Virtual reality-based interventions enhance functional status, balance and muscle strength in patients with ankle instability: a systematic review with meta-analysis of randomized controlled trials

Abstract

INTRODUCTION

Ankle instability reduces functional status in athletes and overall population after orthopedic ankle injuries. Complementary to physical therapy approaches, virtual reality-based interventions (VRBI) that promote exercises in ludic and gamified environments, are being used in reducing ankle instability during ankle rehabilitation. The aim of this systematic review with meta-analysis was to assess the effectiveness of VRBI in improving ankle function status, postural balance and muscle strength in patients with ankle instability.

EVIDENCE ACQUISITION

A systematic review with meta-analysis, previously registered in PROSPERO (CRD42023460797) was conducted aligned with the PRISMA guidelines. Literature search was performed in PubMed Medline, SCOPUS, WOS, CINAHL and PEDro without year of publication and language restrictions up to November 2024. Randomized controlled trials (RCTs) including patients with ankle instability which assessed the effectiveness of VRBI, compared to others, on ankle function status, postural balance, muscle strength, were included. Methodological quality and risk of bias were assessed using the PEDro Scale. Pooled effect was calculated with the Cohen’s standardized mean difference (SMD) and its 95% confidence interval (95% CI) in random-effects models, or mean difference (MD) if studies used the same measure. All methodological phases of this meta-analysis were conducted by peers.

EVIDENCE SYNTHESIS

Twelve RCTs, providing data from 555 participants with ankle instability (89% males with a mean age of 21.8±6.2 years), were included. The mean methodological quality of the studies included was moderate (5.8±1.3 in the PEDro scale). Our meta-analysis showed that VRBI is largely effective in increasing ankle function status (SMD=0.86; 95% CI 0.24 to 1.49; P=0.007), static (SMD=1.6; 95% CI 0.74 to 2.4; P<0.001) and dynamic balance (SMD=0.81; 95% CI 0.36 to 1.25; P<0.001), and strength of muscles involved in plantar flexion (MD=5.76; 95% CI 0.22 to 11.3; P=0.042) and eversion movements (MD=1.7; 95% CI 0.5 to 2.9; P=0.005).

CONCLUSIONS

VRBI is a safe and effective therapy for improving ankle function status, static and dynamic balance and strength of the muscles involved in plantar flexion eversion movements in patients with ankle instability.

Introduction

Functional ankle instability (FAI), as defined by Freeman (1965), is characterized by a subjective feeling of the ankle joint “giving way.”^1-3 Some studies have equated FAI with chronic ankle instability (CAI),^4-6 a complex pathological condition affecting both athletes and the general population with a prevalence ranging from 15% to 48%,⁷ being more common among female athletes than male athletes.⁸ CAI is the most prevalent consequence of acute lateral ankle sprains (LAS).⁹ The high recurrence rate of LAS (18-47% of cases) is closely linked to the presence of CAI in 40-70% of patients who have experienced an LAS within a year of the initial injury.^{10, 11} CAI involves intricate neurological and biomechanical factors. Neurologically, there is a disruption of motor and somatosensory inputs that reduces the ability to process sensory information from mechanoreceptors (proprioceptors and exteroceptors) in the joint capsule, muscles, tendons, and ligaments of the foot.¹² Consequently, the efferent response from the central nervous system, crucial for postural control and ankle joint stability, becomes impaired, leading to ineffective movement strategies.¹³ Deficits in anticipatory postural control during movements, gait, or instability situations contribute to the persistence of this pathological condition associated with ankle joint instability and recurrent LAS. Individuals with CAI often adopt compensatory biomechanical strategies to maintain ankle stability, such as reduced ankle dorsiflexion and increased ankle inversion movements during gait and functional activities. Pain in the ankle area, ankle muscle weakness, restricted ankle range of motion (ROM), impaired proprioception and neuromuscular control, and a subjective perception of ankle instability and frailty can lead to recurrent LAS, constituting the most disabling consequences experienced by patients with CAI.^{14, 15} Patients with CAI have reported diminished postural balance in addition to reduced gray matter volume in the somatosensory cortex (parietal lobe) and thalamus compared to healthy subjects.¹⁶ Additionally, ankle joint laxity, which can lead to excessive ROM after LAS, may contribute to post-traumatic ankle osteoarthritis,¹⁷ further perpetuating the clinical condition. As a result, these patients often experience reduced functional status in the ankle, impaired balance control, and a decline in physical activity level and quality of life.¹⁸

