Effects of exercise regimens on balance ability in older patients with osteoporosis: a systematic review and Bayesian network meta-analysis of randomized controlled trials

Abstract

To determine the optimal exercise regimen for improving balance and preventing severe falls in older patients with osteoporosis (OP). Four databases were searched until September 1, 2025. The risk of bias was assessed using Cochrane ROB2. Outcomes included Berg Balance Scale (BBS, primary), timed up and go test (TUG), bone mineral density (BMD), one-leg stand test (OLS), and number of falls. A Bayesian network meta-analysis in R4.4.1/GeMTC synthesized effects, presented as MD (95% CrI) and ranked by SUCRA. Analysis of 22 RCTs (n=1538) versus usual care showed virtual reality (VR) most effective for BBS (MD 9.2, 95% CrI 7.2, 11; SUCRA 99.66%) and TUG (MD -4.6, 95% CrI -5.8, -3.3; SUCRA 98.51%); balance training+resistance training+aerobics (BT + RT + aerobics) best improved BMD (MD 0.016, 95% CrI 0.012, 0.020; SUCRA 72.38%); trampoline best improved OLS (MD 8.8, 95% CrI 1.7, 5.5; SUCRA 72.38%); RT most significantly reduced falls (MD 0.29, 95% CrI 0.100, 0.68; SUCRA 84.86%). For older OP patients, VR is most effective in improving overall balance and mobility, with reliable evidence. Combination training, trampoline, and RT exhibit potential benefits for BMD, OLS, and fall prevention, respectively. Due to the limited amount of evidence and network strength, however, these interventions cannot yet be considered definitive clinical recommendations. More high-quality direct comparisons are required for further validation in the future.

1 Introduction

Osteoporosis (OP) is a systemic metabolic disease featured with reduced bone mass and bone microarchitecture abnormalities, and it is prevalent in the older group (World congress on osteoporosis, osteoarthritis and musculoskeletal diseases (WCO-IOF-ESCEO 2025), 2025). As global aging accelerates, the prevalence of OP keeps rising, making it a significant public health issue (Wang and Hu, 2023). OP not only increases bone fragility but is also frequently accompanied by muscle atrophy, diminished neuromuscular control, and reduced proprioception, impairing patients’ balance ability (Paramento et al., 2025). Balance impairment is a primary cause of falls in older adults (Gaspar and Lapão, 2021), and the resulting fractures lead to high disability and mortality, and substantial socioeconomic burdens in older OP patients (Chen et al., 2025; The global, regional, and national burden attributab le to low bone mineral density, 1990-2020: an analysis of a modifiable risk factor from the Global Burden of Disease Study 2021, 2025). Research suggests that interventions in balance ability can improve balance ability and reduce falls in older OP patients (Paramento et al., 2025). Therefore, it is of clinical significance to identify effective strategies for enhancing balance ability and preventing falls in this group.

Exercise therapy (Sahni and Nieves, 2019), neuromuscular taping techniques (Paramento et al., 2025), and medication (Prasad et al., 2025) are available for enhancing balance ability in older OP patients nowadays. Research data on the long-term efficacy of neuromuscular taping techniques remains lacking, with insufficient strength of evidence (Paramento et al., 2025). Medication can directly modulate bone metabolism and synergize with other approaches to reduce the incidence of falls and fractures (Brooke-Wavell et al., 2022). However, long-term medication may carry adverse effects such as hepatic and renal burden, and increase economic burden. Moreover, drugs can enhance bone mineral density (BMD), but they are less effective in improving balance ability, muscle strength, and function, and thus cannot directly reduce falls (Giangregorio and Ponzano, 2022). Exercise therapy is considered a core non-pharmacological approach due to its multi-targeted effects, high safety profile, and good cost-effectiveness (Brooke-Wavell et al., 2022; Zhang et al., 2022; Gao et al., 2023; Tao et al., 2024). Evidence suggests that exercise training, such as robot-assisted or virtual reality (VR) balance training (BT), can effectively increase the Berg Balance Scale (BBS) score (Krohn et al., 2024; Hao et al., 2025) and shorten the timed up and go test (TUG) time by enhancing lower limb muscle strength and coordination (Gil et al., 2021). Furthermore, exercise positively modulates bone metabolism and improves BMD by mechanical loading and secretion of myogenic factors (e.g., irisin), and also directly enhances neuromuscular control and static one-leg stand ability (Wu et al., 2023; Hu et al., 2024; Xiong et al., 2025). With these changes in the skeletal, muscular, and nervous systems, older OP patients ultimately experience enhanced overall postural stability and balance, effectively reducing falls (Vonstad et al., 2022).

Despite substantial evidence, available studies have key limitations: The relative efficacy in ameliorating the aforementioned outcome metrics remains unclear across exercise regimens. It is difficult for traditional reviews and meta-analyses to quantitatively rank and optimize exercise regimens in the absence of head-to-head comparisons. To address these issues, we conducted this systematic review and Bayesian network meta-analysis (NMA). It directly compared the relative effects of different exercise regimens on balance ability (e.g., BBS, TUG), BMD, and falls in older OP patients, and ranked the effects of interventions by synthesizing and quantifying available evidence of randomized controlled trials (RCTs). The findings are expected to provide a high-level, evidence-based basis for developing the optimal individualized exercise regimens in clinical practice.

2 Materials and methods

An NMA was conducted following the PRISMA guidelines. It was registered with PROSPERO (https://www.crd.york.ac.uk/prospero/) (registration No. CRD420251158469).

2.1 Search strategy

We searched Embase, Cochrane, PubMed, and Web of Science from inception until September 1, 2025, with primary search terms “Osteoporosis”, “Exercise”, “Balance”, and “Older adults” (Appendix 1). No language restrictions were applied, but the English abstract should be obtained for non-English-language articles.

2.2 Study screening

After duplicate publications were removed using EndNote 21, two reviewers (Liu XY and Chang MJ) independently examined the title and abstract to identify relevant studies. Then the full text was downloaded and screened independently by the two reviewers, and the results were summarized by a third reviewer. The included studies comprised older OP patients, and interventions included trampoline, pulsed electromagnetic fields (PEMF), resistance training (RT), BT, aerobics, aquatic exercise (AE), VR, whole body vibration exercise (WBV), BT + RT + aerobics, BT + RT, BT + cognitive tasks (CT), BT + WBV, BT + aerobics, and usual care (Control). The outcomes were the BBS score, TUG, BMD, OLS, and number of falls. RCTs that at least met the criteria for patient characteristics and interventions were included.

2.3 Outcome metrics

The BBS is a core tool for clinically assessing comprehensive balance function, widely used for fall risk assessment in older stroke and Parkinson’s disease patients undergoing neurological rehabilitation. It comprises 14 items, each scored 0–4 points, with a total score of 56 points (Downs et al., 2013). Higher scores correspond to better balance function (Zhang et al., 2024).

TUG fully assesses mobility and fall risk by measuring the time (seconds) required for subjects to rise from a chair, walk 3 meters, turn around, walk back, and sit down. Longer time suggests a higher likelihood of functional decline or disability (Gatenio-Hefling et al., 2025). TUG is simple to complete and time-efficient, making it a commonly used tool for evaluating functional mobility in older adults and patients with neurological disorders.

