Scoping review of VR exposure therapy for social anxiety and how it could be adapted for stuttering

Background

Approximately 46% of people who stutter (PWS) meet diagnostic criteria for social anxiety disorder, compared with around 4% of fluent speakers. For PWS, social anxiety is typically embedded in speech and communication and revolves around the expectation that others will react negatively to one’s stutter. Cognitive behavioral therapy has been shown to reduce social anxiety in PWS, but at the time of this review no published trial had specifically evaluated virtual reality exposure therapy (VRET) for social anxiety in stuttering. Chard and van Zalk set out to synthesize what could be learned from VRET trials for social anxiety in the general adult population, and to propose how the most relevant design choices could be adapted for PWS.

What the researchers did

The authors conducted a scoping review following the PRISMA-ScR guidelines. Searches were run on 22 September 2021 across Web of Science, Scopus, and PsycINFO/PsycARTICLES using three search strings combining virtual reality, social anxiety, stuttering, exposure, and therapy/treatment terms. After deduplication and screening (751 records screened, 46 full-text articles assessed), 12 studies met the inclusion criteria: peer-reviewed, English-language, randomized or quasi-randomized trials of VRET targeting social anxiety in adults, with a non-VRET comparison condition and a validated outcome measure. The included studies were Anderson et al. (2013, 2017), Bouchard et al. (2017), Harris et al. (2002), Kampmann et al. (2016), Klinger et al. (2005), Lindner et al. (2019), Reeves et al. (2021), Robillard et al. (2010), Safir et al. (2012), Wallach et al. (2009), and Zainal et al. (2021).

For each study the authors extracted 17 design and methodology variables including: sample size and randomization, social anxiety measures, target (general social anxiety vs. public-speaking anxiety), exposure framework (emotional processing vs. inhibitory learning vs. unspecified), session count and follow-up, delivery mode (remote vs. in-person), therapist role (human vs. virtual vs. none), environment type (computer-generated vs. 360° video), use of facial expressions, head-mounted display, and whether VRET was integrated into a wider CBT protocol. The narrative synthesis then discussed how each design distinction might be adapted for stuttering, with no quantitative comparison of effect sizes across studies.

The authors note that two prior exposure-therapy studies have been conducted with PWS (Scheurich et al., 2019, using in vivo exposure; Walkom, 2016, using VRET, cited as reference 38) but that methodological limitations - including no validated social anxiety outcome in Walkom’s study and a “chill session” retreat option that may have reinforced avoidance - precluded their inclusion in the scoping review’s primary corpus.

What they found

The 12 reviewed studies were conducted across seven countries (USA 4, Canada 2, Israel 2, Netherlands 1, France 1, Sweden 1, UK 1) and published between 2002 and 2021. Seven targeted public-speaking anxiety only; five targeted general social anxiety using multi-domain scenarios. Two studies explicitly used the inhibitory learning approach (Bouchard et al., 2017; Lindner et al., 2019), two used emotional processing (Kampmann et al., 2016; Zainal et al., 2021), and the others did not specify. Three studies used 360° video environments; eight used computer-generated environments (the twelfth used a computer-monitor display without an HMD). All studies but one (Reeves et al., 2021) used a human therapist; one study (Zainal et al., 2021) used a fully automated voice-over virtual therapist; the rest used clinical psychologists or psychology trainees.

VRET was associated with reductions in social anxiety relative to waitlist across most of the reviewed studies, with at least one notable exception (Harris et al., 2002, where post-treatment differences between VRET and waitlist were small). Where VRET was compared to in vivo exposure or in-vivo-CBT, outcomes were broadly comparable; some studies found long-term advantages for non-VR CBT. Findings on fear of negative evaluation - a core component of social anxiety - were mixed: some VRET protocols reduced it, others did not, with the available evidence suggesting that the inclusion of visible facial expressions and the type of social interaction (one-to-one vs. audience) may matter more than the specific exposure framework used.

For people who stutter, the authors propose - based on this synthesis - several design adaptations worth testing empirically. Multi-domain scenarios are likely to fit PWS better than public-speaking-only protocols, given that PWS report anxiety across many speaking situations including telephone calls (which the authors flag as potentially a distinguishable sub-type worth its own module). Inhibitory learning frameworks are proposed as a way to integrate stuttering-specific safety behaviors and expectations into exposure exercises and to address continued negative reactions post-treatment - a particular concern for PWS, whose speech can be a source of repeated real-world negative evaluation. Automated VRET with a virtual therapist - represented in the review by a single study (Zainal et al., 2021) - is proposed as a way of lowering access barriers and potentially integrating with speech therapy, although no such combined protocol has been evaluated.

Why this matters

This is the first scoping review to focus on adapting VRET design specifically for people who stutter. Its findings are useful as a design playbook for any future stuttering-specific VRET protocol, and as a reference for clinicians evaluating commercial VR therapy products against the design choices the literature has examined. The review does not, however, establish that VRET is effective for social anxiety in PWS - no such trial existed at the time of the review, and the subsequent pilot RCT from the same authors (Chard et al., 2023) returned a null result on primary outcomes.

Limitations

The authors explicitly flag the following limitations:

• No quantitative analysis. A scoping review approach was chosen “based on the relatively small literature on VRET protocols for social anxiety and PWS.” The authors state they “cannot directly compare the effectiveness of different treatments and the techniques they use.” Effect sizes across studies were not pooled.
• Inconsistent reporting in primary studies. Several included studies did not specify exposure framework, facial-expression use, or HMD model. This may have caused details to be missed in synthesis. The authors flag the need for standardized reporting in VRET research.
• Scope limited to adults. Children and adolescents were excluded because of developmental differences in social anxiety trajectories and the small number of pediatric VRET protocols.
• Emerging techniques excluded. Cloud-based VRET and fMRI-based VRET studies were excluded as outside scope.
• Design hypotheses, not validated protocols. All recommendations for adapting VRET to PWS are author proposals, not empirically tested protocols. The authors are explicit that PWS-specific empirical research is required before any of these design choices can be recommended clinically.