for a subgroup analysis.
1) Subgroup analysis
Based on the measurement tools, the outcome variables for anxiety were divided into the generalized anxiety, social anxiety, and relational anxiety subgroups. The analysis of three VR interventions showed that the effect size for reducing generalized anxiety was SMD=0.16 (95% CI=-0.84 to 1.15), which was not statistically significant. Heterogeneity was also high (I2=81%).
However, for social anxiety, the effect size was SMD= -0.56 (95% CI=-0.87 to -0.25), which was statistically significant. The five studies included in the analysis showed effects in the same direction (I2=0%). For relational anxiety, the effect size was SMD=-0.83 (95% CI=-1.49, to -0.16), based on the results of the five studies that showed the largest effect size (Figure 3).
5. Effect Size of Secondary Outcome
Secondary outcomes included depression and negative evaluations. Depression had a moderate effect size; however, it was not statistically significant. Negative evaluations had a statistically significant large effect size.
1) Depression
The analysis of the effect size of two studies reporting on depression showed an SMD of -0.40 (95% CI=-0.84 to 0.04), indicating a medium effect size with no heterogeneity. However, these results were not statistically significant (Figure 4).
2) Negative evaluation
Three studies were included to analyze the effect size on negative evaluation. The analysis showed an SMD of −0.75 (95% CI=-1.20 to -0.31), indicating a large, statistically significant effect size. The heterogeneity was also low (I2=0%) (Figure 5).
6. Publication Bias
To evaluate publication bias, a funnel plot was used for visualization. As the funnel shape did not show excessive asymmetry, the risk of publication bias was considered to be low (Figure 6).
DISCUSSION
The results of this study's systematic review and meta-analysis are discussed here. These studies were particularly significant for patients experiencing mental health disorders due to social isolation during the COVID-19 pandemic. The studies were conducted in Korea, the UK, the US, and Switzerland, with a total of 286 participants. The participants included in these studies were predom-inantly patients diagnosed with social anxiety and performance anxiety based on DSM criteria. They underwent exposure therapy in various VR scenarios (classroom, cafe, phone call situation, presentation, business scenario, etc.), where they had opportunities to respond, interact, and perform presentations. This approach reflects an exposure-based therapeutic method aimed at alleviating social anxiety symptoms through experiences similar to real-life social interactions.
The findings of several studies indicate that while anxiety can be reduced in specific VR scenarios, the impact on overall social anxiety symptoms remains limited [27,31]. This suggests the need for a more personalized approach, targeting the unique social situations that trigger anxiety for each individual, similar to the tailored interventions used for speech anxiety. Moreover, elements such as real-ism, immersion, and potential issues like motion sickness —identified as limitations in VR studies should be care-fully considered to enhance the effectiveness of VR-based exposure therapy [28,32].
A systematic review of the risk of bias showed that using the Cochrane Collaboration's RoB (Risk of Bias) tool, the bias risk of individual studies was assessed. Four of the six studies showed low bias risk across all domains, while two indicated uncertain bias risk. Particularly, there was a lack of detailed methods in randomized controlled trials (RCTs), which led to uncertain bias risk in some domains. Additionally, there were many instances of uncertain bias risk in the process of implementing the intervention, and incomplete outcome reporting was observed, with a lack of clarity regarding the completeness of the data.
The results of the meta-analysis of six VR interventions (participants, N=144) were confirmed. The primary outcome, anxiety, showed a medium effect size for the VR interventions compared with the control group. In the subgroup analysis, relational anxiety was found to have the largest effect. Social anxiety showed a statistically significant effect size, whereas generalized anxiety did not. The secondary outcome, depression, showed a medium effect size but was not statistically significant, whereas negative evaluation showed a large, statistically significant effect size.
These results are similar to those of previous meta-anal-yses of VR interventions conducted among individuals with mental disorders. This demonstrates that interventions such as exposure therapy using VR are effective for anxiety-related disorders [19].
A meta-analysis conducted by Opriș et al. (2012) demonstrated results similar to those of the present study, showing that virtual reality exposure therapy (VRET) is effective in reducing social anxiety symptoms [26]. The study reported a large effect size (d=1.01, 95% CI [0.69, 1.33]), indicating a significant effect of VRET compared to the control group. Similarly, a meta-analysis by Kampmann et al. [12] (2016) reported comparable results, with an effect size g=0.82, 95% CI [0.13, 1.51], further confirming the efficacy of VRET in alleviating social anxiety symptoms.
However, both Opriș and Kampmann's studies [12,26] found no significant difference between VRET and traditional in vivo exposure therapy. This suggests that while VRET is effective, its benefits are comparable to those of traditional exposure therapy. These findings align with the results of the present study, where social anxiety showed a statistically significant effect size, but generalized anxiety did not, indicating that VRET may be more effective for specific anxiety disorders, particularly social anxiety.
