Efficacy of Virtual Reality Exposure Therapy Scripts for Social Phobia

Social anxiety disorder (SAD) is a widespread mental health problem (Kessler et al., 2012), marked by persistent and intense fear and avoidance of interpersonal situations that may be evaluative (American Psychiatric Association, 2013). SAD is acknowledged to be a complex condition that can be chronic or unremitting (Vriends, Bolt, \& Kunz, 2014). Relatedly, persons with SAD tend to display compromised work or school functioning, relationship satisfaction, and quality of life (Teo, Lerrigo, \& Rogers, 2013), as well as increased comorbid physical and mental health problems (Jacobson \& Newman, 2017; Schmidt, Keough, Hunter, \& Funk, 2008). Therefore, designing and delivering efficacious treatments for SAD is important.

Novel state-of-the-art psychotherapies for SAD have been developed. These include cognitive bias modification approaches (e.g., video feedback, imagery rescripting), cognitive behavioral therapies (CBT; e.g., social skills training, behavioral experiments), internet CBT, and neuroenhancers (e.g., D-cycloserine, yohimbine) (Stangier, 2016). Exposure therapy is a well-established, gold-standard efficacious specific CBT component for SAD (Otte, 2011). Such exposure therapy is grounded in emotional processing theory (Foa \& Kozak, 1986), which proposes that fear memories can be interpreted as structures that store data concerning events, responses, and meaning propositions. By activating and altering fear structures, it is thought that exposure therapy can ease emotional processing through gradual and recurrent experiences of distress-evoking yet potentially rewarding and physically safe social interactions (Zalta \& Foa, 2012). Specifically, exposure therapy is supposed to activate the fear structure by involving the person with SAD in various anxiety-provoking social situations to adjust the fear structure via habituation (reduced fear after repeated encounters of phobic situations) and extinction (unlearning of acquired fear connections; Furlong, Richardson, \& McNally, 2016). On the other hand, it is plausible that even if within- or between-session habituation did not occur, extinction learning might be achieved via practicing greater fear tolerance in diverse anxiety-evoking social situations (cf. inhibitory learning theory; Craske et al., 2008). Relatedly, the cognitive model of SAD (Leigh \& Clark, 2018) posits that exposure therapy works by improving psychological flexibility and other (vs. self-) focused attention as well as eliminating safety behaviors and threat hypervigilance, during anxiety-inducing social events.

Exposure therapy can be carried out in vivo (directly engaging with fear-evoking situations) or in vitro or imaginarily (continually visualizing exposure scenarios). However, in vivo exposure therapy can be time-consuming, costly, and how events unfold during exposure can neither be predicted nor controlled. Also, in vitro exposure therapy can be difficult and ineffective for highly-avoidant persons who lack the ability to envision specific scenes in clear detail and to stay in the imagined anxiety-inducing scenarios long enough for the imaginal exposure therapy to be effective (Hembree \& Cahill, 2007; Vrielynck \& Philippot, 2009). Thus, virtual reality exposure therapy (VRE) has been designed as a means for persons with SAD to privately immerse in myriad computer-generated fear-provoking social situations that mimic the natural environment (Anderson, Rothbaum, \& Hodges, 2003). VRE offers latitude for the psychologist to construct exposure situations, receives more acceptance from patients, provides greater privacy, and has lower dropout rates than in vivo exposure therapy (Garcia-Palacios, Botella, Hoffman, \& Fabregat, 2007; Krijn, Emmelkamp, Olafsson, \& Biemond, 2004). Further, most clinicians have viewed VRE favorably (Lindner et al., 2019a).

To date, seven randomized controlled trials (RCTs) have examined the impact of therapist-directed VRE compared to waiting list (WL) for SAD. An early study showed that four therapist-facilitated 12-to-15-minute sessions of VRE (vs. WL) efficaciously reduced public speaking phobia for highly anxious college students from baseline to post-treatment (Harris, Kemmerling, \& North, 2002); despite this reduction, VRE was not superior to WL in lowering SAD symptoms. Another four-session VRE (vs. WL) that instructed participants to repeatedly read a children's book to a virtual audience observed reductions in public speaking fear across time (Lister, Piercey, \& Joordens, 2010); however, there were no changes on trait anxiety or social phobia symptoms. Relatedly, compared to WL, a therapist-guided, 12-session, VRE with various fear-inducing interpersonal situations led to decrements in SAD fear and avoidance severity, as well as public speaking phobia, among young adult females with public speaking anxiety (Wallach, Safir, \& Bar-Zvi, 2011). Further, some sustained treatment gains at 1-year follow-up were observed (Safir, Wallach, \& Bar-Zvi, 2012); however, high-anxious participants were not assessed for diagnoses of SAD in that study. In a similar nine-session, therapist-directed VRE (vs. WL) RCT for public speaking anxiety that additionally recruited patients with SAD and assigned between-session homework, large effect sizes for treatment efficacy were observed from baseline to 1-year follow-up (Anderson et al., 2013); nonetheless, findings may not extend beyond persons who had expressed public speaking fear as their main problem. Relatedly, VRE (vs. WL) was more effective in reducing avoidance and unhelpful beliefs as well as increasing speech duration in persons with SAD whose interpersonal fears extended to numerous contexts (Kampmann et al., 2016). Additionally, cognitive behavioral therapy (CBT) plus VRE (vs. WL) targeting multiple social fears was more effective at decreasing SAD severity (Bouchard et al., 2017); however, as it examined VRE blended with CBT, the efficacy of self-guided VRE as a stand-alone treatment could not be deduced.

