This prospective randomized controlled crossover study took place in the gastrointestinal (GI) endoscopy unit’s preprocedural areas at the University of California, San Diego Health’s Jacobs Medical Center and Hillcrest Medical Center between September 2024 and December 2024. Three attending anesthesiologists (authors BF, MT, and JLT) approached patients scheduled for endoscopic GI procedures and obtained informed consent for monitored anesthesia care 2 times, once without wearing a headset and once while wearing the Apple Vision Pro. A survey was administered to the patient after each encounter to evaluate their perception of their anesthesiologist’s compassion and empathy, connectedness, communication, and eye contact.

This study used a convenience sample of patients scheduled for nonemergent endoscopic GI procedures at the University of California, San Diego Health. Inclusion criteria required that patients be at least 18 years of age and able to converse, read, and write in English. Exclusion criteria were patients belonging to the following vulnerable populations: children under 18 years of age, pregnant individuals, incarcerated individuals, and those with an unhoused status. Also excluded were patients undergoing an emergent procedure or who, based on the judgment of the anesthesiologist, lacked capacity to provide informed consent to participate in the study. Participants who completed the study protocol were compensated for their time and effort with a US $5 gift card.

Each participant completed 2 consecutive informed consent encounters: 1 in which the physician did not wear the Apple Vision Pro headset and 1 in which the physician wore the headset. Participants were randomly assigned to 1 of 2 encounter sequences (with the headset first or without the headset first). In both sequences, an anesthesiologist conducted a standard preprocedure evaluation including medical history, physical examination, and informed consent discussion for monitored anesthesia care. Every participant then underwent a repeat informed consent discussion under the alternate condition. Both encounters were conducted sequentially, with only a 30-second interval between them; the physician briefly exited and then re-entered the patient area to either put on the headset or remove it. When the Apple Vision Pro headset was used first, it was removed during the physical examination section to allow for an adequate clinical assessment. Although sequence order was randomized to minimize order effects, sequence assignment was not recorded for individual participants. Participants were informed before the encounters that the study aimed to assess patient perceptions of physician communication with and without the use of an AR or VR headset. They were not told that the headset was expected to improve clinical care but were made aware that it is an established technology currently used in health care settings and that the study sought to understand their perspectives on its use.

Prior to patient encounters, the Apple Vision Pro headset was configured to display a realistic representation of the physician’s eyes using the device’s built-in face scanning technology (Figure 1). This feature was implemented to maintain a simulacrum of eye contact during patient interactions.

Nonidentifying demographic data were obtained from each participant, including their age, gender, and race and ethnicity, as well as basic questions regarding their prior experience with AR and VR, if any. Immediately following both encounters, patients completed the Sinclair Compassion Questionnaire (SCQ), a validated 15-item instrument created by Sinclair et al [28,29] that measures patient-perceived compassion in health care encounters and includes assessments of clinician attentiveness, empathy, and caring behaviors. Each item was rated on a 5-point Likert scale ranging from “strongly disagree” (1) to “strongly agree” (5). The SCQ is scored by calculating the mean score of all items, with higher scores indicating greater reported compassion. Patients completed the SCQ once for each encounter (headset and no headset). In addition, the patients completed a separate questionnaire (Multimedia Appendix 1) more specific to this study that assessed, from the patient perspective, patients’ and physicians’ ability to focus during each conversation, their satisfaction with eye contact made, how realistic EyeSight felt, and how concerned they were about their privacy during each encounter. They also responded to 5 questions assessing their general beliefs about technology in health care, whether AR or VR can improve health care delivery, and whether informed consent should be required from patients prior to using AR or VR during clinical encounters.

Survey responses were analyzed to compare patient perceptions between the with-headset and without-headset conditions. Due to the paired nature of the crossover design and the nonnormal distribution of the Likert-scale responses, the Wilcoxon signed-rank test was used to compare SCQ scores and individual survey items between conditions [30]. Effect sizes were calculated as Z/√N, where Z is the standardized test statistic and N is the number of paired observations, and interpreted according to the criteria by Cohen [31] (small: r=0.1; medium: r=0.3; large r=0.5). All statistical analyses were conducted using R (version 4.3.1; R Foundation for Statistical Computing). All P values were 2 tailed, and a P value of less than .05 was considered significant.

