Inclusion criteria 1)

Randomized controlled trials (RCTs) comparing balance exercises to conventional exercises or no therapy and examining functional results in participants with knee OA were examined.

Exclusion criteria 1)

Conference abstracts, case reports, observational studies, and clinical commentaries were not included.

“ PubMed” (PubMed, RRID:SCR_004846), “ Scopus”, “ Web of Science”, “ Cochrane” (Cochrane Library, RRID:SCR_013000) and “ PEDro” were searched in January 2022 for relevant articles that established the efficacy of various methods of balance training for participants with knee OA. The studies featured were written in English. Two independent investigators conducted the search, which included a combination of two primary keywords: “Knee OA” (population) AND “Balance.” The Boolean operators “AND” and “OR” were used to combine the two terms. The search techniques were adjusted based on the databases. The publication dates were not confined, and the review included works published between January 2000 and December 2021.

Conference abstracts, case reports, observational studies, and clinical commentaries were not included. Systemic arthritic illnesses, tibial osteotomy, hip or knee joint replacement, and any other muscle or neurological ailment that may affect the lower extremity and impair balance were excluded. RCTs comparing balance exercises to conventional exercises or no therapy and examining functional results in participants with knee OA were examined. Participants in the chosen studies could be either male or female, of any age or severity level. Articles with a PEDro score ⁹ of five or above were considered. Two independent reviewers screened the articles to check if they met the inclusion criteria, conflicts between the reviewers was resolved by an intervening third reviewer.

The included articles were RCTs, therefore the risk of bias was done using the PEDro score. The PEDro scoring system consists of a checklist of 10 scored yes-or-no questions pertaining to the internal validity and the statistical information provided. PEDro scores of 0-3 were considered ‘poor’, 4-5 ‘fair’, 6-8 ‘good’, and 9-10 ‘excellent’.

Standard mean difference was used as an effect measure for comparison between balance treatment and routine rehabilitation.

The search was conducted by two independent reviewers (DT, SR) on various databases, following which all the identified studies were imported into online software Rayyan QCRI (Rayyan QCRI, RRID:SCR_017584). The titles and abstracts were also screened by two reviewers. For any ambiguities in the studies, consensus was obtained by discussing with the third reviewer (AP). The eligibility assessment under the inclusion-exclusion criteria was carried out by reviewing full-text articles.

The first reviewer obtained data from the included articles, which was then substantiated by a second reviewer and were entered into a standard form developed for the review. Information about the authors, journal, year of publication, characteristics of the subjects (age, inclusion criteria, gender, sample size), method (i.e., design, subjects, intervention, measures), outcome assessed, details of the interventions (parameters, frequency, intensity, type, time) and comparison groups were extracted from the included articles. Quantitative analysis, for the homogenous outcomes, i.e., VAS, TUG and WOMAC, was done. The pooled estimates of effect size were calculated using the effects model. These pooled estimations will subsequently be depicted in forest plots.

The first two reviewers independently completed a procedural quality assessment of the studies based on the PEDro scale, and papers with a PEDro score of less than five were eliminated. The fourth reviewer resolved any doubts about the study’s quality (SP). Each question on the PEDro scale evaluates the statistical significance and internal validity of the trials. Studies having a score of greater than five out of 10 ( Table 1) were determined to have high procedural quality, and the study included 22 articles.

Search results: We identified 22 studies for the systematic review and 17 studies for the meta-analysis out of 835 identified through the database screening. The results of the search and selection process are presented in the PRISMA flow diagram ( Figure 1). The included articles were published between January 2000 and December 2021.

Study designs: All the included articles were RCTs.