BMD is a crucial indicator for skeletal health, which quantifies bone strength by measuring bone mineral content per unit area or volume. Reduced BMD is a significant predictor of fracture risks (The global, regional, and national burden attributab le to low bone mineral density, 1990-2020: an analysis of a modifiable risk factor from the Global Burden of Disease Study 2021, 2025). Therefore, changes in BMD are often tracked to objectively assess the progression of skeletal diseases such as OP and the effect of medication (Banefelt et al., 2022).

OLS assesses the subject’s static balance by recording the duration of successful posture maintenance or whether falls occur (Araujo et al., 2022). It can not only rapidly evaluate balance function but also reveal asymmetry in lower limb strength and coordination in a simple way. Therefore, it is frequently used for preliminary fall risk assessment in community-dwelling older adults (Bendrik et al., 2024).

The number of falls refers to the number of individuals experiencing falls within a defined time period, commonly used for health monitoring in older groups. Tracking changes in the number of falls allows for quantitative assessment of fall risks, guidance of the development and implementation of preventive interventions, and evaluation of the effectiveness of related regimens (Chen et al., 2025).

2.4 Eligibility criteria

Inclusion criteria: (1) Population: older patients (≥60 years) diagnosed with OP; (2) Intervention: exercise regimens; (3) Outcomes: metrics for balance ability, such as BBS and TUG; (4) Study design: RCTs.

Exclusion criteria: (1) Non-adult or animal studies; (2) unclear or non-target populations; (3) unclear interventions or drug interventions; (4) unavailable full text or no relevant data; (5) non-RCTs.

2.5 Data extraction

Two reviewers (Liu XY and Chang MG) reviewed the included studies and extracted data: first author, publication year, country, sample size, age, sex, BMI, interventions, controls, number of subjects in the test and control groups, study type, outcome metrics (BBS, TUG, BMD, OLS, and number of falls), and follow-up period. The mean and standard deviation (SD) pre- and post-intervention were extracted in the test and control groups. Change-from-baseline values with SD were prioritized, but if they were not reported, MD was calculated by pre- and post-intervention measurements. When SD was unavailable, we calculated standard error, 95% confidence interval (CI), range, and interquartile range (IQR). When IQR was reported, it was used as the mean and IQR/1.135 as the SD. If the Min-Max median was reported, it was not statistically analyzed. Given different scoring scales across outcome metrics, the scoring data of outcome metrics were pooled using MD. Falls, a dichotomous variable, were analyzed using the number of falls in the test and control groups post-intervention. Notably, due to variations in follow-up durations across studies and the potential impact of baseline levels of some outcome metrics (e.g., BBS, TUG) on effect sizes, a random-effects model was incorporated in the meta-analysis to partially account for heterogeneity. However, the potential impact of different follow-up durations and baseline levels on the comparability of MD still needs to be considered during result interpretation.

2.6 Risk of bias assessment

The included studies were assessed for the RoB using the Cochrane ROB2 tool from the randomization process, deviations from intended interventions, missing outcome data, measurement of outcome, and selection of reported results. Each domain was rated as low risk, some concern, or high risk. Two reviewers (Liu XY and Chang MG) independently assessed study quality. Any discrepancy was settled by consultation with a third reviewer (Zheng XM).

2.7 Data synthesis and analysis

R4.4.1 and GeMTC were utilized for NMA modeling. All data analyses were completed based on a Bayesian random-effects model supported by the R “GeMTC” and “rjags” packages. The posterior distribution of parameters was simulated via MCMC, and the Bayesian approach was used to provide probability distributions for effect size estimates. A Bayesian NMA model was ultimately created with 25,000 iterations, 250,000 simulations, and 10-fold sampling. Consistency and inconsistency tests were carried out. When closed loops were formed in the network diagram, local inconsistencies between direct and indirect evidence were assessed by the node-splitting method. Model convergence was evaluated using Brooks-Gelman-Rubin statistics combined with trace plots and density plots. Effect sizes were calculated for MD with 95% CrI by a Bayesian random-effects model. Given that falls constitute a dichotomous variable (presence or absence), a Bayesian framework-based binomial likelihood model was utilized, and the log link function was used to create a fixed-effects NMA model. Results were presented as RR with 95% CI. Comparison results were visualized by forest plots and summary tables. An effect was considered significant when its 95% CrI did not include zero. The efficacy of interventions was ranked by the surface under the cumulative ranking curve (SUCRA) value (0-100%), with higher values indicating better efficacy. Notably, SUCRA rankings reveal the probability of each intervention being the optimal intervention rather than the absolute value of effect size, and it cannot be simply equated with clinical superiority. When CIs overlap among multiple interventions, minor differences in SUCRA may reflect sampling error rather than the superiority or inferiority of interventions. Therefore, this study prioritized effect sizes with CIs, and did not rank the interventions with overlapping CIs. The relative relationships among interventions were visualized by a network diagram, where nodes represented interventions and edges represented direct comparisons.

3 Results

3.1 Search results

We initially retrieved 3798 reports, of which 1269 duplicate publications were excluded. After title and abstract review, the full text of 230 eligible studies was examined, and the ineligible studies were removed due to unavailable full text (n=54), ineligible study types (n=9), ineligible interventions (n=8), non-older OP patients (n=62), no outcome metrics for balance (n=32), no outcome data (n=25), duplicate data (n=11), non-applicable data (n=6), and withdrawn article (n=1). Finally, 22 studies were included following the PRISMA flowchart (Figure 1).

Figure 1

3.2 Study characteristics

Twenty-two studies (Vaillant et al., 2006; Madureira et al., 2007; Swanenburg et al., 2007; Smulders et al., 2010; Teixeira et al., 2010; Murtezani et al., 2014; Olsen and Bergland, 2014; Hakestad et al., 2015; Konak et al., 2016; Ruzene et al., 2016; Oksuz and Unal, 2017; Dizdar et al., 2018; Miko et al., 2018; Posch et al., 2019; Ramos et al., 2019; Stanghelle et al., 2020; FilipoviĆ et al., 2021; Zhang et al., 2022; Zhao et al., 2023; Mewara and Chhajed, 2024; Sangtarash et al., 2024; Yilmaz and Kösehasanoğulları, 2024) were included, involving 1538 older OP patients, and 13 interventions (trampoline, PEMF, BT, RT, AE, WBV, VR, aerobics, BT + RT, BT + aerobics, BT + RT + aerobics, BT + WBV, BT + CT) and usual care (Control). All RCTs were from Europe, South America, and Asia, and the subjects were aged ≥60 years, mostly with an excessive BMI. BMD was reported primarily in the lumbar spine, as well as the hip and femur. The follow-up duration varied greatly, ranging from immediate assessments to long-term follow-ups, with 3–6 month short- to medium-term follow-ups being most common (Table 1). Variations were present in training dosage across interventions. For example, BT + RT + aerobics lasted for 5.5–40 weeks, 2–3 sessions per week and 35–90 min per session, with full supervision supplemented by partial supervision. VR training lasted for 12–52 weeks, 3 sessions per week and 45–50 minutes per session, with full supervision. The specific training dosage is summarized in Table 2.

