Virtual reality increases pressure pain threshold and lowers anxiety in children compared with control and non‐immersive control—A randomized, crossover trial

Virtual reality (VR) is a promising non‐pharmacological pain intervention because it may not only distract but also modulate pain by immersing the user in a three‐dimensional 360° alternate reality. In children, VR has been reported to reduce clinical pain and anxiety during medical procedures. However, the effect of immersive VR on pain and anxiety remains to be investigated in randomized controlled trials (RCT). The aim of the present crossover RCT was to assess the effect of VR on pressure pain threshold (PPT) and anxiety level measured with the modified Yale Preoperative Anxiety Scale (mYPAS) in children in a controlled experimental setting.


| INTRODUCTION
Immersive virtual reality (VR) is a promising method to lower pain and anxiety (Chan et al., 2018). It immerses the users in a three-dimensional 360° alternate reality by use of a headset with a pair of goggles with sensors that track the head movement. Furthermore, users can interact with the virtual reality with a controller (Arane et al., 2017). VR may reduce pain by limiting the processing of pain signals by stimulating the visual cortex simultaneous with other senses (Colloca et al., 2020). Another effect of VR may be through distraction, redirecting attention away from the painful stimulus and hence affecting both pain perception and situational anxiety (Gomez-Polo et al., 2021).
VR is progressively being utilized in a clinical paediatric setting as non-pharmacological pain and anxiety management and seems to be beneficial for children, although well-controlled experimental studies are pending (Chan et al., 2019;Eijlers et al., 2019;Jivraj et al., 2020). In a meta-analysis analysing the clinical efficacy of VR for a mixture of adult and paediatric acute procedural pain, 16 randomized studies suggest a reduction in pain scores with VR (Chan et al., 2018). However, after subgroup analysis VR showed no effect for minor surgical procedures or burns wound care, but a favourable effect on needle pain and in burns physical therapy, indicating that the effect of VR varies depending on population and clinical scenario. This emerging body of clinical studies examining the effect of VR on acute or procedural pain/anxiety in various populations should be built on a foundation of well-controlled experimental studies. In addition, a high degree of bias and heterogeneity of clinical VR studies calls for large well-controlled and methodologically sound studies validating the use of VR in an experimental controlled setting eliminating the element of painful medical procedures before widespread clinical use of VR is recommended. High-quality assessment methods for the independent effect of VR on PPT and anxiety in a crossover study design render the possibility for optimal control conditions in an experimental setting instead of the inconsistencies of standard of care, which often serves as control group in clinical studies. In a crossover design, each participant acts as their own control typically with two interventions in two sequences (2 × 2). However, a more advanced randomized crossover design secures extensive control conditions controlling for both no digital intervention/small talk and non-immersive 2D video on a tablet as well as both interactive VR gaming and 3D VR video (Dwan et al., 2019;Mulkey et al., 2019).
In adults, a well-controlled study has found that immersive VR increases heat-pain tolerance and decreases situational anxiety (Colloca et al., 2020). In children, interactive and passive distractive VR increases pain tolerance and threshold; however, using older VR technology (Dahlquist et al., 2007;Law et al., 2011). Pain tolerance thresholds is less ethically feasible or sensitive in children; therefore, pain thresholds appear to be a better measure as it is less affected by psychological factors such as willingness to tolerate pain stimuli. As both situational anxiety and threshold for pain are decisive as to how a child perceives a clinical painful stimuli, a controlled experimental setup seems to be relevant using algometry and the modified Yale Preoperative Anxiety Scale (mYPAS) (Kain et al., 1997;Nikolajsen et al., 2011;Pedersen et al., 2020;Skovby et al., 2014).
The aim of the present study was to assess the effect of immersive VR on PPT and mYPAS in a controlled experimental setting utilizing a within-subject crossover trial with four interventions. The hypothesis was that VR (VR gaming and immersive VR video) have positive modulatory effects on pain threshold and anxiety level (PPT increases and mYPAS decreases) compared with control interventions (2D video and small talk).

