Virtual reality environment for exposure therapy in obsessive–compulsive disorder: a validation study

Obsessive–compulsive disorder (OCD) is characterised by recurrent, repetitive, and unwanted thoughts or impulses triggering significant anxiety. Exposure and response prevention is currently the first-line therapy for OCD. The goal of this validation study was to confirm the potential of the VR house environment that incorporates OCD-specific items that cluster around major symptom dimensions: ‘contamination’, ‘symmetry’, ‘checking’ and ‘hoarding’ to induce anxiety and compulsive behaviour in patients with OCD. We assessed a sample of OCD patients (n = 44) that was compared to a group of healthy controls (n = 31). The severity of OCD symptoms was assessed in all subjects. During a single session, participants were asked to approach a set of 10 stimuli (covering four OCD dimensions) and rate their current intensity of distress/anxiety and compulsive tendencies (scales 0–5) provoked by observing each stimulus. Before and after the VR exposure, participants completed questionnaires assessing subjective levels of anxiety (before/after VR exposure), their sense of presence in VR and experienced simulator sickness. The results show that the OCD group reports elevated levels of distress and compulsive behaviour when confronted with VR exposure stimuli compared to the control group, but no increase in anxiety levels has been observed after the VR exposure. The subjective ratings of provoked distress and compulsive behaviour are not associated with severity of OCD symptoms, perceived sense of presence, association with cybersickness symptoms is weak. Our data suggest that the VR house environment is a suitable tool for VR exposure therapy in OCD patients as it demonstrates OCD symptom provocation relevant for individual patients.