3.3 RoB

Excel was used for data processing and chart generation for all studies (Vaillant et al., 2006; Madureira et al., 2007; Swanenburg et al., 2007; Smulders et al., 2010; Teixeira et al., 2010; Murtezani et al., 2014; Olsen and Bergland, 2014; Hakestad et al., 2015; Konak et al., 2016; Ruzene et al., 2016; Oksuz and Unal, 2017; Dizdar et al., 2018; Miko et al., 2018; Posch et al., 2019; Ramos et al., 2019; Stanghelle et al., 2020; FilipoviĆ et al., 2021; Zhang et al., 2022; Zhao et al., 2023; Mewara and Chhajed, 2024; Sangtarash et al., 2024; Yilmaz and Kösehasanoğulları, 2024). For the randomization process, one study was assessed as some concern due to no reporting of the implementation method. For the deviations from intended interventions, some studies lacked descriptions of blinding or failed to meet requirements for blinding, but they were assessed as low risk because most deviations were attributed to the study environment and did not affect outcomes or interventions. Two studies were assessed as some concern because the deviation affected outcome measurement. One study was assessed as high risk because the deviation affected outcome measurement and whether it was homogeneous between groups remained unclear. For the missing outcome data, all or nearly all outcome data were available for most studies; one study had approximately 17% follow-up attrition but performed imputation, thus rated as some concern. Another study had nearly 20% follow-up attrition, also rated as some concern. For measurement of outcome, one study was rated as some concern because the assessors might have been aware of the intervention. For the selection of reported results, all studies described the pre-specified analysis plans and were rated as low risk (Figure 2 and 3).

Figure 2

Figure 3

3.4 Assessment results of transitivity assumption

The validity of NMA relies on transitivity assumptions, i.e., clinical and methodological characteristics of studies should be comparable when conducting indirect comparisons between different interventions. To validate this assumption, the distribution of key effect modifiers across different intervention nodes was compared. Baseline information on age, BBB scores, and body mass index in intervention groups is shown in Table 1. The results revealed that the mean values of these key indicators were similar across intervention groups, and the ranges of individual variation reflected by SD were highly overlapped. This suggests that the patient populations in the included studies shared similar clinical characteristics, satisfying the prerequisite for transitivity. To further assess comparability of training intensity, the dosage parameters (frequency, duration, and supervision mode) of each intervention were standardized and compared (Table 2). The results revealed that the training intensity exhibited certain heterogeneity across nodes, but its distribution was generally balanced with no systematic bias. We considered that the integrated effect of multi-component interventions is not a simple summation of single-component interventions and should not be separated for analysis. Moreover, treating multi-component interventions as independent nodes better aligns with the pattern of intervention protocols in clinical practice. Therefore, this study incorporated both single- and multi-component interventions as independent nodes into the network. However, training intensity, specific implementation settings, supervision formats, and compliance varied across interventions, which might still influence the transitivity assumption. Besides, the node-splitting analysis revealed no statistically significant inconsistency between direct and indirect comparisons of the outcome metrics with closed loops formed (BBS and TUG) (P>0.05; Supplementary Tables 1 and 2). This further supported the transitivity assumption at the statistical level. Due to the extremely limited number of studies and sparse network structure for the three outcomes (BMD, OLS, and number of falls), effective subgroup analyses or meta-regression could be conducted for detailed variables such as training intensity, specific movement types, and training dosage. Consequently, the ranking results for these three outcomes should be regarded as exploratory findings, necessitating further validation by high-quality, large-sample RCTs.

3.5 Statistical analysis results

3.5.1 Effects of different interventions on BBS scores in older OP patients

The BBS score was reported in 11 studies involving 574 patients. The interventions comprised BT + RT + aerobics, RT, BT, BT + CT, aerobics, AE, VR, WBV, and BT + WBV. Compared with usual care, VR was most effective in increasing the BBS score (MD 9.2, 95% CrI 7.2, 11) (SUCRA 99.66%). Additionally, BT + CT (MD 5.2, 95% CrI 3.7, 6.8) (SUCRA 80.00%). Although VR had the highest rank in terms of the SUCRA value, overlapping CIs in some comparisons suggested that the true differences in interventions might be less significant. Evidence revealed a minimal clinically important difference (MCID) of 7-11.5 points for BBS in older OP patients (Tamura et al., 2022; Kobayashi et al., 2024), and the effect size of VR fell within this range. BT + RT + aerobics (MD 5.2, 95% CrI 2.2, 8.2) (SUCRA 74.74%), BT (MD 4.1, 95% CrI 2.7, 5.6) (SUCRA 60.60%), BT + WBV (MD 3.9, 95% CrI 1.9, 5.9) (SUCRA 55.72%), RT (MD 2.4, 95% CrI 1.7, 3.2) (SUCRA 35.85%), aerobics (MD 1.8, 95% CrI 1.3, 2.3) (SUCRA 20.85%), and WBV (MD 1.6, 95% CrI 0.28, 2.9) (SUCRA 19.22%) all significantly increased the BBS score. However, their ranking differences should be interpreted carefully as the CIs of these interventions partially overlapped with VR. AE (MD 3.8, 95% CrI -0.64, 8.3) (SUCRA 52.70%) was less effective. Four closed loops were formed in the network: BT - RT - aerobics; BT - RT - usual care; BT - aerobics - usual care; RT - aerobics - usual care. The inconsistency test revealed P-values >0.05 for all comparisons of interventions, suggesting no inconsistency. The results were visualized by the network diagram (Figure 4), forest plot (Figure 5), league table (Supplementary Figure 1), and inconsistency test diagram (Supplementary Table 1; Supplementary Figure 2).

Figure 4

Figure 5

3.5.2 Effects of different interventions on TUG time in older OP patients

The TUG was reported in 14 studies involving 1030 patients. The interventions comprised trampoline, PEMF, BT + RT + aerobics, RT, BT, BT + RT, BT + CT, aerobics, VR, WBV, and BT + aerobics. Compared with usual care, VR was most effective in shortening TUG time (MD -4.6, 95% CrI -5.8, -3.3) (SUCRA 98.51%). Additionally, BT (MD -4.0, 95% CrI -5.2, -2.9) (SUCRA 84.18%), RT (MD -3.9, 95% CrI -4.8, -2.9) (SUCRA 78.69%), BT + CT (MD -3.9, 95% CrI -5.2, -2.7) (SUCRA 78.61%), aerobics (MD -3.3, 95% CrI -4.5, -2.1) (SUCRA 61.66%), BT + aerobics (MD -2.8, 95% CrI -3.8, -1.9) (SUCRA 51.91%), WBV (MD -3.0, 95% CrI -4.3, -1.7) (SUCRA 51.27%), PEMF (MD -2.0, 95% CrI -3.7, -0.34) (SUCRA 33.67%), trampoline (MD -1.5, 95% CrI -2.1, -0.96) (SUCRA 25.86%), BT + RT + aerobics (MD -1.5, 95% CrI -1.8, -1.2) (SUCRA 23.90%), and BT + RT (MD -1.0, 95% CrI -1.5, -0.55) (SUCRA 11.66%) all significantly shortened TUG time. However, the CIs of interventions overlapped, so the subtle differences in SUCRA rankings lacked statistical significance. Two closed loops were formed in the network: BT - RT - aerobics; BT - RT - usual care. The BT - RT - aerobics closed loop originated from a three-arm RCT, which did not establish independent direct and indirect evidence paths. Thus, the rigorous inconsistency test was precluded. The inconsistency test revealed P-values >0.05 for all comparisons of interventions, suggesting no inconsistency. The results were visualized by the network diagram (Figure 6), forest plot (Figure 7), league table (Supplementary Figure 3), and inconsistency test diagram (Supplementary Table 2; Supplementary Figure 4).