Optimal rehabilitation for CAI should encompass an initial assessment, followed by early-phase rehabilitation focused on proprioceptive and balance deficits, progressing to strength and neuromuscular training. The final phase should involve advanced functional training related to activities of daily living or specific sports skills.¹³ Physiotherapy plays an essential role in the rehabilitation of CAI symptoms using a widely variety of conservative interventions, including passive mobilizations, manual therapy,¹⁹ neuromuscular electrical stimulation,^{20, 21} and others related such as taping whole body vibration.²² However, active approaches that require active movement exercises in diverse environmental conditions are generally recommended.²³ Recent literature demonstrates that physical and active training programs combining balance, strengthening and proprioceptive exercises increase ankle function with lower risk of recurrence.^24-26 Nonetheless, individuals often report that some of these recommended exercises are repetitive and monotonous.²⁷ Additionally, a lack of correlation between physiotherapy and long-term subjective recovery has been observed,²⁸ potentially due to decreased patient motivation and low adherence.

Virtual reality (VR), a concept pioneered by Ivan Sutherland in the 1960s,²⁹ enables users to visualize and interact with virtual and realistic environments through advanced hardware and software.³⁰ The effectiveness VR devices as an effective therapy approach, also called virtual-reality based interventions (VRBI), has been assessed for different fields of rehabilitation.³¹ VR devices provides interactive, gamified simulations of virtual environments perceived as real by users, fostering interaction with virtual objects. According to the level of immersion and presence, VR systems can be categorized as non-immersive (NIVR), semi-immersive and immersive (IVR),³² being NIVR and IVR the most used in musculoskeletal rehabilitation. On the one hand, NIVR systems, often utilizing commercial gaming consoles like Nintendo, Xbox, or Biodex, project virtual environments onto traditional two-dimensional screens, with user interaction facilitated by joysticks or hand controllers.³³ Due to their lower cost and ease of use, these systems are widely adopted in rehabilitation settings. Conversely, IVR systems, such as the Meta Quest 2 and 3 or Apple Vision Pro, provides the most realistic experiences to the users employing head-mounted displays to present three-dimensional virtual environments in 360°, offering enhanced levels of interaction, immersion, and presence.³⁴ In rehabilitation, VRBI offer several advantages, including the ability to deliver functional activities in gamified, engaging virtual environments within rehabilitation centers or remotely (tele-rehabilitation) increasing the adherence of the patients to the rehabilitation. Besides, VRBI can be tailored to individual patient needs, allowing for precise dosing of intensity, exposure time, and activity difficulty.³⁵

Integrating VRBI, through exergames or commercial video games that require physical movement into conventional physiotherapy, may accelerate recovery in patients with CAI. Numerous studies have demonstrated the effectiveness of VRBI in some orthopedic lower limb injuries, such as knee and hip arthroplasty,^{36, 37} osteoarthritis,³⁸ anterior cruciate ligament rupture,³⁹ and others. In the rehabilitation of the orthopedic ankle joint injuries, such as LAS or ankle instability, VRBI has emerged as a therapeutic promising tool.⁴⁰ A 2023 systematic review by Elaraby et al., encompassing 10 studies (4 of which focused in patients with CAI were included in a meta-analysis) revealed that VRBI was more effective than conventional physiotherapy in improving static balance but not dynamic balance in CAI.⁴¹ The authors suggested that VRBI could be beneficial for enhancing ankle function, although this conclusion was not directly supported by the meta-analysis. However, future meta-analyses, focusing only on patients with CAI/FAI, may yield different results due to the inclusion of newer studies and potentially overlooked studies from the original review. Therefore, the aim of our systematic review with meta-analysis was to assess the effectiveness of VRBI in improving ankle function status, postural balance and muscle strength in patients with ankle instability.

Evidence acquisition

Preliminary design and guidelines

To conduct this systematic review with meta-analysis, the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines⁴² and the Cochrane Handbook for Systematic Reviews of Interventions⁴³ guidelines were followed. Besides, the overall methodological quality of this review was evaluated using the AMSTAR 2 checklist⁴⁴ and its protocol was previously registered in PROSPERO: CRD Finally, the methodological protocol of this review obtained the following registration number in PROSPERO: CRD42023460797.

