Design and Evaluation of Virtual Reality Environments for Claustrophobia

Claustrophobia is defined as intense fear and avoidance of enclosed and small spaces with no way to escape. According to the Diagnostic and Statistical Manual of Mental Disorders (APA, 2013), claustrophobia is a type of specific phobia, but it can also be considered a type or prodromal stage of agoraphobia (Öst, Alm, Brandberg, & Breitholtz, 2001). To qualify as a phobic disorder, the experienced fear must be excessive or unreasonable and the severity of the anxiety or anticipation, and avoidance, must interfere significantly with the person's life (APA, 2013). About 12.5% of the population suffers from this fear and the majority are women (Vadakkan & Siddiqui, 2022). Feared situations involve, in a claustrophobic person, a strong urge to escape, with the fear of suffocating and the fear of restriction (e.g., restricted hand movements) being referred to as the two main dimensions of this disorder (Rachman, 1997). Typical situations that trigger anxiety are windowless or locked rooms, elevators, underground spaces, tunnels, subway trains, mine tours, and airplanes (Wiederhold & Bouchard, 2014). While being exposed to these situations, people with claustrophobic fear experience various emotional and physiological symptoms such as an intense urge to leave, severe anxiety, trembling, difficulty breathing, sweating, etc. (Vadakkan & Siddiqui, 2022). Anxiety and claustrophobia are also common during MRI procedures. It is estimated that about two million scans worldwide cannot be performed annually due to claustrophobic fear, with one out of 100 people experiencing a claustrophobic reaction requiring premature termination of the procedure (Munn, Moola, Lisy, Riitano, & Murphy, 2015).

Exposure-based treatments have been proven effective, especially in the therapy of anxiety disorders, with in vivo exposure (IVE) being the gold standard for phobia treatment. Virtual reality (VR)-based exposure technique offers the possibility of a virtual representation of the feared stimuli and allows exposure in a safe and controlled environment. There is already evidence showing a positive outcome of the VR-based exposure in the treatment of various phobias and other anxiety disorders, including PTSD, panic disorder, etc. Few existing meta-analyses comparing IVE with virtual reality exposure therapy (VRET) are showing that the effectiveness of these approaches is at least comparable. Results can differ depending on the number of sessions, type of treated phobia, or availability of follow-up data. A combination of both approaches (VRET and IVE) appears to be useful, but VRET can also be combined with other methods, for example, with CBT or cognitive training (Freitas et al., 2021; Powers & Emmelkamp, 2008; Wechsler, Kümpers, & Mühlberger, 2019; Wolitzky-Taylor, Horowitz, Powers, & Telch, 2008).

A few virtual environments (VE) for claustrophobia have been developed in previous studies that tested their feasibility and exposure potential (e.g., Botella, Baños, Villa, Perpiñá, & García-Palacios, 2000; Bruce & Regenbrecht, 2009; Rahani, Vard, & Najafi, 2018). The sample sizes ranged from two (Bouchard, Robillard, Larouche, & Loranger, 2012), to four (Botella at al., 2000), six (Malbos, Mestre, Note, & Gellato, 2008), and fourteen (total sample of 33; Rahani et al., 2018) in clinical samples, and from sixteen (Bruce & Regenbrecht, 2009) to eighteen (Christofi & Michael-Grigoriou, 2016) in nonclinical samples. The study by Rahani et al. (2018) is the only one including a control group of volunteers with no fear of enclosed space. Nevertheless, this study does not compare the effect of different environments on explored factors (playability and anxiety level) and at this point aims to decrease users’ anxiety after playing the game in which the user can freely choose between two chapters (elevator or MRI room). Virtual elevators, small rooms, corridors, tunnels, or caves are typically used in this type of VR study (for summary on previous studies, see Rahani et al., 2018). Bruce and Regenbrecht (2009) designed an environment of four interconnected rooms with an increasing number of claustrophobic cues, with the first room being spacious and bright and the last room being small, dark, and windowless.

Even though these clinical studies show promising results by means of reduction of discomfort in repeated exposures, the studies report very small samples and to our knowledge, no environments have been constructed specifically for long-term, exposure-based interventions. We argue that more diverse environments combining various provoking stimuli should be designed as a base for future randomized controlled trials. In addition, the existing virtual environments do not offer many opportunities for direct interaction with the present environmental features (such as pressing the elevator's button to use the elevator or when located in the basement, pressing the bell when stuck in the elevator), exposure to social constraints, and fear of losing control. With regard to the prevalence of claustrophobia, estimated as the third most common phobia (Stinson et al., 2007), and the variety of its potential triggers, we decided to implement a set of various cues and scenarios into a single virtual city environment to overcome some of these gaps.

