Osteoarthritis of the knee is a painful chronic condition that is highly prevalent and disabling [1-3]. In 2008, the prevalence was estimated to be 13.9% among adults aged 25 years and older and 33.6% in those aged 65 years and older, corresponding to approximately 27 million Americans with knee osteoarthritis (KOA) [4]. KOA prevalence is rising [5], as are rates of total knee replacement [6,7]. The direct and indirect economic costs of treating KOA are substantial and include surgical costs associated with TKA, disability among people with painful and limiting disease, and loss of work-related productivity. Direct and indirect lifetime costs for persons diagnosed with KOA were estimated in 2013 to be US $140,300 [8].

Nonpharmacologic therapies (eg, exercise) are first-line treatments for KOA due to their benefits and low risks, particularly compared with opioids and invasive procedures [9-12]. Pharmacologic therapies are also frequently used for KOA but carry well-described adverse effects [13]. When these therapies yield insufficient improvement in symptoms, patients may undergo invasive procedures, such as injections using corticosteroids or viscosupplementation, or surgical procedures, such as total knee arthroplasty (TKA), all of which are associated with risks of AEs. Nondrug treatments (eg, exercise) often require long-term adherence, making patient engagement a key factor in their effectiveness and durability. Similar nondrug treatments are also used in post-TKA rehabilitation as an integral part of postprocedure recovery.

Extended reality (XR) technology can deliver pain interventions using various clinical mechanisms, including pain self-management education, psychological skills, passive distraction, and digitally augmented exercise (DAE) [14,15]. The popularity of XR technologies, decreasing costs, increased availability, and innovations in XR hardware and software have contributed to increased interest in this relatively new application. XR can be categorized according to the level of immersion experienced by the user in the digital environment [15]. Virtual reality (VR) presents the highest (full) level of immersion within an interactive digital environment and describes a state in which the user is fully enclosed within an artificial virtual space in which the visual sense is sealed off from the physical environment [15]. For example, users may undergo guided relaxation training taking place in digitally created settings, such as a peaceful forest or relaxing seaside environment, that are visually separated from the real-world physical environment. Full immersion is typically achieved with the use of a headset. Nonimmersive technologies have been used to facilitate DAE and have historically been evaluated in the context of XR but do not incorporate the user’s physical world into the visual experience and are referred to as interactive digital rehabilitation (IDR) for the purposes of this report. These screen-based technologies are best showcased in gaming systems like Microsoft Kinect and Nintendo Wii (Nintendo Co, Ltd) [16]. XR interventions have demonstrated utility for reducing acute pain through distraction (eg, during dental procedures) [17], but whether these interventions improve outcomes when integrated into treatment plans for chronic musculoskeletal pain conditions remains unclear. Prior systematic reviews of XR for KOA [18,19] and post-TKA rehabilitation [20,21] have often pooled heterogeneous XR technologies without stratifying results by level of immersion. Results at varying follow-up durations have also been pooled, and there has been a lack of systematic evaluation of adverse events (AEs). An updated synthesis is warranted given the rapid expansion of the XR evidence base in recent years and the increasing availability of XR devices in clinical and home settings.

In this review, the objective was to synthesize available evidence on the benefits and harms of XR interventions for the treatment of chronic KOA pain and symptomatic recovery after TKA, with results stratified by level of immersion (VR vs IDR), clinical mechanism, and follow-up duration and with certainty of evidence summarized using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) criteria.

We present findings on KOA and post-TKA that were part of a larger systematic review undertaken to evaluate the evidence on XR therapies for treating chronic and acute pain conditions [22]. Thus, the results included here focus on the following key question: What are the benefits and harms of XR interventions for the treatment of chronic knee pain due to osteoarthritis or for recovery after TKA?

A preregistered protocol for the larger review can be found on PROSPERO (International Prospective Register of Systematic Reviews; registration number CRD42023439903). This review is reported in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 statement [23], with search reporting following the PRISMA-S (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Literature Search Extension) [24]. The PRISMA 2020 and PRISMA-S checklists are provided as Checklist 1. There were no deviations from the registered protocol with respect to the eligible populations, interventions, comparators, outcomes, or analytic approach reported in this manuscript.