Introduction
Obsessive-compulsive disorder (OCD) is a severe neuropsychiatric disorder defined by the presence of obsessions (intrusive recurrent and persistent thoughts, images, or urges) and compulsive repetitive behaviours (e.g. washing, checking, arranging things in a specific way) or mental acts (e.g. counting) (APA 2013). Compulsions are often performed to prevent obsessive thoughts from appearing or make them disappear and temporarily reduce anxiety related to them (APA 2013). OCD symptoms (OCS) induce personal distress, they are time-consuming and can strongly interfere with personal, social, and professional life. These symptoms are very heterogeneous, a dimensional approach (Summerfeldt et al. 1999) to OCS has been therefore proposed in clinical research. Various dimensional models with variable number of dimensions have been proposed (for meta-analyses, see Bloch et al. 2008), but the traditionally accepted approach includes four dimensions of 'contamination/cleaning' (C), 'symmetry/ordering' (S), 'fearof-harm/checking' (CH), and 'hoarding' (H). However, for OCD patients, it is quite common to present with overlapping symptoms and each patient can score in one or more symptom dimensions (Mataix-Cols et al 2005;Mataix-Cols 2006). In a study of 139 patients with OCD, half of the patients had one dimension (50.4%) and the frequency of patients with two and three dimensions was 24.5% and 1 3 15.8%, respectively, and of those with four, five, and six dimensions was 6.5%, 1.4%, and 1.4%, respectively (Li et al 2009). Across various countries, the reported frequency of symptom dimensions for obsession were in the following range: contamination 37-60%, symmetry 19-55%, aggressive thoughts 8-56%, and hoarding 11-28%. The frequency of symptom dimensions for compulsions was washing 42-72%, checking 16-72%, repeating 22-68%, and hoarding 0.5-45% (overview in Li et al 2009;Matsunaga et al 2008).
OCD diagnosis in individuals is established by using standard criteria provided by diagnostic systems, specifically International Classification of Diseases (ICD) and Diagnostic and Statistical Manual of Mental Disorders (DSM). These systems are related, but not identical. Hoarding may have some features distinct from other dimensions and even from OCD itself, which is reflected in current DSM-V, where hoarding has been separated from obsessive-compulsive disorder but remained in the same diagnostic class of obsessive-compulsive and related disorders. However, in the still valid ICD-10 nomenclature, compulsive hoarding falls into the diagnostic category of OCD, reflecting commonalities on key diagnostic validators (compulsive behaviour and distress related to barriers to such behaviour) and possible shared genetic factors (WHO 2022). For the purpose of this study, the ICD-10 nomenclature, currently valid in the Czech Republic, was used, and hoarding disorder was included in our OCD sample.
First-line treatments for OCD are antidepressant medication (serotonin reuptake inhibitors) and cognitive-behavioural therapy (CBT) incorporating in-vivo exposure and response prevention (ERP) techniques (Hirschtritt et al. 2017). CBT and ERP are widely recommended in clinical guidelines for treatment of OCD (e.g. Fineberg et al 2020). ERP method involves progressive, deliberate, and voluntary exposure to stimuli triggering obsessive thoughts and anxiety while knowingly averting compulsive, neutralisation or avoidance behaviour (response prevention) and it represents a very effective CBT technique for treating OCD (Hezel and Simpson 2019). Recent theory suggests that ERP facilitates new adaptive responses by inhibitory learning that promotes new association with feared stimuli competing with the former nonadaptive response that does not disappear completely (Craske et al. 2014). By withholding compulsive reactions, OCD patients are able to learn new associations to feared stimuli (Powers et al. 2007).
Due to potential episodic or chronic course of the illness and limited availability of specialised CBT programmes for OCD patients (McKay et al. 2015), this creates an increasing need for personalised medicine (incl. therapeutic tools) that is currently on rise and is aimed to increase the number of patients who respond to therapy.
Virtual reality exposure therapy (VRET) seems to be a good alternative to in vivo (standard) exposure techniques applied in various anxiety disorders and OCD. In VRET, the person is immersed in a virtual environment whose characteristics directly confront her or him with a feared situation or unpleasant stimuli that correspond with symptoms-triggering real-life situations (Abramowitz et al. 2019). There is already strong evidence showing that VRET can effectively treat anxiety disorders such as specific phobias or PTSD (for a review, see Maples-Keller et al. 2017). Enrichment of the standard CBT program with VRET techniques could potentially increase its efficacy in OCD, as a similar enhancement effect was already reported for VRET-enriched CBT programs in anxiety disorders (Wu et al 2021). Moreover, often, it is easier, economical, and most importantly also safer to access these stimuli in a virtual environment than in real life. VRET technique enables the patient to initiate exposure therapy in a safe environment as a form of preparation for the real-life exposure performed in the CBT.
VR-related factors such as sense of presence and cybersickness could be associated with how the VRET technique is perceived and rated by the treated individual (Weech et al 2019). Sense of presence is broadly described as a sensation of "being there" in the virtual environment (Barfield et al. 1995;Usoh et al. 2000), the situation when the participant forgets that his experience is mediated by technology (Lombard and Ditton 1997); whereas cybersickness is defined as a visual-vestibular conflict of human perception, leading to nausea, headache, or dizziness (McCauley and Sharkey 1992). Potential effects of these concepts on the VR experience are opposite (Weech et al. 2019). A meta-analysis aimed at VRET in phobias reported evidence showing an existing association between sense of presence and anxiety induced by VRET. In means of cybersickness, a study by Ling et al (2011) provides evidence that cybersickness reports measured in VRET environments may partly be explained by feelings of anxiety rather than cybersickness per se, as such symptoms are not reported in the same individuals in a neutral context.
The first study assessing VRET in OCD ) showed that after provocation by a virtual scenario, patients reported higher anxiety before checking and performed more checking behaviours. These findings demonstrated that some of the behavioural patterns of OCD patients can be replicated in VR. Similarly, Laforest et al. (2016a, b) assessed contamination fears and compulsions by comparing OCD patients and healthy controls immersed in a neutral virtual environment (clean room) and in filthy public restrooms, showing anxiety increase (assessed by self-report and heart rate measure) among OCD patients. Cárdenas and colleagues (2012) later developed virtual public toilets and other environments (virtual bus, a restaurant, and a bedroom). The above-listed studies suggest that immersive VR environments presenting OCD-specific stimuli, can induce anxiety in OCD patients and can be potentially used for exposure therapy (Bouchard et al. 2019). A current study by Cullen et al. (2021) compared VR and in vivo contamination-related ERP in a kitchen and a bathroom environment with graded exposure tasks in OCD. Their findings using subjective and objective measures of distress (heart and respiration rates) indicate that both virtual and in vivo ERP sessions provoke consistent anxiety profiles across an exposure hierarchy.
To our knowledge, there is only one pilot study assessing the efficacy of VRET in OCD (Laforest et al. 2016a, b) examining the effectiveness of a VRET-enhanced CBT program in three patients with contamination OCD subtype. The study shows a significant reduction in intensity and severity of OCS in all participants, but the global improvement was apparent only for two of them, as the OCS of the third patient were less related to toilets. The necessity of intensifying the consequences of actions visible in VR environments and creating a narrative specific for individual patients has been pointed out. Recently, a VR game has been developed (van Bennekom et al. 2021) to provoke and assess OCD symptoms from three symptom dimensions (including items such as burning gas, dirty sink, and messy table). Nevertheless, this game is not fully interactive, a first-person video of intervention or checking behaviour is shown when a participant chooses to check. To our knowledge, a complex VR environment that would incorporate all OCD dimensions and its primary purpose is exposure therapy has not been created yet. To overcome the problem of symptoms' heterogeneity, we, therefore, designed a complex VR environment in concordance with the dimensional model and its dimensions (Francová et al. 2019;Fajnerova et al. 2021) that allows us to select and freely combine stimuli from various dimensions.
The main aim of the presented study was to assess the validity of a preselected set of OCD-relevant stimuli presented in the virtual house environment to provoke symptoms in OCD patients. To this end, individual stimuli were evaluated (by means of induced anxiety and urge for compulsive behaviour) by a group of OCD patients and a group of healthy control subjects, compared using the betweensubjects design. Subjective level of anxiety was rated by all subjects prior to and after this VR experience to address its short-term effects. In addition, we aimed to test the association between VR exposure ratings and the severity of OCD symptoms and VR-related experiences such as sense of presence and cybersickness.