Literature search: databases, other sources and search terms

The literature search was carried out by two authors (MGS and ICP), independently, in PubMed Medline, Scopus, Web of Science (WOS), Scopus, CINAHL Complete and PEDro. (Physiotherapy Evidence Database) between September and November 2024, without applying filters related to the year of publication and language with the aim to avoid missing potential studies to be included. Additionally, other sources such as the reference-lists of previously published studies, proceedings or congress abstracts. The PICOS tool was taken into account to design the search question:⁴⁵ population (individuals with ankle instability), intervention (VRBI), comparison (conventional therapy [CT] or usual care), outcomes (see variables subsection) and study design (randomized controlled trials [RCTs] or RCTs pilot). The search strategy was conducted selecting only two conditions from PICOS tool, intervention and population, with the aim to increase the sensitivity of the search, retrieving major number of potential studies to be included. Therefore, keywords employed in the search, according to MeSH thesaurus, were “ankle instability,” “virtual reality” and “virtual reality-based therapy,” and with related entry-terms (Supplementary Digital Material 1: Supplementary Table I). Boolean operators “AND” and “OR” were used in the searches. Finally, all this process was supervised by a third expertise author (EOG).

Study selection: inclusion and exclusion criteria

All studies retrieved from literature search were screened by title and abstract by two authors (MGS and ICP), independently. The Cohen’s kappa coefficient (κ),⁴⁶ that can range from non-existent (κ=0) to excellent (0.8<κ≤ 1).⁴⁷ A third author was in charge to solve doubts between authors (EOG).

To include a study in the meta-analysis, it had to meet all PICOS criteria: Population, patients with ankle instability; intervention, VRBI; comparison, CT or usual care; outcomes, balance skills, muscle strength, muscular performance and satisfaction with the therapy; and study design (RCTs or RCTs pilots). As the last inclusion criteria, all studies included must provide quantitative data (mean and standard deviation) of the variables of interest to carry out the meta-analysis. As exclusion criteria we propose to exclude studies in which the sample was not homogeneous (mixing ankle instability with others lower limb diseases).

Data extraction

The data from the studies included in this meta-analysis was carried out by two authors (MGS and MCOP), independently, using a standardized data collection-form. A third author participated in this search solving disagreements between the involved authors. The extracted were categorized according to: overall characteristics (authorship, year of publication, country, setting, funding and blinding); characteristics of the sample (number of groups, number of participants per group, age and sex); characteristics of the VRBI (modality of VR device used, specific game’s name, number of sessions, sessions per week and duration of each session); and characteristics of the comparison intervention (type of the intervention and number of sessions, sessions per week and duration of each session). Related to the outcomes were compiled the name of variable, the measurement test employed, the mean and standard deviation post-intervention to conduct the meta-analysis and the time-point in which assessment was made.

Outcomes

In this systematic review with meta-analysis the following outcomes were assessed: 1) ankle function status, defined as the degree of disability of the ankle joint; 2) postural balance, focusing on two specific dimensions: static (ability to maintain the body in position without movement or displacement) and dynamic balance (ability to maintain standing and stable during movement or displacement); 3) muscle strength of the muscles involved in plantar flexion, dorsiflexion, eversion and inversion; 4) overall lower limb muscular performance.

Assessment of the methodological quality, risk of bias and quality of evidence

These assessments were carried out separately by two authors (MCLR and ADF), and disagreements were consulted by a third author (ICP). First, the methodological quality of the studies included was assessed using the PEDro Scale, that comprises 11 items.⁴⁸ The total score obtained for one study ranges from 0 to 10 adding the individual scores of items 2 to 11 (item 1 only related to external validity). According to the score, the methodological quality can be excellent (10-9 points), good (8-6 points), moderate (5-4 points), and poor (3 points or less).⁴⁹ Besides, items 2-3, 5-6, and 7 can be useful to recognize the presence of selection, performance and detection biases, respectively.

Second, the level of evidence reported by each meta-analysis was assessed applying the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) statement⁵⁰ and the checklist of Meader (2014).⁵¹ To determine the level of evidence, five items are taken into account: risk of bias of each study included, and four measures derived for the meta-analysis, such as inconsistency, imprecision, evidence indirectness and risk of publication bias. Therefore, the level of evidence of each finding’s meta-analysis can be quantified as: 1) high, if findings are robust; 2) moderate, if results may change including new studies; 3) low, if the level of confidence in our pooled effect is very slight; and 4) very low, when any effect estimation is robust. For each item that was not met, the level of evidence was downgraded one level.

