We performed a prospective, mixed methods, nonrandomized feasibility study of patients with critical illness using immersive VR headsets. The study was conducted in 2 ICUs (a 25-bed medical ICU and a 16-bed cancer-specialty ICU that cares for both medical and surgical critically ill patients with cancer) at a single institution in Salt Lake City, Utah, United States. Reporting follows the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidance for observational research (Multimedia Appendix 1)[19].

The study was approved by the University of Utah Institutional Review Board (00170975). Patients were individually consented with a waiver of the requirement for written documentation of consent. Patients were not offered compensation for participation. Deidentified data were recorded in case report forms that were later digitized for analysis.

Patients who were potentially eligible for study inclusion were identified by attending physicians after daily rounds on days when study staff were available for enrollment. To be included, patients needed to be 18 years or older, admitted to the ICU, free from delirium [20] (as assessed by their providers, nurses, and able to pass an additional attention screen), and able to consent on their own behalf. Exclusion criteria included severe visual or auditory impairments (eg, legal blindness or deafness), isolation precautions for infection, recent condition that could be potentially exacerbated by VR (eg, seizure, uncontrolled nausea, traumatic brain injury, history of psychosis, or admission for a mental health crisis), or other craniofacial injury prohibiting headset use. Additionally, patients under current use of an orofacial mask to deliver positive airway pressure ventilation were excluded as the mask precluded headset use; however, intubated patients, patients receiving high-flow nasal cannula, and patients receiving oxygen via a regular face mask were eligible to participate.

After patients were identified as potential candidates, nursing staff were approached to identify any conflicting patient-care tasks (eg, physical therapy or travel for diagnostic testing or procedures). Patients were then approached about whether they were interested in trying an experimental immersive VR therapy. Patient demographics, reasons for admission, and comorbidities were assessed by chart review of clinical notes. All patients who were identified as potential research participants by their attending physicians were screened for inclusion and approached if eligible.

Participating patients were presented with 3 visual analog scales (1-10) asking them to rate their current overall well-being, level of anxiety, and level of pain. Preintervention heart rate signals were recorded using a BIOPAC MP160 system for 5 minutes prior to VR initiation (Figure 1).

Patients were offered 1 of 3 commercially available VR experiences delivered by a Meta Quest Pro headset (Meta): an urban travel experience (YouTube VR; Google LLC), a nature experience (Nature Treks VR; GreenerGames), or a synthetic landscape experience (TRIPP). In all 3 scenarios, the experiences involved passive exploration of the environment and did not involve use of the hand controllers (Figure 2). Patients planned to use the VR headset for at least 5 minutes, with an option to continue for up to 15 minutes if desired. Physiologic recording was continued for 5 minutes after VR use was complete.

After completing the VR experience, 5 minutes of postintervention vital signs recording and visual analog scale assessments of well-being, anxiety, and pain were administered. These scales were modeled on the Visual Analog Mood Scale [21], Numeric Visual Analog Anxiety Scale [22], and Verbal Numerical Rating Scale for pain [23], respectively (Multimedia Appendix 2).

Lastly, a semistructured qualitative interview was conducted to elicit feedback. A moderator guide (Multimedia Appendix 3) consisting of open-ended questions aimed at understanding the patient’s overall experience using the VR headset was administered by a trained researcher. The responses for each participant were transcribed concurrently by 2 researchers in individual Microsoft Word documents. The participant responses were deidentified to maintain patient confidentiality.

The qualitative data from the semistructured interviews were analyzed using thematic analysis, a 6-phase method for organizing, identifying, and summarizing patterns and themes [24]. In the first phase, known as data familiarization, researchers carefully reviewed, extracted, and organized the textual data in Microsoft Excel. The second phase involved generating initial codes for the data using a grounded theory approach, which explores participants’ attitudes, beliefs, norms, and processes to develop hypotheses from the data rather than testing preexisting hypotheses [25]. In the third phase, the assigned codes were aggregated to uncover underlying patterns, themes, and subthemes. Phases 4 and 5 focused on reviewing, refining, and defining these themes, subthemes, and codes. Finally, in phase 6, the results of the thematic analysis were summarized and reported.

A sample size of 20 participating patients was targeted in accordance with rule-of-thumb guidance for a pilot study emphasizing qualitative assessment and protocol feasibility [26]. Descriptive statistics of participants and nonparticipants in VR were used to describe the population of interest. For physiologic signals, the first 2 minutes of pre-VR recording were compared to the first 2 minutes after the VR experience. Kubios automatic beat detection software was used to preprocess the heart rate data [27]. Heart rate variability was characterized by the square root of the Baevsky stress index (SI) [28]. Pre-post comparisons of vital sign data and mood assessments were performed using paired 2-sided t tests. Statistical analyses were performed using Stata version 18 (StataCorp) and the code is openly available on GitHub [29].