Figure 6

Figure 7

3.5.3 Effects of different interventions on BMD in older OP patients

The BMD was reported in four studies involving 319 patients. The interventions included trampoline, BT + RT + aerobics, BT + RT, and AE. Compared with usual care, BT + RT + aerobics performed best in improving BMD (MD 0.016, 95% CrI 0.012, 0.020) (SUCRA 72.38%). Additionally, trampoline (MD 0.026, 95% CrI -0.012, 0.064) (SUCRA 81.79%), BT + RT (MD 0.0078, 95% CrI -0.041, 0.056) (SUCRA 53.31%), and AE (MD -0.027, 95% CrI -0.074, 0.020) (SUCRA 9.16%) did not improve BMD. Although the pooled effect size of trampoline was larger (MD 0.026), its 95% CI included invalid values (-0.012, 0.064), indicating a non-robust finding that trampoline should not be regarded as superior to BT + RT + aerobics. Only four studies were included for BMD, resulting in a sparse network structure with no closed loop formed. Consequently, the consistency test was not conducted, and the SUCRA ranking was significantly affected by sampling error. Given the limited strength of evidence, these interventions cannot yet be considered a definitive clinical recommendation. The results were visualized by the network diagram (Figure 8), forest plot (Figure 9), and league table (Supplementary Figure 5).

Figure 8

Figure 9

3.5.4 Effects of different interventions on OLS in older OP patients

Two studies reported OLS, involving 136 patients. The interventions included trampoline and BT + RT + aerobics. Compared with usual care, trampoline performed best in improving OLS (MD 8.8, 95% CrI 3.9, 14.0) (SUCRA 98.60%). Additionally, BT + RT + aerobics also significantly regulated OLS (MD 3.6, 95% CrI 1.7, 5.5) (SUCRA 51.39%). Only two studies were included for OLS, resulting in an extremely limited network structure. Direct comparisons could only be made with usual care, precluding complex indirect comparisons. The SUCRA ranking was highly susceptible to incidental effects and should not serve as a basis for judging the superiority or inferiority of interventions. This finding is merely suggestive and far from the strength required for clinical recommendations. More RCTs are required to validate this finding. The results were visualized by the network diagram (Figure 10), forest plot (Figure 11), and league table (Supplementary Figure 6).

Figure 10

Figure 11

3.5.5 Effects of different interventions on falls in older OP patients

Seven studies reported the incidence of falls, involving 586 patients. The interventions included BT + RT + aerobics, RT, VR, and BT + aerobics. Compared with usual care, RT was most effective in reducing falls (RR 0.29, 95% CrI 0.100, 0.68) (SUCRA 84.86%). Additionally, BT + RT + aerobics also reduced falls (RR 0.73, 95% CrI 0.54, 0.99) (SUCRA 39.78%), but VR (RR 0.44, 95% CrI 0.014, 5.3) (SUCRA 58.45%) and BT + aerobics (RR 0.52, 95% CrI 0.19, 1.3) (SUCRA 57.99%) were less effective. Although seven studies reported falls, direct comparisons of different interventions remained scarce. The network was weakly connected, sufficient indirect comparison was lacking, and no closed loop was formed, thus precluding the inconsistency test. In this case, the SUCRA ranking only reflected the model’s estimated probability rather than directly representing the definite clinical superiority of interventions. Studies exhibited heterogeneity in the definition of falls and follow-up durations. Therefore, the preventive effect of RT on falls should be regarded as preliminary evidence, requiring more high-quality studies with standardized outcome reporting. The results were visualized by the network diagram (Figure 12), forest plot (Figure 13), and league table (Supplementary Figure 7).

Figure 12

Figure 13

4 Discussion

This Bayesian NMA fully compared the effects of 13 exercise regimens on balance ability and related outcomes in older OP patients. Different exercise regimens demonstrated varying effects on outcome metrics. Specifically, VR ranked higher in improving overall balance function (BBS) and functional mobility (TUG), with effect sizes reaching the lower limit of MCID. Trampoline exhibited potential for enhancing static balance (OLS), but the strength of evidence was extremely low. BT + RT + aerobics produced a significant effect on increasing BMD. Furthermore, RT alone showed a statistically significant association with reduced fall risk. These findings offered guidance for the direction of clinical exercise prescription development. However, the strength of evidence considerably varied across outcome metrics and warrants stratified interpretation in subsequent discussions.

The BBS score was reported in 11 studies involving 574 patients, and the TUG was reported in 14 studies involving 1030 patients. The results showed that VR outperformed usual care in improving both BBS and TUG. The possible reason is that in an immersive, task-oriented environment, VR can stimulate cognitive engagement in physical activity, promote neuroplasticity, and optimize multisensory integration and neuromuscular coordination. In this way, efficient training of dynamic balance and executive function can be achieved (Barbanchon et al., 2024; Intziegianni et al., 2024). In clinical care, nurses or rehabilitation therapists can utilize VR devices to design training tasks simulating daily scenarios such as stepping over barriers or bending to pick up objects. In this way, neuromuscular control and cognitive-motor integration can be directly trained within a safe, controlled environment (Rezaei et al., 2023). Moreover, patients can embrace higher practical ability and confidence to handle complex environments by enhancing training motivation and relieving fear (Cortés-Pérez et al., 2025). Ultimately, VR improves overall function better than traditional methods (Zhao et al., 2023). In contrast, AE was less effective in improving BBS. The possible reason is that weight-bearing and balance challenges are reduced due to its buoyancy, and neuromuscular control required for walking fails to be sufficiently stimulated (Deng et al., 2024). Similarly, PEMF and some single training ameliorated TUG, but they may not be optimal choices for complex, coordinated tasks of functional mobility. From a perspective of clinical practicality, VR devices are expensive and have high requirements for service environment and technical support, thus restricting their accessibility in resource-constrained medical settings or families. Therefore, VR remains the preferred choice for balance improvement in well-equipped hospitals or rehabilitation centers. In community or primary care settings, physical functional training simulating VR principles can be implemented under the guidance of healthcare staff as a cost-effective alternative. As proposed by Lars et al., VR can be utilized for precise assessment and complex scenario training in specialized institutions, while simplified training protocol can be used in primary care settings (Donath et al., 2016). This low-cost, highly accessible human-guided training mode can be compared with VR for efficacy in future studies, and cost-effectiveness analysis can be performed. The findings are expected to providing more universally applicable protocols in clinical practice across resource contexts.