Statistical analysis

The meta-analysis was conducted by two authors (EOG and MGS) using Comprehensive Meta-Analysis version 4 (Biostat, Englewood, NY, USA).⁵² The meta-analysis of an outcome only was performed if at least 2 comparisons were provided.⁴³ The Cohen’s standardized mean difference with its 95% confidence interval (95% CI) was used as pooled effect measure when studies in a random-effects model for continuous data.^{53, 54} The effect size was interpreted according Kinney et al. interpretation proposed for rehabilitation studies: null (SMD 0), small (SMD 0.08-0.15), medium (SMD 0.19-0.36) and big (SMD >0.4).⁵⁵ Each meta-analysis was graphically displayed in the forest plot.⁵⁶ Secondly, the mean difference (MD) was used to estimate the pooled effect when the studies included in the meta-analysis reported the same outcome measured with the same test. As risk of publication bias is recommendable to be evaluated using more than one method, when 10 or less studies are included, we assessed it using the following tests. The funnel plot, P value for Egger test and the trim-and-fill estimation were used to determine the risk of publication bias.^57-59 Trim-and-fill allows to determine the adjusted pooled effect taking into account the risk of publication bias. Besides, if the difference between original and adjusted pooled effect is major 10%, the level of evidence will be downgraded one level.⁶⁰ Inconsistency or heterogeneity was calculated with the degree of inconsistency of Higgins, the χ2 test and its P value.⁶¹ According to this, Heterogeneity can be large (I2>50%), medium (I2 50-25%), low (I2 25-5%) or null I2<5%).⁶² Finally, as additional analysis, a sensitivity analysis using the leave-one-out method was conducted to assess the contribution of each study to the pooled effect.

Evidence synthesis

Study selection

Literature search retrieved 191 records (189 from the five databases consulted, and two from other sources). After removing duplicates, 122 references were screened by title and abstract of which 99 were excluded for not being relevant, and 11 for not meeting the inclusion criteria (reasons in Figure 1). The inter-rater agreement in the study selection process was excellent (κ=0.95). Finally, 12 studies^63-74 were included in the present systematic review with meta-analysis. PRISMA flow diagram, in Figure 1, summarizes the study selection process.

Figure 1.

—PRISMA flow chart of included studies.

Characteristics of the studies included in review

The 12 RCTs included in this review were carried out between 2016 and 2023 in counties such as South Korea,^{68, 69, 72-74} Thailand,⁶³ the UK and Australia,⁶⁴ Iran,^{65, 70, 71} Switzerland,⁶⁶ and Egypt.⁶⁷ These RCTs provided data from 555 participants (89% males) with ankle instability with a mean age of 21.8±6.2 years. All patients included have a history of previous ankle sprain. All patients included prese Five RCTs^{65, 67, 70, 71, 73} provided data from athletes with ankle instability and seven for non-athlete’s patients with ankle instability. Additionally, in 8 RCTs patients reported FAI and CAI in four. Of all, 286 patients received VRBI, and 269 comprised the control intervention. VRBI included non-immersive VR devices, being Nintendo Wii Fit the most frequent, and immersive VR head-mounted displays. Control interventions were comprised by conventional ankle exercises and usual daily activities. The duration of VRBI was heterogeneous from 3 weeks to 3 months, the most common exposition to VRBI was three times per week and 30 minutes per session. All RCTs provided data from immediate assessment post-intervention, and only two RCTs provided data from assessment follow-up (4 weeks after intervention).^{65, 70} Finally, seven RCTs^{63, 65, 66, 70, 72-74} received external funding to develop the research. Supplementary Digital Material 2 (Supplementary Table II) shows more detailed characteristics of the RCTs included in this systematic review with meta-analysis.

Assessment of the methodological quality and risk of bias

The mean methodological quality of the RCTs included was moderate, showing a mean score in PEDro Scale of 5.8±1.3 points. The PEDro score of 5 RCTs^{63-65, 68, 70} was confirmed in the PEDro database. In four RCTs^{64, 65, 69, 74} (33% of all) the methodological quality was good, moderate in seven RCTs^{40, 63, 66, 67, 70, 71, 73} (58%), and poor in one RCT (9%).⁶⁸ The mean risk of bias in these RCTs was medium, identifying selection, performance and detection as the most reported biases. Selection bias was identified in nine RCTs^{63, 65, 67-71, 73, 74} due to item 3 was not met; performance bias was present in all RCTs^63-74 due to participants and therapists were not blinded (items 5 and 6); and detection bias, due to an inadequate blinding of the evaluators, was present in seven RCTs.^{40, 64, 67, 68, 71, 73, 74} Table I shows the PEDro score and the biases reported for each RCT included in this systematic review with meta-analysis.

Table I. —PEDro score of included studies.^63-74.

