More than 60% of Americans will face a period of significant disability before death, incurring a diminished quality of life from the impact of serious illness, including multimorbidity [1-3]. Serious illness is defined as a health condition with a high risk of mortality that either negatively impacts daily function or quality of life or strains a person’s caregivers [4]. Serious illnesses include those of indeterminate but limited prognosis and those that have a detrimental effect on the quality of life of patients and their caregivers. They encompass malignancy, organ failure, or frailty, a complex state of vulnerability that is often associated with dementia [4,5]. Serious illnesses are associated with burdensome symptoms and functional decline and include many leading causes of death, such as stroke, as well as multimorbidity [6]. Patients also face a significant financial burden because of medical treatment [7]. These conditions interfere with a person’s ability to engage in activities they once enjoyed and impose emotional and physical burdens on family members and other caregivers; these conditions are of growing importance with the population aging [4].

Functional outcomes are a useful adjunct to other quality of life–relevant measures to understand the impact of serious illness on a person’s life [8]. Activities of daily living (ADLs) include basic functions such as dressing, maintaining personal cleanliness, toileting, transferring, and eating. The impairment of 1 to 2 ADLs occurs in most Americans during the years before death, and ≥3 ADL impairments are associated with institutionalization. Important goals for patients include dying at home and maintaining cleanliness, both of which are abetted by functional status or require dedicated caregiving if adequate support is unavailable [9]. Evaluating functional outcomes in patients with serious illness is especially important because it directly reflects their quality of life and their ability to perform ADLs [10].

Virtual reality (VR) is one among emerging interventions (eg, smart technologies) with the potential to improve a patient’s quality of life, patient-reported outcomes, psychological outcomes, and functional outcomes for persons living with serious illness [11-14]. VR offers a computer-generated, stereoscopically rendered, 3D visual environment, often with complementary sound, which responds continuously to a patient’s movement. Immersive VR creates a psychological experience of treating the virtual simulation as a real experience. The sensory experience is thus distinct from nonimmersive reality, such as augmented reality that allows the participant to see computer-generated images superimposed on the real-world visual field [15,16]. Immersion and distraction therapies have been shown to be helpful in improving quality of life and functional outcomes. The use of VR in health care for serious illness care has been expanding to address pain, anxiety, and other needs in palliative care and hospice settings [17,18]. There is a need to understand the quality of research and VR’s efficacy, especially in randomized controlled trials (RCTs), as it applies to older, seriously ill adults.

Several recent reviews have explored content salient to our interest, although not in the specific context of applying VR to palliative care and serious illness [19-21]. We aim to determine the extent to which rigorous studies have focused on the rapidly growing population of older adults, the quality of that research, the extent to which it adheres to recently accepted VR research standards, and the characteristics of that research (eg, common outcomes) [22-24]. Two secondary aims include assessing the components of interventions in RCTs associated with improved quality of life and functional outcomes, as well as assessing the extent to which the current RCTs related to VR adhere to recommended consensus standards for high-quality VR research.

Our review adhered to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) protocols statement and Enhancing the Quality and Transparency of Health Research guidelines [25,26]. We registered our protocol to the PROSPERO (CRD42022346178) [27].

Included in this systematic review are studies published in peer-reviewed journals up through May 2024 that meet the following criteria organized by population, intervention, comparator, outcomes, timing, and setting framework (Textbox 1) [28]. Functional outcomes include patient-reported outcomes relating to the quality of life and patient-demonstrated outcomes (eg, arm curl, chair stand, back scratch, chair sit, reach, walk test, overall cognition, and memory) [29,30]. Biological intermediaries (eg, functional magnetic resonance imaging scans, cortisol levels, blood test values, and balance) that do not directly demonstrate patient function are not included as patient functional outcomes [31]. Eligibility criteria are described using the population, intervention, comparator, outcomes, timing, and setting framework (Textbox 1).

Concepts included in the literature search were RCTs, VR, and serious illness as shown in Multimedia Appendix 1. We searched PubMed, Embase, and CINAHL for studies published at any time and identified 12,814 articles (n=10,799, 84.28% articles after duplicates were removed). During the title and abstract screening and full-text screening phases, 3 coauthor reviewers screened each study and were blinded to each other’s decisions (BM, AK, and SF). During the title and abstract screening, we resolved conflicts predominantly by a “gold standard” reviewer (KL or KG), occasionally by majority consensus. During the full-text screening, reasons for exclusion were identified. Only the gold standard reviewer (KL or KG) resolved conflicts during full-text screening. During full-text screening, if any systematic reviews met inclusion criteria, we added their included articles to the title and abstract screening phase. We used the Covidence (Veritas Health Innovation) software to generate a PRISMA diagram to track studies at each stage of the review (Multimedia Appendix 2 [25]) [32].