Most of the VRE conducted so far has been therapist-led but attempts toward decreasing therapist participation have recently been underway. For example, one-session of minimally-therapist-assisted VRE (vs. in vivo exposure therapy) produced comparable reduction in behavioral avoidance frequency and fear severity in spider phobic individuals (Miloff et al., 2019). Likewise, completely self-guided, smartphone app-delivered VRE (vs. WL) resulted in large decreases in fear of heights for persons with acrophobia (Donker et al., 2019), with strong engagement across unique exposure scenarios that evoked varying degrees of fear (Donker, van Klaveren, Cornelisz, Kok, \& van Gelder, 2020). In addition, the large effect of a relatively inexpensive, one-session, self-directed, Google Cardboard headset-delivered VRE for public speaking phobia did not differ significantly from a therapist-guided treatment (Lindner et al., 2019b). Such efforts are important to test further as VR can enhance the delivery of self-help exposure therapy to persons with SAD by decreasing costs and increasing access when trained therapists are not available. For example, it could decrease costs if used as part of a stepped care model. All patients within a health management organization (HMO) who could benefit from VRE, could first be assigned to complete self-guided VRE within a clinic. The clinic might contain multiple VRE equipped rooms and overseen by one minimum wage paid staff person who serves as a technician troubleshooting any mechanical faults with the equipment. If this was not sufficient to lead to remission the patient would then be provided more intensive in-person CBT with a well-trained and qualified CBT practitioner (Boeldt, McMahon, McFaul, \& Greenleaf, 2019; Newman, Szkodny, Llera, \& Przeworski, 2011). Alternatively self-help VRE equipment could be rented and/or purchased and used at home as the price of VRE devices has been declining across time (Bun, Gorski, Grajewski, Wichniarek, \& Zawadzki, 2017).

Accordingly, building on prior work, we aimed to examine the efficacy of self-guided VRE for persons with SAD. There are several innovative aspects of the VRE used in the current study. First, our VRE had a virtual therapist built into it who auditorily guided participants through the exposure exercises and who conveyed CBT principles (see online supplementary material (OSM) for therapist script examples). We are aware of only one prior self-help VRE controlled trial that included a virtual therapist and this was for a treatment for acrophobia (Donker et al., 2019). The use of the virtual therapist also adds to prior work on non-self-help VREs (Anderson et al., 2013; Bouchard et al., 2017; Harris et al., 2002). Relatedly, the VRE included a built-in system to deliver and record subjective units of distress (SUDS) ratings. Further, it detected if the participant talked in response to the lifelike human character (professional actors were hired to play these roles) and the characters would wait until the participant provided verbal output before continuing the preprogrammed interaction). The current study also augments prior self-guided VRE controlled trials that did not contain cutting-edge, omnidirectional (360°), and dynamic video technology driven by machine learning optimization algorithms, biometrics, and programming rules (Lindner et al., 2019b; Lister et al., 2010). Also, to our knowledge, no other self-help controlled VRE study for SAD conducted exposure to bidirectional social interactions (Lindner et al., 2019b; Lister et al., 2010). Moreover, our study provided between-session exposure homework exercises to consolidate learning; a procedure mirroring face-to-face CBT but not conducted by any other self-help VRE study to date and omitted by more than half of the prior non-self-help studies on VRE for anxiety disorders thus far (refer to meta-analysis by Benbow \& Anderson, 2019).

We hypothesized that from baseline to post-treatment, self-guided VRE would result in substantially greater decrease in primary outcomes (SAD severity and job interview anxiety) when compared to WL. Further, we hypothesized that VRE (vs. WL) would lead to larger reductions in secondary outcomes, namely trait worry and depression symptom severity. In addition, we anticipated that pre-post-treatment gains would be maintained across 3-month follow-up (3MFU) and 6-month follow-up (6MFU) periods. Lastly, we conducted exploratory analyses to examine whether there would be enhancement in presence (i.e., sense of virtual world mimicking real-life scenarios) and reduction in cybersickness across VRE sessions.