Participants: A total of 1456 people were involved in the included studies. The participants ranged in age from 30 to 65 years. The study encompassed all Kellgren Lawrence severity levels of knee OA. ¹⁰

Intervention: The intervention group in eight of the 22 trials received isolated balance training programs with no other form of exercise. Whereas 14 out of the 22 studies provided balance training along with some other form of exercises. Two trials provided agility training to the experimental group, ^{11 , 12} whereas the experimental group in the other two studies received Wu Qin xi Qigong and Wuqinxi exercise programs. ^{13 , 14} Two trials used virtual reality and virtual feedback to give balance training, ^{15 – 17} while the third employed Kinesio tapping to improve stability in subjects with knee OA. ¹⁸ In one study, body weight (BW) training was paired with needle knife therapy as a treatment modality. ¹⁹

Control intervention: The participants in the control groups received routine rehabilitation and strengthening exercises.

Outcomes: The included studies analyzed the following outcomes: VAS, ^{11 , 15 , 17 , 20 – 24} WOMAC, TUG,  self-reported knee joint instability, Berg Balance Scale (BBS), Lequesne index, knee joint proprioception, Tinetti’s Performance Oriented Mobility Assessment (POMA) scale and static postural stability. Eight of the included studies analyzed pain using the VAS, ^{11 , 15 , 17 , 20 – 24} and 12 studies analyzed the function using WOMAC score. ^{11 – 13 , 15 , 17 , 19 , 20 , 25 – 30} TUG score was analyzed by six studies. ^{11 , 14 , 17 , 20 , 31 – 33}

Study characteristics: The size of the samples ranged from 30 to 190 subjects. All the subjects included in the trial were analyzed based on the American College of Rheumatology criteria. ³⁴ All the studies used the Kellgren Lawrence score for radiological grading.

The effects of interventions are explained in Table 2.

Effect of intervention for multiple groups: Rogers et al., ²⁹ conducted a study to compare the efficacy of home-based kinesthesia, balance and agility exercises to resistance exercise, a combination of kinesthesia and resistance exercises, and no exercise. WOMAC pain and physical function significantly improved for all four groups after eight weeks of intervention, with no group differences, although the exercise groups improved more between the midpoint and the eight weeks follow-up, the control group did not. A study conducted by Lin et al., ¹⁶ conducted a study with 89 individuals in which they examined proprioceptive function between computerized proprioception facilitation exercise (CPFE) and closed kinetic chain exercise (CKCE) in subjects with knee OA. After an eight-week exercise intervention, both the CPFE and CKCE groups demonstrated a reduction in knee position error and WOMAC functional scores, as well as increased walking speeds (p = 0.016) on four distinct terrains.

Rahlf et al., ³⁰ studied the effects of Kinesio taping (KT) on pain and function in patients with knee OA. The WOMAC subscales of pain showed significant differences. However, the effects of KT were more noticeable. Post hoc analysis revealed significant WOMAC differences between the tape and control groups (pain: p = 0.047, function: p = 0.004), the tape and sham groups (pain: p = 0.05, function: p = 0.03), and no difference between the sham and control groups. Trans et al., ²⁶ studied the effects of whole-body vibration training on proprioception in patients with knee OA. The balance board with built-in vibration (VibF) group improved significantly in the threshold for detection of passive movement (TDPM) when compared to the control group, with a mean difference of -0.59 seconds (95% CI, 1.13 to 0.05; p = 0.0326). When compared to the control group, the balance board with stable vibration platform (VibM) group improved by 0.52 seconds (95% CI, 1.04 to 0.00; p = 0.0511). There was no discernible difference between the VibM and VibF groups in the above study.

Another study, conducted by Braghin et al., ²⁵ analyzed the effect of exercise on balance and function in people with knee OA. Two of the three groups did physical exercises for 50-60 minutes every day for around eight weeks. The intragroup analysis of the WOMAC questionnaire revealed no differences between the asymptomatic and control groups, whereas the symptomatic group had significantly lower pain and functional outcome scores.