We built an abstraction form through an iterative process (Multimedia Appendix 3). All aspects of the included article interventions were recorded, including what the intervention entailed; the type of VR and media (eg, type of scenery or activity) used; any training, supervision, or assistance provided during VR use; the duration of sessions; frequency of sessions; and duration of intervention. The main outcomes collected were related to patient quality of life–related outcomes. Data abstraction for the included studies was done through Covidence with 2 reviewers per article. Two reviewers (BM, AK, or NI) abstracted each article independently and resolved all abstraction conflicts through a consensus discussion.

Risk assessment for bias was done using the Cochrane Risk of Bias tool for RCTs [33]. Two reviewers (SF, AK, or BM) independently assessed the risk of bias in the domains of randomization, allocation concealment, blinding, accounting of patients and outcome events, and selective outcome reporting bias. Any disagreements were discussed by each of the 2 reviewers, and a consensus was reached.

We performed a narrative synthesis of the data abstracted due to the high degree of heterogeneity of the results, as each included article reported different forms of patient functional outcomes. We synthesized each VR-based intervention and extracted the purpose of VR within the intervention and whether there was any training, supervision, or assistance provided in the use of VR. To better understand the individual differences that may affect the efficacy of VR-based interventions, we looked at the average patient age, SD and range, type of serious illness, patient gender, patient ethnicity, and study location for each of the studies. We characterized each study’s adherence to recent consensus standards for high-quality VR research using the definitions below [22].

A total of 12,621 unique titles and abstracts were dual screened by coauthors as per the inclusion and exclusion criteria. A total of 152 full-text articles were retrieved and further assessed for eligibility. Of the 152 full-text articles reviewed, 128 (84.2%) studies were excluded for reasons noted in the PRISMA flow diagram (Figure 1 [34]). Finally, we included 24 total studies in our review [35-58], of which 21 (88%) reported an improvement in at least 1 patient functional outcome [35-43,45-48,51-58]. The included articles are summarized in Table 1.

In total, the studies included 1225 participants. Of the 24 included studies, 3 (12%) included patients with an average age of <50 years [43,49,58]; 8 (33%) included patients with an average age of 51 to 60 years [35,36,38-41,46,52,55,57]; 5 (21%) included patients with an average age of 61 to 65 years [42,45,48,51,52]; 3 (12%) included patients with an average age of 66 to 70 years [44,50,56]; and 1 (4%) included patients with an average age of >75 years [37]. Two studies did not report the average age of the patients, as listed in Table 2 [47,54]. No studies reported how many patients in the sample were aged >65 years. A total of 16 (67%) of the 24 included studies had a high risk of bias (Multimedia Appendix 4 [35-58]).

Stroke, cancer, and cardiovascular disease were the most common types of serious illnesses for which VR interventions were used to impact the outcomes related to patients’ quality of life. Out of 24 included studies, 9 (38%) focused on patients who experienced stroke, which included both ischemic stroke and hemorrhagic stroke [37,40-43,46,48,49,53], and 9 (38%) included patients who were diagnosed with some type of cancer, varying in the stages of cancer in each patient [35,36,38,42,43,47,52,55-58]. In 5 (21%) of the 24 studies, only patients who were diagnosed with breast cancer were included; 1 study (4%) included patients with colorectal cancer, and 1 (4%) included patients with heterogeneous types of cancer. Four (17%) out of 24 studies included patients who had some form of cardiovascular disease, including heart failure and other conditions that required cardiac surgery [39,44,45,50]. One study included patients with chronic obstructive pulmonary disease (COPD), and another included patients who reported pain during their hospitalization [51,54].

In 9 (38%) out of 24 included studies, the main purpose of VR was to improve mobility and strength in the patient, ranging from upper extremity, lower extremity, and pelvic or abdominal region to a combination of these regions [37,40-43,46,48,49,53]. All of these mobility studies also included patients who experienced strokes. The VR media used in these studies encouraged active participation from the patients in various tasks and games. Some examples of VR media used to help improve mobility and strength in patients’ poststroke treatment were daily exercises that encouraged increased range of motion, strength for specific movements, and reach and grasp through the use of balls. A total of 8 (89%) of the 9 studies that used VR for the purpose of improving mobility and strength improved patient quality of life–related outcomes [37,40-43,46,48,53].

