The study was conducted following the current version of the Declaration of Helsinki. All subjects participating in the study were recruited through the University of Bern, and the study was approved by the Ethics Committee of the Canton of Bern, Switzerland (KEK-Nr. 2017-02195). All subjects signed written informed consent before inclusion. Furthermore, specific consent was obtained to publish identifying information and images in an online open-access publication. The main exclusion criteria were auditory and visual impairments and being below the age of 18.

The study was conducted in a one-bed cubicle in the ICU of the University Hospital Bern. To stimulate the subjects, either a head-mounted display (HTC Vive, High Tech Computer Corporation, Taoyuan, Taiwan) or a classical ICU TV screen (MediTec TV, Bewatec, Telgte, Germany) was used. The head-mounted-display had a resolution of 1080 × 1200 pixels per eye, a field of view of 110 degrees, and a refresh rate of 90 hertz (Hz), whereas the ICU TV screen had an 11.5-inch display and an aspect ratio of 4:3.

Subjects were stimulated two times with the head-mounted display (ie, virtual nature and urban VR stimulation, Figure 1) and once with the gold standard (control condition), a classical ICU TV screen (movie), with each interaction lasting 10 minutes. The nature VR stimulation consisted of a large island surrounded by water, with green hill areas, beaches, forests, plantations, and several animals such as elephants, giraffes, dolphins, birds, and butterflies. The designed urban VR stimulation consisted of a busy downtown, followed by a more relaxed old town. Pedestrians, cars, birds, aircrafts, and a construction site were among the many objects that could be found in the environment. In both cases, subjects were walking on a predefined path. In the control condition, a neutral documentary movie about intuition was presented (ie, combination of nature and urban scenes).

Physiological parameters (ie, noninvasive arterial blood pressure, heart rate, and respiratory rate) were monitored by the in-house monitor system of the University Hospital (Carescape Monitor B650, GE Healthcare, Little Chalfont, United Kingdom). Respiration and heartrate were measured at a frequency of 240 Hz, whereas blood pressure was measured every second minute. To simulate a real ICU scenario, the physiological monitor system constantly produced alarm sounds (eg, simulated arrhythmia). In addition, a noise meter was set up next to the subject to monitor and guarantee the overall noise level in the room was above 50 decibels. The bed was tilted up so that the subjects were in an upright position. The system was approved by the University Hospital’s medical-technical department, except for use in critically ill patients and subjects (ie, hygiene and medical eligibility approved). To evaluate the perceived restorativeness of each stimulation, the PRS-11[23] (measured by a seven-point scale from not at all to completely) and for discomfort the SSQ [24] (measured by a four-point scale, from none to severe) were used. Both score scales were normalized on a score scale between zero and one.

To account for the influence of the order of the stimulations, the subjects were randomly assigned in a pseudo cross-validation to one of the three groups with different stimulation sequences. First, subjects were instructed and prepared (ie, lying on the bed, Figure 2) for the experiment, followed by a recovery phase of 10 minutes to stabilize physiological parameters at subjects’ baseline. Next, subjects were assigned randomly to one of the three groups and stimulated with the three environments with a break of approximately 10 minutes in between. After each stimulation, they were asked to fill out a questionnaire of the perceived restorativeness and discomfort. Before the stimulation, subjects filled out a questionnaire about demographics. All instructions were given prior to the stimulation, and participants could move their head to explore the environment while seated. For each participant, the experiment lasted approximately 80 minutes.

To analyze the calming and relaxing effect (ie, negative time effect is defined as the reduction of physiological parameters during the course of the stimulation) of an environment, the correlation between time and the physiological measurement was calculated and tested on their significance using a one-tailed t test against zero. To analyze the differences in questionnaire ratings between the stimulations, one-way analysis of variance (ANOVA) was used, followed by a post hoc analysis using paired t tests and Bonferroni correction for P value adjustment. The analysis of the physiological parameters and the questionnaires was conducted by using R for statistics (The R Foundation, Vienna, Austria) and MATLAB 2018b (MathWorks, Natick, USA).

A video of the two VR stimulations and the movie about intuition can be found on our website [25].

All relevant codes and data supporting the findings are presented in this paper. Further code and data of the study are available upon request.

A total of 45 (22 male and 23 female) adults between the ages of 22 and 87 (mean 59; SD 16) participated in the study. Overall, 29 participants grew up in the countryside and 16 in an urban environment, and they were all in good health. Analysis using ANOVA showed a significant difference between the three stimulation environments (F₂=15.86; P<.001). The nature VR stimulation had the highest perceived restoration (mean 0.773; SD 0.142) and was closest to the maximum of the score scale (with none=0 and high=1), followed by urban VR (mean 0.65715; SD 0.187) and ICU TV stimulation (mean 0.5854; SD 0.136). A post hoc analysis showed a significant difference between nature VR and both urban VR (P=.002) and ICU TV (P<.001), but not between urban VR and ICU TV stimulation (P=.12).