Study sample
Prior to study initiation, ethical approval has been obtained for all protocols from the NIMH Institutional Review Board-Ethical Committee. All subjects read and signed written informed consent. No clinical trial has been registered. In total, 44 patients diagnosed with obsessive-compulsive disorder F42.X (F420 (N = 4), F421 (N = 4), F422 (N = 36)) and 31 healthy subjects participated in this validation study. The patients were diagnosed based on ICD-10 nomenclature criteria, which are up-to-date and valid in the Czech Republic. No dropout appeared in the study. All patients were recruited during initial phases of the 6-week long treatment program which is based on group cognitive-behavioural therapy (inpatient psychiatric care or daycare) in the National Institute of Mental Health (NIMH) in Klecany. The patients were approached during the second week of the treatment, the stimuli validation was completed as a part of the initial evaluation with the aim of selecting stimuli suitable for potential therapeutic use. Participation in the study was voluntary, without any remuneration, and it did not affect the standard treatment provided to the patients. The patients diagnosed with OCD showed mild to severe OCD symptoms and were treated for OCD (illness duration range 0-25 years, Median: 11). All patients were medicated with antidepressant medication based on clinical judgement of the attending psychiatrist. The OCD subgroups' sizes recruited in this study represent the distribution of OCD dimensions and their incidence in the CBT program run at NIMH, with the most common predominant dimension being "contamination/cleaning" (N = 23; 52%), followed by "control/checking" (N = 14; 32%), "symmetry" (N = 3; 7%) and hoarding (N = 4; 9%). The symptom dimensions in our sample are in line with the above mentioned range (see Introduction section). However, the dimensions (primary and secondary) identified in individual patients did overlap in most of the patients in some typical combinations; a single dimension has been present in a subsample of 16 patients.
The control group subjects were recruited by using a snowball sampling method to match the demographics of OCD patients; therefore, the groups are comparable in means of demographic characteristics of age and education (level 1-elementary school, 5-university degree). The male/female ratio was also comparable in both groups. See Table 1 for more details.