There was some overlap in the protocols and diseases of the additional studies. Out of 24 studies, 12 (50%) used VR distraction therapy to reduce pain in patients with cancer or, cardiovascular disease or hospitalized patients [35,36,38,39,42-44,47,50,52,54-58]. A total of 9 (38%) out of 24 studies used VR to reduce anxiety in patients with cancer, cardiovascular disease, or COPD [36,38,42,43,50-52,55-58]. In studies aiming to reduce pain and anxiety, patients experienced immersive natural scenery and comforting music while comfortably sitting or lying down. In another study, to reduce anxiety, patients experienced the hospital environment and could visualize and review all the steps of an upcoming procedure [56]. In another study, patients with cardiovascular disease went through a visual guided meditation track with the VR media for patients with cardiovascular disease [45].

In 18 (75%) out of 24 studies, staff or physical therapists either trained patients on VR use before the intervention or supervised or assisted them through the experiment. Supervision was defined as the research team watching over the participants, ensuring their use and adherence to study guidelines. However, assistance was defined as the research team supporting patients during the intervention without constant watching throughout the experiment. A total of 5 (20.8%) out of 24 studies included VR training, supervision, and assistance, all of which reported improvements in patient quality of life–related outcomes [38,40,43,46,54]. Of 24 studies, 7 (29.2%) included VR supervision and assistance only [36,37,41,44,47,53,57], and 2 (8.3%) included VR training and assistance only [39,49]. Two (8.3%) out of 24 studies included only VR supervision while patients were using VR [35,55]. In studies that included a variation of personnel working with the patient, only 1 staff member physically accompanied the patient in the room, where they freely communicated with the patient to fulfill their role. During a typical training, the staff member would walk the patient through the VR activity and teach them how to adjust the headset. Out of 24 studies, 6 (25%) did not describe training, supervision, or assistance with the use of VR [42,45,48,50,51,56].

The frequency and duration of VR experience varied based on the intervention purpose; interventions aiming to improve pain or anxiety were shorter. while those aiming to improve mobility and strength were longer. Out of the 24 included studies, 9 (38%) studies with pain or anxiety reduction as a goal, involved only 1 VR session, ranging from 8 minutes to 4 hours (ie, patients could use VR at any time within a 4-hour window) [35,38,39,44,47,50,52,56]. A total of 7 (78%) of these 9 studies had improvements in pain and anxiety metrics, such as the Hospital Anxiety and Depression Scale (HADS), State Anxiety Index, or the Numeric Pain Scale (NPS). One study, which conducted a single 30-minute meditation session, showed improvement in sleep scores of patients with cardiovascular disease [45]. In 1 case where VR was used for 10 minutes per session, for 3 sessions per day for 2 days, patients experienced statistically significant improvement in the NPS (P<.04) [54]. Patients in another study who used VR for 20 minutes per session for 5 sessions per week for 2 weeks had statistically significant improvement in the HADS (P<.001) [51].

In the 9 studies aimed at improving mobility and strength in patients who had a stroke, VR was used for a longer duration than in pain and anxiety studies. Out of 9 studies, 6 (67%) lasted between 2 and 4 weeks [37,42,43,46,49,53], while 3 (33%) lasted between 5 and 9 weeks [40,41,48]. A total of 8 (89%) of these 9 studies showed an improvement in the patient’s functional capacity measured via the Fugl-Meyer upper extremity assessment, proprioception, or orientation metrics.

As shown in Table 3, 7 studies adhered to all the 8 gold standard VR consensus standards criteria (100%) [41,43,44,48,51,52,54], 10 adhered to 7 criteria (88%) [36,37,39,40,42,46,55-58], 6 adhered to 6 criteria (75%) [35,38,47,49,50,53], and 1 adhered to 5 criteria (62%). The reasons for not adhering to the criteria were inadequate blinding, concealment of allocation, reporting of the trial, and a priori failure to justify study duration. The studies that did not report the trial via World Health Organization International Clinical Trials or Clinicaltrials.gov originated from countries where it may not be standard practice to report the clinical trial before the study initiation. All 24 assessed studies received the institutional review board or ethics committee approval. A total of 4 (17%) of the 24 included articles were published before the 2019 study by Birckhead et al [22] that defined the gold standard VR consensus standards criteria.