Study	Items											Total	Quality	Biases
	1	2	3	4	5	6	7	8	9	10	11
Baek69	Y	Y	N	Y	N	N	Y	Y	Y	Y	Y	7/10	Good	Selection and performance
Chuadthong63*	N	Y	N	N	N	N	Y	Y	Y	Y	Y	7/10	Moderate	Selection and performance
Forsyth64*	Y	Y	Y	Y	N	N	N	Y	N	Y	Y	6/10	Good	Performance and detection
Kim72	Y	Y	Y	Y	N	N	N	N	N	Y	Y	5/10	Moderate	Performance and detection
Kim74	Y	Y	N	Y	N	N	N	Y	N	Y	Y	5/10	Good	Selection and performance
Mohammadi70*	N	Y	N	Y	N	N	Y	Y	Y	Y	Y	7/10	Moderate	Selection and performance
Mohammadi65*	N	Y	N	Y	N	N	Y	Y	Y	Y	Y	7/10	Good	Selection and performance
Nam68*	N	Y	N	N	N	N	N	N	Y	Y	Y	3/10	Poor	Selection, performance and detection
Punt66	Y	Y	Y	Y	N	N	Y	N	Y	Y	Y	7/10	Moderate	Performance
Ranjbarzadeh Yamchi71	Y	Y	N	Y	N	N	N	Y	N	Y	Y	5/10	Moderate	Selection, performance and detection
Shousha67	Y	Y	N	Y	N	N	N	Y	N	Y	Y	5/10	Moderate	Selection, performance and detection
Yang73	Y	Y	N	Y	N	N	N	Y	N	Y	Y	5/10	Moderate	Selection, performance and detection

1: Eligibility criteria; 2: random allocation; 3: concealed allocation; 4: baseline comparability; 5: blind subjects; 6: blind therapists; 7: blind assessors: 8. measures of at least one key outcome were obtained from more than 85% of the subjects initially allocated to groups; 9: intention-to-treat analysis; 10: between-group comparisons; 11: point estimates and variability. The eligibility criteria item does not contribute to the total score. *Confirmed in the PEDro database: https://pedro.org.au/spanish/.

Outcomes synthesis and meta-analyses

Ankle function status

Six RCTs^63-68 with 8 independent comparisons provided data from 399 participants (49.9 per study) to assess the effectiveness of VRBI in improving ankle function status using the Cumberland Ankle Instability Tool (CAIT),^{64, 65, 67, 68} and the Foot and Ankle Ability Measure (FAAM).^{63, 66} The meta-analysis showed moderate-quality evidence of a large effect (SMD=0.86; 95% CI 0.24 to 1.49; P=0.007) (Figure 2) favors VRBI in improving ankle function status without heterogeneity (I2=8.5%; Q=7.65; df=7; P=0.36).

Figure 2.

—Forest plot for the effect of virtual reality-based interventions on ankle function status.

Taking into account the possible risk of publication bias identified (Egger P=0.09), trim-and-fill calculation showed an overestimated effect of VRBI (adjusted SMD=0.98; 95% CI 0.38 to 1.58) demonstrating that the original effects was 13% underestimated due to publication bias (Supplementary Digital Material 3: Supplementary Figure 1). Sensitivity analysis showed an equal contribution of each study in pooled effect size.

Postural balance: static and dynamic balance

On the one hand, to assess the effectiveness of VRBI in improving static balance, eight RCTs^{63, 67-69, 71-74} with 11 independent comparisons, providing data from 394 participants (35.8 per study) using the Romberg test,⁶⁹ the single leg stance test (SLST),⁶³ the average value of the Biodex Balance System^® (BBS),^{72, 74} the overall stability index and limits of stability (LOS) values in static conditions,^{67, 68, 73} and the stork test,⁷¹ were included. Our findings showed a large effect (SMD=1.6; 95% CI 0.74 to 2.4; P<0.001; I2=2.2%; Q=10.23; df=10; P=0.42) favors VRBI (Figure 3) without risk of publication bias (Egger P=0.18). Sensitivity analysis did not show variations.

Figure 3.

—Forest plot for the effect of virtual reality-based interventions on static and dynamic balance.

On the other hand, the effectiveness of VRBI in improving dynamic balance was assessed including seven RCTs^{63-65, 69, 71, 72, 74} with 8 independent comparisons that provided data from 287 participants (35.9 per study). The Y-balance test,^{69, 71} the Star Excursion Balance Test (SEBT),^{64, 65} the balance dimension of the Bruininks-Oseretsky Test of Motor Proficiency (BOT-2)⁶³ and the average value of the BBS^® for dynamic conditions,^{72, 74} were the measurements for dynamic balance assessment used by the RCTs included in this meta-analysis. This meta-analysis reported a large effect (SMD=0.81; 95% CI 0.36 to 1.25; P<0.001; I2=0%; Q=6.67; df=7; P=0.46) favors to VRBI (Figure 3) without risk of publication bias (Egger P=0.32). Sensitivity analysis did not showed variations.