The effects of exercise therapy on bone structure (BMD) and static balance (OLS) were also specific to regimens. Four studies reported BMD, involving 319 patients. They identified BT + RT + aerobics as the optimal intervention, which worked by fully stimulating osteoblast activity under diverse, integrated mechanical loads (Hejazi et al., 2022; Altai et al., 2024). This finding aligned with a study by Borja Sañudo et al. on the mechanism of exercise intervention in OP (Sañudo et al., 2025). However, the robustness of these conclusions was restricted by the following factors: Only four studies were included, resulting in a sparse network structure, and training dosage varied across studies. Therefore, the improvement effect of combination training on BMD should be regarded as an exploratory finding. Combination training can serve as a reference for long-term bone health maintenance but requires individualized adjustment in clinical practice based on the patient’s specific circumstances. In addition, two studies reported OLS, involving 136 patients. Trampoline was identified as an effective intervention, as postural control can be continuously challenged by the unstable surface of a trampoline, thereby significantly activating core and ankle-stabilizing muscles. This is an underlying mechanism for its significant effect on static balance (Posch et al., 2019; Tay et al., 2019). Nevertheless, extreme caution is needed when applying trampoline to older OP patients. The reason is that the conclusion was based solely on two small-sample studies, with an extremely weak evidence base. OP patients face potential risks of vertebral compression fractures, falls, and joint injuries during trampoline, and insufficient safety data are available on its routine use in vulnerable populations. Therefore, trampoline should not be recommended as an independent or routine intervention and requires further exploration.

Fall incidents were evaluated in seven studies involving 586 patients. RT showed a statistically significant association with reduced fall risk, whose core mechanism lies in rapidly enhancing the absolute strength and explosive power of lower limb and core muscles (da Rosa Orssatto et al., 2019; González-Gálvez et al., 2024). It offers a critical physiological foundation for effective protective responses to unexpected loss of balance (Kitada et al., 2025). Notably, VR and BT + aerobics did not significantly prevent falls. Prevention of falls relies not only on improved balance and cognition but more on maximal muscle strength that supports rapid responses to sudden events. VR may not fully train rapid integration of vestibular sensation and proprioception, while BT + aerobics emphasizes regular movement and lacks stress training for unexpected imbalance scenarios (Sadeghi et al., 2021; Ketterer et al., 2024). As a result, they are both less effective in preventing falls. In clinical practice, RT possesses high cost-effectiveness and can greatly improve symptoms and enhance muscle strength without the need for expensive devices. Moreover, it can also be easily mastered and taught by healthcare providers while being highly adaptable for home use (Ferrari et al., 2024). Designing progressive plans and incorporating home exercises (e.g., resistance bands) should be able to effectively enhance patients’ long-term adherence. Jackson et al. also argued that regardless of the exercise regimen, sufficient progressive RT should be included (Fyfe et al., 2022). In practice, nurses can guide patients with RT as the core component and encourage their families to participate in safe home exercises under professional supervision. In this way, social and family support can be systematically integrated throughout the rehabilitation process. However, the following methodological limitations warrant close attention: First, no closed loop was formed in the network, precluding the inconsistency test. Second, heterogeneity was present in the definition of fall and follow-up durations. Therefore, the preventive effect of RT on falls should be regarded as preliminary evidence rather than a definitive conclusion. Nevertheless, based on the definite physiological basis of muscle strengthening, progressive RT is still recommended as a basic component of all exercise prescriptions. RT can be implemented by progressive home exercise to enhance accessibility.

Our findings aligned with and further expanded the available evidence. Multiple studies and guidelines recommend a multi-modal exercise regimen (Zhao et al., 2017; Brooke-Wavell et al., 2022), consistent with our conclusion that combination training is most effective in improving BMD. However, this Bayesian NMA fully quantified and ranked the efficacy of exercise regimens in older OP patients, achieving more valuable synthesis of evidence for clinical decision-making. Although this study strictly followed the PRISMA-NMA guidelines, some limitations remained. First, heterogeneity was found in exercise regimens regarding training intensity, frequency, and supervision conditions. Its impact on key outcomes (e.g., BBS, TUG) was controllable, but it may amplify bias in indirect comparison of outcomes with sparser network structures (e.g., BMD, OLS, and falls). Second, only BBS was discussed in conjunction with MCID, so whether some statistically significant intervention effects have clinical practical value remains to be verified. Moreover, the vast majority of studies had durations of interventions ≤12 months, precluding understanding of long-term effects and safety. Key indicators for scientific implementation, such as safety, cost, and acceptability, were not reported in the original studies, preventing a comprehensive assessment of the clinical net benefit of different interventions. Finally, this study utilized change-from-baseline values and post-intervention measurements when pooling effect sizes, and follow-up durations varied significantly across studies (ranging from immediate assessment to 12 months). Although the random-effects model partially accounted for inter-study heterogeneity, the potential impact of different follow-up durations and baseline levels on MD comparability remained unresolved. Nevertheless, the robustness of the study conclusions was not undermined by these limitations, and the scope of application and evidence strength were clarified. The core value of this study lies in systematically synthesizing available evidence to identify relative advantages of different interventions in terms of outcome metrics. This guided the direction of future head-to-head comparative RCTs, long-follow-up designs, standardized outcome reporting, and scientific research. Future high-quality RCTs with larger samples and longer follow-up can focus on comparing VR with traditional training, the dose-response relation of RT, and the individualized application of combination training across populations with different fracture risks.

5 Conclusion

This study systematically compared the relative effects of different exercise regimens on older OP patients. The results suggested that the benefits of exercise therapies were protocol-specific. VR might enhance balance, RT might offer advantages in reducing fall risk, and BT + RT + aerobics exhibited potential benefits in improving BMD. However, given the limited number of included studies, sparse network structures, and overlapping CIs of some interventions, current findings should be treated cautiously. This study also had limitations: Due to short follow-up periods, inferences about long-term efficacy, sustainability, and safety of interventions were precluded. Some studies had RoB due to small sample sizes and inadequate blinding. No quantitative analyses were conducted on key indicators for scientific implementation, such as compliance, adverse events, and economic feasibility. Therefore, our findings were only a relative comparison based on available evidence and were insufficient to serve as a basis for developing specific clinical pathways. Future rigorously designed high-quality RCTs with large sample sizes and longer follow-up are needed to further validate the long-term efficacy and safety of different interventions, and identify appropriate populations. Moreover, key indicators for scientific implementation, such as compliance and cost-effectiveness, can be deeply explored to facilitate the translation of interventions from evidence to clinical practice.