We found 24 RCTs of VR for older adults with serious illness that encompassed diverse conditions, such as stroke, cancer, cardiovascular disease, and COPD. Among these studies, VR has been applied to improve anxiety and pain, as measured by the HADS, State Anxiety Inventory, or NPS, or mobility and strength, as measured by the Fugl-Meyer upper extremity assessment, proprioception, or orientation metrics. In this review, short VR experiences were typically used to improve outcomes related to anxiety and pain, while longer ones were used to improve mobility and strength after stroke. Training, supervision, and assistance provided were described in 5 studies [38,40,43,46,54]. Of the included RCTs, all 24 studies addressed older adults; however, only a few focused entirely on the older adult population: 3 studies (12%) included adults with an average age of 66 to 70 years [44,50,56], and only 1 study (4%) included older adults with an average age of >75 years [37]. From the 24 RCTs included in this systematic review, 7 (29%) studies adhered to all 8 gold standard VR consensus standards criteria, and all studies (100%) adhered to at least 5 criteria.

Although there are emerging bodies of evidence on VR use for acute pain, physical therapy, and mental health, we found relatively few studies that met our criteria, focused on symptomatic and functional aspects of serious illness or their use for older adults [59-61]. Acute pain and pain associated with serious illness are distinct, as suggested by the total pain model for serious illness by Saunders; are multifactorial; and explicitly consider physical, emotional, social, and existential factors in worsening symptoms [62,63]. Function is also impaired to a greater degree in older adults, and the average American lives with 1 to 2 ADL disabilities during the last 2 years of life [64]. Comorbidity is common among older adults, and underlying conditions and treatments can interact to exacerbate symptoms [65]. Owing to this, the same interventions used to manage acute pain may not be efficacious for managing pain associated with serious illness, especially among older adults with serious illnesses.

The practicality and efficacy of VR for older patients may be limited by difficulty using the technology, cost, and limited accessibility in medical settings [66]. For example, an approximately 1- to 2-pound VR headset may be relatively heavy, which is challenging for patients with frailty, as well as those with a serious illness [67]. In many cases, older patients may tend to be unfamiliar with new technologies and, therefore, may require supervision and assistance. This role can be fulfilled with caregivers, friends, or clinicians. In addition, immersive VR can cause motion sickness, which is a concern for older adults with serious illnesses, and are costlier than nonimmersive therapies [68]. Unfortunately, 5 (21%) of the 24 assessed studies lacked descriptions of supervision, training, or assistance provided, which may limit the ability of other researchers or providers to translate the results to their settings [45,48,50,51,56]. Therefore, VR assessments in older adults should include whether and how users were trained and supported, how their clinical condition impacted their ability to use the system, and how their ability to independently use VR was verified.

The extent to which VR treatment efficacy is reliant on media content remains unclear. This is particularly important because VR audio and visual media content varies generally within and between different VR systems. Furthermore, different clinical conditions and intervention goals may lead to the use of specific VR resources, interactive functionality, interactivity, and content. As a result, there are fundamental limitations for researchers and clinicians in understanding the mechanisms and efficacy of VR due to the opaqueness or lack of content descriptions in many publications [69]. In addition, subtleties in achieving high presence or in how perceived reality is portrayed may significantly impact the clinical impact and patient satisfaction. Therefore, better-published specifications of VR media, such as soundscapes, public accessibility, and other features, are required to better evaluate and clinically apply VR to patients [11].

This evaluation is limited by the sparse availability of specific VR content that was used in the previous studies. In addition, the outcomes related to patients’ quality of life in the different studies were heterogeneous, and therefore, a quantitative synthesis was not possible. These factors, combined, limited the ability to characterize the effects of specific VR features on patient outcomes. However, this may be a representation of an evolving field with limited reporting guidelines. In addition, this study focused specifically on immersive VR, and future research can further examine the effectiveness of different types of VR. One of the original aims of this study was to compare individual differences in VR effectiveness (eg, by gender, race, and ethnicity); however, this was not possible due to a lack of sufficient detail in included studies. Nonetheless, this systematic review provides researchers and clinicians with a detailed overview of the current state of the literature for older adults with serious illness, as well as an awareness and emphasis on the need to follow standard best practices to enhance VR research rigor. Future directions to build upon this systematic review include evaluating different types of VR media and frequency and duration to target improvements in specific outcomes in patients with serious illness. Future studies may also consider more structured supervision and training for VR use for older adults and determine whether this leads to increased compliance and efficacy of VR to improve outcomes in serious illness. Finally, future research can focus on making VR more accessible to everyone, including older adults [70].

Our systematic review identified a growing body of research evaluating VR apps for older adults with serious illness through RCTs. The literature review illustrates the potential of VR in this population and finds promising but limited extant evidence. These limitations also highlight an important gap in the development and understanding of using VR to improve health and wellness, which are crucial to address for rapidly aging populations.