This study is anchored in the theoretical construct that the inherent characteristics of a simulation modality may influence participant attention and behavior. In addition, understanding that influence is key to ensuring that specific learning objectives are feasible to accomplish with the use of a given simulation modality. This aligns with the educational theory of constructivism, which posits that learners actively build knowledge by engaging in “hands-on” experiences, experimentation, and reflection within authentic contexts [17]. In a VR simulation, these authentic contexts are approximated by realistic clinical environments that include realistic digital patients and equipment to encourage exploration, role adoption, and collaborative decision-making. However, the same immersive features that make VR engaging (eg, richly detailed visuals, interactivity, and collaboration) may also impose additional cognitive demands (eg, technical work-arounds or deviations from reality) that could influence the construction of knowledge in unintended ways. The impact of those cognitive demands can best be understood through the lens of cognitive load theory, which highlights the limited capacity of working memory and its subsequent impact on knowledge or skill acquisition [16]. With VR simulations, expanded data sources in the form of virtual patients, equipment, environments, or other assets will influence the cognitive load burden on participants and potentially influence their behavior and subsequent learning. At the same time, novel controls and interface complexity may inadvertently add additional cognitive load, drawing learners’ mental resources away from essential clinical tasks and therefore potentially diminishing their bandwidth for the desired learning objectives. The integration of constructivism and cognitive load theory allows for a theoretical framework that aims to capture how participants effectively manage mental resources while constructing new knowledge through immersion, as well as identify when extraneous load hinders that process. This blended theoretical lens guided our data analysis and interpretation of observed learner behaviors in VR-based pediatric resuscitation simulations.

We used video-based focused ethnography [18,19] to study a cohort of video-recorded VR simulations. Focused ethnography is a methodological adaptation of traditional ethnography, offering a targeted and time-efficient approach to studying specific phenomena within specific contexts. This research strategy prioritizes depth of understanding over breadth, using concentrated data collection and a clear research question to describe particular social, cultural, or organizational processes. For this study, the goal was to understand how VR-based simulation affects participant attention and behavior. With this goal in mind, the research team was primed with the initial research question [19]: “How does a virtual reality simulation influence participant attention and behavior, and how might these changes impact educational goals?”

During this focused exploration, the team first identified and classified the data, then described and analyzed the specific behaviors, and finally moved to pattern explanation [18,19]. This study was approved by the Institutional Review Board at Cincinnati Children’s Hospital Medical Center. It was funded, in part, by the Laerdal Foundation.

Our research team, composed of an interdisciplinary group with combined expertise in critical care (DL, KE, and MZ), resuscitation (DL and MZ), simulation (DL, KE, and JS), and medical education (DL, JS, and MZ), reviewed a series of VR simulation recordings captured during a validation study of a specific VR curriculum [20]. The simulation took place in a virtual environment that mirrored a typical high-acuity room, as might exist in an emergency department or intensive care unit. Participants in the simulation included pediatric critical care and emergency medicine attendings and fellows, and pediatric intensive care unit and emergency department nurses. Each physician-nurse dyad had no previous exposure to the scenario and completed it only once.

The virtual patient was located on a hospital bed within a room that included a vital sign monitor, an intravenous pole with fluids, an oxygen delivery device, a stethoscope, a thermometer, a syringe for drawing blood samples, and a large fluid-filled syringe for triggering administration of a fluid bolus. A representative first-person perspective of the VR environment can be found in Figure 1. The virtual patient was an 8-year-old male with pneumonia and sepsis. The rationale for this scenario was that pneumonia frequently progresses to sepsis, and key stakeholders regarded this as a realistic scenario for clinical assessment and management training. The clinical status of the virtual patient changed throughout the simulation, conveying alterations in mental status (ranging from conversant to altered), perfusion (mottled skin that progressed to poor perfusion and cyanosis), and respiratory status (superimposed retractions and tachypnea).

The scenario was coded in Unity (Unity Technologies) and accessed via an Oculus Rift headset connected to a VR-capable laptop, which allowed 2 participants within the same simulation to move freely and interact with the scenario and each other, including communicating regarding shared findings in real time. The purpose of the simulation was to assess the participants’ ability to recognize, describe, and begin to treat a patient with developing shock. A detailed description of this VR simulation was previously described by Zackoff et al [20]. Participants received a 5-minute orientation to navigation in the VR environment and the functionality of the virtual patient and equipment from a trained VR simulation facilitator. They then received the case description and were instructed to begin their assessment and management of the patient as they would in real life. The VR simulation facilitator was present in the VR environment for the orientation and then exited the VR but remained present in real life to address questions and provide feedback throughout the remainder of the simulation session.