This study aimed to design and test a set of scenarios situated in a single immersive VE to evaluate their feasibility in claustrophobic subjects, by means of measuring the provoked anxiety, induced sense of presence, and potential cybersickness symptoms in comparison to nonphobic individuals. In addition, as a base for a future clinical trial, we aimed to demonstrate the effectiveness of inducing anxiety in the three interactive scenarios in terms of anxiety provoked in claustrophobic individuals, when evaluated in standardized levels of difficulty. By comparing this standardized set of graduated situations and evaluating their potential for future use, we would be able to obtain a base for preparing therapy protocols to assess other factors such as the number of treatment sessions.

Eighteen volunteers participated in this study: 7 men (ages 20–36) and 11 women (ages 18–50). They were assigned to an experimental group (n = 10) or a control group (n = 8), depending on their reported fear of enclosed spaces, using the ICD-10 diagnostic criteria as follows: (a) intense fear or anxiety when exposed to small or enclosed space, this fear is (b) irrational, and (c) the person tends to avoid feared situations in daily life (interference with daily life). Participants from the control group did not meet any of these criteria. The sample was recruited from a general population through study advertising on the web page of the National Institute of Mental Health in Klecany, Czechia. The ethics approval was obtained from the internal Ethics Review Committee before the initiation of the study (see sociodemographic characteristics in Table 1).

The primary outcome measures were the State-Trait Anxiety Inventory (form STAI-XI), a commonly used measure of trait and state anxiety consisting of 20 self-report items rated on a 4-point Lickert scale, and the Subjective Unit of Distress Scale (SUDS), a self-report scale ranging from 1 to 10 (1 = “no distress,” 10 = “extreme distress”). The SUDS scale was used to measure the subjective level of anxiety currently experienced by the participant. This scale is commonly used during exposure sessions.

Before the experiment, participants were asked to rate their level of claustrophobic fear using the Claustrophobia Questionnaire (CLQ). This instrument is a reliable and sensitive measure of claustrophobia and its component fears (of suffocation and restriction) consisting of 26 items (Radomsky, Rachman, Thordarson, McIsaac, & Teachman, 2001).

After the experiment, sense of presence in VE was assessed using the Igroup Presence Questionnaire (IPQ). This instrument is used to measure the sense of presence in a virtual environment. This questionnaire consists of 14 items and has three subscales: spatial presence (sense of being physically present in the VE), involvement (attention devoted to the VE), and experienced realism (subjective experience of realism in the VE) (Berki, 2019). In addition, the single Overall presence item and the three subscores (IPQ factors) were evaluated as the mean values of the relevant IPQ items (Spatial presence: items 2–6; Involvement: items 7–10; Experienced realism: items 11–14). Finally, the Simulator Sickness Questionnaire (SSQ) (Kennedy et al., 1993) measured users’ level of sickness symptoms caused by virtual reality (VR). It is a subjective cybersickness self-report scale that assesses 16 symptoms of discomfort such as headache or dizziness, divided into three subcategories: nausea, oculomotor symptoms, and disorientation. The revised scoring method suggested by Bouchard et al. (2021) was applied to take into account the effects of anxiety induced by the VR-based exposure on evaluated cybersickness symptoms.

The HTC Vive Pro Head-Mounted Display (HMD) was used to immerse the participant into the virtual experience while providing accurate tracking of the person's head and hand movements. The virtual scene was presented from the first-person perspective and the participant used the HTC Vive controllers for movement and interaction in the virtual environment. The VR-based exposure therapy application (VRET-City) simulates a large virtual city environment developed in collaboration with the Czech Institute of Informatics, Robotics and Cybernetics (CIIRC CTU, Prague) for exposure therapy in different types of phobias (mainly specific phobias, social phobias, and agoraphobia), including several environments that can be combined in a single scenario (e.g., subway, skyscraper, etc.). For the described study, only environments with claustrophobia-inducing cues were used (elevator, subway, and underground parking lot).