References

• 1

  AltaiZ.HayfordC. F.PhillipsA. T. M.MoranJ.ZhaiX.LiewB. X. W. (2024). Lower limb joint loading during high-impact activities: implication for bone health. JBMR Plus8, ziae119. doi: 10.1093/jbmrpl/ziae119

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 2

  AraujoC. G.de SouzaE. S. C. G.LaukkanenJ. A.Fiatarone SinghM.KunutsorS. K.MyersJ.et al. (2022). Successful 10-second one-legged stance performance predicts survival in middle-aged and older individuals. Br. J. Sports Med.56, 975–980. doi: 10.1136/bjsports-2021-105360

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 3

  (2025). World congress on osteoporosis, osteoarthritis and musculoskeletal diseases (WCO-IOF-ESCEO 2025). Aging Clin. Exp. Res.37, 279. doi: 10.1007/s40520-025-03119-z

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 4

  BanefeltJ.TimoshankoJ.SöreskogE.OrtsäterG.MoayyeriA.ÅkessonK. E.et al. (2022). Total hip bone mineral density as an indicator of fracture risk in bisphosphonate-treated patients in a real-world setting. J. Bone Miner Res.37, 52–58. doi: 10.1002/jbmr.4448

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 5

  BarbanchonC.MourauxD.BaudryS. (2024). Repeated exposure to virtual reality decreases reliance on visual inputs for balance control in healthy adults. Hum. Mov Sci.96, 103236. doi: 10.1016/j.humov.2024.103236

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 6

  BendrikR.SundströmB.BrömsK.EmtnerM.KallingsL. V.PetersonM. (2024). One leg testing in hip and knee osteoarthritis: A comparison with a two-leg oriented functional outcome measure and self-reported functional measures. Osteoarthritis Cartilage.32, 937–942. doi: 10.1016/j.joca.2024.03.111

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 7

  Brooke-WavellK.SkeltonD. A.BarkerK. L.ClarkE. M.De BiaseS.ArnoldS.et al. (2022). Strong, steady and straight: UK consensus statement on physical activity and exercise for osteoporosis. Br. J. Sports Med.56, 837–846. doi: 10.1136/bjsports-2021-104634

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 8

  ChenY.DaiF.HuangS.QiD.PengC.ZhangA.et al. (2025). Global, regional, and national burden of falls among older adults: findings from the Global Burden of Disease Study 2021 and Projections to 2040. NPJ Aging.11, 85. doi: 10.1038/s41514-025-00275-4

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 9

  Cortés-PérezI.Díaz-FernándezÁOsuna-PérezM. C.García-LópezH.Romero-Del-ReyR.Obrero-GaitánE. (2025). Virtual reality-based therapy improves balance, quality of life, and mitigates pain and fear of falling in women with bone mineral density loss: A meta-analysis of randomized controlled trials. Life (Basel).15, 1654. doi: 10.3390/life15111654

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 10

  da Rosa OrssattoL. B.de la Rocha FreitasC.ShieldA. J.Silveira PintoR.TrajanoG. S. (2019). Effects of resistance training concentric velocity on older adults’ functional capacity: A systematic review and meta-analysis of randomised trials. Exp. Gerontol.127, 110731. doi: 10.1016/j.exger.2019.110731

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 11

  DengY.TangZ.YangZ.ChaiQ.LuW.CaiY.et al. (2024). Comparing the effects of aquatic-based exercise and land-based exercise on balance in older adults: a systematic review and meta-analysis. Eur. Rev. Aging Phys. Act.21, 13. doi: 10.1186/s11556-024-00349-4

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 12

  DizdarM.IrdeselJ. F.DizdarO. S.TopsaçM. (2018). Effects of balance-coordination, strengthening, and aerobic exercises to prevent falls in postmenopausal patients with osteoporosis: A 6-month randomized parallel prospective study. J. Aging Phys. Act.26, 41–51. doi: 10.1123/japa.2016-0284

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 13

  DonathL.RösslerR.FaudeO. (2016). Effects of virtual reality training (Exergaming) compared to alternative exercise training and passive control on standing balance and functional mobility in healthy community-dwelling seniors: A meta-analytical review. Sports Med.46, 1293–1309. doi: 10.1007/s40279-016-0485-1

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 14

  DownsS.MarquezJ.ChiarelliP. (2013). The Berg Balance Scale has high intra- and inter-rater reliability but absolute reliability varies across the scale: a systematic review. J. Physiother.59, 93–99. doi: 10.1016/s1836-9553(13)70161-9

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 15

  FerrariL.BochicchioG.BottariA.ScartonA.CavedonV.MilaneseC.et al. (2024). Feasibility and effectiveness of a 6-month, home-based, resistance exercise delivered by a remote technological solution in healthy older adults. Arch. Gerontol Geriatr.127, 105559. doi: 10.1016/j.archger.2024.105559

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 16

  FilipoviĆT. N.LazoviĆM. P.BackoviĆA. N.FilipoviĆA. N.IgnjatoviĆA. M.DimitrijeviĆS. S.et al. (2021). A 12-week exercise program improves functional status in postmenopausal osteoporotic women: randomized controlled study. Eur. J. Phys. Rehabil. Med.57, 120–130. doi: 10.23736/s1973-9087.20.06149-3

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 17

  FyfeJ. J.HamiltonD. L.DalyR. M. (2022). Minimal-dose resistance training for improving muscle mass, strength, and function: A narrative review of current evidence and practical considerations. Sports Med.52, 463–479. doi: 10.1007/s40279-021-01605-8

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 18

  GaoH.ZhaoY.ZhaoL.WangZ.YanK.GaoB.et al. (2023). The role of oxidative stress in multiple exercise-regulated bone homeostasis. Aging Dis.14, 1555–1582. doi: 10.14336/ad.2023.0223

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 19

  GasparA. G. M.LapãoL. V. (2021). eHealth for addressing balance disorders in the elderly: systematic review. J. Med. Internet Res.23, e22215. doi: 10.2196/22215

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 20

  Gatenio-HeflingO.Tzemah-ShaharR.AsrafK.DilianO.GilE.AgmonM. (2025). Revisiting the “Timed Up and Go” test: a 12-s cut-off can predict Hospitalization Associated Functional Decline in older adults. Geroscience.47, 1039–1048. doi: 10.1007/s11357-024-01280-3

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 21

  GBD 2021 Low Bone Mineral Density Collaborators. (2025). The global, regional, and national burden attributable to low bone mineral density, 1990-2020: an analysis of a modifiable risk factor from the Global Burden of Disease Study 2021. Lancet Rheumatol.7, e873–ee94. doi: 10.1016/s2665-9913(25)00105-5

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 22

  GiangregorioL. M.PonzanoM. (2022). Exercise and physical activity in individuals at risk of fracture. Best Pract. Res. Clin. Endocrinol. Metab.36, 101613. doi: 10.1016/j.beem.2021.101613

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 23

  GilS.KirwanJ. P.MuraiI. H.DantasW. S.Merege-FilhoC. A. A.GhoshS.et al. (2021). A randomized clinical trial on the effects of exercise on muscle remodelling following bariatric surgery. J. Cachexia Sarcopenia Muscle.12, 1440–1455. doi: 10.1002/jcsm.12815

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 24

  González-GálvezN.Moreno-TorresJ. M.Vaquero-CristóbalR. (2024). Resistance training effects on healthy postmenopausal women: a systematic review with meta-analysis. Climacteric.27, 296–304. doi: 10.1080/13697137.2024.2310521