First-person audiovisual data recorded from the perspective of both users (a nurse and a physician) were stored in a password-protected, encrypted server, annotated via Vimeo (Vimeo, Inc), and coded in Microsoft Excel. Multiple audiovisual feeds, capturing the perspective of both participants within the virtual environment, allowed for data triangulation [21].

Considering reflexivity [22] and the desire for analytic triangulation [21], the research team was composed of a heterogeneous group of experts in resuscitation team leadership and medical education (DL), resuscitation bedside nursing (KE), and SBME (JS). DL is a practicing pediatric critical care physician as well as a simulation educator. KE is a simulation educator and former critical care nurse. JS is a simulation educator and a former pediatric emergency department nurse. The data analysis was also overseen by MZ, a pediatric critical care physician and education scientist with expertise in designing, implementing, and evaluating SBME using innovative modalities such as VR and AR. MZ served as a senior consultant, providing guidance and expertise at scheduled intervals and during critical decision points where consensus within the research team was required.

The research team reviewed the data, primed with the following generative question [23], “How does a virtual reality simulation influence participant attention and behavior, and how might these influences impact achievable educational goals?”

Using a focused ethnographic approach is particularly suited for foundational research in which little is known about how subjects interact with a new environment (or, more specifically, a technology that places subjects in a new environment). Rather than measure changes against a specific baseline or control group, a focused ethnographic approach provides foundational descriptions of behaviors [18]. Instead of benchmarking against manikin-based or live simulations, our goal was to capture and interpret emergent behaviors, areas of challenge, and decision-making processes that may be uniquely influenced by VR-based training.

To investigate participant attention and behavior within a virtual environment, we conducted a 3-phase analysis of VR simulations [18,24]. We followed a focused ethnographic strategy [18,19,24] incorporating inductive coding techniques [25]. This approach was chosen to (1) provide in-depth, context-specific insights into participant behavior in VR and (2) allow unanticipated themes to emerge from the data rather than imposing pre-existing hypotheses.

We used multiple strategies to enhance the credibility and dependability of our findings. First, we used analyst triangulation [21,28] by involving researchers with diverse backgrounds in critical care, nursing, and simulation education. Each analyst independently reviewed and coded selected video segments, then convened to discuss and reconcile any discrepancies. This iterative group process helped refine the codebook, ensuring that multiple perspectives were integrated and minimizing the influence of individual bias. We also maintained reflexive memos [22] to document personal assumptions or theoretical leanings; these memos were revisited throughout data analysis to encourage ongoing reflection on how our backgrounds might shape interpretations.

In addition, we created an audit trail [29] that outlined key decisions made during coding and theme development. This trail included records of coding updates, consensus discussions, and rationale for code merging or splitting, providing a transparent account of how the analysis evolved. By combining triangulation, reflexivity, and rigorous documentation, we sought to strengthen the trustworthiness of our findings and facilitate replicability of the study’s analytic approach.

The primary study and this secondary analysis received approval from the Cincinnati Children’s Hospital Institutional Review Board, which granted a waiver of documentation of informed consent in accordance with 45 CFR 46.116(d). This waiver was permissible because the research posed no more than minimal risk to participants, did not compromise their rights and welfare, could not be feasibly conducted without the waiver, and would provide subjects with additional relevant information after participation, when appropriate. Given its educational focus and the absence of risk to participants, the study met these criteria. Participation in the simulations was voluntary, with no compensation provided. All data on participant performance were stored securely on a password-protected server.

The ethnographic analysis was conducted on the recordings of 15 interprofessional teams of 2 users (a nurse and a physician) who completed the VR simulation. The simulations were recorded from both the perspective of the nurse and the team lead. This analysis revealed 3 major themes and associated subthemes that offer insights into the interplay between simulation modality and participant attention and behavior (Table 1).

Theme 1, Source of Truth, outlines the hierarchy of informational sources for participants analyzing clinical scenarios, including subthemes related to initial reliance on virtual cues and dynamic reliance based on responses. Theme 2, Cognitive Focus, describes the interplay between thinking, communicating, and doing, with subthemes of dynamic task burden based upon scenario complexity and the cognitive demands of the virtual environment. Theme 3, Fidelity Breakers, explores moments where the boundary between real life and simulation becomes most pronounced, leading participants to deviate from their typical behavior in real-life encounters. Subthemes within this theme relate to navigational and interactional challenges.