For generalized anxiety, results similar to those from single intervention studies were observed [27,28]. This may be due to the fact that general anxiety encompasses a broader range of concerns, and short-term VR interventions targeting social situations may not have a clearly de-fined impact on such diverse anxiety factors.
Since the primary symptom of social anxiety disorder is the fear of exposure to social situations, VR-based exposure therapy has shown effectiveness in reducing social anxiety [29-31] as well as anxiety in relational contexts [30,31]. Exposure therapy, a fundamental component of CBT [10-12], enables individuals to experience realistic exposure to feared social scenarios through VR simulations [30,32], which can lead to a reduction in anxiety. Regarding depression, previous meta-analyses of social anxiety disorder also found that VR interventions did not show statistically significant effect sizes, consistent with the results of this study. However, these meta-analyses indicated that VR interventions are effective in treating depression in individuals with social anxiety disorder using internet-based CBT [12]. Additionally, meta-analyses have reported that such VR interventions typically include stress-relief programs, relaxation exercises, physical activity, and CBT programs [33]. Given the nature of depression, nonpharmacological interventions often involve cognitive therapy, interpersonal therapy, emotional expression, exercise, self-esteem enhancement, and opportunities for self-realization. As these interventions are often based on social activities, the VR interventions in this study, primarily involving exposure to social situations, may not have been as effective.
Negative evaluation was found to have a statistically significant effect in this study. Individuals with Social Anxiety Disorder (SAD) experience significant anxiety in social situations where they may be closely observed by others; they are also highly sensitive to negative evaluations and social rejection [1,34]. Individuals with inter-nalized shame are likely to experience anxiety in interpersonal relationships based on their perception of negative evaluations from others [35]. Therefore, VR interventions that directly mediate situations are thought to be particularly effective in addressing anxiety and negative evaluations in interpersonal relationships rather than general anxiety. In summary, while VRET appears to be a promising treatment for social anxiety and negative evaluation, its efficacy for generalized anxiety and depression remains less clear. Future studies could benefit from in-corporating a wider range of therapeutic techniques to ad-dress the diverse symptoms associated with these conditions.
1. Limitations
Sample Size: The relatively small number of participants in the studies included in this meta-analysis limits the generalizability of the findings. A small sample size can affect the statistical power and reliability of the results, making it difficult to extend the conclusions to a broader population. Cases in which VR was combined with other traditional interventions or where both the experimental and control groups received VR interventions were excluded to clarify the effects of the VR interventions. This reduced the number of studies included but allowed for a clearer understanding of the effects of VR interventions on anxiety.
Heterogeneity: This study included anxiety about public speaking and stuttering, which are representative symptoms of social anxiety disorder, to confirm anxiety about VR interventions. However, a direct confirmation of public speaking anxiety has not yet been conducted. Future research and analyses specific to individuals with public speaking anxiety are necessary. Each study identified social anxiety disorder through surveys and other methods; however, not all the participants were diagnosed with social anxiety disorder. Although this allowed for a larger number of participants, only diagnosed participants should be analyzed for clinical applications.
Lack of Long-term Follow-up: One major limitation of this study is the lack of long-term follow-up data. While the short-term effects of VRET were demonstrated, it remains unclear whether these effects are sustained over time. Future research should include extended follow-up periods to better assess the long-term efficacy of VR interventions.
2. Clinical Significance
A Practical Alternative for Social Anxiety: The findings of this study suggest that VR-based interventions could serve as a practical and effective alternative for individuals with social anxiety disorder, particularly those un-able to access face-to-face therapy due to mobility constraints, geographical barriers, or social stigma. This is es-pecially relevant during periods of restricted access to in-person therapy, such as the COVID-19 pandemic.
Targeting Sensitivity to Negative Evaluation: Individuals with social anxiety disorder are often highly sensitive to negative evaluation in social contexts. VR interventions, which simulate real-life social interactions in a controlled environment, appear particularly effective in reducing this sensitivity, making them a promising treatment option for addressing interpersonal anxieties.
Non-Pharmacological Treatment Option: VR-based interventions represent a viable non-pharmacological treatment for social anxiety disorders. By focusing on exposure and desensitization without the long-term side effects associated with pharmacotherapy, VR interventions provide an alternative therapeutic option that may be preferred by patients seeking non-medication-based treatments.
CONCLUSION
This study conducted a systematic review and meta-analysis of research involving VR interventions in participants with social anxiety disorder. The analysis of six RCT studies confirmed that while VR interventions had no significant effect on general anxiety, they were effective in reducing relational and social anxiety. In the case of depression, a medium effect size was observed, though it was not statistically significant. However, negative evaluation showed a statistically significant effect. This study has several strengths, including that the reviewed studies were RCTs, providing the highest level of evidence for the effectiveness of VR interventions. This means that VR interventions can be used for individuals with social anxiety who cannot receive face-to-face therapy because of mobility, geographical, or situational constraints (such as in-fectious diseases) or who do not wish to receive face-to-face therapy because of stigma.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation 
Print 