Eligible studies evaluated XR interventions for adults with symptomatic KOA or those who had undergone TKA. Prioritized outcomes were pain-related functioning or interference, pain intensity or severity, and AEs. Other eligible outcomes included pain global change, pain catastrophizing, and kinesiophobia, quality of life, opioid dose or use, physical performance, and adherence. AEs were defined broadly as any untoward medical event occurring during or following the intervention and, for XR interventions, specifically included cybersickness (nausea, dizziness, eye strain, or disorientation related to visual-vestibular mismatch), motion sickness, musculoskeletal injury, and falls. All results compatible with each outcome as they pertained specifically to KOA or post-TKA recovery at any timepoint were sought.

MEDLINE, Embase, Scopus, CINAHL, and PsycINFO were searched from inception to May 2023 using subject headings and keywords for VR, exergaming, and pain, along with additional terms for conditions in which pain is a predominant symptom (eg, arthritis). Duplicate records were removed prior to abstract screening using deduplication tools in EndNote (Clarivate) [25] and DistillerSR (Evidence Partners) [26]. Titles and abstracts were screened in DistillerSR by 2 separate reviewers (2 separate reviewers for each reference: combination of CK, DDP, DE, CC, MA, AL, KU, NZ) with agreement from both required to exclude. Included abstracts were moved to full-text review, with eligibility again requiring consensus of 2 separate reviewers. Disagreements were resolved with input from additional team members. Excluded references are shown in Multimedia Appendix 1. Database searches were not adapted from any previous review. Relevant systematic reviews identified during the database searches were hand-searched for additional articles. No search peer-review process was undertaken, and no additional sources were purposively searched or consulted. Searches were updated in December 2025 using the original search terms, with the exception of the population of interest, which was narrowed to focus solely on KOA and TKA. Multimedia Appendices 2 and 3 provide detailed study eligibility criteria and complete search strategies. The study screening process and literature flow are depicted using the PRISMA 2020 flow diagram [23].

Data were abstracted by 1 reviewer and verified by a second reviewer. Abstracted data included participant characteristics and inclusion/exclusion criteria; intervention characteristics (technology and devices used, content and goals of intervention); study design and settings; and findings for primary and secondary outcomes (eg, baseline and follow-up means and SDs, change in scores and SDs of changes, and P values calculated by studies). Data were abstracted into forms constructed in DistillerSR.

Risk of bias (RoB) assessments were conducted independently by 2 researchers using the Cochrane Risk of Bias 2.0 tool [27], which was also adapted to DistillerSR. Discrepancies were resolved by consensus, with input from other parties for resolution as needed. RoB assessments for each included study [28-49] are shown in Figure 1.

We synthesized findings from eligible studies qualitatively and, where ≥3 sufficiently similar studies provided appropriate outcome data at comparable time points, performed random-effects meta-analyses using the Hartung-Knapp-Sidik-Jonkman method [50-52] using meta, metafor, and pimeta packages and R (version 4.3.1; R Foundation for Statistical Computing) to generate forest plots [53]. We examined intervention, comparator, and participant characteristics to determine whether studies were similar, including first grouping studies by clinical indication (ie, osteoarthritis, knee pain, or post-TKA), intervention classification as VR or IDR (according to the framework proposed by Rauschnabel et al [15] and Spiegal et al [16]) and clinical mechanism (distraction or DAE). Distraction refers to the effect of directing attentional resources of the user away from unpleasant symptoms or stimuli, involving techniques such as music, imagery, or relaxation [54,55]. DAE entails an interactive process through which the user is given instruction or a demonstration of a target exercise before performing the exercise themselves [46]. Author-reported XR equipment and devices are shown in Multimedia Appendix 4.

For efficacy outcomes (measures described in Table 1), we assessed between-group differences in mean changes of continuous outcomes (ie, difference in change scores [Diff Δ]), preferentially as standardized effect sizes (Diff Δ/SD of change) [56]. To calculate Diff Δ, we subtracted the mean change in the comparator group from the mean change in the XR intervention group (Δ_XR–Δ_C). Thus, for outcome measures in which lower scores are better (eg, pain intensity or severity), a negative value for Diff Δ indicates greater improvement in the XR intervention group. Table 1 provides a description of the standardized measures reported by studies and the interpretation of scores (eg, higher score is better or worse).

In determining whether there were between-group differences in efficacy outcomes, we preferentially examined standardized Diff Δ and applied recommended thresholds (eg, <0.2 indicates no between-group difference) [56]. If standardized Diff Δ was not reported and could not be calculated (eg, due to lack of SD or confidence intervals), we considered study analyses and interpretations of findings, as well as actual Diff Δ values, to determine whether there were between-group differences. No sensitivity analyses were conducted.