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 25

  HakestadK. A.TorstveitM. K.NordslettenL.RisbergM. A. (2015). Effect of exercises with weight vests and a patient education programme for women with osteopenia and a healed wrist fracture: a randomized, controlled trial of the OsteoACTIVE programme. BMC Musculoskelet Disord.16, 352. doi: 10.1186/s12891-015-0811-z

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 26

  HaoQ. H.QiuM. M.WangJ.TuY.LvZ. H.ZhuT. M. (2025). The effect of lower limb rehabilitation robot on lower limb -motor function in stroke patients: a systematic review and meta-analysis. Syst. Rev.14, 70. doi: 10.1186/s13643-025-02759-6

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 27

  HejaziK.AskariR.HofmeisterM. (2022). Effects of physical exercise on bone mineral density in older postmenopausal women: a systematic review and meta-analysis of randomized controlled trials. Arch. Osteoporos.17, 102. doi: 10.1007/s11657-022-01140-7

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 28

  HuX.WangZ.WangW.CuiP.KongC.ChenX.et al. (2024). Irisin as an agent for protecting against osteoporosis: A review of the current mechanisms and pathways. J. Adv. Res.62, 175–186. doi: 10.1016/j.jare.2023.09.001

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 29

  IntziegianniK.SarensM.TsivitanidouO.NisiotisL.KovacsK.NagyE.et al. (2024). Investigation of neuromuscular activation in older female adults during a dynamic and challenging virtual reality activity: A cross-sectional study. J. Funct. Morphol Kinesiol9, 143. doi: 10.3390/jfmk9030143

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 30

  KettererJ.GollhoferA.RinghofS.AssländerL.GranacherU.GehringD. (2024). Effects of balance training with visual input manipulations on balance performance and sensory integration in healthy young adults: a randomized controlled trial. Sci. Rep.14, 28589. doi: 10.1038/s41598-024-79736-x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 31

  KitadaT.UmegakiH.AkimaH.IshidaK.NakatochiM.InoueA.et al. (2025). Effects of weight-bearing resistance training with explosive motions on the rate of force development in community-dwelling older adults: A randomized controlled trial. J. Phys. Act Health22, 706–715. doi: 10.1123/jpah.2024-0431

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 32

  KobayashiS.MiyataK.TamuraS.TakedaR.IwamotoH. (2024). Minimal important change in the Berg Balance Scale in older women with vertebral compression fractures: A retrospective multicenter study. Pm R.16, 715–722. doi: 10.1002/pmrj.13092

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 33

  KonakH. E.KibarS.ErginE. S. (2016). The effect of single-task and dual-task balance exercise programs on balance performance in adults with osteoporosis: a randomized controlled preliminary trial. Osteoporos Int.27, 3271–3278. doi: 10.1007/s00198-016-3644-1

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 34

  KrohnM.RintalaA.ImmonenJ.SjögrenT. (2024). The effectiveness of therapeutic exercise interventions with virtual reality on balance and walking among persons with chronic stroke: systematic review, meta-analysis, and meta-regression of randomized controlled trials. J. Med. Internet Res.26, e59136. doi: 10.2196/59136

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 35

  MadureiraM. M.TakayamaL.GallinaroA. L.CaparboV. F.CostaR. A.PereiraR. M. (2007). Balance training program is highly effective in improving functional status and reducing the risk of falls in elderly women with osteoporosis: a randomized controlled trial. Osteoporos Int.18, 419–425. doi: 10.1007/s00198-006-0252-5

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 36

  MewaraN.ChhajedB. K. B. (2024). Evaluating the impact of A community based, peer-led exercise program on the prevention of osteoporotic fractures among elderly women. J. Popul Ther. Clin. Pharmacol.8, 1139–1147. doi: 10.53555/jptcp.v31i8.7540

  • CrossRef
  • Google Scholar

• 37

  MikoI.SzerbI.SzerbA.BenderT.PoorG. (2018). Effect of a balance-training programme on postural balance, aerobic capacity and frequency of falls in women with osteoporosis: A randomized controlled trial. J. Rehabil. Med.50, 542–547. doi: 10.2340/16501977-2349

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 38

  MurtezaniA.NevzatiA.IbraimiZ.SllamnikuS.MekaV. S.AbaziN. (2014). The effect of land versus aquatic exercise program on bone mineral density and physical function in postmenopausal women with osteoporosis: a randomized controlled trial. Ortop Traumatol Rehabil.16, 319–325. doi: 10.5604/15093492.1112533

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 39

  OksuzS.UnalE. (2017). The effect of the clinical pilates exercises on kinesiophobia and other symptoms related to osteoporosis: Randomised controlled trial. Complement Ther. Clin. Pract.26, 68–72. doi: 10.1016/j.ctcp.2016.12.001

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 40

  OlsenC. F.BerglandA. (2014). The effect of exercise and education on fear of falling in elderly women with osteoporosis and a history of vertebral fracture: results of a randomized controlled trial. Osteoporos Int.25, 2017–2025. doi: 10.1007/s00198-014-2724-3

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 41

  ParamentoM.PassarottoE.AgostiniM.FormaggioE.ContessaP.BerlingieriC.et al. (2025). EEG markers of improved proprioception and posture in older women with osteoporosis. IEEE Trans. Neural Syst. Rehabil. Eng.33, 3837–3844. doi: 10.1109/tnsre.2025.3609562

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 42

  PoschM.SchranzA.LenerM.TecklenburgK.BurtscherM.RuedlG.et al. (2019). Effectiveness of a mini-trampoline training program on balance and functional mobility, gait performance, strength, fear of falling and bone mineral density in older women with osteopenia. Clin. Interv Aging.14, 2281–2293. doi: 10.2147/cia.S230008

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 43

  PrasadT. N.FloraS. S.BhadadaS. K.YadavU.PalR. (2025). Anti-resorptive and anabolic therapies improve Falls Risk Assessment Score (FRAS) in postmenopausal women with type 2 diabetes mellitus. Osteoporos Int.36, 1633–1639. doi: 10.1007/s00198-025-07599-7

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 44

  RamosL. A. X.RodriguesF. T. M.ShirahigeL.de Fátima Alcântara BarrosM.de CarvalhoA. G. C.GuerinoM. R.et al. (2019). A single whole body vibration session influences quadriceps muscle strength, functional mobility and balance of elderly with osteopenia and/or osteoporosis? Pragmatic clinical trial. J. Diabetes Metab. Disord.18, 73–80. doi: 10.1007/s40200-019-00392-4

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 45

  RezaeiM. K.TorkamanG.BahramiF.BayatN. (2023). The effect of six week virtual reality training on the improvement of functional balance in women with type-I osteoporosis: A preliminary study. Sport Sci. Health19, 185–194. doi: 10.1007/s11332-022-01018-8

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 46

  RuzeneJ. R. S.MorcelliM. H.NavegaM. T. (2016). Equilibrium in women with osteoporosis submitted to balance training with and without an oscillating vibratory pole. J. Bodyw Mov Ther.20, 35–41. doi: 10.1016/j.jbmt.2015.03.005