VRET technical equipment and software
The VR scenarios are created using the VR game engine Unity®Pro (https:// unity 3d. com/). The head-mounted display (HMD) HTC Vive Pro was used for presentation of VR exposure scenarios, providing accurate tracking of the person's head and hand movements by using HTC Vive controllers for locomotion and interaction in the virtual environment. The participant experiences the virtual scene from a first-person perspective with virtual hands that simulate life-like hand presence. Both active and passive (teleportation) movements were enabled in the virtual scene.
The VRET software enables active interaction with objects located in the environment using controllers. The VR scene shows a family house (with a bedroom, kitchen, etc.). The VR environment contains more than 40 various stimuli that can be freely combined in a form of complex scenarios.

VRET experimental design
Both groups were assessed using a set of 10 VR exposure stimuli to demonstrate their validity and thus the potential of VR exposure scenarios, to provoke anxiety and compulsive behaviour in OCD patients compared to healthy subjects. The participant is immersed into a virtual home environment that is specifically designed to provoke OCS typical for home settings with scenarios mainly focusing on evoking anxiety and compulsive behaviour in dimensions of "contamination/cleaning", "fear-of-harm/checking ", and "symmetry/ordering". To address the VRET validity, a standardised set containing 10 stimuli that OCD patients often excessively respond to was used as a short exposure scenario covering all OCD dimensions. The therapist guides the patients to reach individual stimuli one-by-one and asks them to verbally rate their current level of anxiety (scale 0-5) and the need/urge for compulsive behaviour elicited by observing this object. The observation time has not been limited, but patients were asked to rate their anxiety around 10-15 s after approaching each stimulus. The set included three stimuli for every dimension (C, S, CH), except hoarding (H) with only one stimulus present. The patients reached them in predefined instead of randomised order, because they logically corresponded to the virtual house floor plan (e.g. the bathroom close to the main door, the living room following the kitchen, etc.) Stimuli were selected based on a feasibility study conducted in a smaller sample of ten OCD patients. Initially, approximately 19 stimuli were identified and tested in this pilot feasibility study (Taranzová 2021). From this larger set, only the 10 most anxiety-eliciting stimuli were selected to be used during validation in the final stimuli set, mainly due to the time demands of the examination. The selected set of tested items ensured that each dimension was represented, and all stimuli were sufficiently specific to OCD (we, therefore, excluded particularly disturbing stimuli, e.g. faecal matter, which may elicit disgust even in the control group). The whole VR immersion lasted for about 20-25 min without breaks (Fig. 1).

Psychiatric scales and psychometric methods
To assess the severity of OC symptoms, the Yale-Brown Obsessive Compulsive Scale (Y-BOCS; Goodman et al  Usoh et al. 2000) with 6 questions rated using Likert scale 1-7, with total score range 0-6 calculated as count of '6' or '7' scores amongst the 6 questions and cybersickness using the SSQ questionnaire with total score up to 236 (Simulator Sickness questionnaire; Kennedy et al. 1993;Nesbit et al. 2017).

Study procedure
After giving informed written consent, each subject was first evaluated by the psychiatric scales (Y-BOCS, BAI) assessing present symptomatology. During the VRET session, the participant first filled in the STAI-6 questionnaire and then he/she moved through an immersive VR environment to meet and assess all components of the 10-stimuli set (for detailed description see Table 2). The participants moved through individual rooms of the house environment and consecutively visited individual stimuli one-by-one. After the VR exposure, the participants filled in STAI-6 and VR presence and cybersickness questionnaires.