In this feasibility study, volunteers reporting fear of enclosed spaces and volunteers with no reported discomfort were assigned to an experimental or a control group. During a single session, they were exposed to VEs of the three provoking scenarios in randomized order. Before the session, the severity of claustrophobia (CLQ), and the tendency to exhibit avoidance behavior were assessed. General anxiety level was assessed before and after the session (STAI-XI). The sense of presence (IPQ) and cybersickness (SSQ) were assessed at the end of the session. Based on the procedure of Bouchard et al. (2012), VR exposures in individual claustrophobia-inducing environments lasted about 10 minutes, including short breaks (2 minutes) between each immersion, located in the neutral environment (virtual city bench on an empty street, see Figure 1). All scenarios contained a situation of restricted freedom of movement (being trapped in an enclosed environment) of the same duration (120s); for a detailed description of each scenario's steps, see Table 2. The therapist encouraged participants to enter feared situations and to self-report their current distress levels using the Subjective Unit of Distress Scale (SUDS) at specific points (e.g., a few seconds after a stressful moment occurred). Given the nontherapeutic character of sessions, presented situations are rather basic with a moderate level of difficulty.

Descriptive statistics of primary and secondary outcomes are described in Tables 3 and 4. The difference in claustrophobic fear between both groups was tested using the nonparametric Mann-Whitney U test (M-W). The group difference in CLQ scores (evaluated prior to group assignment) was statistically significant (M-W U = 9.5, p = 0.008) with higher scores in the experimental group (see Table 3).

The tested subjects reported on individual items of the SSQ symptoms in a range from none (0) to slight (2). The group difference in the SSQ (see Table 4) score between the experimental and the control group was also tested using the nonparametric Mann-Whitney U test that did not show significant group effect for the SSQ total raw score (M-W U = 27, p > 0.05) and the Oculomotor factor (M-W U = 37.5, p > 0.05), while we observed a significant group effect in the Nausea factor (M-W U = 16, p = 0.03). When controlled for the potential anxiety impact on this factor (SSQ-Nausea-Anx), the group difference was not significant.

Participants were also asked about their previous experience with VR and their self-perceived level of computer literacy. A majority of them (15 out of 18) already had previous experience with VR, and no one rated their computer literacy as low (33.3% high, 66.7% medium).

This study aimed to design and consequently validate various immersive virtual environments (VEs) for exposure therapy in claustrophobia. Even though some VEs for claustrophobia have been created and small sample or case studies showed marked improvements, except for one study (Rahani et al., 2018), the combinations of experimental and control groups are still missing. We address the complexity of claustrophobic fears by evaluating different scenarios with a range of potentially triggering situations (Wiederhold & Bouchard, 2014), every one including a moment of failure. The scenarios also include virtual humans and realistic visual and sound effects. In this pilot study, we compared two groups of subjects (phobic and nonphobic) in subjective outcome measures.

All tested VEs (elevator, subway, underground parking lot) demonstrated a gradually increasing distress induced by the standardized levels of difficulty created and tested for each scenario. When comparing the three virtual environments, the most anxiety-provoking situation was being stuck in an elevator, probably related to avoidance of elevator rides, which can be quite easily achieved in real life and is common in people with claustrophobia. The least provoking environment seems to be the subway, which is, based on our case studies, more anxiety-inducing for agoraphobic patients. It cannot be ruled out that the lower distress ratings in the subway environment could be related to the presence of virtual humans used to simulate an ecologically valid situation, as their presence could give the impression of security to a certain extent. However, individual verbal reports of the participants do not indicate that and suggest the necessity to use this public transportation when living in a big city as a reason. The underground parking lot induces intermediate anxiety. Interestingly, participants from the experimental group commented on ventilators in the underground parking lot with fear of suffocation due to lack of air, compared to the control group, which didn't pay attention to this environmental feature. The ventilators were not originally intended as anxiety-inducing stimuli. Nevertheless, they represent a potentially relevant stimulus we can emphasize more in future clinical studies, making the underground environment even more challenging.

The general anxiety levels (STAI-XI) reported before and after the exposure did not differ significantly between the groups. Importantly, contrary to our expectations, we did not observe the increase in anxiety level after the exposure to claustrophobic scenarios. In contrast, the measured anxiety showed slight but nonsignificant decrease after the exposure in both tested groups (in line with another study using VR-based exposure in claustrophobia; Rahani et al., 2018). This could be related to anticipation anxiety, often reported in people with phobia (Straube, Mentzel, & Miltner, 2007).