Muscle strength in foot movements (plantar flexion, dorsiflexion, eversion and inversion)

Two RCTs^{40, 69} with three independent comparisons provided data to assess the muscle strength in each foot movement using the dynamometer as measurement tool in both studies. per movement with data from 103 patients (34.3 per study). Our meta-analysis revealed statistical significant differences favors VRBI, in increasing the muscle strength for plantar flexion (MD=5.76; 95% CI 0.22 to 11.3; P=0.042: I2=23.7%; Q=2.8; df=2; P=0.25; I2=63.7%; Q=6.2; df=2; P=0.05) and eversion movements (MD=1.7; 95% CI 0.5 to 2.9; P=0.005; I2=7.9%; Q=2.2; df=2; P=0.34) (Figure 4) without risk of publication bias (Egger P=0.37 and P=0.18, respectively). However, no statistically significant differences between therapies were found in increasing muscle strength for dorsiflexion (MD=2.22; 95% CI -0.08 to 4.53; P=0.06; I2=77.6%; Q=8.9; df=2; P=0.01), and inversion movements (MD=0.8; 95% CI -0.72 to 2.4; P=0.3; I2=0%; Q=0.93; df=2; P<0.001).

Figure 4.

—Forest plot for the effect of virtual reality-based interventions on muscle strength of foot movements.

Lower limb muscular performance

Four studies^{63, 70, 71, 73} with four independent comparisons provided data from 162 participants (40.5 per study) to assess the effectiveness of VRBI in increasing lower limb muscular performance using the standing long jump dimension of the BOT-2,⁶³ the Single hop test (SHT),⁷⁰ the Triple hop test (THT),⁷¹ and the stance long jump test (SLJT).⁷³ The meta-analysis did not report statistically significant differences (SMD=0.7; 95% CI -0.4 to 1.76; P=0.22; I2=50%; Q=5.9; df=3; P=0.11) between VRBI and controls (Figure 5). The present risk of publication bias (Egger P=0.01) maintained the non-existent differences between therapies (Supplementary Digital Material 3: Supplementary Figure 2). No differences were found after sensitivity analysis.

Figure 5.

—Forest plot for the effect of virtual reality-based interventions on overall limb muscular performance.

Discussion

The use of VRBI in rehabilitation, utilizing commercial video games, exergames, or serious games, has shown promise as an effective therapy for certain neurological and musculoskeletal diseases, particularly those affecting the lower limbs.^{75, 76} CAI is a common disability among individuals who have suffered ankle injuries, mainly LAS.⁷⁷ VRBI could potentially benefit these patients by reducing disability and improving functional status. While previous research, including a systematic review with meta-analysis, has suggested the effectiveness of VRBI in orthopedic ankle injuries,⁴¹ the evidence for its impact on CAI remains limited. Our hypothesis was that a more recent literature search could identify additional studies not included in the previous review, as well as more recent publications. This could help fill knowledge gaps and provide a more comprehensive understanding of VRBI’s efficacy. Therefore, the primary objective of our systematic review and meta-analysis was to systematically collect and analyze all relevant RCTs to assess the effectiveness of VRBI in improving ankle functional status, postural balance, and muscle strength in patients with ankle instability. By updating the literature search, we identified 12 RCTs^63-74 involving patients with CAI or FAI who underwent VRBI compared to CT or usual care. Focusing on patients with ankle instability, this meta-analysis includes 8 RCTs more participants than the previous review, allowing for a more robust analysis of additional relevant variables such as ankle function status and muscle strength. Briefly, our meta-analysis revealed that VRBI holds promise as an effective intervention for enhancing ankle function status, postural balance, and muscle strength in plantar flexion and eversion movements in patients with ankle instability.

One of the primary outcomes assessed in this meta-analysis was functional ankle function status. Our findings demonstrate that VRBI can significantly reduce functional ankle functional status in these patients. This study highlights a novel and significant contribution to existing literature by revealing, for the first time in a meta-analysis, the efficacy of VRBI in specifically improving functional ankle instability. Unlike previous systematic reviews that only qualitatively suggested a possible improvement of functional instability in various orthopedic ankle injuries,⁴¹ our meta-analysis provides clinical evidence of the benefits of VRBI in this specific aspect. The ability of VRBI to create safe and personalized virtual environments allows patients to perform progressive and challenging functional exercises related to daily tasks, facilitating the transfer of acquired skills to activities of daily living.⁷⁸ VRBI exercises are dynamic and require postural responses to maintain ankle stability, thereby increasing proprioception.⁷⁰ This enhances the nervous system’s ability to perceive and anticipate destabilizing stimuli during walking, running, or functional activities, resulting in improved neuromuscular postural control of the ankle, due to a possible reduction in simple and choice reaction times as reported Mohammadi et al.⁷⁰ This can be translated into an enhanced ability to perform activities of daily living safely and efficiently by reducing the risk of ankle sprains. This finding highlights the potential of VRBI as an innovative and effective therapeutic tool for the rehabilitation of CAI.