| Participants
Seventy-two children (6-14 years old) were enrolled after assent from the children and written consent by both parents. Children were recruited in the orthopaedic outpatient clinic at Aarhus University Hospital during 2021, and the study took place when the clinical outpatient visit had come to an end in a controlled experimental setting. The speciality of children orthopaedics consists of many subspecialties, for example trauma sequelae, congenital deformities, growth disturbances and physiological deformities all needing highly specialized evaluation in our children orthopaedic outpatient clinic. Examples of referral diagnoses of children included in present study are genu valgum or varus, anisomelia, foot deformities such anxiety level decrease compared with extensive control conditions. Paediatric immersive VR is effective, feasible and valid for non-pharmacological pain and anxiety management. All efforts to reach the goal that no child should experience pain or anxiety when exposed to medical procedures. as calcaneovalgus or varus, idiopathic toe-walking, physiological anatomical foot anomalies, in-toing or out-toing or other referral diagnoses not substantiated at the consultation. Only, participants without chronic pain or recurrent pain were included in the current study.
Ninety-seven children were invited for participation in the outpatient clinic and 74 accepted. Twenty-three children and parents declined participation due to lack of time or logistics (17), parental dislike of virtual media (1) or with no explanation (5) (Figure 1). Two malfunctions of the hardware resulted in exclusion of two children, which were replaced by two additional participants. Inclusion criteria were age between 6 and 14 years, normal cognitive, visual and auditory functions. Exclusion criteria were outpatient visit with invasive procedure (blood sample, cast removal and Kirschner wire removal), language difficulties, severe motion sickness, epilepsy, cognitive impairments impeding pain assessment based on selfreport, severe coagulopathies, infection in the anatomical region of the PPT measurements, intake of any pain/anxiety medication on the day of examination. The child and parents were screened for eligibility by verbal inquiry, for F I G U R E 1 CONSORT 2010 Flow Diagram based on the CONSORT 2010 statement: extension to randomized crossover trials. The four interventions were Control, 2D video, VR Video and VR Game randomized into 24 sequences. example 'Do you suffer from motion sickness?' If exclusion criteria were reported, the child was not included in the study. Other medications were not exclusion criteria. No changes in the clinical assessment or planning were made due to the present study.
Baseline demographics of the participants are given in Table 1 listed by 24 randomization groups. All included patients were covered by the Danish Patient Assurance, and no salary or other payment was offered to either patients or study personnel. The study complies with the Declaration of the World Medical Association and was approved by the local ethical committee (1-10-72-168-21), the Danish Data Protection Agency and the GDPR regulative. The present study was not a medical interventional RCT influencing the treatment of patients, and hence was not registered at Clini calTr ials.gov or similar registries.

| Randomized, controlled crossover design
A randomized crossover study design testing four interventions within each of the 72 participants was applied. Three participants were randomly allocated to the 24 possible sequences (4*3*2 = 24) by drawing sealed envelopes. This is a more advanced study design compared with the simple 2 × 2 crossover design, but with the same strength of allowing paired analysis of all four interventions ( Figure 1) (Dwan et al., 2019).
The sample size of n = 72 was based on a calculation of a paired analysis of normative data with the primary outcome with a mean population PPT of healthy children (183.1 kPa, SD: 90.7; Nikolajsen et al., 2011). The expected PPT difference between the control and the VR intervention was set to 75 kPa. With 90% power and 95% significance level (α = 0.05, β = 0.1), a sample size of 62 children would be required. However, a total sample size of 72 children was enrolled to ensure equal groups of three participants in each of the 24 sequences of the four interventions. Randomization into all 24 possible sequences allowed us to estimate a potential carry-over effect from one intervention to another and to calculate the true effect of the intervention.

| Interventions
The four interventions, each lasting 4 minutes were: 1. VR Game (Figure 2a): In the VR game, the person playing is situated in an animated immersive world sitting in a safari jeep, driving through a safari park. The aim of the game was to take photographs of the animals on both sides of the jeep. Points are given according to the quality of the picture taken.
If the entire animal is in the picture, if the animal is in motion and if several animals are in the picture, more points are given. Before the game starts, the hand and controller motion for lifting the camera, looking through the objective and taking a photograph is tested. The game allows 100 pictures, which was not reached. The participants were instructed in the rules of the game and encouraged to take many photographs. The VR game is a part of the SyncVR Relax & Distract application developed by SyncVR Medical DK ApS® and is CE-marked as a medical device.

| Study procedure
During all four interventions, a standardized setup was applied. The child was placed in a comfortable, rotating chair adjacent to its parents in a closed undisturbed room. The parents were instructed only to interact with the child if triggered by the child itself. The data were obtained in a standard examination room behind closed door with no external disturbances. To adjust for personal preferences, a safari theme was introduced for all interventions except for the control. During the 2D video, a tablet was placed on a table in front of the child and the child was instructed to leave the tablet untouched. During the VR video, controller was not necessary, so only the VR headset and headphones were applied ( Figure 2b). The child was told that animals could be all around them and was reminded of the rotating seat. For both the VR video and VR game, the investigators could track the child's visual experience  through the spectator function on the tablet. During the VR game, the handheld controller was placed in the child's dominant hand. Controller use during the game was introduced before the VR headset and headphones was put on the child (Figure 2b). Equipment specifications are a VR headset (type: Pico G2 4K Enterprise), Spectating tablet (type: Samsung A7) and Noise-cancelling headphones (type: MPOW) by SyncVR Medical DK ApS®, Denmark (CVR 42404276).