We did not find any significant group difference regarding reported sense of presence in tested virtual environments; however, the items of the Spatial presence subscale and the single Overall presence item show some difference (higher values) in phobic participants. The Spatial presence refers to the sense of being physically present in the VE, which could be specific when claustrophobia is induced in VEs, even though they might not be perceived as realistic. This is in line with the strong correlation identified between the IPQ Spatial presence (mean score) and induced distress (SUDS) reported by the phobic subjects in some of the virtual environments. This association is most pronounced in the least provoking environment of the Underground parking lot and in contrast the weakest association was found for the most provoking Elevator environment, probably due to low variability of the responses of phobic participants in this smallest environment.

In terms of negative side effects of the immersive experience, both groups reported minimal or no cybersickness symptoms. The total SSQ score did not differ between the groups, but nausea-related symptoms were evaluated as stronger in the phobic subjects. However, we argue that these symptoms should not be attributed to the VEs, as they might be associated with the anxiety induced in the virtual environments. This is supported by the results obtained using the alternative scoring system suggested by Bouchard et al. (2021), which caused a considerable drop in the previously identified group difference.

Despite the limited sample size, the presented findings indicate the feasibility of the presented virtual environments and their ability to induce claustrophobic fear. These results will create a base for a larger clinical study exploring the therapeutic potential of our VRET-City application.

The small sample size represents the main study limitation that could affect the study results. However, the obtained results that are consistent with the hypothesized effects support the study's findings. Another limitation of the study is the presence of virtual humans in one of the evaluated scenarios, in contrast to their absence in the other two scenarios. This decision was based on simulation of realistic scenes, but it could potentially affect the distress ratings obtained from the claustrophobic participants (e.g., it might lower the distress from being restricted when in presence of virtual humans). Moreover, we did not perform a structured diagnostic interview (only basic ICD-10 criteria have been verified) that would allow us to rule out a potential overlap with agoraphobia that could be affecting obtained results.

These limitations will be addressed in the future clinical study including a larger sample. A differential diagnosis of claustrophobia, agoraphobia, or other similar phobias (e.g., cleithrophobia) will be performed. Future assessment of the environments should also consider presence or exclusion of virtual human(s) in individual scenarios, as this factor might potentially affect the level of anxiety perceived in claustrophobic or agoraphobic patients. Our environment connecting various virtual scenes with or without virtual characters might be suitable in future clinical work. Even though agoraphobia and claustrophobia are two distinct conditions, some authors consider claustrophobia a prodromal stage of agoraphobia (Öst, Alm, Brandberg, & Breitholtz, 2001). Specifically, it would be easy to add more virtual characters in the virtual scenes to evoke more crowdedness, which is typically associated with agoraphobia. Finally, the VR-based exposure should be combined with the measurement of physiological signals (such as ECG and skin conductance) that was partially implemented and tested in this study with a small subset of participants (data are not reported).

The feasibility study evaluated a set of immersive VEs of an elevator, a subway, and an underground space in claustrophobic and nonphobic individuals. The study results indicate that the designed VEs have the potential to provoke claustrophobic fear in phobic individuals and might be suitable for virtual exposure therapy in claustrophobia. We aimed to create an immersive experience and our findings suggest that this set of environments induced an overall sense of presence and particularly spatial presence, which was associated with the distress induced in phobic subjects in individual environments. Moreover, the perceived cybersickness symptoms were only minimal and could be mostly attributed to anxiety symptoms provoked by the VEs. Finally, individual VEs showed various levels of induced stress, but each environment tested in several standardized levels of difficulty demonstrated a gradually increasing distress induced by the VEs. This is extremely important, as it creates a base for the design of therapeutic protocols that could be utilized in a future clinical trial addressing the effectiveness of exposure therapy in the targeted clinical population. Future studies must pay attention to the factor of virtual characters, as their presence could modify the provoking potential of individual environments. However, the use of the social element opens up new possibilities for a potential therapeutic use in agoraphobia.

The VE design development and presented study was supported by the project No. TL03000223 “Exposure to stressful situations in a virtual city environment” (VRET-City) funded by the Technology Agency of the Czech Republic in the program ÉTA 3. We thank the collaborating team of Czech Institute of Informatics, Robotics and Cybernetics of The Czech Technical University led by J. Husák for creating the design and functionality of the virtual city environment.