Reporting bias was not formally assessed quantitatively (eg, funnel plot asymmetry tests) because no individual meta-analytic comparison included ≥10 studies, below which such tests are considered unreliable [57]. We instead considered the potential for missing studies qualitatively, including inspection of trial registries and reference lists of prior systematic reviews identified during the searches.

We assessed certainty of evidence (COE) for 3 prioritized outcomes, including pain-related functioning or interference, pain intensity or severity, and AEs. We rated the COE for these outcomes separately for VR and IDR interventions by clinical mechanism (DAE, distraction, or psychoeducation) and by condition (ie, KOA pain and post-TKA pain and rehabilitation). We used GRADE methodology [58] to rate overall COE as high, moderate, low, or very low (Multimedia Appendix 5). For each prioritized outcome, we used the GRADEpro Guideline Development Tool (GDT) [59] to systematically evaluate 5 domains: study limitations, imprecision, inconsistency, indirectness, and other considerations. For presentation of anticipated absolute effects, data from exemplar studies used to illustrate the range and direction of findings at time points were selected on the basis of (1) lower RoB ratings and (2) the largest sample size. Certainty of evidence tables are found in Multimedia Appendices 6 and 7.

We identified 21 studies [28-49] (reported in 22 publications) that evaluated the use of XR interventions for patients with KOA (k=12) or post-TKA (k=9; Figure 2), as summarized in Table 2. The studies were mostly small to moderate in size (k=18 with a total number of participants n≤100), with a few that were larger (k=2 with n=101‐200 and k=1 with n>200). Follow-up duration ranged from 2 weeks to 6 months with most studies (k=16) having ≤3 months of follow-up. The primary clinical mechanism was consistent throughout the studies, with 19 studies [28-32,34-39,41-49] using DAE and 1 study [33] using distraction and 1 [40] using psychoeducation. No studies reported on pain, global change, or opioid use. All but one study [47] was conducted outside of the United States (k=20), and most included middle-aged (k=10 aged 30‐64 years) and older adults (k=7 aged ≥65 years). One study included only young adults with posttraumatic osteoarthritis [42], and mean age was not reported in k=3 studies [39,43,46]. Studies were rated as having a high RoB (k=10) [28,30,31,33,37,41,43-46], some concerns for RoB (k=11) [32,34-36,38-40,42,47-49], or a low RoB (k=1) [29]. Detailed RoB assessments are shown in Figure 1. Detailed trial characteristics are found in Multimedia Appendices 8 and 9.

Below, we provide results first for KOA studies and then for rehabilitation post-TKA. Findings for each condition are grouped according to intervention type (VR or IDR) and clinical mechanism. Detailed results are found in Multimedia Appendices 10 and 11.

Twelve trials evaluated XR interventions for chronic knee pain due to osteoarthritis. Studies involved VR (k=2) [39,45] or IDR (k=10) [28-30,32,34,36,38,41,42,44,49] interventions compared with conventional therapy. Trial characteristics and main findings are summarized in Table 3.

Nine trials evaluated XR for rehabilitation after TKA. Four trials examined VR interventions, incorporating modalities of DAE [37,43], passive distraction [33], and psychoeducation [40]. Five trials examined IDR [31,35,46-48], all of which evaluated the effectiveness of DAE. Trial characteristics and main findings are summarized in Table 4. Certainty of evidence tables are found in (Multimedia Appendix 7). Below, we first describe findings for VR intervention trials, organized by clinical mechanism, and then present findings for IDR studies.

Compared with standard rehabilitation, IDR-DAE may result in better pain-related functioning at 3‐4 months (low COE) and probably results in better functioning at 6 months (moderate COE and 1 randomized controlled trial [RCT]). IDR may result in little to no difference in pain intensity at 2‐3 weeks (low COE), 3‐4 months (low COE), or 6 months (moderate COE), and the evidence for AEs remains very uncertain (Multimedia Appendix 7). Five trials evaluated IDR interventions for patients post-TKA [31,35,46-48]. Three used Microsoft Kinect-based systems [31,35,48] and 2 used other sensor technologies that monitored participant movements and provided digital feedback [46,47] (Table 4). Studies ranged from small to large (n=52‐306) and follow-up ranged from 2 weeks to 6 months. Four [31,35,47,48] were home-based, and one [46] was clinic-based. Four studies [31,35,47,48] reported on pain-related functioning, and 4 [35,46-48] reported on pain intensity; only one reported on AEs. One study [31] included patients who had undergone total arthroplasty of the knee or hip.