Participants face novel challenges in a VR medical simulation that they may not have encountered in previous traditional simulation experiences. They are immersed in a completely virtual environment, with real-world objects replaced with digital replicas. This immersion requires participants to reorient themselves to the virtual room and their ability to move and interact within it. We gained insights into how participants navigate this new environment and identify reliable sources of information to guide their actions within the simulated medical case.

Participant attention is finite, and the dynamic interplay between the VR environment and participants’ cognitive load unfolded during the simulated scenario. By introducing an entirely virtual world and placing participants in it, VR introduces unique features that may impact the cognitive load and task burden placed upon the participant [30,31]. We observed how the virtual environment affected participants’ ability to process information, make decisions, and execute actions within the simulation. These effects occurred both as an intentional part of the design and as unintended consequences of the VR simulation’s technical limitations.

Factors intrinsic to the VR environment appeared to challenge participant immersion and disrupt their sense of “being there.” These disruptions, termed Fidelity Breakers, appeared to break participants’ engagement with the simulation and shift their focus from clinical reasoning to technical troubleshooting. The identified Fidelity Breakers fell into the 2 main categories of navigational challenges and interactional challenges.

This study identified 3 major themes regarding the impact of a headset-based VR simulation on participant attention and behavior. VR was found to influence participants to rely primarily on the virtual patient for clinical data and decision-making; however, the participants’ reliance on “Theme 1: Source of Truth” shifted to the cardiorespiratory monitor when the expected patient response to interventions did not occur. In addition, learner behaviors in the VR environment closely reflected the natural progression of patient care, moving from initial assessment to formulating and implementing a treatment plan, allowing closer approximation of the “Theme 2: Cognitive Focus” experienced during clinical care. Finally, participants faced navigational and procedural challenges that were “Theme 3: Fidelity Breakers,” skewing participant attention to the simulation’s technical details rather than the intended learning objectives.

Practicing medicine in a completely virtual environment is an unfamiliar experience for most medical staff and can be disorienting. In this constructed environment, participants primarily relied on the patient examination as their source of objective data, with focus shifting to the cardiorespiratory monitor only when the virtual patient did not respond as expected to interventions. This grounding of the patient as the center of the simulation, as opposed to a primary reliance on ancillary data, represents a key potential benefit of VR training—reinforcing the patient examination as central in the practice of medicine. Although these observations suggest that VR might help draw learners’ attention to direct patient cues, this conclusion is tentative without a direct comparison to other simulation modalities. While VR could reinforce the patient examination as central in the practice of medicine, further research should verify whether this pattern is unique to VR or equally present in other simulation approaches.

Though participants focused on the patient examination, this examination was at times limited. Due to the unique learning curves and inherent technological limitations for movement within the environment and physical manipulation of the patient and equipment, participants often relied primarily on audiovisual cues. In addition, technological glitches causing accidental teleporting and clipping resulted in participants electing to limit their movement around the room, favoring leaning and reaching over moving around the environment, as highlighted through the examples listed in Table 1, Fidelity Breakers. These challenges and resultant learner adaptations suggest that VR might have limitations for learning objectives focused on examination maneuvers or procedures requiring “hands-on” skill development. Until there is no need for technical workarounds or deviations from reality to perform skills in VR, as observed in this study, experience with these tasks in VR may not support skill acquisition that can be directly applied to real-life clinical care. In addition, the cognitive load added by navigating these workarounds may distract from or prevent the acquisition of other learning objectives. Alternatively, these skills may be better learned, practiced, or assessed with alternative modes of simulation, such as computerized manikin-based or AR-enhanced simulation, and thus specific studies examining specific skill acquisition in different modes of simulation are needed.

A virtual patient has the advantage of being able to portray a broad and realistic range of physical findings, as well as dynamic responses to therapy. However, inherent to VR is the requirement of creating digital content for the participant’s entire field of view and expected actions—extending to the environment, equipment, and the interactions with said equipment. Though education and design teams can meticulously craft a virtual environment, any deviations from the real-world environment risk distracting the participant from the primary objectives of the training exercise. This dissonance extends from observable deviations from the real world to divergences in how to perform tasks. Although this VR simulation was designed to focus on recognizing and treating shock, participants were at times distracted by activities outside of that direct focus, such as needing to relearn how to administer a fluid bolus in VR. This intervention was intentionally simplified to reduce the technical learning curve of the simulation, but this simplification had the unintended effect of distracting the learners from the primary focus of the simulation. Future work might explore balancing necessary simplifications with realistic task fidelity to minimize extraneous cognitive load in VR.