Statistical analysis
All data were pre-processed and analysed using statistical software Statistica 64 (TIBCO software Inc.). Most of the tested variables did not show normal distributions tested by the Shapiro-Wilk test. The data were analysed using nonparametric statistical methods (significance level set at p = 0.05). The Chi-square test was used to compare the sex distribution in the two groups. Simple group differences were analysed using Mann-Whitney U (M-W U) test, and potential relationships between individual measures were analysed using Spearman's correlations. In addition, the effect of VR exposure on repeated evaluation of anxiety (STAI-6) was analysed using the Wilcoxon Matched Pairs test separately for each group.

Study sample specification and measured variables
In total, 44 OCD patients and 31 healthy volunteers were included in the statistical analysis. The groups (OCD vs. HC) differed in several measures (see Table 3). The symptom severity score rated using Y-BOCS was significantly higher in the OCD group in comparison with the control group. Both groups also differed in anxiety level measured using BAI (anxiety present in the last week) and STAI-6 (anxiety rated prior and after the VR exposure). The groups do not differ in subjective ratings of presence (SUS) and cybersickness (SSQ).

SUDs and SUCs score differences
First, we tested the assumption that stimuli displayed in a virtual house environment can effectively simulate an exposure situation and thus induce anxiety or distress measured by Subjective Units of Distress Scale (SUDs) and urge for compulsive behaviour (SUCs) in patients while immersed into the selected exposure scenario (set of stimuli). Our assumption was that the scenarios/stimuli selected in the virtual environment are relevant for OCD and would lead to discomfort in the experimental group, but in contrast, would Note, the total SUDs/SUCs score was calculated as a cumulative score of individual ratings (0-5) for the whole set of 10 stimuli (summary score range 0-50) and this cumulative score is applied in all following statistical analysis. Note, we included the complete OCD sample (n = 44) in the statistical analysis regardless of the predominant symptom dimension determined in individual patients, as we were only concerned with the difference in reported SUDs compared to matched healthy volunteers (n = 31). The group difference was tested using the Mann-Whitney U test. The SUDs rating scores were significantly higher (U = 302.5, p < 0.001, d cohen = 1.18) for OCD patients (Mean (SD) = 14.74 (11.09), Median = 13, range 0-41) compared to healthy subjects (Mean (SD) = 4.39 (5.73), Median = 2, range 0-21), see Fig. 2 left.

Symptoms severity and its relation to SUDs and SUCs ratings in OCD sample
Based on the previous studies (e.g. , symptom severity rated by the Y-BOCS scale in OCD patients was expected to be significantly correlated to the degree of subjectively reported distress on the SUDs/ SUCs rating scale. We tested this assumption in our data  . 2 Group comparison of the VR exposure induced anxiety/distress (SUDs) and compulsive urges (SUCs). Note: The scores were measured for each stimuli using a subjective verbal scale (range 0-5) rated by OCD patients and controls (HC). Total scores were calculated as cumulative scores for the whole set of 10 stimuli (range 0-50) using the Spearman correlation analysis. Only the OCD sample was analysed instead of the whole sample to prevent false positives driven by observed group differences in the correlated variables (see Supplementary Table 1 for the whole sample correlations). No significant correlation was revealed for Y-BOCS total score when tested in the OCD group for SUDs (r s = − 0.05, p = 0.766) and SUCs ratings, respectively (r s = − 0.08, p = 0.635); however, the obsession symptom severity subscale was negatively correlated with SUDs (r s = − 0.37, p = 0.014) and SUCs (r s = − 0.39, p = 0.009), respectively. No correlation was found between compulsive symptoms severity and the SUCs/SUCs ratings. Similarly, stimuli ratings SUDs/ SUCs and BAI show no significant association with SUDs (r = − 0.138, p-value = 0.402) and SUCs, respectively (r = 0.014, p-value = 0.934). Importantly, the anxiety state evaluated using STAI-6 before/after the VR experience did not correlate with the total SUDs and SUCs scores rated during this experience showing no significant association p > 0.05 (SUDs: r = − 0.07/− 0.13; SUCs: r = 0.04/− 0.04).