In all three environments, nausea, oculomotor problems, and disorientation were close to the minimum of the score scale (Figure 3). The only significant difference in discomfort between the environments was found in disorientation (F₂=3.67; P=.03).

At the beginning of the recovery phase, respiration rate was an average of 19.35 Breaths/minute (SD 5.08), heart rate was an average of 69.88 Beats/minute (SD 11.03), and the mean blood pressure was 94.3 mmHg (SD 11.81). As shown in Figure 4, nature and urban VR stimulation were associated with a reduction in respiration rate, whereas during ICU TV stimulation respiration rate increased. The negative correlation of respiration rate and time (negative time effect) was highest and only significant in nature VR stimulation (95% CI –Inf to –0.024; t ₄₂=−2.13; P=.02), but neither in urban VR stimulation (95% CI –Inf to 0.052; t ₄₀=−0.69; P=.25) nor in ICU TV stimulation (95% CI –Inf to 0.132; t ₄₁=1.31; P=.90). With respect to heart rate, only nature VR decreased during the course of the stimulation, whereas in urban VR and ICU TV stimulation heart rate remained more or less constant. However, there was a significant negative time effect in heart rate in nature VR stimulation (95% CI –Inf to –0.177, t₂=−7.53; P<.001), in urban VR stimulation (95% CI –Inf to 0.161; t ₄₀=−7.35; P<.001) and in ICU TV stimulation (95% CI –Inf to 0.141; t ₃₈=1.31; P<.001). Furthermore, the mean blood pressure did not increase or decrease during any of the three stimulations.

In this study, we investigated the subjective restorative and relaxing effect of three different stimulation methods (dynamic, virtual, nature and urban environments presented inside a head-mounted display and a neutral video on an ICU TV screen) in healthy subjects. In line with our hypothesis, there was a significantly higher restorative and relaxing effect in the virtual nature environment compared to the virtual urban environment and the ICU TV screen. Additionally, in all cases, the three different methods evoked none to minor negative reactions.

The main finding in line with the attention restoration theory was that the highest perceived restorative effect was found in virtual nature stimulation, followed by the virtual urban stimulation and the movie presented on the ICU TV screen [19,26]. Furthermore, the three different stimulation methods showed none to minor negative side effects, like nausea, oculomotor problems, or disorientation. The slightly higher discomfort in the virtual urban environment was possibly due to unexpected 90 degree turns (ie, when walking around a building corner), whereas in the virtual nature environment no turns were present.

The second main finding was that the nature VR stimulation with head-mounted display revealed the highest and only significant reduction in respiration rate. In case of heart rate, only the nature VR stimulation showed a relaxing effect, whereas in mean blood pressure no effect was found. Overall, physiological parameters showed that nature VR stimulation by using head-mounted displays revealed the greatest effect on the parasympathetic nervous system and thus was in line with the stress reduction theory suggestion that natural environments are the most restorative environment [15,16,27].

The effect on the parasympathetic nervous system may not result just from the natural environment, but may also be influenced by the head-mounted display and the noise-cancelling headphones, which helped to reduce the sensory overload (eg, alarms, noise) and deprivation (eg, missing daylight) in the ICU. One may argue that this effect should also be present in the urban environment, but there is a significant difference between the stimuli of these environments. While the sound of the natural environment mostly consisted of crashing waves and animal sounds, the soundscape of the urban scene was overloaded with typical city sounds [28]. These sounds may be as irrelevant and annoying as the noises in the ICU and therefore also cause sensory overload. Furthermore, due to memories and past experiences in the different environments, not all participants will respond the same to the different stimuli. Therefore, if in the past participants had a bad experience with certain stimuli inside a natural (eg, animals) or an urban environment (eg, car accident) it may evoke bad memories and thus influence physiological parameters negatively, with the opposite being true for positive experiences.

One of the limitations of this study is that all subjects were healthy subjects, so it is unclear whether the results could be transferred to critically ill patients. Another limitation was that cultural, living environment, and educational differences between the participants were recorded as a covariate but were not considered. Subjects from a rural origin may react differently to the stimulation and have different preferences than a subject who was born and grew up in a city. Furthermore, the different movement patterns inside the environments might have an influence on the perceived restorativeness.

Overall, three environments that were different in technical setup and design were tested safely and successfully. We showed that VR stimulation by using a head-mounted display in combination with a dynamic, virtual, natural environment had the highest restorative effect on the physiological and psychological state of a subject. Furthermore, it was shown that the effect was not simply due to the isolation from the stressful environment but actually due to the composition of the natural environment. Therefore, the method of VR nature stimulation by using a head-mounted display may have potential as an early-intervention method directly used in the ICU to reduce sensory overload and deprivation and thus prevent neurocognitive late effects.