In this feasibility study of commercially available immersive VR use among critically ill patients, we found that most patients did not have prior experience with VR technology but had high levels of interest and acceptance of the technology. Participants commented on VR’s potential to alleviate cognitive and emotional symptoms. Changes in heart rate variability were consistent with increased relaxation. Only minor technical challenges and no severe adverse effects were noted.

Prior work has explored the potential for immersive VR for several ICU use cases [1]. However, this previous work reported widely variable rates of uptake and hinted at a “digital divide,” where older patients may be less interested in new technology [30]. In contrast, we found high levels of participation among patients of all ages. Thus, VR might be considered for study even in settings that frequently care for older adults. We found that patients who had previously used VR were less likely to participate in our study. However, this might indicate that the novelty of the experience drove participation in this feasibility study. Interest may improve in prior users if VR were offered as a validated treatment. Only a minority of patients who did not participate gave reasons related to the VR itself.

After use, several participants commented on the potential of the therapy to address anxiety and foster relaxation or calmness. This was corroborated by the improvement in vital sign correlates of relaxation, such as improved heart rate variability. Some [6], but not all [18], prior studies of VR have shown consistent changes in vital signs.

Participants also highlighted the potential of VR to distract from pain, which is consistent with current guidance from the Society of Critical Care Medicine that recommends consideration of “cybertherapy [VR]” for this purpose [31]. The high usability scores and low rate of technical challenges with “off-the-shelf” commercially available options suggests that extensive customization is not a prerequisite for use. Furthermore, we did not observe claustrophobia, nausea, or “cybersickness” in any patients. Cybersickness may be less common with more modern VR headset technology that minimizes latency [32] and discordance between virtual and actual head positioning [33], which could explain why this was not encountered in our study. In contrast to prior work suggesting that nature scenes may maximize relaxation [15], travel was the most frequent VR experience choice among our participants. The potential for VR to enable “escape” from the ICU was also frequently mentioned in qualitative interviews.

Strengths of this study include the assessment of both quantitative and qualitative dimensions of the VR experience, which both support the feasibility and potential utility of using VR in this group. In addition, there were few exclusion criteria and thus the sample of patients is expected to broadly represent characteristics of critically ill patients without delirium. Lastly, we demonstrated and reported an off-the-shelf and replicable experimental setup that investigators can use as a starting point for future studies.

However, several limitations also deserve mention. First, nonrepresentative sampling of research participants may have occurred due to only enrolling patients during select times when study staff were available. Despite exhaustive screening of patients who were identified as potential candidates during these times, we cannot fully describe what characteristics may have influenced which patients were judged to be potential candidates for the study by attending physicians. Second, there was no control group. Time trends and the influence of conversing with study staff may also contribute to changes in symptoms and vital signs, although we attempted to stabilize physiologic trends with a 5-minute run-in period and structured interview guides were used to focus the conversation. Third, prior work suggested that VR’s effectiveness correlates with the degree of immersion [34], which we did not directly assess. However, we chose not to attempt to eliminate distractions to better emulate the conditions and degree of immersion expected in actual use. Fourth, our quantitative outcome scales are slightly modified from previously validated work, and thus the reported effect sizes should not be compared to other interventions or established minimally important differences. Lastly, the uptake of VR may differ when it is offered as a treatment for specific conditions as opposed to offering the opportunity to help assess feasibility, particularly among patients who have previously used the technology.

This work has several important implications for the study of VR in the ICU. We used commercially available technology and found that the sessions were acceptable to patients. This suggests that customization of either software or hardware is not necessarily required for some VR use cases. Furthermore, we found high interest among critically ill patients. Lastly, we encountered minimal disruptions to patient care or study protocols, with 95% of the patients completing the experience and 90% completing all study assessments. This suggests that protocols to study the impact of VR can be integrated into usual ICU care with little impingement on clinical workflows. Evaluation of the usability in patients excluded from the current work, such as those with mild delirium or more severe emotional symptoms, could help establish the potential of studying immersive VR in additional high-risk populations.

We found that a relatively short session of off-the-shelf immersive VR is acceptable to critically ill patients; resulted in improved pain, anxiety, and overall mood scores; and did not result in side effects or present major technical challenges. This work suggests that studies on the effect of VR on patient-relevant outcomes in the context of critical illness are feasible. Investigators can consider VR study protocols that do not involve substantial technology customization or large changes to patient care workflows.