VR presence and cybersickness in respect to SUDs scores rated by OCD patients
Previous studies also suggested an association between the SUDs/SUCs ratings and the sense of presence in VR evaluated by the patients (Belloch et al. 2014;Laforest et al. 2016a, b). Therefore, we were interested in whether reported exposure-related anxiety (using SUDs rating) and need for compulsive behaviour associated with presented stimuli (SUCs rating) were related to the sense of presence patients experienced in the virtual environment (by means of the VR presence evaluated by the SUS scale). Spearman's correlation coefficient tested for the SUS questionnaire measuring subjective sense of presence in the virtual environment, showed no significant correlation for the rated variables (SUDs: r = − 0.19, with a p = 0.289; SUCs: r = − 0.08, p = 0.658). Please note that the data are only provided for a subset of 32 OCD patients due to incomplete or partially invalid data from VR feedback questionnaires in 12 patients. Finally, we investigated whether virtual environmentinduced sickness (measured via SSQ) may impact how the VR experience has been rated by OCD patients (in means of presence, SUDs and SUCs scores). No significant correlation was found between the tested variables (SUS and SSQ: r = 0.264 (p = 0.144). Similarly, no significant correlation was found between SSQ and SUDs score (r = 0.31, p = 0.087). (See Supplementary Table 1 for the whole sample correlations).

Subjective anxiety score differences
We further hypothesised that the difference in anxiety ratings would be evident in the STAI-6 questionnaire administered to both groups before and after VR exposure. In line with SUDs ratings, we expected the group of OCD patients to show higher scores in the subjective rating of anxiety state reflected by the STAI-6 questionnaire after exposure when compared to the group of healthy volunteers. Moreover, we hypothesised the STAI-6 score will be increased in OCD patients after the VR exposure session (compared to baseline evaluation prior to the VR session), in contrast to healthy subjects that should not exhibit any changes in a rated anxiety state. Both, group effect and interaction group*STAI-6-changes have been expected.
The Mann-Whitney U test has been used to test the group differences for both STAI-6-pre/post variables, showing significant group effect with (p < 0.001) for STAI-6-pre and STAI-6-post variables (see Table 3 for results). The potential change in the evaluation of anxiety (STAI-6-pre vs. STAI-6-post, within effect) has been compared using the Wilcoxon Matched Pairs Test. Contrary to our expectation, no significant change has been identified in OCD patients (T = 213.5, p = 0.061), while a significant decrease has been reported by the control subjects (T = 60, p = 0.006). For the distribution of the two scores in individual groups see Fig. 3.   Fig. 3 Group comparison of the STAI-6 subjective anxiety scores evaluated by OCD patients and control subjects (HC) before (pre) and after (post) VR exposure session 1 3