Postural balance, and its different dimensions such as static or dynamic balance, is a parameter affected in patients with CAI.⁷⁹ Our meta-analysis revealed that VRBI is highly effective in improving both static and dynamic balance in these patients. Regarding static balance, our findings corroborate, complement, and strengthen the evidence reported by Elaraby et al.,⁴¹ who, based on three RCTs,^{68, 72, 74} demonstrated the efficacy of VRBI in improving static balance. However, our results cannot be directly compared from those Elaraby et al., due to they included one study that did not provide data from patients with CAI. However, considering it, our meta-analysis disagree with Elaraby, reporting the large effectiveness of VRBI in increasing dynamic balance in these patients. This discrepancy can be explained by the inclusion of 5 additional RCTs^{63-65, 69, 71} in our meta-analysis, excluding Vernadakis et al.,⁸⁰ as well as by the characteristics of the VR systems used in the included studies. The use of more dynamic VR systems, such as Nintendo Wii Fit, in contrast to others with a more pronounced static component, which could explain the greater improvement in dynamic balance observed in our meta-analysis. The results help to confirm our hypothesis that the incorporation of new RCTs can update and refine previous findings. In improving postural balance, numerous studies have demonstrated the efficacy of VRBI across different diseases.^{81, 82} The multisensory stimulation provided by VRBI, including visual, auditory, and proprioceptive feedback, challenges the nervous system, promoting neuroplasticity.⁸³ The repetitive practice of balance tasks in a gamified VR environments often performed in an upright position, providing multisensory stimulation including visual, auditory, and proprioceptive feedback, enhances postural input, potentially leading to the formation of new synaptic connections or the strengthening of existing ones, favors balance-related neuroplasticity.⁸⁴ Mao et al., reported that VRBI could active cerebral areas related to the integration and processing of balance inputs, improving spatial orientation and motor function in patients.⁸⁵ These neural adaptations could underlie the observed improvements in static and dynamic balance. While our review did not specifically examine the impact of VRBI on gait, the enhanced postural control resulting from VRBI suggests a potential positive effect on gait function, warranting further investigation.⁸⁶

In contrast to the findings related to ankle function static and postural balance, our meta-analysis did not provide clear evidence that VRBI can increase muscle strength. Our meta-analysis reported that VRBI can be effective in improving muscle strength of plantar flexion and eversion movements. However, VRBI did not demonstrate statistically significant effectiveness in improving overall lower limb muscular performance. The limited number of studies included in these meta-analyses could explain these results, and future studies using more active VR devices may yield different outcomes. Given these findings, it may be more appropriate to recommend VRBI for recovering ankle instability and postural balance, while relying more on CT for increasing muscle strength. A potential progression in the recovery of CAI could involve initiating rehabilitation with traditional CT to enhance lower limb muscle strength, with a particular focus on ankle muscles. Subsequently, a more active approach, incorporating functional exercises using VRBI, could be implemented to improve proprioception and reduce instability, leading to enhanced balance.

Finally, our meta-analysis encourages the clinicians in employing VRBI in their clinical practice of these patients. Firstly, beyond the numerous advantages of VRBI, such as its multisensory component, the literature indicates that this therapy is safe for the management of ankle instability, due to the absence of adverse events.^{63, 64, 66} This safety profile allows clinicians to confidently recommend its use in home-based rehabilitation programs. One strength of VRBI is that it can be used to increase the adherence of these patients to our therapies. Although active exercise is recommended or CAI recovery, in some musculoskeletal conditions between 65-90% of patients do not comply the therapy proposed.⁸⁷ Adherence to rehabilitation represents a challenge for clinicians, when the success is closely linked to factors like satisfaction and enjoyment, among others such as perceived barriers or own beliefs.⁸⁸ VRBI have emerged as a potential tool to enhance these factors due to its ludic character. Previous research, such as the 2008 study by Brumels et al., has demonstrated the potential of video games to increase patient motivation and adherence.⁸⁹ Several studies have reported high levels of satisfaction and enjoyment carrying out the ankle rehabilitation undergoing VR devices.^{63, 64} To confirm this benefit of VRBI, more studies can be performed. One of the key advantages of VRBI is its ability to create immersive, gamified environments that can make therapy more engaging and motivating. By transforming therapeutic exercises into interactive, challenging tasks, VRBI can improve patient adherence and outcomes. In summary, VRBI can be integrated into conventional clinical practice as an adjunctive therapeutic resource. Incorporating VRBI to perform functional activities in simulated clinical environments could be a valuable option to conclude the treatment session, following proprioceptive and strength training of the involved musculature.