Four studies [31,35,47,48] evaluated pain-related functioning using WOMAC [31,48], Oxford Knee Score (OKS) [35], or KOOS [47]. At 3‐4 months, findings broadly favored IDR (low COE). Eichler et al [31] (Germany, high RoB; N=111) and Shim et al [48] (South Korea, some concerns RoB; N=56) both found WOMAC improvement in both groups at 3 months, with small reductions favoring IDR (eg, standardized Diff Δ −0.29 [48]). Likewise, Janhunen et al [35] (Finland, some concerns RoB; N=52) reported greater OKS improvement in the IDR group at 4 months (standardized Diff Δ 0.3). Prvu Bettger et al [47] (United States, some concerns RoB; N=306)—the largest study in the review and the only US-based trial—found greater KOOS improvement in the traditional PT group at 6 weeks (Diff Δ −1.8) but a difference favoring IDR at 12 weeks (Diff Δ 1.4). Notably, Prvu Bettger et al [47] was industry-funded (ReflexionHealth), and the primary outcome was total health care costs rather than pain or function. At 24 weeks, Shim et al [48] (moderate COE) reported continued WOMAC improvement favoring IDR (Diff Δ −2.8)—the longest follow-up and the highest certainty rating for any IDR outcome in the post-TKA analysis. Piqueras et al [46] (Spain, high RoB; N=181) also assessed participants using WOMAC scores. The authors, however, did not provide follow-up scores, although they stated there were “no significant differences” between groups. This study primarily compared delivery modality (remote vs face-to-face) rather than IDR content.

Four trials assessed pain intensity using VAS or NRS. The evidence is very uncertain on the effect of IDR in pain intensity at 2‐3 weeks (very low COE; k=2) and at 3‐4 months (low COE; k=4). At 6 months (1 RCT), IDR probably resulted in little to no difference (moderate COE; k=1). Pooled analysis at 3‐4 months (k=3) showed no differences between groups (low COE) in reduction of pain intensity (SMD −0.12; 95% CI −0.75 to 0.52; PI −1.63 to 1.27; Figure 4; Multimedia Appendix 7) [35,47,48]. While not included in the pooled analysis due to intervention characteristics, Piqueras et al [46] showed greater VAS reduction in the control arm at 3 months (standardized Diff Δ 0.22; P=.28).

Only Prvu Bettger et al [47] reported on AEs, assessing falls during 12 weeks postdischarge; 19% (n=27) of the IDR group versus 15% (n=20) of the control group experienced one or more falls. Causality was not assessed; falls may reflect age and disease status rather than the intervention (very low COE). No other AE types were assessed in any of the 5 [31,35,46-48] IDR post-TKA studies. With respect to other outcomes, 2 studies assessed quality of life. Shim et al [48] reported EQ-5D scores at 3, 12, and 24 weeks with modest improvements in both groups and no clear between-group difference. Eichler et al [31] reported 36-item Short Form health survey (SF-36) physical component scores (PCS) improving comparably in both groups at 3 months (standardized Diff Δ −0.04), with no change in mental component scores (standardized Diff Δ −0.24). All 5 studies [31,35,46-48] evaluated physical performance using TUG, 6-minute walk, Short Performance Physical Battery (SPPB), ROM, gait speed, and/or strength measures. Results were heterogeneous; Janhunen et al [35] showed greater TUG improvement in the IDR group at 4 months (standardized Diff Δ −0.71; P=.04), whereas Piqueras et al [46] reported greater TUG improvement in the control group at 3 months (standardized Diff Δ 0.51, P=.02). Other physical performance measures showed either no differences or small inconsistent improvements between groups.

This systematic review evaluated the benefits and harms of XR interventions for chronic KOA pain and for symptomatic recovery after TKA, with findings stratified by level of immersion, clinical mechanism, and follow-up duration. We identified 21 eligible RCTs evaluating XR interventions for KOA pain or post-TKA rehabilitation. Despite a substantially expanded evidence base compared with previous reviews, the overall body of evidence remains limited by small study sizes, heterogeneity in intervention design and comparators, and lack of AE reporting in post-TKA. Most studies were rated as having high (k=9) or some concerns (k=11) for RoB, with only one study rated low. Findings should therefore be interpreted in light of the predominantly high or some concerns of RoB among the included studies, the low to very low GRADE certainty of evidence for most pooled outcomes, as well as the heterogeneity across interventions and follow-up durations. Pooled point estimates with 95% CIs reflect the average effect across the included studies, while prediction intervals describe the plausible range of true effects in similar future settings [60].