Discussion
The presented study confirmed the potential of the newly designed VR home environment to provoke typical manifestations in OCD patients in means of rated distress (anxiety) and the urge to perform compulsive behaviour. OCD patients rated the set of 10 provoking stimuli as more distressing and more provoking (in means of the SUDS and SUCS ratings as seen in Fig. 2) in means of compulsive rituals (including avoidance) in contrast to control subjects that rated none or minimal distress and reported no tendency for compulsions in the same virtual scenario. This is in line with our preliminary data suggesting that the VR house environment could represent a useful tool for VR exposure therapy in OCD patients (Fajnerova et al. 2021). The complex VR household demonstrated symptom provocation specific to OCD in selected set of stimuli in line with previous studies that validated various environments for VRET in OCD (e.g. van Bennekom et al. 2017;Laforest et al. 2016a, b;Belloch et al. 2014;. Note that cumulative SUDS/SUCS scores have been used for the statistical analysis instead of mean scores. This decision is based on the high variability of stimulus evaluation in OCD patients, and on the contrary, low variability in control subjects, who often rated the distress caused by stimuli as 0. We argue that the cumulative score expresses the different representations of individual dimensions/subtypes in OCD sample more adequately and the effect of their combinations in the OCD sample, which we believe a simple average could not address properly. The present combinations of symptom dimensions are clearly visible in the cumulative SUDS/SUCS scores that show strong variability in the tested OCD sample, ranging from 0 in "pure mental compulsions'' (such as counting and other magic rituals, which might not be triggered by a particular stimuli) group up to the cumulative score of 42 suggesting multiple dimensions overlap in this patient.
In contrast to some of the previous studies (e.g. Kim et al 2010), SUDs ratings of the VR exposure stimuli were not positively associated with the OCD symptom severity evaluated by Y-BOCS or general anxiety score (BAI). This missing association is not surprising in our study sample as the SUDs ratings were calculated as a cumulative score for a set of 10 stimuli presented, where only a small subset of them was associated with the symptom dimension(s) identified in individual patients. This is in line with the findings of the study by van Bennekom et al. (2021) which also combined various subtypes of OCD and found no correlation between the emotional responses and virtual compulsions with Y-BOCS scores. A relatively small sample size, however, does not allow us to test this assumption by dividing the SUDs ratings into four clusters and evaluating their association with symptoms severity for the respective subset of patients. This matter should be addressed in future studies.
Contrary to our expectations, the anxiety subjectively evaluated using STAI-6 before and after the VR house exposure showed a decrease in total STAI-6 score that was significant in the healthy control group, but not in OCD sample. This finding is in line with verbal reports obtained from some of the participants, indicating that the anxiety evaluation performed before VR exposure is often associated with some anticipatory stress present due to expectations and fear of the unknown (potentially related to limited/lack of previous VR experience), they might encounter in the VR environment. On the other hand, after the exposure session, the feeling of satisfaction and/or relaxation associated with the completed task and well-managed exposure was often described. This observation could explain the contradictory data and missing correlation between the SUDs/SUCs ratings and the STAI anxiety levels evaluated after the VR experience.
Our study addressed the previously reported direct association between the symptom provocation in VR exposure scenarios and the sense of presence in VR (SUS scale) evaluated by the patients (Belloch et al. 2014;Laforest et al. 2016a, b;Inozu et al 2021). In contrast to previous studies, we did not identify positive correlation between the provoked symptom ratings (both in means of distress and compulsive behaviour) and the sense of presence experienced by the OCD patients during the VR exposure. The reason for the missing association between provoked symptoms and presence could be related to the mixed scenario including stimuli of all OCD subtypes and also the nature of enrolled OCD patients with overlapping symptom dimensions, which did not occur in the previous studies. Moreover, this correlation can be affected by various moderating factors and has been reported much stronger in the clinical population of phobic individuals than in nonclinical samples (Ling et al 2014).
To control the potential negative effect of cybersickness on the rated experience, we tested the correlation between the rated simulator-sickness questionnaire score and the SUS, SUDs, and SUCs scores. No significant association was revealed for any of the tested variables. Moreover, the rated simulator sickness can be considered mild and SSQ scores did not differ between the OCD patients and healthy control subjects. These results suggest that the cybersickness did not negatively affect observed findings. Interestingly, our preliminary data report that analysed a subset of the presented sample showed increased cybersickness in OCD patients when compared to healthy controls. It was previously suggested that this could be related to the unpleasant experience of the VR exposure with OCD-relevant symptoms ) and that VR-induced nausea and visuomotor disturbances reported by the patients could be confused with anxiety symptoms (perspiration, general discomfort, difficulty concentrating, etc. Bouchard et al. 2011). Even though our data show no significant correlation, the p-value of 0.087 obtained for positive correlation (r = 0.31) between reported cybersickness and compulsive behaviour provoked by the exposure stimuli indicates that the suggested association might be relevant in OCD and could be potentially related to disgust.