Jan et al., ³³ conducted a study to assess the effects of weight-bearing (WB) and non-weight bearing (NWB) exercise on walking speed, position sensing, and function in people with knee OA. The WOMAC function and walking times on four different terrains improved significantly in the WB group (p = 0.08), but not in the control group. When compared to the control group, both intervention groups improved in WOMAC function (p = 0.08). However, there was no difference between the NWB and WB groups. When compared to the NWB and control groups, the WB group improved more in walking speed on the spongy surface, figure of eight, and positioning error (p = 0.08). The subjects in the NWB exercise group improved their walking speed up and down the stairs. There was no difference in walking speed on level ground between groups, nor was there a difference in walking speed on a figure of eight or a spongy surface, nor was there a difference in reposition error between NWB and the control group.

Meta-analysis was considered for 15 of the included studies, due to heterogeneity in the outcomes. The meta-analysis was done for the common outcomes of VAS, TUG and WOMAC.

Meta-analysis was conducted for 15 of the 22 included studies. The common outcomes analyzed were VAS scores, WOMAC score and the TUG score.

For pain, ( Figure 2) eight studies were analyzed for the VAS scores, pre-and post-intervention. Heterogeneity [ I ²] was 92% (p < 0.05). The mean difference was -0.79 with (95% CI, -1.59 to 0.01) for the intervention versus the control group ( Table 3).

For functional outcome, seven studies were analyzed using the WOMAC ( Figure 3) scores the heterogeneity [ I ²] was 81% (p < 0.0001). The mean difference was -0.02 (95% CI, -0.44 to 0.40) for the intervention against the control group ( Table 4).

Another functional parameter was analyzed using the TUG ( Figure 4) score. Five studies were analyzed and the heterogeneity [ I ²] was 95% (p < 0.0001). The mean difference was -1.71 (95% CI, -3.09 to -0.33) for the intervention against the control group ( Table 5).

According to the findings of the included studies and their meta-analysis, balance-based exercises help to reduce pain and improve functional outcomes in people with balance alteration following knee OA. The above values are displayed in a small confidence interval range, indicating the analyses’ validity and sensitivity, as well as the significant influence. Furthermore, the random-effects model used provided accurate results by using sample size and standard error. The meta-analysis likewise comes up with a positive conclusion balance-based exercise. Even though the key outcome measures of VAS, WOMAC and TUG scores were homogeneous, the analyzed studies differed in the mode and duration of intervention and hence a meta-analysis was conducted using the random-effects model.

The studies in included our review were of moderate to good methodological quality, however, risk of bias was noted in terms, of blinding of subjects, therapist and assessors. A detailed description of the same is given in Table 1.

Certainty of evidence is low to moderate due to the quality of the included studies and higher heterogeneity.

As the current systematic review includes data from studies with varied sample sizes, the treatment impact could be overestimated. Additionally, comparing outcome measures was difficult due to the diversity of the assessment procedures utilized between studies. A thorough literature search conducted to expand empirical understanding yielded various areas of recommendation. To begin, a larger, suitably powered RCT should be done to offer critical data on the type and dosage of exercises needed. Second, we advocate conducting another systematic review to look into other factors that influence balance, such as OA symptoms (functional limitation, pain, stiffness), quality-of-life domains, and other functional measures.

All data underlying the results are available as part of the article and no additional source data are required.

Open Science Framework: Effectiveness of balance training on pain and functional outcomes in knee osteoarthritis - a systematic review and meta-analysis. https://doi.org/10.17605/OSF.IO/4H7TQ. ⁴³

This project contains the following extended data: -

DATA EX TABLE.docx (Effects of intervention of the included studies)

Data are available under the terms of the Creative Commons Zero “No rights reserved” data waiver (CC0 1.0 Public domain dedication).

Open Science Framework: PRISMA checklist for ‘Effectiveness of balance training on pain and functional outcomes in knee osteoarthritis: A systematic review and meta-analysis’. https://doi.org/10.17605/OSF.IO/4H7TQ. ⁴³

Data are available under the terms of the Creative Commons Zero “No rights reserved” data waiver (CC0 1.0 Public domain dedication).