| Outcome measures
Immediately before each intervention and after exposure to the intervention for 4 minutes, pressure pain threshold (PPT), heart rate and modified Yale Preoperative Anxiety Scale (mYPAS) were assessed (Figure 1). PPT was the primary outcome and mYPAS, heart rate, NRS and questionnaire were secondary outcomes.
The primary outcome, PPT, was obtained using a handheld algometer (Algometer®, Somedic Sales, Hörby, Sweden) with an application rate of 20 kPa/s and is validated for paediatric PPT assessment (Figure 3). PPT is defined as the point when an increasing pressure (by the algometer) reaches a point where the perception of pressure begins to be felt as pain. Measuring the pain threshold rather than the pain level or pain intensity may provide us with a deeper understanding, how VR modulates the pain response in children. The children were instructed to say 'stop' during the PPT assessment, when the sensation of pressure changed to a sensation of pain. The maximum applied pressure was subsequently displayed on the algometer (Figure 3). To decrease the child's fear of the examination, a test PPT assessment was made. The average of two measurements on the thenar of the hand with the probe perpendicular to the first metacarpal bone was calculated. PPT assessment in children younger than 6 years old has not been validated, and adolescents are assumed to have adult pain thresholds. For all children, a standardized verbal information of the PPT assessments was given before and during the study (Nikolajsen et al., 2011). The same investigator made all measurements (LKP).
Anxiety level was assessed using the validated Danish version of the modified Yale Preoperative Anxiety Scale F I G U R E 2 During four interventions, each child was exposed to in a randomized setup in this within-subject study. The present figure depicts four different children during one of the four interventions they each underwent. (a) VR Game: Screenshots of the safari game, in which the child takes photographs of animals in a 360-degree immersive world. (b) VR Video: the child looks at animals during a 360-degree immersive safari (screenshots), but does not use a controller. (c) 2D Video: the child watches the same safari video on a tablet without headphones. (d) Control-small-talk: A child during the control period small-talking with one of the investigators. The same investigator administered all four interventions (LKP). Informed consent for recognizable photographs was sought and granted. (mYPAS), which is validated for paediatric situational anxiety during medical situations (Kain et al., 1997;Skovby et al., 2014). The mYPAS consists of five categories (activity, emotional expressivity, state of arousal, vocalization and use of parents), which define 22 specific behaviours indicating anxiety in a child. Each category has different numbers of items (four or six), hence partial weights are used for a total score ranging from 23.33 (lowest level of anxiety) to 100 (highest level of anxiety). The same investigator made all assessments (LYVF). Blinding of the investigator with regard to intervention was not possible, due to noticeable differences in the interventions. Heart rate was assessed by oximetry on the nondominant hand (WristOx 2 , Model 3150, Nonin), and the heart rate was read and recorded prior to each PPT assessment.
At baseline and after the examination, the children were asked to rate their level of pain using a Numerical Rating Scale (NRS), where 0 reflects no pain and 10 the maximal level of pain.
A non-validated questionnaire was verbally administered at baseline and end of study. The items of the non-validated questionnaires included previous experiences with VR, whether they saw themselves as 'gamers', estimated weekly hours of digital gaming and tendency for motion sickness (Figure 1). Data will be published independently.

| Statistical analyses
Data are shown as either mean (95% confidence interval (CI)) or median (range) and differences are compared using analysis of variance (ANOVA) with the Holm-Sidak multiple comparison test limiting type 1 errors (McHugh, 2011). The repeated measures multivariate ANOVA in the present study is used with PPT and mYPAS as the repeated measures dependent variables. All statistical analysis were performed using Prism or STATA version 16 with a significance level at 0.05.

| RESULTS
Complete data of 72 children (33 girls and 39 boys) were obtained for analysis. The median age was 10.5 years (range 6-14), which followed a normal distribution as examined by Q-Q plots. There were no significant differences in age or gender in the randomization groups (Table 1). PPT assessment was well-tolerated by all children.
VR game and VR video significantly increased PPT with a mean difference of 136 kPa (CI 112; 161), p < 0.0001 and 122 kPa (CI 91-153), p < 0.0001 respectively. Nonimmersive 2D video had a minor, but statistically significant effect on PPT: 47 kPa (CI 24-69), p = 0.0002, while the control (small talk) did not statistically significantly affect the mean PPT 17 kPa (CI -2; 35), p = 0.0746 (Figure 4). The carry-over effect between interventions was minimal and was estimated by comparing the mean PPT difference in the randomized sequence, that is PPT diff (control, 1st) vs.