VR-based simulation demands additional orientation and training that differ from what is typically provided in manikin-based scenarios. Because certain actions—such as administering a fluid bolus in this scenario (refer to Table 1, N10)—are not replicated 1:1 from real-life procedures, participants must allocate some cognitive load to learning the VR mechanics. This interface-based burden can divert attention from the core clinical objectives, underscoring the importance of structured, VR-specific training sessions before learners engage in complex clinical tasks, and considering these extraneous loads when targeting specific learning objectives.

Decision-making when treating critical illnesses must be quick to be effective. In VR simulations, team leads made decisions based on a combination of group consensus, individual opinions, and real-life clinical algorithms. Group discussions were frequent, and team communication dynamics appeared to significantly impact decision-making. The VR scenarios included only two participants (plus a facilitator), making decision-making conversational. However, nonverbal communication was limited for this VR simulation, as only participants’ virtual hands and glasses were visualized, which limited assessment of body language and eliminated the ability for eye contact. For this specific VR simulation, the modality encouraged verbal communication around shared observations and findings but lacked feasible options for nonverbal communication. While this reliance on explicit verbal communication may facilitate clarity in certain respects, we cannot determine whether it is more frequent or effective than in traditional simulations without further comparative research. Finally, despite the presence of ongoing communication, individual parallel assessments continued, suggesting that participants processed information independently. A more complex environment or scenario, with potentially a larger team of clinicians and staff, might result in different communication dynamics than what we observed. Assessing the attention and behavior of larger teams in more complex clinical contexts represents key future work for our team.

While our study sheds light on important aspects of VR training for pediatric resuscitation, several limitations warrant consideration. First, the data originated from a single institution over a limited timeframe, encompassing only 15 simulations. While this sample size is relatively small, the participants represent a significant portion of the pediatric resuscitation team at a major academic medical center, potentially fostering generalizability to similar institutions. In addition, data saturation was achieved after analyzing 15 scenarios, indicating that further analyses would not likely yield novel findings. Second, the focused ethnography approach involves inductive analysis, meaning the researchers’ experience and expertise are inherently woven into the methodology. This strengthens the analysis by adding depth and richness to the conclusions drawn. Notably, the research team was purposefully assembled to leverage their expertise in simulation and resuscitation, allowing them to inject diverse perspectives into data analysis and enrich the interpretation of findings.

However, this methodological approach lacks a direct comparator or control group. Because our objective was to describe behaviors in a previously unexamined context, fully immersive VR, we did not measure change from a standard baseline (eg, manikin-based or live simulations). This limitation was partially mitigated by providing rich, context-specific observations from multiple angles, coupled with cross-disciplinary analyst triangulation. Future studies may build on these findings by including mixed method comparisons across different simulation platforms to quantify how VR might augment or hinder specific learning objectives. In addition, there is currently no literature that describes the focus of interdisciplinary teams during real-life clinical care—a key future direction for our team to enhance our ability to align participant behaviors during simulation with those expected during real life.

Finally, this study used a single VR scenario, and thus the conclusions drawn must be considered in relation to the specific technical design and functionality of that scenario. However, while there are a range of VR systems and approaches at various stages of development and availability, the foundational principles of navigating and interacting in a virtual world are likely similar across platforms. As VR technology advances, limitations in movement within the virtual space may be resolved. For example, wireless head-mounted displays allow for easier movement in the VR space by untethering users from stationary computers. Yet even at its best, VR can only approximate—rather than fully replicate—the fidelity of genuine hands-on, kinesthetic practice. Further research can help clarify how these core elements influence participants’ clinical reasoning, teamwork, and skill acquisition, enabling educators to better tailor VR-based simulations to specific learning objectives.

The conclusions drawn from this study provide a comprehensive understanding of how a VR-based simulation may influence participant attention and behavior, revealing both strengths and limitations of this training modality. Our findings indicate that VR simulations create a unique environment where participants predominantly rely on virtual patient assessments and audiovisual cues. This setup prioritizes patient-centered data over ancillary information, fostering focused clinical decision-making. However, the constraints of VR, such as limited physical interaction and navigational challenges, can impact the authenticity of the training experience, suggesting that VR may be better suited for clinical assessment or cognitive objectives over physical skill–based objectives.

As the landscape of SBME evolves, it is crucial to recognize the potential and limitations of VR simulations compared to other modalities. The ability to create a fully immersive, yet controlled, environment offers significant training advantages for objectives that focus on rapid decision-making and patient assessment. Nevertheless, the technological constraints that limit physical interactions highlight the persistent need for complementary simulation modalities, such as AR and computerized manikin-based simulations, which may better support hands-on skill development. Next steps include the delineation of how the growing armamentarium of simulation modalities and tools can be optimally leveraged to best train a broad range of clinical skills.