A comprehensive, systematic literature search was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Matthew et al., 2021). Four major electronic databases were searched from inception to 1 June 2025: PubMed, Web of Science Core Collection, Cochrane Central Register of Controlled Trials (CENTRAL), and EMBASE. The search strategy employed a combination of Medical Subject Headings (MeSH) terms, EMTREE terms (for EMBASE), and relevant keywords related to the population, intervention, and study design. Key concepts included: Population: Terms related to the target population were combined with OR: “burn” OR “thermal injur*” OR “child*” OR “pediatric” OR “paediatric” OR “adolescen*” OR “youth”. Intervention: Terms describing the intervention were combined with OR: “virtual reality” OR “VR” OR “augmented reality” OR “AR” OR “mixed reality” OR “immersive technology” OR “computer simulation” OR “distraction therapy”. Outcome: Terms for relevant outcomes were combined with OR: “pain” OR “pain intensity” OR “distress” OR “anxiety” OR “range of motion” OR “ROM” OR “rehabilitation” OR “physiotherapy” OR “procedure time” OR “heart rate”. Study Design: Terms specifying the study design were combined with OR: “randomized controlled trial” OR “RCT”. Crucially, the four distinct concepts (Population, Intervention, Outcome, Study Design) were combined using the Boolean operator AND. Thus, the overall search logic was: (Population terms) AND (Intervention terms) AND (Outcome terms) AND (Study Design terms). The details search terms combinations were included in Table 1. Additional searches were conducted as follows: manually screening the reference lists of all included studies and relevant systematic reviews, searching clinical trial registries (ClinicalTrials.gov, WHO ICTRP), and contacting experts in the field to identify unpublished or ongoing studies. All identified records were imported into EndNote software for deduplication and screening.

Two independent reviewers (initials blinded) screened titles and abstracts against predefined inclusion criteria. Full texts of potentially eligible records were then assessed independently by the same reviewers. Disagreements were resolved through discussion or consultation with a third reviewer. Inclusion Criteria: (Za et al., 2024): Study Design: RCT, including parallel-group, cross-over, or cluster-RCTs, regardless of blinding; (van der Heijden et al., 2018); Population: Pediatric patients (aged ≤18 years) with acute or sub-acute burn injuries undergoing any of the following procedures: wound care (e.g., cleansing, debridement), dressing changes, physical therapy, occupational therapy, or other active rehabilitation sessions targeting functional recovery (Ciornei et al., 2023). Intervention: Any form of VR, including immersive VR (head-mounted displays), semi-immersive VR (e.g., large projection screens), or augmented reality (AR), used during the target procedure as an adjunctive intervention (Paul et al., 2021). Comparator: Standard care. Standard care, as defined by the control groups in the included studies, refers to the routine clinical management of pediatric burn procedures at participating institutions. It encompassed the following interventions according to institutional protocols: Pharmacological analgesia, including systemic analgesics (e.g., opioids such as morphine or fentanyl) or non-opioids (e.g., acetaminophen or ibuprofen), as well as topical anesthetics (e.g., lidocaine); Conventional non-pharmacological distraction, involving passive techniques (e.g., television, music) or active strategies (e.g., interactive toys, nurse-led play therapy); and Combined approaches that integrate pharmacological and behavioral interventions (Raith and Hochhaus, 2004). Outcomes: Studies must report data on primary outcome: pain intensity. Studies reporting data on any secondary outcome were also included: anxiety, range of motion (ROM). procedure time, heart rate (HR) measured via pulse oximetry or ECG (beats per minute - bpm) (Mendell, 2014). Language: English language publications. Exclusion Criteria: Non-randomized studies (e.g., case reports, case series, observational studies). Studies exclusively on adults (>18 years) or where pediatric data could not be separated. Studies using VR for purposes other than during the target procedure (e.g., preoperative preparation, psychological therapy sessions separate from wound care/rehab). Studies lacking a relevant comparator group and studies that do not report quantitative data for at least one outcome.

Two independent reviewers extracted data onto a standardized, piloted electronic form developed in Covidence and Microsoft Excel. Extracted data encompassed the following categories: study characteristics (e.g., author, publication year, country, study design, sample size per group, participant age range), intervention details (e.g., type of virtual reality technology, procedural specifics [burn wound care, dressing change, or physical therapy]), and associated outcomes. Corresponding authors were contacted via email twice over 2 weeks to request missing data or clarifications.

The methodological quality of each included RCT was independently assessed by two reviewers using the revised Cochrane Risk of Bias tool for randomized trials (RoB 2.0). This tool evaluates bias across five domains (Za et al., 2024): Bias arising from the randomization process (van der Heijden et al., 2018). Bias due to deviations from intended interventions (effect of assignment to intervention) (Ciornei et al., 2023). Bias due to missing outcome data (Paul et al., 2021). Bias in measurement of the outcome (Raith and Hochhaus, 2004). Bias in selection of the reported result. Judgments (“Low risk of bias”, “Some concerns”, or “High risk of bias”) were made for each domain following the RoB 2.0 algorithm, leading to an overall risk of bias judgment for each study and for each outcome. Disagreements were resolved through discussion or with a third reviewer.

Meta-analyses were performed using Stata version 18.0, employing random-effects models to account for anticipated clinical and methodological heterogeneity. Statistical heterogeneity was determined using Q-test and the I² statistic. The endpoints were analyzed as continuous outcomes. Standardized Mean Differences (SMD) with 95% Confidence Intervals (CI) were calculated using Hedges’ *g* to account for different measurement scales (e.g., VAS vs FPS-R for pain). Sensitivity analyses were performed to assess the robustness of the results. Forest plots were generated for primary outcome meta-analyses.