F I G U R E 4
Main results. Mean before and after values of PPT and mYPAS in the four interventions with differences and vertical lines indicating 95% CI. Mean differences (95% CI) are given above. Note, the four interventions are depicted ordered by effect size. However, the sequence in the trial was completely random for the 72 participants.
Anxiety levels improved during VR game (mYPAS diff: −7 points (−8 to −5), p < 0.0001) and VR video (mYPAS diff: −6 points (CI -7; −4), p < 0.0001). No difference in anxiety level was observed during the 2D video (mYPAS diff: −1 point (CI −2; 1), p = 0.5695) or the control (mYPAS diff: 0 points (CI −1; 1), p = 0.9140) (Figure 4). Regarding anxiety levels, a carry-over effect between the first intervention to the subsequent interventions was noted. In the second, third and fourth allocation, mYPAS data were similar to the total mYPAS scores, indicating that the allocation order only influenced the mYPAS in the first allocation and no considerable carry-over effect on mYPAS was seen. When analysing the heart rate (HR) of the participants during the four interventions, the VR game showed the highest increase, though not significant (HR diff (VR game): 1.5 bpm (CI 0.1; 2.9), p = 0.1196). First allocation data did not show significant heart rate changes. The level of pain during the entire study period did not change (NRS PRE = 1.0; NRS POST = 1.0; NRS DIFF = 0.0; 95% CI: −0.3;0.4; p = 0.6677) substantiating that no experimental pain was induced.
No correlations were found between background characteristics and the effect of VR on pressure pain threshold and anxiety.