For adults with KOA, IDR interventions showed better pain-related functioning (k=5; low COE) and lower pain intensity (k=4; low COE) at 6‐8 weeks, compared with conventional therapy. The observed standardized effect sizes fall in the small-to-moderate range [56]. A consistent finding in both KOA and TKA studies was that both groups improved substantially over time, as assessed with established outcome measures (WOMAC, KOOS, and VAS), and met minimal clinically important differences (MCIDs) for these various measures (Multimedia Appendix 13 [61-65]). However, observed differences in change scores between groups (Diff Δ) typically were small and much less than MCID, though there should be caution in using MCID to interpret between-group differences [66]. Among the 10 IDR studies [28-30,32,34,36,38,41,42,44,49], the single low-RoB trial [29] did not report on WOMAC, and thus the pooled estimate of effects on pain-related functioning depended on data for studies with high or some concerns for RoB. VR-DAE for KOA (k=2) produced inconsistent results: short-term clinic-based benefits reported by Özlü et al [45] and null findings in a longer home-based trial reported by Lo et al [39] raise questions about the sustainability and generalizability of VR effects for KOA. Heterogeneity in intervention content (gaming vs guided strengthening), setting (clinic vs home), and duration (3 vs 12 weeks) limited comparability and synthesis.

Post-TKA evidence similarly indicates that IDR may result in better pain-related functioning at 3‐4 months (low COE), and this benefit probably persists at 6 months (moderate COE; 1 RCT)—a trajectory suggesting durable functional gains from IDR-guided rehabilitation. For pain intensity, the pooled meta-analysis at 3‐4 months showed little to no difference between groups (k=3). At 6 months, pain intensity probably showed little to no difference (moderate COE; 1 RCT).

Our results broadly align with other recent reviews undertaken to evaluate effects of XR technologies for KOA. Wei et al [18] evaluated immersive and nonimmersive VR technologies for patients with KOA and found improvements in pain and WOMAC total scores (moderate and low COE). Byra et al [19] evaluated rehabilitation using VR and exergames in patients with KOA and found “no conclusive evidence” that XR interventions were more beneficial than standard rehabilitation. Both reviews included studies evaluating a range of XR technologies encompassing both VR and nonfully immersive modalities, although only Wei et al [18] stratified outcomes by degree of immersion. Wei et al [18] also presented subgroup analyses for pain and WOMAC scores by duration of therapy; however, in contrast with our review, neither of the previous reviews stratified results by duration of follow-up, which is of particular importance when assessing effects of an intervention for a chronic pain condition [67]. Neither review evaluated AEs. Two recent reviews evaluated XR for post-TKA but neither stratified results by level of immersion. Su et al [20] and Peng et al [21] reported similar results for XR versus conventional rehabilitation on measures of pain and functioning (WOMAC). Pain scores improved in intervention arms at early time points, ≤2 weeks and ≤1 month, respectively, but neither review reported positive results at later time points. WOMAC scores showed improvement at 1 month in both reviews but disagreed at later time points (>1‐6 months) and showed considerable heterogeneity. Though a relatively small number of studies were included in Su et al [20] (k=14) and Peng et al [21] (k=8), only the latter commented on the quality of evidence (Physiotherapy Evidence Database [PEDro] scale for differences in pain (low-quality) and function (high-quality) scores. Neither review evaluated AEs. By stratifying findings by level of immersion (VR vs IDR), clinical mechanism, and follow-up duration, and by appraising certainty of evidence using GRADE and systematically evaluating AEs alongside efficacy, the present review extends and complements these prior syntheses and helps clarify which XR modalities and indications are best supported by the current evidence.

The lack of high-quality (ie, low RoB) studies evaluating XR interventions for KOA and post-TKA recovery is an important gap to address in order to improve care of these populations. XR interventions are being widely tested and deployed in a variety of other acute and chronic pain conditions, but insurance coverage for potentially beneficial interventions will be contingent on sufficient evidence to demonstrate benefit (and possibly cost-effectiveness) [68-71]. Given the rapid development, deployment of, and diminishing costs of XR hardware and software, there is also a critical need for rigorous evaluation of associated AEs relative to their benefits so we can safely deploy these devices in clinical settings.