The PRISMA flow diagram (Figure 1) outlines the systematic screening process. The initial search yielded 749 records. Following deduplication and title/abstract screening, 49 full-text articles were evaluated for eligibility. Ultimately, 16 RCTs (Debashish et al., 2005; Engle Angela et al., 2007; Jonathan et al., 2008; Yuko et al., 2010; Belinda et al., 2012; Debra et al., 2014; Yun et al., 2015; Waked and Eid, 2019; Hunter et al., 2021; Christelle et al., 2020; Henry et al., 2021; Rania et al., 2021; Merve and Zeynep, 2022; Ümmühan and TBJJBCR, 2024; Megan et al., 2022; Abd Elrazek and Ghada, 2020)published between 2005 and 2024 were included. These studies involved pediatric burn patients from Australia (3 studies) (Debashish et al., 2005; Jonathan et al., 2008; Belinda et al., 2012), North America (6 studies) (Yuko et al., 2010; Debra et al., 2014; Hunter et al., 2021; Christelle et al., 2020; Henry et al., 2021; Megan et al., 2022), Egypt (3 studies) (Waked and Eid, 2019; Rania et al., 2021; Abd Elrazek and Ghada, 2020), and Asia (4 studies) (Engle Angela et al., 2007; Yun et al., 2015; Merve and Zeynep, 2022; Ümmühan and TBJJBCR, 2024). The study designs consisted of 4 cross-over RCTs and 12 parallel RCTs. The key characteristics of the included studies encompassed multiple variables: among the procedures, 3 focused on physical therapy, 2 on burn wound care, 1 on burn wound debridement, and the remaining 10 on burn dressing changes. VR applications included both immersive and non-immersive VR games and VR videos. All studies reported data on pain intensity. Additionally, 2 studies each provided data on anxiety (Merve and Zeynep, 2022; Ümmühan and TBJJBCR, 2024)and procedure time (Yun et al., 2015; Abd Elrazek and Ghada, 2020), 2 studies focused on range of ROM (Waked and Eid, 2019; Rania et al., 2021), and 4 studies reported HR data (Hunter et al., 2021; Merve and Zeynep, 2022; Ümmühan and TBJJBCR, 2024; Abd Elrazek and Ghada, 2020).

Pain intensity was assessed in all sixteen studies using the following measurement tools: the adolescent pediatric pain tool (APPT), the adolescent pediatric pain tool word graphic rating scale (APPT-WGRS), the self-report Faces Scale (FACES), the Face, leg, activity, cry and consolability (FLACC), the faces pain scale-revised (FPS-R), the graphic rating scale (GRS), the numeric rating scale (NRS), the wong baker faces rating scale (WBFRS), and the visual analog scale (VAS). Two studies assessed anxiety descriptors using the following measurement tools: the faces anxiety scale (FAS) and the children’s fear scale (CFS). Procedure time was assessed in the two studies, ROM was assessed in the two studies, HR was assessed in four studies (Table 2).

Among the 16 included RCTs, risk of bias across seven domains (Figure 2) showed variable profiles. Random sequence generation, incomplete outcome data, selective reporting, and other bias mostly presented low risk, ensuring sound randomization, data handling, and transparency. Allocation concealment had some unclear risk, raising doubts about selection bias prevention. Blinding of participants and personnel carried high risk due to VR’s unblindable immersive nature, potentially biasing behavior and reporting. Blinding of outcome assessment mixed low and unclear risks, leaving uncertainty about assessor blinding. While randomization and data integrity were well - controlled, performance bias from unblinded exposure and unclear detection bias threaten reliability, especially for subjective outcomes (e.g., pain, anxiety). Future trials should prioritize blinded assessment strategies to address these gaps.

Analysis of the primary outcome showed that VR intervention significantly reduced pain intensity compared to standard care, with a pooled standardized mean difference (SMD) of −1.58 (95% confidence interval [CI]: −2.11, −1.05), although substantial heterogeneity was observed (I² = 90.0%, P < 0.1). Subgroup analysis by procedure revealed that in the burn dressing change subgroup, VR demonstrated an SMD of −1.56 (95% CI: −2.19, −0.93; I² = 87.9%, P < 0.1); the physical therapy subgroup showed a stronger effect (SMD = −2.79, 95% CI: −5.45, −0.13) but with significantly increased heterogeneity (I² = 96.6%, P < 0.1). The burn wound care subgroup showed no significant effect (SMD = −0.56, 95% CI: −1.01, −0.10; I² = 36.0%, P = 0.210), with reduced heterogeneity (Figure 3).

For secondary outcomes, VR intervention significantly alleviated anxiety compared to standard care (SMD = −1.89, 95% CI: −3.13, −0.65; I² = 88.3%, P < 0.1), though high heterogeneity persisted (Figure 4). Procedure time was shorter in the VR group (SMD = −1.09, 95% CI: −1.76, −0.42; I² = 67.8%, P < 0.1), with moderate heterogeneity (Figure 5). No significant differences were found between VR and standard care in HR (Figure 6) changes (SMD = −0.64, 95% CI: −0.89, −0.39; I² = 0.0%, P = 0.433) or ROM (Figure 7) changes (SMD = 1.40, 95% CI: 0.90, 1.90; I² = 0.0%, P = 0.614), both showing no heterogeneity.

Results found removing each study did not change the pooled effect size of VR on pain intensity of burns in children (95% CI: −2.11 to −1.05) (Figure 8).

Although visual inspection of the funnel plot suggested a potential publication bias (Figure 9), the result of Begg’s statistical test did not reach significance (P = 0.012).