| DISCUSSION AND CONCLUSIONS
In the present study, immersive VR (Video and Game) markedly increased the pain threshold and decreased anxiety levels in an experimental well-controlled setting. 2D Video had a minute effect on pain threshold but did not relief anxiety. Small talk (control) had no statistically significant effect on these outcome measures. The positive effect of immersive VR on pain threshold and anxiety level was highly significant for VR Video and VR Game. VR Video might be better for invasive medical procedures than VR Game in which sudden movements of both upper extremities were observed. This makes VR Video more suited for application during medical interventions, for example intravenous catheters but also casting of fractures or removal of percutaneous Kirschner wires in the outpatient clinic (passive distraction). Opposite, VR Game showed a greater absolute change in both pain threshold and anxiety compared with VR Video, possibly reflecting both a modulation of pain perception and distraction due to the interactivity of the game (active distraction). Both VR modalities are immersive proving the hypothesis that immersive solutions are more effective compared with the non-immersive 2D video.
Previous studies investigating the procedural effect of VR on children have found a beneficial effect on childassessed pain level. However, compared with standard of care and without well-controlled designs. The present study focused on the threshold for pain and anxiety level in children under experimental settings and under wellcontrolled conditions. In addition, the study shows a positive and individual effect of VR on both pain and anxiety in children and validates VR as a non-pharmacological pain and anxiety management tool. Opposite to previous studies, present study uses extensive control conditions (2D Video and small talk/control) substantiating that the effect of VR on PPT and anxiety is independent and present when only changing one constant at a time. Studies have described abnormal PPTs in children with diseases; however, few studies have investigated whether the PPT can be modulated in children under experimental condition and excluding adolescent and adult populations (King et al., 2017;Riquelme et al., 2016;Scheper et al., 2017).
In conformity with Colloca et al., the autonomic, affective and non-pharmacological pain and anxiety influence of immersive VR translate well into a paediatric population (Colloca et al., 2020). Studies in paediatric medical procedures indicate a positive effect on pain and anxiety as reported in a meta-analysis by Eijlers et al (Eijlers et al., 2019); however, only limited data were available for anxiety assessments. Jivraj et al (Jivraj et al., 2020) used VR during cast removal in children and found a significant positive effect on both intra-and post-procedural anxiety assessed by the objective CEMS (Children's Emotional Manifestations Scale). The complex interaction between pain, anxiety, catastrophizing and other individual factors of children are still an enigma and most paediatric VR studies focus on child-assessed procedural pain. Walther-Larsen et al. (Walther-Larsen et al., 2019) found no difference in pain scores during venous cannulation in children randomized to control or VR use. However, the pain level was child-assessed and recorded as recall pain 15 min postprocedure. Gomez-Polo et al (Gomez-Polo et al., 2021) focus on anxiety and behaviour in children requiring several dental treatments. They report that the use of VR significantly reduces child-assessed procedural anxiety and increases positive behaviour, though only subjective assessment methods were used and states that a strong correlation between anxiety and the child's perception of pain exists. This underlines the importance of present study investigating this relationship by use of objective and validated assessment methods. Colloca et al. (Colloca et al., 2020) have found a significant increase in heat-pain threshold and tolerance in the immersive VR modalities in a within-subject study design on adults using both immersive VR (ocean and opera themes) and 2D control. It concludes that immersive VR in adults increases the pain tolerance threshold, but also led to an improvement of mood, situational anxiety and pain unpleasantness. Present study corroborates these findings by showing that PPT increases and anxiety decreases in a paediatric population during immersive VR. Furthermore, our study also includes a small-talk control period strengthening the design in addition to avoiding pain tolerance assessments in children. This is not only due to both ethical considerations but also due to pain thresholds being less affected by physiological factors such as willingness to tolerate the pain (Nikolajsen et al., 2011).
The crossover design strengthens the study by allowing paired comparison of data and ensures transparency according to the extension to the CONSORT statement (Dwan et al., 2019). When applying an advanced crossover study, heterogeneities of the population are inherently controlled for and the statistical power is higher compared with a parallel group design (Arhold & Betensky, 2018). Commonly used outcomes in VR studies often fail to accommodate the heterogeneous nature of both pain level, pain threshold, situational anxiety and catastrophizing. Dwan et al. (Dwan et al., 2019) define carry-over effect as when the effect of the first intervention persists into the second (third or fourth) period and define a period effect as when the outcome of interest changes with time irrespective of treatment effect. Since this study mostly adheres to an advanced crossover design with four interventions, a statistical omnibus test for separability to assess the treatment effects independently of carry-over effects or linear regression, mixed models with special software (Wellek & Blettner, 2012), were not feasible. We accounted for both time effects and carryover effect by design with 24 randomized sequences of the four interventions. In addition, the primary outcome, PPT, showed similar start values for all four interventions when observing the total value compared with the first, second, third and fourth allocation data indicating that the effect of the previous intervention do not persist into the following intervention, rejecting the presence of a significant carry-over effect.
Regarding the anxiety score, mYPAS, all children, regardless of which intervention they were randomized to receive first, had a higher anxiety level before the first intervention, which was not present for the second, third or fourth allocation. This is not interpreted as carry-over effect, but rather as sign of a higher level of situational anxiety for all children before the start of the study protocol, potentially due to the experimental study setting the child was put in despite attempts to create a tranquil environment. However, this situational anxiety was resolved quickly and no dropouts during the study interventions was seen. The entire study protocol took approximately 25 minutes, in which timeframe neither the threshold for pain nor the level of anxiety is expected to significantly change without intervention ruling out a potential time effect. However, as stated by Wellek et al. (Wellek & Blettner, 2012), a simple example of a period effect may be familiarization with the study situation. In traditional crossover studies, a washout period between treatment periods is needed to allow the treatment to wear off (i.e. medications to be cleared from the body) before the next treatment period. As stated by Chan et al (Chan et al., 2018), it seems unlikely that VR would have a persistent effect. In the present study, the effect on PPT by the VR and control interventions returned to the start-out point in the short period between interventions indicating an adequate washout period. This complies with Dwan et al. (Dwan et al., 2019), who states that crossover trials are most appropriate when the intervention effect is reversible and short lived.
No significant limitations, risks or side effects are present. External validity is good for children (6-14 years) and results may be generalizable to procedural, acute and chronic pain. Blinding of the assessor of PPT and mYPAS was impossible. In addition, a safari theme with different animals was chosen for all interventions, presuming a general interest of the topic in the present age group. On the contrary, we did encounter a small ceiling effect regarding age with a median age of 10.5 years (range 6-14 years). PPT assessment in children younger than 6 years is not validated and patients older than 14 years is assumed to have adult pain perception and response to VR, hence present age group are required to assess paediatric pain perception and response to VR. A low level of nausea and discomfort was reported. Other studies have shown a risk of nausea related to VR.
In conclusion, immersive VR can modulate a child's threshold for pain and anxiety level. Results are significant. The threshold for pain increases and anxiety lowers during immersive and interactive VR use in children assessed under experimental and well-controlled conditions. Hence, VR may play a role in the modulation of pain thresholds and is beneficial for non-pharmacological pain and anxiety management in children. An understanding of the relationship between pain, anxiety and personal factors (personality/behaviour, values, prior experiences and culture) is an important perspective in the future research of VR in paediatric non-pharmacological pain and anxiety management as well as research into the effect of personalized VR content. All efforts to reach the common goal that no child should experience pain or anxiety when exposed to medical procedures.