Study limitations and future directions
To our knowledge, this is the first study validating VR exposure scenarios in OCD patients of all subtypes. Despite sufficiently large sample sizes of the experimental and control groups, individual OCD subtypes were not represented equally as their proportions correspond to the incidence of individual OCD subtypes in the CBT program provided at NIMH. Moreover, as only the predominant/primary OCD dimension was used to divide patients into 4 subgroups, some more frequent combinations of subtypes present in OCD patients were not taken into consideration and due to the overlap of symptom dimensions in a subset of patients, the symptom dimensions could not be analysed separately. Future studies should address the validation of individual subsets of stimuli in the OCD house virtual environment in individual subtypes of OCD.
Another study limitation is the predetermined order applied in the evaluation of exposure stimuli instead of randomised order approach. The reason for this decision was our aim to achieve naturalistic movement around the house environment that would not allow the tested subject to walk around some of the stimuli without noticing them; therefore, these were approached in the order based on their localisation in the house environment. However, this predefined order could affect the subjective ratings of the presented stimuli; therefore, the individual ratings were not evaluated separately.
Moreover, missing data from questionnaires related to the VR experience in means of sense of presence and cybersickness in a subset of patients could due to a smaller sample size affect related results, which should be therefore considered with caution. Another limitation is the missing report on previous experiences with VR, as the lack of VR experience could affect the anticipatory stress measured using STAI prior to the experience. To our knowledge, our sample included only a very small subsample of participants with prior VR experience. However, this information was provided only verbally and was not recorded for all subjects. This should be controlled in future studies.
Objective physiological measures (heart rate, skin conductance, etc.) should be included in future studies to support the subjective evaluations provided by the patients.
Finally, future studies should also investigate the validity of the complete set of stimuli (above 40) present in our VR house environment and compare them to in vivo exposure. Moreover, a randomised clinical trial aimed at VRET effectiveness with gradually increasing intensity of stimuli should be conducted using the presented VR house environment, including a follow-up 6 to 12 months after the treatment completion to address long-term outcomes and effect on quality of life.

Conclusions
The presented study confirmed the potential of the VR home environment to evoke typical manifestations in OCD patients in contrast to control subjects that rated no or minimal distress in the same virtual scenarios. This is in line with our preliminary data suggesting that the VR house environment could represent a useful tool for VR exposure therapy in OCD patients. Using a VR home environment, we demonstrated symptom provocation specific to OCD.
The state anxiety (measured using STAI-6) associated with VR exposure rated prior to and after the session was significantly higher in patients than in control subjects. Contrary to our expectations, the anxiety level rated before and after the VR exposure session showed a decrease in the total score in the healthy controls but not in OCD patients. This finding may be potentially related to strong expectations and anticipatory anxiety due to limited VR experience before the VR exposure session and subsequent satisfaction after the task was completed.
Subjective ratings of distress and compulsive behaviour provoked during the VR exposure stimuli were not associated with symptom severity (Y-BOCS, BAI). Also, previous studies addressing this matter bring contradictory results reporting present or missing association. We argue that correlation between OC symptoms and rated VRET experience can be identified only if subjectively relevant stimuli were used for each patient or in case of unified OCD samples targeting only a single dimension. Unfortunately, the present study design with data obtained in a mixed cohort of OCD subtypes, did not allow us to carefully investigate this.
Finally, we did not identify any significant association between SUDs/SUCs and the sense of presence reported in some previous studies. Cybersickness symptoms were not significantly related to the evaluation of VRET experience (provoked symptoms), however, some mild association could be observed with the provoked compulsive behaviour.
In conclusion, the presented data confirm the validation of the VR house environment for the exposure therapy in OCD patients. Future work should address some of the issues emerging in this study. Particularly, validity of the whole set of various scenarios with respect to the dimensional approach (each set addressing a single OCD subtype) should be evaluated in a sample of OCD patients including bigger subgroups of all subtypes/dimensions. Moreover, a randomised clinical study aimed at VRET effectiveness should be conducted using the presented VR house environment. In the future, the complex VRET environment may enable the inclusion of new procedures, for example monitoring of physiological data to determine more precisely the induced mental and physical state of the subject (Meehan et al 2002) and lead thus to much-needed progress in current treatment methods for OCD.