The study protocol was reviewed by the director of the Office of Institutional Review Board (IRB) Administration, IRB chair, or IRB chair’s designee at the University of California, San Diego, and was determined to be exempt from institutional review (810467) under Title 45 of the Code of Federal Regulations, part 46.104(d) [32].

This study included 60 participants with a mean age of 59 (SD 15) years, ranging from 18 to 91 years (Table 1). A total of 53.3% (n=32) of the participants were female. The racial and ethnic breakdown was 80% (n=48) White, 8.3% (n=5) Asian, 6.7% (n=4) Hispanic or Latino, and 5% (n=3) Black. Approximately 73% (n=44) of the participants had no prior experience with AR or VR, and only 16.7% (n=10) owned an AR or VR headset.

Participants reported consistently high ratings across all empathy and compassion measures for both with-headset and without-headset conditions; however, mean SCQ scores were statistically significantly higher for the without-headset condition (4.69, SD 0.48) when compared to the with-headset condition (4.42, SD 0.87) using the Wilcoxon signed-rank test (P=.003 [28,29]).

The difference between responses to custom survey questions in the headset and no-headset conditions are shown in Table 2. Participants reported significantly reduced ability to focus on their conversation with the anesthesiologist when they were wearing the Apple Vision Pro (P<.001), with a large effect size (r=0.614). Patients also perceived that their anesthesiologist was less focused while wearing the Apple Vision Pro (P<.001), with a medium effect size (r=0.454). The most pronounced difference was observed with regard to patients’ satisfaction with eye contact made with their physician, which decreased significantly when they wore the Apple Vision Pro (P<.001; large effect size: r=0.74). While statistically significant differences were observed between conditions, overall ratings across both conditions remained generally high.

Regarding the Apple Vision Pro’s EyeSight feature, 53.3% (32/60) of the patients agreed or strongly agreed that the representation of the physician’s eyes on the outward-facing display was realistic, whereas 35% (21/60) disagreed or strongly disagreed. One-quarter (15/60, 25%) of patients reported that they would be opposed to their physician wearing the headset if it were completely occlusive and did not allow for simulated eye contact (Figure 2).

Custom questions regarding the patients’ overall opinions on the role of technology in health care revealed high levels of technology acceptance (Figure 3). In total, 85% (51/60) of the patients agreed or strongly agreed that technology can make a physician’s job easier, with 76.7% (46/60) agreeing that AR or VR can make a physician’s job easier during patient encounters. A total of 18.3% (11/60) of the patients were opposed or strongly opposed to their physician wearing an AR or VR headset at their next encounter. When the question was slightly changed to ask whether they would still be opposed if their physician told them they preferred to use it, 13.3% (8/60) of the patients were opposed.

Just over half (31/60, 51.7%) of participants believed that physicians should be required to obtain consent from patients before wearing an AR or VR headset during an encounter. Patients reported slightly increased privacy concerns during the headset encounter compared to no headset (P=.002; r=0.42); however, this was primarily caused by an increased number of patients responding neutrally (3 out of 5) to the following statement: “I was concerned about my privacy.” One patient in the headset condition agreed with the statement, and no patients strongly agreed.

We found that the use of an AR or VR headset negatively impacted perceptions of provider empathy and communication. While participants in both conditions reported relatively high SCQ ratings, the with-headset encounters received statistically significantly lower SCQ scores than the without-headset encounters, suggesting that the technology may create subtle barriers to physician-patient connection. We hypothesize that this decline in perceived compassion could stem from reduced eye contact, altered nonverbal communication, or a sense of detachment caused by the headset’s presence.

EyeSight received mixed reviews from participants in this study as just over half of patients found EyeSight to be realistic (32/60, 53%) and felt satisfied with their ability to maintain eye contact with their physician (33/60, 53%). Although we did not evaluate a fully occlusive headset in our study, 25% (15/60) of the participants reported that they would be opposed to their physician wearing one. Our findings are relatively consistent with those of Vergari et al [33], who conducted a pilot study showing that interactions with the Apple Vision Pro were perceived as more engaging and comprehensible compared to those using the Meta Quest 3, a fully occlusive headset, but less than face-to-face interactions.