Defining the characteristics of interventions by level of immersion (VR vs IDR) can be challenging, as existing frameworks for XR technologies sometimes differ in where boundaries are drawn [72]. Although most previous systematic reviews of XR interventions for pain did not stratify results by level of immersion, we sought to operationalize the distinction between full immersion (ie, VR) and nonimmersive digitally augmented experiences (ie, IDR) in order to provide greater clarity on benefits and harms attributable to each group. To do this, we relied on author descriptions of interventions, which were sometimes limited. To ensure that a broad range of clinically relevant XR interventions were evaluated, we included IDR interventions that many would consider minimally immersive (eg, Microsoft Kinect), which could potentially dilute the impact of more immersive XR interventions (that were not VR). Finally, we limited eligibility to English-language studies, but we did identify studies conducted in predominantly non–English speaking countries.

The evidence on XR interventions for pain related to KOA or TKA is hampered by small sample sizes, short duration of follow-up, and lack of robust reporting on AEs. Methodological concerns were frequently encountered in the included studies, including most frequently missing outcome data and deviations from the intended interventions. Higher-quality studies addressing these concerns are needed to reliably compare the effects of XR to non-XR therapies.

Consistent with the early state of the science in an emerging field, most of the studies included in this report, along with 3-quarters of ongoing or recently completed trials on XR interventions for chronic pain, were small (n<100) [73]. Small sample sizes limit the effects of randomization on balancing baseline measures and unmeasured confounding [74]. As described above, we compared between-group change in outcome measures to avoid imbalances created by small sample size and imbalanced confounders, but between-group change is still subject to some amount of confounding. Future studies with sample sizes sufficient to effectively balance participants by measured and unmeasured confounders will facilitate rigorous assessment of the efficacy of IDR interventions for KOA and recovery post-TKA.

Included studies also were limited to short-term follow-up, with most (n=16) only reporting on outcomes at less than 6 months. This renders confidence in the effects of XR in the longer-term treatment of KOA, a chronic disease, uncertain. Further, assessment of whether XR interventions improve medium- and long-term outcomes post-TKA is important when considering the added cost, complexity, and potential for AEs entailed by XR. Longer trials are needed to adequately assess the effects of XR interventions on symptoms of KOA and recovery after TKA.

Lack of reporting of AEs is a critical gap to address in future research, particularly for VR interventions and for post-TKA populations, as AEs are an important component of the patient experience and often influence whether someone will start or continue an intervention [75]. AEs should be assessed systematically and reported for each arm and involve participant interviews with open-ended questions and/or checklists [76]. Accurate observation of rare but potentially serious AEs will require substantially larger studies but is an essential safety assessment, particularly for the TKA population performing intensive daily exercise programming early after surgery.

Finally, only 6 [33,37,39,40,43,45] of 21 included studies evaluated the effects of more immersive (VR) technologies. This is a critical evidence gap given the pace at which VR technologies are evolving and the high likelihood that they will be applied to areas of clinical need as the barriers to developing high-quality software to support them are lessened. Further, no study compared VR and IDR interventions to explore the potential for relative effects of higher levels of intervention immersion. These comparative effectiveness studies are needed given the increased complexity and cost of VR relative to IDR interventions.

In patients with KOA, IDR interventions may result in small improvements in pain-related functioning and pain intensity at 6‐8 weeks compared with conventional therapy (low COE). Evidence for VR interventions in KOA is limited to 2 trials with inconsistent results. For post-TKA rehabilitation, IDR may result in better pain-related functioning at 3‐4 months (low COE), with benefits probably persisting at 6 months (moderate COE; 1 RCT). IDR may result in little to no difference in pain intensity at 2‐3 weeks and 3‐4 months (low COE) and probably results in little to no difference at 6 months (moderate COE; 1 RCT). VR-delivered psychoeducation showed promise for post-TKA pain reduction in a single trial (moderate COE) but requires replication. AE reporting was absent in the majority of studies but suggested acceptably low rates in those that did (high and low COE for KOA and post-TKA, respectively). Larger high-quality RCTs with longer follow-up, standardized outcome reporting, and systematic safety monitoring are needed before XR interventions can be recommended as standard-of-care adjuncts for recovery after both KOA and TKA. By organizing findings around level of immersion, clinical mechanism, and follow-up duration, as well as systematically appraising AEs alongside efficacy outcomes, this review provides clinicians, patients, and policymakers with an actionable framework for translating current evidence into future practice. The most defensible near-term clinical implication is that IDR-based exercise is reasonable to consider as an adjunct to conventional KOA care in real-world settings, while broader VR adoption and routine integration of XR into post-TKA rehabilitation should await additional high-quality evidence with longer follow-up periods and systematic safety monitoring.