Limitations of the study

While this meta-analysis provides valuable insights into clinical practice, several limitations must be considered. First, the relatively small number of studies included in certain meta-analyses and the limited sample sizes of individual studies can affect the precision and generalizability of the findings. A second limitation was the moderate methodological quality and the medium risk of bias in the RCTs included in this meta-analysis. In general, risk of bias can alter the quality of evidence and the pooled effect, potentially leading to under- or overestimation. Specifically, selection bias (related to inadequate randomization concealment in some RCTs) is highly related with the generalization of the findings in the population, and to a lesser extent, with an overestimation of the treatment effect.⁹⁰ However, performance bias (due to the inability to blind participants and therapists) is present in all RCTs and significantly influences the results. Detection bias, also prevalent (due to the inability to blind evaluators in over half of the RCTs), can be closely linked to the accuracy of VRBI effects. These two last biases may result in an overestimation of our findings.⁹¹ Third, the risk of publication bias, particularly reported in the functional ankle instability meta-analysis, may underestimate the effect of VRBI. While it is generally recommended to include at least 10 studies to confidently assess publication bias, the novelty of VRBI limited the availability of such many studies in any meta-analysis. Furthermore, the heterogeneity observed in the interventions and variables employed across studies limits direct comparison of results and may affect the validity and generalizability of conclusions. Specifically, variability in VRBI devices and protocols (NIVR and IVR) hinders comparison between studies. Nevertheless, some uniformity has been found in the frequency and duration of weekly sessions (three 30-minute sessions). Measurement of variables, except for muscle strength in foot movements’ assessment, is performed using a combination of measurement tools. To combine them, necessitates the use of the Standardized Mean Difference (SMD) as the effect measure, following recommendations from the Cochrane Collaboration. This practice is common in physiotherapy research, where the diversity of treatment parameters and outcomes makes comparison between studies and the performance of systematic reviews challenging. Finally, two limitations should be considered. First, a meta-analysis of other relevant variables, such as gait and related parameters, was not possible due to a lack of available data in the included studies. Second, the effectiveness of VRBI over time (follow-up) could not be assessed for the same reason. To address the limitations identified in this meta-analysis, future research should prioritize rigorous and standardized methodological designs. We strongly encourage authors to increase sample sizes, assess gait and satisfaction/adherence variables, and standardize VRBI (Virtual Reality-Based Intervention) protocols, particularly regarding the number of sessions, weeks of intervention, sessions per week, and duration of each session. This standardization would enhance study comparability and facilitate integration into future meta-analyses, leading to more generalizable findings and higher-quality evidence, ultimately supporting evidence-based clinical decision-making. Additionally, it is crucial to incorporate follow-up assessments in RCTs to determine the long-term maintenance of VRBI’s beneficial effects.

While this meta-analysis demonstrates the effectiveness of VRBI in treating ankle instability, further investigation is warranted to explore optimal treatment parameters, including gait. Specifically, future studies should examine the comparative effectiveness of different VR modalities, determine the most suitable treatment protocols, and evaluate the potential synergistic effects of combining VRBI with other interventions, such as strength and proprioceptive training of the involved musculature. The integration of muscle strengthening through resistance training with the multisensory stimulation provided by VRBI may be pivotal in optimizing ankle instability management. In particular, it would be highly valuable to investigate the combined effects of these treatments with and without footwear, and on both stable and unstable surfaces.

Conclusions

This systematic review with meta-analysis evaluates the efficacy of VRBI in the rehabilitation of ankle instability, including the major number of RCTs to date. Our meta-analysis demonstrates that VRBI is largely effective in improving functional ankle instability, static balance, and dynamic balance in these patients. While VRBI appears to be effective in increasing muscle strength for plantar flexion eversion movements, no statistically significant differences were found between in improving overall lower limb muscular performance. These findings strengthen the evidence base presented by previous reviews and provide clinical relevance for rehabilitation practitioners. To further enhance the strength and generalizability of these findings, future RCTs should involve larger sample sizes and strive for greater homogeneity in intervention protocols and assessment tools.

Supplementary Digital Material 1

Supplementary Table I

Literature search strategies in databases.

Supplementary Digital Material 2

Supplementary Table II

Characteristics of included studies.^63-74

Supplementary Digital Material 3

Supplementary Figure 1

Funnel plot for the effect of VRBI on functional ankle instability.

Supplementary Figure 2

Funnel plot for the effect of VRBI on lower limb muscular performance.