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 47

  SadeghiH.JehuD. A.DaneshjooA.ShakoorE.RazeghiM.AmaniA.et al. (2021). Effects of 8 weeks of balance training, virtual reality training, and combined exercise on lower limb muscle strength, balance, and functional mobility among older men: A randomized controlled trial. Sports Health13, 606–612. doi: 10.1177/1941738120986803

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 48

  SahniP.NievesJ. W. (2019). Determining the effects of a 4-week structured strength and flexibility exercise program on functional status of subjects with osteoporosis. Hss J.15, 241–246. doi: 10.1007/s11420-019-09686-y

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 49

  SangtarashF.ShadmehrA.SadeghiA.FereydounniaS. (2024). Comparison between the effects of pulsed electromagnetic fields versus resistance exercises on muscle strength, thoracic kyphosis, balance, fear of falling and quality of life in postmenopausal osteoporosis women. Iran Red Crescent Med. J.26, 1–9. doi: 10.22034/ircmj.2024.473059.1388

  • CrossRef
  • Google Scholar

• 50

  SañudoB.Reverte-PagolaG.MaherC.GodinoJ.CarrascoL.OviedoM.et al. (2025). Effectiveness of an mHealth-based impact exercise program for bone health in postmenopausal women: a randomised controlled trial protocol. BMC Public Health25, 2426. doi: 10.1186/s12889-025-22576-9

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 51

  SmuldersE.WeerdesteynV.GroenB. E.DuysensJ.EijsboutsA.LaanR.et al. (2010). Efficacy of a short multidisciplinary falls prevention program for elderly persons with osteoporosis and a fall history: a randomized controlled trial. Arch. Phys. Med. Rehabil.91, 1705–1711. doi: 10.1016/j.apmr.2010.08.004

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 52

  StanghelleB.BentzenH.GiangregorioL.PrippA. H.SkeltonD. A.BerglandA. (2020). Effects of a resistance and balance exercise programme on physical fitness, health-related quality of life and fear of falling in older women with osteoporosis and vertebral fracture: a randomized controlled trial. Osteoporos Int.31, 1069–1078. doi: 10.1007/s00198-019-05256-4

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 53

  SwanenburgJ.de BruinE. D.StauffacherM.MulderT.UebelhartD. (2007). Effects of exercise and nutrition on postural balance and risk of falling in elderly people with decreased bone mineral density: randomized controlled trial pilot study. Clin. Rehabil.21, 523–534. doi: 10.1177/0269215507075206

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 54

  TamuraS.MiyataK.KobayashiS.TakedaR.IwamotoH. (2022). Minimal clinically important difference of the Berg Balance Scale score in older adults with hip fractures. Disabil. Rehabil.44, 6432–6437. doi: 10.1080/09638288.2021.1962993

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 55

  TaoL.WangJ.WangK.LiuQ.LiH.XuS.et al. (2024). Exerkine FNDC5/irisin-enriched exosomes promote proliferation and inhibit ferroptosis of osteoblasts through interaction with Caveolin-1. Aging Cell.23, e14181. doi: 10.1111/acel.14181

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 56

  TayZ. M.LinW. H.KeeY. H.KongP. W. (2019). Trampoline versus resistance training in young adults: effects on knee muscles strength and balance. Res. Q Exerc Sport.90, 452–460. doi: 10.1080/02701367.2019.1616045

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 57

  TeixeiraL. E.SilvaK. N.ImotoA. M.TeixeiraT. J.KayoA. H.Montenegro-RodriguesR.et al. (2010). Progressive load training for the quadriceps muscle associated with proprioception exercises for the prevention of falls in postmenopausal women with osteoporosis: a randomized controlled trial. Osteoporos Int.21, 589–596. doi: 10.1007/s00198-009-1002-2

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 58

  VaillantJ.VuillermeN.MartignéP.Caillat-MiousseJ. L.ParisotJ.NougierV.et al. (2006). Balance, aging, and osteoporosis: effects of cognitive exercises combined with physiotherapy. Joint Bone Spine.73, 414–418. doi: 10.1016/j.jbspin.2005.07.003

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 59

  VonstadE. K.BachK.VereijkenB.SuX.NilsenJ. H. (2022). Performance of machine learning models in estimation of ground reaction forces during balance exergaming. J. Neuroeng Rehabil.19, 18. doi: 10.1186/s12984-022-00998-5

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 60

  WangK.HuS. (2023). The synergistic effects of polyphenols and intestinal microbiota on osteoporosis. Front. Immunol.14. doi: 10.3389/fimmu.2023.1285621

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 61

  WuJ.HuM.JiangH.MaJ.XieC.ZhangZ.et al. (2023). Endothelial cell-derived lactate triggers bone mesenchymal stem cell histone lactylation to attenuate osteoporosis. Adv. Sci. (Weinh).10, e2301300. doi: 10.1002/advs.202301300

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 62

  XiongZ.SongY.WuJ.LuW.XiongJ.ZhuM.et al. (2025). Bidirectional regulation factor of bone marrow mesenchymal stromal cells differentiation: a focus on bone-fat balance in osteoporosis. Stem Cell Res. Ther.17, 14. doi: 10.1186/s13287-025-04838-w

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 63

  YilmazN.KösehasanoğullarıM. (2024). The effectiveness of virtual reality exercise games on balance functions and fear of falling in women with osteoporosis. Rheumatol Int.44, 1071–1076. doi: 10.1007/s00296-024-05569-6

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 64

  ZhangY. W.CaoM. M.LiY. J.ChenX. X.YuQ.RuiY. F. (2022). A narrative review of the moderating effects and repercussion of exercise intervention on osteoporosis: ingenious involvement of gut microbiota and its metabolites. J. Transl. Med.20, 490. doi: 10.1186/s12967-022-03700-4

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 65

  ZhangF.WangZ.SuH.ZhaoH.LuW.ZhouW.et al. (2022). Effect of a home-based resistance exercise program in elderly participants with osteoporosis: a randomized controlled trial. Osteoporos Int.33, 1937–1947. doi: 10.1007/s00198-022-06456-1

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 66

  ZhangY.ZhaoW.WanC.WuX.HuangJ.WangX.et al. (2024). Exoskeleton rehabilitation robot training for balance and lower limb function in sub-acute stroke patients: a pilot, randomized controlled trial. J. Neuroeng Rehabil.21, 98. doi: 10.1186/s12984-024-01391-0

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 67

  ZhaoR.ZhangM.ZhangQ. (2017). The effectiveness of combined exercise interventions for preventing postmenopausal bone loss: A systematic review and meta-analysis. J. Orthop Sports Phys. Ther.47, 241–251. doi: 10.2519/jospt.2017.6969

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 68

  ZhaoR.ZhaoX.GuanJ.ZhangC.ZhuK. (2023). The effect of virtual reality technology on anti-fall ability and bone mineral density of the elderly with osteoporosis in an elderly care institution. Eur. J. Med. Res.28, 204. doi: 10.1186/s40001-023-01165-9

  • Pubmed Abstract
  • CrossRef
  • Google Scholar