The findings of the technology acceptance questionnaire suggest that patients remain open to the use of technology in health care workflows, with most patients recognizing its potential to enhance physician efficiency and health care delivery. Specifically, over three-quarters of the participants agreed that technology in general can improve health care delivery (51/60, 85%) and that AR and VR specifically could enhance physician capabilities (46/60, 77%).

Despite this overall acceptance, some reservations remain regarding the direct implementation of AR and VR headsets in patient encounters. A subset of patients opposed physician use of the headset, and over half (31/60, 51.7%) of patients believed that physicians should obtain consent before wearing the headset during an appointment, highlighting the importance of maintaining patient autonomy in technology integration. Notably, privacy concerns were minimal, with only 1.7% (1/60) of the patients expressing concern while the physician used the Apple Vision Pro. These findings underscore the need for clear communication between physicians and patients regarding the purpose and benefits of AR and VR technology while also ensuring that workflows integrating this technology incorporate patient preferences and privacy-preserving frameworks.

This study offers several insights into how AR technology may be more effectively integrated into clinical encounters. As 16.7% (10/60) of the patients would be opposed to their physician wearing the Apple Vision Pro at their next visit and 25% (15/60) would be opposed to their physician wearing a fully occlusive headset, the simulated eye contact provided by the EyeSight feature may be slightly preferable to a fully occlusive headset; however, it likely does not fully preserve the benefits of natural eye contact [33]. Finally, the justification for physician use of these technologies, including measured or perceived benefits, and an acknowledgment of the potential for friction in the patient-physician relationship should be integrated into larger discussions of informed consent. Such transparency may reduce patient confusion or discomfort, allowing the technology to be a supportive element of care rather than a distraction.

While the Apple Vision Pro certainly holds potential to enhance physician workflows and patient care, we believe that its success may depend on thoughtful design, clinician preparation, and open communication with patients.

This was a single-center study at an academic institution, which may limit the generalizability of these results. Participant demographics were skewed relative to the local patient population (eg, 48/60, 80% of the participants identified as White). We studied a convenience sample of patients undergoing endoscopic GI procedures due to the high turnover of patients in one day and the relatively lower acuity of cases; however, this likely skewed the age range of our participants (53/60, 88.3% were over 40 years of age, and 30/60, 50% were over 60 years of age). Generalizability is also limited by our study evaluating brief preprocedural encounters; these results may not be generalizable to longer patient encounters, such as in a primary care clinic. The short interval between the headset and no-headset interactions may have introduced a carryover effect into our data as there was no washout period. Finally, the novelty effect associated with new technologies warrants further longitudinal studies to assess whether reactions may normalize over time as adoption increases.

Future research should build on this study by incorporating a larger and more diverse patient population to better understand how patients of different ages and from different backgrounds perceive the use of AR and VR in health care. Comparing additional headset designs—including AR glasses (eg, Meta Orion), an AR headset with transparent lenses (eg, Microsoft HoloLens), or a fully occlusive VR headset (eg, Meta Quest)—could help determine which type of design best balances physician needs and patient comfort with or trust in their physician. Additionally, further investigation is needed to develop strategies that minimize perceived reductions in physician attention and empathy. Future studies should also more deeply assess AR and VR’s influence on the physician-patient relationship over time by including validated scales of therapeutic alliance and communication. Finally, long-term studies are also needed to follow how patient opinions change over time as availability, affordability, and adoption of this technology increases.

As AR and VR technologies are adopted into clinical settings, their impact on the empathy and communication from physicians must be carefully considered. Statistically significantly lower SCQ scores in the headset condition suggest that, without thoughtful design and implementation, AR and VR headsets may introduce unintended barriers to patient focus and engagement. However, patients remain open-minded with regard to the implementation of technology, including AR and VR, into clinical practice, and most would not oppose its use if it helps their physician perform their job better or more easily. Careful implementation and ongoing research will be essential to fully realize the potential of AR and VR in clinical settings while ensuring that patient-centered care and high-quality communication remain a top priority.